article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
pulsar psr b1913 + 16 ( psr j1915 + 1606 ) was the first binary pulsar to be discovered @xcite .
its 7.75-hr binary period and 300 km s@xmath2 projected orbital velocity suggested that it would exhibit a rich set of potentially measurable relativistic effects , and that has turned out to be the case .
subsequent studies have shown that the pulsar s companion must also be a neutron star . with tidal and rotational stellar distortions eliminated as complicating factors for any dynamical analysis , the system is an essentially clean laboratory for testing relativistic gravity . in this respect ,
the largest remaining complications depend on reference - frame accelerations related to the structure and dynamics of our galaxy . over the past 35 years , we and our colleagues have timed psr b1913 + 16 at arecibo observatory with steadily improving equipment and analysis techniques . among the best known results are measurement of the general relativistic advance of periastron at a rate some 35,000 times that of mercury in the solar system @xcite ; the effect of gravitational radiation damping , causing a measurable rate of orbital decay @xcite ; and detection of changes in the pulse shape , resulting from geodetic spin precession @xcite .
our results continue to be fully consistent with general relativity and have placed strong constraints on alternative , previously viable , relativistic theories of gravity @xcite .
the purpose of this paper is to provide our latest analysis of the pulsar and its orbit , based on timing observations from 1974 through 2006 .
our highest quality dataset consists of 7650 five - minute integrations of the pulse profile at wavelengths near 20 cm , each subsequently reduced to a topocentric time - of - arrival ( toa ) measurement at arecibo observatory .
these data span the years 1981 through 2006 ; earlier measurements are consistent with these but ( owing to their much larger uncertainties ) have little effect on the final results . in 2004
we made a special effort to observe regularly throughout the year , so as to improve quality of the astrometric results .
parameters of the observing equipment and associated measurement uncertainties are presented in table [ table : obsys ] .
further details of the princeton mark i observing system are given by @xcite , while the mark ii and iii are described in @xcite .
the latest system , the wideband arecibo pulsar processor ( wapp ) , consists of four spectrometers optimized for pulsar work , each having 512 frequency channels across a 100 mhz bandwidth and synchronously accumulating the pulsar signal into 1024 phase bins across the full pulsar period .
data were analyzed with the program tempo , using the jpl de405 solar system ephemeris and the relativistic timing model of @xcite .
the damour - deruelle approach is particularly desirable because it leaves open the question of the correct relativistic theory of gravity : one solves for phenomenological parameters whose precise relation to a gravitational theory can be investigated later .
further discussion of the damour - deruelle model and its application to binary pulsars in tempo can be found in @xcite .
important quantities determined in this way include those related to the pulsar s celestial position , spin , and orbit .
the first category includes right ascension @xmath7 , declination @xmath8 , proper motions @xmath9 and @xmath10 , epoch @xmath11 , pulse repetition frequency @xmath12 , spindown rate @xmath13 , and a glitch epoch and frequency discontinuity @xmath14 .
fitted orbital parameters include five keplerian quantities : the projected semimajor axis of the pulsar orbit @xmath15 , eccentricity @xmath16 , epoch of periastron passage @xmath17 , period @xmath18 , and longitude of periastron @xmath19 ; and the following relativistic or `` post - keplerian '' parameters : average rate of periastron advance @xmath20 , variations in gravitational redshift and time dilation @xmath21 , and orbital period derivative @xmath22 . as with most other pulsars that have been timed carefully over several decades , a number of `` nuisance parameters '' must also be measured to account for unmodeled long - term timing irregularities .
these extra fitted terms have no clear physical interpretation beyond being somehow related to the poorly understood structure and dynamics of the spinning neutron star .
their mathematical form is somewhat arbitrary ; as described further below , we have experimented with a number of higher - order time derivatives of the pulsar rotation frequency , and one additional frequency discontinuity .
the astrometric and basic spin parameters determined from the full data set are listed in table [ table : astromfits ] .
a significant result for proper motion has been obtained for the first time .
while the pulsar position is determined from the annual variation of toas as the earth moves about its orbit , proper motion is determined from a secular variation of this annual signal . timing noise or other systematic effects
can contaminate such measurements .
we now believe that the previously reported value of psr b1913 + 16 s proper motion @xcite was biased by timing noise , a problem which was exacerbated by the concentrated but @xmath23biennially spaced observing campaigns frequently employed for this source . to circumvent these problems , we observed psr b1913 + 16 several times over the course of calendar year 2004 to achieve thorough data coverage around the earth s orbit . by merging the 2004 data with observations made in 1985 - 1988 , which also had good coverage throughout those years , we have now obtained a robust measurement of the pulsar s proper motion .
see section [ sec : vel ] for further analyses of the proper motion result . as in other pulsars , the measured spindown rate @xmath24 is assumed to be the result of an electromagnetic braking torque on the spinning , strongly magnetized neutron star .
its value is deterministic and ( at the level of accuracy quoted in table 2 ) independent of the span of data over which it is fitted .
in addition , the pulsar experienced a well - defined `` classical '' glitch in may 2003 .
the magnitude @xmath25 of the event is smaller than almost all seen in the population of normal ( not recycled ) pulsars , with only some of the glitches in psr b0355 + 54 having a comparably small magnitude @xcite .
ours is only the second glitch to be detected in a recycled pulsar , with the other ( the smallest known among all pulsars ) occurring in the millisecond pulsar psr b1821@xmath2624 in globular cluster m28 @xcite . the remaining timing parameters higher - order derivatives @xmath27 , @xmath28 , , and ( in some fits ) another small frequency discontinuity in mid-1992 were introduced to the timing solution in an _ ad hoc _ manner , in order to `` whiten '' the remaining post - fit residuals .
although we offer no clear or unique physical interpretation for these parameters , their combined effects are almost certainly a consequence of stochastic timing - noise processes in the neutron star interior @xcite . unlike the case of the well - fitted may 2003 glitch , the values of these fitted parameters are not independent of the data span analyzed and can not be expected to extrapolate the timing behavior accurately to future epochs .
identifying the mid-1992 behavior as a discrete event is highly uncertain , in part because of coarse sampling around that time .
our data can be fit almost as well by introducing several additional frequency derivatives instead of a second discrete event .
similar results were obtained by fitting multiple harmonically related sinusoids to the timing noise with the routine `` fitwaves '' @xcite of program tempo2 @xcite , instead of the multiple frequency derivatives described above .
since we are unable to converge on a unique glitch parameter solution for the mid-1992 behavior , we do not include this event in table [ table : astromfits ] .
fitted orbital parameters for psr b1913 + 16 are listed in table [ table : orbfits ] . as we have emphasized before @xcite
, values for each of the damour - deruelle post - keplerian parameters expected in general relativity can be expressed in terms of the keplerian parameters and the initially unknown masses of the pulsar and its companion , @xmath29 and @xmath30 .
the appropriate expressions for @xmath20 and @xmath21 are @xmath31^{2/3 } { \rm deg~yr}^{-1 } , \\
\gamma & = & g^{2/3 } \
c^{-2 } \ e \
( p_b / 2 \pi)^{1/3 } \ m_2 \ ( m_1 + 2 m_2 ) \ ( m_1 + m_2)^{-4/3 } \nonumber \\ & = & 0.002936679(2 ) \ \left[\frac{m_2 \ ( m_1 + 2 m_2 ) \ ( m_1 + m_2)^{-4/3}}{m_\sun^{2/3}}\right]~{\rm s}.\end{aligned}\ ] ] in the second line of each equation
we have substituted values for @xmath18 and @xmath16 from table [ table : orbfits ] , and used the constants @xmath32 s and 1 julian yr @xmath33 s. the figures in parentheses represent uncertainties in the last quoted digit , determined by propagating the uncertainties listed in table [ table : orbfits ] . in each case , the uncertainties are dominated by the experimental uncertainty in orbital eccentricity , @xmath16 .
( 1 ) may be solved for the total mass of the psr b1913 + 16 system , yielding @xmath34 . the additional constraint privided by eq .
( 2 ) permits a solution for each star s mass individually , @xmath4 and @xmath35 . as far as we know , these are the most accurately determined stellar masses outside the solar system .
it is interesting to note that since the value of newton s constant @xmath36 is known to a fractional accuracy of only @xmath37 , @xmath38 can be expressed more accurately in solar masses than in grams . according to general relativity
a binary star system should radiate energy in the form of gravitational waves .
peters and matthews ( 1963 ) showed that the resulting rate of change in orbital period should be @xmath39 \end{aligned}\ ] ] inserting values obtained for @xmath29 and @xmath30 and propagating uncertainties appropriately , we obtain the general relativistic predicted value @xmath40 equations ( 3 ) and ( 4 ) apply in the orbiting system s reference frame .
relative acceleration of that frame with respect to the solar system barycenter will cause a small additional contribution to the observed @xmath41 .
@xcite presented a detailed discussion of this effect and other possible contributions to @xmath41 .
recent progress in determining the galactic - structure parameters allows us to update the relevant quantities and compute a new value for the kinematic correction to @xmath41 . using @xmath42 kpc for the distance to the galactic center and @xmath43 km s@xmath2 for the circular velocity of the local standard of rest @xcite , and @xmath44 kpc for the pulsar distance @xcite
, we obtain the kinematic contribution , @xmath45 : @xmath46 thus , we find the ratio of observed to predicted rate of orbital period decay to be @xmath47 agreement between the observed orbital decay and the general relativistic prediction is illustrated in fig .
[ fig : parabola ] , which shows how excess orbital phase ( relative to an unchanging orbit ) has accumulated since the pulsar s discovery in 1974 .
we note that the overall experimental uncertainty embodied in eq .
( 6 ) is now dominated by uncertainties in the galactic parameters and pulsar distance , not the pulsar timing measurements .
even better agreement between observed and expected values of @xmath41 would be obtained if the true value of @xmath48 or @xmath49 were slightly smaller , or @xmath50 slightly larger .
for example , observed and expected values agree if @xmath51 kpc , which is within the @xcite error envelope .
it will be interesting to see whether improved future estimates of these quantities will show one or more of these conditions to be true .
two other relativistic phenomena are potentially measurable in the psr b1913 + 16 system : geodetic precession and gravitational propagation delay .
spin - orbit coupling should cause the pulsar s spin axis to precess @xcite , which should lead to observable pulse shape changes . @xcite
first detected such changes , which were observed and modeled further by @xcite .
@xcite and @xcite found that the pulsar beam is elongated in the latitude direction and becomes wider in longitude with increasing distance from the beam axis in latitude .
these models suggest that in the next decade or so , precession may move the beam far enough that the pulsar will become unobservable from earth for some decades , before eventually returning to view .
[ sec : propresults ] the excess propagation delay @xcite caused by passage of pulsar signals through the curved spacetime of the companion is largest at the pulsar s superior conjunction .
the maximum amplitude varies with time because the impact parameter at superior conjunction depends strongly on the current value of @xmath52 . in this respect
the orbital geometry was particularly unfavorable in the mid-1990s ( see damour & taylor 1992 ) , but in coming years the propagation delay should start to become observable . @xcite
characterize the measurable quantities as range @xmath53 and shape @xmath54 , of the shapiro delay , where @xmath55 is the orbital inclination .
as orbital precession carries our line of sight deeper into the companion s gravitational well , future observations should permit the robust measurement of these two parameters , and hence two additional tests of relativistic theories of gravity @xcite .
our pulsar proper motion measurement ( section [ sec : astroresults ] ) , combined with the distance estimate discussed in section [ sec : radresults ] , corresponds to a transverse velocity ( with respect to the solar system barycenter ) of 75 km@xmath56s@xmath2 with a galactic position angle of @xmath57 ; i.e. , directed 36 degrees above the galactic plane .
the @xmath58 distance uncertainty places similar limits on velocity accuracies .
we can now estimate two components of the pulsar systemic velocity in _ its own _ standard of rest by combining the measured pulsar transverse velocity and distance , the solar motion with respect to our local standard of rest @xcite , and galactic quantities @xmath48 and @xmath50 .
the third component of motion , which is inaccessible via proper motion measurements , lies close to the direction of galactic rotation at the pulsar s position .
the pulsar s galactic planar and polar velocity components relative to its standard of rest are 247 km / s almost directly away from the galactic center and 51 km / s toward the galactic north pole , respectively .
( this is significantly larger than the measured velocity in the solar system barycenter frame because the pulsar s standard of rest velocity fortuitously cancels much of the pulsar s peculiar velocity with respect to it . )
the systemic velocity of b1913 + 16 is significantly larger than other well - measured double neutron star binary system velocities , including the j0737 - 3037 ( transverse velocity 10 km / s ; stairs et al .
2006 ) , j1518 + 4905 ( transverse velocity 25 km / s ; janssen et al .
2008 ) , and b1534 + 12 ( transverse velocity 122 km / s ; thorsett et al .
2005 ) systems .
we have analyzed the full set of arecibo timing data on pulsar b1913 + 16 to derive the best values of all measurable quantities .
a significant proper motion has finally been determined .
a small glitch was observed in the pulsar s timing behavior , the second known glitch in the population of recycled pulsars
. the measured rate of orbital period decay continues to be almost precisely the value predicted by general relativity , providing conclusive evidence for the existence of gravitational radiation .
uncertainties in galactic accelerations now dominate the error budget in @xmath59 , and are likely to do so until the pulsar distance can be measured more accurately .
we expect that the shapiro gravitational propagation delay will yield additional tests of relativistic gravity within a few more years .
damour , t. 2009 , in _ physics of relativistic objects in compact binaries : from birth to coalescence _ , eds . m. colpi , u. moschella , p. casella , a. possenti , & v. gorini ; springer netherlands .
astrophysics and space science library , 359 , 1 ; preprint available at arxiv:0707.0749 [ gr - qc ] lcccccc mark i & 1981 1984 & 16 & 64 & 125 & 20 & 1719 + mark ii & 1985 1989 & 8 & 32 & 125 & 31 & 1015 + mark iii & 1988 2003 & 40 & 32 & 640 & 16 & 2473 + wapp & 2003 2006 & 100 & 512 & 58 & 14 & 2443 + [ table : obsys ] lll @xmath11 ( mjd ) & 52984.0 + @xmath7 ( j2000 ) & @xmath60 & + @xmath8 ( j2000 ) & 1606273871(13 ) & + @xmath9 ( mas yr@xmath2 ) & @xmath261.43(13 ) & + @xmath10 ( mas yr@xmath2 ) & @xmath260.70(13 ) & + @xmath12 ( s@xmath2 ) & 16.94053778563(15 ) & + @xmath13 ( s@xmath61 ) & @xmath262.4761(9 ) @xmath62 & + glitch epoch ( mjd ) & 52770(20 ) + @xmath14 ( s@xmath2 ) & 6.2(2 ) @xmath63 + [ table : astromfits ] ll @xmath64 ( mjd ) & 52144.90097841(4 ) + @xmath15 ( s ) & 2.341782(3 ) + @xmath16 & 0.6171334(5 ) + @xmath18 ( d ) & 0.322997448911(4 ) + @xmath19 ( deg ) & 292.54472(6 ) + @xmath65 ( deg / yr ) & 4.226598(5 ) + @xmath21 ( ms ) & 4.2992(8 ) + @xmath66 & @xmath262.423(1 ) @xmath67 + [ table : orbfits ] | we present results of more than three decades of timing measurements of the first known binary pulsar , psr b1913 + 16 .
like most other pulsars , its rotational behavior over such long time scales is significantly affected by small - scale irregularities not explicitly accounted for in a deterministic model .
nevertheless , the physically important astrometric , spin , and orbital parameters are well determined and well decoupled from the timing noise .
we have determined a significant result for proper motion , @xmath0 , @xmath1 mas yr@xmath2 .
the pulsar exhibited a small timing glitch in may 2003 , with @xmath3 , and a smaller timing peculiarity in mid-1992 .
a relativistic solution for orbital parameters yields improved mass estimates for the pulsar and its companion , @xmath4 and @xmath5 .
the system s orbital period has been decreasing at a rate @xmath6 times that predicted as a result of gravitational radiation damping in general relativity . as we have shown before
, this result provides conclusive evidence for the existence of gravitational radiation as predicted by einstein s theory . |
as the field of quantum chaos @xcite has emerged over the last few decades , much research has been directed towards the quantal counterparts of classically chaotic hamiltonian systems @xcite . amongst these systems , one - dimensional , periodically kicked quantum systems have proved a popular target for study .
the dynamics of these systems , which are described by discrete time evolution operators acting over the period of one or more kicks , often display quantum resonances and antiresonances ; changes in the degree of localization of the dynamics which are strongly dependent on the value of an effective planck constant .
advances in the cooling and trapping of atoms and ions has paved the way for substantial direct experimental investigation of such archetypal quantum - chaotic systems , particularly the @xmath0-kicked rotor @xcite , and also the @xmath0-kicked accelerator @xcite .
another possibility is the @xmath0-kicked harmonic oscillator ( kho ) @xcite ; many kicked atom ( ion ) configurations involve trapping potentials which are to some degree harmonic , making the dynamics of the kho a relevant consideration for a wide variety of experiments .
the classical kho was originally developed as a model for the interaction of charged particles with electromagnetic fields @xcite .
the frequency degeneracy of orbits in the harmonic oscillator renders the kam theorem inapplicable to the kho ( i.e. , all invariant tori of the harmonic oscillator may be destroyed at any finite perturbation strength ) @xcite . in cases where the kicking period rationally divides the natural oscillator period this allows the kho to display exotic non - kam chaos , manifest as interlinked regions of chaotic dynamics forming _
stochastic web _ structures throughout phase space @xcite . the quantum kho has also been studied in considerable depth @xcite ; particularly in the case where it is kicked four times per oscillator period , when it can be related to the kicked harper model @xcite .
depending on the relative alignment of the kicking potential with respect to the trapping potential , the kho can display either quantum resonances ( even kicking potentials ) or quantum antiresonances ( odd kicking potentials ) @xcite .
in this paper , we consider the effects of quantum resonances that may be observed in an atom - optical kho . in section [ sec : aokho ] we outline our model of the atom - optical kho and introduce the kick - to - kick floquet operator . in section [ sec : origin ]
we derive the structure of the quantum resonances and their effects on the dynamics .
we consider the relation between the quantum resonances , the eigenstates of the floquet operator for @xmath1 kicks , and that floquet operator s expansion in terms of phase space displacement operators ; this leads to an alternative , physically intuitive explanation for the occurrence of quantum resonances . in section [ sec : manifestation ] we consider the measurable manifestation of quantum resonant dynamics in the atom - optical kho , including the experimental accessibility of the necessary parameter regimes .
section [ sec : conclusions ] consists of the conclusions , which are then followed by three technical appendices .
we consider a two - level atom ( or ensemble of noninteracting atoms ) or ion , subject to tight radial confinement and relatively loose harmonic axial confinement , and periodically driven by pulses from an off - resonant laser standing wave .
we assume the pulse duration to be sufficiently short for the particle centre of mass displacement during the pulse to be negligible relative to the period of the standing wave ( raman - nath regime ) . to a good approximation
, the excited internal state can be eliminated , and the system can be described by the one - dimensional single particle hamiltonian @xcite @xmath2 describing the centre of mass motion in the axial direction only . here @xmath3 is the particle mass , and @xmath4 the frequency of the harmonic trapping potential ; the kicks are pulses of duration @xmath5 , repeated with period @xmath6 , from the laser standing wave , detuned by @xmath7 from the two - level transition and with rabi frequency @xmath8 , where @xmath9 and @xmath10 is the laser wavelength .
is therefore equal to two photon recoils . ]
we have set the origin of the @xmath11-coordinate to be at the minimum of the harmonic potential , and choose the laser standing wave to be aligned symmetrically about this minimum .
the resulting cosine potential is an even function in hamiltonian ( [ eqn : quantum_hamiltonian ] ) ; hence , we can expect quantum resonances , i.e. , particular parameter regimes associated with ballistic growth in the system energy , to be observable @xcite . shifting the standing wave by @xmath12 yields a sine potential an odd function , producing antiresonances in parameter regimes where quantum resonances occur for a cosine potential .
when the alignment of the laser standing wave with respect to the trapping minimum lies between these extremes , intermediate quantum resonant behaviour can occur @xcite .
it is therefore not essential for the laser standing wave to be aligned perfectly symmetrically in order for quantum resonant behaviour to be observable , although from now on we assume this , for simplicity .
more important is that any kick - to - kick variations in the standing wave alignment be negligible compared to the standing wave period , otherwise the resulting randomness introduced into the dynamics will destroy the desired quantum resonant behaviour .
the time evolution of the quantum kho is described by a floquet ( discrete time evolution ) operator @xmath13 which maps the state of the system at a time immediately prior to one kick onto the state at a time immediately prior to the next .
it is convenient to describe the floquet operator @xmath13 in terms of the harmonic oscillator creation and annihilation operators @xmath14 , @xmath15 , where @xmath16 .
hence , @xmath17/\sqrt{2}\eta^2 } , \label{eqn : floquet_operator}\ ] ] where @xmath18 is the dimensionless kicking period , @xmath19 is the dimensionless kicking strength , and @xmath20 is the lamb - dicke parameter , which characterizes the scale of the harmonic ground state relative to the laser wavelength . the parameters @xmath21 and @xmath22 emerge naturally from a rescaling to dimensionless dynamical variables of the kho classical map @xcite , whereas @xmath23 now assumes the role of an effective planck constant @xcite . we confine our attention to when the kicking period rationally divides the natural period of the oscillator , i.e. , when @xmath24 for coprime natural numbers @xmath25 , @xmath1 . except in the special cases where
@xmath26 or @xmath27 , the phase space of the classical kho is then spanned by a _
stochastic web _ @xcite , i.e. , interconnected regions of chaotic dynamics penetrating through all of phase space . if @xmath28 the classical phase space has a discrete translational ( crystal ) symmetry ; otherwise the stochastic web has a quasicrystal structure .
these crystal symmetries are reflected in the quantum regime through the existence of infinite sets of unitary displacement operators is the conventional complex index of a coherent state , such that @xmath29 and @xmath30 . ]
@xmath31 commuting with @xmath32 , the evolution operator for one natural oscillator period @xcite .
this implies that eigenstates of @xmath32 for @xmath33 have a comparable crystal structure , and , importantly , are consequently _ extended _ in phase space . in the non - trivial crystal cases
@xmath34 when the displacement operators form a two - dimensional crystal lattice in phase space the set of displacements commuting with @xmath32 is @xmath35 , where @xmath36 for arbitrary integer @xmath37 , @xmath38 . ) the translations ( [ eqn : classical_q=4_group ] ) and ( [ eqn : classical_q=36_group ] ) with @xmath39 set to unity @xcite . ]
the general study of quantum resonances in systems with periodic classical phase spaces has been carried out in considerable theoretical detail by dana @xcite .
our intention here is to provide an accessible derivation of the conditions under which quantum resonances occur specifically for the atom - optical kho , as described in sec .
[ sec : aokho ] . with this in mind
, we substitute @xmath24 into ( [ eqn : floquet_operator ] ) , and rearrange @xmath32 into the product @xmath40/\sqrt{2}\eta^2 } e^{-i(\hat{a}^\dagger \hat{a } + 1/2)2\pi jr / q}. \label{eqn : product_rearrangement}\ ] ] we use the identity @xmath41 to transform each term of ( [ eqn : product_rearrangement ] ) , and write @xmath32 as @xmath42 / \sqrt{2 } \eta^2 \right ) , \label{eqn : q - kicks - q - axes}\ ] ] that is , as a sequence of @xmath1 cosine kicks directed along @xmath1 axes equally distributed in angle about the origin in phase space . using the identity @xcite @xmath43 ( the @xmath44 are @xmath45th - order cylindrical bessel functions )
, we now expand @xmath32 in terms of products of unitary displacement operators : @xmath46 taking the commutation relation for the unitary displacement operators @xmath47 @xcite , it follows that for @xmath48 , @xmath49 , then @xmath50 ( i.e. , the displacement operators commute ) whenever @xmath51}{q } \right ) = 2\pi w , \label{eqn : master_resonance_condition}\ ] ] for integer @xmath52 .
if ( [ eqn : master_resonance_condition ] ) is fulfilled for all @xmath53 it follows that all displacement operators in ( [ eqn : ultimate_displacement_expansion ] ) mutually commute , and hence all of the product terms in ( [ eqn : q - kicks - q - axes ] ) must mutually commute . hence , under these conditions , the @xmath54th power of @xmath32 will describe a @xmath1-fold diffraction exactly equivalent to ( [ eqn : q - kicks - q - axes ] ) but with amplified kicking strength @xmath55 : @xmath56 / \sqrt{2 } \eta^2 \right ) .
\label{eqn : many - kick - reduction}\ ] ] this is analogous to the case where the quantum resonance condition is fulfilled for the quantum @xmath0-kicked rotor , when the free evolution operator between kicks collapses to the identity , and the evolution induced by @xmath54 iterations of the floquet operator is equivalent to that caused by a single kick with @xmath54 times the kicking strength @xcite .
the values of @xmath23 for which ( [ eqn : master_resonance_condition ] ) is fulfilled therefore determine when quantum resonances occur in the kho .
in summary , ( [ eqn : master_resonance_condition ] ) is fulfilled for : all values of @xmath57 when @xmath58 ; @xmath57 equal to a multiple of @xmath59 when @xmath60 ; and @xmath57 equal to a multiple of @xmath61 when @xmath62 ( see appendix [ app : derivation ] ) . for @xmath63 it is impossible to satisfy ( [ eqn : master_resonance_condition ] ) . note that , for @xmath58 , ( [ eqn : q - kicks - q - axes ] ) reduces to being exactly equivalent to what one would expect for the quantum @xmath0-kicked rotor fulfilling the conditions necessary for quantum resonance ; the free evolution between kicks has no net effect , and the ( time - periodic ) dynamics can be equivalently described by a single amplified kick . however , the fact that the free evolution between kicks has no net effect is equally true classically ( for @xmath26 this simply means that after one oscillator period between kicks , the oscillator has returned to its original location in phase space ) , and one must therefore conclude that quantum resonances do not occur for @xmath58 , as is consistent with the dynamics independence of the effective planck constant @xmath23 .
consequently the kho displays quantum resonances only if @xmath64 , and for @xmath57 equal to multiples of the principal quantum resonance values : @xmath59 when @xmath60 , and @xmath61 when @xmath65 or @xmath66 . at further rational multiples @xmath67 of the principal quantum resonance values ,
commutativity between kicks depends on the value of @xmath68 in ( [ eqn : master_resonance_condition ] ) , and only occurs for some terms within the summation in ( [ eqn : ultimate_displacement_expansion ] ) , generating higher order quantum resonances . the proportion of terms for which the displacements commute decreases with the denominator @xmath69 , approaching zero as @xmath67 approaches irrationality .
when @xmath64 , the set of unitary displacements in ( [ eqn : ultimate_displacement_expansion ] ) describe a two - dimensional crystal lattice in phase space . in analogy with section [ subsec : crystal_quasicrystal_symmetry ]
we denote this set @xmath70 , where @xmath71 comparing ( [ eqn : quantum_q=4_set ] ) and ( [ eqn : quantum_q=36_set ] ) with ( [ eqn : classical_q=4_group ] ) and ( [ eqn : classical_q=36_group ] ) , one sees that , at the _ principal quantum resonance values _
[ @xmath72 ( @xmath60 ) , or @xmath73 ( @xmath74 ) ] , the set @xmath70 of displacements constituting @xmath32 is identical to the set @xmath35 of translational symmetries of the eigenstates of @xmath32 .
moreover , for @xmath57 equal to rational multiples @xmath67 of these principal values , @xmath75 remains an element of @xmath35 whenever @xmath37 and @xmath38 are integer multiples of @xmath69 .
this ( complete or partial ) equivalence of @xmath70 and @xmath35 offers a physical explanation for quantum resonances in the kho ; at quantum resonance values of @xmath57 the displacements constituting each kick are translational symmetries of @xmath32 s eigenstates , and consequently commute with @xmath32 . in the context of cold ,
trapped atoms or ions , the initial motional state of the atom(s ) or ion in any experiment will generally be as close to a zero - temperature state as is practicable for the specific experimental configuration .
this provides a clear motivation to study the system evolution when the initial condition is a harmonic oscillator ground state . in the case of an initial atomic bose - einstein condensate
, this would describe the initial condensate mode if atom - atom interactions could be neglected , which could potentially be achieved with the aid of magnetic fields to exploit a feshbach resonance @xcite .
coherent states are harmonic oscillator ground states which have been displaced in phase space , and as such can in principle be accessed by displacing an initial ground state in momentum ( e.g. , by application of a bragg pulse ) or position ( e.g. , by suddenly changing the trap minimum ) , possibly in combination with a free evolution subsequent to the displacement . in what follows we therefore consider the evolution of an initial coherent state analytically , which , in the case of quantum resonance , proves to be comparatively tractable .
we begin by applying the identity ( [ eqn : jacobi_anger ] ) to the floquet operator ( [ eqn : floquet_operator ] ) , yielding @xmath76 where we have introduced @xmath77 .
will usually be a positive quantity , as this corresponds to red detuning . ]
it follows that applying @xmath13 to an arbitrary initial coherent state @xmath78 will yield a weighted superposition of regularly spaced coherent states .
more generally , we will show that if the state of the system @xmath79 after @xmath45 applications of ( [ eqn : action_on_expansion_1 ] ) takes the form of a `` lattice state '' @xmath80}_{m , n } { { \hat{d}\left ( i\eta \left[m + n e^{-i2\pi r / q } \right ] \right ) } } e^{-i\pi jr / q } { { |\alpha e^{-i2\pi jr / q}\rangle}},\ ] ] then @xmath81 will take an essentially similar form .
the @xmath82 in ( [ eqn : lattice_expansion_def ] ) are all coherent states , and the @xmath83}_{m , n}$ ] a set of complex coefficients .
applying ( [ eqn : action_on_expansion_1 ] ) to ( [ eqn : lattice_expansion_def ] ) , and using the identities @xcite @xmath84 and @xmath85 , now yields , in the first instance , @xmath86}_{m , n } e^{i k
n \eta^2 \sin(2\pi r / q ) } \\ & \times { { \hat{d}\left ( i\eta \left[(m+k ) e^{-i2\pi r / q } + n e^{-i4\pi r / q } \right ] \right ) } } \\ & \times e^{-i(j+1)r / q } { { |\alpha e^{-i2\pi(j+1)r / q}\rangle}}. \end{split}\ ] ] when @xmath87 , @xmath88 can be written as an integer superposition of unity and @xmath89 , allowing ( [ eqn : action_on_expansion_2 ] ) to be simplified . from now on we assume @xmath90 to illustrate the procedure : @xmath91 using ( [ eqn : vector_relations ] ) to make replacements in ( [ eqn : action_on_expansion_2 ] ) , one obtains @xmath92}_{m , n } e^{i k n
\eta^2 \sin(2\pi /q ) } \\ & \times { { \hat{d}\left ( i\eta \left[-n + ( m+n\xi_{q}+k ) e^{-i2\pi /q } \right ] \right ) } } \\ & \times e^{-i(j+1)/q } { { |\alpha e^{-i2\pi(j+1)/q}\rangle } } , \end{split}\ ] ] where we have introduced @xmath93 effecting a change of indices such that @xmath94 and @xmath95 now gives @xmath96}_{m\xi_{q}+n - k ,- m } e^{-i k m \eta^2 \sin(2\pi /q ) } \\ & \times { { \hat{d}\left ( i\eta \left[m + n e^{-i2\pi /q } \right ] \right)}}e^{-i(j+1)/q } { { |\alpha e^{-i2\pi(j+1)/q}\rangle } } , \end{split}\ ] ] where we have dropped the primes , for convenience . with ( [ eqn : action_on_expansion_4 ] )
we have an expression of the same form as ( [ eqn : lattice_expansion_def ] ) , but for the state @xmath97 , and can obtain the evolution resulting from @xmath13 as a mapping of the lattice coefficients : @xmath98}_{m , n } = \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) m^{[j]}_{m\xi_{q}+n - k ,- m } e^{-i k m
\eta^2 \sin(2\pi /q)}.\ ] ] the coefficient mapping ( [ eqn : coefficient_mapping ] ) fully describes the time evolution of lattice states in general .
the coherent state at the center of the lattice state when @xmath99 can be chosen arbitrarily . hence , choosing an initial set of coefficients @xmath100}_{m , n } = \delta_{m,0 } \delta_{n,0 } , \label{eqn : single_coherent_state}\ ] ] amounts to following the evolution of an initial coherent state @xmath78 .
the general evolution of the coefficients over many kicks described by ( [ eqn : coefficient_mapping ] ) yields an ever - lengthening product of nested sums that is not especially insightful . in the case of quantum resonance , however , @xmath101 , which simplifies the final exponent in ( [ eqn : coefficient_mapping ] ) , allowing a simpler expression for evolution over many kicks to be extracted . in the next section
we follow the evolution of an initial coherent state , for @xmath60 and @xmath102 , by repeatedly applying the mapping ( [ eqn : coefficient_mapping ] ) .
our initial condition is a single coherent state , as described by ( [ eqn : single_coherent_state ] ) .
we first rearrange the indices to obtain @xmath103}_{n - k ,- m } = \delta_{n - k,0 } \delta_{-m,0 } = \delta_{n , k } \delta_{m,0}$ ] , which we substitute into ( [ eqn : coefficient_mapping ] ) , yielding @xmath104}_{m , n } = \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) \delta_{n , k } \delta_{m,0 } e^{-ikm\pi } = i^{n } j_{n}(\zeta ) \delta_{m,0 } e^{-imn\pi}.\ ] ] we now repeat this procedure ; rearranging the indices and using the identity @xmath105 yields @xmath104}_{n - k ,- m}= i^{m } j_{m}(\zeta ) \delta_{n , k } e^{i(n - k)m\pi},\ ] ] which we substitute into ( [ eqn : coefficient_mapping ] ) : @xmath106}_{m , n } & = i^{m } j_{m}(\zeta ) i^{n } j_{n}(\zeta ) e^{-imn\pi } , \label{eqn : m2_final}\\ \rightarrow m^{[2]}_{n - k ,- m } & = i^{n - k } j_{n - k}(\zeta ) i^{m } j_{m}(\zeta)e^{i(n - k)m\pi}. \label{eqn : m2}\end{aligned}\ ] ] we next substitute ( [ eqn : m2 ] ) into ( [ eqn : coefficient_mapping ] ) , accompanied by a change of indices @xmath107 and use of @xmath108 : @xmath109}_{m , n } & = i^{m } j_{m}(\zeta ) i^n e^{imn\pi } \sum_{k=-\infty}^{\infty } j_{k}(\zeta ) j_{n+k}(\zeta ) e^{ik\pi } .
\label{eqn : pre_graf } \end{split}\ ] ] the summation over @xmath37 can be reduced using graf s addition theorem ( appendix [ app : graf ] ) to give @xmath110}_{m , n } = i^{m } j_{m}(\zeta ) i^{n } j_{n}(2\zeta ) e^{imn\pi}.\ ] ] we repeat the process , again with a change of indices @xmath107 followed by the application of graf s theorem : @xmath111}_{n - k ,- m } & = i^{n - k } j_{n - k}(\zeta ) i^{m } j_{m}(2\zeta ) e^{-i(n - k)m\pi},\\ \begin{split } \rightarrow m^{[4]}_{m , n } & = i^{m }
j_{m}(2\zeta ) i^n e^{-imn\pi } \sum_{k=-\infty}^{\infty }
j_{k}(\zeta ) j_{n+k}(\zeta ) e^{ik\pi } \\ & = i^{m } j_{m}(2\zeta ) i^{n } j_{n}(2\zeta ) e^{-imn\pi}. \end{split } \label{eqn : m4_final}\end{aligned}\ ] ] the final expression obtained for @xmath112}_{m , n}$ ] ( [ eqn : m4_final ] ) is identical to that for @xmath113}_{m , n}$ ] ( [ eqn : m2_final ] ) but with @xmath114 consistently replaced by @xmath115 .
further evolution will see such cycles repeated ; the time evolution of the coefficients is entirely contained within the changing arguments of the bessel functions : @xmath116}_{m , n } = ( -1)^{mn } i^m j_{m}(c_m^{[n]}\zeta)i^n j_{n}(c_n^{[n]}\zeta ) , \label{eqn : final_result}\ ] ] where @xmath117 } & = \left\ { \begin{array}{l l } 2^{(n-2)/2 } & \quad \mbox{$n$ even}\\ 2^{(n-3)/2 } & \quad \mbox{$n$ odd } \end{array } , \right .
\\ c_n^{[n ] } & = \left\ { \begin{array}{l l } 2^{(n-2)/2 } & \quad \mbox{$n$ even}\\ 2^{(n-1)/2 } & \quad \mbox{$n$ odd } \end{array}. \right.\end{aligned}\ ] ] as shown in fig .
[ fig : lattice_phases ] , in the case of quantum resonance the phases of the lattice state coefficients [ eq . ( [ eqn : final_result ] ) ]
take an extremely simple form , which repeats regularly across all of phase space .
we emphasize that this holds for any initial coherent state .
we have chosen the case @xmath118 for simplicity , however the above analysis holds for all @xmath119 . for more general @xmath39
, each application of graf s theorem introduces a @xmath37-dependent phase ( see appendix [ app : graf ] ) , preventing an end result as straightforward as ( [ eqn : final_result ] ) . as a consequence , no such regularity of the phases , as shown in fig .
[ fig : lattice_phases ] , is apparent for non - resonant values of @xmath23 beyond the first two kicks . } _ { m , n}$ ] as given by ( [ eqn : final_result ] ) , which describe the state of a kho ( with @xmath60 ) after the application of @xmath120 ( @xmath121 ) kicks to an initial coherent state , for the quantum resonant case @xmath102 .
the phases shown neglect the parity of the bessel functions in the coefficient [ that is , we actually plot @xmath122}_{m , n } / j_{m}(c_m^{[n]}\zeta ) j_{n}(c_n^{[n]}\zeta ) = ( -1)^{mn}i^{m+n}$ ] ] so that the pattern remains constant for all @xmath123 . ] in these cases the locations in phase space of the coherent states making up the lattice states ( [ eqn : lattice_expansion_def ] ) are defined in terms of @xmath124 and @xmath125 , which represent two non - orthogonal directions in the complex plane .
this non - orthogonality is necessary to derive the coefficient mapping ( [ eqn : coefficient_mapping ] ) , but it adds complexity to the evolution of the coefficients for a coherent state initial condition .
in contrast to the case @xmath60 , where all coefficients @xmath83}_{m , n}$ ] are a direct function of @xmath126 and @xmath127 at quantum resonance , in the cases @xmath65 and @xmath66 the coefficients @xmath83}_{m , n}$ ] for @xmath128 can only be expressed as a summation over products of three bessel functions , even at quantum resonance . despite this added complexity ,
the cases @xmath65 and @xmath66 remain tractable at quantum resonance .
there is a three - step cycle in the coefficients , analogous to the two - step cycle in the coefficients occurring when @xmath60 , with the coefficients @xmath129}_{m , n}$ ] being equal to @xmath130}_{m , n}$ ] but with the arguments of all bessel functions doubled . a derivation for the case @xmath66 is given in appendix [ app : evo ] .
functions @xmath131 @xcite for the kho ( with @xmath60 and @xmath132 ) evolved from the harmonic oscillator ground state .
we contrast the non - resonant case , where @xmath133 [ @xmath134 is the golden ratio ] for ( a ) @xmath135 kicks and ( b ) @xmath136 kicks , with the resonant case @xmath137 for ( c ) @xmath135 and ( d ) @xmath136 .
although @xmath138 close to the origin in ( b ) , our scale is fixed at this maximum to aid comparison between the images . ] as for the atom - optical @xmath0-kicked rotor @xcite or @xmath0-kicked accelerator @xcite systems , the primary observable effect of the quantum resonances in the kho is an enhanced absorption of energy from the kicking potential .
this is visible as an enhanced rate of expansion of the oscillator s state through phase space , as can be seen in fig .
[ fig : q - functions ] , which shows @xmath139 functions ( husimi distributions ) @xcite @xmath140 evolved from an initial harmonic oscillator ground state under resonant and non - resonant conditions .
the evident symmetry between the position and momentum quadratures of the @xmath139 functions in fig . [ fig : q - functions ] mean _ in situ _ measurement of the position density or time - of - flight determination of the momentum density would both provide equally effective signatures of quantum resonant evolution , and the @xmath139 function itself may also in principle be determined directly @xcite .
note , however , that in any experiment the harmonic potential can not be infinite in extent ; after a large number of kicks the state of the oscillator will extend beyond the harmonic region .
consequently , the subsequent time evolution will depart from the predictions of our , entirely harmonic , model . in the kho , the expansion through phase space changes from diffusive ( energy growing linearly in time ) to ballistic ( energy growing quadratically in time ) at the quantum resonance values . in fig .
[ fig : resonances](a ) we have plotted the number of kicks required for the kho to achieve particular mean energy values , as a function of @xmath23 when starting from an initial ground state .
this reveals distinct minima in the number of kicks required when @xmath141 , i.e. , a quantum resonant value .
other , less prominent minima are associated with higher - order resonances , i.e. , when @xmath23 is a rational multiple of @xmath59 , with the rate of energy absorption falling rapidly as the rational denominator increases .
figure [ fig : resonances](b ) shows explicitly how the mean energy grows quadratically as a function of the number of kicks when @xmath142 or @xmath143 ( ballistic expansion ) , quite distinct to the two cases we have shown when @xmath23 is an irrational multiple of @xmath59 ( diffusive expansion ) . , @xmath144 , and the initial condition is the harmonic oscillator ground state .
( a ) plots of the number of kicks required to reach a mean energy of @xmath145 ( lower line ) and @xmath146 ( upper line ) as a function of @xmath23 .
( b ) mean energy as a function of @xmath120 for @xmath147 , @xmath143 , @xmath148 , and @xmath149 [ @xmath134 is the golden ratio ] . ]
the quantum resonances are also apparent in the quasienergy spectra of @xmath13 , i.e. , the values @xmath150 such that @xmath151 is an eigenvalue of @xmath13 . taken as a function of @xmath57 , the quasienergy spectrum of the kho is a hofstadter butterfly - type multifractal @xcite , as shown in fig .
[ fig : floquet_spectrum ] .
the theory of butterfly - type spectra in kicked quantum maps has been extensively studied @xcite : in summary , a pure point spectral component is associated with localized behaviour , whilst singularly continuous and absolutely continuous spectral components are associated with diffusive and ballistic expansion respectively @xcite . for the kho ( when @xmath34 ) this means that the quasienergy spectrum of @xmath13 switches between being singularly continuous at irrational multiples , and absolutely continuous at rational multiples of the primary quantum resonant values of @xmath57 . kicked atom - optical systems
have been proposed as candidates for direct observation of butterfly - type spectra @xcite , and the atom - optical kho considered here presents another system in which this might be realised .
the kho also offers the opportunity to directly explore the changes in the nature of the butterfly spectrum across the quantum resonance / antiresonance transition ( that is , as a function of the relative alignment of the kicking and trapping potentials ) .
these changes have been explored theoretically for the case @xmath60 @xcite .
to observe , for example , the principal quantum resonance in the case where @xmath60 requires a lamb - dicke parameter of @xmath152 , and hence a suitable combination of harmonic trapping angular frequency @xmath4 and wavevector @xmath153 .
this could be realized in a linear ion trap with relatively weak axial confinement , perhaps in conjunction with a standing wave formed by two counterpropagating laser beams overlapping at an acute angle .
typical experimental configurations involving laser cooled alkali atoms within a magnetic trap yield much larger values of @xmath39 ; a configuration equivalent to that used by duffy _
et al_. would require a trap frequency of @xmath154 khz @xcite .
such frequencies are nevertheless within reach of recent generations of magnetic chip traps @xcite .
alternatively , a potentially quite flexible set - up could be achieved by generating the kicking field with a co@xmath155 laser ( wavelength @xmath156 ) . applying the kicking field directly along the axis of harmonic confinement , values of the lamb - dicke parameter around the principal quantum resonances ( e.g. , from @xmath157 up to @xmath158 )
could be reached using trap frequencies ranging from 836 hz ( @xmath159cs ) up to 40200 hz ( @xmath160na ) , and introducing the kicking field at an angle to the trap axis would allow these frequencies to be increased .
kicking strengths would be low because of the extreme detuning of the co@xmath155 laser , but the quantum resonances will still be manifest at any value of @xmath22 after a sufficient number of kicks .
in this paper we have considered the characteristic quantum resonances occurring in an atom - optical @xmath0-kicked harmonic oscillator ( kho ) , a system in which typically alkali metal atom(s ) or an alkaline earth ion are trapped in an approximately one - dimensional harmonic potential and periodically driven by pulses of an off - resonant laser standing wave .
we have demonstrated that the atom - optical kho displays quantum resonances dependent on the value of @xmath57 , where @xmath39 is the lamb - dicke parameter , equal to the ratio of the width of the ground state of the harmonic trapping potential to the wavelength of the laser standing wave . when the oscillator is kicked a rational number of times @xmath161 per oscillator period , with @xmath1 equal to @xmath162 , @xmath163 or @xmath164 , quantum resonances occur . at these values of @xmath1
the eigenstates of the floquet operator @xmath32 also possess a crystal phase space symmetry .
we have shown that the quantum resonances occur at natural number multiples of a primary resonance value : @xmath102 in the case @xmath60 , and @xmath73 in the cases @xmath65 and @xmath66 . at these values of @xmath57
all @xmath1 kicks within @xmath32 commute .
further , secondary , quantum resonances occur for @xmath57 equal to non - integer rational multiples of the primary values ; at these values of @xmath57 there is a partial commutativity between the kicks constituting @xmath32 .
we have demonstrated that the quantum resonances can be interpreted in terms of the relation between @xmath32 and its own eigenstates ; specifically , the quantum resonances occur when @xmath32 can be decomposed in terms of a set of displacement operators which is equivalent to the set of crystal symmetries of the eigenstates .
we have also derived an expression for the time evolution of a coherent state in the cases @xmath165 by recasting the dynamics in terms of a mapping on a two - dimensional lattice of coefficients . at quantum resonance values of @xmath57 the evolution of these coefficients
is greatly simplified ; we have shown that in this case there is a direct expression for all the coefficients at a given step .
finally , we have considered the manifestation of quantum resonances in an atom - optical kho , in particular the ballistic increase in the mean energy associated with quantum resonances , and the quasienergy spectrum . to this end
we have also discussed experimentally reasonable parameter regimes in which to observe quantum resonant dynamics in the atom - optical @xmath0-kicked harmonic oscillator .
we would like to thank c. s. adams , k. r. helmerson , i. g. hughes , w. d. phillips , and s. l. rolston for stimulating and enlightening discussions , together with the uk epsrc ( grant no .
ep / d032970/1 ) and durham university ( t.p.b . ) for support .
( the vertical position of points ) in the cases @xmath65 , @xmath163 , @xmath166 , and @xmath164 . ] because we are free to choose the sign of @xmath52 on the right side of ( [ eqn : master_resonance_condition ] ) we can , without loss of generality , simplify that commutation condition to @xmath167 , \ , w = 0,1,2,\ldots .
\label{eqn : new_master_resonance_condition}\ ] ] quantum resonances occur when the condition ( [ eqn : new_master_resonance_condition ] ) is satisfied for all possible @xmath45 . ignoring the trivial cases @xmath26 and @xmath168 , where @xmath169 for all possible @xmath45 , @xmath170 takes either @xmath171 values other than zero ( even @xmath1 ) , or @xmath172 values other than zero ( odd @xmath1 ) .
if we denote these nonzero values by @xmath173 [ where @xmath127 ranges from @xmath174 to @xmath171 ( even @xmath1 ) , or from @xmath174 to @xmath172 ( odd @xmath1 ) ] , the condition for quantum resonance becomes @xmath175 because the @xmath173 are the moduli of sines equally distributed in angle , they can not share a common factor other than unity .
consequently , condition ( [ eqn : simple_condition ] ) can be satisfied for all @xmath173 only if all of the @xmath173 are equal .
the cases of interest are enumerated pictorially in figure [ fig : commutation_diagram ] , where we see that : ( a ) @xmath176 for @xmath65 ( @xmath177 at quantum resonance ) ; ( b ) @xmath178 for @xmath60 ( @xmath119 at quantum resonance ) ; ( c ) there is more than one value of @xmath173 for @xmath179 ( there are no quantum resonances ) ; ( d ) @xmath176 for @xmath66 ( @xmath177 at quantum resonance ) .
figure [ fig : commutation_diagram](d ) also illustrates the absence of quantum resonances for @xmath180 ; the addition of another distinct point on the circumference of the circle shown necessitates at least a second value of @xmath173 .
) for two bessel functions of integer order and equal , real , positive argument @xmath114 . adapted from @xcite . ]
graf s addition theorem for two bessel functions of integer order and equal , real argument @xmath114 is @xcite @xmath181 where @xmath182 and @xmath183 are related to @xmath114 and @xmath184 by @xmath185 } , \\
\zeta [ 1 - \cos(\alpha ) ] & = \zeta^\prime \cos(\chi ) , \\
\zeta \sin(\alpha ) & = \zeta^\prime \sin(\chi).\end{aligned}\ ] ] these relations are shown graphically for positive @xmath114 in fig .
[ fig : graf_diagram ] , for @xmath186 . in the quantum resonant case we consider in section [ subsubsec : q4 ] ,
consequently @xmath188 , and we obtain @xmath189
in the case @xmath66 , and at quantum resonance @xmath190 the mapping ( [ eqn : coefficient_mapping ] ) becomes @xmath191}_{m , n } = \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) m^{[j]}_{m+n - k ,- m } e^{-i k m \pi}. \label{eqn : q=6mapping}\ ] ] as in the case @xmath60 , we begin with the single coherent state initial condition @xmath100}_{m , n } = \delta_{m,0 } \delta_{n,0}.\ ] ] replacing the indices in the correct fashion we get @xmath100}_{m+n - k ,- m }
= \delta_{m+n , k } \delta_{m,0}\ ] ] and thus , substituting this into the mapping ( [ eqn : q=6mapping ] ) , @xmath104}_{m , n } = i^{m+n } j_{m+n}(\zeta ) \delta_{m,0 } e^{-i m(m+n ) \pi}. \label{eqn : q=6m1}\ ] ] again , as in the case @xmath60 , we observe the more general pattern of evolution by repeating this process until we obtain the coefficients @xmath192}_{m , n}$ ] .
we begin by rearranging ( [ eqn : q=6m1 ] ) to get @xmath104}_{m+n - k ,- m } = i^{n - k } j_{n - k}(\zeta ) \delta_{m+n , k } e^{-i ( m+n - k)(n - k ) \pi}.\ ] ] substituting this into the mapping ( [ eqn : q=6mapping ] ) produces @xmath193}_{m , n } = i^{m+n } j_{m+n}(\zeta ) i^{m } j_{m}(\zeta ) e^{-i m(m+n ) \pi } , \label{eqn : q=6m2}\ ] ] where we have used the identity @xmath194 . rearranging ( [ eqn : q=6m2 ] )
we find that @xmath193}_{m+n - k ,- m
} = i^{n - k } j_{n - k}(\zeta ) i^{m+n - k } j_{m+n - k}(\zeta ) e^{-i ( mn + n^2 + k^2 + mk ) \pi},\ ] ] and hence @xmath110}_{m , n } = \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) i^{n - k } j_{n - k}(\zeta ) i^{m+n - k } j_{m+n - k}(\zeta ) e^{-i ( mn + n^2 + k^2)\pi}. \label{eqn : q=6m3}\ ] ] graf s addition theorem ( appendix [ app : graf ] ) can not be directly applied to ( [ eqn : q=6m3 ] ) , as there are three bessel functions with @xmath37-dependent indices .
however , we accept this summation as the final form of @xmath195}_{m , n}$ ] and continue with the process , using @xmath196 as the new index of summation . rearranging ( [ eqn : q=6m3 ] )
we have @xmath109}_{m+n - k^\prime ,- m } = & \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(\zeta ) i^{n - k - k^\prime } \\ & \times j_{n - k - k^\prime}(\zeta ) e^{i [ ( m+n - k^\prime)m -m^2 - k^2 ] \pi } , \end{split}\ ] ] which , when substituted into the mapping ( [ eqn : q=6mapping ] ) , produces @xmath197}_{m , n } = & \sum_{k=-\infty}^{\infty } \sum_{k^\prime=-\infty}^{\infty } i^{k^\prime } j_{k^\prime}(\zeta ) i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(\zeta ) \\ & \times
i^{n - k - k^\prime } j_{n - k - k^\prime}(\zeta ) e^{i ( mn - k^2 ) \pi } , \\ = & \sum_{k=-\infty}^{\infty }
i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(\zeta ) i^{n - k } e^{i(mn - k^2)\pi } \\ & \times \sum_{k^\prime=-\infty}^{\infty } j_{k^\prime}(\zeta ) j_{n - k+k^\prime}(\zeta ) e^{ik^\prime \pi } , \\
= & \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(\zeta ) i^{n - k } j_{n - k}(2\zeta ) e^{i(mn - k^2)\pi } , \end{split}\ ] ] where we have used graf s theorem to eliminate the summation over @xmath196 . rearranging this
we obtain @xmath197}_{m+n - k^\prime ,- m } = & \sum_{k=-\infty}^{\infty } i^{k } j_{k}(\zeta ) i^{m+n - k^\prime+k } j_{m+n - k^\prime+k}(\zeta ) \\ & \times i^{m+k } j_{m+k}(2\zeta ) e^{-i[m(m+n - k^\prime ) + k^2]\pi } , \end{split}\ ] ] and hence @xmath198}_{m , n } = & \sum_{k=-\infty}^{\infty } \sum_{k^\prime=-\infty}^{\infty } i^{k^\prime } j_{k^\prime}(\zeta ) i^{k } j_{k}(\zeta ) i^{m+n - k^\prime+k } j_{m+n - k^\prime+k}(\zeta ) \\ & \times i^{m+k } j_{m+k}(2\zeta ) e^{-i[m(m+n ) + k^2]\pi } \\ = &
\sum_{k=-\infty}^{\infty}i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(2\zeta ) i^{m+n+k } e^{-i[m(m+n ) + k^2]\pi } \\ & \times \sum_{k^\prime=-\infty}^{\infty } j_{k^\prime}(\zeta ) j_{m+n+k^\prime+k}(\zeta ) e^{ik^\prime \pi}\\ = & \sum_{k=-\infty}^{\infty}i^{k } j_{k}(\zeta ) i^{m+k } j_{m+k}(2\zeta ) i^{m+n+k } j_{m+n+k}(2\zeta)\\ & \times e^{-i[m(m+n ) + k^2]\pi}. \end{split}\ ] ] repeating the process one final time we have @xmath198}_{m+n - k^\prime ,- m } = & \sum_{k=-\infty}^{\infty}i^{k } j_{k}(\zeta ) i^{m+n - k^\prime+k } j_{m+n - k^\prime+k}(2\zeta ) \nonumber \\ & \times i^{n - k^\prime+k } j_{n - k^\prime+k}(2\zeta ) e^{-i(mn + n^2 + { k^\prime}^2 + k^2 -mk^\prime)\pi } , \end{split}\ ] ] and hence @xmath199}_{m , n } = & \sum_{k=-\infty}^{\infty } \sum_{k^\prime=-\infty}^{\infty } i^{k^\prime } j_{k^\prime}(\zeta)i^{k } j_{k}(\zeta ) i^{m+n - k^\prime+k } j_{m+n - k^\prime+k}(2\zeta ) \\ & \times i^{n - k^\prime+k } j_{n - k^\prime+k}(2\zeta ) e^{-i(mn + n^2 + { k^\prime}^2 + k^2)\pi}. \label{eqn : q=6m6_1 } \end{split}\ ] ] if we define @xmath200 , then ( [ eqn : q=6m6_1 ] ) takes the form @xmath199}_{m ,
n } = & \sum_{\delta=-\infty}^{\infty } i^{m+n+\delta } j_{m+n+\delta}(2\zeta ) i^{n+\delta } j_{n+\delta}(2\zeta ) \\ & \times i^\delta e^{-i(mn + n^2 + \delta^2)\pi}\sum_{k^\prime=-\infty}^{\infty } j_{k^\prime}(\zeta ) j_{k^\prime+\delta } ( \zeta ) e^{i k^\prime \pi}. \label{eqn : q=6m6_2 } \end{split}\ ] ] by applying graf s theorem to the summation over @xmath196 in ( [ eqn : q=6m6_2 ] ) , and then applying the transformation @xmath201 we obtain @xmath199}_{m , n } = & \sum_{\delta=-\infty}^{\infty } i^\delta j_{\delta}(2\zeta ) i^{n-\delta } j_{n-\delta}(2\zeta ) \\ & \times i^{m+n-\delta } j_{m+n-\delta}(2\zeta ) e^{-i(mn + n^2 + \delta^2)\pi } , \end{split}\ ] ] that is , an analogue of @xmath195}_{m , n}$ ] with @xmath114 replaced by @xmath115 and @xmath37 replaced by @xmath7 . as in the case @xmath60 , the evolution from this point on will continue cyclically . | under certain conditions , the quantum @xmath0-kicked harmonic oscillator displays quantum resonances .
we consider an atom - optical realization of the @xmath0-kicked harmonic oscillator , and present a theoretical discussion of the quantum resonances that could be observed in such a system .
having outlined our model of the physical system we derive the values at which quantum resonances occur and relate these to potential experimental parameters .
we discuss the observable effects of the quantum resonances , using the results of numerical simulations .
we develop a physical explanation for the quantum resonances based on symmetries shared between the classical phase space and the quantum - mechanical time evolution operator .
we explore the evolution of coherent states in the system by reformulating the dynamics in terms of a mapping over an infinite , two - dimensional set of coefficients , from which we derive an analytic expression for the evolution of a coherent state at quantum resonance . |
the dispersive properties of capillary - gravity waves are responsible for the complicated wave pattern generated at the free surface of a still liquid by a disturbance moving with a velocity @xmath0 greater than the minimum phase speed @xmath1 , where @xmath2 is the gravity , @xmath3 is the surface tension and @xmath4 the density of the fluid @xcite .
the disturbance may be produced by a small object partially immersed in the liquid or by the application of an external surface pressure distribution @xcite .
the waves generated by the moving perturbation propagate momentum to infinity and , consequently , the disturbance experiences a drag @xmath5 called the wave resistance @xcite . for @xmath6
the wave resistance is equal to zero since , in this case , no propagating long - range waves are generated by the disturbance @xcite .
a few years ago , it was predicted that the wave resistance corresponding to a surface pressure distribution symmetrical about a point should be discontinuous at @xmath7 @xcite .
more precisely , if @xmath8 is the the total vertical force exerted on the fluid surface , the wave resistance is expected to reach a finite value @xmath9 for @xmath10 . for an object much smaller than the capillary length @xmath11 , the discontinuity @xmath12 is given by : @xmath13 experimentally
, the onset of the wave - resistance due to the generation of capillary - gravity waves by a partially immersed moving object was studied recently by two independent groups @xcite . while browaeys _ et al .
_ @xcite indeed find a discontinuous behaviour of the wave - resistance at @xmath7 in agreement with the theoretical predictions , burghelea _ et al .
_ @xcite observe a smooth transition .
the discrepancy between the theoretical analysis of @xcite and the experimental results of @xcite might be due to the fact that the experimental setup of burghelea _ et al .
_ uses a feedback loop to keep the object at a constant depth while the analysis of @xcite assumes that the vertical component @xmath8 of the force exerted by the disturbance on the fluid does not depend on the velocity @xmath0 ( we might call such an analysis a fixed force " analysis ) . in order to check this proposition ,
we perform in this letter a fixed - depth " calculation of the wave - drag close to the onset threshold .
a somewhat similar analysis was performed in the large velocity limit by sun and keller @xcite .
we will show that such a calculation indeed yields a cancellation of the vertical force at @xmath7 , that is to say , according to equation ( [ pgg ] ) , a smoothing of the discontinuity . immersed at a depth @xmath14 .
the velocity of the object is @xmath0 .
we decompose the force exerted by the fluid on the object in a component orthogonal to the free surface ( @xmath8 ) and a component parallel to it ( @xmath15 ) . ]
we take the @xmath16 plane as the equilibrium surface of the fluid .
the immersed object exerts a stress at the fluid surface that can be considered equivalent to a pressure field @xmath17 @xcite that travels over the surface with a velocity @xmath0 in the @xmath18 direction .
we assume @xmath19 ( the fourier transform of @xmath17 ) to be of the form : @xmath20 where @xmath21 and @xmath22 . in this case
, @xmath19 is isotropic @xcite and @xmath8 is the total vertical force exerted on the fluid . within the framework of rayleigh s linearized theory of capillary - gravity waves ,
the fourier transform @xmath23 of the free surface displacement @xmath24 is related to the pressure field through @xcite : @xmath25 where @xmath26 is the free dispersion relation , @xmath27 and @xmath28 is the kinematic viscosity of the fluid . let us suppose the object is located at the origin of the moving frame .
if @xmath14 is its depth , the free surface displacement @xmath29 must be @xmath30 under the pinpoint ( here we suppose the object sufficiently hydrophobic and @xmath14 not to large so that the pinpoint does not pierce the surface ) .
this leads to the following normalization condition : @xmath31 the value of @xmath14 as a function of @xmath8 can be readily calculated using equations ( [ linearreponse ] ) and ( [ boundary ] ) : @xmath32 with @xmath33 finally , the drag - force @xmath34 is calculated by simply integrating the pressure force over the free surface @xcite . @xmath35 this yields , using the explicit expression ( [ linearreponse ] ) for @xmath36 , @xmath37 with @xmath38 according to equation ( [ formaldrag ] ) , the integral @xmath39 describes the fixed - force behaviour of the wave - resistance .
due to the symmetry of @xmath40 , @xmath39 is parallel to @xmath41 and we shall henceforth set @xmath42 , where @xmath43 is the unit vector parallel to the velocity of the object .
the authors of @xcite studied the properties of @xmath44 in the case of a non - viscous fluid for which they showed that : \(a ) @xmath45 for @xmath6 ; \(b ) @xmath44 is discontinuous for @xmath10 , with @xmath46 \(c ) in the large velocity limit @xmath47 in the case of a fixed depth analysis , @xmath8 becomes a function of @xmath0 .
using equations ( [ formaldepth ] ) , we can rewrite the wave - resistance as @xmath48 in general we have to rely on numerics to calculate the integrals @xmath49 and @xmath44 .
typical results are presented on figure [ fig2 ] for an objet of size 0.1 mm immersed in water , and for a step - like function @xmath40 equal to 1 for @xmath50 and @xmath51 otherwise .
we first observe on fig .
( [ fig2]-a ) that @xmath49 increases sharply near the threshold .
this leads to two rather different behaviours for @xmath44 and @xmath15 as shown on fig .
( [ fig2]-b ) : while @xmath44 exhibits a discontinuity close to @xmath7 @xcite , the fixed - depth wave - drag @xmath52 cancels smoothly at the critical velocity . .
( b ) comparison of @xmath53 ( dashed line ) and @xmath54 ( full line ) describing respectively the fixed - force and fixed - depth behaviour of the wave - resistance .
we clearly see that contrary to the fixed - force analysis , the fixed - depth calculation does not yield any discontinuity at the threshold . here
@xmath55 , @xmath56 , @xmath57 and @xmath58 . ]
the characteristic features displayed by the plots of fig . [ fig2 ]
can be captured by a zero - viscosity analysis . setting @xmath59 , equation ( [ normalize ] )
can be simplified as : @xmath60 where @xmath61 denotes the cauchy principal value of the integral .
the integral is calculated in polar coordinates @xmath62 , where @xmath63 is the angle of @xmath64 with respect to @xmath65 .
introducing the function @xmath66 defined by : @xmath67 where @xmath68 and @xmath69 is the mach " number , equation ( [ visczero ] ) can then be rewritten as : @xmath70 using the residue theorem @xcite , we get : @xmath71 where @xmath72 the is the heaviside step - function .
the variations of @xmath73 with @xmath74 are plotted on fig .
[ fig1 ] : @xmath73 reaches its minimum value @xmath75 for @xmath76 .
it shows that equation @xmath77 has two solutions @xmath78 and @xmath79 , with @xmath80 , if @xmath0 is larger than the critical velocity @xmath81 , and none if @xmath6 .
( see text ) with k. since @xmath73 reaches a minimum for @xmath76 with @xmath82 , equation @xmath83 presents solutions for @xmath84 . in this case
, there are two possible values @xmath78 and @xmath79 with @xmath80 . ] for @xmath85 , @xmath49 evaluates to @xmath86 . \label{integral}\ ] ] the above integrals can be calculated in the two limiting cases @xmath87 ( _ i.e. _ @xmath88 ) and @xmath89 ( _ i.e. _ @xmath90 ) . for large @xmath91 ,
the integrals are restricted to either large or small values of @xmath74 .
if @xmath40 vanishes faster than @xmath92 for large @xmath74 , we can show that the small @xmath74 contribution dominates . in this region ,
the dispersion relation is dominated by gravity waves , so that we can approximate @xmath73 by @xmath93 . a straightforward calculation
then yields the following asymptotic expansion for @xmath49 : @xmath94 combining this result with ( [ lambdalarge ] ) and [ formalfixeddepth ] , we get : @xmath95 let s now focus on the case @xmath0 close to @xmath81 .
we set : @xmath96 and @xmath97 using the fact that @xmath78 and @xmath79 are the roots of @xmath98 , we can rewrite @xmath99 as : @xmath100 when @xmath101 , the @xmath102 term cancels in @xmath76 .
since all other terms are regular , we can write at the leading order : @xmath103 this latter integral is readily calculated and gets a very simple form in the limit @xmath104 : @xmath105 since @xmath73 is invariant by the transformation @xmath106 , we see that @xmath107 and @xmath108 have the same asymptotic behaviour for @xmath104 . in this limit
we have @xmath109 , hence : @xmath110 in the case of an object of size @xmath111 , the width of @xmath40 is about @xmath112 . if we choose @xmath111 much smaller than the capillary length @xmath113 , we can approximate @xmath114 by @xmath22 . in this limit ,
equation ( [ approxboundary ] ) takes the following form : @xmath115 if we combine equations ( [ lambdathreshold ] ) and ( [ mainresult ] ) , we see that slightly above the threshold , the wave - resistance behaves like : @xmath116 equation ( [ scaling ] ) constitutes the main result of this paper .
first , we notice that for small objects this relation is , as expected , independent of the actual shape of the pressure field .
second , and more important , it shows the wave resistance @xmath15 cancels out at @xmath7 .
this smearing is due to the cancellation of the vertical force @xmath8 near the threshold that we get from the behaviour of @xmath49 .
in this paper we have shown that in a fixed - depth situation the discontinuity of the drag - force calculated in @xcite vanishes and is replaced by a smooth variation , in accordance with the experimental results found in @xcite . a quantitative comparison between the present analysis and the data of ref .
@xcite is however more involved since experiments from @xcite were performed in narrow channel geometry @xcite with objects of size comparable with the capillary length . to recover the scaling relation observed experimentally
, it would also be necessary to take into account both the variations of the shape of @xmath40 with @xmath0 and the reflections of the waves on the walls of the channel .
the present study suggests nevertheless that to fully test relation ( [ pgg ] ) , experiments need to be devised that would measure both @xmath15 and @xmath8 . not that this assumption is not entirely correct in the presence of viscosity since it leads to a slip of the object with respect to the fluid surface . in this case , a force field parallel to the free surface should be added to take into account the viscous drag .
however , this new term leads to the well known stokes drag , that can substracted from experimental results , and so can be omitted in our analysis . | we study the onset of the wave - resistance due to the generation of capillary - gravity waves by a partially immersed moving object in the case where the object is hold at a fixed immersion depth .
we show that , in this case , the wave resistance varies continuously with the velocity , in qualitative accordance with recent experiments by burghelea _
et al . _
( phys .
rev .
lett . * 86 * , 2557 ( 2001 ) ) . |
one of the striking features of ordinary quantum mechanics is its linear structure . given two possible states of a system there is a notion of the sum of these two states .
no such notion exists in classical mechanics .
the phase space of classical systems does not have the structure of a vector space .
in general we can not add the states of a classical system . since quantum mechanics
is the more fundamental theory the question arises of how this non - linear property of classical physics arises from the linear theory .
this problem is usually called the measurement problem since it is in the process of a measurement on the system that the problem requires a solution . the solution to the problem
is usually referred to as an interpretation . in this article
we argue that no additional interpretation of ordinary quantum mechanics is required if only one carefully defines what is meant by a classical state .
the relation of classicality to quantum mechanics is often blurred by the fact that we have constructed the quantum theory from the classical theory by a process of quantization .
this means that the states of the classical system are also possible states of the quantum system .
we regard this fact as one of the major reason for why the measurement problem exists .
if the classical states are known from the outset the question arises of how the quantum system assumes one of these states . since the quantum system might be in a superposition of these classical states the measurement problem can not be escaped . in our presentation
we thus want to be more careful .
we present quantum mechanics and its relation to classical mechanics in three steps . in the first step ( section [ sec : qm ] )
we present the basic setup of quantum mechanics . here
we adhere to the usual linear structure .
the state space is given by a hilbert space and the evolution is given by a unitary one - parameter group generated by a self - adjoint hamiltonian .
we will not introduce any notion of classicality at this point and we do not talk about measurements yet . in the second step ( section [ sec : class ] ) we then introduce the notion of a classical property .
the important point here is that it is a quantum system that has this classical property .
no notion of classicality outside of quantum mechanics needs to be introduced . in the last step ( section [ sec : trans ] )
we then look at the transition from quantum mechanical states to classical states .
this is where one usually encounters the measurement problem .
we will show how with our definition of classicality the measurement problem does not arise . in the concluding section [ sec : concl ] we comment on what these views of quantum mechanics mean for the search for a quantum theory of gravity .
the basic setup we will use is that of standard quantum mechanics .
we thus start with a hilbert space @xmath0 .
an example of such a hilbert space could be the space of @xmath1 qbits : @xmath2 on this hilbert space the evolution of the system is described by a hamiltonian @xmath3 .
it generates time evolution through the unitary @xmath4 this is the basic setup we will use .
it is completely linear and deterministic .
we will not add a collapse postulate that breaks the linearity .
we will also not give a probabilistic interpretation of the states in @xmath0 or postulate the born rule . the most important difference to the standard formulation of quantum mechanics is that we do not have an a priori notion of classicality .
when setting up a quantum mechanical calculation the hilbert space is often of the form @xmath5 an example is @xmath6 , where @xmath7 represents the space of positions of a particle .
we do not want to start with such an identification of what a classical state is .
the reason is that such an identification forces the measurement problem on us .
if one assumes it from the outset the question for how the classical world arises is fixed : how does the state of the system become @xmath8 given that the initial state of the system was a superposition of states of this form one encounters the measurement problem .
the inclusion of a classical world into the basic formulation of the theory is also unsatisfactory in that a less fundamental theory is needed to formulate the more fundamental one .
we want to avoid this mix of ontologies and start with a quantum theory that has no notion of classicality built in .
given that we have no a priori notion of classicality we now have to define what we mean by classicality .
the definition we will use is the following : a system has a _ classical property _ @xmath9
if the free energy describing the system is of the form @xmath10 for some constant @xmath11 . if a system has a classical property @xmath9 we can introduce a generalized force @xmath12 in solid state physics this force is referred to as generalized elasticity or rigidity . a large class of examples of systems with classical properties is provided by systems that have undergone symmetry breaking .
the order parameter that is non - zero in the symmetry broken phase is the classical property .
the easiest example here is spin chain where the classical property is the magnetization @xmath13 .
symmetry breaking does not exhaust the class of examples of classical properties .
the formation of a surface in the gas to liquid transition is an example that does not fit neatly into the symmetry breaking paradigm . here
the classical property is the surface @xmath14 .
the corresponding generalized force is the surface tension .
defined this way we see that a classical property is a _ property of a large quantum system_. the presence of the generalized force @xmath15 makes the classical property of the system observation independent .
the interaction with a system that has a classical property @xmath16 is effectively described using this classical property .
to infer what the magnetization @xmath17 of a spin chain is one might probe it with another spin chain with magnetization @xmath18 . their interaction is best described by a term of the form @xmath19 .
this probing does not destroy the classical property .
it becomes in this sense an objective , observer and observation independent , property .
with this new definition of classicality we can now ask the question what happens when a classical property emerges .
the key observation is that this transition from quantum to classical is discontinuous and very sensitive to the environment .
we claim that it is the statistic nature of the environment that enters here that is responsible for the probabilistic nature of quantum mechanics . of the system
does not depend on @xmath16 . in the transition it acquires a dependence on @xmath16 .
this transition is naturally discontinuous and very sensitive to the state of the environment as is the motion of the ball in this classical image .
it is here that the probabilistic nature of quantum mechanics emerges . ]
before the system acquires a classical property @xmath9 the free energy describing the system does not depend on @xmath16 .
after the system has acquired the classical property @xmath9 it depends on @xmath16.this change in behavior of the free energy is by necessity discontinuous .
it is also very sensitive to the state of the environment .
we see that the environment has three roles that are not usually acknowledged . because the system after the transition is in a state of lower entropy and energy the environment functions as a dump for the excess energy and entropy .
the environment is further required to bring the system close to a transition point .
this might be achieved by adjusting pressure or temperature to appropriate values .
finally the environment is responsible for the stochastic character of the transition . to the last point
one might object that the environment should be symmetric and is thus unable to choose a particular classical property .
the environment is indeed symmetric but it is only symmetric in an ergodic sense .
if we denote by @xmath20 an element of the symmetry group of the environment and by @xmath21 the state of the environment at a particular time then in general we will have @xmath22 the state of the environment is non - symmetric because of the existence of non - symmetric fluctuations .
if we define a time averaged state @xmath23 by @xmath24 then we will have indeed @xmath25 for @xmath26 large enough .
this is because the non - symmetric fluctuations cancel each other .
the presence of these non - symmetric fluctuations together with a transition that is very sensitive to the environment gives rise to the probabilistic appearance of classical properties .
one can compare this situation to the situation in usual quantum mechanics . in the copenhagen interpretation of quantum mechanics the environment of the quantum system
is provided by a classical observer .
it is the contact with this classical observer that leads to the collapse of the wavefunction and the introduction of probability . in our approach
the environment is a quantum ensemble .
probability is introduced by the non - symmetric fluctuations of the environment .
the obvious advantage of our approach is that no split between observer and system needs to be postulated .
the many body system achieves classicality with a quantum environment .
probabilistic behavior in the transition from quantum to classical can thus not be avoided .
can one also calculate the corresponding probabilities ? given the structure of hilbert spaces and inner products that we have assumed from the outset this is indeed possible provided one makes some assumption about the nature of the interaction of the system and the environment .
details can be found in @xcite .
in this article we have argued that the measurement problem in quantum mechanics can be traced to the appearance of a classical world in the initial formulation of the theory .
if instead a definition of classicality in quantum mechanical terms is adopted the measurement problem can be resolved in the linear setup .
the resolution of the problem does not require the introduction of a split between the observer and the object .
probabilities arise because the nature of the transition from quantum to classical amplifies the fluctuations in the environment .
probabilities thus have the same origin in quantum mechanics as elsewhere in physics : the lack of knowledge .
the view we are proposing here relies on reasoning common in statistical physics .
the world is as it is because this is the most likely way for it to be .
the situation in quantum mechanics can be compared to the situation it statistical mechanics at the beginning of the last century . there existed a conceptual tension between the second law of thermodynamics which stated that the entropy of a system never decreases and poincar recurrence which implied that a system always returns to its initial conditions .
no function could be both monotonically increasing and periodic .
the solution to this problem was given by the ehrenfests @xcite who argued that it was overwhelmingly unlikely for the system to return to its initial state .
we have proposed a similar statistical solution to the measurement problem .
the occurrence of superpositions of classical objects is not impossible but just very unlikely .
john bell called a solution of this kind a solution for all practical purposes@xcite and introduced the shorthand fapp for it .
it is clear that he intended it as a derogatory term .
he was after a more fundamental explanation .
we have argued that there does not exist a more fundamental solution then the one we have proposed here .
we will have to do with a solution fapp .
this view of the foundations of quantum mechanics also sheds light onto the search for a quantum theory of gravity .
one lesson is : do not quantize .
if classicality arises as it is proposed here then the more fundamental quantum theory is not obtained by quantizing a classical limit .
the search for quantum gravity should thus not start with the classical theory and try to quantize it .
instead one should start with a purely quantum theory .
the construction of quantum theories without a classical theory to start with has begun only recently .
examples are the theories constructed by x .-
wen @xcite and s. lloyd @xcite .
a second lesson is that certain concepts that make sense in a classical world might not have a meaning in the more fundamental quantum world .
concepts like locality , position , and momentum might only be strictly applicable and meaningful in a classical context .
the author would like to thank lee smolin , chris beetle , chris isham , fay dowker , fotini markopoulou , hans westman , joy christian , and lucien hardy for comments and discussions .
part of the work presented in this article was carried out at the perimeter institute for theoretical physics .
the author is grateful for the institutes hospitality .
this work was presented at the dice 2006 conference in piombino .
the author thanks thomas elze for organizing the workshop and the opportunity to speak . 9
dreyer o 2006 emergent probabilities in quantum mechanics _
quant - ph/0603202 .
ehrenfest p and ehrenfest t 2002 _ the conceptual foundations of the statistical approach in mechanics _ ( dover ) .
bell j s 1990 _ phys .
world _ * 3 * 33 .
this article is reprinted in bell j s 2004 _ speakable and unspeakable in quantum mechanics _ ( cambridge university press ) .
wen x - g 2004 _ quantum field theory of many - body systems _
( oxford university press ) .
lloyd s 2005 the computational universe : quantum gravity from quantum computation _ preprint _ quant - ph/0501135 . | in this article we propose a solution to the measurement problem in quantum mechanics .
we point out that the measurement problem can be traced to an a priori notion of classicality in the formulation of quantum mechanics .
if this notion of classicality is dropped and instead classicality is defined in purely quantum mechanical terms the measurement problem can be avoided .
we give such a definition of classicality .
it identifies classicality as a property of large quantum system .
we show how the probabilistic nature of quantum mechanics is a result of this notion of classicality .
we also comment on what the implications of this view are for the search of a quantum theory of gravity . |
model photospheres for late - type stars are a fundamental cornerstone of modern astronomy .
the sun , our closest star , belongs to this stellar category and its atmosphere provides an exceptional plasma laboratory to study detailed physical processes .
late - type stars are also the most numerous group in the galaxy .
they not only dominate its present dynamics , but their photospheric abundances constitute a unique tool for tracing its chemical evolution history .
furthermore , the oldest unevolved stars in the galaxy , which provide information about the physical conditions at early epochs and constraints on models of primordial nucleosynthesis , are also late - type stars . in this century
, considerable efforts have been devoted to theoretical and empirical modeling of late - type stellar atmospheres . in this respect ,
particular success was achieved in the seventies , and most of the models used nowadays are mainly based on work carried out in that decade . for a comprehensive review of the state of the art in the modeling of photospheres of f- and later - type stars see , for example , the discussion by gustafsson & jrgensen ( 1994 ) and avrett ( 1996 ) .
classical theoretical models are usually computed under the simplifying assumptions of hydrostatic equilibrium , plane - parallel stratification , local thermodynamic equilibrium ( lte ) , and conservation of energy flux ( treating convection with the mixing length approximation ) .
high - resolution , high signal - to - noise ratio spectra over extensive spectral ranges can be used to test the validity of existing theoretical models and to better constrain the observational features which have to be reproduced ( line profiles and asymmetries , excitation and ionization equilibria of different elements , stellar fluxes , etc . ) .
these data can also be used to construct models by using observed features to find a temperature depth ( @xmath0 ) relation .
semi - empirical models of this kind have been obtained for the solar photosphere , where the present - day knowledge of the abundances of different elements , as well as its extended disk , allow a better observational constraint .
the model constructed by holweger & mller ( 1974 ) is a good example of this class and has been widely used not only for the sun but also for other late - type stars by scaling its @xmath0 relation to different stellar effective temperatures .
although the spatially resolved information that can be obtained for the sun is lacking for other late - type stars , spectral lines do contain enough information to determine reliable values for the physical magnitudes governing the state of a stellar atmosphere , and this approach has been applied , for instance , to the modelling of giant stars such as arcturus ( mckle et al .
1975 ) and pollux ( ruland et al . 1980 ) .
the so - called spectroscopic _ inversion _ methods are aimed at obtaining the model atmosphere from which the synthetic line profiles best match the observed ones ( usually with a least - squares criterion ) . the availability of high - quality data ( observed spectra and line parameters ) and modern computers make possible these inversions by taking into account the information contained in the whole line profile , which can potentially provide not only the temperature stratification in the atmosphere but also insights into other physical magnitudes such as velocity and magnetic fields .
this paper is part of a wider project aimed at better understanding the photospheres of late - type stars of different metallicities for use in studies of galactic chemical evolution and primordial nucleosynthesis . among the different approaches envisaged within this project ,
which incorporates the study of theoretical model atmospheres and the presence and extension of line asymmetries in the photospheres of metal - poor stars , we present here a successful attempt to obtain a semi - empirical solar model photosphere by inverting observed clean line profiles with accurate oscillator strengths .
this kind of information is also available for other late - type stars ( with different levels of metallicity ) , and what we learn from modeling the sun will help us when trying to apply the method to other stars .
section 2 provides the details of the inversion code , which is based on a well - tested inversion method for the stokes parameter vector ; 3 shows its application to the solar case and the comparison of its results with those associated with other well - known theoretical and semi - empirical models .
we describe in 4 the potential use and limitations in applying the code to other late - type stars , and the main conclusions and the perspectives for the near future are summarized in 5 .
the miss ( multiline inversion of stellar spectra ) code described in this paper is an adaptation of a previous inversion method of stokes vector spectra : the sir method ( ruiz cobo & del toro iniesta 1992 ; see also del toro iniesta & ruiz cobo 1995 , 1996 , 1997 ) .
the sir code has recently been applied to the study of different solar structures observed in polarized light , such as sunspots ( collados et al .
1994 ; westendorp plaza et al .
1997a , b , c ) , and unresolved magnetic elements in facular regions ( bellot rubio , ruiz cobo & collados 1997 ) , and for non - polarized light to structures such as penumbrae ( del toro iniesta , tarbell & ruiz cobo 1994 ) , solar granulation ( ruiz cobo et al . 1995 , 1996 ; rodrguez hidalgo , ruiz cobo & collados 1996 ) , and the solar 5-min oscillation ( rodrguez hidalgo , ruiz cobo & collados 1997 ; ruiz cobo , rodrguez hidalgo & collados 1997 ) . a study of its performance versus other inversion techniques
can be found in wood & fox ( 1995 ) and westendorp plaza et al .
( 1998 ) .
our inversion method iteratively modifies an initial model atmosphere until an optimal fit of the observed profiles is reached .
the model atmosphere contains a set of values of temperature , microturbulence , and line - of - sight velocity evaluated on an equally spaced logarithmic scale of the optical depth of the continuum at 5000 ( @xmath1 ) , together with a set of optical depth - independent parameters such as rotation velocity ( @xmath2 ) , macroturbulent velocity amplitude , and chemical abundances .
gravity is fixed and the gas and electronic pressures are computed to satisfy hydrostatic equilibrium , the ideal gas equation of state , and the saha equation . in this section
we will pay special attention to the differential features between our inversion code ( miss ) and the progenitor ( sir ) , describing its particularities within three subsections dedicated to hypotheses and model parameterization , spectral synthesis , and inversion algorithm . in order to simplify the synthesis process , as well as to keep the number of free parameters small enough , a set of hypotheses
has been assumed : the lte approximation , a steady - state , plane - parallel , one - dimensional atmosphere , solid body rotation , hydrostatic equilibrium , radial bulk velocity , and negligible magnetic field . to evaluate the continuum absorption coefficient , contributions from h , he , h@xmath3 , he@xmath3 , h@xmath4 , h@xmath5 , c , mg , na , and scattering terms have been taken into account . other atomic and molecular line opacity sources , which are neglected here , can also be included . to obtain an accurate solution of the transfer equation the number of points ( optical depth values ) through the atmosphere has to be about 50 .
considering the values for every one of the depth - dependent quantities as free parameters would imply solving a system of more than 150 dimensions . to simplify the numerical problem , following the sir strategy ,
the perturbations of depth - dependent quantities are evaluated only in a discrete number of points in @xmath1 , called _
nodes_. the entire stratifications of the various parameters are thus obtained by adding the result of a spline interpolation of the perturbations at the nodes to the initial stratification . for each parameter a different number of nodes can be selected by the user .
experience from the quoted applications of sir reveals that the optimum solution is reached by steadily increasing the degree of complexity of perturbations ( i.e. , the number of nodes ) from one iteration to the next , until no significant improvement in the fit between the observed and synthesized profiles is reached .
provided the quantity and quality of observational data allow a corresponding increase in the number of free parameters , some of the previous hypotheses can be relaxed in the framework of the present miss code : the homogeneous atmosphere can be substituted for a two - component atmosphere in order to take the effect of convective motions into account ; the hydrostatic - equilibrium hypothesis can be ignored by considering the electron - pressure stratification as a free parameter ; the application of miss to the determination of magnetic fields is straightforward , since this code is a simplified version of sir , a procedure for stokes vector inversion . nevertheless , as will be shown below , the amount of information we shall be able to extract is strongly limited by problems deriving from the accuracy of atomic parameters ( mainly oscillator strengths ) , as well as from the presence of observational noise .
therefore , up to the present time , the search for inhomogeneous , magnetized or far from hydrostatic equilibrium models is beyond our reach .
concerning the lte approach , a project is at present in progress for inversion of the stokes parameter vector under the following hypotheses : a steady - state atmosphere , complete redistribution on scattering , ls - coupling , and no quantum interference between zeeman states ( socas - navarro et al .
1996 , socas - navarro , ruiz cobo & trujillo bueno 1998 ) .
a typical application of miss to spectroscopic observations implies the solution of the radiative transfer equation a large number of times .
there are several reasons for this : the inversion problem must be solved iteratively due to the non - linear dependence of the spectral profiles on the physical quantities ; to evaluate the stellar flux , the solution of the radiation transfer equation must be calculated for several lines of sight ; and finally many spectral lines are needed to obtain a reliable model . therefore , spectral synthesis is the most time - consuming part of the code , and it is of paramount importance to have a fast integration method .
miss uses a new one , recently developed by bellot rubio , ruiz cobo & collados ( 1998 ) , which is based on the taylor expansion of the stokes parameter vector ( only the specific intensity in our case ) to fourth order in depth .
compared to other methods , the new technique turns out to be superior in terms of speed and accuracy .
it also gives an approximation to response functions ( rfs , the derivatives of specific intensity with respect to the perturbation of each free parameter ; see ruiz cobo & del toro iniesta 1992 , 1994 ) , which are a fundamental ingredient of the inversion algorithm .
the following quadrature has been used to evaluate the energy flux , @xmath6 , at a given wavelength , @xmath7 , from the specific intensity , @xmath8 , computed at each point @xmath9 across the stellar disk : @xmath10 where @xmath11 are the abscissae and @xmath12 are the weighting factors for gaussian integration . using solar model atmospheres from the literature , it has been numerically checked that a quadrature of order @xmath13 is enough to guarantee that the integration precision exceeds the observational data accuracy .
when taking into account the influence of the instrumental profile , @xmath14 , macroturbulence , @xmath15 , and rotation , @xmath16 , for the i - th annulus ( see appendix a for details ) , equation ( [ cuadr ] ) becomes : @xmath17 \ast p(\lambda ) \ast m(\lambda),\ ] ] where `` @xmath18 '' means convolution .
the instrumental profile , including telescope and spectrograph , can be provided either as input or approximated by a gaussian in which @xmath19 depends on the spectral resolution .
the macroturbulence profile is a gaussian with @xmath19 proportional to the amplitude of the macroturbulent velocity .
the algorithm is the same as used by sir , and a detailed description of it can be found in the above mentioned references .
a brief overview is presented below .
inversion of observed data is performed by using a non - linear least - squares algorithm which involves the minimization of a merit function : @xmath20 ^ 2,\ ] ] where index @xmath21 stands for the wavelength samples , indices `` obs '' and
`` syn '' refer to observed and synthetic data , respectively , and @xmath22 is the number of degrees of freedom .
minimization of the merit function is carried out through a marquardt algorithm ( marquardt 1963 ) , modified by applying a singular value decomposition ( svd ; press et al 1986 ) .
this algorithm requires the evaluation of the derivatives of @xmath23 with respect to each free parameter .
these derivatives can be directly expressed in terms of response functions , which are calculated by applying a quadrature ( eq . [ [ cuadr ] ] ) to the rf of the specific intensity provided by the integration method .
elimination of singularities ( via svd in this case ) is essential because certain free parameters ( e.g. , temperature in layers where the line profile has no sensitivity ) may produce a null effect on the emergent flux and consequently make the problem undetermined . to summarize , the steps of the procedure are : \1 .
choice of a guess model and a number of nodes for each free parameter .
calculation of flux , rf , and @xmath23 .
iterative evaluation of a new model after application of the marquardt algorithm , until minimization of @xmath23 is reached .
subsequent modification of the number of nodes and iteration from step 2 until @xmath23 ceases to decrease significantly .
the sir method has been successfully applied to the sun taking advantage of its spatially resolved disk and the high quality of solar observations ( see references in 2 ) .
the sun is also the best known late - type star and provides a unique opportunity to field - test the potential of the miss technique for the sun seen as a star .
the widely used high - resolution solar flux atlas of kurucz et al .
( 1984 ) was selected as the source of the line profiles .
its complete spectral coverage in the 2960130000 region , with a very high signal - to - noise ratio ( @xmath24 2500 ) , and the extremely high spectral resolution ( @xmath25 400000 ) makes it ideal for our purposes .
furthermore , allende prieto & garca lpez ( 1998 ) have recently verified the reliability of the wavelength calibration of this atlas by measuring the central wavelength shifts of 1446 lines with accurate laboratory determinations .
it was required that the input lines to be included in the inversion be selected from the compilation of solar lines by meylan et al .
( 1993 ) , who identified clean line wings in the same atlas by fitting voigt profiles .
we also impose the condition that the atomic transition probability had been measured at the oxford furnace ( see , e.g. , blackwell & shallis 1979 ) .
their oscillator strengths ( @xmath26 ) are usually recognized as the most accurate ones present in the literature .
several tests were carried out to study the dependence of the method on the accuracy of the oscillator strengths used .
it turned out that when artificially introducing errors in the @xmath27 values larger than @xmath24 5 % the inversion returns errors larger than 2% in the temperature stratification .
furthermore , special attention was given to the chemical elements selected .
we mainly focused on lines of elements which show an agreement between their photospheric and meteoritic abundances , as it is more likely for the abundance errors to be smaller .
although iron does not satisfy this condition , it was felt that the significant increase in the number of useful lines justified its inclusion .
finally , in an effort to minimize the problems related to the lack of knowledge of the collisional damping of the lines , features stronger that 80 m ( equivalent width ) were rejected .
the method is able to extract information only on the photospheric region corresponding to the depth range where the employed lines are formed . to improve the sampling of the photospheric domain , whole line profiles or just single wings
were accepted as input data following the considerations of the line cleanness by meylan et al .
( 1993 ) .
a total of 40 absorption lines produced by neutral iron , titanium , chromium and calcium entered the inversion program .
the wings of the stronger @xmath7 6162 were included with the aim of extending the depth coverage .
this is one of the few spectral lines for which the collisional enhancement of the wings is well known ( see allende prieto , garca lpez & trujillo bueno 1997 ) , there being a reasonable agreement between theoretical analyses and semi - empirical measurements .
this consideration is mandatory in order to wide the input information without introducing new uncertainties .
the collisional width of the line was interpreted in terms of collisions with neutral hydrogen , as computed by spielfiedel et al .
table 1 lists the lines entering the inversion and their parameters .
an attempt was made to employ the spectral information contained in both the continuum and the line shapes .
the observed continuum may differ strongly from the true one in the blue ( @xmath28 5000 ) due to the line absorption , so we can determine only a lower limit for the flux at these wavelengths .
even the measurement of the pseudo - continuum is a non - trivial matter : the discrepancies among different authors are as high as 8% ( see , e.g. , burlov - vasiljev , gurtovenko & matvejev 1995 ) . the slope of the continuum is closely linked to the temperature gradient in the photosphere in such a way that the employment of the information stored in the continuum requires highly accurate spectrophotometry .
this led us to exclude the absolute calibration of the line profiles entering the inversion code , and only line profiles normalized to their local continuum were used .
starting from an isothermal model photosphere , step by step the depth dependence of the temperature is allowed to be modified in a successively increasing number of nodes .
no vertical velocity gradient was allowed for , and the microturbulence was assumed to be constant with depth .
a gaussian profile was employed to represent the instrumental profile .
three snapshots of the inversion process are shown in fig . 1 ( ordered a , b , and c ) .
the evolution of the synthetic profiles for several of the lines included in the inversion is shown and compared with the solar observations ( dots ) .
a constant - temperature model photosphere ( dotted line ) does not produce any line absorption , while the final model ( solid line ) provides lines which get very close to the observed features .
starting from the solar abundances listed by anders & grevesse ( 1989 ) , the successive modification of the abundance of the considered elements was allowed for , providing an excellent final fit .
this resulted in no variation of any of the abundances except for iron , which dropped from the initial value log n(fe ) @xmath29 7.67 to 7.48 ( where log n(h ) @xmath29 12 ) , very close to the meteoritic abundance .
using the synodic equatorial velocity of 1.88 km s@xmath30 and the microturbulence of 0.6 km s@xmath30 ( consistent with takeda 1995 ) , the code arrives at 1.7 km s@xmath30 for the macroturbulent velocity . as an example
, one of the lines employed in the inversion procedure is displayed in more detail in fig .
2 , comparing the final synthetic profile ( solid line ) with the solar feature ( circles ) .
the point has been reached where no closer fit is possible without taking the line asymmetries into account .
the observed asymmetries have their origin mainly in the correlation between temperature and convective velocities , as observed in the solar granulation , although blends , isotopic or hyperfine splitting might be responsible for particular cases ( gray 1980 ; dravins , lindegren & nordlund 1981 ; kurucz 1993 ) . a more complex treatment than a plane - parallel one - dimensional static single - component model photosphere
is required to reproduce it ( e.g. , dravins & nordlund 1990a , b ; rast et al .
1993 ) .
figure 3 shows the final model photosphere , which we will refer to as miss hereafter , compared with other well - known models from the literature : kurucz ( 1992 ) , holweger & mller ( 1974 ; hereafter holmu ) , vernazza , avrett & loeser ( 1981 ; hereafter valc ) , and gingerich et al .
( 1971 ; harvard smithsonian reference atmosphere , hereafter hsra ) .
kurucz s model is based on theoretical calculations assuming lte , hydrostatic equilibrium and taking the line blanketing into account .
the model of howeger & mller was empirically adjusted to reproduce continuum observations in the optical and near infrared , as well as equivalent widths and residual intensities of spectral lines .
the valc empirical model was based on observations of the ultraviolet continuum , taking into account the departures from lte . and
hsra was basically designed to reproduce observations of the solar continuum .
the error bars shown for the miss model have been computed following the recipe described in ruiz cobo et al .
( 1997 ) and limit the depth range where the lines employed sample the atmosphere ; that is , where the model is meaningful .
the miss model shows , in general , a very similar temperature dependence to that of the other models , tending to be cooler in the upper part of the photosphere and hotter in the inner part . insight into the quality of the model photosphere obtained can be achieved by comparing the behavior of predicted and observed features not included in the empirical modeling . choosing what we consider as the most significant features ,
systematic comparisons with other well - established model photospheres for the sun are shown below .
the center - to - limb variation observed in the sun maps the temperature stratification of the solar photosphere and hence is probably one of the best tools for testing a one - dimensional model photosphere .
figure 4 shows the comparison at @xmath31 3500 , 4163 , 4774 , 5799 , 7487 , and 10990 between polynomial fits to the continuum observations of neckel & labs ( 1994 ) and the predictions from the different models considered .
the difference between the observed and computed intensities against @xmath9 ( the cosine of the position angle on the solar surface with respect to our line of sight ) is plotted , normalized to their corresponding values at the disk center .
holmu reproduces in detail the observed variation at @xmath32 @xmath9 m , the agreement being worse at shorter wavelengths . to make a fully legitimate comparison at short optical wavelengths ( @xmath28 5000 )
, it will be necessary to clarify first whether or not the thousands of lines in the blue part of the optical spectrum do seriously change the observed limb - darkening from its shape in the true continuum .
hsra provides the best general comparison at all wavelengths , while the limb - darkening predicted by the miss model keeps close to the observations for the whole optical range .
there are observations available for the variation of line profiles across the disk .
balthasar ( 1988 ) lists the center - to - limb variation for the equivalent widths of 143 lines , and shows that the behavior of the different lines is not homogeneous .
a comparison between the observations and predictions from the model atmospheres is shown in figure 5 .
the variation of the equivalent widths normalized to the center of the disk is shown for three lines in common between our line list and balthasar s measurements .
it can be seen how the absolute synthetic equivalent widths ( with values in the range 70100 m ) are underestimated with respect to the observed ones , arising from the use of the van der waals approach for the collisional enhancement of the line wings .
although the chemical abundances of the different elements are able to alter the curves , and the temperature dependence of the collisional width is more likely to be t@xmath33 , rather than the van der waals dependence of t@xmath34 ( e.g. , allende prieto et al .
1997 ) , we deduce from the figure that the model obtained from the inversion behaves remarkably better than the others . caution
must be taken regarding the hsra and valc models : both were constructed under nlte and the synthetic profiles shown here were calculated under lte . a more precise way of carrying out this comparison would be to use directly the wing profiles of lines with accurate broadening parameters , such as the sodium d lines and the @xmath35@xmath36 lines at @xmath31 6102.72 , 6122.22 , and 6162.17 .
as previously noted , although there is some disagreement in the literature , the absolute solar flux can be used as a lower limit for the true continuum .
this may help to discard model atmospheres predicting lower true continuum fluxes than the measured pseudo - continuum .
figure 6 shows the comparison between the measurements by neckel & labs ( 1984 ; triangles ) , the pseudo - continuum estimated from the correction of the line absorption observed in the high - resolution solar spectrum of kurucz et al .
( 1984 ; dots ) , and the theoretical true continua as predicted by several classical solar models .
kurucz ( 1992 ) claims that his solar model reproduces the solar observations of neckel & labs ( 1984 ) with high accuracy and , according to this , the true continuum calculated by taking no line absorption into account is well over the pseudo - continuum .
that is not the case for the continua associated with two of the empirical models , hsra and valc , being the continuum predicted by holmu higher than the pseudo - continuum but not as much as kurucz s model .
the model we obtained from only the inversion of line profiles gives rise to a continuum very similar ( somewhat brighter ) to that from the theoretical model by kurucz .
this is a remarkable property , because the miss model was constructed from line profiles normalized to the local continuum without considering any data on the absolute flux at any wavelength or any other photometric information .
the @xmath7 6162.17 line was already included in the inversion procedure and its synthetic profile obtained using the miss model provides an excellent fit to the observed solar atlas , as can be seen in the upper panel of figure 7 .
this is also the case for the other two lines corresponding to the @xmath35@xmath36 multiplet .
it is also worthwhile noting the remarkably good fit obtained for other lines surrounding the @xmath7 6162.17 line , not considered in the inversion , for which no necessarily high - quality 6161 and 6162 lines ] transition probabilities have been used .
another important comparison associated with strong features is shown in the lower panel of figure 7 .
the predicted wings of the sodium d doublet , which are supposed to cover a wide depth range of the photosphere and be formed under the lte approximation ( covino et al .
1993 ) , nicely reproduce the observed solar profiles .
apart from the lines included in the inversion , it is of great interest to compare predicted and observed profiles of other lines for which accurate transition probabilities are available .
a very suitable group of lines for this purpose are the lines measured by obrian et al .
( 1991 ) with an accuracy better than 5 % , which belong to the meylan et al .
( 1993 ) selection , but are not included in the oxford line lists .
figure 8 shows this comparison for three lines located at @xmath31 4635.85 , 6027.05 , and 6481.87 .
we have assumed an iron abundance of log n(fe)@xmath29 7.48 , a macroturbulent velocity of 1.7 km s@xmath30 ( the values obtained from the inversion ) , and a constant microturbulence of 0.6 km s@xmath30 .
the profiles were computed in lte in all cases .
it can be seen that the best fits correspond to miss , followed very closely by valc , and then by hsra ( although these are nlte models ) , kurucz , and holmu , respectively .
it can be concluded then that the miss model reproduces , in general , several key observational constraints with a quality similar to that of other well - known models and behaves better than several of these models when focusing on particular comparisons .
as explained in the introduction , the final goal of this work is to develop a spectroscopic tool for deriving semi - empirical model photospheres for late - type stars in general , and for metal - poor stars in particular . in principle
, the information used here for the solar case ( normalized line profiles with very high resolution and high signal - to - noise ratio ) can be also available for other bright stars .
allende prieto et al . (
1995 ) have shown preliminary observations of a metal - poor star obtained with @xmath37 and s / n@xmath24 300600 .
observations of the same and other stars with slightly higher resolution and better signal - to - noise ratios have been already carried out and are being processed .
these values of @xmath38 and s / n are , however , significantly smaller than those associated with the solar atlas we have used .
furthermore , the wavelength coverage achieved in the stellar observations is much smaller than that available for the sun .
it will be necessary then to test in detail what happens in the inversion when using different numbers of lines combined with spectra of lower quality .
this is , in any case , a practical problem which can be solved by using ultra - high resolution spectrographs to observe very bright objects . a different problem associated with this semi - empirical procedure is related to the spectral synthesis itself in terms of knowledge of the atomic and molecular opacities ( especially for metal - poor stars ) , effects of non - lte , convection and overshooting , etc .
a way for partially minimizing these effects would be to use as many lines as possible with accurate transition probabilities and damping constants , and also to focus as much as possible on near - infrared lines where the opacity problems decrease .
the multiline inversion of stellar spectra ( miss ) procedure developed here has been proved able to find an lte temperature stratification for the solar photosphere from normalized high - resolution line profiles which reproduces the solar continuum and the limb - darkening sufficiently well and is in excellent agreement with observed strong and weak spectral lines with accurate line data
. a better comparison of its performance in reproducing the limb - darkening of the wings of strong lines with accurate damping constants is now being studied .
it is found that none of the input abundances ( anders & grevesse 1989 ) for chromium , calcium , and titanium need to be changed to find the best fit to the observations , but the iron abundance is preferred to vary from the initially assumed value ( 7.67 , in the scale where the hydrogen abundance is 12 ) to the lower ( approximately meteoritic ) abundance 7.48
. a better fit to the solar flux line spectrum would require the abandonment of the hypothesis of hydrostatic equilibrium and the introduction of multi - component models or velocity fields inducing asymmetries in the line profiles .
our work in the near future will consider this improvement .
this procedure of semi - empirical modeling of stellar photospheres can potentially be applied to other late - type stars .
while other classical methods of empirical modeling would require observations not available for stars ( the limb - darkening can not be measured yet , and flux calibrated spectra are affected by a number of larger uncertainties ) , the normalized line profiles used by miss are already at hand .
we express our gratitude to l. r. bellot rubio , m. collados , e. simmoneau , and j. c. del toro iniesta for interesting comments on the draft .
we also thank t. meylan , who kindly provided a digital version of his line list .
this article has been corrected for english and style by terry mahoney ( research division , iac ) .
nos / kitt peak fts data used here were produced by nsf / noao .
this research has made use of the vald database and was partially supported by the spanish dges under projects pb95 - 1132-c02 - 01 and pb95 - 0028 .
the effect of the rotation on the spectral profile has been estimated by taking advantage of the knowledge of the specific intensity of the local continuum for each spectral line @xmath39 , where @xmath40 is the number of points of the quadrature used .
let @xmath41 be the normalized radius corresponding to the point with cosine of the heliocentric angle @xmath11 .
the stellar disk is considered divided into @xmath40 annuli with internal and external radii @xmath42 and @xmath43 , respectively , defined as follows : @xmath44 , @xmath45 , @xmath46 , and @xmath47 .
let us assume that @xmath48 varies linearly between @xmath11 and @xmath49 , in the following way : the line - of - sight component of the rotation velocity will be the same for all points located at an equal distance @xmath51 from the central meridian : @xmath52 .
the flux crossing the fraction of area of the @xmath53-th annulus at a distance between @xmath51 and @xmath54 is given by @xmath55 + 2 \ ; b_i ( y_i - y_{i-1}),\ ] ] allende prieto , c. , garca lpez , r. j. , lambert , d. l. , & gustafsson , b. 1995 , in iau symp .
176 , stellar surface structure , poster proceedings , ed .
k. g. strassmeier ( vienna : institut fr astronomie der universitt wien ) , 107 rodrguez hidalgo , i. , ruiz cobo , b. , & collados , m. 1996 , in asp conf .
109 , cool stars , stellar systems , and the sun , 9@xmath62 cambridge workshop , ed .
r. pallavicini , & a. k. dupree ( san francisco : asp ) , 155 ruiz cobo , b. , del toro iniesta , j. c. , rodrguez hidalgo , i. , collados , m. , & snchez almeida , j. 1996 , in asp conf .
109 , cool stars , stellar systems , and the sun , 9@xmath62 cambridge workshop asp conference , eds .
r. pallavicini & a.k .
dupree ( san francisco : asp ) , 155 del toro iniesta , j. c. , & ruiz cobo , b. 1995 , in la polarimtrie , outil pour ltude de lactivit magntique solaire et stellaire , eds .
n. mein & s. sahal - brchot ( paris : observatoire de paris - meudon ) , 127 westendorp plaza , c. , del toro iniesta , j. c. , ruiz cobo , b. , martnez pillet , v. , lites , b. w. , & skumanich , a. 1997a , in 1@xmath63 advances in solar physics euroconference .
advances in the physics of sunspots , eds . b. schmieder , j. c. del toro iniesta , & m. vzquez , asp conf .
118 , 197 westendorp plaza , c. , del toro iniesta j. c. , ruiz cobo , b. , martnez pillet , v. , lites , b. w. , & skumanich , a. 1997b , in 1@xmath63 advances in solar physics euroconference .
advances in the physics of sunspots , eds . b. schmieder , j.c .
del toro iniesta , & m. vzquez , asp conf .
118 , 202 cccccccc & 4578.557 & 2.52 & 0.697 & & 5922.115 & 1.05 & 1.410 + & 5512.986 & 2.93 & 0.447 & & 6092.799 & 1.89 & 1.323 + & 6161.297 & 2.52 & 1.266 & & 6258.109 & 1.44 & 0.299 + & 6162.183 & 1.89 & 0.097 & & 6303.762 & 1.44 & 1.510 + & 6166.441 & 2.52 & 1.142 & & 6312.244 & 1.46 & 1.496 + & 6455.604 & 2.52 & 1.290 & & 7357.735 & 1.44 & 1.066 + & 6499.656 & 2.52 & 0.818 & & 4602.006 & 1.61 & 3.150 + & 4801.028 & 3.12 & 0.131 & & 5225.533 & 0.11 & 4.790 + & 4964.931 & 0.94 & 2.527 & & 5247.057 & 0.09 & 4.950 + & 5272.002 & 3.45 & 0.422 & & 5916.254 & 2.45 & 2.990 + & 5300.751 & 0.98 & 2.129 & & 5956.700 & 0.86 & 4.610 + & 5312.859 & 3.45 & 0.562 & & 6082.715 & 2.22 & 3.570 + & 5787.922 & 3.32 & 0.083 & & 6151.623 & 2.18 & 3.300 + & 7355.899 & 2.89 & 0.285 & & 6173.342 & 2.22 & 2.880 + & 4758.122 & 2.25 & 0.481 & & 6200.321 & 2.61 & 2.440 + & 4759.274 & 2.25 & 0.570 & & 6297.801 & 2.22 & 2.740 + & 5113.445 & 1.44 & 0.727 & & 6481.878 & 2.28 & 2.980 + & 5295.781 & 1.05 & 1.577 & & 6498.945 & 0.96 & 4.700 + & 5490.154 & 1.46 & 0.877 & & 6750.161 & 2.42 & 2.620 + & 5866.457 & 1.07 & 0.784 & & 6978.861 & 2.48 & 2.500 + | an inversion technique has been developed to recover lte one - dimensional model photospheres for late - type stars from very high resolution , high signal - to - noise stellar line profiles .
it is successfully applied to the sun by using a set of clean , , , and normalized line profiles with accurate transition probabilities and taking advantage of the well understood collisional enhancement of the wings of the line at 6162 .
line and continuum center - to - limb variations , continuum flux , and wings of strong metal lines are synthesized by means of the model obtained and are compared with solar observations , as well as with predictions from other well known theoretical and empirical solar models , showing the reliability of the inversion procedure .
the prospects for and limitations of the application of this method to other late - type stars are discussed . |
fluctuation theorems or fluctuation relations @xcite are a variety of exact equalities about statistics of entropy production or dissipated work that are held even in far from equilibrium regimes . in near - equilibrium region
, these relations reduce to the famous fluctuation - dissipation theorems ( fdts ) @xcite .
the discovery of these fluctuation relations significantly advances our understanding about nonequilibrium physics , and particularly about the second law of thermodynamics of small systems @xcite .
the fluctuation relations are very relevant with the concept of time reversal @xcite .
for instance , under the framework of markovian stochastic systems , previous work has proved that a majority of them can be derived by a ratio of the probability densities of observing a trajectory in a original system and the reversed trajectory in the time - reversed system .
very recently , chetrite and gawedzki @xcite further elaborated this observation and presented a generalized time reversal definition on continuous diffusion processes .
different from conventional definition of time reversal as simply changing time parameter in a stochastic dynamics into minus , they explicitly defined time reversal as a splitting of drift vector into irreversible and reversible parts which possess distinct rules under the transformation of @xmath0@xmath1@xmath2 .
because of freedom of the splitting , a variety of time reversal and corresponding fluctuation relations are obtained , e.g. , the hatano - sasa equality @xcite arising from a novel time reversal with nonzero reversible drift . the importance of the generalized time reversal was shown again when we understood the origin of a generalized integral fluctuation relation ( gifr ) on general diffusion processes @xcite .
in addition to the continuous diffusion process , another typical and important markovian process is master equation with discrete states and continuous time @xcite .
although the latter is more general than the former in principle , due to their highly formal analogy , many results and evaluations about the fluctuation relations in the diffusion processes could be established correspondingly in the master equation systems @xcite .
nevertheless , whether a similar time reversal definition exists and how to define it are not investigated yet .
we must emphasize that it is not trivial as one might think of at first glance . except for positive transition rates , in these master equation systems
there are not quantities such as drift vector and diffusion matrix that have intuitive rules of transformation as time reversed @xcite .
somewhat surprisingly , here we will show that an physically relevant answer indeed exists .
the organization of this work is as follows . in sec .
[ review ] , we briefly review a gifr in the master equation systems that we found very recently @xcite .
the reason we use the gifr rather than other famous fluctuation relations is its generality .
additionally , the necessary of extending the conventional time reversal will be brought forth naturally in deriving the gifr . in order to interpret an unknown matrix in this relation , in sec .
[ splitting ] we present a generalized time reversal in these systems as a splitting of rate matrix into reversal and irreversible parts .
the consequences of this new definition will be discussed in sec .
[ discussion ] , which includes reinvestigation of the fluctuation relations from a point of view of the splitting , generalization of total entropy production and conditions of the detailed balance .
section [ conclusion ] is the summary .
assume a markovian process with discrete states and continuous time is described by a master equation @xcite @xmath3_n , \label{forwardeq}\end{aligned}\ ] ] where @xmath4 is the state index which may be a vector , the @xmath5-dimensional column vector @xmath6@xmath7@xmath8 is the probability of the system at individual states at time @xmath0 , and the matrix element @xmath9=@xmath10@xmath110 ( @xmath12@xmath13@xmath4 ) is the time dependent or independent rate and @xmath14=@xmath15@xmath16 .
given a normalized positive column vector @xmath17=@xmath18 and a @xmath5@xmath19@xmath5 matrix @xmath20 whose elements @xmath21@xmath7@xmath22 ( @xmath23 ) satisfy condition @xmath24@xmath25@xmath22@xmath26 and @xmath27@xmath7@xmath28 , we found that the inner product @xmath29 ( @xmath30@xmath31@xmath0 ) is @xmath30-invariable if the vector @xmath32 satisfies a perturbed backward equation @xmath33 , \label{backwardeq}\end{aligned}\ ] ] where the final condition @xmath34=@xmath35 , and @xmath5-dimension column vector @xmath36 .
this is easily proved by noticing the time differential property and the transpose property of a matrix . employing the feynman - kac and girsanov formulas for discrete jump processes @xcite ,
the solution of ( [ backwardeq ] ) has a path integral representation given by @xmath37({\bf x}(\tau))d\tau } d_{{\textbf x}(t)}\rangle={\textbf f}^{\rm t}(t ) { \textbf d}\end{aligned}\ ] ] and the integrant in the functional is @xmath38&=&f_{\textbf x(\tau)}^{-1}\left[-\partial_{\tau}\textbf f+\textbf
h\textbf f+\textbf a\textbf 1\right]_{{\textbf x}(\tau)}+{\cal q}[f_{\textbf x(\tau)}^{-1}\textbf a]\label{functional}\end{aligned}\ ] ] with @xmath39&=&-b_{{\textbf x}(\tau){\textbf x}(\tau)}-\ln\left[1+\frac{b_{{\textbf x}(\tau^+){\textbf x}(\tau^-)}(\tau)}{h_{{\textbf x}(\tau^+){\textbf x}(\tau^-)}(\tau ) } \right]\sum_{i=1}^k\delta(\tau-\tau_i),\label{functionalg}\end{aligned}\ ] ] where @xmath40 @xmath41 is the expectation over all trajectories @xmath42 generated from the system ( [ forwardeq ] ) with a fixed state @xmath12 at initial time 0 , @xmath43 is the discrete state of the system at instant time @xmath44 , @xmath45 and @xmath46 represent the states just before and after a jump occurring at time @xmath44 , respectively , and we have assumed the jumps occur @xmath47 times for a trajectory . by selecting different @xmath48 and @xmath49 , the gifr ( [ gift ] ) may be reduced into different fluctuation relations in the literature @xcite .
although ( [ backwardeq ] ) seems complicated , in fact it could be arranged into a concise form : @xmath50_{\tilde{n}},\label{timereversal}\end{aligned}\ ] ] where the elements of the probability @xmath51 are @xmath52^{-1}f_n(t')v_n(t')\end{aligned}\ ] ] with @xmath53=@xmath0@xmath15@xmath30 , and the elements of the matrix @xmath54 are @xmath55 \label{ratetimereversal}\end{aligned}\ ] ] for @xmath12@xmath13@xmath4 and @xmath56@xmath7@xmath57 , and @xmath58 represents an index whose components are the same or the minus of themselves depending on whether they are even or odd under time reversal @xmath0@xmath1@xmath2 . considering that the parameter @xmath53 is analogous to a reversed time and @xmath59@xmath7@xmath60 as specifically selecting @xmath61
whose elements are the probability flux @xmath62_{mn}$]=@xmath63@xmath15@xmath64 , we simply name ( [ timereversal ] ) as generalized time reversal of the original system ( [ forwardeq ] ) . however , why @xmath61 is almost arbitrary instead of equaling probability flux only was not fully understood previously .
hence , defining the generalized time reversal from a different point of view seems very essential .
let us begin with two general matrix identities : @xmath65}_n-{[{\textbf m}{\bf a}]}_nb_n+a_n{[{\textbf m}^{\rm t}{\bf b}]}_n&=&[{\textbf s}^{\rm t}{\bf b}]_n-[{\textbf s}{\textbf 1}]_nb_n,\label{identity1 } \\ { [ { \textbf m } ( { \bf a. b})]}_n+{[{\textbf m}{\bf a}]}_nb_n - a_n{[{\textbf m}^{\rm t}{\bf b}]}_n&=&[{\textbf j}^{\rm t}{\bf b}]_n-[{\textbf j}{\bf 1}]_nb_n\label{identity2 } , \end{aligned}\ ] ] where both @xmath66 and @xmath67 are @xmath5-dimensional vectors , @xmath68 is @xmath5@xmath19@xmath5 * matrix with @xmath69@xmath7@xmath70 * , @xmath71=@xmath72 represents an array multiplication of the vectors , and the matrixes @xmath73 and @xmath74 are constructed by @xmath68 and @xmath66 : @xmath75 they are symmetric and antisymmetric , respectively .
proving ( [ identity1 ] ) and ( [ identity2 ] ) is straightforward .
now we assume that the rate matrix can be splitted into a sum of reversible " and irreversible " matrixes , namely @xmath76=@xmath77+@xmath78 .
particularly , we require @xmath79@xmath11@xmath80 ( @xmath12@xmath13@xmath4 ) , the reason of which will be seen shortly .
it is easily to see that @xmath79 is always greater than zero for @xmath10@xmath110 , though the sign of @xmath79 is indefinite .
we further assume @xmath81 is transformed into @xmath82=@xmath83+@xmath84 under time reversal and @xmath85 obviously , time - reversed matrix @xmath86 still possesses rate interpretation . using these definitions and identities we rewrite the right - hand side of ( [ timereversal ] ) as @xmath87_{\tilde{n}}&=&\frac{1}{{\bf f}(t){\bf d}^{\rm t}}\{[\widetilde{\bf h}^{\rm irr}({\bf f.v})]_{\tilde{n}}+[\widetilde{\bf h}^{\rm rev}({\bf f.v})]_{\tilde{n}}\}\\ & = & \frac{1}{{\bf f}(t){\bf d}^{\rm t}}\{({\textbf h}^{\rm t}{\textbf v})_n-({\textbf h}{\textbf f})_nv_n-[(\textbf j^{\rm irr}-\textbf s^{\rm rev})\textbf 1]_nv_n\nonumber\\ & & \hspace{1.2cm}+[(\textbf j^{\rm irr}-\textbf s^{\rm rev})^{\rm t}\textbf v]_n\},\nonumber\end{aligned}\ ] ] where @xmath88 comparing ( [ expandingreversedh ] ) with ( [ backwardeq ] ) , we immediately find that @xmath89_{mn}&=&({\bf j}^{\rm irr})_{mn}-({\bf s}^{\rm rev})_{mn}\\ \label{amatrixdef2}&=&[{\bf j(f)}]_{mn}-2h^{\rm rev}_{nm}f_m,\end{aligned}\ ] ] ( [ amatrixdef2 ] ) implies why the matrix @xmath61 is almost arbitrary : the reversible matrix @xmath77 to be determined is responsible for this freedom .
in addition , we also notice that the condition of @xmath24@xmath25@xmath22@xmath110 mentioned in last section is identical with @xmath79@xmath11@xmath90 here . in practice , we may prior know the matrix @xmath61 or time - reversed rate matrix @xmath91 . under these circumstances ,
the splitting can be constructed conversely as @xmath92 or @xmath93/2,\\ \label{splitdefbyreveredhirr } h^{\rm irr}_{mn}(t')=[h_{mn}(t')+\widetilde h_{\tilde{m}\tilde{n}}(s)]/2.\end{aligned}\ ] ] regardless of how a splitting of the rate matrix is achieved , substituting ( [ amatrixdef1 ] ) into the integrand ( [ functional ] ) we can obtain its new expression given by $ ] and @xmath94 $ ] without explicit statement , which would not result in confusion because of ( [ amatrixdef1 ] ) . ]
@xmath95 & = & -\frac{d}{d\tau}\ln f_{{\bf x}(\tau)}(\tau)+{\cal j}[\textbf 1,\textbf h^{\rm irr},\textbf h^{\rm rev}],\end{aligned}\ ] ] where the second term on the right - hand side is @xmath96\sum^k_{i=1}\delta(\tau-\tau_i).\end{aligned}\ ] ] ( [ newfunctional1 ] ) is a consequence of the arbitrariness of @xmath48 in ( [ reversedprobdefinition ] ) , a brief explanation of which is given in appendix a. noting ( [ trajectorymediaentropy ] ) also satisfies a fluctuation relation if the initial distribution of the system is uniform .
( [ newfunctional1 ] ) and ( [ trajectorymediaentropy ] ) are the central results of this work .
because the physical interpretation of the gifr ( [ gift ] ) , its connection with the other fluctuation relations , and its detailed version have been partially investigated @xcite , in the remainder we only present several new results obtained from a point of view of the splitting . if @xmath48 is the system s probability vector @xmath97 and the splitting is prior known from physical consideration , e.g. , time reversibility below , ( [ newfunctional1 ] ) is the balance equation of trajectory entropy : the first term on the right - hand side is the change in trajectory system entropy , the second term is the change in trajectory environmental entropy along a specified trajectory , and the left - hand side is the total trajectory entropy .
this interpretation has been widely accepted @xcite . to our knowledge , however , the expression of the trajectory environmental entropy ( [ newfunctional2 ] ) that involves both even and odd discrete variables is firstly given here .
this new formula also reminds us that the total entropy production for the master equation systems with irreversible and reversible rate matrixes is @xmath98 the inequality may be proved easily by using ( @xmath99@xmath100)@xmath101@xmath102 or using the jensen inequality for the gifr ( [ gift ] ) .
the expression of ( [ entropyprod ] ) is significantly distinct from the classical entropy production formula given by schnakenberg @xcite that involves only even variables . for a time - independent master equation system that has equilibrium state @xmath103 ,
we may physically require it invariable under time - reversal , namely , @xmath104@xmath7@xmath81 .
according to ( [ splitdefbyreversedhrev ] ) and ( [ splitdefbyreveredhirr ] ) an unique splitting is then @xmath105 under this circumstance the irreversible and reversible parts have distinctive properties : @xmath106 obviously , if the discrete master equation system involves only even states , which is exclusively the object investigated in numerous references @xcite , the reversible rate matrix must vanish . because the entropy production is zero when the system is in equilibrium state , according to ( [ entropyprod ] ) , all terms on its right hand side
must vanish .
hence we obtain conditions of detailed balance on the rate matrix and the state which are respectively @xmath107 we must emphasize that , except for the last equation ( [ steadycondition ] ) that is from the steady - state requirement and was ignored in previous literature @xcite , the former two equations are fully equivalent with the conventional detailed balance condition @xmath108 which may be easily checked using ( [ splitdefbyreversibleh ] ) .
we see an unusual side of a time - reversible master equation system with nonvanishing @xmath109 : the probability flux @xmath110 between two states @xmath12 and @xmath4 may be not zero even if the system is in an equilibrium state ; both the irreversible and reversible rate matrixes have contributes to the total entropy production .
the latter point can be seen more clearly when we rewrite ( [ entropyprod ] ) as @xmath111 by taking ( [ detailbalancerev])-([steadycondition ] ) into account . to our best knowledge
, there are very fewer stochastic jump processes with both even and odd variables in the literature . in appendix
b we present a simple mathematical model to exemplify their intriguing features . for a time - dependent master equation system , we may select the simplest case with @xmath61=0 .
then ( [ splitdefbyarev ] ) and ( [ splitdefbyairr ] ) become @xmath112 we used a new notation @xmath113 instead of @xmath114 to indicate the speciality of this splitting . under this case ,
( [ newfunctional1 ] ) is @xmath115 & = & -\frac{d}{d\tau}\ln f_{{\bf x}(\tau)}(\tau)+{\cal j}[\textbf 1,{\cal h}^{\rm irr},{\cal h}^{\rm rev}],\end{aligned}\ ] ] and @xmath116=&&-\frac{1}{f_{{\bf x}(\tau)}}\sum_{m\neq { { \bf x}}(\tau)}[h_{{\bf x}(\tau)m}f_m - h_{m{\bf x}(\tau)}f_{{\bf x}(\tau)}]\nonumber\\ & & -\ln\frac { f_{\bf x(\tau^-)}}{f_{\bf x(\tau^+ ) } } \sum^k_{i=1}\delta(\tau-\tau_i).\end{aligned}\ ] ] if one further selects @xmath48 to be the instant equilibrium solution @xmath117 or the instant nonequilibrium steady - state solution @xmath118 of the master equation if it has , the gifr is respectively the famous jarzynski equality about dissipated work @xcite or hatano - sasa equality about excess entropy @xcite .
noting the first term of the right - hand side of the above equation in both cases vanishes because of the steady - state condition .
it is worth pointing out that for the former equality , the detailed balance conditions ( [ detailbalancerev ] ) and ( [ detailbalanceirr ] ) imply that such a splitting is trivial because of @xmath119=@xmath120 and @xmath121=@xmath122 , and ( [ jehefunctional ] ) has other three different but equivalent expressions : @xmath123\nonumber\\&&=\ln
\frac { h^{\rm irr}_{{\bf x(\tau^-)}{\bf x(\tau^+)}}-h^{\rm rev}_{{\bf x(\tau^-)}{\bf x(\tau^+)}}}{h^{\rm irr}_{{\bf x(\tau^+)}{\bf x(\tau^-)}}+h^{\rm rev}_{{\bf x(\tau^+)}{\bf x(\tau^-)}}}.\end{aligned}\ ] ] on the other hand , if @xmath124@xmath7@xmath70 the integrand ( [ functional ] ) can be decomposed into @xmath125={\cal j}[{\bf f},{\bf 0}]+{\cal q}[f_{\textbf x(\tau)}^{-1}\textbf a].\end{aligned}\ ] ] this relationship is very intriguing .
in addition that the three terms above satisfy integral fluctuation relation @xcite simultaneously , all of them have important physical implications in the nonequilibrium steady - state thermodynamics @xcite .
here we do not repeat previous interpretations but present a simple understanding from a point of view of the splitting . for a nonequilibrium master equation system with @xmath126@xmath7@xmath127 and
assuming it has unique steady - state as external time - dependent parameter fixed , we have @xmath128=&-&\frac{d}{d\tau}\ln \frac{p_{{\bf x}(\tau)}}{{p}_{{\bf x}(\tau)}^{\rm ss } } + { \cal j}[{\bf p}^{ss},{\bf j(p^{\rm ss})}]\\ & = & -\frac{d}{d\tau}\ln\frac { p_{{\bf x}(\tau)}}{{p}_{{\bf x}(\tau)}^{\rm ss}}+{\cal j}[{\bf p}^{ss},{\bf 0}]+{\cal q}[\frac{{\bf j}(p^{\rm ss})}{p_{\bf x(\tau)}^{\rm ss}}]\label{entropyproductionfaces2}\\ & = & -\frac{d}{d\tau}\ln p_{{\bf x}(\tau)}+{\cal j}[{\bf 1},{\cal h}^{\rm irr } , { \cal h}^{\rm rev}]+{\cal q}[\frac{{\bf j}(p^{\rm ss})}{p_{\bf x(\tau)}^{\rm ss}}],\label{entropyproductionfaces3}\end{aligned}\ ] ] where ( [ newfunctional1 ] ) , ( [ newfunctional1neqss ] ) , ( [ decompositionj ] ) and steady - state condition @xmath129=0 are used .
( [ entropyproductionfaces2 ] ) implies that trajectory total entropy is the sum of trajectory relative entropy , trajectory excess entropy @xcite , and trajectory housekeeping entropy @xcite .
particularly , the sum of the first two terms in ( [ entropyproductionfaces3 ] ) that was called as nonadiabatic trajectory entropy @xcite is just @xmath130 $ ] , the ensemble average of which is @xmath131\rangle = & -&\sum_{m\neq n}\sum_nh_{nm}p_m\ln\frac{p^{\rm ss}_mp_n}{p_n^{\rm ss}p_m}\geq 0\end{aligned}\ ] ] according to ( [ entropyprod ] ) . in ref .
@xcite this quantity was also named as free energy dissipation .
motivated by the important idea of defining time reversal as a splitting of drift vector in continuous diffusion process , we present the same effort in the master equation system with discrete states .
different from the former , we define the time reversal in this system as a splitting of the rate matrix into irreversible and reversible parts . even that we find very analogous formulas and results
are revealed in these two systems .
e.g. ( [ trajectorymediaentropy ] ) corresponding ( 7.6 ) in the work of chetrite and gawedzki @xcite or ( 28 ) in the work of us @xcite .
the advantages of introducing this definition are obvious .
first , we explain the origin of the matrix @xmath132 in the gifr , which was somewhat mysterious to us before starting this work .
second , a variety of fluctuation relations in the master equation systems are unified into various rate matrix splitting .
this point was not acknowledged previously .
additionally , the relationships among these fluctuation relations become very clear from this splitting viewpoint , e.g. ( [ entropyproductionfaces2 ] ) and ( [ entropyproductionfaces3 ] ) .
finally , this definition also reminds us the importance of the reversible part of the rate matrix .
for instance , the expression of the total entropy production must be modified appropriately . to our knowledge ,
its existence and implications in the master equation systems were almost completely ignored for a very long time . + this work was supported in part by state key laboratory of software development environment sklsde-2011zx-20 and the national science foundation of china under grant no .
for simplicity in notations , we study a specific case of ( [ backwardeq ] ) with a final condition is @xmath35=1 . under this circumstance
the perturbed backward master equation has the time - reversal explanation ( [ timereversal ] ) and @xmath133 noting that the initial condition of @xmath134 is now @xmath135 .
it is worth emphasizing that we can replace the vector @xmath136 above by other normalized positive vectors , e.g. @xmath137=@xmath138 .
then we have a new relation @xmath139 where @xmath140 satisfies the same ( [ backwardeq ] ) except that @xmath141 therein including those in the matrix @xmath142 ( [ amatrixdef1 ] ) are substituted by @xmath143 and the finial condition becomes @xmath144=@xmath145 . according to ( [ reversedprobdefinition1 ] ) , ( [ reversedprobdefinition2 ] ) , and the path integral representation ( [ gift ] )
, we immediately obtain @xmath146({\bf x}(\tau))d\tau } \frac{f_{{\textbf x}(t)}(t)}{f_n(0 ) } \rangle= ^{n,0}\langle e^{-{\int_0^t\cal j}[\textbf f,\textbf a]({\bf x}(\tau))d\tau } \rangle\end{aligned}\ ] ] because of @xmath147$]=@xmath148 $ ] ( [ newfunctional1 ] ) is proved .
and @xmath149 ) variables , ( b ) jump process with only even variables ( 1 , 1@xmath150 , 2 , and 2@xmath150 ) .
the letters @xmath151 ( @xmath110 ) are respective rates . ] to show the unusual features of jump processes with both even and odd variables , we develop a simple model ; see fig .
( [ figure1])(a ) . assuming that the variable @xmath152=1 and @xmath153 are even and @xmath154 and @xmath149 are odd , under time reversal ( @xmath0@xmath155@xmath2 ) we have @xmath156 and @xmath157 .
although this model has eight rate parameters , they would not to be independent of each other if we required the model to have a steady - state solution satisfying the detailed balance principle . using ( [ detailbalancerev])-([steadycondition ] ) and performing a simple calculation , we find these rates must obey the following three constraints : @xmath158 the first equation is a consequence of ( [ detailbalancerev ] ) or ( [ detailbalanceirr ] ) , while the last two equations are from ( [ steadycondition ] ) .
the equilibrium solutions can be easily obtained , which are @xmath159 in addition , the total entropy production ( [ entropyproddetailed ] ) is @xmath160\ln\frac{p_{(1,\uparrow)}}{p_{(1,\downarrow)}}+[d_2p_{2,\uparrow)}-u_2p_{(2,\downarrow)}]\ln\frac{p_{(2,\uparrow)}}{p_{(2,\downarrow)}}\nonumber\\ & + & [ rp_{(1,\uparrow)}-lp_{(2,\uparrow ) } ] \ln\frac{(r+r')p_{(1,\uparrow)}}{(l+l')p_{(2,\uparrow)}}+[r'p_{(1,\downarrow)}-l'p_{(2,\downarrow ) } ] \ln\frac{(r+r')p_{(1,\downarrow)}}{(l+l')p_{(2,\downarrow)}},\end{aligned}\ ] ] where @xmath161 and @xmath162 are the transient probabilities of the model at distinct states started from an initial distribution .
reader may easily check that , if @xmath163=@xmath164 , the probability currents @xmath74 or the terms before these logarithmic functions above are nonzero
. it would be interesting to compare these results with those obtained in a conventional jump processes with only even variables , e.g. fig .
( [ figure1])(b ) .
its states are respectively marked by the variables 1 , 2 , 1@xmath150 , and 2@xmath150 , all of which are even under time reversal .
because of @xmath165=@xmath166 in the model ( a ) , here we additionally require @xmath167=@xmath168 .
hence , we obtain three constraints on the rates given by @xmath169 the equilibrium solutions @xmath170 and the classical total entropy production @xcite @xmath171 where @xmath172 and @xmath173 ( @xmath1741,2 ) are the transient probabilities of the model ( b ) at individual states started from an initial distribution . obviously , in the model ( a ) if one did not take the odd variable into account and naively used ( [ schnakenbergep ] ) , namely replacing @xmath175 and @xmath173 therein by @xmath161 and @xmath162 , respectively , he or she would find that the classical formula would not vanish as the system reaches the equilibrium states ( [ equilibriumsolutionsevenodd1 ] ) and ( [ equilibriumsolutionsevenodd2 ] ) . a numerical example to confirm these results
is shown in fig .
( [ figure2 ] ) . ) ( a ) and ( b ) the entropy production rates calculated by the classical schnakenbergs formula ( [ schnakenbergep ] ) and ( [ entropyproddetailed ] ) proposed in this work .
we choose @xmath176=1s@xmath177 , @xmath178=0.2s@xmath177 , @xmath179=0.4s@xmath177 , @xmath180=0.5s@xmath177 ; @xmath181=0.7s@xmath177 ; @xmath182=0.4s@xmath177 , @xmath183=0.2s@xmath177 , and @xmath184=0.8s@xmath177 , which satisfy ( [ contraintsevenodd1])-([contraintsevenodd3 ] ) .
the initial conditions are @xmath185=1.0 , and the others vanish . ]
99 bochkov g n and kuzovlev yu e 1977 _ sov .
. jetp _ * 45 * 125 | motivated by recent progresses in nonequilibrium fluctuation relations , we present a generalized time reversal for stochastic master equation systems with discrete states that is defined as a splitting of the rate matrix into irreversible and reversible parts .
an immediate advantage of this definition is that a variety of fluctuation relations can be attributed to different matrix splitting .
additionally , we also find that , the accustomed total entropy production formula and conditions of the detailed balance must be modified appropriately to account for the presence of the reversible part , which was completely ignored in the past a long time . |
large configuration interaction ( ci ) calculations of the helium ground state are performed here in order to more precisely elucidate the convergence properties of the ci expansion for this atom .
the general properties of the ci expansion have been known since the seminal work of schwartz @xcite , which provided the underlying foundation for the later analytic and computational investigations @xcite .
the ci expansion using single center orbitals is slowly convergent with respect to @xmath2 , the maximum angular momentum of any orbital included in the ci expansion . in particular , the leading term to the energy increment
is expected to behave at high @xmath2 as : @xmath10 although there have been a number of very large ci calculations performed on helium , all of the earlier calculations using analytic basis sets treat the higher @xmath2 contributions to the energy with less precision than the low @xmath2 contributions . typically , the number of radial orbitals for the high @xmath0 are smaller than the number of low @xmath0 .
the justification for this is that the high @xmath0 partial waves make a smaller contribution to the energy and other expectation values of than the low @xmath0 orbitals . at first sight
this approach would seem reasonable for obtaining estimates of the total energy .
however , this approach does lead to problems when studying the convergence properties of ci expansion itself . here
it is necessary to ensure that the successive contributions to the energy are obtained with the same _ relative _ accuracy and this can hardly be guaranteed with a radial basis that decreases in size as @xmath0 increases .
indeed , the evidence suggests that the dimension of the radial basis should be increased as @xmath2 increases if the relative accuracy of the energy is to be maintained @xcite .
the convergence problems present in ci calculations of atomic and molecular structure are also present in a much severe manner in ci calculations of the positron - atom problem .
the ci method has recently been applied to the study of positronic atoms ( electronically stable states consisting of a positron bound to an atom ) @xcite and also to positron - atom scattering states @xcite .
the attractive electron - positron interaction leads to the formation of a ps cluster ( i.e. something akin to a positronium atom ) in the outer valence region of the atom @xcite .
the accurate representation of a ps cluster using only single particle orbitals centered on the nucleus requires the inclusion of orbitals with much higher angular momenta than a roughly equivalent electron - only calculation @xcite . in the most extreme case
so far considered , namely @xmath11li , a calculation with @xmath12 was required before the energy had decreased enough to establish binding .
given that helium is described as slowly convergent @xcite , one struggles to find an adjective that could characterize the convergence properties of positronic systems !
the two most important expectation values for positronic systems are the energy , and the rate for electron - positron annihilation .
the annihilation rate , which is proportional to the expectation of the electron - positron @xmath1-function , has the inconvenient property that it is even more slowly convergent than the energy with respect to orbital angular momentum .
one has successive increments decreasing at high @xmath2 according to @xcite : @xmath13 to put this in perspective , it would take a calculation with @xmath14 to recover 99@xmath6 of the psh annihilation rate @xcite .
in addition to the slow convergence with @xmath2 , the @xmath1-function operator also exhibits very slow convergence with respect to the radial basis @xcite . in the present work ,
large basis ci calculations of the he ground state are performed in order to more exactly understand the convergence of the ci expansion .
since the properties of the he ground state are known to high precision it is a very useful laboratory system with which to test methods of extrapolating the radial and partial wave expansions to completion .
the insights obtained from helium should then be applicable to positronic systems and also possibly give additional guidance about how to approach purely electronic systems . besides looking at the energy , the convergence of the ci expansion of the electron - electron @xmath1-function expectation value
is also studied due to its relation with the electron - positron annihilation operator ( which is also a @xmath1-function ) .
it should be noted that the @xmath1-function operator also appears in the breit - pauli relativistic correction as the two - body darwin interaction @xcite .
the present work builds on an earlier investigation that studied the convergence of the radial basis in a simplified model of the helium atom which only included @xmath15 orbitals @xcite .
the ci wave function in a single - center basis is a linear combination of anti - symmetrised two - electron states with the usual clebsch - gordan coupling coefficients , @xmath16 the functions @xmath17 are single electron orbitals written as a product of a radial function and a spherical harmonic : @xmath18 all observable quantities can be defined symbolically as @xmath19 where @xmath20 is the increment to the observable that occurs when the maximum orbital angular momentum is increased from @xmath21 to @xmath2 , e.g. @xmath22 hence , one can write formally @xmath23 the first term on the right hand side will be determined by explicit computation while the second term must be estimated .
the problem confronting all single center calculations is that part of @xmath24 arises from terms with @xmath0-values that are not included in the largest explicit calculation .
the two expectation values that were investigated were that of the energy @xmath25 and the electron - electron @xmath1-function , @xmath26 . for helium , terms with @xmath27
contribute only 0.033@xmath6 of the total energy . for purely electronic systems
these higher @xmath0 terms make a small ( but slowly convergent ) correction to the total energy and other expectation values .
the extrapolation schemes used later in this paper have their basis in the work of schwartz @xcite , hill @xcite and kutzelnigg and associates @xcite .
analytic work indicates that the energy increments are given by @xmath28 where @xmath29 given a two - body wave function @xmath30 .
no expressions for @xmath31 have been presented . at large @xmath2
, one expects the energy increments to be well described by eq .
( [ pe1 ] ) .
for the @xmath1-function one can write @xmath32 where @xmath33 is believed @xcite to be @xmath34 ( ottschofski and kutzelnigg give a formula similar to this for the leading relativistic contribution to the energy of two - electron atoms .
we have assumed the slow @xmath35 convergence is due to the two - electron darwin term ) .
it should be noted that gribakin and ludlow @xcite have also derived an expression equivalent to eq .
( [ ad ] ) in the context of positron annihilation .
the numerical value was taken from a variational wave function of the he ground state with a basis of 250 explicitly correlated gaussians and an energy of -2.9037243752 hartree .
lcccccccc & & + @xmath2 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath36 & @xmath37 & @xmath38 & @xmath39 + 0 & 4.8 & 20 & -2.879 028 507 & 0.155 789 346 & 8.6 & 44 & -2.879 028 760 & 0.155 766 769 + 1 & 7.8 & 40 & -2.900 515 873 & 0.128 501 540 & 11.6 & 80 & -2.900 516 228 & 0.128 460 082 + 2 & 10.1 & 60 & -2.902 766 378 & 0.120 923 186 & 14.4 & 115 & -2.902 766 823 & 0.120 862 126 + 3 & 12.1 & 80 & -2.903 320 527 & 0.117 264 315 & 17.2 & 150 & -2.903 321 045 & 0.117 183 496 + 4 & 14.0 & 100 & -2.903 517 973 & 0.115 104 494 & 19.2 & 185 & -2.903 518 552 & 0.115 004 651 + 5 & 15.5 & 120 & -2.903 605 022 & 0.113 681 991 & 21.2 & 220 & -2.903 605 654 & 0.113 563 078 + 6 & 17.1 & 140 & -2.903 649 142 & 0.112 676 622 & 22.8 & 255 & -2.903 649 820 & 0.112 539 353 + 7 & 18.7 & 160 & -2.903 673 821 & 0.111 930 245 & 24.8 & 290 & -2.903 674 539 & 0.111 775 243 + 8 & 20.1 & 180 & -2.903 688 677 & 0.111 355 981 & 26.5 & 325 & -2.903 689 430 & 0.111 183 690 + 9 & 21.5 & 200 & -2.903 698 142 & 0.110 901 652 & 28.0 & 360 & -2.903 698 926 & 0.110 712 681 + 10 & 22.9 & 220 & -2.903 704 451 & 0.110 534 160 & 29.5 & 395 & -2.903 705 263 & 0.110 329 155 + 11 & 24.2 & 240 & -2.903 708 815 & 0.110 231 642 & 31.0 & 430 & -2.903 709 652 & 0.110 011 117 + 12 & 25.5 & 260 & -2.903 711 927 & 0.109 978 870 & 32.5 & 465 & -2.903 712 786 & 0.109 743 369 + & -2.903 724 377 034 & 0.106 345 371 & & & & + + & -2.903 723 421 & 0.106 943 & & & -2.903 724 362 & 0.106 527 + & -2.903 723 252 & 0.107 178 & & & -2.903 724 249 & 0.106 630 + & -2.903 723 205 & 0.107 334 & & & -2.903 724 240 & 0.106 698 + as can be imagined there have been a number of very large ci calculations upon the helium ground state that have addressed the convergence issue @xcite .
these calculations can be roughly divided into two classes , those that represented the radial wave function on a grid or used piecewise polynomials @xcite , and those that describe the radial wave function as a linear combination of analytic basis functions @xcite .
the first systematic calculation was the seminal investigation by carroll , silverstone and metzger ( csm ) @xcite who used a piece - wise polynomial basis to construct a natural orbital expansion . besides performing some very large calculations they also estimated the completeness limit of their radial basis .
the largest explicit calculation by csm will be termed the csm calculation while the extrapolated calculation will be denoted as csm@xmath40 . despite their importance
, these calculations have been largely superseded by the grid - based calculation of salomonson and oster ( so ) @xcite and the b - spline calculation of decleva , lisini and venuti ( dlv ) @xcite .
the so calculation obtained energies , @xmath38 accurate to about 10@xmath8 hartree by extrapolating the radial basis to the variational limit .
this extreme level of accuracy has not been achievable with the three calculations that used slater type orbitals ( sto ) to represent the radial wave function @xcite .
linear dependence problems become severe as the basis set is expanded toward completeness . indeed
, recourse was made to very high precision ( real*24 ) arithmetic in the sims and hagstrom ( sh ) calculation @xcite which is the largest calculation of this type so far reported .
lcccccc & & & & + @xmath2 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath38 & @xmath39 + 0 & 4.8 & 35 & -2.879 028 716 & 0.155 774 273 & -2.879 028 766 & 0.155 763 804 + 1 & 7.8 & 70 & -2.900 516 172 & 0.128 472 171 & -2.900 516 246 & 0.128 451 020 + 2 & 10.1 & 105 & -2.902 766 757 & 0.120 878 722 & -2.902 766 852 & 0.120 845 876 + 3 & 12.1 & 140 & -2.903 320 971 & 0.117 204 759 & -2.903 321 084 & 0.117 159 843 + 4 & 14.0 & 175 & -2.903 518 472 & 0.115 030 202 & -2.903 518 601 & 0.114 973 165 + 5 & 15.5 & 210 & -2.903 605 568 & 0.113 593 010 & -2.903 605 710 & 0.113 523 543 + 6 & 17.1 & 245 & -2.903 649 729 & 0.112 573 360 & -2.903 649 884 & 0.112 491 488 + 7 & 18.7 & 280 & -2.903 674 443 & 0.111 813 216 & -2.903 674 609 & 0.111 719 074 + 8 & 20.1 & 315 & -2.903 689 330 & 0.111 225 558 & -2.903 689 506 & 0.111 119 165 + 9 & 21.5 & 350 & -2.903 698 823 & 0.110 758 273 & -2.903 699 007 & 0.110 639 719 + 10 & 22.9 & 385 & -2.903 705 157 & 0.110 378 300 & -2.903 705 349 & 0.110 247 727 + 11 & 24.2 & 420 & -2.903 709 543 & 0.110 063 722 & -2.903 709 741 & 0.109 921 236 + 12 & 25.5 & 455 & -2.903 712 675 & 0.109 799 333 & -2.903 712 882 & 0.109 645 079 + + & -2.903 724 243 & 0.106 881 & -2.903 724 476 & 0.106 328 + & -2.903 724 123 & 0.106 757 & -2.903 724 378 & 0.106 341 + & -2.903 724 109 & 0.106 847 & -2.903 724 384 & 0.106 334 +
the present calculations use a basis set consisting of laguerre type orbitals ( ltos ) @xcite .
the ltos of a given @xmath0 are chosen to have a common exponential parameter which means they are automatically orthogonal .
hence , the basis can be expanded toward completeness without causing any linear dependence problems .
the ci basis can be characterized by the index @xmath2 , the maximum orbital angular momentum of any single electron orbital included in the expansion of the wave function .
it should be noted that all matrix elements were evaluated using gaussian quadrature even though the basis functions have an analytical form @xcite .
lcccccc @xmath2 & 35lto & 35lto@xmath41 & sh @xcite & csm@xmath40 @xcite & so@xmath40 @xcite & dlv @xcite + @xmath42 & -2.879 028 760 & -2.879 028 766 & -2.879 028 757 & -2.879 028 765 & -2.879 028 77 & -2.879 028 767 + @xmath43 & -2.900 516 228 & -2.900 516 246 & -2.900 516 220 & -2.900 516 220 & -2.900 516 25 & -2.900 516 245 + @xmath44 & -2.902 766 823 & -2.902 766 853 & -2.902 766 805 & -2.902 766 822 & -2.902 766 85 & -2.902 766 849 + @xmath45 & -2.903 321 045 & -2.903 321 084 & -2.903 321 016 & -2.903 321 079 & -2.903 321 09 & -2.903 321 079 + @xmath46 & -2.903 518 552 & -2.903 518 601 & -2.903 518 465 & -2.903 518 598 & -2.903 518 60 & -2.903 518 600 + @xmath47 & -2.903 605 654 & -2.903 605 710 & -2.903 605 515 & -2.903 605 71 & -2.903 605 72 & -2.903 605 97 + @xmath48 & -2.903 649 820 & -2.903 649 884 & -2.903 649 644 & -2.903 649 88 & -2.903 649 89 & -2.903 650 24 + @xmath49 & -2.903 674 539 & -2.903 674 609 & -2.903 674 328 & -2.903 674 59 & -2.903 674 62 & -2.903 675 01 + @xmath50 & -2.903 689 430 & -2.903 689 505 & -2.903 689 193 & -2.903 689 47 & -2.903 689 52 & -2.903 689 93 + @xmath51 & -2.903 698 926 & -2.903 699 006 & -2.903 698 656 & -2.903 698 95 & -2.903 699 02 & -2.903 699 44 + @xmath52 & -2.903 705 263 & -2.903 705 349 & -2.903 704 974 & -2.903 705 27 & -2.903 705 37 & -2.903 705 79 + @xmath53 & -2.903 709 652 & -2.903 709 742 & -2.903 709 325 & -2.903 709 64 & & -2.903 710 19 + @xmath54 & -2.903 712 786 & -2.903 712 882 & -2.903 712 433 & & & -2.903 713 33 + + @xmath55 & -0.021 487 468 & -0.021 487 480 & -0.021 487 463 & -0.021 487 455 & -0.021 487 48 & -0.021 487 478 + @xmath56 & -0.002 250 594 & -0.002 250 606 & -0.002 250 585 & -0.002 250 662 & -0.002 250 61 & -0.002 250 604 + @xmath57 & -0.000 554 223 & -0.000 554 232 & -0.000 554 211 & -0.000 554 197 & -0.000 554 23 & -0.000 554 230 + @xmath58 & -0.000 197 507 & -0.000 197 516 & -0.000 197 449 & -0.000 197 519 & -0.000 197 52 & -0.000 197 521 + @xmath59 & -0.000 087 102 & -0.000 087 109 & -0.000 087 050 & -0.000 087 112 & -0.000 087 11 & -0.000 087 37 + @xmath60 & -0.000 044 166 & -0.000 044 174 & -0.000 044 129 & -0.000 044 17 & -0.000 044 18 & -0.000 044 27 + @xmath61 & -0.000 024 719 & -0.000 024 725 & -0.000 024 683 & -0.000 024 71 & -0.000 024 73 & -0.000 024 77 + @xmath62 & -0.000 014 891 & -0.000 014 896 & -0.000 014 866 & -0.000 014 88 & -0.000 014 90 & -0.000 014 92 + @xmath63 & -0.000 009 496 & -0.000 009 501 & -0.000 009 463 & -0.000 009 48 & -0.000 009 50 & -0.000 009 51 + @xmath64 & -0.000 006 337 & -0.000 006 342 & -0.000 006 318 & -0.000 006 32 & -0.000 006 35 & -0.000 006 35 + @xmath65 & -0.000 004 389 & -0.000 004 394 & -0.000 004 351 & -0.000 004 37 & & -0.000 004 40 + @xmath66 & -0.000 003 134 & -0.000 003 139 & -0.000 003 108 & & & -0.000 003 14 + three sets of calculations have been performed for the he ground state . in the first set
, there were 20 ltos per @xmath0 with the largest calculation including orbitals up to @xmath67 .
the lto exponents for a given @xmath0 were the same and the values of the exponents were optimized in a quasi - perturbative fashion .
the exponents for @xmath68 and 2 orbitals were optimized in a ci calculation with all 60 orbitals .
the exponents for @xmath27 were optimized separately for each @xmath0 with ci calculations that also included the @xmath69 orbitals .
once the exponents were optimized , a sequence of calculations to give the @xmath38 and @xmath39 for successive @xmath2 were carried out .
the basis is denoted the 20lto basis and the results of the calculations with this basis are reported in table [ hetab1 ] .
the second set of calculations were much larger .
here there were 35 ltos per @xmath0 with the exception of @xmath70 and 1 where respectively 44 and 36 ltos were used respectively .
the orbital exponents were optimized for each @xmath0 in a manner similar to that described above and the calculations were taken to @xmath71 .
a total of 465 single electron orbitals were included in the largest calculation , which required the diagonalization of a hamiltonian matrix of dimension 8586 .
this calculation was an example of a very large explicit calculation .
the basis is denoted the 35lto basis and the results of the calculations with this basis are reported in table [ hetab1 ] .
the idea behind the third calculation was to exploit extrapolation techniques to estimate the variational limit for each partial wave . a sequence of calculations with 32 , 33 , 34 and 35 ltos per @xmath0 was done for a basis that was defined with the same exponential parameters as the 20lto calculation .
the number of basis functions were varied so that all partial waves had the same basis dimension . optimizing the lto basis for the largest radial basis has been shown to result in distortions in the convergence pattern with respect to the number of radial basis functions .
this can be avoided if the basis is optimized in a basis that has at least 10 fewer ltos per @xmath0 than the active calculation @xcite .
the variational limit for the radial basis can be estimated by fitting the increments to @xmath7 and @xmath5 to the inverse series @xcite @xmath72 it is possible to estimate the @xmath73 limits for the radial basis once the @xmath74 , @xmath75 , @xmath76 , @xmath77 coefficients have been determined .
a two - term series was used for both eqs .
( [ eseries2 ] ) and ( [ dseries2 ] ) .
it would have been preferable to use three - term series but the impact of round - off error rendered this impractical .
the basis for this set of calculations is denoted the 35lto@xmath78 basis , while the basis including the @xmath73 correction is termed the 35lto@xmath41 basis .
the energies and expectation values for these two calculations are listed in table [ hetab3 ] .
the validity of these results can be tested by examination of the energy increments of large ci calculations of helium . besides the present calculations , data from a number of previous ci calculations have been used . as a function of @xmath2 for the different ci calculations of the he ground state energy as listed in table [ hetab2 ] .
, width=321 ] table [ hetab2 ] gives the energies of the present 35lto and 35lto@xmath41 basis sets , along with the sh , csm@xmath40 , so@xmath40 and dlv calculations .
these same sets of data are also presented as energy differences between consecutive calculations @xmath79 .
the energies of the sh calculation , which used the even - tempered sto basis , are consistently the worst , and are @xmath80 hartree larger than the 35lto calculation at @xmath71 . even though csm@xmath40 does attempt to achieve the variational limit for each @xmath2 , in reality it is only about as good as the 35lto calculation .
indeed for @xmath81 , the csm@xmath40 values of @xmath79 were smaller than those of the 35lto calculation .
the present 35lto@xmath41 , so@xmath40 and dlv calculations are in agreement to 10@xmath8 hartree ( or better ) for @xmath82 .
this is expected , since all three calculations are large and use extrapolation techniques to achieve the variational limit .
the energy difference between the 35lto and 35lto@xmath41 energies gives an indication of the incompleteness of the 35lto basis and by @xmath71 the difference is @xmath83 hartree .
the good agreement between the 35lto@xmath41 , so@xmath40 and dlv energies is not present for @xmath84 .
although the 35lto@xmath41 and so@xmath40 energies generally agree at the level of 10@xmath8 hartree , it is seen the dlv @xmath79 increments are larger than these two other calculations .
for example , dlv give @xmath85 hartree which is about @xmath86 hartree larger than the 35lto@xmath41 and so@xmath40 increments .
it has also been noticed that dlv do overstate the accuracy of their calculation , they assert an accuracy of @xmath87 hartree
. however , this accuracy is based on a calculation which gives @xmath88 hartree ( 2nd column of table iv of @xcite ) , and this energy is in error by @xmath89 hartree ! a useful way to scrutinize the partial wave series is to assume a power law decay of the form @xmath90 and determine the value of @xmath91 for a succession of three @xmath92 values using @xmath93 the exponent derived from the energy increments is @xmath94 while the exponent derived from the @xmath1-function increments is @xmath95 .
one expects @xmath96 @xcite and @xmath97 as @xmath9 @xcite , in agreement with eqs .
( [ eseries ] ) and ( [ dseries ] ) .
the values of @xmath94 for the he energies presented in table [ hetab2 ] are plotted in figure [ pehe ] as a function of @xmath2 .
one of the noticeable features of figure [ pehe ] are the irregularities in some of the calculations , e.g. the so@xmath40 , csm@xmath40 and 35lto@xmath41 calculations .
the fluctuations in the present 35lto@xmath41 curve are due to the impact of round - off error on the radial extrapolations . the determination of the coefficients in eq .
( [ eseries2 ] ) involves the subtraction of the energies for calculations that differ by a single lto .
the resulting energy differences are very small and therefore are susceptible to the essentially random errors resulting from round - off that gradually accumulate during the course of the computations .
the irregularities in the csm@xmath40 and so@xmath40 curves are a consequence of the number of digits at which the energies were published @xcite .
plots of @xmath94 vs @xmath2 were examined ( but not plotted in figure [ pehe ] ) for some calculations @xcite that used an sto basis set .
these plots of @xmath94 showed much larger fluctuations than any of the calculations depicted in figure [ pehe ] .
as a function of @xmath2 for the lto calculations of the he ground state @xmath5 . , width=321 ] the smaller 20lto and csm@xmath40 calculations had plots of @xmath94 vs @xmath2 that tended to level out at @xmath98 .
indeed , the tendency for 20lto trajectory to curve up indicates that the successive @xmath79 increments are decreasing too quickly at the higher @xmath2 values .
the larger so@xmath40 , dlv , 35lto and 35lto@xmath41 calculations have @xmath94 versus @xmath2 trajectories that steadily decrease with increasing @xmath2 and appear to be approaching the expected limit of @xmath99 although this is obscured somewhat for the so@xmath40 and 35lto@xmath40 curves .
it will be demonstrated later that the behavior of the 20lto and csm@xmath40 curves is due to slower convergence of the radial basis at high @xmath0 . the tendency for @xmath94 to approach the limiting value of 4 from above is a consequence of the fact that the @xmath100 and @xmath101 coefficients of eq .
( [ eseries ] ) have the same sign . the coefficients @xmath100 and @xmath101 are derived from second and third order perturbation theory respectively @xcite and have the same sign due to repulsive nature of the electron - electron interaction .
one surmises that a mixed electron - positron system , with its attractive electron - positron interaction should have @xmath96 from below , and this is indeed the case @xcite .
the incremental exponent for the @xmath1-function , @xmath95 , is shown in figure [ pdhe ] for the 20lto , 35lto and 35lto@xmath41 basis sets .
it should be noted that the values of @xmath95 were sensitive to the precision of the calculation .
originally , the diagonalization of the hamiltonian was performed using the davidson algorithm @xcite . however , this method could not give @xmath1-function expectation values to better than 8 significant figures ( irrespective of the convergence tolerance for the energy ) .
this lead to noticeable fluctuations in the @xmath95 versus @xmath2 plot .
the diagonalization was subsequently performed using the eispack libraries , reducing the size of the fluctuations .
the trajectories of the 20lto and 35lto calculations do not appear to be approaching the @xmath97 limit as @xmath9 .
the 20lto curve has a @xmath95 trajectory that diverges from 2 for @xmath102 while the 35lto curve diverges from 2 for @xmath81 .
the 20lto basis gives @xmath103 at @xmath71 while the 35lto basis gives @xmath104 at this @xmath2 value .
however , the plot of @xmath95 based on the 35lto@xmath41 sequence does exhibit the correct qualitative behavior as @xmath2 increases .
the @xmath73 corrections have a larger impact on @xmath39 than on @xmath38 since the former converges as @xmath105 while the latter converges as @xmath106 .
the @xmath71 value of @xmath95 was 2.008 .
: 20lto ratios of the increments to @xmath38 and @xmath39 ( refer to eq .
( [ ratio ] ) ) as a function of @xmath2 for the calculations of the he ground state .
, width=321 ] the behavior exhibited in figures [ pehe ] and [ pdhe ] can be attributed to the convergence of the radial basis .
a larger radial basis is required to predict successive @xmath107 increments as @xmath2 increases .
the ratios @xmath108 gives a measure of the relative impact of the 20lto , 35lto and 35lto@xmath41 basis sets to a @xmath2 energy increment .
one can define a similar ratio , @xmath109 , for the @xmath1-function @xmath110 increments .
the behavior of these ratios versus @xmath2 are depicted in figure [ rde ] . both @xmath111 and @xmath109 increase steadily with @xmath2 .
the slower convergence of the energy at higher @xmath2 is also evident in table i of @xcite and explicit comment about this point has been made previously @xcite .
the @xmath39 expectation value is much more sensitive to the increase in the dimension of the radial basis and there was a 9.3@xmath6 increase in @xmath112 between the 20lto and 35lto@xmath41 calculations .
the corresponding increase in the @xmath66 was only @xmath113 .
this extra sensitivity of @xmath39 is something we have noticed in calculations of positron annihilation rates in positron - atom systems @xcite even though explicit mention of this point has not been made .
it is apparent from figures [ pehe ] and [ rde ] that including the radial extrapolations has resulted in irregularities appearing in the 35lto@xmath41 energy sequence .
these irregularities are of order 10@xmath114 hartree at @xmath115 and should be removed before the @xmath9 corrections are determined .
examination of figure [ rde ] suggested that @xmath116 . accordingly a fit of @xmath117 to a @xmath118 functional form was performed over the @xmath119 $ ] interval .
an adjustment to the @xmath38 energy sequence was made once @xmath120 , @xmath121 and @xmath122 were fixed .
the values of @xmath122 was approximately @xmath123 .
the 35lto@xmath41 values of @xmath38 and @xmath79 given in table [ hetab2 ] are those of the smoothed energy sequence .
the largest change to any of the energies was @xmath124 hartree .
limit for the he ground state energy @xmath25 using three different methods to complete the partial wave series .
the input @xmath38 upon which the extrapolations were based were those of the 35lto and the smoothed 35lto@xmath41 calculations .
the horizontal line shows the exact helium energy @xcite . , width=321 ] one of the major aims of this paper was to determine whether it is possible to extract the @xmath9 limit from a finite sequence of calculations . to this end , fits of inverse power series of different lengths are made to sequences of @xmath92 data , and then those inverse power series are summed to infinity .
equations ( [ eseries ] ) and ( [ dseries ] ) are the working equations .
fits are performed retaining just the leading order term ( method 1 ) , the first two terms ( method 2 ) , and the first three terms ( method 3 ) of these series .
the fits of these equations use the minimum information necessary .
so , method 1 , which only retains the first @xmath100 term of eq .
( [ eseries ] ) , only requires two successive values of @xmath38 to determine @xmath100 .
three successive values of @xmath92 are used to determine @xmath125 and @xmath126 when the two leading terms of eq .
( [ eseries ] ) or eq .
( [ dseries ] ) are used .
four successive values of @xmath92 are used to determine @xmath125 , @xmath126 and @xmath127 when the three leading terms of eq .
( [ eseries ] ) or eq .
( [ dseries ] ) are used .
the fits to determine @xmath125 and/or @xmath126 and/or @xmath127 can be done to different sequences of @xmath2 values as a self - consistency to check that the two - term fits to the @xmath128 or @xmath129 sets of @xmath130 give answers that are numerically close .
once the coefficients of the inverse power series have been determined , the @xmath9 contribution is determined by a two - step procedure .
firstly , the series ( [ eseries ] ) and ( [ dseries ] ) are summed explicitly up to @xmath131 . the remainder from @xmath132
is determined using the approximate result : @xmath133 eq .
( [ bettertail ] ) can be regarded as an approximation to the @xmath134 integral using the mid - point rule .
this approximation is accurate to 0.1@xmath6 for @xmath135 and @xmath136 .
figures [ heeinf ] and [ hedinf ] show the behavior of the extrapolated @xmath137 and @xmath1-function for the three different extrapolations as a function of @xmath2 .
tables [ hetab1 ] and [ hetab3 ] gives estimates of @xmath138 and @xmath139 using the calculated values at the largest possible @xmath2 values to determine the @xmath9 corrections .
limit for the he ground state @xmath140 using three different methods to complete the partial wave series .
the horizontal line shows the value of drake @xcite .
, width=321 ] figure [ heeinf ] shows that the quality of the 35lto energy extrapolation using method 1 is inferior to methods 2 and 3 which give @xmath138 energies in agreement which each other at the 10@xmath114 hartree level for @xmath141 .
however , using the 35lto energies in conjunction with methods 2 and 3 gives @xmath138 values that are too large by about @xmath4 hartree .
this is a consequence of using a large but not quite complete radial basis .
the use of the 35lto@xmath41 energies results in an energy limit that is an order of magnitude more precise than those of the 35lto basis . using method 2 for the @xmath142 35lto@xmath143 energies gave @xmath138 = -2.903 724 378 hartree , an energy that is in error by 10@xmath114 hartree .
the @xmath9 corrections were only made using method 2 since the more sophisticated method 3 is more sensitive to the imperfections of the smoothed data sets .
the smoothed energy sequence is probably not a perfect reproduction of the actual sequence and there is a tendency for the @xmath138 limit to be more negative than the exact energy .
the method 3 estimate of @xmath138 , at @xmath71 , namely -2.903 724 384 hartree , is about 10@xmath8 more negative than the exact energy .
a similar level of accuracy was achieved in the earlier so@xmath40 calculation , their estimate of the energy in the @xmath144 limit was -2.903 724 39 hartree @xcite .
the difficulties in obtaining sub 0.1@xmath6 accuracy in @xmath139 for the 35lto sequence are readily apparent from figure [ hedinf ] .
as one increases @xmath2 , the estimates of @xmath139 also increase and the discrepancy with the accurate value of drake @xcite gets larger .
the ultimate accuracy achievable for the 35lto basis is between 0.1 and 0.5@xmath6 .
the apparent superiority of method 1 for the 35lto basis arises because errors resulting from a finite dimension radial basis act to partially cancel errors that arise from this least sophisticated @xmath9 extrapolation . however , usage of the 35lto@xmath143 sequence permitted a much more accurate extrapolation to the @xmath9 limit .
the 35lto@xmath145 basis gives estimates of @xmath139 that are two orders of magnitude more precise .
the method 2 extrapolation was only @xmath146 @xmath147 larger than the exact value @xcite .
while the radial extrapolations did introduce fluctuations into the @xmath39 values , the relative size of the individual @xmath148 increments were much larger than the @xmath79 increments and thus did not lead to fluctuations in @xmath139 as long as method 2 was used for the angular extrapolations .
however , the use of method 3 did result in fluctuations of order 10@xmath149 @xmath147 in @xmath139 so is not depicted in figure [ hedinf ] . as extracted from sequences of @xmath38 data .
the horizontal line shows the value of from eq .
( [ ae ] ) , namely -0.074226 .
estimates of @xmath100 from method 2 are drawn with dashed lines while estimates of @xmath100 from method 3 are drawn with solid lines .
the dlv data analyzed here was solely from their table iv of ref.@xcite , since this avoided the discontinuity at @xmath150 and gave a smooth curve .
, width=321 ] the coefficients of the asymptotic forms , ( [ eseries ] ) and ( [ dseries ] ) are known _ a - priori _ from eqs .
( [ ae ] ) , ( [ be ] ) and ( [ ad ] ) .
estimates of these parameters are also obtained during the fit of the inverse power series to a set of @xmath38 or @xmath39 .
an ideal consistency check would be estimates of @xmath100 and @xmath33 that steadily approached -0.074226 and -0.04287 as @xmath2 increased and as the number of terms included in eqs .
( [ eseries ] ) and ( [ dseries ] ) increased .
unfortunately , this has not yet been achieved .
the least squares analysis of the csm@xmath40 energies gave @xmath151 and @xmath152 @xcite .
however , this value of @xmath100 is only achieved when using @xmath38 for @xmath153 $ ] .
the very large calculations of dlv reported @xmath154 and @xmath155 @xcite .
however , a cursory examination of figure [ aecoef ] which depicts values of @xmath100 obtained from three successive @xmath38 energies demonstrates that their value of @xmath100 is not converging to -0.074226 with increasing @xmath2 . applying the more sophisticated 3-term inverse power series to the dlv energies leads to an @xmath100 that exhibits a 4@xmath6 variation between @xmath156 and @xmath157 .
figure [ aecoef ] also shows the variation in @xmath100 when fitting the 35lto and 35lto@xmath41 energy sequences to eq .
( [ eseries ] ) .
fits were performed with both methods 2 and 3 for the 35lto energy sets , and only with method 2 ( for reasons discussed earlier ) to the 35lto@xmath41 sequence .
the @xmath100 coefficients for a given method are computed using the minimum range of @xmath2 values that permitted the unique determination of the coefficients .
application of method 2 to the 35lto data reveals that @xmath100 achieves a minimum value of @xmath158 -0.07413 at @xmath136 before increasing at larger @xmath2 .
application of method 3 results in values of @xmath100 that are clearly not approaching the correct value .
this should be expected since it has been demonstrated that the 35lto @xmath79 are increasingly underestimated as @xmath2 increases .
it would therefore be hoped that values of @xmath159 extracted from the 35lto@xmath41 would show better convergence to the expected limit as @xmath2 increases .
this expectation has only been partly realized , there are indications that @xmath100 may be converging to the correct value , but the application of smoothing has probably introduced a systematic bias that resulted in a tendency to overestimate the magnitude of @xmath100 . figure [ adcoef ] shows the values @xmath33 as obtained from the 35lto basis using methods 1 , 2 and 3 as a function of @xmath2 .
none of the calculations using the 35lto basis resulted in an @xmath33 vs @xmath2 curve that approached the correct value as @xmath2 increased .
this is another manifestation of the very slow convergence of @xmath39 with respect to the dimension of the radial basis set .
there was a significant improvement when @xmath33 was extracted from the 35lto@xmath41 sequence using method 2 . in this case
, @xmath33 does appear to be converging to the expected value of -0.04287 and at @xmath71 one obtains @xmath160 the small irregularities in the 35lto@xmath41 @xmath39 sequence resulted in irregularities in @xmath33 when using the more sophisticated and sensitive method 3 fit ; so this was not depicted in figure [ adcoef ] .
it is also worth noting that @xmath33 was also subject to irregularities of @xmath161 when the davidson method was used to diagonalize the hamiltonian and generate the ground state wave function . as extracted from sequences of @xmath39 data .
the horizontal line shows the value of eq .
( [ ad ] ) , namely @xmath162 . , width=321 ]
results of a set of very large ci calculations of the he ground state have been presented .
the largest explicit ci calculation reported here with a minimum of 35 lto s per @xmath0 gave an energy that was accurate to @xmath163 hartree . including energy corrections obtained from the 2- and 3-term inverse power series in @xmath2 resulted in a he ground state energy that was accurate to @xmath164 hartree . improved accuracy
required the use of extrapolations in the radial basis set to get an estimate of the variational limit for @xmath38 .
this permitted the he ground state energy to be predicted to better than 10@xmath8 hartree , an improvement of a factor of 1000 over the largest explicit calculation .
the main impediments to more refined predictions of the he ground state are those due to round - off errors . estimating the coefficients of
the inverse power series involves manipulating very small energy differences which will be sensitive to round - off errors .
the fluctuations in the radial extrapolation were about @xmath124 hartree at @xmath71 . while this in itself is not that bad , these fluctuations are magnified by an order of magnitude when the angular momentum extrapolation is then done .
the impact of the fluctuations was somewhat mitigated by the introduction of a smoothing procedure , at the cost of introducing a small systematic error .
the prediction of the electron - electron @xmath1-function was considerably more difficult due to the @xmath165 convergence . in this case , the explicit calculation was accurate to 3@xmath6 at @xmath71 .
application of the inverse power series ( method 2 ) to include higher @xmath2 contributions improved the accuracy to 0.3@xmath6 .
the main reason for the low accuracy was the slow convergence with respect to the number of radial basis functions .
the relative accuracy of successive @xmath148 increments decreases as @xmath2 increases if the number of radial basis functions per @xmath0 is kept the same . once again , extrapolating the radial basis to the variational limit lead to an improved prediction of @xmath5 .
the best ci estimate of @xmath166 @xmath147 was within 0.01@xmath6 of the close to exact variational estimate @xcite .
the extrapolations of @xmath5 were less susceptible to round - off error simply because the @xmath148 increments were larger .
while the use of extrapolations did improve the quality of the calculation , the full potential of the method has not been realized due to round - off error .
the radial matrix elements are evaluated with gaussian quadratures and the achievable precision for the larger calculations is about 10@xmath167 hartree .
this accuracy could be improved by the either the development of a convenient analytic form for the electron - electron matrix elements or the usage of quadruple precision arithmetic .
this would then permit the use of inverse power series with more terms leading to improved radial and angular extrapolations .
for example , an accuracy of 10@xmath168 hartree was achievable for a ci calculation restricted to @xmath70 orbitals @xcite .
these results have implications for the prediction of the annihilation rate of positronic atoms from single - center ci type calculations @xcite .
some sort of extrapolation in @xmath2 is needed to determine the energy and more particularly the annihilation rate .
one way to minimize the impact of the extrapolation in @xmath2 is to run the calculation to the highest possible angular momentum .
however , the high @xmath2 parts of the annihilation rate will tend to be increasingly underestimated as @xmath2 increases unless accurate estimates of the radial variational limit can be made .
since this can now be achieved for a laguerre basis @xcite , it eminently conceivable that estimates of the annihilation rate at better than @xmath169 accuracies will be achievable for single - center basis sets .
this work was supported by a research grant from the australian research council .
the authors would like to thank shane caple for providing access to extra computing resources .
one author ( mb ) would like to thank prof .
cliff surko for his , and his group s , summer hospitality at ucsd while this paper was in preparation . | the configuration interaction ( ci ) method using a very large laguerre orbital basis is applied to the calculation of the he ground state .
the largest calculations included a minimum of 35 radial orbitals for each @xmath0 ranging from 0 to 12 resulting in basis sets in excess of 400 orbitals .
the convergence of the energy and electron - electron @xmath1-function with respect to @xmath2 ( the maximum angular momenta of the orbitals included in the ci expansion ) were investigated in detail .
extrapolations to the limit of infinite in angular momentum using expansions of the type @xmath3 , gave an energy accurate to @xmath4 hartree and a value of @xmath5 accurate to about 0.5@xmath6 .
improved estimates of @xmath7 and @xmath5 , accurate to 10@xmath8 hartree and 0.01@xmath6 respectively , were obtained when extrapolations to an infinite radial basis were done prior to the determination of the @xmath9 limit .
round - off errors were the main impediment to achieving even higher precision since determination of the radial and angular limits required the manipulation of very small energy and @xmath5 differences . |
understanding quantum gravity ( qg ) in finite spacetime regions is the greatest conceptual problem facing physics today .
the observed positive value of the cosmological constant implies that it can not be sidestepped by resorting to experiments at infinity , since infinity is not accessible . in this paper
, we advocate an approach to the issue that involves restricting to _ null - bounded _ finite regions ,
i.e. to boundary data on lightlike hypersurfaces .
the paper is composed of two parts . in the first part ( section [ sec : why_null ] ) , we present an abstract argument in favor of null boundary data .
we list some conceptual difficulties that arise with respect to finite - region qg , and label them as either `` hard '' or `` secondary '' .
the `` hard '' issues have to do directly with the finite information bound @xcite implied by black hole thermodynamics .
this renders obsolete the notion of infinitely precise observables @xcite , a situation that no one knows how to deal with . in particular
, it s likely that quantum mechanics should be modified , since at present it deals with infinitely precise real numbers , such as probabilities and couplings .
in contrast , the `` secondary '' difficulties do not directly touch on the issue of finite precision , though they are sometimes conflated with it .
an example is the issue of invariance under transverse diffeomorphisms .
our goal in section [ sec : why_null ] is to show that the `` secondary '' problems can be removed by restricting the discussion to null boundary data . in this way
, we propose to disentangle them from the `` hard '' problem , making the latter easier to approach .
the bulk of our reasoning does nt favor a particular type of qg theory , apart from the use of null boundaries .
nevertheless , in section [ sec : why_null : discrete ] we make several arguments in favor of discrete geometries , in particular in the context of black hole creation and evaporation . in the second part of the paper ( sections [ sec : class]-[sec : gravity ] )
, we take baby steps in the proposed direction .
quantum gravity aside , the notion of restricting field theory to null boundary data has far - reaching consequences .
we set out to explore these on the safe ground of classical field theory .
in classical physics , the dynamics can be encoded in the on - shell action functional . in quantum theory , the ( effective ) on - shell action is upgraded into transition amplitudes , from which , again , all observables can be derived .
the off - shell action , on the other hand , is upgraded into contributions to the path integral
. these can be rearranged , and are therefore less physical than the overall amplitudes .
our strategy , then , is to treat the on - shell action as the basic `` observable '' in classical field theory , and to ask what features of the system can be derived from it .
we will find that certain basic properties of field theory are no longer manifest in the null setup .
the reason is that configuration variables at different points of a closed boundary are no longer independent .
the resulting breakdown of standard features of field theory is potentially exciting , for reasons we will now discuss . for a similar discussion , see section 8 of @xcite . with each conceptual step in physics ,
old certainties are exposed as either approximate or observer - dependent . in the process ,
the number of genuine , invariant features of the world is reduced . knowing which concepts are lost at each step
can be a valuable guide towards the new physics .
in particular , it pays to find a description of the _ old _ physics in which the doomed concepts are no longer obvious .
a good example is the minimal - action formulation of classical mechanics .
it makes determinism non - obvious , and in doing so paves the way towards quantum mechanics .
another example is the pre - symplectic formulation of hamiltonian mechanics , in which time loses its special role . at present
, we are in search of a conceptual framework for quantum gravity .
there is a list of features of known physics that are expected to fade away in a qg world .
first and foremost , space is expected to lose its infinite information - carrying capacity : black hole thermodynamics implies that the number of bits in a finite region is also finite , and proportional to the surface area .
one also expects a loss of _ locality_.
this appears to be a necessary result of the information bound @xcite , and is strongly implied by a holographic picture of the degrees of freedom @xcite . at the very least , the accuracy with which one can measure a would - be pointlike observable now depends on the overall surface area of the experimentally available region .
note that this notion of non - locality is stronger than a simple planck - scale resolution limit : it implies influences over an arbitrarily large distance . also , it is of course stronger than the classical notion of `` no pointlike observables '' in gravity .
the latter is due to the lack of a background geometry in which pointlike objects can be located .
that problem can be resolved by restricting to observables on the same hypersurface as the boundary data ( which is not necessarily at the boundary of spacetime ) .
these observables are then local , in the sense that gr is a local field theory ; we will elaborate on this in section [ sec : class : scalar_locality ] .
another notion that must apparently be abandoned in qg is that of conserved global charges @xcite .
one way of stating the argument is that a black hole of fixed size can have an arbitrary value of a would - be global charge .
this constitutes an unbounded amount of information , contradicting the holographic bound .
it is therefore desirable to find a formulation of known physics in which bulk degrees of freedom , locality and global charges - the features that are expected to disappear in qg - are no longer manifest .
this is precisely what happens when we examine classical field theory in terms of on - shell actions in null - bounded regions ( though our exploration is in no way a complete _ reformulation _ of classical field theory ) .
generically , we find that manifest locality is lost , and that global charges can no longer be defined .
also , in special cases such as free scalars and pure gravity , we find an apparent `` holographic '' reduction in the degrees of freedom . on a broader level
, these properties can all be traced to a failure of the usual machinery of classical mechanics .
this parallels the failure of quantum mechanics that is expected in finite - region qg .
we stress that all of the above takes place for classical field theories , where it is merely an artifact of a restrictive formalism . nonetheless , together with the general arguments from section [ sec :
why_null ] , these parallels build up a case for null - boundary observables as the correct approach to qg in finite regions . as a curious spin - off
, our investigation of the null - bounded gr action in section [ sec : gravity ] reveals an imaginary contribution ( eq . ) to the gibbons - hawking boundary term .
this contribution reproduces the bekenstein - hawking formula for black hole entropy @xcite , both in pure gr and in the presence of matter .
imaginary actions in gr are not a new notion , dating back to the analysis of black holes through wick rotation @xcite .
our derivation reveals this feature in lorentzian language , with no restriction to stationary metrics .
an imaginary action term is in fact present for generic boundaries in lorentzian spacetime .
however , its locality structure makes it irrelevant for action variations , in all but the null case .
the relevant notions of locality are discussed in section [ sec : class : scalar_locality ] .
section [ sec : discuss ] contains a discussion and outlook on the classical results , along with some comments on the content of the quantum theory .
in this section , we motivate the use of finite null - bounded regions in quantum gravity . this motivation was the starting point of the present work . through supergravity ,
string theory and ads / cft , various aspects of quantum gravity have become accessible . in particular , ads / cft @xcite provides a handle on non - perturbative dynamics .
however , this understanding remains limited to observables at infinity .
there is not even a clear definition of observables for a qg experiment in a local region of spacetime .
indeed , the entropy - bound arguments mentioned in the introduction suggest that a paradigm shift is necessary before such experiments can be properly discussed . due to the finite amount of information in a region with given surface area , it seems that the very notion of an observable ( or a quantum - mechanical probability ) as an infinitely precise real number must be discarded .
one may think that this problem is not very serious .
indeed , physics always deals with idealizations .
so perhaps in quantum gravity , an ideal , well - defined experiment can only be performed at infinity .
then realistic , finitely extended experiments are understood as approximations , as they already are in many other ways . however , this reasoning collapses in the face of a positive cosmological constant . just as @xmath0 provides a fundamental , finite limit on the accuracy of classical physics , so does @xmath1 provide a finite limit on the size of causally available spatial regions .
the entropy available to any observer is then limited by the de - sitter horizon area @xcite .
arbitrarily precise observables at infinity no longer provide a valid idealization of real experiments , and the hard issues concerning finite regions must be addressed .
discrete qg models , such as the spinfoam formulation @xcite of loop quantum gravity ( lqg ) @xcite , are at first sight well - suited to address dynamics in finite regions . however , the physical meaning of e.g. spinfoam amplitudes is far from trivial .
the relevant difficulties can be formulated in a model - independent way .
these difficulties are sometimes seen as ruling out any `` simple - minded '' attempt to discuss finite - region qg .
the first and foremost problem arises from the freedom of transverse diffeomorphisms , i.e. displacements of the boundary in perpendicular to itself . as a result of this redundancy
, different boundary data may correspond to different slices of the same history , i.e. to the same state .
in canonical gr , this redundant degree of freedom is generated by the hamiltonian constraint , which also encodes the dynamics .
the `` physical '' state space is obtained from the `` kinematical '' space of codimension-1 boundary data by imposing the hamiltonian constraint , i.e. by solving the theory . while this does nt invalidate spinfoam - like approaches
, it means that the theory s formulation is only distantly related to its physical content , i.e. to the structure of the true state space and observables on it .
this is the strong form of the statement that `` local observables are not diff - invariant '' .
the weak form , concerning _ pointlike _
observables , arises already in classical gr , and is addressed by restricting to observables on the boundary - data hypersurface .
if we consider boundary data on finite _ spacelike_-bounded regions , as is the case with spinfoams , another problem arises .
boundary data on a spacelike hypersurface implies a simultaneous measurement of fields within a codimension-1 chunk of space .
this contradicts the holographic principle , which should only allow a simultaneous measurement on a codimension-2 surface .
again , this argument does nt invalidate spinfoam - like models , because the codimension-1 boundary data is merely `` kinematical '' . upon solving the theory ,
the `` physical '' state space may turn out to be much smaller , in accord with the holographic bound .
however , this again means ( at best ) that the formalism is far removed from the reality it describes . one may try to avoid this conflict with holography by switching to timelike boundaries . in that case , measuring the boundary amounts to a _ sequence _ of simultaneous measurements , each one only capturing a codimension-2 slice .
however , timelike boundaries suffer from a different problem : they are not causally closed .
this problem is perhaps less severe than the previous ones , and arises already in classical field theory . on a timelike boundary , one can never fully predict future measurements from a set of previous ones : the next slice to be measured is always subject to new outside influences . as a result ,
the configuration variables on each new slice are arbitrary . from these and from earlier measurements ,
the momentum variables can then be deduced .
this situation contrasts with the more standard case of a spacelike boundary .
there , the phase - space data on the boundary s `` initial half '' can be used to fully predict the data on the `` final half '' . to summarize
, we ve outlined two problems with quantum gravity in finite spacelike - bounded regions : the transverse diffeomorphism redundancy and an apparent dimensional clash with holography .
timelike boundaries suffer too from the diffeomorphism issue , and also from a lack of full predictivity ( which bears no specific relation to gravity ) .
it s worth noting how the latter two problems are avoided in ads / cft , where a timelike asymptotic boundary is used .
first , transverse diffeomorphisms simplify at infinity , becoming conformal transformations of the boundary .
these can be easily incorporated as a redundancy of a quantum theory .
second , the boundary configuration variables , which can not be predicted from past measurements , assume the role of _ background _ parameters , since it costs infinite energy to change them .
the momentum degrees of freedom , which remain dynamical , can then be predicted from their values on an initial slice .
we will now demonstrate that , at least conceptually , the `` secondary '' problems from section [ sec : why_null : secondary ] can be avoided by using null boundaries .
we stress that an actual model of this kind does not exist , and it s not clear what it would look like , even in the context of classical field theory .
that question is explored on a preliminary level in sections [ sec : class]-[sec : gravity ] .
the nature of the quantum theory s `` output '' is briefly discussed in section [ sec : discuss ] .
we begin by describing the structure of null - bounded regions .
consider a spacetime metric not too different from minkowski space . a closed null boundary is then uniquely determined by a closed codimension-2 spacelike surface .
we refer to this surface as the boundary s `` equator '' .
the surface has precisely two lightrays orthogonal to it at each point .
these two rays generate the lightcone in the 1 + 1d transverse timelike plane . at each point of the equator , we choose the future - pointing half of one ray and the past - pointing half of the other .
specifically , we choose the `` ingoing '' half - rays , i.e. those along which the area decreases , and which eventually intersect .
thus , we limit ourselves to `` normal '' equators in the sense of @xcite , and do not consider trapped or anti - trapped ones .
we terminate each ray upon its first intersection with another ray .
the null boundary is then given by the union of these rays .
it consists of two smooth null hypersurfaces , or `` lightsheets '' .
generically , the lightray intersections away from the equator constitute a pair of codimension-2 spacelike surfaces - one for the past - ingoing rays , and one for the future - ingoing ones .
we refer to these surfaces as the past and future `` tips '' , respectively .
the boundary consists of an initial and a final half .
the initial ( final ) half - boundary stretches from the equator to the past ( future ) tip . see figure [ fig : boundary ] .
if the metric is a solution of gr with suitable energy conditions , then the boundary area decreases monotonously from the equator towards the tips @xcite .
the past and future tips are then anti - trapped and trapped surfaces , respectively . in the degenerate case where the past and future tips are single points , the two half - boundaries become lightcones .
the full boundary is then a causal diamond @xcite
. dimensions with trivial topology ( but not in the simple examples in section [ sec : class ] ) , the left and right sides of the figure are understood to be connected.,title="fig : " ] + the above recipe differs from that in @xcite , in that we terminate each lightray upon its first intersection . in @xcite ,
the rays are continued until they reach zero - area caustics - surfaces where they intersect with _ neighboring _ rays .
this choice is not suitable for us , since it does nt result in a closed hypersurface .
furthermore , we will see in section [ sec : gravity ] that nonvanishing tip areas lead to an entropy - like imaginary contribution to the gr action .
the tips and the caustics coincide at the tip surfaces edges . with sufficient curvature or a non - trivial topology , existence and
uniqueness issues arise .
for instance , in a de - sitter universe , equators above a certain size do not generate closed boundaries , because the lightrays never intersect .
another possibility is a disjoint equator . for a simple example , consider a spacetime of the form @xmath2 , with @xmath3 closed and spacelike . in this spacetime , consider the boundary @xmath4 , where @xmath5 is a 1 + 1d causal diamond .
the `` equator '' of @xmath5 consists of two points , so the equator of @xmath4 consists of two copies of @xmath3 .
the rays from each connected equator component to each tip generate a smooth lightsheet . in this case , then , the closed null boundary consists of four separate lightsheets .
we can now see how the null boundary structure avoids all the difficulties from section [ sec : why_null : secondary ] .
first , there is no extra degree of freedom allowing for transverse displacements of the boundary . in a given spacetime ,
the entire boundary is fixed by its codimension-2 equator .
indeed , any lightsheet consists of null geodesics , and is fully determined by a codimension-2 slice , up to a discrete choice for the rays initial direction .
another way to view the disappearance of the redundancy is that it gets swallowed by the constraint equation @xmath6 that requires the boundary normal to be null .
yet another intuition is that the hamiltonian constraint refers to displacements of the boundary _ in perpendicular _ to itself .
now , on a lightsheet , the normal direction is also tangent .
thus , perpendicular displacements become intrinsic diffeomorphisms , which do not alter the boundary s location in spacetime .
this implies that null boundaries are a more subtle and natural way to remove the redundancy than a conventional gauge choice .
we note a peculiar difference between the null - boundary setup and canonical gr . in the latter ,
the hamiltonian constraint plays a dual role : it encodes the theory s dynamics , as well as the redundancy under transverse displacements . with null boundaries ,
the redundancy is automatically taken care of , so the dynamics must be provided separately , e.g. in path - integral form .
this sort of `` splitting '' of an object with dual roles into two separate objects is a recurring feature in geometry and field theory on lightsheets .
the other problems from section [ sec : why_null : secondary ] evaporate similarly .
as in the timelike case , the preparation and measurement of a null boundary state proceed in a sequence of codimension-2 snapshots , never requiring an instantaneous codimension-1 measurement . in this sense , we can avoid the clash with holography that afflicts spacelike boundaries . on the other hand , as in the spacelike case , a null boundary is causally closed : full knowledge of the fields on the initial half is sufficient to deduce their values on the final half . in this way , the predictivity issue associated with timelike boundaries is also avoided .
the restriction to null boundaries is holographic in yet another sense , closely related to the demise of the hamiltonian constraint . in ordinary field theory
, once we refuse the non - predictivity of timelike boundaries , the state of a finite region can be described in terms of any spacelike hypersurface that terminates on the same codimension-2 surface .
the `` region '' is thus effectively _ defined _ by its closed codimension-2 boundary .
the limiting case of these spacelike hypersurfaces is just the null boundary defined above , with the codimension-2 surface as its equator .
the null boundary is thus the _ outermost _ boundary of a causally closed region in spacetime . by restricting to it ,
we are saying in some sense that the region can only be discussed `` from the outside '' .
this statement has a holographic ring to it , though of course it should nt be confused with a reduction in the degrees of freedom .
it also releases us from providing a unitary description of intermediate measurements , such as spacelike intermediate states .
this may relieve some of the difficulties concerning black holes and information ( see the recent discussion @xcite ) . in the above
, we advocated the use of null boundaries for finite - region qg .
we would now like to go further , and argue that this idea might be best implemented in a discrete model .
first , a closed null boundary can not be everywhere smooth .
like any closed hypersurface of fixed causal type in spacetime , it must contain codimension-2 `` corners '' , where the boundary is not differentiable .
in fact , one may argue that even smooth closed boundaries contain degenerate surfaces in the _ metric _ sense - the loci of signature flips . in the null case ,
the corner surfaces are the lightsheet ( self-)intersections at the equators and tips .
now , if such singular objects are inevitable in our description of the geometry , then it s better if they are the rule , not the exception . in this way , we are led to a picture in which the building blocks of spacetime are codimension-1 lightsheets intersecting at codimension-2 surfaces .
this is a natural situation for discrete models , particularly for spinfoams .
to take this point further , it s desirable for bulk histories to be described in the same terms as the boundary data .
this allows an interpretation of bulk histories as collections of on - shell states . for qg with boundary data on a set of intersecting lightsheets ,
it is therefore natural to define bulk geometries also in terms of intersecting lightsheets .
this again implies a discretization of the bulk spacetime region .
we emphasize that bulk histories are to be viewed as _ collections _
, not _ sequences _ of on - shell states .
we are not proposing to view `` jagged '' null slices as intermediate states of the full region - that would be tantamount to allowing spacelike states back in . instead
, we should view closed null boundaries within the bulk as on - shell states for the _ _ sub__regions defined by their equators .
the discretization we are suggesting here must differ in nature from that in e.g. regge calculus @xcite .
there , spacetime is constructed out of _ flat _ hyperfaces and their intersections .
in contrast , using flat lightsheets , one can not compose even a simple shape in minkowski space , such as a lightcone .
in particular , flat lightsheets can not accommodate area expansion or contraction .
thus , the discrete lightsheet elements must be allowed some rudimentary form of curvature .
for instance , one may specify an initial and final area for each element , allowing the two to differ .
constructing bulk geometries out of intersecting lightsheets makes sense also from a different perspective .
specifically , it seems to offer a compelling picture for the role of black holes .
consider a sum over bulk geometries within some null - bounded experimental region .
generically , these geometries will contain black holes , which are created and then evaporated through hawking radiation .
at least semiclassically , these black holes contain singularities .
due to evaporation , the singularities are not eternal .
thus , there exist finite `` hidden '' regions in the bulk , within which all causal lines fall into the singularity , failing to emerge outside .
the boundary of each hidden region is a causal horizon , i.e. a null hypersurface .
it is defined similarly to an event horizon , but with asymptotic infinity replaced by the boundary of the experimental region .
now , the quantum field fluctuations within the hidden regions should not affect any outside observables .
in other words , they should nt contribute to the path integral .
we are thus led to postulate that each hidden region should enter the sum - over - histories as a single discrete piece , with no resolution of its interior . favoring rules over exceptions , we arrive again at a picture of bulk geometries discretized into null - bounded cells .
having come this far , we can also reverse the logic , and interpret the interior of _ every _ discrete cell as a short - lived black hole .
this picture offers a neat realization of the notion that spacetime discreteness is due to horizons at the planck scale . in the above discussion
, the null boundary of every bulk cell is _ closed _ : it never falls into the singularity , which lies entirely on the inside ( with the possible exception of its edge ) . for consistency
, we should demand the same for the outer boundary of the experimental region .
in other words , the outer boundary should be a closed null hypersurface that never hits a singularity .
it is tempting to relate this to our insistence on non - trapped equators .
however , we are not aware of general results that relate the two notions . a possible independent way to enforce closure is to spread the boundary data throughout the lightsheets , which would imply that they extend as far as they should .
this contrasts with the usual practice for null boundaries , which is to place all the boundary data on either the initial or the final half .
note that there may well be a singularity _ outside _ the experimental region .
in particular , the region may lie entirely or partially inside the horizon of a large black hole .
all we require is that before they hit a singularity , the boundary lightsheets must `` close '' by intersecting .
the picture sketched above resembles a form of black hole complementarity @xcite . a black hole s interior can affect an observation , but only if the observer himself ( here , the boundary of the experimental region ) falls inside . from outside a black hole , only the horizon is relevant , with no resolution of the interior .
note also that singularities never explicitly appear .
they are always either outside the experiment , or within hidden regions . a boundary that hits a singularity corresponds to an experiment that could nt be completed , and therefore does nt need to be described .
in this section , we explore the use of null finite boundaries in the context of classical field theory . as explained in the introduction , we consider the on - shell action as the `` observable '' from which all physics is to be derived .
thus , we ll be discussing on - shell action functionals with null boundary data .
the goal is to see which features of field theory can be recovered from such actions alone . in effect , we are doing classical field theory with our hands tied behind our back .
the significance of this exercise is that in qg , we may be truly restricted to null - boundary observables , as argued in section [ sec : why_null ] .
as we will see , the action functional for null - bounded regions has a number of nonstandard properties .
these can all be related to a breakdown of the usual notion of canonical conjugates .
the underlying reason for that is the `` halving '' of the boundary degrees of freedom that takes place in the lightlike case . in relativistic field theory ,
lightsheets are characteristic hypersurfaces of the field equations . as a result ,
when providing boundary data on a lightsheet , we need half as many variables per point as we would on a non - null boundary ( assuming all redundancies are accounted for ) @xcite . to understand this heuristically
, we note that the conjugate momentum of a field @xmath7 is usually related to its derivative @xmath8 along the boundary s normal .
but the normal to a lightsheet is also tangent to it .
therefore , the value of @xmath7 on the lightsheet automatically contains the derivative @xmath8 .
the momentum data is thus included in the configuration data , which explains the `` halving '' of the degrees of freedom .
the above already suggests that something unusual should happen to the notion of canonical conjugates .
the point can be sharpened , as follows .
due to the `` halved '' degrees of freedom , a classical solution is now determined by the configuration data on just the initial half - boundary .
in particular , the data on the final half is determined as well . as a result
, we can not vary the configuration data at a boundary point without affecting it at other points .
but such localized variations are essential for the standard properties of canonical conjugates , as derived from the action
. this point can be easily missed if one proceeds formally with the lagrangian derivation , habitually treating variations at the initial and final half - boundaries as independent
. there is an important difference between our approach and what is normally meant by `` physics on null hypersurfaces '' .
usually , one takes e.g. a formula for the charge current , and projects it onto a null hypersurface ( which may or may not be at asymptotic infinity ) . the formula itself
, however , is obtained by conventional methods of field theory , which ultimately rely on action variations with non - null boundary data .
this approach can be termed `` null later '' : the null nature of the boundary enters late in the conceptual chain .
in contrast , our present approach is `` null first '' : we wish to use only concepts that can be derived from the actions of null - bounded regions .
thus , we do not take for granted the usual constructs of field theory , even when they have a sensible null limit .
the issue is whether they can be _ derived _ from null - bounded actions alone . on the other hand ,
when calculating the null - bounded actions themselves , we _ will _ use standard tools such as field equations , and we will take limits of non - null action formulas . this double standard is inevitable , since we do not have a complete reformulation of field theory .
we arrive at the null - bounded on - shell actions by whatever means we can , then forget the means and study the actions in their own right . as a final remark , we will not consider hamiltonian approaches to null - bounded field theory .
the reason is that there does nt seem to be a good notion of hamiltonian flow .
indeed , there is no way to `` evolve '' the initial half - boundary such that it remains null and still terminates on the equator , i.e. still spans the same causal region .
this of course echoes the arguments made in section [ sec : why_null : null ] . with no way to define the dynamics in hamiltonian terms , we must resort to the action functional .
we begin by considering the simplest field theory - a free scalar field in a @xmath9-dimensional background spacetime .
we define the metric signature as mostly - plus .
we take the field @xmath7 to be complex ; the results for a real @xmath7 are similar . the off - shell action in a spacetime region @xmath10 reads : @xmath11 the field equation is : @xmath12 integrating by parts and using the field equation , the action reduces on - shell to a boundary term : @xmath13 here , @xmath14 is the ( appropriately densitized ) covector associated with the codimension-1 boundary @xmath15 .
the indices @xmath16 span the boundary s tangent space . in the null case , the codimension-1 vector density @xmath17 is tangent to the boundary , and is directed along its generating lightrays .
it can therefore be written with a boundary - tangent index , as @xmath18 .
in fact , @xmath18 is the `` area current '' of the boundary lightsheet - the null analogue of a volume density . since the sign of @xmath19 in must be chosen as outgoing , we get ( for a mostly - plus metric ) that @xmath18 is future - pointing on the initial lightsheets and past - pointing on the final ones . in other words , @xmath18 points away from tips and towards equators
. see figure [ fig : boundary ] .
the on - shell action now reads : @xmath20 the first term in can be integrated , giving the fluxes of @xmath21 through the codimension-2 lightsheet intersections .
up to a sign , the flux of @xmath18 through any surface is the surface area density @xmath22 , where @xmath23 is the induced codimension-2 spacelike metric .
the sign is determined by the orientation of @xmath18 .
from the discussion above , it is positive for equator surfaces and negative for tips .
each surface contributes to the integral _ twice _ - once for every lightsheet that ends on it .
our on - shell action finally becomes : @xmath24 the divergence @xmath25 in the first term is the lightsheet s area expansion rate .
formula is remarkably simple .
it is a boundary integral of @xmath26 , with an algebraic kernel determined by the geometry .
this simplicity is somewhat deceptive , since the values of @xmath7 on the entire boundary are not independent .
the values on the final half - boundary are ( linear ) functionals of the values on the initial half .
this makes the action _ non - local _ , as we will now discuss .
let us briefly review what is meant by `` locality '' of an on - shell action .
viewed purely as a functional of boundary configuration data , the action @xmath27 is in fact typically non - local . a contribution that is local in the boundary data can usually be discarded , as it affects neither the dynamics nor the poisson brackets . at first sight
, this would suggest that our free - scalar action is trivial , and can be set to zero .
however , in the null setup , if our goal is e.g. to derive canonical momenta from @xmath27 , terms of this type can not be discarded .
the reason is that , as mentioned above , the boundary configuration fields are not independent in the null setup .
the action is a non - trivial functional of the _ independent _ boundary data , such as the values of @xmath7 on the initial half - boundary .
if @xmath27 is usually a non - local functional of boundary data , how does it encode the locality of the field theory ?
the answer is that canonical momenta , i.e. the variations of @xmath27 with respect to boundary fields , are local _ in terms of the bulk solution _ : they can be determined from the fields and their bulk derivatives at the variation point .
it is this notion of locality that is violated in the null setup .
we will now clarify this statement through examples . consider the simplest of free scalar theories : a massless scalar field in 1 + 1d minkowski space .
we choose null coordinates @xmath28 , so that @xmath29 .
the field equation reads simply @xmath30 . without loss of generality
, we can choose our null - bounded region as the diamond @xmath31 ( lorentz boosts preserve only the area @xmath32 ) .
the solution within the diamond is fully determined by the boundary data on the initial edges : @xmath33 with the constraint @xmath34 .
the solution for this boundary data reads : @xmath35 let us evaluate the action for this setup . in 1 + 1d , the `` area expansion rate '' vanishes
identically , so the first term does nt contribute .
the past and future `` tip surfaces '' are the points @xmath36 and @xmath37 , respectively .
the `` equator surfaces '' are the two points @xmath38 and @xmath39 .
the `` area density '' @xmath22 and the codimension-2 integrals are trivial .
we get : @xmath40 on - shell , the value @xmath41 at the future tip is a function of the @xmath7 values at the other three corners : @xmath42 using this relation , the action can be expressed in terms of @xmath43 alone : @xmath44 this expression for @xmath27 in terms of independent boundary data is non - local , in the following sense .
consider e.g. the derivative of @xmath27 with respect to the boundary value @xmath45 : @xmath46 this derivative involves field values far from the point @xmath38 .
as discussed above , it s not enough that depends on _ boundary _ data away from @xmath38 .
the non - locality statement rests on the fact that the `` momentum '' can not be determined from the field _ in a small bulk neighborhood _ of the variation point .
indeed , the solution in the bulk neighborhood of @xmath38 can only tell us about @xmath47 around @xmath48 and @xmath49 around @xmath50 , as well as their complex conjugates . on the other hand ,
the `` momentum '' depends on @xmath51 at @xmath52 .
a second example will allow us to sharpen this notion of non - locality .
consider a _ massive _ free scalar in 1 + 1d .
the boundary data can be chosen as before .
the field equation reads : @xmath53 this can be solved recursively , order by order in @xmath54 .
the solution reads : @xmath55 the expression for the on - shell action still holds .
this time , since information does nt travel only at the speed of light , the value @xmath41 at the future tip can not be deduced from the @xmath7 values at the other corners . thus , the four field values appearing in are independent . at first sight
, this means that the action is a local function of independent boundary data .
this would imply that the non - locality of @xmath27 is peculiar to the massless case .
however , a slightly refined notion of non - locality _ does _ hold for the massive scalar , and in fact seems to be generic . to formulate this notion
, we will need the concept of a _ local set _ of boundary data .
we define this as a collection of boundary data with connected support that determines a unique bulk solution .
the data set is a valid example .
however , the field values at the four corners , though independent , are not part of any local set .
indeed , such a set would have to involve at least three of the diamond s edges , and the @xmath7 values on three edges overspecify the solution .
if we insist on a local data set such as @xmath56 , then the field value @xmath41 in becomes a non - trivial function of the boundary data : @xmath57 plugging this into , we get the action in terms of @xmath56 .
its variation with respect to e.g. a value @xmath47 with @xmath58 reads : @xmath59 the factor @xmath60 is a field - independent function of the coordinate @xmath61 . on the other hand ,
the factor @xmath62 , given by the complex conjugate of , depends on the boundary data @xmath63 in the entire range @xmath64 . in particular , it depends on field values far from the variation point @xmath65 .
again , the non - trivial statement is that @xmath66 ca nt be determined from a small _ bulk _ neighborhood of @xmath65 .
for instance , if the variation point is near the past tip , @xmath66 still depends on @xmath63 values near the equator , which can not be determined from the tip s neighborhood .
to summarize , the on - shell action in a null - bounded region is non - local in the following sense : there is no local ( i.e. connectedly supported ) set of independent boundary data , such that the action variation with respect to it can be determined from the bulk solution in a small neighborhood of the variation point .
consider again the expression for the null - bounded on - shell action of a free scalar .
when the boundary s geometry is such that the expansion rate @xmath25 vanishes , the action is reduced to codimension-2 integrals at the lightsheet intersections . for instance , in the 1 + 1d cases studied above , the codimension-2 `` surfaces '' are the diamond s corners , and the action is determined by the field values on them , as in .
if one believes that all physics should be contained in the on - shell action , then this looks like a holographic reduction in the degrees of freedom
. we will refer to this property as `` explicit holography '' . unlike the non - locality discussed in section [ sec : class : scalar_locality ] , this is nt a general property of null boundary data . as far as we can tell , it only applies to free scalars with @xmath67 and to pure gravity ( see section [ sec : gravity ] ) .
there is another sense in which the free scalar action with @xmath67 is `` holographic '' .
recall that the codimension-2 boundary data on which the second term in depends is nt part of a local data set .
one can ask whether some form of `` holography '' would still take place in terms of a local set . for the massless free 1 + 1d scalar ,
the answer is clearly positive .
indeed , the action is written in terms of the local boundary data , and still depends only on field values at isolated points , i.e. at codimension-2 `` surfaces '' . however
, this does nt seem to happen in any other setup .
for instance , eqs . and show that for a massive scalar , the action depends on the entire codimension-1 local set @xmath56 .
but if we vary the action with respect to this local set , as in , we find a codimension-_two _ set of `` canonical momenta '' .
indeed , tells us that the `` momenta '' @xmath68 at different @xmath61 s are all given by the same functional @xmath62 , up to field - independent coefficients @xmath60 .
this is in fact a general feature of free scalar actions with @xmath67 . to see this , note that the value of @xmath7 at each point is a linear functional of the boundary data .
the second term in is a quadratic form over a codimension-2 set of such functionals .
it follows that this same set of functionals spans all the derivatives of @xmath27 .
if the action variations with respect to independent local boundary data are spanned by a codimension-2 set of `` canonical momenta '' , we will say that the action exhibits `` local - set holography '' .
we saw that for free scalars with @xmath67 , this property follows from `` explicit holography '' and from the linearity of the field equations . in section
[ sec : class : maxwell ] , we will show that the `` explicit holography '' is not a necessary condition .
it seems likely that linearity _ is _ necessary , but we will not attempt to prove this .
so far , we ve explicitly calculated @xmath27 in terms of local data sets only for 1 + 1d scalars . in higher dimensions ,
the boundary s expansion rate no longer vanishes automatically .
in particular , the `` holographic '' properties described above are not expected to hold for a causal diamond in @xmath69 dimensions .
when we do have a higher - dimensional boundary with @xmath67 , the free scalar theory ( whether massless or massive ) is qualitatively similar to the 1 + 1d massive case .
regardless of mass , propagation along codimension-2 spacelike surfaces makes the motion in the 1 + 1d transverse plane slower than light .
the simplest case is a flat spacetime with topology @xmath70 , where @xmath71 is a torus .
we can then choose a boundary of the form @xmath72 , where @xmath5 is a 1 + 1d causal diamond .
using null coordinates @xmath28 in @xmath73 , we fourier - transform the fields along the compact coordinates @xmath74 : @xmath75 where @xmath76 is the area of the codimension-2 torus . from here on , the results are completely analogous to those in section [ sec : class : scalar_locality ] .
the field equation reads : @xmath77 the on - shell action reads : @xmath78 a local set of boundary data is given by : @xmath79 with @xmath80 .
the future - tip values @xmath81 are expressed in terms of the local set as : @xmath82 as before , the variations of @xmath27 with respect to the local set are spanned by a codimension-2 set of `` canonical momenta '' . these consist of the boundary data @xmath83 and the linear functional @xmath81 , together with their complex conjugates , for each codimension-2 momentum @xmath84 .
a non-``holographic '' example can be easily constructed by perturbing the @xmath72 boundary shape .
consider a small constant expansion rate @xmath85 on the initial half - boundary , keeping @xmath67 on the final half . to accommodate this ,
the bulk metric should be modified by varying the @xmath71 volume as a function of position in @xmath73 .
`` explicit holography '' is then trivially lost , since the first term in no longer vanishes .
specifically , it is given by @xmath86 times the integral of @xmath26 over the initial half - boundary .
now , the values of @xmath7 on the initial half - boundary are a local data set .
thus , the first term in is a non - degenerate quadratic form over a codimension-1 local set .
the second term is a quadratic form of codimension-2 `` rank '' , by the same arguments as before .
the `` rank '' of the overall action is therefore dominated by the first term .
as before , this `` rank '' is also the `` number '' of independent canonical momenta .
we conclude that there is a codimension-1 set of independent momenta .
thus , `` local - set holography '' is also absent .
let us now turn to the on - shell null - bounded action of free maxwell theory in a background spacetime . in the simplest 2 + 1d geometry , we will find that `` local - set holography '' is satisfied , but `` explicit holography '' is not .
this implies that `` explicit holography '' is a special property of free scalars ( together with pure gravity - see section [ sec : gravity ] ) .
our derivation of `` local - set holography '' makes use of the spacetime dimension .
it may be overturned in 3 + 1d or higher , as well as in more complicated 2 + 1d geometries . as with the scalar theory
, we will also find that the action is non - local , in the sense of section [ sec : class : scalar_locality ] .
though we will demonstrate this only in the 2 + 1d calculation , we expect this property to hold generally .
a free maxwell field has the following action and field equations : @xmath87 } = 0 \ .
\label{eq : max_equations}\end{aligned}\ ] ] the field strength @xmath88 can be expressed in terms of a gauge potential as @xmath89}$ ] . on - shell , the action can then be written as a boundary term : @xmath90 where we again used the area current @xmath91 , oriented as in section [ sec : class : scalar_action ] . in the free scalar theory , for special boundary shapes
, we could further reduce the action to a sum of codimension-2 integrals .
this can not be done with the maxwell action , except in the degenerate 1 + 1d case .
thus , in the language of section [ sec : class : scalar_holography ] , free maxwell theory is not `` explicitly holographic '' .
we now turn to examine the simplest lower - dimensional setups . in 1 + 1d , the theory is trivial , with no dynamical degrees of freedom .
in coordinates @xmath28 , the field equations set @xmath92 to a constant . the action is simply @xmath93 times the spacetime volume of the region .
needless to say , the action s dependence on just a single field value @xmath94 should not be confused with the `` holography '' exhibited by free scalars .
the simplest non - trivial setup is a 2 + 1d flat spacetime with topology @xmath95 - a special case of the toroidal topology @xmath70 . as in section
[ sec : class : scalar_holography ] , we choose a boundary of the form @xmath96 , and use the same coordinate conventions .
we fourier - transform along the compact coordinate @xmath97 , imposing the reality condition @xmath98 . for each wavenumber @xmath99
, the action reduces to a sum over the four edges of the diamond @xmath31 .
the maxwell equations read : @xmath100 a local set of boundary data is given by : @xmath101 note that the two values @xmath102 and @xmath103 , though complex , correspond to two real data pieces , due to the reality condition @xmath104 . for @xmath105
, we have a single real data piece @xmath106 .
the same applies to the functions @xmath107 and @xmath108 .
to proceed , we must distinguish between the @xmath105 and @xmath109 modes . for @xmath105 , eqs . have the simple solution : @xmath110 the contribution of this mode to the action can now be found directly from : @xmath111 note that @xmath106 , @xmath112 and @xmath113 each contain an implicit @xmath97 integral .
the first term in is analogous to the 1 + 1d result .
the second term is the interesting one .
it should be contrasted with the free - scalar expression . by the same arguments as for the scalar theory ,
the second term is non - local in the @xmath28 plane .
both terms are consistent with `` local - set holography '' , being simple products of linear functionals .
however , the second term is _ not _ consistent with `` explicit holography '' : the integrals in can not be reduced to contributions from the diamond s corners .
indeed , at the corners , the solution is only sensitive to @xmath112 and @xmath113 at the edges of their argument s range .
let us now consider the @xmath109 modes
. these will be qualitatively similar to the free scalar case . for @xmath109
, we can choose a gauge where @xmath114 vanishes ( this is impossible for @xmath105 , due to the non - trivial holonomy around the circle ) . with this choice
, we have the following relations between the potentials and the field strengths : @xmath115 lowering indices with @xmath116 and keeping in mind the orientation of @xmath18 , the contribution to the action of each @xmath109 mode becomes : @xmath117 with @xmath118 . using the field equations , we can rewrite as : @xmath119 together with the complex - conjugate contribution , this reduces to a sum of codimension-2 terms : @xmath120 the full action is thus given by : @xmath121 we see that the @xmath109 terms are `` explicitly holographic '' . from this and from the theory s linearity , it follows that they also satisfy `` local - set holography '' . for completeness
, we should express the @xmath122 values in in terms of the local data set .
order by order in @xmath99 , the full solution of the field equations reads : @xmath123 we can now extract the @xmath122 values appearing in the action : @xmath124 the @xmath109 terms in the action are non - local in the @xmath28 plane , just like the @xmath105 terms .
we will not dwell on this , since the non - locality of @xmath125 is enough to render the whole action non - local . to conclude
, we have analyzed the on - shell null - bounded action of a free maxwell field in a simplistic 2 + 1d geometry .
this action is non - local ( in both the @xmath105 and @xmath109 pieces , though one is enough ) , exhibits `` local - set holography '' ( in both the @xmath105 and @xmath109 pieces ) , but is not `` explicitly holographic '' ( due to the @xmath105 piece ) .
we now come to the third property of null - bounded actions advertised in the introduction : they can not be used to derive global conserved charges . by now , this should not come as a big surprise . indeed , conserved charges are a special kind of canonical momenta , and we ve already seen that those behave very differently in the null formalism .
to derive the global charge of some finite - region classical solution , one usually performs the associated symmetry transformation on the boundary data .
the action variation in response to transforming the fields at a boundary point gives the charge density at that point .
the variation from transforming the entire initial ( or final ) boundary data gives the total charge .
of course , the variation from transforming _ both _ the initial and final data is zero , since this amounts to a global symmetry applied to the entire solution . in the null setup
, this recipe no longer works .
the reason , as usual , is that the boundary values of configuration fields are not independent .
one can not vary them at one point without affecting other points .
in particular , if we try to derive the total charge by transforming the initial boundary data , we will always get zero .
this is because the initial data fully determines the solution , including the final data .
a transformation of the initial data is thus a global symmetry operation , which leaves the action invariant .
there exist other methods for defining global charges , which do extend to the null case .
one can perform a transformation with an @xmath126-dependent parameter @xmath86 _ in the bulk _ , and then derive the current @xmath127 as the coefficient of @xmath128 in the off - shell action variation .
alternatively , one can turn on a small external gauge field @xmath129 in the bulk , and derive @xmath127 as the coefficient of @xmath129 in the on - shell action .
however , we consider both methods unsatisfactory . at the level of classical field theory , the first method requires off - shell actions , while the second requires fictitious fields .
neither method derives the global charge from the on - shell action of the actual field theory in question . at the quantum level
, both the off - shell and the external - field method translate into expectation values of bulk operators . in quantum gravity , which is our true object of interest , such quantities are problematic , since there is no background spacetime in which bulk operators may be localized .
we conclude that in the null formalism , global charges can not be derived in a satisfactory way .
the above arguments can be tested on the simple case of a free massless scalar in 1 + 1d ( section [ sec : class : scalar_locality ] ) .
consider a single complex scalar , with a global charge that generates phase rotations . without our self - imposed restrictions , it s easy to find the charge current .
one can just use the general formula derived from non - null boundaries ( the `` null - later approach '' ) , or rederive it using e.g. the off - shell method . integrating this current over the initial half - boundary ,
we get the following expression for the charge in terms of initial data : @xmath130 unlike similar previous expressions , this one ca nt be integrated into codimension-2 terms . in other words , it genuinely depends on the values of @xmath7 on the entire half - boundary . in contrast , the on - shell action is only sensitive to the @xmath7 values at the corners @xmath131 .
we conclude that the charge can not be derived from the action .
this conclusion does nt assume any particular _ method _ of derivation .
the action just does nt contain the necessary information .
while global charges are expected to disappear in qg , gauged charges are not .
the crucial difference is that a gauged charge can be measured `` holographically '' via the gauss law on a codimension-2 enclosing surface , whereas a global charge must be integrated over a codimension-1 volume slice .
more precisely , we expect _ abelian _ gauged charges to survive .
a non - abelian gauss law is problematic , since the flux can not be integrated gauge - covariantly over an extended surface ( in ads / cft , this problem is solved by a preferred gauge on the boundary ) .
we would like the `` null - first '' formalism of classical field theory to reflect the properties of qg in this context as well .
thus , we d like to be able to derive gauged abelian charges .
this appears to be possible , at least in a large class of theories that includes gr and the bosonic sector of the standard model lagrangian .
though the construction proposed below `` bends the rules '' somewhat , it does nt seem to violate the philosophy of this paper . with non - null boundary data ,
a gauged charge can be derived from the action by varying the boundary value of the gauge potential . in standard maxwell theory ,
the resulting action variation is the boundary value of the electric field .
more generally , it is the noether potential for the gauge symmetry @xcite .
the charge enclosed by a codimension-2 surface on the boundary is given , via gauss s law , by the flux of this quantity . with a null boundary , this general procedure no longer works , for the usual reason :
the boundary field values can not be varied independently .
we will now propose an alternative method to derive the gauged charge , which makes use of the boundary s null nature .
it involves a sort of gauge transformation , which affects the action without perturbing the bulk solution .
we take the charged fields to be scalars and denote them by @xmath7 .
they may carry internal indices , which we suppress in the notation .
the gauge potential @xmath129 enters the lagrangian @xmath132 through covariant derivatives @xmath133 and through the field strength @xmath88 .
we assume that the lagrangian is quadratic in derivatives , so that boundary problems are well - posed .
in particular , it is quadratic in the derivatives @xmath133 of the charged fields .
we assume that the second - order terms take the form @xmath134 , so that the gradient indices are contracted with each other .
we also assume that linear terms in @xmath133 are not multiplied by any other derivatives .
now , consider a classical solution of the field theory in a null - bounded region .
let us focus on one of the boundary lightsheets . in its neighborhood
, we define a coordinate @xmath135 such that the lightsheet is at @xmath136 , with the bulk at @xmath137 .
the lightsheet coordinates are denoted by @xmath138 .
let us now vary the gauge potential as follows : @xmath139 here , @xmath86 is a constant .
the `` delta function '' @xmath140 has a unit integral when integrated up to the boundary from the inside . in components
, we have @xmath141 and @xmath142 .
this change in @xmath129 effectively offsets the boundary lightsheet from the bulk by a gauge phase @xmath86 .
let us analyze the effects of on the classical solution and its action .
we argue that the bulk field configuration with the variation remains a solution , with no need for any compensating variations .
first , notice that the only component of @xmath143 that gets altered by is @xmath144 .
therefore , the field strength @xmath89}$ ] remains unchanged .
@xmath145 itself enters the action through the kinetic terms of the charged fields . by the assumptions above
, @xmath146 is a linear function of @xmath147 , with derivative - free coefficients .
now , @xmath147 is a derivative along the lightsheet normal , which is also tangent due to its null nature .
thus , the action variation due to involves only configuration field values on the boundary itself . as a result ,
the stationarity condition on the off - shell action is unaffected .
we conclude that the variation does nt disturb the classical solution . as a corollary ,
the on - shell action variation @xmath148 due to is simply an integral of @xmath149 over the lightsheet .
but the latter is just the current component @xmath150 , i.e. the flux of @xmath127 through the lightsheet .
adding up the variations @xmath148 for the initial ( or final ) lightsheets , we get the total charge in the null - bounded region . in a sense , our derivation can be viewed as a special case of the standard gauss - law method . in this perspective , the transformation on e.g. the final half - boundary is seen as the propagation of a change @xmath151 to the _ initial _ data on the equator .
now , at the equator , what looks like the transverse component @xmath145 on the final lightsheet is an _ intrinsic _ component @xmath152 on the initial lightsheet , transverse to the equator itself .
the charge is then viewed as the action variation in response to a change in @xmath152 on ( and transversely to ) an enveloping surface , just as in the standard gauss law .
this is of course a whimsical interpretation , since is a gauge transformation , and can not be truly said to `` propagate '' .
in section [ sec : class ] , we calculated some on - shell null - bounded actions for free field theories .
we arrived at two kinds of results . using the field equations to integrate by parts , but without solving them explicitly , we arrived at functionals such as and . by explicitly solving the field equations
, we then arrived at expressions such as , giving the action in terms of a local data set . in this section
, we consider the on - shell , null - bounded action of general relativity .
it is nt feasible to solve the field equations in sufficient generality , so we wo nt be able to express the action in terms of a local set
. however , it is possible to arrive at analogues of eqs . and , especially for pure gravity .
it turns out that the pure gr action is similar to the free scalar action , but is even simpler .
in particular , it reduces to a sum of codimension-2 integrals , this time without any assumptions on the null boundary s shape .
we also identify a universal imaginary term that reproduces the bekenstein - hawking entropy .
this term is accompanied by a divergence in the action s real part .
we begin with the gr action , complete with the gibbons - hawking boundary term @xcite . for non - null boundaries ,
this reads : @xmath153 in the bulk term , we allowed for a cosmological constant @xmath154 and a matter lagrangian @xmath155 . in the boundary term , @xmath156 is the determinant of the ( spacelike or timelike ) boundary metric ; @xmath157 is the trace of the extrinsic curvature , defined using the outgoing unit covector ( which may correspond to either the outgoing or ingoing normal _ vector _ , depending on signature ) .
we will focus on the boundary term in . in pure gr , without matter or cosmological
constant , this gives the entire action , since the bulk term vanishes on - shell . in any case
, we will see that the boundary term has special features - an imaginary part and a divergent real part - that distinguish it from the bulk term .
adapting the boundary term to the null setup is a bit non - trivial . indeed ,
neither factor in the product @xmath158 has a well - defined finite limit in the null case .
the volume density @xmath159 vanishes , and there is no unit normal with which to define @xmath157 .
however , the product itself does have a null limit .
it is given by the area expansion rate @xmath25 , where @xmath18 is the boundary area current .
intuitively , both quantities describe volume or area increase along the boundary s normal .
a more rigorous argument is given below .
first , extend the boundary hypersurface into a family of foliating hypersurfaces in a small neighborhood of the point of interest .
they should all have the same signature , e.g. in the null case they should all be null . also , in the non - null case , choose the hypersurfaces to be evenly spaced , so that the normal distance from each to its neighbor is constant .
one can then define a coordinate @xmath135 labeling the hypersurfaces , such that the outgoing covector @xmath160 has a constant ( not necessarily unit ) norm . in the null case
, that norm goes to zero . as before ,
we denote the coordinates within each hypersurface by @xmath138 .
it is now easy to show that in the adapted coordinates @xmath161 , both @xmath158 ( in the non - null case ) and @xmath25 ( in the null case ) are given by @xmath162 .
we conclude that @xmath25 is the correct null limit of @xmath158 .
the gibbons - hawking integral over the boundary lightsheets therefore reads : @xmath163 this integral is immediately reduced to a linear combination of the lightsheet intersection areas ( see section [ sec : class : scalar_action ] for the sign considerations etc . ) : @xmath164 where @xmath165 , @xmath166 and @xmath167 are the equator , initial tip and final tip areas , respectively .
this remarkable formula , like the free - scalar result , is a local functional of the boundary configuration fields - in this case , the intrinsic metric . as discussed in section [ sec : class : scalar_locality ] , in the null case , such an action should not be considered trivial .
the additional contributions derived in section [ sec : gravity : corner ] below will also be linear combinations of the intersection areas . in the jargon of section [ sec : class : scalar_holography ] , this means that the action of pure gr is `` explicitly holographic '' .
its behavior in terms of independent boundary data is much more difficult to study , due to the theory s non - linearity .
the reasonable bet for a generic interacting theory is that the action is non - local ( section [ sec : class : scalar_locality ] ) , while `` local - set holography '' ( section [ sec : class : scalar_holography ] ) is absent .
also , the `` explicit holography '' of is apparently restricted to pure gravity .
generically , we expect that the bulk term in can not even be reduced to a codimension-1 integral .
the contribution arises from area expansion towards the equator .
alternatively , it can be viewed as the bending of the lightsheet normal , as one moves transversely to its generating lightrays .
there is no proper sense in which a lightsheet bends _ along _ its generating rays , since the latter are geodesics .
however , this is no longer true when two lightsheets intersect .
at the intersection , the normal vector must instantaneously `` bend '' as one passes from one lightsheet to the other .
let us analyze the resulting contribution to the boundary action . in euclidean signature ,
the contribution to the boundary term from hypersurface intersections is well - known @xcite .
such `` corners '' can be understood as limiting cases of fast - bending smooth curves .
this translates into a delta - function in the extrinsic curvature , which has a well - defined integral .
the corner contribution to the action is then @xmath168 , where @xmath169 is the codimension-2 area of the `` corner '' , and @xmath170 is the bending angle . in the lorentzian case ,
complications arise , since angles ( i.e. boost parameters ) can now diverge .
this is particularly true for null directions .
a consistent assignment of angles @xmath171 for the entire 1 + 1d plane is given in appendix c of @xcite .
in each quadrant of the lorentzian plane , the boost angle runs between @xmath172 and @xmath173 , diverging at the null lines .
when crossing a null line counter - clockwise into a neighboring quadrant , the angle picks up an _ imaginary contribution _ @xmath174 , without changing the sign of the infinite real part .
see figure [ fig : angles_plane ] .
thus , the total angle for a complete counter - clockwise circuit is @xmath175 , in place of the euclidean @xmath176 .
the virtue of this assignment is that @xmath177 and @xmath178 are given consistently by : @xmath179 where we take @xmath180 for negative @xmath181 . in other words
, we are analytically continuing away from the @xmath182 quadrant .
the procedure is unique up to complex conjugation .
we will see that the above choice is preferred over its complex conjugate by the requirement that quantum amplitudes @xmath183 do not explode exponentially .
our conventions differ from @xcite by an overall factor of @xmath184 .
in particular , we treat boost angles within a quadrant as real . in the lorentzian plane , according to eq . .
this is one of two complex - conjugate choices , distinguished by the sign convention below .
the horizontal and vertical axes are @xmath97 and @xmath185 , respectively .
the angles are defined up to integer multiples of @xmath186.,title="fig : " ] + let us apply this recipe to the normal direction on our null boundary and its corners . as with ordinary corners , we can imagine them as fast - bending smooth curves .
we then have momentarily a spacelike normal on the equator and timelike normals at the tips . to keep track of the angles , it s best to consider a 1 + 1d version of the boundary , where the rays actually close into a diamond .
see figure [ fig : angles_boundary ] .
the higher - dimensional result is obtained by inserting area factors where necessary . as an intermediate step
, one can think of a boundary with @xmath4 topology ( section [ sec : why_null : null ] ) .
, as applied to the null boundary normals from figure [ fig : boundary ] . for clarity , spacelike and timelike normals
are included at the equator and tips .
they arise if one replaces the corners with smooth curves .
empty circles denote the `` signature flip '' surfaces , where the boost angle crosses into a neighboring quadrant .
the path integral favors placing them infinitesimally close to the tips .
the `` corner '' contributions to the action can be read off by traveling counter - clockwise and picking up angle differences times @xmath187 factors.,title="fig : " ] + each corner contributes a divergent term to the action , due to the infinite boost between one null direction and the other .
the sum of these divergent terms reads : @xmath188 this is the same as @xmath189 from , but with a factor of @xmath172 .
the signs in can be read off from the angle differences in figure [ fig : angles_boundary ] , as one walks along it clockwise .
this can be verified by evaluating @xmath157 on smooth `` zoomed - in '' versions of the equator and tips , i.e. on convex timelike and spacelike curves , respectively .
we note that when @xmath190 , as is always the case in 1 + 1d , @xmath191 may have a finite value .
the normal angle should also get imaginary contributions of @xmath174 at signature flips , i.e. when it crosses into a different quadrant of the lorentzian plane .
all we can say from figure [ fig : angles_boundary ] is that this happens _ somewhere _ along the null extent of each lightray .
overall , there should be two surfaces , one on the initial half - boundary and one on the final half , where the signature flip takes place .
let us denote their areas by @xmath192 and @xmath193 , respectively .
applying the same counter - clockwise recipe for the signs , we find the following imaginary contribution to the action : @xmath194 the full boundary term in is given by @xmath195 .
the bulk term is generically real and finite on - shell .
therefore , @xmath191 is the divergent part of the total action , while @xmath196 provides the imaginary part .
the complete gr action thus satisfies : @xmath197 it remains to fix @xmath192 and @xmath193 . to do this
, we can imagine a path integral over the different possibilities .
the imaginary action @xmath198 suppresses the amplitudes @xmath183 exponentially , as @xmath199 . like in euclidean path integrals , this selects the smallest possible imaginary action .
thus , we should give @xmath192 and @xmath193 the smallest possible values .
now , it is known @xcite that on - shell , the boundary area decreases monotonously from the equator towards the tips ( and will continue decreasing until a caustic is reached ) .
thus , the `` flip '' surfaces must be chosen next to the tips .
this fixes @xmath200 and @xmath201 .
the factor of 2 arises from the two rays intersecting at each tip point .
alternatively , the boundary area infinitesimally close to the tip is twice the tip area .
substituting into , we arrive at the following result in terms of the _ average _ tip area : @xmath202 this is remarkably similar to the bekenstein - hawking entropy formula .
indeed , an imaginary ( or euclidean ) action is usually identified with the logarithm of a partition function .
this is the same as the entropy , up to terms arising from conserved charges , such as @xmath203 for energy and @xmath204 for angular momentum . for a euclidean black hole
, the latter can be viewed as contributions from the boundary at infinity , while the entropy is a contribution from the horizon @xcite .
this might explain why the imaginary action gives the entropy formula directly : for our boundary structure , there is no `` term at infinity '' to consider .
it should be noted that contributions of the form - are not unique to null boundaries .
any closed hypersurface in spacetime contains infinite boosts and signature flips .
the consequences for more general boundaries are addressed in a separate work @xcite .
it turns out that the action s imaginary part remains related to the black hole entropy formula , including in higher - order lovelock gravity .
this coincidence was explained in @xcite by relating the lorentzian calculation to a euclidean one .
in this paper , we argued that restricting to null boundaries may be a useful step in the grand project of finite - region quantum gravity .
we set out to explore this idea in the context of classical field theory .
the expected breakdown of conventional observables in finite - region qg is then mirrored by a breakdown of the standard tools of classical mechanics .
somewhat fancifully , we identified certain artifacts in the null version of classical field theory with certain features that are expected of quantum gravity - non - locality , holography and the absence of global charges .
finally , while exploring the gr action for null boundaries , we found an imaginary term that resembles the bekenstein - hawking entropy . in our interpretation of the on - shell actions
, we claimed that terms such as and , which are local in configuration data , must be taken seriously in the null case , and should in fact be treated as `` non - local '' .
a possible objection is that such terms can be added to the action _ before _ taking the null limit , rendering them ambiguous .
however , it seems hard to create such ambiguity in practice . to be concrete ,
let us demand that any new boundary term must be the null limit of a well - defined , analytical expression in the non - null case . for classical field theory ,
this looks like a sensible requirement .
first , note that and themselves arise from terms that involve normal derivatives . outside the null limit ,
those are _ not _ local in configuration data , and are therefore not arbitrary .
if we try to add boundary terms _ without _ normal derivatives , there appear to be two possible outcomes .
they either vanish in the null limit ( due to the vanishing volume density ) , or become ill - defined ( due to factors of the inverse boundary metric ) .
one may also consider introducing codimension-2 terms such as or directly into the non - null action , by assigning them to corners , or to loci of signature flips .
however , such singling out of special surfaces in the general action formula can be ruled out by analyticity .
the most significant result of this work is arguably the gr `` entropy '' formula .
its derivation bears some resemblance to the analysis of euclidean black holes @xcite .
there as well as here , an imaginary action is derived through analytical continuation . however , in our derivation , there is no wick rotation of the entire metric , and thus no stationarity requirement .
instead , one can view the imaginary boosts during signature flips as resulting from `` localized wick rotations '' .
another distinguishing feature of our calculation is that it makes no reference to asymptotic infinity .
the entropy is obtained directly , without involving thermodynamic potentials ( associated with stationarity ) or conserved charges ( associated with boundary terms at infinity ) .
a curious feature of the `` entropy '' result is that the relevant areas are those of the _ smallest _ boundary sections .
this contrasts with the notion that the entropy bound is concerned with the largest available area , in particular in the null context @xcite .
the picture suggested by seems more compelling : the smallest area defines a bottleneck on the available information .
for instance , a causal diamond with vanishing tip areas @xmath205 is associated with a pointlike observer @xcite .
now , a truly pointlike observer can not carry any information at all .
in particular , he can not design an experiment in his causal diamond , or collect the results of measurements . from this perspective
, it stands to reason that the entropy bound for a causal diamond should be zero .
an important sanity check for our `` entropy '' calculation is to apply it to a stationary black hole solution .
this turns out to be possible , and produces the correct result @xcite .
we conclude with a comment on the nature of the quantum theory which was advocated in section [ sec : why_null ] .
one s first impulse would be to assign a hilbert space to the initial and final half - boundaries , and to think of a path integral as providing an `` s - matrix '' of transition amplitudes .
however , this seems problematic .
the issue is with the operational meaning of hilbert products between states on the same hypersurface . in ordinary qft
, they can be viewed as a limiting case of transition amplitudes between states on different hypersurfaces , obtained as the hypersurfaces approach one another . in the null setup
, this procedure fails , due to the rigidity of lightsheets : one can not have two infinitesimally close lightsheets with support on the same equator .
there are only transition amplitudes between the initial and final lightsheets , with no intermediates .
these amplitudes define a map linking the initial and final states , but not a hilbert product for either . in @xcite ,
a similar situation was described , with the past and future infinities of de - sitter in place of our initial and final lightsheets . there , a hilbert product was constructed by using also the cpt map between initial and final states
. this does not seem to work in our setup , since an irregular equator would not be invariant under cpt .
this failure in constructing a hilbert space is consistent with the expected failure of standard quantum mechanics in finite - region qg .
the nature of the theory s correct output is an open question .
a variant of these remarks applies also to lqg and spinfoams .
there , one can not understand the hilbert product as a limit of transition amplitudes because the boundary hypersurfaces are _ too free _ to move around : the initial and final half - boundaries can approach each other in principle , but it is nt clear how to _ make _ them approach each other using gr - like boundary data .
this work is supported in part by the nsf grant phy-1205388 and the eberly research funds of penn state .
the early stages were carried out in tel aviv university , supported in part by the israeli science foundation center of excellence , the us - israel binational science foundation ( bsf ) , the german - israeli foundation ( gif ) and the buchmann scholarship fund . | we argue that finite - region observables in quantum gravity are best approached in terms of boundary data on null hypersurfaces .
this has far - reaching effects on the basic notions of classical and quantum mechanics , such as hamiltonians and canonical conjugates .
such radical properties are not unexpected in finite - region quantum gravity .
we are thus motivated to reformulate field theory in terms of null boundary data . as a starting point
, we consider the on - shell action functional for classical field theory in finite null - bounded regions .
closed - form results are obtained for free scalars and for maxwell fields .
the action of classical gravity is also discussed , to the extent possible without solving the field equations . these action functionals exhibit non - locality and , in special cases , a `` holographic '' reduction of the degrees of freedom .
also , they can not be used to define global charges .
whereas for ordinary field theory these are just artifacts of a restrictive formalism , in quantum gravity they are expected to be genuine features .
this further supports a connection between quantum gravity and null - boundary observables . in our treatment of the gr action
, we identify a universal imaginary term that reproduces the bekenstein entropy formula . |
while neutrino oscillation experiments have successfully shown that neutrinos change flavor , and therefore have non - zero mass , the absolute mass scale remains unknown .
oscillation experiments have accurately measured two independent mass differences between the three known mass states .
so far the data only provide upper limits on the neutrino mass , but experiments continue to probe lower mass ranges . if zero - neutrino double - beta - decay is measured , it can be interpreted as a measurement of neutrino mass , provided that the decay occurs via straight - forward weak processes .
neutrino mass limits can be implied from cosmological data , as well , due to the effects of neutrino mass on structure formation in the universe .
the third method is using single beta decays .
the oscillation data indicate that the neutrino mass measured by beta decay experiments , @xmath0 , must satisfy either @xmath1 or @xmath2 , depending on the ordering of the mass states .
the most successful direct measurements of the neutrino mass to - date have been performed using tritium beta decays .
neutrino mass has an effect on the kinematics of decay process @xcite .
while the neutrinos themselves are difficult to measure , the energy of the outgoing electrons can be precisely determined .
the neutrino mass can then be inferred from the shape of the electron energy spectrum .
the limits on the neutrino mass as measured with tritium beta decays have steadily improved over several generations of experiments .
the next - generation tritium experiment , katrin , is currently under construction .
it anticipates having a sensitivity of 200 mev @xcite .
the lower limits for the neutrino mass from oscillation experiments provide a strong motivation for probing to lower neutrino masses .
however , with katrin , the technologies used in spectrometer - type tritium experiments have been pushed to their current practical limits .
a new technique is needed to push the mass limits lower .
the technique proposed is the use of cyclotron radiation emitted by the beta decay electrons to measure their energy spectrum .
to determine an electron s energy from its cyclotron radiation , we plan on measuring the frequency of that radiation ; fortunately , the frequency of radiation can be measured very precisely , making the technique an attractive method for a potential future neutrino mass measurement .
the best prospects for directly measuring the mass of the neutrino , and the current limits for such measurements , come from the use of beta decays .
the neutrino mass is probed by carefully measuring the energy of the outgoing electron from the decay . the energy spectrum for beta - decay electrons
is given in eq .
[ eq : betadecayspectrum ] : @xmath3 the fermi function , @xmath4 , takes into account the coulomb interactions of the electron with the recoiling nucleus ; @xmath5 is the proton number of the final - state nucleus , @xmath6 is the electron s kinetic energy , @xmath7 is the electron s momentum , @xmath8 is the q - value of the decay , and @xmath9 are the elements of the pmns matrix for neutrino mass states @xmath10 , @xmath11 .
the only dependence on the neutrino mass comes from the phase - space factor .
it is independent of all other properties of the neutrino , including whether neutrinos are majorana or dirac particles .
two primary techniques have been considered for measuring the neutrino mass with beta decays .
cryogenic bolometers are being employed by the mare collaboration @xcite to measure the heat deposited in the calorimeters by the beta decays of re .
the first iteration of the experiment aims to confirm the measurements made by the mainz and troitsk experiments and place a limit at the ev level .
the follow - up experiment , mare - ii , will reach for the sub - ev level .
the second technique involves the use of a spectrometer to precisely select high - energy electrons from tritium decays . the most recent experiments to use this technique are the mainz and troitsk experiments .
they placed similar limits on the neutrino mass : @xmath12 @xcite .
katrin , aims to lower that limit by an order of magnitude , to 200 mev ( 90% cl ) @xcite .
katrin is currently under construction in karlsruhe , germany . in aiming to bring the mass limit down to 200 mev
, katrin will have reached the practical limits of the spectrometer method .
the new technique for determining the electron energy , using cyclotron radiation , is described in the following section .
tritium has several advantages with respect to its use as a source for measuring the neutrino mass .
it is a simple nucleus , so the matrix element of the decay is relatively easy to calculate .
the q - value for the reaction is relatively low , at 18.575 ev , and the half - life is only 12.3 years . since the spectrometer technique is reaching its practical technological limit
, it is worth investigating new techniques that can further take advantage of tritium s convenient properties .
the project 8 collaboration proposes an alternate method of measuring the electron energies : measure the cyclotron radiation emitted by the electrons spiraling around magnetic field lines .
an enclosed volume of tritium is placed in a uniform magnetic field . as the tritium nuclei decay
, the electrons will spiral around the magnetic fields lines .
as they do so , they are being accelerated , and therefore emit cyclotron radiation .
the frequency of that radiation is proportional to the magnetic field strength , and inversely proportional to the electron s kinetic energy : @xmath13 by measuring the frequency of the cyclotron radiation , one can measure the electron s kinetic energy without interfering with the electron itself . using a 1-t magnetic field ,
the endpoint of the tritium spectrum falls around 26 ghz .
this is a frequency band of significant commercial interest ; we can therefore take advantage of commercially - available antennas and detectors . the power emitted
as cyclotron radiation depends both on the relativistic velocity electron , @xmath14 , and the angle at which it is emitted relative to the direction of the magnetic field , @xmath15 : @xmath16 the electrons that radiate the most power will be the easiest to detect , because the signal - to - noise ratio will be higher .
equation [ eq : power ] shows that the power will be greatest for electrons with @xmath17 .
conveniently , these electrons also travel the slowest along the magnetic field lines , increasing the amount of time they can be observed .
several configurations for the experiment are under consideration .
figure [ fig : simulationsetup ] shows one possible setup .
the tritium is contained within a chamber inside of the bore of a 1-t magnet .
an antenna array is arranged around the outside of the source chamber .
as the electrons radiate , the microwave radiation passes through the chamber wall and is detected by the antenna array .
the signals from the individual antennas in the array can be summed coherently , since multiple antennas will be viewing each electron at the same time . one possible setup for an experiment measuring the cyclotron radiation from tritium beta decay electrons in a magnetic field .
this design was used in an initial simulation to examine the potential for this type of experiment @xcite .
the tritium gas is enclosed in a chamber that is transparent to microwaves .
the chamber is then surrounded by an antenna array , all of which is placed inside the bore of a 1 t magnet.,width=480 ] figure [ fig : powerspectrum ] shows the power spectrum from a simulation of 100,000 tritium decays in 30 @xmath18s in the hypothetical experiment in fig . [
fig : simulationsetup ] .
since frequency is inversely proportional to electron energy , the rare high - energy electrons are at lower frequency , near 26 ghz .
low - energy electrons , making up the vast majority of the spectrum , are piled up towards 27 ghz .
the pileup from low - energy electrons is a possible complication : unlike spectrometer - type experiments , we do not have an intrinsic method for rejecting low - energy electrons . instead , we can narrow the bandwidth such that the event rate is low enough that individual events can be identified . simulated power spectrum from the hypothetical experiment shown in fig . [
fig : simulationsetup ] .
@xmath19 beta decays were simulated over 30 @xmath18s in a 10-m - long uniform magnet ( @xmath20 ) .
the vertical arrow indicates the location of the tritium beta - decay endpoint in frequency space.,width=480 ] the primary concern for making a precise electron energy measurement is the ability to measure frequency precisely .
the desired energy precision is therefore the place to start in considering the requirements for this type of experiment . to achieve the necessary energy precision , @xmath21
, we need to achieve a relative frequency precision of @xmath22 .
katrin is designed to achieve @xmath23 ; for project 8 to achieve a similar accuracy means that @xmath24 .
this accuracy is reasonable with current technologies . with a 1-t magnetic field , @xmath25 at 26 ghz .
the desired frequency accuracy determines for how long we must be able to observe single electrons . to have a frequency resolution of @xmath26
, we must measure each electron for @xmath27 . with the design parameters discussed above
, the electrons must be coherently measured for at least 20 @xmath18s .
the minimum measurement time places constraints on a number of physical parameters of the experiment
. the gas density must be low enough that , on average , 18.6 kev electrons can travel for @xmath28 without scattering .
furthermore , the experiment must be large enough so that the electron can be tracked continuously .
the signal detected for a single electron may be more complicated than the single frequency at which the cyclotron radiation is emitted .
in particular , the detected signal can include a doppler shift due to the velocity of the electron parallel to the magnetic field , @xmath29 , a dependence on the electron - antenna distance , and effects from the angular dependence of the power distribution of the radiation .
the way these effects are represented in the data depends strongly on the antenna configuration . for the hypothetical experiment shown in fig .
[ fig : simulationsetup ] , if the signals from the different antennas are summed coherently , there will be sideband peaks from the doppler shift . fig . [
fig : zoomedpowerspectrum ] zooms in on the high - energy ( low - frequency ) region of the power spectrum shown previously . in this simulation
a triplet of peaks from a single high - energy electron are easily distinguishable .
though it is an antenna - design - dependent effect , the triplet of peaks due to the doppler shift could be a convenient tool for tagging electrons . zooming in on the beta - decay endpoint of the power spectrum from fig .
[ fig : powerspectrum ] .
the triplet of peaks from a high - energy electron are easily distinguished from the background.,width=480 ] the project 8 collaboration has put together a prototype experiment to explore the practical use of cyclotron radiation as a method for measuring the decay - electron energy . the initial goal of the prototype is to verify that we can , in fact , detect the cyclotron radiation from a single electron .
we will use a @xmath30kr radioactive source , which emits a monoenergetic electron .
this excited nucleus emits a 17.8 kev electron , and has a half - life of 1.83 hours .
the source is a good stand - in for tritium : it is gaseous , emitting the electrons isotropically , and the electron energy is close to the tritium - decay endpoint . figure [ fig : prototype ] shows a diagram of the prototype , which is located at the university of washington , in seattle , wa .
a superconducting solenoid provides the 1-t magnetic field .
the electrons are trapped in a small ( @xmath31 ) magnetic bottle in the bore of the magnet .
the magnetic field traps the electrons in the horizontal plane ; a current loop within the bore of the magnet decreases the field slightly in a small volume , trapping the electrons vertically . whether or not electrons are trapped depends on the depth of the magnetic bottle potential , and the pitch angle of the electrons .
electrons with large @xmath32 will be trapped .
fortunately , these electrons also emit the most power as cyclotron radiation . for the initial setup of the prototype , only electrons with large pitch angles ( @xmath33 )
are trapped .
though this angle selection severely limits the number of electrons we will detect , it maximizes the signal - to - noise ratio .
diagram of the prototype experiment located at the university of washington , in seattle , wa .
this particular design shows a two - wire antenna to detect the cyclotron radiation , though other designs are also being considered and tested.,width=288 ] the ability to open the magnetic bottle by turning off or reversing the current in the current loop will allow us to confirm that we are indeed trapping electrons , and accurately measure the noise levels .
in addition to detecting the cyclotron radiation , we will employ more traditional means of detecting electrons to monitor the presence of @xmath30kr in the trap , and verify that electrons are actually trapped .
multiple antenna configurations will be investigated .
the diagram in fig .
[ fig : prototype ] shows a two - wire antenna .
the two wires act as a waveguide .
the cyclotron radiation couples to the transmission modes of the waveguide , and is propagated to the ends .
one end is attached to a low - noise cryogenic amplifier , after which the signal is transmitted out of the magnet bore , and mixed to a lower frequency .
a second antenna design being considered is a rectangular waveguide .
this design is particularly attractive because the gas can be confined within the waveguide horizontally , and with kapton windows in the vertical direction .
the rectangular - waveguide antenna is capable of detecting signals emitted anywhere within the gas volume , whereas the two - wire antenna is most sensitive to electrons between the wires .
once we have shown that we can detect single electrons using their cyclotron radiation , we will investigate the energy resolution achievable with this type of setup .
the @xmath30kr source is particularly useful for this , since the electrons are monoenergetic .
we plan to use a tunable microwave source to perform power - vs .- frequency calibrations .
finally , we want to demonstrate that the signal from cyclotron radiation can be used to identify electrons and determine their energy without additional detection methods .
though for the initial stages of the prototype the data acquisition is untriggered , we will need to develop the ability to recognize electrons and trigger the recording of data . at the time of this writing ,
the prototype has successfully collected an initial set of data using the rectangular waveguide antenna .
analysis of the data is ongoing , and more data will be collected in the near future .
tritium beta decay experiments are currently the best method for directly probing the neutrino mass .
the katrin experiment has a sensitivity of 200 mev , but it has reached the practical technological limits for spectrometer - type experiments . if we want to examine below the hundred - mev range , we need a new technique . project 8 aims to use radiofrequency techniques to measure the beta - decay electron energy .
a prototype experiment is currently in use to examine the feasibility of the technique , and to provide some initial insight into the possible sensitivity that might be achieved . | it is well known that the neutrino masses affect the shape of the energy spectrum of tritium beta - decay electrons .
however , experiments have yet to measure that distortion .
the project 8 experiment proposes to measure the spectral distortion in a novel way : using radio - frequency techniques to detect and measure the energies of the beta - decay electrons .
we plan on measuring the radiation created from the cyclotron motion of the electrons in a strong magnetic field .
i will report on the status of a prototype that is designed to demonstrate single - electron detection at energies near the tritium endpoint , 18.6 kev .
i will also discuss the possibilities for scaling up to a neutrino - mass experiment . |
neurons in the brain are connected with each other and send short electrical pulses ( action potentials or spikes ) along those connections . despite the fact that there are correlations between the type of connections and the type of neurons @xcite , it is fair to say that neurons fall essentially into two classes , excitatory and inhibitory , and that the connectivity in a local population of several thousand cortical neurons is close to random .
networks with fixed random connectivity can , in principle , contain loops of varying size , which could sustain the flow of transient information signals over times that are long compared to the intrinsic time constants of the network elements , i.e. , the neurons . in neuroscience and related fields , elementary considerations on information flow in random networks
have inspired ideas as diverse as synfire chain activity @xcite , reverberations @xcite , liquid computing @xcite , echo state machines @xcite , or computing at the edge of chaos @xcite .
on the other hand , random networks have also been studied intensively by the physics community , in the context of diluted spin glasses @xcite , formal neural networks @xcite or automata @xcite and limiting cases have been identified for which exact solutions are known . in particular , in the limit of asymmetric networks with low connectivity , mean - field dynamics becomes exact @xcite .
more recently these approaches have been extended to the case of random networks of spiking neurons in continuous time @xcite . in this paper , we will compare simulations of a random network of excitatory and inhibitory neurons with the mean - field solutions valid in the low - connectivity limit and evaluate the performance of such networks on a simple information buffering task that can be seen as a minimal and necessary requirement for more complex computational tasks @xcite which a neural network might have to solve .
more precisely , the task consists in reconstructing a time - dependent input @xmath0 by reading out the activity of the network at a later time @xmath1 .
we will see that performance in the information buffering task is best at the phase transition that is marked by a rapid increase in both the _ macroscopic _ activity variable and the lyapunov exponent characterizing the _ microscopic _ state indicating transition to chaos .
surprisingly , if the same time - dependent input @xmath2 is shared by all neurons in the network , an information readout based only on the macroscopic variable is as good as a readout that is based on the output pulses of all @xmath3 neurons .
however , if the input is only given to a small group of neurons a detailed readout conveys more information than a macroscopic one suggesting that loops in the network connectivity might indeed play a role .
we consider a network of @xmath3 leaky integrate - and - fire units ( neurons ) with fixed random connectivity .
80 percent of the neurons are taken as excitatory and the remaining 20 percent inhibitory .
independent of the network size ( @xmath4 ) , each neuron @xmath5 in our simulation receives input from @xmath6 excitatory and @xmath7 inhibitory units ( presynaptic neurons ) , which are chosen at random amongst the @xmath8 other neurons in the network .
the ensemble of neurons that are presynaptic to neuron @xmath5 is denoted by @xmath9 and the efficacy @xmath10 of a connection from a presynaptic neuron @xmath11 to a postsynaptic neuron @xmath5 is @xmath12 if @xmath13 is excitatory , and @xmath14 if @xmath13 is inhibitory .
each neuron is described by a linear equation combined with a threshold . in the subthreshold regime
the membrane potential follows the differential equation @xmath15 where @xmath16ms is the effective membrane time constant and @xmath17 external input .
the recurrent input @xmath18 neuron @xmath5 receives from the network is @xmath19 where @xmath20 is the time neuron @xmath13 fires its @xmath21th spike and @xmath22ms is a short transmission delay . a spike from an excitatory ( inhibitory )
neuron @xmath11 causes a jump in the membrane potential of neuron @xmath5 of @xmath23mv ( @xmath24mv ) .
if the membrane potential @xmath25 reaches a threshold @xmath26mv , a spike of neuron @xmath5 is recorded and its membrane potential is reset to @xmath27 .
integration restarts after an absolute refractory period of @xmath28ms .
the external input @xmath17 can be separated into two components .
first , we inject a time dependent test signal @xmath29 which is generated as follows : the total simulation time is broken into segments of duration @xmath30ms . during each segment of length @xmath31
the input is kept constant . at the transition to the next segment
, a new value of @xmath32 is drawn from a uniform distribution over the interval @xmath33 $ ] , i.e. the signal distribution has a standard deviation of @xmath34 . by construction the signal at time @xmath35 provides no information about the signal at @xmath36 for @xmath37 .
more precisely , the autocorrelation @xmath38 of the signal has a triangular shape and is strictly zero for @xmath39 .
second , the network of @xmath3 neurons is considered as part of a larger brain structure . to mimic input from excitatory ` background ' neurons that are not modelled explicitly , we assume stochastic spike arrival described by a poisson process of total rate @xmath40 . for the sake of simplicity , we assume that the efficacy @xmath41 of background spikes is identical to that of the recurrent connections within the network .
we approximate the background input @xmath42 by a gaussian white noise with mean @xmath43 and standard deviation @xmath44 , where @xmath40 is the background spike arrival rate .
this approximation is valid under the assumption that a neuron receives a large number of presynaptic contributions per unit time , each generating a change in the membrane potential that is relatively small compared to the firing threshold . to simplify notation we set @xmath45 and @xmath46 .
the values in our simulations are @xmath47 and @xmath48 .
if we replace the signal @xmath49 by gaussian white noise of zero mean and standard deviation @xmath50 a mean - field analysis of the random network defined above can be performed following @xcite .
the macroscopic variable describing the activity of the network is the population rate @xmath51 . in a stationary state of asynchronous neuronal activity ,
the population rate depends on the mean @xmath52 and variance @xmath53 of the total input @xmath54 via the equation @xmath55 the mean - field result is valid in the low connectivity limit @xcite . combining eqs .
( [ input - mean ] ) - ( [ eq2 ] ) , we obtain a self - consistent solution of the population firing rate @xmath51 as a function of the external poisson drive @xmath40 .
the population rate shows a marked increase near @xmath56hz as shown in fig .
[ fig - transition ] which is in the vicinity of a first - order phase transition predicted by the mean - field theory .
if we decrease the amplitude of the input signal @xmath57 the phase transition becomes more pronounced and a regime of coexistence of several solutions appears ( inset ) . as a function of the poisson background drive resulting from eqs .
( [ input - mean]-[eq2 ] ) .
note the sharp transition at @xmath58hz .
the network switches from an almost quiescent state to a state of sustained activity .
dotted line : largest lyapunov exponent as a function of the external drive @xmath40 in a network of 800 neurons . for @xmath59hz
the population rate @xmath51 is low ( quiescent state ) and the largest lyapunov exponent is negative . for a stronger drive
, the exponent switches to a positive value , reflecting the chaotic behaviour of the membrane potential trajectories .
solid line : signal reconstruction error for a delay of 20 ms in a network of 800 neurons .
the error is minimal near the transition from the quiescent to a chaotic regime .
inset : fixed points of the population rate ( eq . [ eq2 ] ) in absence of test signal ( solid line ) , and with increasing signal variance .
the system exhibit a first - order phase transition if the signal is weak . ]
the marked increase in the macroscopic variable @xmath51 is accompanied by a transition of the largest lyapunov exponent from negative to positive values which indicates that the microscopic dynamics becomes chaotic .
the largest lyapunov exponent is defined by @xmath60 where @xmath61 is the difference between a reference trajectory of the @xmath3 variables @xmath62 and test trajectory @xmath63 with slightly different initial conditions . using standard numerical techniques @xcite ,
the largest lyapunov exponent has been estimated from a large number of simulations of a test network and a reference network with identical connectivity .
both networks received identical input ( same realization of the poisson input and signal input ) and the test trajectory was regularly reset close to the reference trajectory .
moreover , we confirmed that not only for background spike input but also for appropriately chosen constant input ( i.e. @xmath64 , @xmath65 ) , the fixed random connectivity is sufficient to generate irregular asynchronous spiking activity @xcite with a positive largest lyapunov exponent ( data not shown ) .
after having characterized the macroscopic and microscopic state of the network , we asked how performance in an information buffering task , inspired by concepts of liquid computing @xcite and echo state machines @xcite , would depend on the network state .
we considered a linear readout unit with dynamics @xmath66 where the sums run over all firing times @xmath67 of all neurons in the network .
@xmath68ms is a short synaptic time constant .
the @xmath69 free parameters @xmath70 ( @xmath71 ) are chosen so as to minimize the signal reconstruction error @xmath72 ^ 2\rangle /\sigma_{\rm sg}^2 $ ] .
parameters were optimized using a first simulation ( learning set ) lasting 100 seconds ( 100000 time steps of simulation ) and were kept fixed afterwards .
the performance measurements reported in this paper are then evaluated on a second simulation of 100 seconds ( test set ) .
simulation results were obtained using the simulation software nest .
the same time - dependent signal @xmath32 was injected into all neurons in the network and the performance evaluated in terms of the signal reconstruction error .
the performance depends on the delay @xmath73 of information buffering which has to be compared with the membrane time constant ( @xmath74ms ) and the autocorrelation of the input @xmath75ms .
overall the signal reconstruction error is relatively high .
as expected , the signal reconstruction error increases if we increase the desired buffer duration from @xmath76ms to @xmath77ms or @xmath78ms ( fig .
[ fig - delays]a ) . as a function of the background firing rate in a network of 800 neurons for three different information buffer delays : @xmath76ms ( solid ) , @xmath77ms ( dashed ) , @xmath78ms ( dotted ) . for sufficiently long delays ,
optimal performance is located near the transition between the quiescent and the chaotic state ; cf fig [ fig - transition ] .
deeper in the chaotic phase the error goes back to the chance level whereas in the almost quiescent regime we can see the effects of overfitting ( @xmath79 ) , because the number of action potentials is insufficient to build an accurate model of the past events .
b. comparison of the errors for different network sizes : @xmath4 neurons ( resp .
dotted , dashed and solid lines ) , for a delay @xmath78ms .
the solid line with filled circles corresponds to the _ macroscopic _ readout of the network of @xmath80 neurons .
the location of minimal error is independent of the number of neurons and coincides with the phase transition .
the vertical shift of the error curves is not significant but due to overfitting because of limited amount of data .
vertical bars indicate the mean difference between errors on the data used for parameter optimisation ( training set ) and that on an independent test set for @xmath80 ( left bar ) , @xmath81 ( 2nd bar ) , @xmath82 ( 3rd bar ) and macroscopic readout ( right bar ) .
a representative curve of errors on the training set for @xmath80 neurons is shown by the dot - dashed line . ] at the same time , the optimal background firing rate @xmath40 to achieve minimal signal reconstruction error shifts towards lower values and is for @xmath78ms very close to the transition between regular and chaotic microscopic dynamics as shown in fig .
[ fig - transition ] .
this result is consistent with the idea of computing at the edge of chaos in cellular automata @xcite or threshold gates @xcite . also ,
similar to the results in discrete - time spin networks @xcite , the information buffering performance does not significantly depend on the number @xmath3 of neurons in the network ; cf .
[ fig - delays]b .
differences are within the statistical variations caused by overfitting on finite data samples .
given that networks states have been classified successfully by macroscopic mean - field equations @xcite , we wondered whether the performance in the above information buffering task can be completely understood in terms of _ macroscopic _ quantities . to answer this question
, we compared the performance using the previous readout unit @xmath83 ( i.e. , a ` microscopic ' readout with @xmath69 free parameters , one per neuron plus an offset ) with that by a simplified readout @xmath84 with two free parameters @xmath85 and @xmath86 only , @xmath87 , i.e. , a readout that uses only the macroscopic population activity .
surprisingly , for a stimulation paradigm where all neurons receive the same time - dependent signal @xmath32 the macroscopic readout performs as well as the microscopic one . in other words ,
connectivity loops between _ specific _ subsets of neurons where signals could circulate for some time seem not to play a role in information buffering .
this suggests that , for signals of sufficiently small amplitude , the information buffering capacity is directly related to the _ macroscopic _ linear response kernel @xmath88 of the network activity , that can , in principle , be calculated from the linearized mean - field equations of the population rate , i.e. , @xmath89 ; cf .
the time constant of the kernel , and hence information buffering delays , become large in the vicinity of a phase transition .
we hypothesized that signal transmission loops in our randomly connected network could manifest themselves more easily if only a small subset of neurons received the input signal .
we therefore selected 20 percent of neurons at random ( group @xmath90 ) and injected an identical time dependent signal @xmath91 into all neurons @xmath92 .
the remaining 80 percent of neurons ( group @xmath93 ) received no signal . in such a network consisting of two groups , signal buffering performance is indeed significantly better than in a single homogeneous group ( fig .
[ fig - twogroups ] ) . on a macroscopic scale
, a network consisting of two groups @xmath90 and @xmath93 can be described by two macroscopic variables , i.e. , the population activities in groups @xmath90 and @xmath93 . in order to evaluate the information contained in the macroscopic population rates
, we used a linear readout unit @xmath94 with three free parameters @xmath95 , @xmath96 and @xmath97 , characterized by the differential equation @xmath98 and proceeded as before . as we can see from fig .
[ fig - twogroups ] , a readout based on the macroscopic activity of the two groups performs significantly worse than the microscopic readout .
this suggests , that for the case when only a small subset of units in a random network receive an input , signal transmission loops , and hence microscopic neuronal dynamics , indeed play a role in short - term information buffering . of the neurons only . a macroscopic readout assuming a single population ( dotted line ) performs well near the phase transition .
however deeper in the chaotic phase it is outperformed by the _ microscopic _ readout ( solid line ) .
a macroscopic readout based on a two - population assumption ( dashed line ) explains only part of the increased performance .
the signal buffering delay for this figure is @xmath78ms . ]
mean - field methods neglect correlations in the input . in random networks
mean - field theory becomes asymptotically correct only in the low - connectivity limit where the probability of closed loops tends to zero @xcite .
however , it is exactly these loops which could give the network the power to buffer information for times significantly longer than the intrinsic time constants of the network elements .
our network is formally not in the low - connectivity limit since the number of neurons @xmath80 is small .
nevertheless , we found that mean - field results can qualitatively predict the rough location of the phase transition of the macroscopic population rate .
moreover , if the input signal is shared between all neurons , a macroscopic readout is sufficient to explain the network performance in an information buffering task .
microscopic properties do , however , play a role if the input is only given to a subset of the network units suggesting that in this case ultra - short term information buffering in connectivity loops is indeed possible . in additional simulations we checked that the maximum delay @xmath73 for which signal reconstruction is feasible is only in the range of 20 - 50ms and hence not significantly different from the intrinsic neuronal time constants .
this suggests that , without slow processes such as synaptic plasticity or neuronal adaptation a purely random network of spiking neurons is not suitable as an information buffer beyond tens of milliseconds . | in randomly connected networks of pulse - coupled elements a time - dependent input signal can be buffered over a short time .
we studied the signal buffering properties in simulated networks as a function of the networks state , characterized by both the lyapunov exponent of the microscopic dynamics and the macroscopic activity derived from mean - field theory .
if all network elements receive the same signal , signal buffering over delays comparable to the intrinsic time constant of the network elements can be explained by macroscopic properties and works best at the phase transition to chaos .
however , if only 20 percent of the network units receive a common time - dependent signal , signal buffering properties improve and can no longer be attributed to the macroscopic dynamics . |
it was recognized already in the sixties that the proton - neutron pairing may play a non - negligible role in the nuclei with equal numbers of protons and neutrons due to the particulary large spatial overlaps of the wave functions of both types of nucleons .
one of the first methods developed to take into account such correlations were bcs - type approaches.@xcite it is however known that the bcs wave function violates the particle number conservation which is especially harmful in the cases of weak pairing and phase transitions . a generic feature of the results of bcs calculations generalized so as to accomodate neutron - proton correlations ( _ i.e. _ that take into account isoscalar and isovector pairs ) in self - conjugate nuclei is a sharp transition from like - particle to @xmath1 pairing modes .
such a scenario is absent in the particle number conserving approaches which allow for the coexistence of different superfluid phases.@xcite in the present study we propose an approach which is free of the deficiencies of the treatments la bogoliubov .
this method , called higher tamm - dancoff approximation ( htda ) , was already applied to study ground and isomeric states in heavier nuclei and to describe pairing correlations in high spin states.@xcite here we extend the htda formalism to take into account the proton - neutron pairing .
the purpose of the method is to describe the pair diffusion phenomenon around the fermi surface as a particular , yet very important , part of some many quasiparticle excitations of the particle - hole type over a slater determinant vacuum @xmath2 .
thus one can consider dealing with a kind of higher tamm - dancoff approximation or equivalently with a highly truncated shell - model calculation .
the latter point of view makes it clear that the success ( _ i.e. , _ the fast convergence in terms of the number of considered quasiparticles ) of such a truncation scheme will depend on the realistic character of the vacuum in use .
such a favorable situation may be expected when the vacuum is defined self - consistently from the mean - field carrying most of the single - particle ( sp ) properties ( even those yielded by the correlations ) associated with a given effective hamiltonian @xmath3 where @xmath4 is the kinetic energy and @xmath5 an effective interaction .
it is advisable to choose for the wave function @xmath2 the hartree - fock solution coming from the hf potential @xmath6 self - consistently obtained from the many - body reduced density matrix @xmath7 of the correlated solution for the desired number of particles and possibly taking into account various constraints ( _ e.g. _ , deformation ) and symmetries ( _ e.g. _ , time - reversal symmetry ) .
we have thus @xmath8 where @xmath9 @xmath6 being the one - body reduction of @xmath5 for @xmath2 .
the @xmath2 vacuum may now serve to construct an orthonormal n - body basis in which we will diagonalize hamiltonian ( [ htot ] ) . in principle to build this basis we should consider the slater determinant corresponding to the 0p0h state and all the particle - hole excitations : 1p1h @xmath10 , 2p2h @xmath11 , 3p3h @xmath12 and so on .
it is however obvious that in practice we need to truncate the space of particle - hole excitations .
based on the former studies in the htda framework@xcite we may assume that the components of the pair excitation type will be dominant in the ground state . in the section [ results ]
we will show for the case of @xmath13 and @xmath14 correlations that the other components are indeed of a minor importance .
the total ground state wave function can be decomposed in the following way ( @xmath15 and @xmath16 denoting two different charge states ) @xmath17 the ensemble of products of slater determinants @xmath18 represents a complete orthogonal basis of the n neutrons and z protons physical space with real coefficients @xmath19 fulfilling the relation @xmath20 which guarantees the normalization of the function ( [ mbstpn ] ) .
it is clear that this function has a good particle number @xmath21 . to obtain the correlated ground state one should perform the diagonalization of the hamiltonian ( [ htot ] ) in the given many - body basis .
let us rewrite the hamiltonian ( [ htot ] ) in the following form @xmath22 where the independent quasiparticle hamiltonian @xmath23 reads @xmath24 where @xmath25 is the particle creation operator @xmath26 for @xmath27 being a particle ( unoccupied ) state and to the anihhilation operator @xmath28 in the case of hole ( occupied ) states .
for @xmath29 we have : @xmath30 or @xmath31 , with @xmath32 indicating the single - particle ( hf ) energy corresponding to the particle or hole case , respectively . as seen from eq .
( [ h2 ] ) , the residual interaction reads @xmath33 the matrix element of the above given hamiltonian in the multi - particle multi - hole basis takes the form @xmath34 where @xmath35 is the total particle - hole excitation energy calculated with respect to the vacuum @xmath36 .
the residual interaction comprising only one- and two - body operators will merely contribute to the matrix elements in cases where the two many - body states are identical or differ by one or two particles .
the many - body matrix elements of @xmath37 can be evaluated with the use of the wick theorem ( for details see ref.@xcite ) and expressed as sums of proper two - body matrix elements of the interaction @xmath5 . in actual calculations with our approach
we have used the skyrme force for the hf part of the problem . since the majority of the skyrme forces is not well suited to reproduce the data in the particle - particle channel we apply a @xmath38-force to define the residual interaction .
using the zero - range force is well established in the context of like - particle and to a lesser extent of the proton - neutron pairing . to assess the validity of our approach at this early stage of the investigation of the proton - neutron correlations within the htda framework a substitution @xmath39 is therefore done for the residual interaction .
the operators @xmath40 project onto spin - isospin subspaces @xmath41 where @xmath42 , @xmath43 are the standard spin and isospin exchange operators .
the main challenge in the exact pairing diagonalizations methods and shell - model like calculations is the large dimension of the matrix to be diagonalized . in the present study we use the lanczs algorithm@xcite within the code developed by parlett and scott@xcite which allows to search for the lowest energy solutions in a reasonable time .
however , as will be shown in the following , the advantage of the htda method is that the amount of the many - body configurations within a given single - particle space that need to be taken into account is quite limited and the matrix diagonalization problem is far from attaining the difficulty level of typical shell - model calculations .
the calculations within the htda framework which we will discuss here are performed for the ground state of the @xmath0ge nucleus as obtained in a skyrme - hf+bcs approach .
we use the siii force for the particle - hole channel , seniority force in the bcs part and volume @xmath38 interaction to calculate two - body matrix elements in the htda part .
the htda results presented here are non self - consistent .
indeed , only one diagonalization of the htda hamiltonian matrix has been carried out for a hamiltonian @xmath44 obtained with the density @xmath7 corresponding to the hf+bcs calculation . in the limiting case of vanishing proton - neutron correlations we obtain two decoupled problems , one for each charge state .
let us discuss first the aspects of the basis truncation , both in single - particle and many - particle many - hole excitations terms .
if the assumptions of the bcs theory are to hold , we may expect that the most important part of the ground state correlations will correspond to the pair excitations , _
i.e. _ the situation where two particles in kramers degenerate hole orbitals are promoted to two kramers degenerate particle levels . in what follows
we will restrict the many - body states first to take into account all possible 1p1h and 2p2h excitations including thus the one pair excitations in addition to the particle - hole vacuum .
in further calculations only the states corresponding to one , two and three pairs excitations will be considered .
.numbers of many - body configurations of different types depending on the number of sp levels in the window . [ cols="^,^,^,^,^,^",options="header " , ] as we already stated we aim at including the @xmath46 pairing correlations within the htda framework . as above
discussed in this preliminary study we limit the configuration space to the so - called one pair excitations in all channels .
( 0p0h)@xmath47 , pn-(1p1h)@xmath48(1p1h)@xmath47 , pp-(0p0h)@xmath49(2p2h)@xmath50 , nn-(2p2h)@xmath49(0p0h)@xmath47 ) in the correlated wave function @xmath51 versus the ratio @xmath52 of the @xmath38-force strengths in isoscalar @xmath53 and isovector @xmath54 channels . ]
since proton - neutron pairs may exist in both @xmath1 and @xmath45 channels the problem of adjusting the strength of the isoscalar pairing arises here .
the interplay of @xmath1 and @xmath45 pairing being rather unclear , we will adopt the same value of the isovector pairing strength for @xmath13 , @xmath14 and @xmath55 channels and will treat the @xmath1 interaction strength as a free parameter
. pairing correlations can be as well obtained in an isospin broken model with @xmath56 , nevertheless we do not see any argument supporting such a choice . ] of the interaction strengths in isoscalar and isovector channels . the results concern calculations with @xmath1 @xmath55 pairs only and with all possible pairs taken into account . ] within a given configuration space and @xmath45 strength adjusted as described in the precedent we perform the calculations as a function of the increasing @xmath1 coupling constant . in fig .
[ fig-1 ] the evolution of the probability of different components is plotted against the ratio @xmath57 the probability of the @xmath58 pair excitation becomes siginificant with respect to that of @xmath13 or @xmath14 pair excitation for @xmath59 which corresponds to the ratio value at which a @xmath55 superfluid solution is found in bcs(@xmath38)-type of calculations in this nucleus.@xcite the gain in energy due to the @xmath55 coupling is non - negligible which can be seen from fig .
[ fig-2 ] where the contribution of different components to the correlation energy is depicted .
it is worth noting that the increase of the energy is of a similar magnitude like that of bcs calculations with the approximate particle number projection of the lipkin - nogami ( ln)@xcite nonetheless a certain gain in the htda correlation energy due to the @xmath55 coupling occurs already in the range of parameters for which no @xmath55 pairing solution is found in the bcs+ln method ( @xmath60 ) .
we have presented a method explicitely conserving particle number in the context of describing isovector and isoscalar pairing correlations .
the inclusion of @xmath1 pairs leads to a constant increase of the absolute value of the correlation energy . the expected benefit of the htda framework in yielding a much smoother decrease of correlations than when using a bogoliubov quasiparticle vacuum in weak pairing regimes is confirmed .
one of the authors ( k.s . ) is indebted to jacek dobaczewski for fruitful discussions .
this work is partly supported by the committee of scientific research ( kbn ) of the polish ministry of science and education under contract no .
1p03b13028 ( 2005 - 2006 ) and the french embassy in poland . | in the present work the so - called higher tamm - dancoff apporximation method is presented for the generalized case of isovector and isoscalar residual interactions treated simultaneously .
the role of different particle - hole excitations and of proton - neutron pairing correlations in the ground state of the self - conjugate @xmath0ge nucleus is discussed . + * pacs : 21.10.hw , 21.60.jz , 21.60.cs * |
there are @xmath4 10 low mass x - ray binaries ( lmxb ) that exhibit absorption dips in x - ray intensity at the orbital period , and it is generally accepted that these are due to absorption in the bulge in the outer accretion disk where the accretion flow from the companion impacts ( white and swank 1982 ) .
xbt0748 - 676 is an important member of the dipping class , discovered and observed extensively with exosat , exhibiting dipping , bursting and eclipsing ( parmar et al .
since then , it has undergone strong variations in brightness .
the _ exosat _
observations showed that both dips and interdips were present , and that the depth and extent of dipping varied markedly from orbit to orbit . eclipses showed that the inclination was high , and during these , about 4% of the non - dip emission in the band 2 - 6 kev remained , indicating the presence of an extended emission component . using the eclipses as fiducial markers , parmar et al .
( 1986 , 1991 ) found a very accurate value for the orbital period and a decreasing period .
asai et al .
( 1992 ) and corbet et al .
( 1994 ) showed evidence that the period is modulated sunusoidally , but this pattern was not continued when rxte data was added ( hertz et al .
however all determinations agree on the period to @xmath4 0.01 s , ie 13766.78 s. the dipping sources do not , in general , have the spectral evolution expected for photoelectric absorption of a single emission component in the bulge in the outer accretion disk , ie a strong hardening of the spectrum .
for example , x1624 - 490 shows a softening of the spectrun ( church and balucinska - church 1995 ) , and x1755 - 338 shows energy independence ( white et al .
1984 ; church and balucinska - church 1993 ) .
we have previously proposed a physical model for the dipping sources which explains the spectral evolution in these two sources ( church and balucinska - church , 1995 ) . in this model ,
emission consists of two components : blackbody emission from the neutron star plus extended comptonised emission from the accretion disk corona ( adc ) represented as a power law at energies well below the comptonisation break .
x1624 - 490 is also important because dipping reaches a `` saturated '' lower level which is strong evidence that two emission components are present ( church and balucinska - church 1995 ) . the differences between the two sources above are mainly due to differences in blackbody temperature . in x1624 - 490 , @xmath0 is relatively high , and as dipping consists predominantly of absorption of the point - like blackbody emission , this leaves the residual spectrum softer . in x1755 - 338 , @xmath0 is smaller , with the peak emission at @xmath4 3 kev , so that removal of the blackbody leaves the spectrum neither harder nor softer in the band 1 - 10 kev .
there is also an important group of dipping sources including xb0748 - 676 , xb1916 - 053 and xb1254 - 690 which have very similar spectral evolution in dipping , in that there is clearly part of the spectrum which is not absorbed .
these sources also , do not fit a simple pattern of photoelectric absorption in the bulge in the outer accretion disk .
spectral evolution in dipping in these sources has been modelled using the `` absorbed plus unabsorbed '' approach ( parmar et al .
1986 , smale et al .
1988 and courvoisier et al .
1986 ) . in the case of xbt0748 - 676 ,
the non - dip spectrum was fitted with a cut - off power law model ( parmar et al .
dip data selected in intensity bands however , revealed a strong low energy excess , and the spectra could be fitted by dividing the non - dip model into two parts with the same form , one strongly absorbed and the other not absorbed .
good fits were obtained with the column density of the unabsorbed part fixed at the non - dip value and but with strongly decreasing normalisation as dipping progresses , whereas the absorbed component had increasing column density but with a normalisation remaining at about the non - dip value .
the origin of the unabsorbed component has often in this group of sources been attributed to electron scattering producing an energy - independent decrease of intensity , although it is not clear that electron scattering should play an important role in the bulge in the outer accretion disk in competition with photoelectric absorption . in the case of xbt0748 - 676 ,
parmar et al . (
1986 ) concluded that the unabsorbed component was probably due to fast variations of column density which could mimic a soft excess . in the case of xb1916 - 053
, we have shown that the asca spectra may be fitted well by our two - component model including progressive covering of the extended comptonised emission ( church et al .
the blackbody emission from the neutron star was covered rapidly .
very good fits were obtained in this way , with large increases in column density @xmath5 for the blackbody and smaller increases in @xmath5 for the comptonised emission , which are averages across the absorber .
the unabsorbed peak in the spectrum is the uncovered part of the power law component , and the increase in the partial covering fraction from zero to unity in deepest dipping is the behavior represented by the decreasing normalisation in `` absorbed + unabsorbed '' modelling .
this is however radically different from that type of modelling .
firstly , we are explicitly using partial covering of extended emission in the model which has not previously been done .
secondly , the emission model for the source has two terms , and is the same model that we have shown gives good explanations of other dipping sources .
dipping can be modelled well by photoelectric absorption alone , without the need for electron scattering , and calculations from the cross sections showed that scattering is not expected to be important in the absorbing bulge in competition with absorption below 10 kev ( church et al .
the main aim of the present work was to test using the asca data whether this approach can also successfully describe spectral evolution in dipping in xbt0748 - 676 .
we present results for the observation of xbt0748 - 676 made on 1993 , may 7th with asca ( tanaka et al .
1994 ) , which lasted 23 hours , during the performance verification phase .
the gis data were screened to remove regions of saa passage , to restrict elevation above the rim of the earth to more than @xmath6 , particle rigidity to more than 6 gev / c , the radiation belt parameter to less than 200 c / s and angular deviation of the telescope pointing to less than @xmath7 .
the calibration source and outer ring were removed from the image , and rise - time rejection applied .
source data were selected from a circle of radius 6 centered on the source .
background data were also obtained from a circular region of radius 6 which provided a sufficiently large background count to use for sensible subtraction from the spectra .
sis0 data were extracted similarly and ` faint ' data converted to ` bright ' mode data .
the data were screened , and cleaned to remove the effects of hot pixels .
the main spectral fitting results were however obtained by adding gis2 and gis3 spectra to improve the quality of the spectra .
the gis2 light curve in the band 0.7 - 10.0 kev is shown in fig . 1 ,
in which 4 bursts and 2 eclipses can be seen .
it is also evident that dipping reached a depth of 100% , for example at @xmath4 60,000 s. the source was faint during the _ asca _ observation at @xmath4 1 c / s ; this would be about 1 c / s in the _ rosat pspc _ , compared with @xmath4 30 c / s obtained in the _ pspc _ during the observation of 1991 nov 11 , for example ( hertz et al .
thus the intensity of the source had decreased by about a factor of 30 .
the orbital period was obtained by carefully measuring the times of eclipse ingress and egress from the gis2 light curve with 4 s binning .
differences in time between eclipse centers gave an orbital period p = @xmath8 s , consistent with the accepted value of @xmath4 13766.78 s. however , in producing our folded light curves , we used the more accurate value .
the bursts , but not the eclipses , were excluded from the data before folding
. of the 6 orbital cycles observed , only 2 were covered well .
when all of the data is included , the folded light curve is distorted from the shape of individual dips ; for example , the shape of the dips is affected by the variability in the width of dipping .
we therefore show in fig .
2 folded light curves using orbits 3 and 5 which are not significantly distorted and which can be seen to be significantly different from the folded light curve including all data ( lower panel ) .
note that in this figure , the dip between phases 0.6 and 0.7 is only part of the main dipping activity which in this source generally extends within an envelope between phases 0.5 and 1.1 . in the lowest energy band 1 - 3 kev , dipping extends to almost all of the orbital cycle , suggesting the presence of absorber spreading around the outer accretion disk .
the dipping at phase 0.6 - 0.7 is 100% deep and relatively steep - sided compared with dipping in the band 3 - 10 kev which is only 80% deep .
thus there is a marked increase in hardness in this source , which contrasts with xb1916 - 053 during the _ asca _ observation , in which the depth of dipping was @xmath4 100% at all energies in the band 1 - 10 kev ( church et al . 1997 ) .
the eclipse in the folded light curve is clear in the center panel in the band 3 10 kev , but is less obvious in the band 1 3 kev because dipping is in progress when the eclipse starts .
gis spectra were derived using the complete observation of 6 orbital cycles as there was little evidence for dependence of the results obtained on whether all 6 orbits were used , or the first 3 , or the last 3 .
firstly , gis2 spectra were accumulated by removing bursts and eclipses and dividing the data into intensity bins using the ranges : 2.4 - 3.2 c / s ( non - dip ) , 1.8 - 2.4 c / s , 1.2 - 1.8 c / s , 0.6 - 1.2 and 0.0 - 0.6 c / s . a consequence of dipping extending to several parts of the orbital cycle was that little non - dip data was available , and care was necessary to identify the correct intensity band for non - dip to avoid including dip data .
the folded light curve in the higher band 3 - 10 kev has a shape more typical of the dipping sources and so was used to determine which data was non - dip .
gis2 and gis3 data could not be selected in the same intenity bands because of count rate differences due to their different offsets from the bore - sight axis , and so gis3 data were selected via time filters corresponding to intensity selections in gis2 .
the gis2 and gis3 spectra were added , systematic errors of 2% were applied conservatively to each channel , and the data regrouped to an appropriate minimum number of counts in each bin .
channels below 0.8 kev and above 10.0 kev were ignored .
an average response function was derived from the gis2 and gis3 responses appropriate to the source position in each detector . finally the gis2 and gis3 background spectra were added .
the non - dip data is of relatively poor quality because of the predominance of dipping during the orbital cycle . in our work on other dipping sources , it was possible to derive well - constrained values of @xmath0 and @xmath2 for our two - component model from the non - dip data , and to fix these in fitting dip data . in this case , we have carried out simultaneous fitting of non - dip data and the 4 dip spectra .
the non - dip spectrum could be fitted by a simple absorbed power law adequately , although this model will not fit the dip spectra because of the absorbed and unabsorbed peaks in the spectrum .
simple thermal models are not capable of fitting the spectra .
thus we next used the two - component model which we have proposed for the dipping sources in general , consisting of a blackbody component from the neutron star plus comptonised emission from the adc with progressive covering of the adc : @xmath9@xmath10 where bb and pl are the blackbody and power law terms , ag is a galactic absorption term , ab are variable absorption terms ( including the galactic term for the blackbody ) , and _
f _ is the covering fraction .
the source emission parameters can not change during dipping and so we require @xmath0 , @xmath2 and the two normalisations to remain constant during dipping .
good fits to the 5 spectra simultaneously were obtained , with @xmath11 for the simultaneous fitting of 312 for 517 degrees of freedom , suggesting that the systematic errors added were pessimistic .
we have also carried out spectral fitting of the sis0 data in the energy band 0.5 - 10.0 kev using the same spectral model .
the model fitted the non - dip and dip data well at all energies above 1 kev , but at @xmath4 0.7 kev , the soft excess reported by thomas et al .
( 1997 ) was seen .
first , we added an absorbed gaussian line to the spectral model above , but found that this could only fit the spectra simultaneously with a decreasing normalisation .
this is exactly the behavior in the continuum emission modelled in the absorbed + unabsorbed approach by a decreasing normalisation , but in our fitting by the partial covering term .
consequently we next fitted a model in which a gaussian line was added to the power law term , thus applying the same partial covering fraction to the line as to the continuum .
this model was clearly able to fit the intensity variation of the excess in dipping adequately .
we further found that no improvement in the fit was obtained by allowing the line normalisation to vary in addition to the effect of the partial covering term .
this strongly suggests that the excess originates in the same region as the comptonised emission , ie in the adc .
the best fit energy from fitting was 0.65 kev , which is the energy of o viii l@xmath12 , although there are several lines within @xmath4 0.1 kev of this energy .
= 90 mm our best fit spectral fitting results from simultaneous fitting the 5 added gis2 and gis3 spectra gave @xmath0 = 1.99 @xmath1 0.16 kev and power law photon index @xmath2 = 1.70 @xmath1 0.16 .
the value of the galactic column density was @xmath13 h atom @xmath14 , in reasonable agreement with the stark et al .
value ( 1992 ) of @xmath15 h atom @xmath14 , and in good agreement with the value of parmar et al .
( 1986 ) of @xmath16 h atom @xmath14 .
the best fit to non - dip and dip data is show in fig . 3 and table 1 .
llrll @xmath17 & @xmath18 & @xmath19 & f + 2.4 - 3.2 & @xmath4 0.2 & [email protected] & 0.0 + 1.8 - 2.4 & @xmath4 0.2 & [email protected] & [email protected] + 1.2 - 1.8 & [email protected] & [email protected] & [email protected] + 0.6 - 1.2 & [email protected] & [email protected] & [email protected] + 0.0 - 0.6 & 79@xmath116 & [email protected] & [email protected] + @xmath21 intensity in c / s ; @xmath22 in units of @xmath23 h atom @xmath14 from table 1 and fig .
3 , it can be seen that the spectral evolution in dipping was fitted by large increases in @xmath5 for the blackbody , togther with smaller increases in @xmath5 for the power law as this was progressively covered by absorber .
the blackbody is expected to measure @xmath5 along a track across the absorber towards its center where the density is high , whereas the smaller @xmath5 of the power law suggests that @xmath5 is being integrated across the density gradient in the absorber . in
deepest dipping for which we have a spectrum , the blackbody is completely absorbed and the partial covering fraction has risen to 95% .
however it is clear that the unabsorbed peak is the 5% of incident power law emission not covered by the absorber whereas the higher energy peak is the residual of the power law covered by absorber after photoelectric absorption .
it is interesting that the blackbody in the first dip spectrum persistently required an @xmath5 of zero in addition to the galactic contribution .
this supports the idea that @xmath4 50% of the extended emission region is covered by absorber before blackbody emission from the neutron star begins to be absorbed , as would be expected .
the two - component model which we have previously proposed for the dipping sources has been shown to fit the source xbt0748 - 676 .
in addition , we have shown that the spectral evolution in dipping can be explained in the same way as we did for xb1916 - 053 , by allowing the extended comptonised component to be progressively covered by the absorber . in a separate analysis of the pv phase data for xbt0748 - 676 , thomas et al .
( 1997 ) use the absorbed + unabsorbed approach for dip data , and find for non - dip emission that a single component generalised thermal model is ` preferred from selected models ' ; however they do not report results using our two - component model , so their results should be seen in this light .
our spectral fitting results give a total absorbed flux of @xmath24 erg @xmath14 @xmath25 in the band 1 - 10 kev , of which the blackbody contribution is 29.9% .
the luminosity of the blackbody emission calculated from the unabsorbed flux in the band 0.01 to 50 kev , and assuming a distance of 10 kpc ( gottwald et al .
1986 ) is @xmath26 erg @xmath25 , and from this , we can calculate the height of the emitting region assuming it is an equatorial strip around the star .
this is found to be very small : 0.09 km , indicative of a very thin accretion disk , consistent with the fading of the source by a factor of about 30 from the _ rosat _ observation of 1991 nov 11 . in the case
that the magnetic field strength of the neutron star were non - zero , the emitting area at the magnetic poles would still be very small : 5.8 @xmath27 .
the duration of eclipse or dip ingress and egress can in principle be used to derive the size of the extended source region , ie the adc . however , in the present observation the source was faint , much fainter than during the _ exosat _ observations , and we have only two dips that are well covered and 2 eclipses , compared with the 9 eclipses used by parmar et al .
( 1986 ) to obtain the duration of eclipse ingress .
the count rate of the source in _ rxte _ was also much higher than in _
asca _ ( hertz et al . 1997 ) . in these circumstances
we can only estimate the size of the adc . using individual dips ,
we get an approximate duration of ingress @xmath28 of about 110 s. in the present case , the angular size of the absorber is known to be greater than that of the source because dipping is 100% deep and so the duration of dip ingress is the time taken for the bulge in the outer disk to cross the diameter of the adc .
we estimate the radius of the disk as @xmath29 cm , and the velocity of the bulge is @xmath30 , where p is the orbital period . from these
we derive the radius of the adc to be @xmath31 cm . from the 2
eclipses , we find that the duration of ingress is between 16 s and 64 s. combining these with a binary separation of the stars of @xmath32 cm , we get a radius of the adc of between @xmath33 cm and @xmath34 cm .
this can be compared with the results from _ exosat _ as follows .
fig . 3 of parmar et al .
( 1986 ) shows the profile of 9 eclipses added by folding on the orbital period , having a steep transition to eclipse and also shoulders to the transition .
the total duration of ingress can be seen to be about 30 s , equivalent to a @xmath35 of @xmath36 cm .
our eclipse ingress times are consistent with their value .
thus our dip and eclipse data suggest a value of @xmath35 between @xmath33 cm and @xmath37 cm , consistent with an extended region above the inner part of the accretion disk .
the spectrum of xbt0748 - 676 is clearly dominated by non - thermal emission which is identified with comptonised emission of the adc , and we can derive some information about the adc from our results .
firstly , the value of the comptonisation y - parameter is found to be @xmath38 1.5 from our power law index of 1.7 ( eg rybicki and lightman 1979 ) .
there is no evidence ( or expectation ) for the comptonisation break being less than 10 kev ; however we can assume that the break energy is similar to that we have determined for the related source xb1916 - 053 using the very broad band of 0.1 - 300 kev of bepposax ( church et al .
1998 ) , ie @xmath4 80 kev . from this value ,
the electron temperature @xmath39 is @xmath4 30 kev , from which we derive values in xbt0748 - 676 of the average optical depth and column density in the adc of 2.5 and @xmath40 @xmath14 .
we have also shown that the low energy feature can be fitted by a gaussian line at 0.65 kev in non - dip and dip spectra , and in addition have shown that it originates in the adc , since it can be fitted by the same progressive factor as the extended continuum emission .
future identification of this line , perhaps as o viii l@xmath12 will , of course , provide direct information on conditions in the adc .
high energy data is clearly required for xbt0748 - 676 , together with higher resolution data below 1 kev to reveal the nature of the line emission .
the explanation of spectral evolution in xbt0748 - 676 is radically different from previously modelling of the source using the absorbed + unabsorbed approach .
previously partial covering has not explicitly been used in modelling , and fitting our two - component model including partial covering of the extended component shows that the unabsorbed emission can be identified with uncovered non - dip emission .
gradual covering is represented in absorbed + unsabsorbed modelling by the decreasing normalisation often attributed to electron scattering .
we do not need to invoke electron scattering in the absorber , so that dipping can be explained by photoelectric absorption alone . in the case of xb1916 - 053 we presented calculations showing that electron scattering is not expected to be important in the bulge in the outer disk at energies below 10 kev ( church et al .
1997 ) . in xbt0748 - 676 ,
the density increase in dipping is less , and so the effects of scattering will be even smaller . | we report results of analysis of the asca observation of 1993 , may 7th of the dipping lmxb source xbt0748 - 676 , and propose a new explanation of the spectral evolution in dipping in this source .
the behavior of the source was very unusual in that , in the band 1 - 3 kev , dipping extended around most of the orbital cycle with almost no non - dip intensity evident and the depth of dipping reached 100% . at higher energies ,
eg 3 - 10 kev , the depth of dipping was less than 100% , and there were marked increases in hardness in dipping . we show that the non - dip and dip spectra in several intensity bands are well fitted using the same physical model that we have previously shown gives good explanations of several dipping sources , consisting of point source blackbody emission from the neutron star , plus extended comptonised emission from the accretion disk corona ( adc ) , with progressive covering of the adc during dipping .
best fit values of @xmath0 = [email protected] kev and power law photon index @xmath2 = [email protected] are found .
the strong excess below 1 kev was well fitted by a gaussian line at 0.65 kev . in dipping ,
good fits were obtained by allowing it to be covered by the same progressive covering factor as the extended continuum emission , providing strong evidence that the line originates in the adc .
our approach of applying the two - component model , and explicitly including progressive covering of the comptonised emission , differs radically from the `` absorbed plus unabsorbed '' approach previously used extensively for xbt0748 - 676 and similar sources , in which the normalisation of the unabsorbed peak in dip spectra is allowed to decrease by a large factor in dipping .
this decrease has often been attributed to the effects of electron scattering . by using our two - component model
we show that the unabsorbed component is the uncovered fraction of the comptonised emission , and in the band 1 - 10 kev , we do not need to invoke electron scattering to explain dipping
. 0.5 cm |
at the end of 1944 , shortly before his death in august 1945 , stefan banach regularly met with andrzej alexiewicz , then a fresh phd from the ( underground at that time ) lvov university . during these meetings
, they discussed a lot of mathematics , and alexiewicz kept a diary whose mathematical part , including questions posed by banach , was published posthumously by alexiewicz s son in @xcite .
the last entry in this diary , dated december 29 , 1944 , reads : in what follows , the claim and the question from this passage will be referred as `` banach s claim '' and `` banach s question '' .
this seems to be the last question of banach which appeared in the `` literature '' ( broadly interpreted ) .
it is interesting that among all other questions posed by banach in his papers , in the scottish book @xcite , in other diary entries of @xcite , in the problem sections of _ fundamenta mathematicae _ and _ colloquium mathematicum _ ( the latter being entered posthumously ) this is the only one which does not belong to the field of analysis .
( the only possible exception is question 47 from @xcite about permutations of infinite matrices , which , incidentally , also deals with ( im)possibility of building a certain class of maps from `` simpler '' ones ) . on the other hand ,
that banach was interested in such sort of questions
belonging , somewhat vaguely , to a crossroad of universal algebra , discrete mathematics , logic , and combinatorics is , perhaps , not accidental at all , as such a crossroad , along the functional analysis , the main banach s occupation , was another `` polish speciality '' at that time ( and long thereafter ) .
it is the purpose of this note to discuss ( and answer ) possible interpretations of this question . for the historical context
we refer the reader to a variety of excellent sources : to @xcite for the last years of banach in lvov during two russian and one german occupations ; to duda @xcite , turowicz @xcite , and to introductory chapters of @xcite for the general atmosphere of mathematical lvov ; and to murawski @xcite and woleski @xcite , @xcite for a unique interaction of mathematics and logic in inter - war poland .
there is at least another mention of a variant of banach s question in the literature , namely by jerzy o , a student at the lvov university during 19371939 , in @xcite . there , also with reference to alexiewicz
, o writes : here by a ternary semigroup one means a set @xmath0 with a ternary map @xmath1 subject to a ternary variant of associativity : @xmath2 for any @xmath3 , and by a reducible ternary semigroup one means a ternary semigroup @xmath4 with multiplication given by @xmath5 where @xmath6 is an ordinary ( i.e. , binary ) semigroup structure on the same underlying set @xmath0 .
o answers the question in affirmative as a consequence of his general results about extensibility of first - order logical models .
later an alternative and more constructive proof was given by monk and sioson ( * ? ? ?
* theorems 1 and 2 ) .
the latter paper contains also examples of ternary ( in fact , @xmath7-ary for any positive integer @xmath7 ) semigroups not representable in the form ( [ eq - star ] ) , thus validating this variant of banach s claim .
however , this variant of banach s question is more narrow in scope than those presented in @xcite . can the latter question be interpreted in a different way ? in the absence of additional qualifications , the most general reading is the following : `` multiplication '' on a set @xmath0 means an arbitrary ternary map ( [ eq - f ] ) , without associativity , or any other , for that matter , constraint .
but what does `` generation '' mean ?
virtually all banach s works are devoted to ( proper ) analysis and/or measure theory ( including the famous banach tarski paradox ) , with , perhaps , only two exceptions : a joint unpublished work with stanisaw mazur on computable analysis ( a constructive approach to analysis ) , and a short paper @xcite , which touches , seemingly , a similar question . there
, banach gives a shorter proof of an earlier result of wacaw sierpiski @xcite : [ th - un ] for any countable set @xmath8 of unary maps @xmath9 on an infinite set @xmath0 , there are two unary maps @xmath10 ( not necessary belonging to @xmath8 ) , such that any map from @xmath8 is a composition of @xmath11 and @xmath12 .
this statement is , essentially , about @xmath13-generation of a countable transformation semigroup , and was apparently rediscovered in the literature several times afterwards .
it admits various generalizations to other semigroups and groups , some of them , especially in topological setting ( where `` generation '' is understood up to the closure ) , were pursued by sierpiski and two banach s students , stanisaw ulam ( phd from the lvov polytechnic , 1933 ) , and jzef schreier ( phd from the lvov university , 1934 ; cf .
guran and prytula @xcite for an interesting discussion , and the bestseller @xcite for ulam s account how his joint work with schreier secured him a place in the harvard society of fellows in 1936 ) .
yet theorem [ th - un ] about unary maps proved by sierpiski and banach admits generalizations in another direction to the multiary maps .
( note the drastic difference between superposition of unary maps , which reduces to a mere composition and hence is associative , and the general case of superposition of multiary maps for which associativity is not even well defined ) .
this may suggest that `` generation '' in banach s question can be interpreted as an arbitrary superposition of maps ( with possible repetition of variables ) . for example , among the ternary maps obtained by superposition of a binary map @xmath11 with itself , are the maps @xmath14 etc . in this general setting
, however , banach s claim becomes false : any map of an arbitrary ( finite ) arity on any set in fact , any countable set of such maps can be generated in this fashion by one binary map ( a multiary analog of theorem [ th - un ] ) .
more precisely : [ th - webb - los ] for any countable set @xmath8 of multiary maps @xmath15 ( of , possibly , different arities ) on a set @xmath0 , there is a binary map @xmath16 ( again , not necessary belonging to @xmath8 ) such that any map from @xmath8 is a superposition of @xmath11 .
this was proved several times in the literature : for the first time , in @xcite for the case of @xmath0 finite ( the required binary map is a multiary generalization of the sheffer stroke ) . in the case of
@xmath0 infinite , a different , and much shorter proof was presented , again , by o @xcite ( a similar , and yet simpler , proof was rediscovered more than half century later by goldstern @xcite ) .
these proofs , in its turn , are based on an another result of sierpiski @xcite to the effect that any single multiary map on an infinite set can be obtained as a superposition of ( possibly , several ) binary maps .
the latter paper of sierpiski appeared around the same time as banach s question under discussion : the same 1945 issue of _ fundamenta mathematicae _ in which the paper was published , the first one after the 6-year break occurred during wwii , contains an announcement about banach s death .
the same paper contains also another elementary , but interesting for us result : [ th - tri ] for any binary bijection @xmath17 on a ( necessary infinite ) set @xmath0 , and any ternary map @xmath18 , there is another binary map @xmath19 such that @xmath20 for any @xmath21 .
( the statement readily generalizes to @xmath7-ary maps @xmath11 ) .
the result was , however , not new at that time : it appeared as the solution to the problem 119 , imaginatively entitled `` are there actually functions of @xmath22 variables ? '' , of part 2 in the first 1925 edition of the famous book @xcite .
sierpiski s interest in this topic stems , evidently , from ( a variant of ) hilbert s 13th problem : does there exist a real continuous function in @xmath22 variables which can not be represented as a superposition of real continuous functions in @xmath13 variables ?
see , for example , his earlier paper @xcite where theorem [ th - tri ] is proved for the case where @xmath0 is the set of real numbers ( albeit without using axiom of choice which is necessary in the general case ) , with reference to the ( in)famously , at that time , erroneous paper by bieberbach about hilbert s 13th problem .
ulam , along with mark kac ( phd from the lvov university , 1937 , with banach as a member of examining committee ) , also an active participant of the lvov mathematical scene until the end of 1930s ( cf .
@xcite , @xcite , and @xcite ) , has at least a cursory interest in hilbert s 13th problem as well . in the collection of ulam s problems (
* chapter iv , problem 2 and chapter vi , problem 5 ) , which he positions as a sort of successor to the scottish book , as well in the joint kac s and ulam s popular book @xcite , they note a famous 19561957 negative solution of the hilbert s 13th problem by kolmogorov and arnold : [ th - kolm - arn ] for any real continuous function in @xmath22 variables @xmath23 \times [ 0,1 ] \times [ 0,1 ] \to \mathbb r$ ] , there are @xmath24 real continuous functions in @xmath13 variables @xmath25 , @xmath26 , such that @xmath27 for any @xmath28 .
kac and ulam proceed then by formulating various extensions of the problem ( for other classes of functions , for functions acting on more complicated spaces , etc . ) . in @xcite ,
ulam acknowledges banach , among others , for `` the pleasure of past collaboration '' , but it is unclear whether this interest in the circle of questions related to hilbert s 13th problem goes back to the lvov years , and if yes , whether it has something to do with banach s question . at any rate , as was noted on several occasions , `` the iteration and composition of functions '' was one of the main leitmotifs of ulam s work during his life ( cf .
, e.g. , @xcite ) .
of course , there is a very long way from theorems [ th - un][th - tri ] to theorem [ th - kolm - arn ] , but they represent development along the same circle of ideas ( arnold himself used to cite theorem [ th - tri ] as one of the early precursors in his popular expositions of the history of hilbert s 13th problem ) . the questions whether that or another set of maps generates all maps within a given class , framed in terms of functional completeness of various multivalued propositional calculi ( `` multivalued logics '' , or `` logistics '' , as it was called then ) and the corresponding truth tables , were also popular among polish logicians at that time . in particular , jan ukasiewicz was concerned about functional completeness of his famous @xmath22-valued logic that is , `` implication '' , the binary map _ _ on the @xmath22-element set @xmath29 given by `` multiplication table '' @xmath30 together with `` negation '' , the unary map @xmath31 defined by @xmath32 and similar systems .
( while ukasiewicz puts emphasis on the philosophical significance of many - valued logics cf .
, e.g. , @xcite most of the questions related to these logics , including question of functional completeness , naturally have a purely formal , mathematical , character ) .
his student jerzy supecki proved in his phd thesis ( cf .
@xcite ) functional incompleteness of the set @xmath33 , and , in the positive direction , established functional completeness of the set @xmath34 , where @xmath35 is the unary map sending all @xmath22 elements to @xmath36 : let @xmath37 be a set of @xmath22 elements . 1 .
there are multiary maps @xmath15 which can not be represented as a superposition of the maps and @xmath31 .
any multiary map @xmath15 is a superposition of the maps , @xmath31 , and @xmath35 .
( it is interesting to note that the same 19361937 issue of _ comptes rendus des sances de la socit des sciences et des lettres de varsovie _ in which supecki s paper was published , contains the paper @xcite of donald l. webb , a fresh phd from caltech and the author of the already mentioned @xcite , in which he extends the results of the latter paper .
this , among other things , suggests that developments in this branch of logical calculi on both sides of atlantic did not proceed in isolation at that time - valued logic as one of the four major contributions `` on the nature of truth '' during the last 6000 years ( cf .
@xcite ) , so it is , perhaps , not accidental that webb , a young californian , published in a relatively obscure warsaw journal . ] ) .
earlier , another ukasiewicz student , mordchaj wajsberg has reported at the ukasiewicz
tarski seminar some related results see a brief description at the very end ( `` anmerkung '' ) of @xcite ( cf . also surma @xcite ) .
in the same paper , more binary maps generating all maps of arbitrary arity on a finite set are presented without proof .
some of the wajsberg s results apparently go back as early as 1927 , i.e. almost a decade before webb .
more logical calculi , both complete and incomplete , were developed around this time , first of all , by emil leon post ( born in poland , otherwise not related to that country ) , as well as by ukasiewicz , supecki , bolesaw sobociski ( another phd student of ukasiewicz ) , eustachy yliski ( professor of the lvov university in 19191941 ) , zygmunt zawirski , and others .
post s purely formal treatment of many - valued logics ( `` a combinatorial scheme '' in the words of mostowski @xcite ) is , perhaps , closer in spirit to mathematical questions considered here than the philosophical attitude of ukasiewicz .
another feature of the work of ukasiewicz s school and polish logicians in general , was the constant quest for minimal , as far as possible , systems of axioms for that or another logical calculus .
minimality was understood both in terms of the number of axioms and their length ( cf .
, e.g. , @xcite and @xcite ) .
banach s claim and question resonate well with this line of thought , as they can be phrased as follows : if a certain @xmath22-term operation in an @xmath7-valued logic is not generated by a @xmath13-term one ( alas , as we have seen , this is wrong ) , the next best thing to ask is to extend the @xmath22-term operation so that it will be generated that way . of course ,
many further results about hilbert s 13th problem , from one side , and functional completeness and incompleteness of various logical calculi , from the other , were obtained afterwards ( cf . , e.g. , the survey by vitushkin @xcite for the former , and the book by lau @xcite for the latter ) , but their discussion will bring us far away from our main topic .
the `` generation '' in the results above is understood in the sense of the theory of clones , i.e. when forming superposition of maps , repeated variables ( and , hence , superpositions of arbitrary length ) are allowed .
( recall that clone is a set of a multiary maps on a fixed set @xmath0 which contains projections on each variable , and is closed with respect to superposition ; cf .
, e.g. , @xcite ) .
thus , banach s claim can not be interpreted in terms of the clone of all maps on the underlying set @xmath0 , as theorem [ th - webb - los ] tells that all the clone , including its ternary fragment , is generated by its binary fragment .
it should be noted that if we assume that @xmath0 possess some additional topological , analytic , order , etc .
structure , and consider not the clone of all maps , but the clone of maps on @xmath0 preserving this structure then theorem [ th - webb - los ] about generation of all the maps by binary ones is no longer true .
some sporadic examples of various degree of sophistication : 1 .
not every real analytic function in @xmath22 variables can be represented as a superposition of analytic functions in two variables ( cf .
@xcite ) a statement made already by hilbert in his original formulation of the 13th problem , whose proof utilizes some counting similar to elementary counting in the case of @xmath0 finite , and also in the case of polynomials over a finite field in example [ ex - gf ] , see below .
. superpositions of smooth functions in @xmath13 variables satisfy various differential equations , not satisfied by all smooth functions in @xmath22 variables ( cf .
solution of problem 119a , part 2 of @xcite for the relevant calculations ) .
algebraic functions in sufficiently high number of variables can not be represented as superposition of algebraic functions in sufficiently low ( in particular , @xmath13 ) number of variables a suite of deep results due to v.i .
arnold and his followers , obtained by interpreting cohomology classes of a suitable braid group as obstructions to such representation ( cf .
napolitano @xcite for references ) .
4 . there is an @xmath38-element poset whose clone of monotone maps can not be generated not only by its binary fragment , but by any finite set of maps ( due to g. tardos , cf .
@xcite ) .
however , as banach is apparently interested in the case of @xmath0 finite , and does not impose on @xmath0 any additional structure , such interpretation seems to be unlikely .
if , however , we will understand the superposition in a more `` operadic - like '' , `` multilinear '' fashion , where each variable occurs only once , the only possible ways to generate the ternary map ( [ eq - f ] ) by a binary one @xmath39 , are : by1 @xmath40 i.e. the same as in the version of the question from the o paper @xcite discussed above , and @xmath41 this is also in line with theorem [ th - tri ] . of course , any map representable in the form ( [ eq - l ] ) gives rise , via permutation of arguments , to a map representable in the form ( [ eq - r ] ) , and vice versa .
indeed , the equality ( [ eq - r ] ) can be rewritten as @xmath42 where @xmath43 and @xmath44 is a permutation from @xmath45 , the symmetric group in @xmath7 variables .
banach is concerned with the case of `` commutative '' maps , which , by all accounts , are @xmath7-ary maps @xmath11 which are stable under any @xmath46 : @xmath47 ( usually , such maps are called symmetric ) .
as it follows from ( [ eq-12 ] ) , a commutative map @xmath11 is representable in the form ( [ eq - l ] ) if and only if it is representable in the form ( [ eq - r ] ) . in this sense
, banach s claim becomes true . in the case of a finite set @xmath0 ,
this is obvious from an elementary counting : the number of binary maps is @xmath48 , so the number of ternary maps of the form ( [ eq - l ] ) and ( [ eq - r ] ) is less than @xmath49 ( less , because these maps may coincide for different @xmath50 s ) , while the number of all ternary maps is @xmath51 . as an aside note
, it seems to be an interesting question to estimate more exactly the number of different maps of the form ( [ eq - l ] ) and ( [ eq - r ] ) on an @xmath7-element set .
a computer count produces the following table , where the second column contains the number of all binary maps on an @xmath7-element set , @xmath52 denotes the number of ternary maps of the form ( [ eq - l ] ) , @xmath53 denotes the number of ternary maps both of the form ( [ eq - l ] ) and ( [ eq - r ] ) , and @xmath54 denotes the number of commutative ternary maps of the form ( [ eq - l ] ) ( which coincides with the number of such maps both of the form ( [ eq - l ] ) and ( [ eq - r ] ) ) ( so , for @xmath55 we have the obvious inequalities @xmath56 ) : [ cols="<,>,>,>,>",options="header " , ] in the given range , @xmath54 turns out to be equal to @xmath57 , the number of commutative ternary maps of the form ( [ eq - l ] ) with _ commutative _ @xmath50 .
we do not know the general formulas for @xmath52 , @xmath53 , @xmath54 , and @xmath57 , and refrain here from making any conjectures , but this seems to be an interesting topic to study further , is available as an ancillary file in the arxiv version of this note . at the time of this writing ,
the @xmath22-term sequences for @xmath52 and @xmath53 are absent in the online encyclopedia of integer sequences . ] .
it is easy to manufacture examples of a ternary map @xmath11 which can not be generated , in the sense of ( [ eq - l ] ) , by any binary operation @xmath50 , thus explicitly confirming banach s claim ( in examples below , the ternary maps are commutative , but they are not representable as superposition , in the sense of ( [ eq - l ] ) , of any binary map , commutative or not )
. let @xmath0 be a set containing more than @xmath13 elements , and @xmath58 be two distinct elements of @xmath0 .
define a ternary map @xmath11 on @xmath0 by @xmath59 suppose ( [ eq - l ] ) holds for some binary map @xmath50 on @xmath0 . if for any two elements @xmath60 , both distinct from @xmath61 , @xmath62 , then for any @xmath22 elements @xmath21 , each distinct from @xmath61 , we have @xmath63 , a contradiction .
hence there are @xmath64 , both distinct from @xmath61 , such that @xmath65 .
then @xmath66 , and then for any @xmath67 , @xmath68 , and , finally , @xmath69 , a contradiction .
[ ex - gf ] let @xmath0 be a set of @xmath70 elements , where @xmath71 is a prime , and @xmath11 a ternary map on @xmath0 .
endow @xmath0 with the structure of the finite field @xmath72 .
the question is whether there exists or not a binary map @xmath12 on @xmath72 such that @xmath73 for any @xmath74 . since , due to lagrange interpolation formula , each @xmath75-ary function on @xmath72 can be represented as a polynomial in @xmath75 variables with coefficients in @xmath72 , and the degree @xmath76 in each variable ( cf . ,
e.g. , lidl
niederreiter @xcite ) , and , moreover , two such polynomial maps are equal if and only if their coefficients are equal , the condition ( [ eq - fg ] ) can be rewritten as a system of polynomial equations in ( polynomial ) coefficients of @xmath12 . suitably choosing @xmath11
, one can obtain a system not having solutions in @xmath72 ( in fact , `` most '' of the @xmath11 s will do , as the system consists of @xmath77 equations in @xmath78 unknowns ) . for example , defining a ternary map @xmath79 by @xmath80 , and writing @xmath81 for some @xmath82 , we arrive at the system @xmath83 which , evidently , does not have solutions .
of course , nothing in these examples is specific to @xmath22 variables , and they can be easily extended to @xmath7-ary maps for arbitrary @xmath7 , and , moreover , to non - representability in the form @xmath84 for two binary maps @xmath50 and @xmath85 , and similar @xmath7-ary expressions .
the following answers the question of banach interpreted as above
i.e. about extensions of arbitrary ternary maps to those having the form ( [ eq - l ] ) in affirmative .
( we deal with arbitrary , not necessary commutative , maps ) .
[ th-1 ] for any ( finite ) set @xmath0 and a ternary map @xmath18 , there is a ( finite ) set @xmath86 and a binary map @xmath87 such that @xmath88 for any @xmath89 .
define @xmath90 as a disjoint union @xmath91 , and the binary map @xmath50 on it as follows : @xmath92 where @xmath89 .
the rest , i.e. @xmath93 and @xmath94 , is defined in an arbitrary way .
this elementary construction has a drawback that it can not be easily modified to obtain similar statements in classes of maps satisfying various conditions .
for example , the so constructed binary map @xmath50 is , generally , neither commutative , nor associative , even if the initial ternary map @xmath11 is . to remedy this
, one may adopt just a slightly more complicated approach , based on various , related , constructions of envelopes of lie and other triple systems , an idea going back to the pioneering paper @xcite of nathan jacobson ( born in warsaw , but otherwise not related to poland ) published only a few years later than the question was posed . in this way
, we get an alternative proof of a positive answer to banach s question in the narrower `` associative '' version , given in the papers @xcite and @xcite as mentioned at the beginning of [ sec - los ] .
our proof differs from both of them and , as we hope , is shorter and simpler . for any ( finite ) commutative ternary semigroup @xmath4
there is a ( finite ) commutative binary semigroup @xmath95 such that @xmath86 and @xmath88 for any @xmath89 .
let @xmath96 be the set consisting of maps @xmath9 of the form @xmath97 for all @xmath60 , and @xmath98 the subsemigroup of the semigroup of all maps @xmath9 with respect to composition , generated by @xmath96 .
note that commutativity and associativity of @xmath11 imply @xmath99 for any @xmath100 , so @xmath98 is a commutative semigroup .
define @xmath90 as a disjoint union @xmath101 , and the binary map @xmath50 on it as follows : @xmath102 for @xmath60 , @xmath103 ( @xmath85 denotes the composition of maps ) . as @xmath11 is commutative , @xmath104 , and
hence @xmath50 is commutative . for any @xmath21 ,
we have @xmath105 it remains to check the associativity of @xmath50 . for @xmath22 terms which all belong to @xmath0 ,
the associativity of @xmath50 follows from the commutativity of @xmath11 .
similarly , for 3 terms which all belong to @xmath98 , the associativity of @xmath50 follows from the associativity of @xmath85 .
if two terms , say , @xmath12 and @xmath106 , belong to @xmath98 , and one , say @xmath107 , belongs to @xmath0 , the associativity of @xmath50 follows from the commutativity of @xmath85 in @xmath98 : @xmath108 in the remaining cases , where one term , @xmath109 ( @xmath64 ) , belongs to @xmath98 , and two terms , @xmath107 and @xmath110 , belong to @xmath0 , assuming additionally @xmath111 , and utilizing commutativity and associativity of @xmath11 , we have : @xmath112 and @xmath113 what completes the proof .
one can deal similarly with not necessary commutative ternary semigroups ( one needs then to consider , instead of @xmath114 , both `` left '' and `` right '' multiplications ) , with ternary groups ( thus recovering a part of post s results @xcite ) , etc .
some of the statements of this section may be suitably extended , via a straightforward iterative procedure , to the maps @xmath11 of arbitrary arity , but we will not venture into this .
thanks are due to martin goldstern , kateryna pavlyk , jan woleski , and especially an anonymous referee for useful remarks which improved the text . the financial support of fapesp ( grant 13/12050 - 2 ) , of the regional authority of the moravian - silesian region ( grant msk 44/3316 ) , and of the ministry of education and science of the republic of kazakhstan ( grant 0828/gf4 ) is gratefully acknowledged .
arnold , _ representation of continuous functions of three variables by the superposition of continuous functions of two variables _ , mat .
sbornik * 48 * ( 1959 ) , 374 ( in russian ) ; english translation : amer .
* 28 * ( 1963 ) , 61147 .
m. goldstern , _ a single binary function is enough _ , proceedings of the 81st workshop on general algebra ( ed .
czermak et al . ) , verlag johannes heyn , klagenfurt , contributions to general algebra * 20 * ( 2012 ) , 3537 .
s. ulam , _ problems in modern mathematics _ , john wiley & sons , 1960 ; reprinted , in a slightly extended form , in 1964 , and in _ sets , numbers , and universes . selected works _ , the mit press , 1974 , 505670 .
m. wajsberg , _ metalogische beitrge _ , wiadomoci matematyczne
* 43 * ( 1937 ) , 131168 ; english translation in @xcite , 285318 ; reprinted in _ logical works _ ( ed . s.j .
surma ) , ossolineum , wrocaw , 1977 , 172200 .
j. woleski , _ mathematical logic in poland 19001939 : people , circles , institutions , ideas _ , modern logic * 5 * ( 1995 ) , 363405 ; reprinted in _ essays in the history of logic and logical philosophy _ , jagiellonian univ .
press , krakw , 1999 . , _ logic and foundations of mathematics in lvov ( 19001939 ) _ , in _ lvov mathematical school in the period 19151945 as seen today _
b. bojarski , j. awrynowicz , and ya.g .
prytula ) , banach center publ .
* 87 * ( 2009 ) , 2744 . | we discuss the last question of banach , posed by him in 1944 , shortly before his death , about extension of a ternary map to superposition of a binary map .
we try to put things into the context of polish mathematics of that time , and touch upon subsequent developments in such diverse areas as multivalued logics , binary and ternary semigroups , theory of clones , and hilbert s 13th problem . making almost a full circle in time ,
we show how variants of banach s question may be settled using a 1949 idea of jacobson about envelopes of lie triple systems . |
in the quest to better understand non perturbative phenomena in string theory , and to find clues as to the full nature of m
theory@xcite , rich solvable examples are of considerable value . the type 0 minimal string theories ( formulated in refs .
@xcite and refs .
@xcite , and recognized as type 0 strings in ref .
@xcite ) have highly tractable non perturbative physics , and despite being exactly solvable contain a rich set of phenomena such as holography and open closed dualities . like the original minimal strings@xcite
, they have bosonic spacetime physics , although they have worldsheet fermions , and may be thought of as two dimensional supergravity coupled to @xmath6 superconformal minimal models , with a diagonal gso projection .
the physics of the minimal strings can be succinctly formulated in terms of associated non
linear ordinary differential equations , known as the string equations .
they furnish asymptotic expansions for the free energy once the boundary conditions are fixed .
these expansions take the form of a string world sheet expansion .
typically ( for the cases of the type 0a and 0b systems coupled to the @xmath7 superconformal minimal models ) there are two perturbative regimes in which such an expansion can be developed .
one is interpreted as purely closed string backgrounds with integer , @xmath8 , units of r r flux , while the other has both open and closed string sectors with @xmath8 d branes in the background d branes or flux units correspond to special soliton solutions , as shown in refs .
@xcite . ] .
remarkably , these models have a non perturbative completion ( first discovered in refs .
@xcite in the context of the type 0a systems ) that connects these two asymptotic regimes , furnishing an example of a so called `` geometric transition '' between open and closed string backgrounds .
while in the case of type 0a , complete ( numerical ) solutions of the exact string equations for any @xmath8 can be found , the connectedness of the type 0b case has been argued for on the basis of a large @xmath8 , t hooft like , limit where the interpolating solution can be found using algebraic techniques@xcite .
this has more than just shades of the ads / cft correspondence@xcite , where in the prototype , the open string physics of @xmath8 d3branes can be rewritten in terms of that of closed strings in ads@xmath9 with @xmath8 units of r r flux . in the present context
we have a precise analogue of this important example .
there is of course a t hooft large @xmath8 limit there too , connecting supergravity to large @xmath8 yang mills . that we have here not just the analogue of the large @xmath8 limit but also knowledge of how to go beyond , may ultimately prove instructive . in ref .
@xcite , we investigated a new system called the dispersive water wave ( dww ) hierarchy@xcite , which we argued should yield new string equations _ via _ a similarity reduction .
these string equations turn out to be a painlev iv hierarchy of equations .
we found that both the type 0a and 0b string theories were found to be naturally embedded within this framework .
in addition to this new non perturbative connection , we found further that the two string theories are merely special points in a much larger tapestry of possibilities that also appear to be string theories .
this is somewhat suggestive of an m theory , now for minimal strings , which is exactly solvable .
it clearly deserves further study . among the other special points which suggested themselves ( using perturbation theory )
were some that we conjectured to be the type iia and iib string theories coupled to the @xmath0 superconformal minimal models .
much of the analysis carried out to support this identification was using perturbative techniques .
the subject of the present paper is to continue the study of this rich system into the non perturbative regime .
we argue that the conjectured type ii theories have non perturbative completions by showing that the corresponding perturbative expansions match onto each other smoothly .
this is accomplished using a combination of analytical and numerical techniques similar to those used for the type 0 theories coupled to the @xmath7 superconformal minimal models .
the outline of the paper is as follows : in sections 2 and 3 , we review essential aspects of the well known type 0 theories and provide a summary of results about the dww system that will be needed later .
we reproduce the analytic ( t hooft limit ) technique used for the type 0b string theory , as in ref .
@xcite , in section 4 to keep the presentation self contained .
we then reformulate it in a manner that allows us to extend the technique to our dww system .
we apply this strategy to our system in section 5 for the first two even dww flows , which contain the type 0 and conjectured type ii theories of interest and find that the type ii theories possess smooth non perturbative solutions .
in addition , we find evidence for the existence of new non perturbative solutions with novel asymptotics . in section 6 , we consider the first ( and simplest ) flow of the dww system .
for this case , we are able to strengthen evidence for the existence of these new non perturbative solutions with exact numerical results .
we end with a brief discussion in section 7 .
an appendix details the expansions arising from the dww system for the first few flows along with some of their essential properties .
we begin with a brief review of the type 0 string theories coupled to the @xmath7 superconformal minimal models of type @xmath10 .
those familiar with these models can proceed directly to the next section .
type 0a string theory coupled to the @xmath7 superconformal minimal models ( first derived and studied in refs.@xcite and identified as type 0a in ref .
@xcite ) is described by the string equation @xmath11 where , for a particular model , @xmath12 is a real function of the real variable @xmath13 , a prime denotes @xmath14 , and @xmath15 and @xmath16 are real constants .
the quantity @xmath17 is a function of @xmath12 and its @xmath13derivatives . in general
@xmath12 additionally depends on couplings @xmath18 associated with flowing between the various models
. then we have @xmath19 where the @xmath20 $ ] are polynomials in @xmath12 and its @xmath13derivatives , called the gelfand
dikii polynomials@xcite .
they are related by a recursion relation ( defining a recursion operator @xmath21 ) @xmath22 and fixed by the value of the constant @xmath23 and the requirement that the rest vanish for vanishing @xmath24 .
the first four are : @xmath25 the @xmath26th model is chosen by setting all the @xmath27 to zero except @xmath28 and @xmath29 this number is chosen so that the coefficient of @xmath30 in @xmath17 is set to @xmath31 . as @xmath32 .
if we had instead chosen @xmath33 , we would have chosen the coefficient of @xmath30 to be unity . ] the flows between various models are captured by the integrable kdv hierarchy@xcite .
the function @xmath12 defines the partition function @xmath34 of the string theory @xmath35 @xmath36 where @xmath37 is the coefficient of the lowest dimension operator in the world
sheet theory .
the asymptotic expansions of the string equations for the first two cases are : @xmath38 @xmath39 @xmath40 @xmath41 the solutions for various @xmath26 for @xmath42 can be numerically and analytically shown to match onto the solution for @xmath43 , providing a unique@xcite non perturbative completion of the theory .
( see figure ( [ fig : plot ] ) for an example of a solution found using numerical methods . )
+ in the @xmath44 regime , @xmath15 represents@xcite the number of background zz d branes @xcite in the model , while in the @xmath45 regime , @xmath15 represents the number of units of r r flux in the background@xcite .
type 0b string theory coupled to the @xmath46 superconformal minimal models @xcite is described by the following string equations@xcite : @xmath47 where the @xmath48 and @xmath49 are polynomials of functions @xmath50 and @xmath51 ( and their derivatives ) , and @xmath16 and @xmath52 are real constants .
the function @xmath53 defines the partition function of the theory @xmath35 @xmath54 where @xmath37 is the coefficient of the lowest dimension operator in the world
sheet theory .
the @xmath55th model is chosen by setting all @xmath56 to zero except @xmath57 and @xmath58 .
these models have an interpretation as type 0b strings coupled to the @xmath59 superconformal minimal models only for even @xmath60 . the flows between various models are captured by the integrable zakharov shabat ( zs ) hierarchy@xcite .
the asymptotic expansions of the string equations ( [ streqn0b ] ) for the first two even @xmath60 are : @xmath61 @xmath62\ , \quad ( x \rightarrow \infty ) \nonumber\\ { \widetilde w}(x ) & = & \frac{\nu q\sqrt{2}}{4|x|^{1/2 } } - \frac{\nu^2 q^2}{4 |x|^2 } + \frac{\nu^3}{|x|^{7/2}}\frac{5\sqrt{2}}{64 } q\left(1 + 4q^2\right)+\cdots \quad ( x \rightarrow -\infty)\end{aligned}\ ] ] @xmath63 @xmath64 in the @xmath45 regime , @xmath52 represents the number of background zz d branes in the model , while in the @xmath65 regime it counts the number of units of r r flux in the background@xcite .
the structure of solutions with increasing @xmath55 is particularly rich @xcite ; the @xmath66 expansion for @xmath45 shown above breaks a @xmath67 symmetry due to the presence of r
r fields in the dual string theory .
unlike the 0a case , the solutions for @xmath68 have not been shown numerically to match onto those for @xmath69 so far .
the highly non - linear nature of the string equations makes it difficult to numerically obtain smooth solutions connecting the two regimes for the 0b case . nevertheless , as argued in ref .
@xcite and reviewed in section [ sec : type0expnmtch ] , these theories can be argued to be non - perturbatively complete in a particular ( t hooft ) limit . for the @xmath70 ( @xmath71 ) model @xmath72 super conformal minimal models
is given by @xmath73 .
for @xmath70 , the central charge of the @xmath74 superconformal minimal model is thus zero and we simply have the pure world sheet supergravity sector . ] , the full non perturbative solution is known since it can be mapped directly to the solution known for the @xmath71 type 0a case @xmath35 the morris map @xcite .
the string equation for the 0a theory becomes the string equation for the 0b theory once one identifies @xmath15 with @xmath52 , but with the sign of @xmath75 reversed .
analogues of the morris map for higher @xmath55 are not known .
the dww system introduced in ref .
@xcite leads to a painlev iv hierarchy of string equations : @xmath76 where @xmath77 and @xmath15 are real constants and @xmath16 plays the same role as for the type 0 theories ( note that here and in the rest of the paper , for any function @xmath78 , @xmath79 will denote @xmath80 ) .
the functions @xmath81 and @xmath82 are generalizations of the two functions @xmath50 and @xmath83 used to describe the 0b theory .
the polynomials @xmath84 $ ] and @xmath85 $ ] are defined by @xmath86\\ k_n[u , v]\end{array}\right ) \
\quad,\ ] ] where @xmath87 $ ] and @xmath88 $ ] are polynomials in @xmath81 , @xmath82 and their derivatives , similar to the polynomials @xmath89 and @xmath90 for the 0b theory .
they satisfy the recursion relation @xmath91\\ k_{n+1}[u , v]\end{array}\right ) \
= r \left(\begin{array}{c } l_n[u , v]\\ k_n[u , v]\end{array}\right ) \
\quad .\end{aligned}\ ] ] where @xmath92 is the recursion operator of the dww hierarchy , given by @xmath93 the first few @xmath94 and @xmath95 are as follows : @xmath96 the @xmath55th model is chosen by setting all @xmath97 equal to zero except for @xmath98 and @xmath58 which is chosen to be a numerical factor to fix the normalization .
the parameter @xmath58 can be replaced by @xmath99 in order to make direct contact with recent literature which discusses this system in a much different ( mathematical ) context@xcite .
analytic expansion solutions to the string equations ( [ dwwstring1 ] ) and ( [ dwwstring2 ] ) for all @xmath55 can be organized into five main classes on the basis of the @xmath100 ( or @xmath101 ) behavior of @xmath81 and @xmath82 , as follows , @xmath102 they follow interesting patterns , while the expansions in class 5 are the analogues of the @xmath67 symmetry - breaking solutions of the 0b theories . ] with increasing @xmath55 and have been explored in detail in ref .
representative solutions for the first few @xmath55 have been provided in the appendix for ease of reference .
in addition to the type 0 theories reviewed earlier , these solutions encode new string theories , some of which were conjectured to be type ii string theories coupled to super
conformal minimal models as reviewed below .
the type 0 theories reviewed earlier can be recovered completely from this system of equations by applying appropriate constraints @xcite . setting @xmath103 and @xmath104 in ( [ dwwstring1 ] ) and ( [ dwwstring2 ] ) results in the type 0a and type 0b theories respectively . ] .
consistency of the two equations then requires that @xmath105 for type 0a and @xmath106 for type 0b , while the free parameter counts the branes and fluxes in the respective theories .
the function @xmath82 ( after appropriate redefinition ) encodes the free energies of the respective theories , according to equations ( [ 0afreeenergy ] ) and ( [ 0bfreeenergy ] ) .
one can also recover solutions corresponding to the type 0 theories by setting @xmath107 or @xmath106 in appropriately chosen expansions from the five classes listed above .
expansions in classes 1 and 2 ( labeled @xmath108 and @xmath109 ) reduce to those of the type 0a theory ( see ( [ 0aexpnsk=1 ] ) and ( [ 0aexpnsk=2 ] ) ) once we set @xmath105 .
this helps fix the directions of the various expansions by requiring that @xmath82 be real , with the result that some expansions are real and others complex for a given @xmath55 .
further details can be found in ref .
@xcite .
a full solution for @xmath82 is constructed by specifying its behavior in the two asymptotic regimes , as seen for the @xmath110 0a theory in figure ( [ fig : plot ] ) .
this 0a solution is obtained from the dww expansions , @xmath108 and @xmath109 , by setting @xmath105 . using this scheme ,
the above expansions can be organized in the form of a square as shown in figure ( [ square2 ] ) .
the four corners of the square correspond to four different string theories .
the top corners represent the known type 0 strings coupled to the @xmath7 @xmath10 superminimal models .
the bottom corners represent two new theories that were conjectured in ref .
@xcite to be type iia and iib string theories coupled to the @xmath0 @xmath1 minimal models ( for even @xmath26 only ) . at each corner ,
one parameter from @xmath111 is frozen to the value indicated on the figure , while the other parameter counts the branes / fluxes in the appropriate regime .
the horizontal sides of the square correspond to expansions in the @xmath112 direction , while the vertical sides represent expansions in the @xmath113 direction . for all @xmath55 , there are two special points on the vertical sides corresponding to @xmath114 , where the expansions @xmath115 and @xmath116 identically vanish . + it must be noted that the square fully exists only for even @xmath117 . for @xmath118
, the expansions labeled @xmath119 become complex as @xmath112 expansions and the expansions labeled @xmath120 do not appear for odd @xmath55 .
the new theories were conjectured to be type ii strings coupled to super minimal matter by matching the genus zero contributions from the asymptotic expansions ( the terms appearing at order @xmath121 in the free energy ) with the continuum calculation of their one loop partition functions . in the absence of
non perturbative constraints ( like @xmath122 for the 0a theory ) leading to the conjectured type ii theories , the corresponding values of @xmath123 and @xmath124 in figure [ square2 ] , were obtained by systematic analysis of the properties of the genus zero contributions @xcite . however , pairing perturbative expansions in this manner does not guarantee the existence of a full non - perturbative solution with the desired properties .
a type of t hooft limit will allow us to argue for the existence of such non - perturbative solutions for the type ii theories .
we demonstrate how this limit works for the two parameter dww system in section ( [ sec : dwwexpnmtch ] ) . before that
, we review the t hooft limit for the known type 0 theories .
the theories encoded by the string equations described so far can be non
perturbatively completed by matching the perturbative solutions for @xmath125 on to those for @xmath126 .
numerical solutions have been found for the string equations for the 0a models that smoothly interpolate between the two asymptotic regimes as discussed earlier .
see , e.g. , figure ( [ fig : plot ] ) .
the string equations for the 0b models , however , are more difficult to analyze numerically and attempts to find smooth numerical solutions similar to the 0a ones have been unsuccessful thus far @xmath127 0b theory , where the numerical @xmath71 type 0a solution can be converted into it @xmath35 the morris map , as mentioned in section [ sec : type0breview ] . ]
this difficulty is inherited by the dww string equations presented in section [ sec : dwwreview ] .
it was shown in ref .
@xcite that the asymptotic expansions for the 0b theories ( for example , those in equations ( [ 0bexpnsm=2 ] ) and ( [ 0bexpnsm=4 ] ) ) match on to each other in a t hooft limit , with the brane / flux counting parameter ( labeled @xmath52 for 0b and @xmath15 for 0a ) taken to be large .
this was then taken to suggest that the theories are non perturbatively complete even when the parameter is finite .
we review this limit in some detail , before analyzing our system in the same limit in section [ sec : dwwexpnmtch ] .
the string equation for the simplest 0b theory with @xmath70 is @xmath128 obtained by using the explicit forms of the polynomials @xmath21 and @xmath129 and eliminating @xmath83 between the two equations ( [ streqn0b ] ) .
consider the limit @xmath130 , @xmath131 with @xmath132 fixed .
this is our t hooft limit with @xmath133 being the t hooft coupling and @xmath52 as counting the number of units of r r flux , this is recognized as the usual t hooft coupling @xmath134 familiar from higher - dimensional string theories .
taking @xmath131 amounts to taking the limit @xmath135 . ] . defining @xmath136 ,
the string equation ( [ streqn0bm=2v2 ] ) becomes @xmath137 in the large @xmath52 limit , the first term containing the derivative can be neglected resulting in an algebraic equation for @xmath138 , @xmath139 for generic @xmath140 this equation has 3 solutions , only _
one _ of which is real for all @xmath140 ( an easy check is to consider @xmath141 ) .
this solution reads @xmath142 \quad.\ ] ] for @xmath143 , the arguments of the square roots are positive and @xmath144 is real .
since there are opposite signs in front of the square roots in the above expression , the half - integer powers of @xmath140 cancel when one expands @xmath144 for @xmath145 .
for @xmath146 the arguments of the square roots are negative and the second and third terms in the solution equation are complex , but @xmath144 is real since all the imaginary contributions cancel out .
the expansions for @xmath144 for large negative and positive @xmath140 are @xmath147 these expansions reproduce the coefficients of the highest powers of @xmath52 in the asymptotic expansions to the string equation ( [ 0bexpnsm=2 ] ) after remembering that @xmath148 .
one smooth function ( [ vsolnm=2 ] ) captures the limiting behavior of large @xmath52 and connects = \int dx \left[\frac{1}{2}r'^{2 } + v(r^2)\right]$ ] whose equation of motion is ( [ streqn0bm=2v2 ] ) .
since this action is bounded below , it is clear that a solution to this equation will exist .
the original matrix model integral dual to the theory is well defined and convergent , so one expects a finite and real answer for the free energy @xmath149 .
it is then natural to expect that equation ( [ streqn0bm=2v2 ] ) will have a unique real and smooth solution . ]
the two asymptotic regions smoothly . this limit has been interpreted in @xcite as a t hooft limit where only spherical topologies with boundaries ( d branes ) or fluxes survive . for @xmath150 ,
the 0b theory exhibits the gross
witten phase transition at @xmath151 ( namely @xmath152 for @xmath153 and @xmath154 for @xmath69 ) .
this transition can be smoothed out by either the genus expansion with @xmath155 or by the expansion in the t hooft parameter for large @xmath52 .
a similar analysis shows that the asymptotic expansions for the @xmath66 theory ( [ 0bexpnsm=4 ] ) match smoothly onto each other .
the analysis is somewhat complicated because @xmath83 can not be eliminated in favor of @xmath50 , unlike the @xmath70 case and one has to use the asymptotic expansions for both @xmath50 and @xmath83 .
the expansions for @xmath83 are @xcite : @xmath156 in the @xmath130 limit with @xmath157 , @xmath158 and @xmath159 , the derivative terms in the string equation for this case are suppressed , resulting in @xmath160 the second equation , on solving for @xmath161 , gives @xmath162 from equation ( [ 0bexpnsm=4omega ] ) , it can be seen that @xmath163 is negative , so that only the solution @xmath164 can be chosen .
such multiple solutions will be prominent in our more general analysis in the next section . ] for real @xmath161 .
substituting @xmath165 into the first equation gives , after some rearrangement : @xmath166 defining @xmath167 reduces the above equation to a quadratic in @xmath168 , @xmath169 solving for @xmath168 as a function of @xmath170 @xmath171 will allow a solution for @xmath172 .
the solution @xmath173 is always positive and non zero as a function of @xmath170 . to get the asymptotics we desire , we focus on @xmath174 , which is negative and zero only for @xmath175 . using @xmath176
it is easy to see that as @xmath177 , @xmath145 and and as @xmath178 , @xmath179 .
as @xmath140 changes continuously , we will always lie on the @xmath180 branch since the two branches never cross .
the expansion for @xmath180 as a function of @xmath140 can then be worked out @xcite to be @xmath181 the coefficients in these expansions agree with the leading powers for large @xmath52 in the expansions of @xmath83 in equation ( [ 0bexpnsm=4omega ] ) and can again be interpreted as a limit in which only spherical topologies with boundaries or fluxes survive .. this shows that an appropriately chosen solution of equation ( [ vheqnm=4 ] ) interpolates between the two asymptotic regimes in the limit of large @xmath52 .
this t hooft limit removes the derivative terms from the string equations , resulting in algebraic equations that are simpler to analyze .
it can then be sensibly thought of as _ algebraic _ limit .
including the derivatives gives terms _ sub leading _ in powers of @xmath52 and presumably does not introduce any singularities that would destroy the smooth interpolation .
leading powers of @xmath52 likely smooth out any sharp transitions ( as we saw for the gross
witten phase transition above ) from large negative @xmath75 to large positive @xmath75 .
it is interesting to work out the t hooft limit of the type 0a theories ( with string equations ( [ streqn0a ] ) ) .
since , as already discussed , solutions of the full equations have been obtained numerically , it is instructive to compare the t hooft limit ( _ i.e. , _ algebraic ) results to the numerical results .
( note that while the last section s type 0b results were a review , the type 0a analysis is presented here for the first time . )
the string equation for this theory is , @xmath182 consider the limit @xmath183 , @xmath184 with @xmath185 fixed .
this is the t hooft limit with @xmath186 being the t hooft coupling .
defining @xmath187 , the above string equation becomes @xmath188 in the large @xmath15 limit , the derivative terms can be neglected to give the simple algebraic equation @xmath189 as for the type 0b case of the last section , the solutions to this cubic equation ) to obtain this equation directly from the corresponding 0b algebraic equation ( [ 0balgeqnm=2 ] ) .
] can be expanded as a taylor series for @xmath190 and seen to reproduce the coefficients of the highest powers of @xmath15 in the asymptotic expansions ( [ 0aexpnsk=1 ] ) .
one smooth function essentially connects the two asymptotic regions , as in the 0b theory .
figure ( [ 0ak1g1 ] ) shows a comparison between the solution to equation ( [ 0aalgeqnforgk=1 ] ) and the solution to the exact string equations ( [ streqn0ak=1 ] ) obtained by numerical methods , both with @xmath191 .
the two solutions deviate slightly from each other in the interior , and asymptotically , they match .
this can be seen in figure ( [ 0ak1g1diff ] ) , which plots the difference between the exact and algebraic solutions , which goes to zero for large @xmath192 . physics sub leading in powers of @xmath15 contribute to the exact solution , while only the highest powers of @xmath15 contribute to the algebraic solution .
the inclusion of the derivative terms does not introduce any singularities .
+ the string equation for the @xmath110 0a theory has @xmath193 . in the large @xmath183 and @xmath184 limit , with @xmath194
held fixed , the derivative terms drop out . defining @xmath195 results in the algebraic equation @xmath196 the solutions to this equation
reproduce the coefficients of the highest powers of @xmath15 in the asymptotic expansions ( [ 0aexpnsk=2 ] ) .
figure ( [ 0ak2g1 ] ) shows a comparison between the solution to equation ( [ 0aalgeqnforgk=2 ] ) and the solution to the exact @xmath110 0a string equation obtained by numerical methods .
the difference between the two solutions is plotted in figure [ 0ak2g1diff ] and can be seen to approach zero for large @xmath192 .
the deviation for finite @xmath75 and asymptotic matching is evident , and can be explained by the additional contributions coming from sub leading powers of @xmath15 in the exact solution .
+ a modification of the above limit ( one that does not affect the physics ) will allow us to apply it to the painlev iv hierarchy of string equations .
the need for such a modification will be explained in the next section .
instead of rescaling variables in the 0b ( 0a ) theory to eliminate @xmath52 ( @xmath15 ) , one can remove the derivatives by first rescaling @xmath197 and then taking the limit @xmath198 in that order , without affecting the physics . note that the rescaled @xmath52 is large , so this is still a large @xmath52 limit .
the limit @xmath199 amounts to taking @xmath135 , as is clear from the definition of @xmath200 ( @xmath201 for @xmath70 and @xmath202 for @xmath203 ) .
this modified limit therefore extracts the physics same as the t hooft limit of the previous subsections . in this modified limit
, the following equations are obtained for the type 0b theories for @xmath70 , @xmath204 and for @xmath66 , @xmath205 these are the same as equations ( [ 0balgeqnm=2 ] ) and ( [ vheqnm=4 ] ) with @xmath206 , @xmath207 and @xmath208 , but with the rescaled parameter @xmath52 explicitly present . for the 0a theory , after rescaling @xmath209 , the following equations are obtained for @xmath71 and @xmath110 , respectively , @xmath210 in fact , for general @xmath26 , the type 0a string equation ( [ streqn0a ] ) effectively reduces to @xmath211 where @xmath212 is obtained from @xmath213 after dropping all the derivatives of @xmath12 .
the string equations of the dww hierarchy ( reviewed in section [ sec : dwwreview ] ) are quite general , encoding the string equations of type 0a and of type 0b as special cases .
this generality , however , comes at a price : all but the simplest cases are too complicated to be solved using standard numerical methods .
this complicates our task of performing a non perturbative analysis of the string theories proposed in ref .
@xcite , which were based on strictly perturbative considerations . to facilitate this analysis
, we examine the dww string equations in a ( modified ) t hooft limit .
the result of this study is considerable evidence that the conjectured type ii theories possess smooth solutions connecting the perturbative expansions .
this analysis also uncovers new non - perturbative solutions , absent from the earlier perturbative analysis , which we suspect may encode new unidentified string theories . in section [ sec : type0expnmtch ]
we described how to take the t hooft limit of the type 0a and type 0b theories .
implicit in these methods was the fact that the type 0 theories each depend on a single parameter , which counts branes or fluxes .
the full dww equations , in contrast , have two free parameters , so it is not immediately clear how to implement the t hooft limit in this more general case . the perturbative analysis of ref .
@xcite suggests that , in addition to the type 0 theories , the full dww string equations also describe type ii theories whose branes and fluxes are counted by a single parameter .
this motivates our assumption that _ new theories are described by a one dimensional subspace of the parameters @xmath77 and @xmath15_. we further restrict our study to subspaces defined by _ linear
_ combinations of @xmath77 and @xmath15 .
this is the simplest possibility and although there could be interesting one - dimensional subspaces corresponding to non - trivial curves in @xmath214 parameter space , we will not consider them here . given this assumption and the scaling prescription described in section [ sec : modlim ] , we can utilize the t hooft limit to analyze the string equations ( [ dwwstring1 ] ) and ( [ dwwstring2 ] ) . for given @xmath215 and @xmath216 , let @xmath217 be the constraint defining the one - dimensional parameter subspace of a new theory under consideration . since the constraint is ( by assumption ) satisfied by the solutions of interest
, we can impose the constraint on the dww equations themselves .
we are left with a single free parameter which we can recale as in section [ sec : modlim ] to take the t hooft limit . in this way
we obtain algebraic equations which are simple to analyze . unfortunately
, in the search for new theories , we do not _ a priori _ know the values of @xmath215 and @xmath216 , so the above method is not directly applicable .
nevertheless , we can proceed by adopting a slightly different perspective .
note that the constraint @xmath218 implies that the free parameter can be taken to be @xmath77 or @xmath15 , or any linear combination of them ( as long as it is independent of the constraint ) .
rescaling the free parameter by @xmath219 is therefore equivalent to rescaling @xmath220 and @xmath221 .
this procedure is independent of the details of the original constraint , so it can be implemented without actually knowing the values of @xmath215 and @xmath216 .
the result is a set of algebraic equations which depends on @xmath77 and @xmath15 .
moreover , the solutions to these equations can be used to deduce information about the original constraint . to see this
, note that since @xmath216 is a constant , after taking the limit the constraint takes the form @xmath222 .
therefore , the original constraint is satisfied by any solution of the algebraic equations which satisfies @xmath222 . turning this around , if a particular solution to the algebraic equations only exists when @xmath222 , then this is a strong indication that it descends from the special solution to the full equations subject to @xmath223 .
that is to say , it is an algebraic approximation to the full solutions of some new string theory .
note that @xmath216 is not determined from this approach , but @xmath215 is . to obtain the algebraic equations ,
we begin with the dww string equations ( [ dwwstring1 ] ) and ( [ dwwstring2 ] ) , @xmath224 in the modified t hooft limit discussed above , @xmath225 the string equations simplify to @xmath226 where @xmath227 and @xmath228 are the polynomials in @xmath229 and @xmath230 obtained from @xmath231 and @xmath232 after dropping all the derivative terms . before analyzing these equations in full generality ,
let us discuss a few special cases to illustrate our methods .
1 . @xmath122 ( type 0a ) + the restriction to type 0a requires @xmath55 to be even , in which case @xmath233 .
thus , setting @xmath122 in first equation of ( [ dwwalgeqns ] ) implies @xmath123 .
this constraint can be written as @xmath234 with @xmath235 .
as described above , this constraint lifts to @xmath236 at the level of the full equations .
if we had nt known the parameter constraint of the 0a theory ( @xmath237 ) , the algebraic equations alone would indicate that @xmath238 , a constant .
+ alternatively one could start by looking for solutions to the equations ( [ dwwalgeqns ] ) with @xmath123 .
a subset of smooth solutions , identified by their boundary behavior , will correspond to asymptotic expansions in classes 1 and 2 with @xmath77 fixed ( recall that these correspond to 0a ) .
the boundary behavior of the smooth solutions is matched with the behavior of expansions in classes 1 and 2 , which are the asymptotic expansions of the type 0a theory .
2 . @xmath239 ( type 0b ) + setting @xmath239 forces @xmath106 and the first equation gives @xmath240 . for even @xmath55 ,
these reduce to the 0b algebraic equations after redefining the variables appropriately .
the constraint @xmath106 can be rewritten as @xmath234 with @xmath241 and lifts to a constraint on the full equations which takes the form @xmath242 , a constant . the true parameter constraint for the full @xmath243 theory is @xmath244 ; the algebraic equations alone do not determine the precise value of @xmath216 .
+ as in the 0a case above , one could look for smooth solutions to the algebraic equations subject to @xmath106 .
if they obey the correct boundary conditions ( expansion classes 1 and 3 ( or 5 ) ) one could identify these as 0b solutions . our strategy to demonstrate the existence of smooth non perturbative solutions to the type 0 and type ii theories and identify potentially new theories
can be summarized as follows .
take the limit ( [ modlimdww ] ) of the full dww equations to obtain a set of algebraic equations .
search for solutions to these equations subject to one of three constraints , and for simplicity we exclude such cases from the present analysis . ] : @xmath245 by matching the asymptotic behavior of these solutions onto the various expansions , we identify the type 0 and type ii string theories .
we also find new solutions with asymptotics different from those of the type 0 or type ii theories .
we speculate that these solutions correspond to some new unknown string theories .
as emphasized above , the precise parameter constraints of these new theories can not be fully determined from the algebraic equations .
the string equations for @xmath70 reduce in the algebraic limit ( [ modlimdww ] ) to @xmath247 they produce a total of nine asymptotic expansions which fall into 4 classes ( see appendix [ appexpns ] ) . as the expansions within each class are related by various @xmath248 symmetries @xcite , we label each expansion only by a subscript specifying its class . the solutions corresponding to the three constraints in parameter space are described below .
1 . @xmath123 + a plot of solutions to the algebraic equations ( [ n=2algeqn ] ) with @xmath123 and @xmath249 is shown in figure ( [ n=2c0 ] ) .
the plot on the right shows two smooth solutions that connect regions of large @xmath113 and @xmath112 . in our conventions , these are @xmath250 and @xmath251 solutions asymptotics @xmath252 and @xmath112 asymptotics @xmath253 by @xmath254 . ] .
the plot on the left shows the solutions which do not join negative asymptotics to positive asymptotics , a feature we expect to find in any solutions describing new string theories .
for this reason , these types of solutions will not be our primary interest .
+ the @xmath255 solutions correspond to the 0a theory ( with @xmath71 ) coupled to pure supergravity .
it is unclear if the @xmath256 solutions correspond to a consistent new theory .
based on the square of theories in figure ( [ square2 ] ) , it is tempting to conclude that this theory might be the type iia string theory coupled to the @xmath257 @xmath1 superconformal minimal model .
however , as mentioned in the discussion below figure ( [ square2 ] ) , the @xmath119 expansion is complex as a @xmath112 expansion for @xmath258 . for the particular case of @xmath70 and @xmath259
, @xmath119 happens to be real , but this is an exception .
we find it unlikely that the corresponding solution to the full equations exists and encodes a consistent theory , but we leave this question for future investigation .
@xmath106 + figure ( [ n=2g0 ] ) shows the algebraic solutions for @xmath244 and @xmath260 .
the right figure shows two smooth solutions , @xmath261 and @xmath262 , connecting the negative and positive regions .
+ the @xmath261 solutions are algebraic approximations to the full solutions of the 0b theory ( with @xmath70 ) coupled to pure supergravity .
analogous to the previous case with @xmath259 , it is unclear if the @xmath263 solutions descend from a consistent new theory .
it is tempting to conclude that these are type iib string theories coupled to the @xmath257 @xmath1 superminimal model , but this is likely incorrect because @xmath119 is complex as a @xmath112 expansion for @xmath118 .
3 . @xmath264 + we consider @xmath265 .
the case of @xmath266 is completely analogous .
figure ( [ n=2ceqg ] ) shows solutions for @xmath267 and @xmath249 .
the solutions in the right figure are @xmath268 and @xmath269 and connect the negative region smoothly to the positive region . + the fixed value of @xmath270 can not be deduced from the algebraic solutions . the string equations in the t hooft limit , after setting @xmath271 , are too long to be reproduced here .
the polynomials @xmath231 and @xmath232 reduce to , @xmath272 from which the corresponding algebraic string equations can be easily obtained using ( [ dwwalgeqns ] ) .
there are a total of twenty five expansions @xcite , falling into five classes .
solutions to the algebraic string equations with the three constraints are shown below . 1 .
@xmath123 + figure ( [ n=4c0 ] ) shows solutions with @xmath259 and @xmath249 .
there are three solutions that interpolate between the negative region and the positive region , @xmath273 , @xmath274 and @xmath275 .
+ the @xmath255 solutions represent type 0a coupled to the @xmath276 @xmath10 superminimal model .
the @xmath275 solutions fit with our conjectured type iia string theory coupled to the @xmath277 @xmath1 superminimal model and are analogues of the @xmath67 symmetry breaking solutions of the 0b theory .
+ the @xmath278 solutions are new .
on setting @xmath259 in the _ full _ expansions , the powers of @xmath15 in both correspond to a parameter counting branes .
an underlying string theory with these solutions , if it exists , would have branes in both asymptotic regimes type expansions ( for @xmath107 ) in both asymptotic directions for @xmath75 . ] .
further work is needed to conclude if such an underlying theory exists .
2 . @xmath106 + figure ( [ n=4g0 ] ) shows solutions with @xmath244 and @xmath279 .
the solutions which interpolate between @xmath113 and @xmath112 are @xmath280 , @xmath274 and @xmath281 .
the @xmath282 solutions correspond to the @xmath67 symmetry breaking solutions of type 0b string theory coupled to the @xmath276 superminimal model , while the @xmath283 solutions fit with our conjectured type iib string theory coupled to the @xmath277 @xmath1 superminimal models .
+ the @xmath284 solutions with @xmath106 are new , similar to those above with @xmath259 .
further work is required to understand the existence and nature of such theories .
@xmath264 + again we consider @xmath265 since the case of @xmath266 is completely analogous .
figure ( [ n=4ceqg ] ) plots the solutions for @xmath279 and @xmath285 .
the four smooth solutions which interpolate between @xmath113 and @xmath112 are two @xmath282 solutions , one @xmath286 solution and one @xmath284 solution .
( the two @xmath282 solutions can be obtained from one another by applying @xmath67 symmetries on the signs of the parameters ) .
all of these solutions are new , and it is possible that they correspond to well - defined underlying string theories , but we leave any definitive claims to future work .
the string equations for @xmath287 are , @xmath288 where we have used the relation @xmath289 .
these equations are simple enough to allow for complete numerical solutions , which will help us classify new non - perturbatively complete solutions .
we will also examine these equations in the limit ( [ modlimdww ] ) , under which they reduce to @xmath290 where we have used @xmath291 .
these equations admit four asymptotic expansions , which we label as : @xmath292 the solutions with the three different parameter constraints are presented below .
1 . @xmath123 + a plot of solutions to the algebraic equations ( [ n=1algeqn ] ) with @xmath123 is shown in figure ( [ c0gm2 ] ) .
there are two smooth solutions interpolating between @xmath112 and @xmath113 , labeled @xmath293 in our convention .
+ + + as outlined in the appendix ( [ braneflux ] ) , @xmath115 with @xmath123 can be interpreted as a flux expansion ( for odd @xmath55 ) in @xmath15 , while @xmath116 with @xmath123 is a brane expansion in @xmath15 .
these interpretations are possible strictly for @xmath123 ; a finite value of @xmath77 results in powers of @xmath200 that do not allow @xmath115 to be interpreted as a flux expansion .
so a string theory corresponding to this solution ( if it exists ) should require @xmath259 in the exact string equations .
2 . @xmath106 + the plots for this case are shown in figure ( [ cm2g0 ] ) .
there are three smooth solutions , two of which are @xmath294 and @xmath295 solutions .
the third solution is parallel to the @xmath75axis and exactly equals @xmath296 and @xmath297 .
it is a 0b solution , the ` topological point ' of the 0b theory @xcite with @xmath298 .
+ the @xmath294 and @xmath283 solutions are new . for any value of @xmath15 ,
@xmath115 and @xmath119 have powers of @xmath77 and @xmath200 consistent with a parameter counting branes .
this suggests that a string theory underlying such solutions ( if it exists ) should have branes in _ both _ asymptotic regimes .
+ + + interestingly , algebraic @xmath294 solutions do not exist for @xmath299 .
nevertheless , we have demonstrated the existence of numerical solutions to the full equations in this and other cases .
see figure ( [ cwithg0full ] ) . these solutions decreases .
there are a number of qualitative features of this family of solutions that are akin to those seen in studies of the type 0a case in refs .
@xcite , when examining the case of @xmath300 .
] were obtained using the @xmath301 algorithm in @xmath302 .
one lesson learned is that the failure to find algebraic solutions in a t hooft limit is not a guarantee that smooth solutions to the full equations do not exist .
+ + 3 . @xmath264
+ figure ( [ c2g2 ] ) shows algebraic solutions where @xmath267 and @xmath249 .
these are @xmath303 and @xmath304 solutions in our notation .
+ the constraint @xmath305 on the algebraic equations lifts to @xmath306 on the full solutions .
the constant @xmath216 can not be determined from algebraic solutions alone . in this case , however , a thorough numerical analysis of the full equations is possible and it shows that @xmath307 .
examples of these numerical solutions are displayed in figure ( [ cplusg ] ) .
+ + + + the numerical solution strongly suggests that such new solutions are not just an artifact of the t hooft limit and should be taken seriously as new examples of non perturbatively complete solutions .
such solutions were not apparent in the perturbative analysis of ref .
@xcite and lead us to believe that the modified t hooft limit presented here is a good way of unearthing new non perturbative solutions .
we have presented a modification of the t hooft limit , first used by the authors of ref .
@xcite , to argue for the existence of new non perturbative solutions to string equations that are difficult to obtain numerically .
we have analyzed the painlev iv hierarchy of string equations introduced in ref .
@xcite in this limit , showing that examples of the conjectured type ii string theories coupled to the @xmath0 superconformal minimal models have well defined non perturbative solutions . as in the case of type 0a ,
higher @xmath26 solutions are likely to exist as a consequence of these@xcite , since the underlying integrable flow structure should evolve lower @xmath26 solutions into higher @xmath26 ones .
although this limit results only in the highest powers of the brane / flux parameter surviving ( so that we have only spherical topologies with boundaries or fluxes ) , it is likely that smooth solutions exist for the full string equations too .
the higher genus terms in the free energy are obtained by including the derivative corrections to the string equations and from the lesson of type 0a ( where we can compare to numerical solutions of the full equations see section [ sec : modlim ] ) it seems that they do not introduce any singular behaviour .
this can presumably be checked through further numerical work for the 0b and conjectured type ii theories .
we have uncovered a number of clear examples of new non perturbative solutions which also seem to be string like .
it would be interesting to determine if these indeed correspond to new string theories .
this could presumably be checked using perturbative techniques of the sort that we used in ref .
@xcite to identify the type ii theories .
we have demonstrated , in at least the @xmath287 case , that such new solutions are not simply an artifact of the t hooft limit , and that smooth solutions to the full equations can be obtained numerically obeying the same parameter constraints .
it would be interesting to find new solutions of this type numerically for higher @xmath55 .
this work was supported by the us department of energy .
we list the asymptotic expansions for @xmath287 , @xmath70 and @xmath66 here . for each @xmath55 , there are multiple expansions within each class that can be related to each other by various @xmath67 symmetries .
we list these explicitly for @xmath287 .
the symmetries for @xmath70 and @xmath66 are similar to those for @xmath287 and can be found in ref .
@xcite .
the other expansions within each class can be obtained by the following symmetry operation : @xmath311 since @xmath115 and @xmath119 contain only even powers of @xmath15 , the are invariant under this map ; however , @xmath312 .
thus there are two expansions in the @xmath116 class , and , together with @xmath115 and @xmath119 , these comprise four total @xmath287 expansions . in this case , there are four relevant classes of expansions @xmath313 here , we see that the string coupling is @xmath314 .
there are @xmath315 distinct expansions related to each other by various discrete symmetries as for @xmath287 .
the following five classes of expansions are obtained : @xmath316 here , the string coupling is @xmath317 .
there are a total of _ twenty five _ expansions related by various @xmath67 symmetries .
these have been explicitly listed in @xcite .
recall the interpretation given to the parameter @xmath15 of the 0a theory ( and also to @xmath52 of the 0b theory ) . in the @xmath318 regime
, @xmath15 represents the number of background zz d branes in the model , with a factor of @xmath15 for each boundary in the worldsheet expansion .
since an orientable surface with odd ( even ) euler characteristic must contain an odd ( even ) number of boundaries , @xmath15 must be raised to an odd ( even ) power if @xmath200 is .
in addition , the power of @xmath15 must be less than or equal to the power of @xmath200 . in the @xmath45 regime
, @xmath15 represents the number of units of r r flux in the background , with @xmath319 appearing when there is an insertion of pure r
so in this case both @xmath15 and @xmath200 should appear with even powers . in applying these observations to the dww expansions above
, we immediately notice the remarkable fact that the various expansions have powers of the parameters which somehow allow for interpretations as counting branes or fluxes .
this is by no means guaranteed , and indeed its occurrence was one of our main motivations for in depth study of the system .
the presence of two parameters , however , leads to a few subtleties .
for example , in some expansions an interpretation in terms of branes is only possible if one of the two parameters is set to zero .
we summarize these below .
* @xmath120 contains powers of @xmath15 consistent with a parameter counting branes .
this remains true for any value of @xmath77 .
* @xmath120 contains powers of @xmath77 consistent with a parameter counting fluxes .
however , for arbitrary @xmath15 , @xmath200 appears with odd powers , inconsistent with our requirements for a description of fluxes , as mentioned above .
this problem is avoided if we set @xmath244 since this forces the odd powers of @xmath200 to vanish . * for even @xmath55 , @xmath115 contains powers of @xmath15 and @xmath200 consistent with a parameter counting fluxes .
this is true for all values of @xmath77 . for odd @xmath55 ,
the powers of @xmath15 are still consistent with the flux interpretation , but there are odd powers of @xmath200 which are inconsistent with fluxes . these odd powers can be removed by setting @xmath259 . * for even @xmath55 , @xmath115 contains powers of @xmath77 and @xmath200 consistent with a parameter counting fluxes .
this is true for all values of @xmath15 . for odd @xmath55 ,
the powers of @xmath77 are consistent with a parameter counting branes . in this interpretation , there are no contributions from surfaces with only one boundary .
* @xmath116 contains powers of @xmath15 consistent with a parameter counting branes .
this is true only for @xmath259 .
* @xmath116 contains powers of @xmath77 consistent with a parameter counting branes .
this is true only for @xmath244 .
* we notice that associating one boundary to each factor of @xmath77 _ and _ @xmath15 also produces a consistent worldsheet expansion .
this encourages us to speculate these expansions might capture @xmath77 and @xmath15 simultaneously counting branes . *
@xmath119 contains powers of @xmath77 consistent with a parameter counting branes .
this remains true for any value of @xmath15 .
* @xmath119 contains powers of @xmath15 consistent with a parameter counting fluxes .
however , for arbitrary @xmath77 , @xmath200 appears with odd powers , inconsistent with our requirements for a description of fluxes , as mentioned above .
this problem is avoided if we set @xmath259 since this forces the odd powers of @xmath200 to vanish .
* @xmath320 contains powers of @xmath15 consistent with a parameter counting branes .
this is true only for @xmath259 .
* @xmath320 contains powers of @xmath77 consistent with a parameter counting branes .
this is true only for @xmath244 . * as for @xmath116 , we notice that associating one boundary to each factor of @xmath77 _ and _ @xmath15 also produces a consistent worldsheet expansion .
this encourages us to speculate these expansions might capture @xmath77 and @xmath15 simultaneously counting branes .
* these expansions do not exist for @xmath321 . c. m. hull and p. k. townsend , `` unity of superstring dualities , '' http://dx.doi.org/10.1016/0550-3213(94)00559-w[_nucl .
phys . _ * b438 * ( 1995 ) 109137 ] , http://arxiv.org/abs/hep-th/9410167[arxiv:hep-th/9410167 ] .
p. k. townsend , `` the eleven - dimensional supermembrane revisited , '' http://dx.doi.org/10.1016/0370-2693(95)00397-4[_phys .
lett . _ * b350 * ( 1995 ) 184187 ] , http://arxiv.org/abs/hep-th/9501068[arxiv:hep-th/9501068 ] .
e. witten , `` string theory dynamics in various dimensions , '' http://dx.doi.org/10.1016/0550-3213(95)00158-o[_nucl .
phys . _ * b443 * ( 1995 ) 85126 ] , http://arxiv.org/abs/hep-th/9503124[arxiv:hep-th/9503124 ] . t. r. morris , `` 2-d quantum gravity , multicritical matter and complex matrices , '' .
fermilab - pub-90 - 136-t .
s. dalley , c. v. johnson , and t. r. morris , `` multicritical complex matrix models and nonperturbative 2- d quantum gravity , '' http://dx.doi.org/10.1016/0550-3213(92)90217-y[_nucl .
phys . _ * b368 * ( 1992 ) 625654 ] .
s. dalley , c. v. johnson , and t. r. morris , `` nonperturbative two - dimensional quantum gravity , '' http://dx.doi.org/10.1016/0550-3213(92)90218-z[_nucl .
_ * b368 * ( 1992 ) 655670 ] . c. v. johnson , t. r. morris , and a. watterstam , `` global kdv flows and stable 2-d quantum gravity , '' http://dx.doi.org/10.1016/0370-2693(92)90112-h[_phys .
lett . _ * b291 * ( 1992 ) 1118 ] , http://arxiv.org/abs/hep-th/9205056[arxiv:hep-th/9205056 ] . s. dalley , c. v. johnson , t. r. morris , and a. watterstam , `` unitary matrix models and 2-d quantum gravity , '' http://dx.doi.org/10.1142/s0217732392002226[_mod
lett . _ * a7 * ( 1992 ) 27532762 ] , http://arxiv.org/abs/hep-th/9206060[arxiv:hep-th/9206060 ] . c. crnkovic , m. r. douglas , and g. w. moore , `` physical solutions for unitary matrix models , '' http://dx.doi.org/10.1016/0550-3213(91)90413-r[_nucl .
* b360 * ( 1991 ) 507523 ] .
t. j. hollowood , l. miramontes , a. pasquinucci , and c. nappi , `` hermitian vs. anti - hermitian 1-matrix models and their hierarchies , '' http://dx.doi.org/10.1016/0550-3213(92)90457-m[_nucl .
phys . _ * b373 * ( 1992 ) 247280 ] , http://arxiv.org/abs/hep-th/9109046[arxiv:hep-th/9109046 ] . i. r. klebanov , j. m. maldacena , and n. seiberg , `` unitary and complex matrix models as 1-d type 0 strings , '' http://dx.doi.org/10.1007/s00220-004-1183-7[_commun .
* 252 * ( 2004 ) 275323 ] , http://arxiv.org/abs/hep-th/0309168[arxiv:hep-th/0309168 ] .
d. j. gross and a. a. migdal , `` nonperturbative two - dimensional quantum gravity , '' http://dx.doi.org/10.1103/physrevlett.64.127[_phys .
* 64 * ( 1990 ) 127 ] .
e. brezin and v. a. kazakov , `` exactly solvable field theories of closed strings , '' http://dx.doi.org/10.1016/0370-2693(90)90818-q[_phys .
* b236 * ( 1990 ) 144150 ] .
m. r. douglas and s. h. shenker , `` strings in less than one - dimension , '' http://dx.doi.org/10.1016/0550-3213(90)90522-f[_nucl .
phys . _ * b335 * ( 1990 ) 635 ] .
j. e. carlisle , c. v. johnson , and j. s. pennington , `` baecklund transformations , d - branes , and fluxes in minimal type 0 strings , '' http://dx.doi.org/10.1088/1751-8113/40/41/013[_j .
phys . _ * a40 * ( 2007 ) 1245112462 ] , http://arxiv.org/abs/hep-th/0501006[arxiv:hep-th/0501006 ] . j. e. carlisle , c. v. johnson , and j. s. pennington , `` d - branes and fluxes in supersymmetric quantum mechanics , '' http://arxiv.org/abs/hep-th/0511002[arxiv:hep-th/0511002 ]
. j. m. maldacena , `` the large n limit of superconformal field theories and supergravity , '' http://dx.doi.org/10.1023/a:1026654312961[_adv .
* 2 * ( 1998 ) 231252 ] , http://arxiv.org/abs/hep-th/9711200[arxiv:hep-th/9711200 ] .
s. s. gubser , i. r. klebanov , and a. m. polyakov , `` gauge theory correlators from non - critical string theory , '' http://dx.doi.org/10.1016/s0370-2693(98)00377-3[_phys .
* b428 * ( 1998 ) 105114 ] , http://arxiv.org/abs/hep-th/9802109[arxiv:hep-th/9802109 ] . e. witten , `` anti - de sitter space and holography , '' _ adv .
* 2 * ( 1998 ) 253291 , http://arxiv.org/abs/hep-th/9802150[arxiv:hep-th/9802150 ] .
r. iyer , c. v. johnson , and j. s. pennington , `` string theory and water waves , '' http://arxiv.org/abs/1002.1120[arxiv:1002.1120 [ hep - th ] ] .
d. kaup , `` finding eigenvalue problems for solving nonlinear evolution equations , '' http://dx.doi.org/10.1143/ptp.54.72[_prog .
* 54 * ( 1975 ) 7278 ] .
r. gordoa , n. joshi , and a. pickering , `` on a generalized @xmath322 dispersive water wave hierarchy , '' http://dx.doi.org/10.2977/prims/1145477227[_publ .
inst . math .
* 37 * ( 2001 ) no . 3 , 327347 ] . i. m. gelfand and l. a. dikii , `` asymptotic behavior of the resolvent of sturm - liouville equations and the algebra of the korteweg - de vries equations , '' _ russ
. math . surveys _
* 30 * ( 1975 ) 77113 .
m. r. douglas , `` strings in less than one dimension and the generalized kdv hierarchies , '' http://dx.doi.org/10.1016/0370-2693(90)91716-o[_phys .
lett . _ * b238 * ( 1990 ) 176 ] .
t. banks , m. r. douglas , n. seiberg , and s. h. shenker , `` microscopic and macroscopic loops in non - perturbative two- dimensional gravity , '' http://dx.doi.org/10.1016/0370-2693(90)91736-u[_phys .
_ * b238 * ( 1990 ) 279 ] .
a. b. zamolodchikov and a. b. zamolodchikov , `` liouville field theory on a pseudosphere , '' http://arxiv.org/abs/hep-th/0101152[arxiv:hep-th/0101152 ] .
v. e. zakharov and a. b. shabat , `` integration of nonlinear equations of mathematical physics by the method of inverse scattering .
ii , '' _ funct .
* 13 * ( 1979 ) 166174 .
t. r. morris , `` multicritical matter from complex matrices , '' _ class .
grav . _ * 9 * ( 1992 ) 18731881 . c. v. johnson , `` string theory without branes , '' http://arxiv.org/abs/hep-th/0610223[arxiv:hep-th/0610223 ] | we use a combination of a t hooft limit and numerical methods to find non perturbative solutions of exactly solvable string theories , showing that perturbative solutions in different asymptotic regimes are connected by smooth interpolating functions .
our earlier perturbative work showed that a large class of minimal string theories arise as special limits of a painlev iv hierarchy of string equations that can be derived by a similarity reduction of the dispersive water wave hierarchy of differential equations .
the hierarchy of string equations contains new perturbative solutions , some of which were conjectured to be the type iia and iib string theories coupled to @xmath0 superconformal minimal models of type @xmath1 .
our present paper shows that these new theories have smooth non - perturbative extensions .
we also find evidence for putative new string theories that were not apparent in the perturbative analysis . * non - perturbative string theory from water waves * * ramakrishnan iyer@xmath2 , clifford v. johnson@xmath3 , jeffrey s. pennington@xmath4 * _ @xmath5department of physics and astronomy _ _ university of southern california _ _ los angeles , ca 90089 - 0484 , u.s.a . _ _ @xmath4slac national accelerator laboratory _ _ stanford university _ _ stanford , ca 94309 , u.s.a . _
@xmath2ramaiyer , @xmath3johnson1 , [ at ] usc.edu ; @xmath4jpennin [ at ] stanford.edu |
generating , preserving and controlling entanglement between two quantum systems is of great importance for quantum communications , teleportation , metrology , cryptology , computing and other operations involving quantum bits ( qubits ) @xcite .
a fundamental problem facing practical applications of entanglement is environment - induced dissipation of the quantum system .
the traditional view @xcite holds that dissipation is necessarily detrimental as it leads to decoherence , and thus to entanglement decay .
however , recently it has been realized that dissipation may have a positive effect as well . in particular
, it has been demonstrated that by engineering dissipation of the environment , quantum systems can be driven into the desired entangled steady state ( ss ) , encoding the outcome of quantum computations @xcite . in this case , the depopulation of the quantum system is compensated by constant pumping with an external electromagnetic field , and properties of the final states depend on the environment to which the systems are coupled . therefore , engineering of the environment seen by qubits is of crucial importance for applications in quantum optics . of particular interest
are structured photonic reservoirs , i.e. structures that have strongly inhomogeneous spatial or spectral distributions of their photonic density of states , as they offer an unprecedented level of control over emitter decay dynamics @xcite .
the efficient dissipative generation of entanglement has been theoretically predicted and experimentally demonstrated for qubits in optical cavities @xcite and photonic crystal systems @xcite . in particular , it was reported @xcite that dissipative preparation of entanglement between qubits in high - finesse optical cavities offers significant advantages for fidelity improvement over the protocols based on unitary dynamics ; decreasing the error by an order of magnitude requires only an order of magnitude increase of the cavity finesse in the former case , compared to a two order of magnitude increase of the finesse required in the latter case .
preparation of entangled states with fidelity ( with respect to the bell state @xmath1 ) exceeding 91.9@xmath2 was experimentally demonstrated @xcite .
though offering strong potential for creating highly - entangled states , the small size of optical cavities is a disadvantage when it comes to transferring quantum states over long distances , which is important for applications in quantum communications . for long distance entanglement , coupling of the qubits to reservoirs supporting propagating photonic modes , such as photonic crystals ,
is useful .
in particularly , efficient transfer of entanglement between two quantum dots ( qds ) placed inside a photonic crystal over a distance 100 times the wavelength of vacuum radiation has been theoretically predicted @xcite . despite the success of using all - dielectric photonic reservoirs for preparation and transfer of entanglement , the large size of such structures is a drawback for their use in nanophotonic integrated circuits . in order to overcome this difficulty ,
significant attention has been recently devoted to plasmonic waveguides as intermediaries for quantum state transfer .
nanoscale localization of plasmonics modes , as well as their relatively long propagation lengths , makes them good candidates for photonic circuits @xcite .
efficient plasmon - mediated entanglement has been predicted for emitters coupled to a metal or metamaterial slab @xcite , infinitely long v - shaped waveguides cut in a flat metal plane @xcite , infinitely long nanowires @xcite , and arrays of metallic nanospheres @xcite .
transfer of entanglement over distances far - exceeding the radiation wavelength was reported @xcite . and 15 nm , respectively ) and ( b ) a v - shaped channel cut in a flat metal plane ( if coupling slots are present , their length , width and depth are 70 nm , 15 nm , and 138 nm , respectively ) .
we consider both infinite- and finite - length waveguides.,width=432 ] one can ask if it is possible to optimize the geometry of a plasmonic waveguide in order to achieve stronger plasmon mediated entanglement between qubits .
moreover , realistic waveguides are never infinite , and can have discontinuities , accidental or by design .
therefore , it is important to understand what role waveguide edges and inhomogeneities can play in entanglement transfer , which is the subject of this paper .
we consider two quantum emitters ( modeled as two - level systems ) placed above 3d waveguides .
since there are a multitude of possible plasmonic waveguide geometries that one can envision , here we restrict our consideration to two waveguides that have been extensively studied in the literature , i.e. a metal nanowire and a v - shaped channel cut in a flat metal plane ( see fig .
[ fig1 ] ) . regarding long - distance entanglement ,
these waveguide geometries have been studied assuming that the nanowire and groove are infinite in their axial directions @xcite .
the main result of this work is to show that realistic finite - length nanowire and groove waveguides , with their associated discontinuities , play a crucial role in the engineering of highly entangled states .
we demonstrate that proper positioning of the emitters with respect to the waveguide edges can lead to a significant increase in entanglement compared to the case of the emitter coupled to an infinite plasmonic waveguide . moreover ,
even for the infinite - length case , discontinuities in the waveguides do not always play a detrimental role ; an increase in entanglement compared to the unperturbed case can be achieved by introducing coupling slots into the structure ( which aid in qubit - waveguide coupling ) . in the following we demonstrate the effect of coupling slots for both finite - length and infinite - length waveguides .
the paper is organized as follows . in section [ sec : theory ] we present the theory describing dissipation - induced entanglement between two emitters coupled to a plasmonic waveguide .
we briefly introduce the concept of the green dyadic ( sec .
[ sec : dyadic ] ) as an entity rigorously describing the spatial and spectral distribution of plasmonic modes in the dissipative structured reservoir formed by the waveguide .
we then present quantum master equations ( sec .
[ sec : transient ] ) that use the medium s green function , which describe the evolution of the density matrix of two emitters , one of which is initially in its excited state , placed near the lossy waveguide .
we solve these equations numerically and calculate the degree of entanglement induced by the plasmonic waveguide .
the entanglement presented in sec .
[ sec : transient ] is a transient one as it decays with time due to depopulation of the emitters excited states .
in order to obtain a steady entangled state , the emitters depopulation has to be compensated by constant pumping with an external electromagnetic field , which is the subject of sec .
[ sec : steady ] .
numerical results for entanglement are presented in sec .
[ sec : numer ] .
transient entanglement between emitters coupled to a silver nanowire or a v - shaped channel waveguide is studied in sec .
[ sec : nanowire ] and sec .
[ sec : groove ] , respectively , and results for steady - state entanglement are presented in sec .
[ sec : steady_numer ] .
in this section we introduce two concepts that are crucial for the theoretical description of the evolution of quantum systems coupled to lossy environments , namely , the dyadic green function @xcite and the quantum master equation @xcite .
we use these concepts in order to describe behavior of two identical two - level emitters ( qubits ) placed above 3d plasmonic waveguides ( see fig .
[ fig1 ] for the waveguide geometries ) . even though the process of spontaneous decay is essentially non - classical , and thus requires a full quantum description , the media the emitter
is coupled to can be rigorously incorporated into the quantum formalism through a classical quantity known as the dyadic green function @xmath3 @xcite .
the electric field green function is a 3-by-3 dyadic , each column of the dyadic being the electric field produced by a classic electric dipole polarized along the corresponding coordinate system axis .
the green dyadic satisfies @xmath4 where @xmath5 are the observation and source point vectors , respectively , @xmath6 is the vacuum wavenumber , @xmath7 is angular frequency , @xmath8 is the speed of light in vacuum , @xmath9 is relative permittivity , and @xmath10 is the unit 3-by-3 dyadic . the solution of can be written as @xmath11 where @xmath12 is the free - space green dyadic ( i.e. the solution of assuming @xmath13 ) , @xmath14 is the dyadic product , and @xmath15 is the `` scattered '' green function , which is the homogeneous solution of , accounting for the electric field scattered by the media .
it should be noted that although the real part of @xmath16 is singular when @xmath17 , the imaginary part is finite , with @xmath18 , @xmath19 .
in contrast to the free - space green dyadic , the scattered green dyadic @xmath15 does not have any singularities .
however , it depends both on the geometry and permittivity of objects filling the space , and thus does not have a closed analytic form in the general case . in order to calculate the green dyadic @xmath20 for the metallic plasmonic waveguides presented in fig .
[ fig1 ] we numerically solved maxwell equations using a commercial finite - difference time - domain method ( fdtd ) from lumerical solutions @xcite .
the importance of the classical green function is that it leads to the dissipative decay rate and coherent coupling terms @xmath21 here @xmath22 is the emitter transition dipole moment , @xmath23 and @xmath24 are the decay rates of the emitter @xmath25 due to its interaction with the reservoir ( which includes the plasmonic system ) , and plasmon mediated interactions with the emitter @xmath26 , respectively .
when the emitter itself is in a lossless region of space ( such as the space above the waveguide)@xmath27 does not have any singularity and thus the decay rate is always finite . the emitters transition frequency shift induced by dipole - dipole coupling
is given by @xmath28 .
we assume that the emitter transition frequency @xmath29 already accounts for the photonic lamb shift which is defined by @xmath30 .
in @xcite it was shown that for an infinite waveguide the best entanglement was obtained when @xmath31 was large and @xmath32 was small ( forming the dissipative regime ) .
for the finite waveguide case this classification does not hold exactly , which is discussed below ( e.g. , see fig .
[ fig2_2 ] ) . in the following we consider two identical emitters located near one of the plasmonic waveguides shown in fig .
[ fig1 ] and assume that the waveguide material is silver , which ia relatively low loss metal .
we choose the dipole transition frequency of the emitters to be @xmath33 thz ( emission wavelength @xmath34 nm ) .
the relative permittivity of silver at the dipole transition frequency is @xmath35 @xcite , and we fully account for the dispersive properties of the metal . for the nanowire waveguide
we choose the radius to be @xmath36 nm , such that the surface plasmon wavelength in the nanowire is @xmath37 nm , and the propagation length is @xmath38 , calculated analytically @xcite . for the groove waveguide ,
the depth of the groove is @xmath39 nm and its opening angle is @xmath40 ; these dimensions were chosen in @xcite to yield almost identical plasmonic wavelengths and propagation lengths for the rod and groove structures ( for the groove @xmath41 nm and @xmath38 ) , facilitating comparison between the waveguides .
the total emitter decay rate can be expressed as @xmath42 , in terms of the following emitter decay channels : ( a ) free - space radiation , @xmath43 , ( b ) joule losses in the metal , @xmath44 , and ( c ) excitation of surface plasmons , @xmath45 .
since radiative decay is not helpful for long - distance entanglement , and ohmic decay is obviously deleterious , it is clear that we want to maximize the plasmon decay channel , which enhances qubit - qubit interaction .
this requires careful positioning of the emitter above the nanowire surface ; if the emitter is too close to the nanowire surface , joule losses dominate the emitter decay ( quenching ) . on the other hand , if the emitter is too far away than it mostly radiates into free space .
this problem was studied in @xcite where it was demonstrated that at @xmath46 nm the optimum distance between the surface of an @xmath47 nm infinitely - long silver nanowire and a tuned emitter is 20 nm ( which maximizes @xmath45 ) , with dipole moment perpendicular to the wire , and for the groove structure the optimum emitter height is 12 nm above the surface of the metal , positioned over the central line of the groove and polarized horizontally .
for the finite nanowire the problem is more complicated and a unique definition of @xmath45 is difficult ( since plasmons reflect and refract from the nanowire edges ) .
we expect that the optimum distance should not be considerably different from the infinite - nanowire case , and we here choose the separation between the emitter and waveguide to be the same as the optimum height for the infinite structure , 20 nm for the nanowire and 12 nm for the groove .
although not shown here , slight improvement in entanglement can be obtained by adjusting the qubit height by a few nm . and decay rate @xmath48 as a function of distance between emitters above an infinite - length silver nanowire with and without coupling slots ( based on scattered green function ) .
the position of one emitter is fixed at @xmath49 and the second emitter position is changing with @xmath50 . slot are located at @xmath51 nm and @xmath52 nm @xmath53 677.5 nm.,width=480 ] although a primary aim of this work is to examine the effect of a finite plasmonic waveguide structure on entanglement , we first show that introducing discontinuities on an infinite waveguide can enhance qubit - waveguide coupling ( i.e. , in the following we refer to these discontinuities as coupling slots ) , thereby enhancing entanglement .
figure [ fig2_1 ] shows @xmath54 and @xmath55 for emitters placed above an infinite - length nanowire , where @xmath50 indicates the lateral separation between emitters .
one emitter is located at @xmath49 , and the position of the other emitter varies from @xmath49 to @xmath56 nm . note that here , and in fig .
[ fig2_2 ] below , we plot the rates based on the scattered green function ; this term is dominate over the vacuum green function for all cases of interest ( otherwise the plasmonic system would not affect the qubits much , such as , e.g. , if the height of the qubits was increased considerably ) , and it avoids the singularity in @xmath57 associated with the vacuum lamb shift .
however , in computing entanglement we use the full ( scattered plus vacuum ) green function .
oscillatory behavior of @xmath58 and @xmath59 as a function of qubit separation distance is evident in fig .
[ fig2_1 ] , with the period of oscillation being approximately equal to the plasmon wavelength . moreover , the positions of the @xmath60 maxima correspond to positions of @xmath59 minima .
thus coherent and dissipative regimes become dominant at different separations between emitters @xcite . for the case of a nanowire with coupling slots ,
the slots are fixed at @xmath51 nm and @xmath52 [email protected] nm ( one slot is to the right of the right emitter , and the other to the left of the left emitter ; @xmath50 refers to emitter - emitter separation ) .
coupling slots can lead to enhancement of entanglement between the emitters by increasing the dissipative decay rate at the emitter positions , as shown in the figure ( for the effect of coupling slots on entanglement , see fig .
[ fig4 ] below ) . by adding judiciously - chosen coupling slots to the infinitely long nanowire we are increasing emitter - waveguide coupling , similar to the improved field - plasmon coupling in the case of a grating , and thereby increasing emitter decay rates @xmath61 and @xmath48 into the plasmon channel .
it should be noted that coupling slots seem to affect only the dissipative regime , as the slot induced modification of the coherent exchange rate seems to be insignificant ( fig . [
fig2_1]a ) .
the normalization constant @xmath62 is the decay rate of the emitter in free - space .
it should be noted that the ratios @xmath54 and @xmath55 do not depend on the emitter dipole moment @xmath22 , since we are in the weak coupling regime .
it is likely that further improvement in emitter - plasmon coupling ( and , subsequently , entanglement ) can be made by optimizing the slot geometry .
in fact , qubit entanglement depends on many parameters even if the dipole transition frequency and nanowire diameter are fixed . in the case of an infinitely - long nanowire
these parameters are the distance between emitters and the nanowire surface , separation between emitters , and position and geometry of the coupling slots . in the case of a finite - length nanowire
we also have to consider the nanowire length , and positions of the emitters and coupling slots relative to the nanowire ends . in theory , optimal values of these parameters could be found though multi - variable optimization .
however , in this work we make no attempt to find optimal values ; we merely show that reasonably - positioned qubits and coupling slots can improve entanglement . .
the plasmon mode is excited by the emitter positioned at @xmath63 nm at a distance @xmath64 nm above the nanowire surface .
( b ) geometry , ( c)-(d ) normalized coupling strength @xmath59 and decay rate @xmath48 as a function of a distance between emitters above the nanowire with and without coupling slots ( based on scattered green function ) .
slot are located at @xmath65 nm and @xmath66 nm=610.5 nm.,width=480 ] for the finite - length nanowire we choose the nanowire length to be @xmath67 nm . we assume one emitter is located above one end of the nanowire ( @xmath68 nm ) .
if coupling slots are present , they are placed 27 nm away from each end of the nanowire ( this location leads to enhanced @xmath58 when qubits are at the nanowire ends ) .
the spatial distribution of the electric field intensity of the plasmon mode is shown in fig .
[ fig2_2]a . in order to calculate this distribution
we assume a classical electric dipole at the first emitter s position .
we can see that the positioned emitters experience maximum intensity of the plasmon electric field .
figures [ fig2_2](b)-(c ) show the dissipative and coherent rates as a function of emitter - emitter separation for the finite - length nanowire .
we can see that due to plasmon reflections from the nanowire edges , positions of maximums of @xmath59 and @xmath48 almost coincide with each other .
thus we can not clearly separate the dissipative and coherent regimes in the emitter dynamics anymore , unlike for the infinite - waveguide case . for emitters positioned over
the nanowire ends , the addition of coupling slots leads to an increase of coupling between emitters and surface plasmons .
moreover , in this case we observe a shift in the position of the maximums of the emitter coupling rates ( which does not occur for the infinite - length structure ) , as the slots change the effective length of the nanowire .
although not showm , similar results and conclusions are obtained for the groove waveguide .
let us consider two identical two - level emitters placed above the plasmonic waveguide ( see fig .
[ fig1 ] ) .
evolution of the system density matrix @xmath69 is described by the von neumann equation , @xmath70,\ ] ] where @xmath71 is the system hamiltonian , including emitters and plasmonic reservoir degrees of freedom , as well as coupling between them @xcite , @xmath72 here the index @xmath73 indicates the emitters , @xmath29 and @xmath74 are the frequency and transition dipole moment between the emitter ground @xmath75 and excited @xmath76 states , @xmath77 and @xmath78 are the emitter lowering and raising operators , respectively .
the operator of the electromagnetic field in the presence of the plasmonic reservoir is @xcite @xmath79 where @xmath80 are operators of elementary excitations of the reservoir , satisfying bosonic commutation relations @xmath81 = \delta_{ij } \delta(\omega_{\lambda}-\omega_{\lambda ' } ) \delta(\textbf{r}-\textbf{r}')$ ] and @xmath82 = 0 $ ] .
tracing out the electromagnetic field degrees of freedom in and using the born , markov and rotating wave approximations , we obtain master equations for the reduced @xmath83 density matrix as @xmath84 where @xmath85,\end{aligned}\ ] ] accounts for the interaction between the emitters and the plasmonic reservoir .
it should be noted that eq .
accounts for the emitter decay due to its interaction with the electromagnetic field of the plasmonic reservoir
. however , depending on the nature of the emitter , there can be other non - radiative channels for emitter decay .
in particular , in the case of qds , electron - phonon interaction leads to non - radiative quenching of emitter luminescence . in order to account for these processes
, we introduce pure dephasing phenomenologically , i.e. , @xmath86 where @xmath87 and where @xmath88 is a phenomenological pure - dephasing rate .
we show later that in the case of a typical qds above a plasmonic waveguide , the dephasing decay rate is generally small compared to the radiative decay rate and thus it s contribution to qd decay dynamics is insubstantial . unless specified otherwise ( e.g. , fig . [ fig6 ] ) , we assume it to be equal to zero .
equations can be solved analytically in the basis @xmath89 , @xmath90 , @xmath91 .
particularly , if initially the only non - zero elements of the density matrix are @xmath92 , @xmath93 , @xmath94 , @xmath95 , than the system of master equations is reduced to @xmath96 where @xmath97 .
this is the case when only one of the emitters is initially excited , such that @xmath98 . in order to characterize entanglement between two emitters we use the concept of concurrence @xcite , which is defined as @xmath99 where @xmath100 are arranged in descending order of the eigenvalues of the matrix @xmath101 , where @xmath102 is the spin - flip density matrix with @xmath103 being the pauli matrix .
concurrence may vary in the range between 0 ( unentangled state ) and 1 ( completely entangled ) , such that values between 0 and 1 correspond to different degrees of entanglement . if the only non - zero elements of the density matrix are defined by eqs . - , then concurrence is given by simplified expression @xmath104 ^ 2 + 4 \mathrm{im } [ \rho_{+-}]^2 } = \frac{e^{-\gamma_{aa } t}}{2 } \sqrt{\left[e^{-\gamma_{ab}t } - e^{\gamma_{ab}t}\right]^2 + 4 \sin^2(2 g_{ab}t)},\end{aligned}\ ] ] where we again assumed that only one of the emitters is initially in the excited state . as one can see from eq .
, the entanglement between two emitters above a plasmonic waveguide decays with time .
in order to obtain a steady entangled state we need to compensate for the depopulation of the emitters excited states ( via coupling to the environment ) by pumping with an external laser field . here
we restrict our consideration to the case when the external pumps are modeled by classical monochromatic waves of frequency @xmath105 , @xmath106 , and assume that each of the emitters can be pumped individually ( i.e. , using focused beams ) . in this case
, the master equation takes the form @xmath107 + \mathcal{k } \rho_s(t ) + \mathcal{l } \rho_s(t ) , \end{aligned}\ ] ] where @xmath108 accounts for the interaction between the classical pump field and qubits , @xmath109 is the effective rabi frequency of the pump , and @xmath110 is a detuning parameter . in the basis @xmath111 , @xmath112 , @xmath113 , @xmath114 we obtain a system of 16 coupled differential equations for the density matrix , which we solve numerically .
for the pumped case , eq . has to be used to calculate concurrence .
in order to calculate the transient concurrence we use the analytic equation .
steady - state concurrence is calculated using and density matrix elements are obtained by solving numerically the system of 16 differential equations . unless specified otherwise ( as in fig .
[ fig6 ] ) , the pure dephasing rate of the emitters , @xmath115 , is equal to zero .
we assume that the emitters are in vacuum ( solid green line ) , near an infinite wire with ( dashed brown line ) and without ( solid brown line ) coupling slots , near a finite wire of length of @xmath116 with ( solid red ) and without ( dashed blue line ) coupling slots . in the case of the finite wire ,
qubits are placed above the nanowire ends , and slots are located 27 nm from each end . the normalization constant @xmath117 is the decay rate of the emitter ( eq . ) for the given geometry .
( b ) time dependence of the population of the emitter initially in the excited , @xmath118 , and ground , @xmath119 , states , for the case of a finite nanowire with coupling slots .
, width=432 ] it was reported in ref .
@xcite that entanglement between emitters coupled to an infinite length plasmonic nanowire exceeds that between emitters in vacuum . as one can see from fig .
[ fig4]a entanglement can be improved even further by introducing discontinuities ( coupling slots ) in the nanowire , as described in sec .
[ sec : dyadic ] .
in particular , using an infinite nanowire with coupling slots ( see fig .
[ fig1 ] and sec .
[ sec : dyadic ] for the details of the slot geometry ) can considerably increase concurrence compared to the case with no slots , due to improved emitter - nanowire coupling .
[ fig4]a also shows results for the concurrence of two qubits near a finite - length nanowire , with length equal to @xmath120 , forming a fabry - prot ( fp ) cavity , with the qubits positioned above the ends of the wire .
it can be seen that by coupling into the fp nanowire resonances , we can considerably improve entanglement compared to the infinite - length case .
figure [ fig4]b shows evolution of the emitters excited state population for the case of a finite nanowire with coupling slots . here
@xmath121 is the probability of the first qubit to be in the excited state and the second qubit to be in ground state , and @xmath122 is similarly defined .
the initial state is @xmath123 .
we can see that decay of the emitter initially in the excited state leads to excitations of the emitter initially in the ground state , creating a transient entangled state , which is decaying with time . .
vacuum case ( dashed green line ) , infinite groove ( dotted black line ) , infinite groove with coupling slots ( dashed blue line ) @xmath124 nm laterally away from the qubits ( outside the space between them ) , and finite groove of length @xmath125 ( solid red line ) .
( b ) time dependence of the population of the emitter initially in the excited , @xmath118 , and ground , @xmath119 , states .
, width=432 ] a ) or finite - length groove ( fig .
[ fig5]a),width=384 ] the concurrence of two qubits separated by a distance @xmath126 nm is shown in fig .
[ fig5]a for an infinite groove , an infinite groove with coupling slots ( slots are positioned as in the nanowire case ) , and a @xmath125 groove . although the finite nanowire considered above had length @xmath127 , for the finite groove the length is @xmath125 because for the groove the modal field is not strong at the groove ends ( although not shown , the modal field distribution in the groove is more cavity - like , with a node at the groove end - wall ) , and so a @xmath125 groove allows the qubits to be separated by @xmath127 and also to be at the modal antinodes .
results for the finite - length groove are shown without slots since for this case coupling slots did not improve concurrence considerably ; for the nanowire the modal field is strong at the wire ends ( fig .
[ fig2_2 ] ) , but it is not exactly a modal antinode because of strong end - diffraction . in this case , coupling slots serve to shift the mode pattern slightly , enhancing qubit - plasmon coupling .
if the qubits are positioned exactly at a modal antinode , which is better defined away from the wire ends , coupling slots have a weaker effect on improving concurrence .
it can be seen that for the infinite groove , the addition of coupling slots leads to improved concurrence , and the fp resonances of the finite - length groove leads to further improvement . as reported in @xcite , for the chosen geometry the groove waveguide leads to stronger entanglement compared to the nanowire ( although this conclusion is likely dependent on metal absorption and geometry ) .
the population dynamics of the qubits are also shown for the finite groove case in fig .
[ fig5]b . in order to study the effect of pure dephasing
, we considered a qd with dipole moment 30 d as our model system .
it was reported @xcite that the dephasing rate , @xmath115 of the qd is equal to 1 @xmath128ev .
we considered emitters placed either above a finite nanowire ( fig .
[ fig4]a ) or a finite groove ( fig . [ fig5]a ) without coupling slots .
the emitter coupling rates for the case of the nanowire are @xmath129ev , @xmath130ev , @xmath131ev , while for the case of the groove waveguide @xmath132ev , @xmath133ev , @xmath134ev .
we can see that in general the pure dephasing decay rate is smaller than the radiative decay rate . even though it affects entanglement detrimentally ,
the decrease in concurrence is rather small ( see fig .
[ fig6 ] ) .
thus we otherwise ignore pure dephasing in our results , although the validity of this assumption depends on the qd system under consideration and the temperature .
up to this point we have considered spontaneous ( or vacuum - induced ) decay which leads to transient entanglement between two coupled qubits .
transient entanglement does not persist due to depopulation of the excited state of the qubits .
thus , in order to obtain a steady entangled state , depopulation has to be compensated by pumping qubits with an external laser pump in resonance with the frequency of the qubits dipole transition .
we assume the pump interacts only with the qubits . in this case
, the concurrence reaches a final , steady state ( ss ) value . .
the system geometry is the same as in figs .
[ fig4],[fig5].,width=528 ] figure [ fig7 ] shows the externally driven concurrence of qubits in the presence of a finite nanowire with length @xmath120 with qubits located at the nanowire ends ( the same geometry as in fig .
[ fig4 ] ) , a finite groove with length @xmath135 , qubits located symmetrically above its surface separated by @xmath120 ( the same geometry as in fig .
[ fig5 ] ) , and the infinite - length cases with the same qubit spacings as in the finite case .
three different coherent pumps differing in the relative phase of the laser field on qubit 1 and 2 have been considered : symmetric pumping indicates identical rabi frequencies , @xmath136 , antisymmetric pumping corresponds to @xmath137 and asymmetric pumping means @xmath138 . from fig .
[ fig7 ] , it is clear that the transient portion of the concurrence is intensity independent , and that the infinite groove has a higher transient peak compared to the infinite nanowire for all pumping regimes .
numerical experimentation shows that larger values of @xmath139 and @xmath60 at the positions of the qubits lead to larger transient concurrence peaks . for
the geometries studied , these mutual rates are larger for the infinite groove than for the infinite nanowire , which facilitates photon exchange between the qubits for the groove guide .
we can also see that the main advantage of a finite waveguide relative to an infinite one is significant improvement in the transient concurrence peak , with more modest improvement in the final , steady - state value .
, as function of qubit separation normalized by the modal wavelength , @xmath140 .
qubits are placed either above a finite nanowire ( right ) or a finite groove ( left ) .
the geometry is the same as in figs .
[ fig4],[fig5 ] .
asymmetric , antisymmetric and symmetric pumping have been considered.,width=528 ] figs .
[ fig8 ] shows the ss value of concurrence , @xmath141 , corresponding to different types of pumping as a function of qubit separation normalized to the plasmon wavelength , @xmath140 for the finite nanowire ( left panels ) and finite groove ( right panels ) .
numerical experimentation indicates that for steady state concurrence the preferred case is when we have large @xmath60 and small @xmath139 , in agreement with @xcite for the case of infinite waveguides .
therefore , the dissipative regime leads to larger transient entanglement peaks for infinite - length waveguides , and larger values of @xmath141 in general , and it is only for transient entanglement on finite - length waveguides that the dissipative and coherent regimes becomes similar . here , we can have steady state concurrence close to the peak of the transient value for relatively weak pumping .
for steady state concurrence , large @xmath139 and @xmath60 are less ideal , because of the tendency to decrease ss concurrence . furthermore
, pump strength can not be too large otherwise the qubits will interact mostly with the pump and become decoupled from each other .
ideally , the pump should be strong enough to keep the system interacting , but small enough for the qubit interaction to dominate the dynamics .
as made clear in fig .
[ fig8 ] , pumping one qubit strongly ( asymmetric pumping ) is possible , but pumping both qubits strongly ( symmetric or antisymmetric pumping ) reduces concurrence .
population dynamics of the qubits are shown in fig .
[ fig9 ] for the case when qubits are separated by @xmath116 and pumped either symmetrically , or only one qubit was pumped , for different intensities .
four density matrix elements are @xmath142 , @xmath143 , @xmath119 , and @xmath118 which are respectively the probability of both qubits to be in ground state , both qubits to be in excited state , the first qubit to be in the ground state and the second to be in the excited state , vice versa .
moreover we have assumed that initial state is a state when @xmath123 .
the qubits dynamics is different for one - field and two - field pumping . for the case of two - field pumping ,
in the steady state we have @xmath144 which is reasonable as we are pumping both systems identically .
we can see that these elements first grow rather fast with the growth of the pump intensity , but then slow down around @xmath145 .
we can also see that the element @xmath143 also grows under the influence of the pump .
but it first grows slower and then starts growing faster around @xmath145 .
therefore , this indicates that under such strong pumping the dynamics of both emitters is mostly defined by external pumps , and not by the qubit - qubit interactions , which is detrimental for entanglement . under the single field illumination the density matrix elements
evolve similarly to the previous case .
however , @xmath146 , because the first qubit that is pumped in this case .
we can see that with the intensity increase all elements grow , however , when the intensity becomes too strong the growth freezes and only @xmath118 grows , indicating that dynamics of first emitter is entirely defined by the pump field , which is also detrimental for entanglement .
( see fig .
[ fig5 ] for details of the geometry ) .
two different regimes of pumping have been considered , symmetric pumping ( two field pumping ) and asymmetric pumping ( single field pumping).,width=528 ]
transient and steady state entanglement of two qubits mediated by two different types of plasmonic waveguides have been presented .
the dynamics of the qubits has been studied using the formalism of the master equations .
the waveguides , acting as photonic reservoirs , were incorporated into the formalism through the classical dyadic green function , accounting for dissipative and coherent interactions between qubits and the reservoir .
the green dyadic has been calculated numerically .
two different methods have been proposed to improve the entanglement between qubits communicating through excited surface plasmonic waves ; using a finite , resonant - length plasmonic waveguide , and including coupling slots near the sites of the qubits to improve the coupling of the qubits to the plasmonic modes .
the calculated results are compared with the infinite - length versions of the waveguides , and the vacuum case , and it is shown that a finite structure or slotted structure can improve the entanglement of appropriately - positioned qubits .
this work was supported by the natural sciences and engineering research council of canada and queen s university . | entanglement between two qubits ( two level atoms ) mediated by surface plasmons in three - dimensional plasmonic waveguides is studied using a quantum master equation formalism .
two types of waveguides , a nanowire and a v - shaped channel cut in a flat metal plane , are considered .
the green functions for the waveguides , which rigorously describes the dissipative qubit environment , are calculated numerically using a direct finite - difference time - domain ( fdtd ) solution of maxwell s equations .
finite - length effects are shown to play a crucial role in enhancing entanglement , and resonant - length plasmonic waveguides can provide higher entanglement between qubits than infinite - length waveguides . it is also shown that coupling slots can improve entanglement via stronger qubit - waveguide coupling , for both the infinite- and finite - waveguide cases
. 10 l. aolita , f. de melo , and l. davidovich , `` open - system dynamics of entanglement : a key issues review , '' rep . prog . phys . *
78 * , 042001 ( 2015 ) .
k. yczkowski , p. horodecki , m. horodecki , and r. horodecki , `` dynamics of quantum entanglement , '' phys .
rev .
a * 65 * , 012101 ( 2001 ) .
f. verstraete , m. m. wolf , and j. ignacio cirac , `` quantum computation and quantum - state engineering driven by dissipation , '' nat . phys . * 5 * , 633636 ( 2009 ) .
j. t. barreiro , p. schindler , o. guhne , t. monz , m. chwalla , c. f. roos , m. hennrich , and r. blatt , `` experimental multiparticle entanglement dynamics induced by decoherence , '' nat .
phys .
* 6 * , 943946 ( 2010 ) .
y. lin , j. p. gaebler , f. reiter , t. r. tan , r. bowler , a. s. sorensen , d. leibfried , and d. j. wineland , `` dissipative production of a maximally entangled steady state of two quantum bits , '' nature * 504 * , 415418 ( 2013 ) .
h. walther , b. t. h. varcoe , b .-
g .
englert , and t. becker , `` cavity quantum electrodynamics , '' rep . prog .
phys . * 69 * , 1325 ( 2006 ) .
p. anger , p. bharadwaj , and l. novotny , `` enhancement and quenching of single - molecule fluorescence , '' phys .
rev .
lett . * 96 * , 113002 ( 2006 ) .
a. v. akimov , a. mukherjee , c. l. yu , d. e. chang , a. s. zibrov , p. r. hemmer , h. park , and m. d. lukin , `` generation of single optical plasmons in metallic nanowires coupled to quantum dots , '' nature * 450 * , 402406 ( 2007 ) .
a. m. nemilentsau , g. y. slepyan , s. a. maksimenko , a. lakhtakia , and s. v. rotkin , `` spontaneous decay of the excited state of an emitter near a finite - length metallic carbon nanotube , '' phys .
rev .
b * 82 * , 235411 ( 2010 ) .
e. forati , g. w. hanson , and s. hughes , `` graphene as a tunable thz reservoir for shaping the mollow triplet of an artificial atom via plasmonic effects , '' phys .
rev .
b * 90 * , 085414 ( 2014 ) .
m. j. kastoryano , f. reiter , and a. s. srensen , `` dissipative preparation of entanglement in optical cavities , '' phys .
rev .
lett .
* 106 * , 090502 ( 2011 ) .
b. casabone , a. stute , k. friebe , b. brandsttter , k. schppert , r. blatt , and t. e. northup , `` heralded entanglement of two ions in an optical cavity , '' phys .
. lett . * 111 * , 100505 ( 2013 ) .
r. reimann , w. alt , t. kampschulte , t. macha , l. ratschbacher , n. thau , s. yoon , and d. meschede , `` cavity - modified collective rayleigh scattering of two atoms , '' phys .
rev .
lett . * 114 * , 023601 ( 2015 ) .
p.
yao and s. hughes , `` macroscopic entanglement and violation of bell s inequalities between two spatially separated quantum dots in a planar photonic crystal system , '' opt . express * 17 * , 1150511514 ( 2009 ) .
j. wolters , j. kabuss , a. knorr , and o. benson , `` deterministic and robust entanglement of nitrogen - vacancy centers using low-@xmath0 photonic - crystal cavities , '' phys .
rev .
a * 89 * , 060303 ( 2014 ) .
g. angelatos and s. hughes , `` entanglement dynamics and multiple mollow triplets between two coupled quantum dots in a nanowire photonic crystal system , '' arxiv e - prints ( 2014 ) .
m. s. tame , k. r. mcenery , s. k. ozdemir , j. lee , s. a. maier , and m. s. kim , `` quantum plasmonics , '' nat . phys .
* 9 * , 329340 ( 2013 ) .
j. xu , m. al - amri , y. yang , s .- y .
zhu , and m. s. zubairy , `` entanglement generation between two atoms via surface modes , '' phys .
rev .
a * 84 * , 032334 ( 2011 ) .
a. gonzalez - tudela , d. martin - cano , e. moreno , l. martin - moreno , c. tejedor , and f. j. garcia - vidal , `` entanglement of two qubits mediated by one - dimensional plasmonic waveguides , '' phys .
rev .
lett . * 106 * , 020501 ( 2011 ) . d. martin - cano , a. gonzalez - tudela , l. martin - moreno , f. j. garcia - vidal , c. tejedor , and e. moreno , `` dissipation - driven generation of two - qubit entanglement mediated by plasmonic waveguides , '' phys . rev .
b * 84 * , 235306 ( 2011 ) .
d. dzsotjan , a. s. srensen , and m. fleischhauer , `` quantum emitters coupled to surface plasmons of a nanowire : a green s function approach , '' phys .
rev .
b * 82 * , 075427 ( 2010 )
. h. zheng and h. u. baranger , `` persistent quantum beats and long - distance entanglement from waveguide - mediated interactions , '' phys .
rev .
lett . * 110 * , 113601 ( 2013 ) . c. gonzalez - ballestero ,
e. moreno , and f. j. garcia - vidal , `` generation , manipulation , and detection of two - qubit entanglement in waveguide qed , '' phys .
rev .
a * 89 * , 042328 ( 2014 ) . c. lee ,
m. tame , c. noh , j. lim , s. a. maier , j. lee , and d. g. angelakis , `` robust - to - loss entanglement generation using a quantum plasmonic nanoparticle array , '' new journal of physics * 15 * , 083017 ( 2013 ) .
h. t. dung , l. knll , and d .-
g .
welsch , `` three - dimensional quantization of the electromagnetic field in dispersive and absorbing inhomogeneous dielectrics , '' phys . rev .
a * 57 * , 3931 ( 1998 ) .
h. t. dung , l. knll , and d .-
g .
welsch , `` spontaneous decay in the presence of dispersing and absorbing bodies : general theory and application to a spherical cavity , '' phys .
rev .
a * 62 * , 053804 ( 2000 ) . http://www.lumerical.com / tcad - products / fdtd/. d. w. lynch and w. hunter , `` comments on the optical constants of metals and an introduction to the data for several metals , '' in `` handbook of optical constants of solids , '' , e. d. palik , ed .
( academic press , burlington , 1997 ) , pp .
275 367 .
l. novotny , `` effective wavelength scaling for optical antennas , '' phys .
rev .
lett . * 98 * , 266802 ( 2007 ) .
w. k. wootters , `` entanglement of formation of an arbitrary state of two qubits , '' phys .
rev .
lett . *
80 * , 22452248 ( 1998 ) . |
collective decision making has been evidenced in many animal species and contexts @xcite including food collection @xcite , problem - solving @xcite , collective movement @xcite or nest site selection @xcite . in this later case ,
social animals have to select a resting sites among severals potential options in a complex environment .
this selection can be made either through individual decisions or complex decision - making processes involving the participation of all individuals @xcite and can be temporary or permanent according to the needs and living style of the considered species . while this process of collective decision has been studied for a long time in social animals that select a permanent home ( social insects for example )
, only few studies address this problem in the case of nomad species that constantly explore their environment such as birds or fish .
in particular , experiments on fish have been generally designed to observe preferences for particular environmental features or landmarks during a relatively short experimental time ( few seconds @xcite , 5 minutes @xcite , 10 minutes @xcite , up to 30 minutes per trial @xcite ) .
these studies have shown for example that landmarks in a bare tank arouse interest and attract the fish @xcite and that variations of the shape of these landmarks can change territory features @xcite . while these studies provide numerous insights on the individual and collective preferences in fish groups , they generally rely on the comparison between two or more qualitatively different alternatives .
thus , the selection of one option is often based on an intrinsic preference for a particular feature in comparison with the others .
such asymmetric choices may hide the collective decision that results from the internal processes of decision - making of the group . here
, our aim is to characterise the collective behaviour of groups of zebrafish swimming in an environment with identical landmarks .
we observe the collective motion of small shoals of different group sizes ( 5 and 10 fish ) and of two different zebrafish strains ( wild type ab or tl ) .
zebrafish are gregarious vertebrate model organisms to study the cohesion of the group and its decision making @xcite .
originating from india where they live in small groups in shallow freshwaters @xcite , the zebrafish is a diurnal species that prefers staying in groups both in nature and in the laboratory @xcite . to be closer to their natural swimming habits
, our experimental method relies on the observation of fish freely swimming in an open environment rather than in a constraining set - up ( i.e. maze as used in @xcite ) .
we observe during one hour each group of fish swimming in a large experimental tank with two spots of interest ( landmarks ) .
the landmarks consist of two striped yellow - green opaque plastic cylinders placed in the water column or two blue transparent floating perspex disks providing shadow .
we choose these colours in the visible spectrum of the zebrafish according to the results of @xcite .
we expect that these landmarks placed in a homogeneous environment could induce a choice of one prefered option by the zebrafish as evidenced for other species faced with identical ressources @xcite . on the one hand , we test with cylinders the effect of visual and physical cues in the water column on collective choices .
since zebrafish are known to swim along the walls of experimental tanks , cylinders could act as such walls in the water column . on the other hand , we test with floating disks the impact of visual and physical cues above water , on collective choices .
we placed disks and cylinders landmarks to see whether and how the two strains of zebrafish will adapt their group behaviour and their preferences for landmarks .
we quantified behavioural properties of groups of zebrafish from high resolution videos with the help of a tracking system and an automated analysis . following the method of @xcite , we track in real time the positions of the fish and automatically compute their probabilities of presence in the tank , their interindividual distances , the number of individual present near the landmarks as well as the duration of their visits .
then , we compare the distributions obtained for each group size and landmark to highlight differences between strains of zebrafish at the collective level .
this methodology based on massive data gathering has now become standard in studies on animal collective behaviour with flies , _ drosophila melanogaster _
@xcite , birds , _ sturnus vulgaris _ @xcite and fish , _ notemigonus crysoleucas _ @xcite .
in the presence of two striped yellow and green cylinders ( see methods ) , the groups of 10 ab zebrafish are mainly present along the wall of the tank and around the cylinders , as shown by their average probability of presence computed for the one hour observation period and 10 replicates ( figure [ fig : all_pdfcylinders ] ) . on the contrary , groups of 5 tl , 10 tl and 5 ab zebrafish
are less observed near the cylinders but are still present along the walls of the tank ( figure [ fig : all_pdfcylinders ] ) .
the probabilities of presence computed for each experiment are shown in the annexe ( figure [ fig : figures1 ] , figure [ fig : figures2 ] , figure [ fig : figures3 ] and figure [ fig : figures4 ] ) .
= 10 measures for 5 ab , 10 ab , 5 tl and 10 tl .
* = @xmath0 , * * = @xmath1 , * * * = @xmath2 , ns = non significant.,scaledwidth=50.0% ] to highlight the differential effect of the cylinders on the tested groups , we measured the proportion of positions that were detected at 25 cm from the centre of the cylinders ( figure [ fig : anova_cylinders_strains_size ] ) .
this measure confirms that groups of 10 ab zebrafish were more present near the cylinders than groups of 5 ab .
in contrast , while groups of 5 tl responded similarly to groups of 5 ab , groups of 10 tl zebrafish were less detected near the cylinders .
a two - way anova ( group size , strain , @xmath3 for each experimental conditions ) indicated a non - significative effect of the group size ( @xmath4 , f = 3.59 , df = 1 ) but a significant effect of the strain ( @xmath2 , f = 43.17 , df = 1 ) and a significant interaction strain / group size ( @xmath2 , f = 35.38 , df = 1 ) on the attractivity of the cylinders .
the interaction effect indicates here a strain - specific effect of the group size on the time spent near the landmarks : groups of 10 ab are more attracted by the cylinders than groups of 5 ab but on the contrary , 10 tl are less detected near the cylinders than groups of 5 tl .
then , we studied in more details the dynamics of the presence near the cylinders of the zebrafish swimming in group of 10 individuals . in the ab strain , the fish form a cohesive group that regularly transits from one landmark to the other at the beginning of the trial .
then , the group starts to split in multiple subgroups and the periodicity of the visits becomes less regular ( figure [ fig : cylindersexploit_temporal]a ) .
these oscillations are also observed for groups of the tl strain but contrarily to ab zebrafish , this phenomenon is observed for the whole experimental time ( figure [ fig : cylindersexploit_temporal]b ) . to quantify the dynamics of these transitions , we computed the number of majority events detected near one of the two landmarks or outside of them .
a majority event was counted when 7 or more individuals were simultaneously present in the same zone independently of the duration of this majority event .
figure [ fig : cylindersexploit_symbdyn]a shows that the median and mean number of majority event is always smaller in the tl strain , but this difference is only significant for the majorities detected near one of the cylinder ( cylinder 1 , mann - whitney u test , u = 26 , @xmath5 ; cylinder 2 , mann - whitney u test , u = 48 , @xmath6 ; outside , mann - whitney u test , u = 38 , @xmath7 ) .
then , we characterised the transitions of the fish from one landmark to the other by analysing the succession of majority events .
in particular , we counted the number of `` collective transitions '' ( two majority events nearby different cylinders separated by a majority outside ) , the number of `` one - by - one transitions '' ( succession of a majority event in one cylinder and a majority event in the other cylinder ) and finally the number of `` collective u - turns '' ( two majority events in the same cylinder separated by a majority outside ) .
this reveals that the main transitions occurring for ab zebrafish are the collective ones while some collective u - turns and almost no individual transitions were detected ( figure [ fig : cylindersexploit_symbdyn]b red ) .
similarly , almost no individual transition were observed for the tl zebrafish that perform mainly collective transitions ( figure [ fig : cylindersexploit_symbdyn]b green ) .
tl zebrafish performed also numerous collective u - turns .
the absence of individual transitions reveals that both strains are mostly swimming in groups but with different collective dynamics .
we compared the results with mann - whitney u tests : `` one - by - one transitions '' ( u = 3.0 , @xmath2 ) and `` collective u - turns '' ( u = 27.5 , @xmath8 ) between ab and tl are significantly different when `` collective transitions '' between ab and tl are not ( u = 40 , @xmath9 ) . finally , we computed the interindividual distances between all fish to characterize the cohesion of the group for both strains and group sizes .
the distribution of all interindividual distances ( figure [ fig : all_interdistdistricylinders ] ) shows that groups of tl zebrafish have a stronger cohesion ( 5 tl : median = 0.12 m and 10 tl : median = 0.14 m ) than the groups of ab zebrafish ( 5 ab : median = 0.27 m and 10 ab : median = 0.23 m ) .
the intra - strain comparison for the two group sizes shows that the distribution significantly differs from each other ( kolmogorov - smirnov test , 5 ab vs 10 ab , d = 0.102 , @xmath2 ; 5 tl vs 10 tl , d = 0.080 , @xmath2 ) .
the inter - strain comparison for similar group sizes also reveals a statistical difference between the distributions ( kolmogorov - smirnov test , 10 ab vs 10 tl , d = 0.184 , @xmath2 ; 5 ab vs 5 tl , d = 0.161 , @xmath2 ) .
the distributions of the average interindividual distance measured at each time step confirms these results ( figure [ fig : all_averageinterdistdistibutioncylinders ] , kolmogorov - smirnov test , 5 ab vs 10 ab , d = 0.433 , @xmath2 ; 5 tl vs 10 tl , d = 0.051 , @xmath2 ; 10 ab vs 10 tl , d = 0.333 , @xmath2 ; 5 ab vs 5 tl , d = 0.464 , @xmath2 . in addition , the analysis of the evolution of the average interindividual distance revealed that the cohesion of the fish decreases for both strains and both group sizes through the time ( figure [ fig : figures11 ] of the annexe ) .
we also placed symmetrically two floating identical perspex transparent blue disks , at 25@xmath10 cm from two opposite corners along the diagonal .
acting as roofs on the water , they create shades that could attract zebrafish .
we did 10 trials on groups of 10 ab and 10 tl zebrafish .
we observed similar behaviours than in the previous experiments with cylinders .
the maximum probability of presence under disks with ab zebrafish reaches 4.5@xmath11 when for tl zebrafish it reaches only 1@xmath11 ( figure [ fig : all_pdfdisks ] ) .
again , tl zebrafish spent the majority of their time near the borders of the tank ( the probabilities of presence for each experiment are shown in the figure [ fig : figures5 ] and figure [ fig : figures6 ] of the annexe ) .
we compared the probability to be near the spots for both strains and both types of landmarks with a two - way anova ( figure [ fig : anova_cylinders_disks_strains_size ] ) .
it revealed that the type of landmarks ( @xmath2 , f = 11.37 , df = 1 ) and the strain of zebrafish affect the attraction ( @xmath2 , f = 102.95 , df = 1 ) while there is also an evidence of an interaction effect between type of landmarks and strains ( @xmath12 , f = 5.67 , df = 1 ) .
thus , while groups of 10 ab are attracted by cylinders as much as by disks , groups of 10 tl are more attracted by cylinders than disks . eventually , tl zebrafish were also significantly more cohesive than the ab zebrafish in the presence of the floating disks , as shown by the distribution of the interindividual distances ( medians for 10 ab : 0.35 m , for 10 tl : 0.23 m , figure [ fig : all_interdistdistridisks ] , kolmogorov - smirnov test , d = 0.135 and @xmath2 ) .
, * * = @xmath1 , * * * = @xmath2 , ns = non significant.,scaledwidth=50.0% ]
we investigated whether collective motion and collective choice can differ in groups of 5 and 10 individuals of the same species ( zebrafish _ danio rerio _ ) but of different strains ( ab versus tl ) .
both strains were laboratory wild types , had the same age and were raised under the same conditions .
we observed zebrafish swimming for one hour in the presence of two identical landmarks ( immersed cylinders or floating disks ) that were put symmetrically in the tank .
although one hour observations show that the zebrafish do not select one of the two landmarks , the fish were mainly swimming together and oscillate from one landmark to the other with short resting times .
thus , while all individuals can be punctually present at the same landmark ( figure [ fig : cylindersexploit_temporal ] , figure [ fig : figures7 ] and figure [ fig : figures8 ] of the annexe ) , the probability of presence computed for the entire experimental time shows that the fish were equally present at both stimuli .
therefore , no collective choice emerged on the long time for both strains of zebrafish and group size .
hence , long time and short time range analyses reveal opposite results and conclusions on collective motions .
our methodology is complementary to typical y - maze experiments .
we extend and compare their conclusions to our observations with repeated interactions between the fish and their environment . during a hour ,
the collective behaviour of zebrafish contrasts with other collective species in which spatial fidelity emerges from the interactions between the individuals that take place in the resting sites ( in cockroaches @xcite , in hymenoptera @xcite ) .
these oscillations from one site to the other could originate from individual differences among group members : _ bold _ and _ shy _ behavioural profiles have been evidenced in zebrafish according to the intrinsic propensity of each fish to explore new environments @xcite .
it also has been identified in other fish species @xcite . in this context , the presence of bolder fish in the group could favour the transition from one spot to the other while groups composed only by shy individuals could show less frequent departures @xcite .
a more detailed analysis show quantitative differences among the two studied strains and group sizes .
concerning the response of the fish to the landmarks , we highlight that groups of 5 ab and tl zebrafish show the same attraction for the cylinders by computing the probability for the fish to be observed near these landmarks .
this attraction increases for groups of 10 ab zebrafish but decreases for groups of 10 tl zebrafish .
this strain difference is also observed in the experiments with floating disks .
in addition , the type of the landmarks seems to be determinant for tl zebrafish as they prefer objects immersed in the water column than objects lying on the surface of the water .
hence , it exists a difference of collective behaviour between the two tested strains of zebrafish .
this different response to heterogeneities may be based on the intrinsic preference of the fish of a particular strain for congeners or for landmarks .
such difference has already been shown in shoaling tendency between several strains of guppies @xcite and zebrafish @xcite , @xcite . in their natural environment ,
fish have to balance the costs of risks and benefits of moving in groups or staying near landmarks @xcite . moving in groups first of all
prevents the fish to be static preys , and second allow spatial recognition and easier food and predators detection .
the drawback is that the takeover of the fish on the territory is punctual and they have less chance to find areas where they can hide .
staying around landmarks gives the fish a feeling of control of the territory and the possibility to hide from predators . in that case
the drawback is that the preys will rarely cross the territory of the fish . regarding the structure of the group , we notice that whatever the group size of tl zebrafish , the cohesion of the group does not change and is always stronger than those of groups of ab zebrafish .
also , the bigger the group of ab zebrafish , the stronger the cohesion .
these differences of group cohesion may be based on physical features differences between ab and tl zebrafish .
cohesion differences could be explained by phenotype differences between the two strains .
some studies demonstrated that a large variability exists in the individual motion and shoaling tendency of the zebrafish according to their age or strain .
for example , adults ab and casper zebrafish swim longer distance than abstrg , ek , tu or wik zebrafish @xcite .
likewise , the interindividual distance between shoal members decreases from 16 body length to 3.5 body length between day 7 and 5 months after fertilization @xcite .
also , it has been demonstrated that the fin size has an impact on the swimming performance and behaviour of the zebrafish @xcite .
starting from the assumptions that ab and tl zebrafish have the same visual acuity and the same lateral line performance , the only source of separation between both strains are their fin lengths and the patterns on the skin : tl zebrafish are homozygous for leo@xmath13 and lof@xmath14 , where leo@xmath13 is a recessive mutation causing spotting in adult zebrafish and lof@xmath14 is a dominant homozygous viable mutation causing long fins @xcite , @xcite .
thus , ab zebrafish have short fins and tl zebrafish show long fins .
it may suggest that tl zebrafish move a higher quantity of water with their long fins when swimming and thus emit a stronger signal of presence ( hydrodynamical signal ) .
thus , it may be easier for conspecifics in the moving shoal to perceive the signal through their lateral line and realign themselves according to their conspecifics . if the realignment becomes easier
, it is simpler for tl zebrafish to keep their position in the shoal , which increases its cohesion . following a similar hypothesis ,
the signal of presence is weaker for ab zebrafish due to their shorter fins .
thus , realignment in the moving shoal is less performant and their cohesion decreases .
each of the two hypotheses could explain the collective behaviours observed during the experiments and nothing prevents merging both of them . in conclusion
, this study demonstrates that behavioural differences exist at the individual and collective levels in the same species of animal .
the analysis of the dynamics reveals that ab and tl zebrafish mainly oscillate in groups between landmarks .
in addition , increasing the size of the group leads to opposite results for the two strains : groups of 10 ab zebrafish are proportionally more detected near the landmarks than groups of 5 ab while groups of 10 tl zebrafish are less attracted by the landmarks than groups of 5 tl .
finally , the two tested zebrafish strains show differences at the structural level : ( 1 ) groups of tl zebrafish are more cohesive than groups of ab zebrafish and ( 2 ) ab zebrafish collective responses to landmarks show that they are generally more present near the cylinders and floating disks than tl zebrafish .
thus , this study provides evidences that zebrafish do not select resting site on the midterm and highlights behavioural differences at the individual and collective levels among the two tested strains of zebrafish .
future studies of collective behaviour should consider the tested strains , the intra - strain composition of the shoals and the duration of each trial .
the experiments performed in this study were conducted under the authorisation of the buffon ethical committee ( registered to the french national ethical committee for animal experiments # 40 ) after submission to the state ethical board for animal experiments .
we acquired 500 adult wild - type zebrafish ( 200 ab strain and 300 tl strain ) from institut curie ( paris ) and raised them in tanks of 60l by groups of 50 .
the zebrafish ab line show a zebra skin , short tail and fin .
the zebrafish tl line show a spotted skin , long tail and fin and barbel .
both strains are 3.5 cm long .
the zebrafish used for the experiments are adult fish between 5 months and 18 months of age . during this period ,
zebrafish show a shoaling tendency allowing study of their collective behaviours .
we kept fish under laboratory condition , @xmath15 , 500@xmath16s salinity with a 9:15 day : night light cycle .
the fish were reared in 55 litres tanks and were fed two times per day ( special diets services sds-400 scientic fish food ) .
water ph is maintained at 7.5 and nitrites ( no@xmath17 ) are below 0.3 mg / l .
we measured the size of the caudal fins of 10 ab ( about 0.4 cm ) and 10 tl ( about 1.1 cm ) zebrafish .
the experimental tank consists in a 1.2 m x 1.2 m tank confined in a 2 m x 2 m x 2.35 m experimental area surrounded by a white sheet , in order to isolate the experiments and homogenise luminosity .
the wall of the experimental tank were covered with white tape and the water column is 6 cm .
water ph is maintained at 7.5 and nitrites ( no@xmath17 ) are below 0.3 mg / l . the experiments with disks were performed in the experimental tank while those with cylinders were performed in a white square arena ( 1 m x 1 m x 0.15 m ) placed in the experimental tank .
groups of zebrafish were randomly formed at the beginning of the experiments .
the experiments were recorded by a high resolution camera ( 2048 x 2048 px , basler scout aca2040 - 25gm ) placed above the experimental tank and recording at 15 fps ( frame per second ) .
luminosity is ensured by 4 fluorescents lamps of 80w placed on each side of the tank , on the floor and directed towards the walls to provide indirect lightning . to trigger interest of fish
, we placed symmetrically in the set - up either two floating disks ( @xmath18 = 20 cm ) or two cylinders ( @xmath18 = 10 cm , height = 15 cm ) surrounded by yellow and green striped tape @xcite . to avoid the presence of a blind zone ,
the cylinders were slightly tilted toward the centre of the tank .
the center of both disks and cylinders are at 25@xmath10 cm from two opposite corners along the diagonal of the tank ( figure [ fig : set - up ] ) .
we recorded the behaviour of zebrafish swimming in the experimental tank during one hour . before the trials ,
the attractive landmarks are put in the setup and fish are placed with a hand net in a cylindrical arena ( 20 cm diameter ) made of plexiglas placed in the centre of our tank . following a five minutes acclimatisation period ,
this arena is removed and the fish are able to freely swim in the experimental arena .
we performed 10 trials for each strain with the floating disks and 10 trials for each combination of parameters ( number of fish x strain ) with the cylinders for a total of 60 experiments .
each fish was never tested twice in the same experimental condition .
the experiments with cylinders were recorded at 15 fps and tracked online by a custom made tracking system based on blob detection .
we call a batch a group of 10 experiments . for these batchs ,
each experiment consists of 540000 positions ( 10 zebrafish x 54000 frames ) and 270000 positions ( 5 zebrafish x 54000 frames ) . for experiments with disks , we faced tracking troubles .
since the fish below the floating disks were difficult to distinguish by the program due to a lack of sufficient contrast , experiments with floating disk were tracked offline by two custom matlab scripts .
a first script automatically identifies the positions of the fish swimming outside of the floating disks by blob detection .
since this method did not allow a perfect detection of all the individuals , we developed a second script that was run after the first one and that plotted the frame where a fish ( or more ) was undetected for the user to manually identify the missing individual(s ) .
it allowed us to identify the fish that were partially hidden during a collision / superposition with another fish or the fish that were situated under the floating disks . since this analysis tool is time - costly , we only analysed 1 fps for all experiments with disks . for these batchs ,
each experiment consists of 36000 positions ( 10 zebrafish x 3600 positions ) .
since our tracking system did not solve collision with accuracy , we did not calculate individual measures but characterised the aggregation level of the group .
the probability of presence of the fish was calculated by the cumulated positions of all individuals along the entire experiment .
we also calculated the distance between the fish and the attractive landmarks as well as the inter - individual distances between the fish and the average inter - individual distance .
finally , we computed the time of shelter occupancy as the time that is spent by the fish at less than 25 cm of the attractive landmarks .
these time sequences were calculated according to the number of fish present near the landmarks .
all scripts were coded in python using scientific and statistic libraries ( numpy , pylab , scilab and matplotlib ) . to compute the number of majority events ,
the number of fish was average over the 15 frames of every second .
this operation garanties that a majority event is ended by the departure of a fish and not by an error of detection during one frame by the tracking system .
figure [ fig : figures9 ] and figure [ fig : figures10 ] of the annexe show the proportions of the durations of the the majority events before and after this interpolation .
.*means of percentages of tracking efficiencies*. [ cols="^,^",options="header " , ] for the figures [ fig : anova_cylinders_strains_size ] and [ fig : anova_cylinders_disks_strains_size ] , 10 measures of means of the probability for different groups of zebrafish to be near the landmarks are ploted .
they have been tested using a two - way anova .
we then compared the data between each group using a one - way anova and finally used a tukey s honest significant difference criterion post - hoc test .
we did these tests on matlab and chose 0.001 as significance level . in the table [
fig : efficiency ] , we show the number of majority events after interpolation of the data at 1 fps . this table is related to the figure [ fig : cylindersexploit_symbdyn ] .
we used mann - whitney u tests to compare the number of events between strains , areas and transition types .
these tests are performed on 10 values of majority events for each strain , area and transition type .
these tests were made with the python package scipy .
we chose 0.001 ( * * * ) , 0.01 ( * * ) and 0.05 ( * ) as significance levels .
for the figure [ fig : all_interdistdistricylinders ] as well as the figures [ fig : all_averageinterdistdistibutioncylinders ] and [ fig : all_interdistdistridisks ] , we compared the distribution with kolmogorov - smirnov tests .
these tests were made with the python package scipy .
we chose 0.001 as significance level .
the authors thank filippo del bene ( institut curie , paris , france ) who provided us the fish observed in the experiments reported in this paper .
this work was supported by european union information and communication technologies project assisibf , fp7-ict - fet-601074 .
the funders had no role in study design , data collection and analysis , decision to publish , or preparation of the manuscript .
supplementary figures of `` strains differences in the collective behaviour of zebrafish ( _ danio rerio _ ) in heterogeneous environment '' .
m. ballerini , n. cabibbo , r. candelier , a. cavagna , e. cisbani , i. giardina , v. lecomte , a. orlandi , g. parisi , a. procaccini , m. viale , and v. zdravkovic .
interaction ruling animal collective behavior depends on topological rather than metric distance : evidence from a field study .
, 105:12321237 , 2007 .
a. strandburg - peshkin , c.r .
twomey , n.w.f .
bode , a.b .
kao , y. katz , c.c ioannou , s.b .
rosenthal , c.j .
torney , h.s .
levin , and i.d .
visual sensory networks and effective information transfer in animal groups .
, 23:r709 , 2013 .
s. millot , m. -l .
bgout and b. chatain risk - taking behaviour variation over time in sea bass _ dicentrarchus labrax _ : effects of day night alternation , fish phenotypic characteristics and selection for growth . 75:17331749 , 2009 . | recent studies show differences in individual motion and shoaling tendency between strains of the same species . here , we analyse the collective motion and the response to visual stimuli in two morphologically different strains ( tl and ab ) of zebrafish . for both strains ,
we observe 10 groups of 5 and 10 zebrafish swimming freely in a large experimental tank with two identical attractive landmarks ( cylinders or disks ) for one hour .
we track the positions of the fish by an automated tracking method and compute several metrics at the group level .
first , the probability of presence shows that both strains avoid free space and are more likely to swim in the vicinity of the walls of the tank and the attractive landmarks .
second , the analysis of landmarks occupancy shows that ab zebrafish are more present in their vicinity than tl zebrafish and that both strains regularly transit from one landmark to the other with no preference on the long duration .
finally , tl zebrafish show a higher cohesion than ab zebrafish .
thus , landmarks and duration of the repicates allow to reveal collective behavioural variabilities among different strains of zebrafish .
these results provide a new insight into the need to take into account individual variability of zebrafish strains for studying collective behaviour . |
throughout this text @xmath9 and @xmath2 are finitely connected jordan domains in the complex plane @xmath10 .
each boundary @xmath11 and @xmath12 consists of @xmath13 disjoint jordan curves .
we shall consider orientation preserving homeomorphisms @xmath14{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\,\mathbb y\,$ ] of finite dirichlet energy @xmath15 = \iint_\mathbb x |\,dh(z)|^2\ ; { \textnormal d}x \,{\textnormal d}y \;<\;\infty , \qquad z = x+iy\ ] ] where @xmath16 stands for the hilbert - schmidt norm of the derivative matrix @xmath17 .
such a mapping has a continuous extension @xmath18{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{{\mathbb{y}}}$ ] which is not necessarily a homeomorphism @xcite . in this paper
we show that @xmath19 can be strongly approximated by homeomorphisms between closed domains , provided @xmath20 is inner chordarc regular , see definition [ ica ]
. in particular , @xmath20 can be a lipschitz domain .
[ main ] let @xmath21 and @xmath20 be finitely connected jordan domains , @xmath20 being inner chordarc .
let @xmath22{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] be a homeomorphism in the sobolev space @xmath23 . then
: a. there exist homeomorphisms @xmath24{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,,\ ; k = 1 , 2 , \dots,\,$ ] that converge to @xmath25{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] uniformly and strongly in @xmath26 .
b. moreover , for every compact subset @xmath27 , we have @xmath28 on @xmath29 , provided @xmath30 is sufficiently large .
if , in addition , @xmath31 is injective on a compact subset @xmath32 , then @xmath33 can be chosen so that @xmath28 on @xmath34 .
the motivation for part ( c ) comes from variational problems for mappings between quadrilaterals , i.e. , jordan domains with four distinguished points which form the injectivity set @xmath35 .
we will pursue such applications of theorem [ main ] elsewhere .
let us compare theorem [ main ] to the known results in this field .
the problem of approximation of sobolev homeomorphisms was raised by j. m. ball and l. c. evans in 1990s @xcite .
there have been several recent advances in this direction @xcite . in particular ,
in @xcite we proved that any @xmath36-homeomorphism @xmath37{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb v$ ] between open subsets of @xmath38 can be approximated in the @xmath36 norm by diffeomorphisms of these open subsets .
the new feature of theorem [ main ] is that approximation takes place also on the boundary , where the original map need not be a homeomorphism .
this explains why theorem [ main ] imposes regularity assumptions on the boundaries of @xmath21 and @xmath20 .
combining theorem [ main ] with ( * ? ? ?
* theorem 1.2 ) yields the following result .
if @xmath39{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] is a @xmath40diffeomorphism in @xmath41 then , in addition to all properties listed in theorem [ main ] , the mappings @xmath42{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] can also be found as @xmath40 -diffeomorphisms .
if @xmath19 is only a homeomorphism , such an approximation by diffeomorphisms is still available , except for the property ( b ) .
one of the fundamental problems in topology is to approximate continuous mappings by homeomorphisms .
the approximation procedures , still only partially understood , have led topologists to the concept of monotone mappings @xcite and somewhat subtle concept of cellular mappings @xcite .
we refer the interested reader to @xcite . in the mathematical theory of hyperelasticity ,
on the other hand , we are concerned with the energy - minimal deformations @xmath43{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{y}}$ ] , so having additional sobolev type regularity . however ,
very often the injectivity of the energy minimal mappings is lost , though they enjoy some features of homeomorphisms , like monotonicity . in particular , the question of approximation of a monotone mapping in the sobolev space @xmath41 by homeomorphisms @xmath44{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{{\mathbb{y}}}$ ] gains interest in the mathematical models of elasticity . a novelty in these directions is the following corollary of theorem [ main ] and ( * ? ? ?
* theorem 1.6 ) .
suppose @xmath39{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] lies in the sobolev space @xmath41 and extends to a continuous monotone map @xmath45{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}$ ] .
then there exist homeomorphisms @xmath24{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,,\ ; k = 1 , 2 , \dots,\,$ ] that converge to @xmath25{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] uniformly and strongly in @xmath26 .
does theorem [ main ] remain valid when target @xmath20 is an arbitrary finitely connected jordan domain ? can theorem [ main ] be extended to sobolev spaces @xmath36 , @xmath46 ? or to dimensions @xmath47 ?
the royden algebra @xmath48 consists of continuous functions @xmath49 which have finite energy .
the norm is given by @xmath50 let @xmath51 denote the completion of @xmath52 in this norm .
any conformal mapping @xmath53{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{x}}\,$ ] between jordan domains induces an isometry @xmath54{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathscr r(\mathbb x')$ ] by the rule @xmath55 .
thus the domain of definition of @xmath7 is permitted to be changed by any conformal transformation .
the following observation will allow us to transform the target .
[ folklore ] let @xmath56{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{y}}\,$ ] be homeomorphisms in @xmath48 and @xmath57{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{{\mathbb{y}}}'\,$ ] be a @xmath58-diffeomorphism that extends to a homeomorphism @xmath59{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{{\mathbb{y}}}'\,$ ] .
assume that both the gradient matrix @xmath60 and its inverse @xmath61 are bounded in @xmath62 .
then @xmath63 in @xmath48 if and only if @xmath64 in @xmath48
. our primary appliance for strong approximation in theorem [ main ] will be local harmonic replacements near the boundaries of @xmath9 .
we will rely on a well - known fact , see @xcite . in a finitely
connected jordan domain @xmath65 , we consider a function @xmath66 .
let @xmath67 denote the continuous harmonic extension of the boundary map @xmath68 into @xmath9 .
then @xmath69 .
moreover , @xmath70\ ; = \,\,\mathscr e_{_\mathbb x } [ g]\;-\ ; \mathscr e_{_\mathbb x } [ h ] \;\leqslant \ ; \mathscr e_{_\mathbb x } [ g]\,.\ ] ] [ ica ] a finitely connected jordan domain @xmath20 is _ inner chordarc _ if there exists a constant @xmath71 with the following property .
suppose that @xmath72 belong to the same boundary component of @xmath20 and @xmath73 is an open jordan arc with endpoints at @xmath74 and @xmath75 .
then the shortest connection from @xmath76 to @xmath77 along @xmath78 has length at most @xmath79 .
inner chordarc domains were studied in geometric function theory since 1980s @xcite .
they are more general than lipschitz domains .
for instance they allow inward cusps and logarithmic spiraling , see figure [ exa1 ] .
[ thmica]@xcite a simply connected jordan domain @xmath80 is inner chordarc if and only if there exists a @xmath58-diffeomorphism @xmath81 from @xmath82 onto the unit disk @xmath83 that extends to a homeomorphism @xmath84{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb d}\,$ ] such that both gradient matrices @xmath85 and @xmath86 are bounded in @xmath82 .
fugure [ exa1 ] illustrates two such mappings from the unit disk onto a non - lipschitz domain : @xmath87 ( the image is a spiral domain ) and @xmath88 ( the image contains an inward cusp ) .
recall that a continuous map @xmath89{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\,\overline{\mathbb y}\,$ ] is monotone if the preimage of every continuum ( connected compact set ) in @xmath90 is a continuum in @xmath91 .
a point @xmath92 is said to be a _ simple value _ of @xmath7 if its preimage @xmath93 is a single point in @xmath94 .
we need an elementary lemma .
[ monotone ] let @xmath95{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] be a homeomorphism between @xmath13-connected jordan domains and suppose that it extends continuously up to the boundary . then the extended map @xmath96{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] is monotone .
furthermore , the inverse map @xmath97{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb x\,$ ] extends continuously up to simple values of @xmath7 which are virtually all points in @xmath98 , except for a countable number of them .
every homeomorphism with finite dirichlet energy has a continuous extension to the boundary @xcite .
precisely , [ extension ] every finite energy homeomorphism @xmath95{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] between @xmath99 -connected jordan domains extends to a continuous map between the closures , again denoted by @xmath96{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] .
this map is monotone , though the inverse @xmath100{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb x\,$ ] may not admit continuous extension to the closure of @xmath2 , unless it also has finite energy .
we shall make use of the following strong version of the rad - kneser - choquet theorem , see @xcite .
[ rkc ] let @xmath101 be a simply connected jordan domain and @xmath102 a bounded convex domain .
suppose we are given a continuous monotone map @xmath103{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\omega\,$ ] , not necessarily a homeomorphism .
then its continuous harmonic extension , denoted by @xmath104 , defines a @xmath40 -diffeomorphism @xmath105{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\omega\,$ ] .
we reserve the notation , @xmath106 @xmath107 the components are numbered so that @xmath19 sends @xmath108 onto @xmath109 for @xmath110 .
figure [ ram ] illustrates a domain @xmath20 that satisfies the assumptions of the theorem .
we recall that the boundary map @xmath111{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb y\,$ ] is injective on a compact subset @xmath112 , possibly empty .
let us fix an @xmath113 and a compact @xmath114 .
we need to construct a homeomorphism @xmath115{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] of sobolev class @xmath116 which coincides with @xmath7 on @xmath117 and @xmath118 hereafter the symbol @xmath119 indicates that the inequality holds with an _ implied _ multiplicative constant .
the implied constants will vary from line to line but remain * independent of @xmath120 * as long as @xmath120 is sufficiently small . to prove we shall set up a chain of homeomorphisms @xmath121{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] in the sobolev space @xmath122 whose continuous extensions , still denoted by @xmath123{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] , satisfy + thus the final term @xmath134 works for the desired homeomorphism @xmath115{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] , by the triangle inequality .
we proceed by induction on @xmath135 .
the induction begins with @xmath136 , which is obvious .
suppose we are given the map @xmath137{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}$ ] for some @xmath138 .
the construction of @xmath139{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb y}\,$ ] will be made in 5 steps .
+ * step i. * ( _ transition to the case when @xmath140 is the outer boundary _ ) . to make this transition rigorous ,
let us perform the following transformations of @xmath90 .
first , we reduce ourselves to the case in which @xmath141 is the outer boundary of @xmath2 by applying an inversion if necessary . once @xmath142 is the outer boundary of @xmath143 we apply theorem [ thmica ] to transform the bounded component of @xmath144 , denoted by @xmath80 , onto the unit disk .
let this transformation be denoted by @xmath145{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb d$ ] .
this map extends to a homeomorphism @xmath146{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb d}$ ] , and has both matrix functions @xmath85 and @xmath86 continuous and bounded in @xmath82 .
let @xmath147 .
by virtue of lemma [ folklore ] the composition with the mapping @xmath148{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{y}}'$ ] transforms converging sequences of mappings @xmath149{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{y}}$ ] into converging sequences of mappings @xmath150{{}_{\!\!\textnormal{onto\,\,}\!\!}}}{\mathbb{y}}'$ ] .
the inverse @xmath151 has the same property ; therefore we can work with the target @xmath152 instead of @xmath20 . in what follows
we simply assume that @xmath153 is the outer boundary of @xmath2 . +
* step ii . *
( _ harmonic replacements near _
@xmath154 ) . the idea is to alter @xmath155 in a thin neighborhood of @xmath154 to gain piecewise harmonicity therein . in this step
we change neither the boundary map @xmath156{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\ , \mathbb y\,$ ] nor the values of @xmath157 on the given compact @xmath158 .
recall from lemma [ monotone ] that all but a countable number of points in @xmath159 are simple values of the map @xmath160{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb t\,$ ] .
they are dense , so one can partition @xmath161 into arbitrarily small _ closed _ circular arcs whose ends are simple values of @xmath155 , @xmath162 let @xmath163 denote the segment of the unit disk ; the open region between the arc @xmath164 and the closed line interval @xmath165 connecting the endpoints of @xmath166 , which we call the base of @xmath167 .
we require the partition of @xmath161 to be fine enough so that the compact set @xmath168 intersects none of the segments @xmath169 . one further restriction on the partition comes from the following observation : the finer the partition the closer to @xmath161 are the segments @xmath163 .
since @xmath170{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y\,$ ] is a homeomorphism , it follows that the preimages of @xmath163 under @xmath155 , denoted by @xmath171 , can be as close to @xmath154 as we wish .
in particular , we may ensure that @xmath172 that is all what we require to determine the partition of @xmath173 .
this partition will remain fixed for the rest of the proof .
now we observe that each @xmath174 is a simply connected jordan domain .
its boundary consists of two closed jordan arcs with common endpoints .
the one in @xmath154 is denoted by @xmath175 and the open arc in @xmath176 is denoted by @xmath177 .
it is at this point that we take advantage of the condition that the endpoints of @xmath166 are simple values of @xmath178 .
this condition implies that the inverse map @xmath179{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\gamma_\kappa } \,$ ] is a homeomorphism . on the other hand
the preimage @xmath180{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\gamma_\kappa \subset \mathfrak x_{\nu + 1}\,$ ] is a closed arc , because of monotonicity of @xmath181{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\mathbb y\,$ ] .
therefore , the open jordan arc @xmath182 and the closed arc @xmath183 form a closed jordan curve ; precisely , the boundary of @xmath174 . in summary ,
* @xmath184{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb s_\kappa\ , , $ ] * @xmath185{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\mathbb s_\kappa\,\;\;\textnormal{is continuous and monotone}$ ] * this latter boundary map is injective on the compact subset + @xmath186 now we appeal to theorem [ rkc ] of rad - kneser - choquet which allows us to replace @xmath187{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb s_\kappa\,$ ] by the harmonic extension of its boundary map @xmath188{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\mathbb s_\kappa\,$ ] . we need to introduce , for a little while , more notation .
* @xmath189{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\,$ ] harmonic extension of @xmath188{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\mathbb s_\kappa\,$ ] * @xmath190{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb y \,$ ] homeomorphism of class @xmath191 , defined by @xmath192 let us estimate the difference @xmath196 in the norm of royden algebra .
first we find that @xmath197 secondly , in view of ( [ energy ] ) , we see that @xmath198 = \sum_{\kappa = 1}^{n } \mathscr e_{_{\mathbb x_\kappa } } [ h^{\nu } - h^{\nu}_\kappa ] \,\leqslant \sum_{\kappa = 1}^{n } \mathscr e_{_{\mathbb x_\kappa } } [ h^{\nu } ] \,\preccurlyeq \varepsilon^{\,2}\ ] ] by ( [ energy ] ) .
hence @xmath199 summarizing , the construction of @xmath200 in an @xmath120 proximity to @xmath201 will be done once the similar construction is in hand for @xmath202 . in what follows , instead of using @xmath202 , we assume that the original map @xmath155 was already harmonic in every @xmath174 .
this simplifies writing and causes no loss of generality . +
* step iii . *
( _ reduction of the domain to the unit disk _ ) .
the idea is to construct for each @xmath203 a sequence of homeomorphisms @xmath204{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\ , , \;j = 1 , 2 , \dots \,,\,$ ] that converge to @xmath205{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\,$ ] uniformly and in @xmath206 .
in addition to that , we require that each @xmath207 agrees with @xmath178 on the compact subset @xmath186 . recall that the boundary map @xmath185{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial \,\mathbb s_\kappa\,$ ] is injective on @xmath208 .
once this is done , the construction of @xmath209 will be completed in the following way : for each @xmath210 we choose and fix @xmath211 sufficiently large so that @xmath212 then we replace each @xmath205{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\,$ ] by @xmath213{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\,$ ] to obtain the desired map @xmath214 thus we are reduced to finding the sequence @xmath215{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\ , , \;j = 1 , 2 , \dots \,$ ] . before proceeding to
somewhat involved computation we need to simplify the domain and the target of @xmath205{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}_\kappa\,$ ] .
since the problem is clearly unaffected by a rotation of the target ( harmonicity of the map is not compromised ) , we may confine ourselves to the segment of the form @xmath216 thus its arc @xmath217 is @xmath218 , the base @xmath219 is @xmath220 and the corners are @xmath221 . regarding the domain @xmath174 , it is legitimate to conformally transform it onto the unit disk @xmath222 ; any conformal mapping between two jordan domains induces an isometry of their royden algebras .
thus we consider a conformal map @xmath223{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb x_\kappa\,$ ] and the pullback @xmath224{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb s\,$ ] .
this is a homeomorphism up to the boundaries .
one extra condition turns out to be useful ; namely , we may choose @xmath225 to be normalized at three boundary points so that the map @xmath226 satisfies , @xmath227 the first two values of @xmath228 are the endpoints of the base @xmath229 and last one is its midpoint .
let us state clearly what we aim to show .
[ propo ] let @xmath230{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb s\,$ ] be an orientation preserving harmonic homeomorphism of finite energy and let its continuous extension @xmath231{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}\,$ ] satisfy .
thus @xmath232 maps the closed arc @xmath233 monotonically onto itself and @xmath234 homeomorphically onto the base @xmath229 .
suppose , in addition , that @xmath235{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c\,$ ] is injective on a compact subset @xmath236 .
then there exist homeomorphisms @xmath237{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb d}\,$ ] converging to @xmath231{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\overline{\mathbb s}\,$ ] uniformly and in the sobolev space @xmath238 .
moreover @xmath239 on @xmath234 and @xmath240 , for @xmath241
. * step iv . *
( _ good approximation of the boundary map @xmath242{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb s\,$ ] _ ) .
the aim is to properly approximate the continuous monotone boundary map @xmath242{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb s\,$ ] by homeomorphisms @xmath243{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb s\,$ ] which agree with @xmath244 on both sets @xmath245 and @xmath246 .
once such an approximation is in hand , we shall extend each @xmath247 harmonically inside the disk .
certainly , uniform convergence @xmath248 on @xmath161 would suffice to deduce uniform convergence in the entire disk , by the maximum principle .
however , it is not obvious at all how to make the approximation of the boundary map in order to control the energy of the extended mappings .
the douglas criterion will come into play .
we need only work to construct homeomorphisms @xmath249{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c\,$ ] , as their values on @xmath250 are already known ; @xmath251{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\,\mathbf t\,$ ] .
let us write @xmath235{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c\,$ ] as @xmath252 , where @xmath253 { \xrightarrow[]{{}_{\!\!\textnormal{onto\,\,}\!\!}}}[-\omega\ , , \omega ] \,$ ] is a nondecreasing continuous function such that @xmath254 and @xmath255 .
we also require that @xmath256 on @xmath236 .
the complement @xmath257 consists of a countable number of disjoint open circular arcs ( components ) whose endpoints belong to @xmath240 .
let @xmath258 be one of such arcs together with its endpoints , where @xmath259 , and similarly for the image @xmath260 .
the latter strict inequality is justified by the fact that @xmath261 and @xmath262 are points in @xmath263 and the map @xmath228 , being injective on @xmath240 , assumes distinct values @xmath264 and @xmath265 at these points .
we now define homeomorphisms between closed arcs @xmath266{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c^{\,\phi(\beta)}_{\phi(\alpha ) } \,$ ] ( @xmath267 ) by the rule ; @xmath268 , where @xmath269\cdot \big [ \phi(\theta ) - \phi(\alpha ) \big]\;+\ ; \frac{\phi(\beta ) -
\phi(\alpha)}{m}\,\cdot \big(\theta - \alpha\big ) \;+\ ; \phi(\alpha)\ ] ] for @xmath270 .
we have @xmath271 and @xmath272 .
the first term defining @xmath273 is nondecreasing in @xmath274 while the second term is strictly increasing . therefore , @xmath266{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c^{\,\phi(\beta)}_{\phi(\alpha ) } \,$ ] is a homeomorphism .
formula ( [ deff / of / fm ] ) , applied to every arc component of @xmath257 , gives homeomorphisms which agree with @xmath228 at the endpoints of the arcs .
we glue them together with @xmath228 at the endpoints to obtain a homeomorphism @xmath275{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb t\,$ ] which coincides with @xmath276{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb t\,$ ] on @xmath277 .
further analysis of @xmath247 is necessary to deduce proper convergence as @xmath278 .
first note that on each arc component of @xmath257 we have @xmath279 hence latexmath:[\[\label{uniform } [ mess ] for all @xmath281 and @xmath282 , we have @xmath283 and @xmath284 belong to the closure of the same component of @xmath285 ; say , @xmath286 and @xmath287 , where @xmath288 .
it follows from formula ( [ deff / of / fm ] ) that @xmath289 0.3 cm _ * case 2 . * _ both @xmath290 and @xmath291 belong to @xmath292 .
we assume that the closed set @xmath293 is not empty .
otherwise , @xmath294 and @xmath284 would belong to the same arc component of @xmath285 .
let us set the notation , @xmath295 @xmath296 and note that the points @xmath297 belong to the closure of one arc component in @xmath285 .
the same applies to the pair @xmath298 .
therefore , using _ case 1 _ , one can write the following chain of inequalities since @xmath300 and @xmath301 lie in the shorter circular arc that connects @xmath294 to @xmath284 , it follows that @xmath302 the same argument applies to the images of these point under the monotone map @xmath303{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathcal c\,$ ] .
thus , @xmath304 and also @xmath305 .
substitute these inequalities into the chain above to conclude with the desired inequality @xmath306 _ * case 3 .
* _ let @xmath309 and @xmath310 .
by symmetry we may take @xmath311 , where @xmath312 .
we may also assume that @xmath313 . otherwise , inequality ( [ estimate for douglas ] ) holds ; namely , @xmath314 .
geometrically , the assumption @xmath315 tells us that @xmath284 can not lay in the lower half of the arc @xmath316 .
thus @xmath291 , where @xmath317 .
let the upper corner of the segment @xmath318 be denoted by @xmath319 .
the location of @xmath320 is restricted to the upper half of the base of the segment @xmath318 , because @xmath321{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\mathbb t \,$ ] is monotone and @xmath322 ( the midpoint of the base ) due to normalization at ( [ bconditions ] ) . regarding the position of @xmath323
, we may assume that this value also lies in the upper half of the arc @xmath292 .
otherwise , we would have @xmath324 while , on the other hand , @xmath325 which implies .
we are ready to complete case 3 .
first we use case 2 and the triangle inequality , @xmath326 then comes a geometric fact about the term within the curled braces .
certainly , we have @xmath327 , because @xmath228 is injective on @xmath328 .
if , incidentally , @xmath329 then the latter estimate yields .
thus we may assume that three points @xmath330 and @xmath331 are vertices of a triangle .
let @xmath332 and @xmath333 . since @xmath334 lies in the arc of @xmath318
, @xmath335 lies in the base of @xmath318 and @xmath336 is the corner of @xmath337 , all of them in the upper half of @xmath318 , it follows ( from geometry of the segment @xmath318 ) that the angle opposite to the side @xmath338 , denoted by @xmath339 , satisfies : @xmath340 .
the law of cosines tells us that @xmath341 hence @xmath342 completing the proof of lemma [ mess ] .
the boundary homeomorphisms @xmath344{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb s\,$ ] will now be extended harmonically inside the unit disk .
we use the same label for the extensions , @xmath345{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\,\overline{\mathbb s } \,$ ] .
these mappings are homeomorphisms , due to theorem [ rkc ] .
since both @xmath228 and @xmath247 are harmonic in @xmath346 , the sequence @xmath247 converges to @xmath228 uniformly in @xmath347 , by the maximum principle .
the key point here is that they also belong to the sobolev space @xmath348 , and converge in the sobolev norm as well . to see this we recall the douglas criterion @xcite which asserts that any function @xmath349 that is continuous on @xmath347 and harmonic in @xmath346 satisfies @xmath350 \,{\stackrel { \textnormal{def}}{=\!\!= } } \ , \iint_\mathbb d |d g|^2 \;=\ , \frac{1 } { 2\,\pi } \underset { \mathbb t \,\times \,\mathbb t } { \;\,\;\int \,\int}\ ; \big{|}\frac{g(\xi ) - g(\zeta)}{\xi - \zeta } \big{|}^2 \ ; |\textnormal d\xi| \!\cdot\!|\textnormal{d}\zeta | \;.\ ] ] recall that by the mappings @xmath344{{}_{\!\!\textnormal{onto\,\,}\!\!}}}\partial\,\mathbb s\,$ ] satisfy @xmath351 where the implied constant does not depend on @xmath352 . by virtue of this implies @xmath353 \;\preccurlyeq\ ; \mathscr e_{_\mathbb d } [ f]+ 1 < \infty\;.\ ] ] therefore @xmath354 have uniformly bounded energy .
it follows that @xmath247 converge to @xmath228 not only uniformly but also weakly in @xmath355 .
in particular , @xmath356 \leqslant \liminf \mathscr e[f_m]\,$ ] .
it is crucial to notice , using dominated convergence theorem , that in fact we have equality @xmath357 \,&= \frac{1 } { 2\,\pi } \underset { \mathbb t \,\times \,\mathbb t } { \;\,\;\int \,\int}\ ; \big{|}\frac{f(\xi ) - f(\zeta)}{\xi - \zeta } \big{|}^2 \ ; |\textnormal d\xi| \!\cdot\!|\textnormal{d}\zeta | \\&= \lim \frac{1 } { 2\,\pi } \underset { \mathbb t \,\times \,\mathbb t } { \;\,\;\int \,\int}\ ; \big{|}\frac{f_m(\xi ) - f_m(\zeta)}{\xi - \zeta } \big{|}^2 \ ; |\textnormal d\xi| \!\cdot\!|\textnormal{d}\zeta | = \lim \mathscr e_{_\mathbb d}[f_m ] . \end{split}\ ] ] j. m. ball , _ singularities and computation of minimizers for variational problems _ , foundations of computational mathematics ( oxford , 1999 ) , 120 , london math . soc . lecture note ser . , 284 , cambridge univ .
press , cambridge , 2001 . j. m. ball , _ progress and puzzles in nonlinear elasticity _ , in `` poly- , quasi- and rank - one convexity in applied mechanics '' ( eds .
j. schrder and p. neff ) , proceedings of cism international centre for mechanical sciences , vol . 516 ( 2010 ) , 115 . l. f. mcauley , _ some fundamental theorems and problems related to monotone mappings _
1971 proc .
first conf . on monotone mappings and open mappings ( suny at binghamton , binghamton , n.y . , 1970 ) .
state univ . of new york at binghamton , n.y . | let @xmath0 and @xmath1 be jordan domains of the same finite connectivity , @xmath2 being inner chordarc regular ( such are lipschitz domains ) .
every homeomorphism @xmath3 in the sobolev space @xmath4 extends to a continuous map @xmath5 .
we prove that there exist homeomorphisms @xmath6 which converge to @xmath7 uniformly and in @xmath8 .
the problem of approximation of sobolev homeomorphisms , raised by j. m. ball and l. c. evans , is deeply rooted in a study of energy - minimal deformations in nonlinear elasticity .
the new feature of our main result is that approximation takes place also on the boundary , where the original map need not be a homeomorphism . |
the initial observation @xcite of a baryonic resonance @xmath0 with strangeness @xmath5 generated an enormous amount of experimental and theoretical activity .
as the minimal quark content of the @xmath0 is @xmath6 , this was the first evidence for exotic hadronic states with more than three quarks , which have long been conjectured to exist @xcite .
there followed a number of experimental results @xcite consistent with a pentaquark resonance close to @xmath7 mev , roughly @xmath8 mev above the kaon - nucleon threshold , with a surprisingly narrow width , possibly even as small as a few mev .
however , there have also been several null experimental results where the @xmath0 resonance has not been found @xcite .
the uncertain experimental status of the pentaquark is an ideal situation for theoretical calculations to have an impact , particularly as the parity of the @xmath0 has not yet been measured ( it is expected to have isospin 0 ) .
in fact , previous to the experimental discovery , it was actually predicted @xcite in the chiral soliton model that a narrow resonance should exist at approximately @xmath9 mev , which was completely consistent with the first observations .
there have since been a number of attempts @xcite to model the behavior of the pentaquark , in particular its very narrow width , for example using quark models , kaon - nucleon bound states and qcd sum rules .
the only known method to derive hadronic properties _ ab initio _ from a theory of strongly - interacting quarks and gluons is lattice qcd .
not surprisingly , there have been a number of lattice studies to see if the @xmath0 can be predicted from qcd .
the first attempts used standard hadron spectroscopy methods to explore pentaquark states in a number of isospin and parity channels , to determine for which quantum numbers could a pentaquark state be consistent with experiment @xcite . it was consistently found that the lightest isospin 0 state is below the lightest isospin 1 state , but there was disagreement on the parity assignment of a possible @xmath0 resonance . there were also studies which did not find a pentaquark resonance , only scattering states of weakly - interacting kaons and nucleons @xcite .
more recent studies which try to disentangle a possible pentaquark from kaon - nucleon states indicate that a pentaquark resonance is not seen @xcite .
however consensus has not yet been reached , as other work @xcite continues to favor a single - particle resonance consistent with the experimental @xmath0 . the search for spin-@xmath10 pentaquarks on the lattice
have also been conducted more recently by a couple of groups ( @xcite ) ; however , also without agreement on their conclusions .
the purpose of this short paper is to try to add to the level of agreement among lattice qcd calculations .
there are countless ways qcd can be regulated using the lattice as a space - time cutoff , but all possible formulations should agree in their physical predictions when the lattice spacing vanishes and the continuum limit is reached .
we use a particular improved lattice qcd action , designed so that physical continuum results can be extracted even on relatively coarse lattices .
because our action is also chirally symmetric at vanishing quark mass ( up to small parametrization errors ) , this allows us to reach light quark masses , which is not possible in many lattice formulations .
we use the correlation matrix method to separate a possible pentaquark resonance from an interacting multiparticle system . as in the other studies
, we use the quenched approximation where dynamical quark effects are omitted , which is an enormous saving in computational difficulty .
although without justification , this approximation empirically gives qualitatively and even quantitatively the correct picture for the hadronic states whose width is narrow .
the expectation is that this will also hold for any possible pentaquark states , if it is indeed as narrow as some experimental results claim .
the paper is organised as follows : in section 2 we give a short description of the lattice action and the details of the numerical simulation . in section 3
we discuss the correlation matrix analysis to separate ground and excited states in various channels . in section 4
we present and discuss our numerical results , followed by our conclusions .
the lattice qcd action we use is based on the real space renormalization group @xcite .
the fixed point action is a close approximation to the exact renormalization group trajectory , so the theory has only a weak dependence on the lattice spacing and continuum results can be extracted even on relatively coarse lattices .
if the continuum limit is reached earlier , this could bring a large reduction in the computational time required to work at a fixed physical volume .
fixed point actions have been constructed for 2- and 4-dimensional systems and in general have shown the expected good behavior @xcite .
in addition , the exact fixed point dirac operator is a solution to the ginsparg - wilson relation @xcite , giving exact chiral symmetry at non - zero lattice spacing @xcite .
the fixed point action is one of the many possible lattice fermion formulations @xcite that have all the desirable chiral properties of the continuum theory , such as massless pions at vanishing quark mass .
the action we use in the simulations is close to an exact ginsparg - wilson solution , hence the chiral properties are almost completely intact .
the fixed point qcd action has been tested in a number of quenched qcd studies @xcite , including hadron spectroscopy , @xmath11-@xmath11 scattering lengths , the chiral condensate , glueball masses , the topological susceptibility , unphysical zero - mode effects in quenched qcd , and properties of the dirac operator eigenvalues .
large - scale hadron spectroscopy studies have shown the improved behavior of the action @xcite .
hadron masses calculated at lattice spacings of @xmath12 and @xmath13 fm have shown only a weak dependence on the lattice cutoff and are closer to continuum - extrapolated values than using the wilson action at @xmath14 fm or the improved staggered action at @xmath15 fm .
pion masses as light as @xmath16 mev could be reached without any indication of exceptional configurations ( i.e. gauge field configurations for which the dirac operator has unphysical accidentally small eigenvalues due to large fluctuations ) , which limit the lightest possible quark mass that can be reached .
this is an indication of the good chiral behavior of the lattice action .
in this paper we study an ensemble of 176 gauge configurations of lattice size @xmath17 which have already been generated for standard hadron spectroscopy @xcite . from the sommer parameter
@xmath18 fm , the lattice spacing is found to be @xmath19 fm , where the systematic error is at the % level and the statistical error is less than @xmath20 .
the gauge configurations are separated by 500 alternating metropolis and pseudo - over - relaxation sweeps .
the gauge configurations are fixed to the coulomb gauge and gaussian - smeared sources of width @xmath21 were used for the quark propagators .
the ensemble covers a range of quarks masses , reaching pseudoscalar meson masses as light as @xmath22 mev .
the usual goal in standard hadron spectroscopy studies is to extract particle masses for the various possible quantum numbers .
the typical method used is to calculate the correlation functions in euclidean time of two- and three - quark operators with the desired quantum numbers .
for example , the pion correlation function is @xmath23 . for large euclidean time
, the correlation function is dominated by the lowest energy state @xmath24 if the correlation function can be measured for sufficiently large time separations , the higher energy states are suppressed and this is a safe method to extract the ground state energy .
this can be seen in the effective mass @xmath25,\ ] ] which develops a plateau when the higher states give only a negligible contribution to the correlation function .
however , if the energy separations are small , the higher energy states die out slowly and contaminate the lowest state .
in addition , if the overlap of the operator with the ground state is accidentally small , it might be necessary to measure the correlation function for very large time separations to suppress the higher states . in these situations , the time separation required for the effective mass to develop a plateau might be unfeasibly large . in principle , one can extract the energy of the excited states by including their contribution to the correlation function .
however , in practice , this requires measuring @xmath26 to greater accuracy than is normally possible .
a more reliable method to extract the energy states is to form a matrix of correlation functions @xmath27 of operators with the same desired quantum numbers @xcite . for an @xmath28 correlation matrix ,
the solution of the generalized eigenvalue problem @xmath29 between timeslices @xmath30 and @xmath31 is used to construct the correlators @xmath32 whose overlap with the @xmath33 lowest energy states is optimal . the contamination of the ground state by higher states is minimized and the excited states themselves can also be extracted if the correlation matrix is measured to sufficient accuracy . the correlation matrix analysis is more reliable for extracting the ground state and it is necessary for getting higher energy states for multiparticle systems . if the @xmath0 state does exist , its quantum numbers will be the same as a tower of kaon - nucleon interacting states .
if the pentaquark is not the lightest state in that channel , at the very least the two lightest states must be extracted to positively identify the @xmath0 . to look for a possible pentaquark state @xmath0 , some five - quark operator with the desired quantum numbers must be chosen . to construct a correlation matrix ,
more than one such operator is required . without much guidance , it is difficult to tell which will have the best overlap with the @xmath0 wave function .
a number of operators have been tried , including ones based on a kaon - nucleon - type construction or using the diquark - diquark - antiquark structure proposed by wilczek and jaffe .
many possible five - quark operators can be related to one another by fierz transformations and all of them must have some overlap with kaon - nucleon states . even with the experience of the previous lattice qcd studies , it is not clear what the best choice is . to allow us to make a direct comparison with other work
, we use the operator originally suggested in @xcite and first used in a lattice simulation in @xcite @xmath34\ { u_e \bar{s}_e i \gamma_5 d_c \mp ( u \leftrightarrow d ) \},\ ] ] as well as the color - rearranged combination @xmath35\ { u_c \bar{s}_e i \gamma_5 d_e \mp ( u \leftrightarrow d ) \},\ ] ] where @xmath36 is the charge - conjugation matrix and the minus and plus signs correspond to the isospin 0 and 1 combinations respectively .
these operators have intrinsic negative parity , however , it is well - known that the correlation functions constructed from these operators receive contributions from negative and positive parity states . to project a particular parity state , an additional factor @xmath37
is included in the correlation function .
the quark propagators were generated from gaussian - smeared sources . using a smeared instead of point source
has previously been shown to accelerate the decay of excited states in correlation functions and to generate effective masses whose plateaux extend to smaller time separation , allowing a more accurate determination of the effective mass .
smearing both the source and the sink in the correlation function has not in general shown much additional improvement . in our study
, we use smeared sources and point sinks .
this asymmetry means the correlation matrix is not hermitian and so left and right eigenvectors must be determined in the generalized eigenvalue problem to construct the optimal correlators : @xmath38
previous lattice studies have examined a variety of isospin and parity channels to find a state compatible with the observed @xmath0 . with the exception of one study ,
a possible signal of a genuine pentaquark state has been found only in the isospin 0 negative parity channel . for the purpose of this short paper
, we will concentrate on these quantum numbers . for this ensemble of gauge configurations ,
quark propagators were calculated for 10 quark masses , with pseudoscalar meson masses ranging from @xmath22 mev to 1390 mev .
although one is ultimately interested in reaching the physical quark masses , it is also instructive to study the mass dependence of hadronic states . to calculate a correlation function of five - quark operators ,
the number of quark propagator contractions is much greater than for two- or three - quark operators .
hence we consider only some of the possible combinations of up - down and strange masses , @xmath39 and @xmath40 , for this ensemble . in quenched qcd ,
where dynamical quark effects are omitted , gauge configurations with topological zero modes of the dirac operator are not suppressed in the ensemble when the quark masses become small .
the zero modes can give large unphysical @xmath41 contributions to correlation functions , which are only suppressed as the volume becomes large . this effect and possible solutions
have been studied extensively for two- and three - quark correlation functions , but not yet for the pentaquark case .
since our physical volume is relatively small , roughly 1.8 fm , this might prevent us from extracting useful information from the lightest quark masses .
we calculate the @xmath42 correlation matrix of the operators @xmath43 and @xmath44 and construct the optimal correlators @xmath45 . near the region where a plateau in the effective mass is observed ,
we fit the diagonal entries of the optimal correlators to a two - exponential ansatz @xmath46 giving us the ground and first excited energies @xmath47 and @xmath48 . the best fit values were obtained using the standard correlated @xmath49 method using the measured covariance to estimate the covariance matrix .
the errors were obtained by a 1024-point bootstrap sampling procedure .
the different quark mass combinations were fitted independently and so correlations between the different quark mass values were not taken into account .
we should note that there are also backward - propagating contributions to the correlators , but these are suppressed by a factor of @xmath50 in the region where the two - exponential fits are performed . in figures [ fig : effmass0 ] and [ fig : effmass1 ] we show typical effective mass plots and the corresponding two - exponential fits for the pentaquark ground and first excited state .
a single - exponential fit can also be used , but only over a shorter time range , which limits the precision of the fitted parameters .
the single- and two - exponential fitting procedures give completely consistent results , as shown in figure [ fig : agreement ] .
to demonstrate the stability of the two - exponential fits , we show in figure [ fig : fits ] the fitted energies for a particular quark mass combination as a function of the minimum timeslice included in the fits .
the dotted lines are the @xmath49 per degree of freedom from the fits and the horizontal solid and dashed lines indicate the kaon - nucleon threshold and lowest non - zero momentum scattering states .
the fitting procedure clearly works well . .
the fitted energy values for the ground and first excited states .
the two states were fitted separately to two exponentials after the diagonalization of the @xmath42 correlation matrix .
[ table:1 ] [ cols=">,>,>,>,>,>,>,>",options="header " , ] the meson and nucleon masses for degenerate @xmath40 and @xmath39 quarks have previously been calculated for this ensemble and are tabulated in @xcite . in tables
[ table:1 ] and [ table:2 ] we give the results for the five - quark ground and first excited state energies , and the pseudoscalar meson mass for non - degenerate quarks .
most of the data are for equal up - down and strange quark masses .
the physical value of the strange quark mass is very close to @xmath51 , so we also consider non - degenerate quark masses with the strange quark mass fixed at this value .
previous lattice qcd and experimental studies have shown that the kaon - nucleon interaction in the isospin 0 s - wave channel is weak @xcite .
therefore the energy of a kaon - nucleon scattering state with equal back - to - back momentum is closely approximated by @xmath52 where the momentum on a finite lattice is quantized @xmath53 . in figure
[ fig : summary ] we plot the five - quark ground and first excited state energies and the three lowest kaon - nucleon scattering state energies as a function of the sum of the nucleon and pseudoscalar meson masses . over a large range of quark masses
, we see that the five - quark energies are in very good agreement with the weakly - interacting scattering states . as the quark masses are reduced , the five - quark excited state is more difficult to determine , but the ground state is still accurately measured . with @xmath40 close to the physical value and @xmath54
, the five - quark ground state agrees well with the kaon - nucleon ground state but we can not make a strong statement about the excited state .
the most important observation is that , for heavier quarks where our results are most accurate , we do not see any indication of a pentaquark resonance state , which should lie between the scattering state energy levels .
the picture for light quarks in consistent with this , where the first excited state lies between the first and second excited kaon - nucleon scattering states , with much larger statistical errors . in principle
, one would like to extrapolate the results to the physical quark masses .
however , as we mentioned previously , quenched qcd correlation functions suffer from unphysical large contributions from topological zero modes at small quark mass .
this complication might obstruct taking the chiral limit on this lattice volume .
in addition , one would like to repeat this calculation for a range of lattice spacings to extrapolate to the continuum limit .
however other hadronic states have shown only a very weak dependence on the lattice cutoff , so we expect this will not change our results significantly .
the effective mass of the optimized pentaquark ground state correlator .
the optimal effective mass as defined in ref .
@xcite is also shown with diamond symbols .
, scaledwidth=110.0%,scaledwidth=95.0% ] the effective mass of the optimized pentaquark first excited state correlator .
the optimal effective mass as defined in ref .
@xcite is also shown with diamond symbols.,scaledwidth=110.0%,scaledwidth=95.0% ] single- and two - exponential fits to the ground state diagonal component of the optimized correlation matrix .
for the single - exponential fit , the fit could only be done starting at large time separations.,scaledwidth=110.0%,scaledwidth=95.0% ] the fitted energies for the pentaquark ground and first excited states and the quality of the fits as a function of @xmath55 , the smallest time separation included in the fit range .
the horizontal solid lines are the kaon - nucleon threshold ( including the error ) and the horizontal dashed lines are the energy levels of kaon - nucleon scattering states.,scaledwidth=105.0%,scaledwidth=90.0% ] the fitted pentaquark ground and first excited state energies as a function of the sum of the nucleon and pseudoscalar meson mass .
the degenerate quark mass data are the dark blue and red circles , the non - degenerate quark mass data the light blue and orange squares .
the hollow symbols are the three lowest kaon - nucleon scattering state energies .
the dashed lines indicate the physical values of @xmath56.,scaledwidth=105.0%,scaledwidth=85.0% ] let us compare our results with other studies .
the first attempts @xcite at pentaquark spectroscopy found an energy state slightly above the kaon - nucleon threshold in the isospin 0 negative parity channel , which looked consistent with the experimental @xmath0 . further studies using bayesian techniques to extract ground and excited states found energies consistent with k - n scattering only @xcite .
scattering states can also be distinguished from resonances via the volume - dependence of the energies , which can also be raised with hybrid boundary conditions @xcite .
the amplitudes @xmath57 of the energy states contributing to the correlation functions also have a distinctive volume - dependence for two - particle scattering .
the correlation matrix method has also been used previously to extract the pentaquark ground and excited state energies @xcite .
there is growing evidence that the isospin 0 negative parity channel , the one thought most likely to describe the experimentally observed @xmath0 , only has only weakly - interacting scattering states and no resonance .
a stronger claim is made in @xcite , that a genuine pentaquark resonance should become a bound state at heavy quark mass , which is clearly not seen in any study .
however , the picture is not completely consistent . in @xcite
the pentaquark first excited energy state determined from the correlation matrix does not have the expected volume - dependence of two - particle scattering in small volumes . in @xcite the amplitudes @xmath57 for the excited state appear to be independent of volume , signalling a resonance rather than a two - particle state , in contradiction to the findings in @xcite .
there is also the competing claim @xcite that the @xmath0 is actually in the isospin 0 positive parity channel .
our conclusions are relatively straightforward .
the correlation matrix technique is designed to extract a number of energy levels in a given channel .
previous evidence suggests that the physical @xmath0 , if it exists , is in the isospin 0 negative parity channel , above the kaon - nucleon threshold . in that case , the minimum requirement from a lattice study is to identify the two lowest energy states .
doing this , we can accurately determine the five - quark energies for a range of quark masses .
we find that the ground and first excited states are completely consistent with weakly - interacting kaon - nucleon scattering states , with no indication of a pentaquark resonance which should lie between them .
although we work at a finite lattice spacing , previous work has shown that this lattice action reproduces extremely well continuum properties of hadronic states , so we do not expect our observation to change significantly in the continuum limit . as the title of this paper suggests , absence of evidence is not evidence of absence and there are many caveats one can include .
this and all other lattice qcd pentaquark studies have ignored the effect of dynamical fermions , which is an unjustified approximation .
the most accurate results ruling out a pentaquark resonance are at unphysically heavy quark masses , requiring an extrapolation to the physical values .
we work at a finite lattice spacing and do not extrapolate to the continuum .
however we do not expect these effects to qualitatively change our conclusions .
we believe a more serious issue is our choice of operators in the correlation matrix , which may have little overlap with a genuine @xmath0 state , if it exists .
we can not rule out this possibility , and there may even be other reasons why we do not see a pentaquark resonance
. however we can only conclude that we find no evidence that qcd predicts the @xmath0 state .
we wish to thank peter hasenfratz , julius kuti , and aneesh manohar for invaluable discussions and support and shoichi sasaki for comments and discussions on our manuscript .
we wish to thank the swiss center for scientific computing in manno , where the numerical simulations were performed .
this work was supported by the iitac prtli initiative , by the department of energy under the grant doe - fg03 - 97er40546 and by the national science foundation under grant no.phy99-07949 .
kh wishes to thank the kitp at the university of california santa barbara and the itp at the university of bern , where this paper was partly written .
t. nakano _ et al .
_ [ leps collaboration ] , phys . rev
. lett .
* 91 * , 012002 ( 2003 ) [ arxiv : hep - ex/0301020 ] .
r. l. jaffe , slac - pub-1774 ( 1976 ) ; d. strottman , phys .
d * 20 * , 748 ( 1979 ) ; a. v. manohar , nucl .
b * 248 * , 19 ( 1984 ) ; m. chemtob , nucl .
b * 256 * ( 1985 ) 600 .
s. stepanyan _ et al . _ [ clas collaboration ] , phys . rev
. lett . *
91 * , 252001 ( 2003 ) [ arxiv : hep - ex/0307018 ] ; j. barth _ et al .
_ [ saphir collaboration ] , [ arxiv : hep - ex/0307083 ] ; v. v. barmin _ et al . _ [ diana collaboration ] , phys . atom .
nucl . * 66 * , 1715 ( 2003 ) [ yad .
fiz . * 66 * , 1763 ( 2003 ) ] [ arxiv : hep - ex/0304040 ] ; a. e. asratyan , a. g. dolgolenko and m. a. kubantsev , phys . atom .
nucl . *
67 * , 682 ( 2004 ) [ yad .
fiz . * 67 * , 704 ( 2004 ) ] [ arxiv : hep - ex/0309042 ] ; v. kubarovsky _ et al . _ [ clas collaboration ] , phys .
* 92 * , 032001 ( 2004 ) [ erratum - ibid .
* 92 * , 049902 ( 2004 ) ] [ arxiv : hep - ex/0311046 ] ; a. airapetian _ et al . _
[ hermes collaboration ] , phys .
b * 585 * , 213 ( 2004 ) [ arxiv : hep - ex/0312044 ] ; s. chekanov _ et al .
_ [ zeus collaboration ] , phys .
b * 591 * , 7 ( 2004 ) [ arxiv : hep - ex/0403051 ] ; a. aleev _ et al .
_ [ svd collaboration ] , [ arxiv : hep - ex/0401024 ] ; m. abdel - bary _ et al . _ [ cosy - tof collaboration ] , phys .
b * 595 * , 127 ( 2004 ) [ arxiv : hep - ex/0403011 ] ; p. z. aslanyan , v. n. emelyanenko and g. g. rikhkvitzkaya , [ arxiv : hep - ex/0403044 ] ; y. a. troyan , a. v. beljaev , a. y. troyan , e. b. plekhanov , a. p. jerusalimov , g. b. piskaleva and s. g. arakelian , [ arxiv : hep - ex/0404003 ] . j. z. bai _ et al . _
[ bes collaboration ] , phys .
d * 70 * , 012004 ( 2004 ) [ arxiv : hep - ex/0402012 ] ; k. t. knopfle , m. zavertyaev and t. zivko [ hera - b collaboration ] , j. phys .
g * 30 * , s1363 ( 2004 ) [ arxiv : hep - ex/0403020 ] ; c. pinkenburg [ phenix collaboration ] , j. phys .
g * 30 * , s1201 ( 2004 ) [ arxiv : nucl - ex/0404001 ] ; m. j. longo _ et al .
_ [ hypercp collaboration ] , phys .
d * 70 * , 111101 ( 2004 ) [ arxiv : hep - ex/0410027 ] ; d. o. litvintsev [ cdf collaboration ] , nucl .
suppl . * 142 * , 374 ( 2005 ) [ arxiv : hep - ex/0410024 ] ; t. berger - hrynova [ babar collaboration ] , [ arxiv : hep - ex/0411017 ] ; k. abe _ et al . _ [ belle collaboration ] , [ arxiv : hep - ex/0411005 ] ; s. r. armstrong , nucl .
phys . proc .
* 142 * , 364 ( 2005 ) [ arxiv : hep - ex/0410080 ] ; p. rossi [ clas collaboration ] , [ arxiv : hep - ex/0409057 ] .
d. diakonov , v. petrov and m. v. polyakov , z. phys .
a * 359 * , 305 ( 1997 ) [ arxiv : hep - ph/9703373 ] .
r. l. jaffe and f. wilczek , phys .
* 91 * , 232003 ( 2003 ) [ arxiv : hep - ph/0307341 ] , phys .
d * 69 * , 114017 ( 2004 ) [ arxiv : hep - ph/0312369 ] ; k. cheung , phys . rev .
d * 69 * , 094029 ( 2004 ) [ arxiv : hep - ph/0308176 ] ; r. d. matheus , f. s. navarra , m. nielsen , r. rodrigues da silva and s. h. lee , phys .
b * 578 * , 323 ( 2004 ) [ arxiv : hep - ph/0309001 ] ; f. stancu and d. o. riska , phys .
b * 575 * , 242 ( 2003 ) [ arxiv : hep - ph/0307010 ] ; m. karliner and h. j. lipkin , phys .
b * 575 * , 249 ( 2003 ) [ arxiv : hep - ph/0402260 ] ; l. y. glozman , phys .
b * 575 * , 18 ( 2003 ) [ arxiv : hep - ph/0308232 ] ; s. capstick , p. r. page and w. roberts , phys .
b * 570 * , 185 ( 2003 ) [ arxiv : hep - ph/0307019 ] ; a. hosaka , phys .
b * 571 * , 55 ( 2003 ) [ arxiv : hep - ph/0307232 ] ; b. k. jennings and k. maltman , phys .
d * 69 * , 094020 ( 2004 ) [ arxiv : hep - ph/0308286 ] ; c. e. carlson , c. d. carone , h. j. kwee and v. nazaryan , phys .
b * 573 * , 101 ( 2003 ) [ arxiv : hep - ph/0307396 ] , phys .
b * 579 * , 52 ( 2004 ) [ arxiv : hep - ph/0310038 ] , phys .
d * 70 * , 037501 ( 2004 ) [ arxiv : hep - ph/0312325 ] ; f. huang , z. y. zhang , y. w. yu and b. s. zou , phys .
b * 586 * , 69 ( 2004 ) [ arxiv : hep - ph/0310040 ] ; r. bijker , m. m. giannini and e. santopinto , eur .
j. a * 22 * , 319 ( 2004 ) [ arxiv : hep - ph/0310281 ] ; n. itzhaki , i. r. klebanov , p. ouyang and l. rastelli , nucl .
b * 684 * , 264 ( 2004 ) [ arxiv : hep - ph/0309305 ] ; d. e. kahana and s. h. kahana , phys .
d * 69 * , 117502 ( 2004 ) [ arxiv : hep - ph/0310026 ] ; f. j. llanes - estrada , e. oset and v. mateu , phys .
c * 69 * , 055203 ( 2004 ) [ arxiv : nucl - th/0311020 ] ; j. sugiyama , t. doi and m. oka , phys .
b * 581 * , 167 ( 2004 ) [ arxiv : hep - ph/0309271 ] .
s. l. zhu , phys .
* 91 * , 232002 ( 2003 ) [ arxiv : hep - ph/0307345 ] .
f. csikor , z. fodor , s. d. katz and t. g. kovacs , jhep * 0311 * , 070 ( 2003 ) [ arxiv : hep - lat/0309090 ] .
s. sasaki , phys .
lett . *
93 * , 152001 ( 2004 ) [ arxiv : hep - lat/0310014 ] .
t. w. chiu and t. h. hsieh , phys .
d * 72 * , 034505 ( 2005 ) [ arxiv : hep - ph/0403020 ] .
n. mathur _ et al .
_ , phys .
d * 70 * , 074508 ( 2004 ) [ arxiv : hep - ph/0406196 ] .
n. ishii , t. doi , h. iida , m. oka , f. okiharu and h. suganuma , phys .
d * 71 * , 034001 ( 2005 ) [ arxiv : hep - lat/0408030 ] .
b. g. lasscock _ et al .
_ , phys .
d * 72 * , 014502 ( 2005 ) [ arxiv : hep - lat/0503008 ] .
f. csikor , z. fodor , s. d. katz , t. g. kovacs and b. c. toth , phys .
d * 73 * , 034506 ( 2006 ) [ arxiv : hep - lat/0503012 ] . c. alexandrou and a. tsapalis , phys .
d * 73 * , 014507 ( 2006 ) [ arxiv : hep - lat/0503013 ] .
t. t. takahashi , t. umeda , t. onogi and t. kunihiro , phys .
d * 71 * , 114509 ( 2005 ) [ arxiv : hep - lat/0503019 ]
. n. ishii , t. doi , y. nemoto , m. oka and h. suganuma , phys .
d * 72 * , 074503 ( 2005 ) [ arxiv : hep - lat/0506022 ] . b. g. lasscock , d. b. leinweber , w. melnitchouk , a. w. thomas , a. g. williams , r. d. young and j. m. zanotti , phys .
d * 72 * , 074507 ( 2005 ) [ arxiv : hep - lat/0504015 ] .
p. hasenfratz and f. niedermayer , nucl .
b * 414 * ( 1994 ) 785 [ arxiv : hep - lat/9308004 ] ; p. hasenfratz , [ arxiv : hep - lat/9803027 ] .
u. j. wiese , phys .
b * 315 * , 417 ( 1993 ) [ arxiv : hep - lat/9306003 ] ; w. bietenholz and u. j. wiese , phys .
b * 378 * , 222 ( 1996 ) [ arxiv : hep - lat/9503022 ] ; t. degrand , a. hasenfratz , p. hasenfratz and f. niedermayer , nucl .
b * 454 * , 587 ( 1995 ) [ arxiv : hep - lat/9506030 ] ; t. degrand , a. hasenfratz , p. hasenfratz and f. niedermayer , nucl .
b * 454 * , 615 ( 1995 ) [ arxiv : hep - lat/9506031 ] ; t. degrand , a. hasenfratz , p. hasenfratz and f. niedermayer , phys . lett .
b * 365 * , 233 ( 1996 ) [ arxiv : hep - lat/9508024 ] ; m. blatter , r. burkhalter , p. hasenfratz and f. niedermayer , phys .
d * 53 * , 923 ( 1996 ) [ arxiv : hep - lat/9508028 ] ; w. bietenholz and u. j. wiese , nucl .
b * 464 * , 319 ( 1996 ) [ arxiv : hep - lat/9510026 ] ; t. degrand , a. hasenfratz and d. c. zhu , nucl .
b * 475 * , 321 ( 1996 ) [ arxiv : hep - lat/9603015 ] ; t. degrand , a. hasenfratz and d. c. zhu , nucl .
b * 478 * , 349 ( 1996 ) [ arxiv : hep - lat/9604018 ] ; t. degrand , a. hasenfratz , p. hasenfratz , p. kunszt and f. niedermayer , nucl .
* 53 * , 942 ( 1997 ) [ arxiv : hep - lat/9608056 ] ; w. bietenholz , r. brower , s. chandrasekharan and u. j. wiese , nucl .
proc . suppl .
* 53 * , 921 ( 1997 ) [ arxiv : hep - lat/9608068 ] ; t. degrand , a. hasenfratz and t. g. kovacs , nucl .
b * 505 * , 417 ( 1997 ) [ arxiv : hep - lat/9705009 ] ; c. b. lang and t. k. pany , nucl .
b * 513 * , 645 ( 1998 ) [ arxiv : hep - lat/9707024 ] ; f. farchioni and v. laliena , nucl .
b * 521 * , 337 ( 1998 ) [ arxiv : hep - lat/9709040 ] ; k. orginos , w. bietenholz , r. brower , s. chandrasekharan and u. j. wiese , nucl .
suppl . * 63 * , 904 ( 1998 ) [ arxiv : hep - lat/9709100 ] ; f. farchioni and v. laliena , phys .
d * 58 * , 054501 ( 1998 ) [ arxiv : hep - lat/9802009 ] ; f. farchioni , i. hip , c. b. lang and m. wohlgenannt , nucl .
suppl . * 73 * , 939 ( 1999 ) [ arxiv : hep - lat/9809049 ] ; w. bietenholz and h. dilger , nucl .
b * 549 * , 335 ( 1999 ) [ arxiv : hep - lat/9812016 ] ; t. bhattacharya , r. gupta and w. j. lee , nucl .
* 83 * , 860 ( 2000 ) [ arxiv : hep - lat/9910046 ] .
p. h. ginsparg and k. g. wilson , phys .
d * 25 * , 2649 ( 1982 ) .
p. hasenfratz , nucl .
suppl . * 63 * , 53 ( 1998 ) [ arxiv : hep - lat/9709110 ] ; p. hasenfratz , v. laliena and f. niedermayer , phys .
b * 427 * , 125 ( 1998 ) [ arxiv : hep - lat/9801021 ] ; p. hasenfratz , nucl .
b * 525 * , 401 ( 1998 ) [ arxiv : hep - lat/9802007 ] .
r. narayanan and h. neuberger , nucl .
b * 443 * , 305 ( 1995 ) , [ arxiv : hep - th/9411108 ] , nucl .
b * 412 * , 574 ( 1994 ) , [ arxiv : hep - lat/9307006 ] , phys .
lett . * 71 * , 3251 ( 1993 ) , [ arxiv : hep - lat/9308011 ] , phys .
b * 302 * , 62 ( 1993 ) , [ arxiv : hep - lat/9212019 ] ; d. b. kaplan , phys .
b * 288 * , 342 ( 1992 ) , [ arxiv : hep - lat/9206013 ] ; y. shamir , nucl .
b * 406 * , 90 ( 1993 ) , [ arxiv : hep - lat/9303005 ] ; v. furman and y. shamir , nucl .
b * 439 * , 54 ( 1995 ) , [ arxiv : hep - lat/9405004 ] ; t. blum _ et al .
_ , phys .
d * 69 * , 074502 ( 2004 ) , [ arxiv : hep - lat/0007038 ] ; c. gattringer , phys
. rev .
d * 63 * , 114501 ( 2001 ) , [ arxiv : hep - lat/0003005 ] ; c. gattringer , i. hip and c. b. lang , nucl .
b * 597 * , 451 ( 2001 ) , [ arxiv : hep - lat/0007042 ] ; c. gattringer , nucl .
suppl . * 119 * , 122 ( 2003 ) , [ arxiv : hep - lat/0208056 ] .
p. hasenfratz , s. hauswirth , k. holland , t. jorg , f. niedermayer and u. wenger , int .
j. mod .
c * 12 * , 691 ( 2001 ) , [ arxiv : hep - lat/0003013 ] ; f. niedermayer , p. rufenacht and u. wenger , nucl .
b * 597 * , 413 ( 2001 ) , [ arxiv : hep - lat/0007007 ] ; p. hasenfratz , s. hauswirth , k. holland , t. jorg , f. niedermayer and u. wenger , nucl .
suppl . *
94 * , 627 ( 2001 ) , [ arxiv : hep - lat/0010061 ] ; f. niedermayer , p. rufenacht and u. wenger , nucl .
phys . proc .
suppl . * 94 * , 636 ( 2001 ) , [ arxiv : hep - lat/0011041 ] ; p. rufenacht and u. wenger , nucl . phys .
b * 616 * , 163 ( 2001 ) , [ arxiv : hep - lat/0108005 ] ; p. hasenfratz , s. hauswirth , k. holland , t. jorg and f. niedermayer , nucl .
suppl . * 106 * , 799 ( 2002 ) , [ arxiv : hep - lat/0109004 ] ; p. hasenfratz , s. hauswirth , k. holland , t. jorg and f. niedermayer , nucl .
suppl . * 106 * , 751 ( 2002 ) [ arxiv : hep - lat/0109007 ] ; s. hauswirth , [ arxiv : hep - lat/0204015 ] ; t. jorg , [ arxiv : hep - lat/0206025 ] ; p. hasenfratz , s. hauswirth , t. jorg , f. niedermayer and k. holland , nucl .
b * 643 * , 280 ( 2002 ) , [ arxiv : hep - lat/0205010 ] ; c. gattringer
_ et al . _
[ bern - graz - regensburg collaboration ] , nucl .
* 119 * , 796 ( 2003 ) , [ arxiv : hep - lat/0209099 ] . c. gattringer _
[ bgr collaboration ] , nucl .
b * 677 * , 3 ( 2004 ) [ arxiv : hep - lat/0307013 ] .
p. hasenfratz , k. j. juge and f. niedermayer [ bern - graz - regensburg collaboration ] , jhep * 0412 * , 030 ( 2004 ) [ arxiv : hep - lat/0411034 ] . c. michael , nucl .
b * 259 * , 58 ( 1985 ) .
m. luscher and u. wolff , nucl .
b * 339 * , 222 ( 1990 ) .
m. fukugita , y. kuramashi , m. okawa , h. mino and a. ukawa , phys .
d * 52 * , 3003 ( 1995 ) , [ arxiv : hep - lat/9501024 ] . j. s. hyslop , r. a. arndt , l. d. roper and r. l. workman , phys .
d * 46 * , 961 ( 1992 ) . | we study the question of whether or not qcd predicts a pentaquark state @xmath0 .
we use the improved , fixed point lattice qcd action which has very little sensitivity to the lattice spacing and also allows us to reach light quark masses . the analysis was performed on a single volume of size @xmath1 with lattice spacing of @xmath2 fm .
we use the correlation matrix method to identify the ground and excited states in the isospin 0 , negative parity channel . in the quenched approximation where dynamical quark effects are omitted , we do not find any evidence for a pentaquark resonance in qcd .
# 1 januaryfebruarymarch aprilmayjunejulyaugustseptemberoctober novemberdecember absence of evidence for pentaquarks on the lattice kieran holland@xmath3 and k. jimmy juge@xmath4 + bgr ( bern - graz - regensburg ) collaboration @xmath3 department of physics , university of california san diego , + 9500 gilman drive , la jolla ca 92093 , usa @xmath4 school of mathematics , trinity college + dublin 2 , ireland |
imaging fabry - perot interferometers are now in common use at several major observatories and operate at both optical and infrared wavelengths .
traditionally , fabry - perots are employed to perform studies of extended gaseous nebulae .
examples include outflow sources ( starburst and active galaxies , herbig - haro systems ) and normal disk galaxies .
some groups have utilised the angular spectral coverage to detect diffuse sources ( hi / ly@xmath4 clouds in optical emission ) , by azimuthally summing deep fabry - perot images ( haffner , reynolds & tufte 1999 ) . more recently , scanning fabry - perots have been used to construct spectral line profiles at many pixel positions across a large format ccd . in many instances ,
these spectra are used simply to obtain kinematic information ( line widths , radial velocities ) in a strong emission line ( _ e.g. _ laval 1987 ; cecil 1988 ; veilleux , bland - hawthorn & cecil 1997 ; shopbell & bland - hawthorn 1998 ) . at lower spectral resolutions , _ tunable filters _ are a special class of fabry - perot , with which a series of narrowband images , stepped in wavelength , can be obtained ( atherton & reay 1981 ; bland - hawthorn & jones 1998a , b ; jones & bland - hawthorn 2001 ) .
alternative modes see the fabry - perot used for time - series imaging ( deutsch , margon & bland - hawthorn 1998 ; tinney & tolley 1999 ) , which we do not discuss here .
it is less common to see these instruments employed as spectro_photometers _ where the observed signal is calibrated to exoatmospheric flux units .
this may be due , in part , to a perception that the nature of the airy function makes fabry - perots unreliable photometers . in this review ,
we describe reliable methods for flux - calibrating fabry - perot data with the aid of worked examples .
the case of tunable imaging is special , in that photometric measurements centre upon the measurement of low resolution spectra and extraction of objects with candidate emission or absorption features . in this sense
it is analogous to multi - slit or multi - fibre spectroscopy .
however , because the tunable filter images _ all _ objects in the field at once , every object is a potential candidate . therefore , data reduction and methods need to be devised so as to be fully automated throughout . in deep extragalactic surveys at high galactic latitude , the number of objects can vary anywhere between a few hundred in high galactic latitude fields to over a thousand in galaxy clusters .
this paper reviews all stages of analysis , discussing different approaches where they exist .
section 2 describes features of tunable filter use and section 3 defines terminology used throughout the rest of the paper .
section 4 discusses different methods for the removal of instrumental signatures while section 5 reviews strategies for the detection and classification of the many objects in a single field . flux calibration issues
are discussed in section 6 and concluding remarks follow .
many of the procedures described here have been written into software ( jones 1999 ) .
interested prospective users are encouraged to contact the authors and visit the taurus tunable filter ( ttf ) web page .
at optical and infrared wavelengths , imaging fabry - perot devices are used in three different ways : ( i ) to obtain a spectrum at each pixel position over a wide field by tuning the etalon , ( ii ) to obtain a single spectrum of a diffuse source which fills a large fraction of the aperture ( from one or more deep frames at the same etalon spacing ) , and ( iii ) to obtain a sequence of monochromatic images within a field defined by the jacquinot spot ( tunable narrowband imaging ) .
figure [ differentmodes ] summarises the different modes and the types of data they produce . in this paper
, we will concentrate on analysis techniques adapted for final case , tunable filter imaging .
however , many of the methods presented herein are common to all three modes of operation . observations of a particular field consist of one of more sets of images called a scan ( or a _ stack _ or _ spectral cube _ ) . a _ scan _ is a series of images ( or _ slices _ ) in which each one has been taken at a slightly different plate spacing .
this is achieved by tuning the filter to a new passband between each exposure . ideally , scans should be repeated three or more times to permit median - filtering of cosmic - ray events and ghost images from slices of common wavelength .
removal of ghosts requires the telescope to be offset between scans and the fabry - perot to be tilted to ensure separation of ghosts and parent sources .
in this situation , the ghosts will shift in the opposite direction to the parent objects and by twice the amount . at a given ( constant ) order of interference , a one - to - one relationship exists between plate spacing and image wavelength . at the telescope ,
the spacing is encoded in software units known as _ z - step values _ , due to the movement of the fabry - perot plates during both tuning and scanning .
wavelength calibration frames take the form of long `` sausage - cubes '' : small images ( few pixels square ) at many wavelength settings ( typically 100 or more ) .
the small image area permits expediency while scanning the plates at sufficiently high sampling resolution .
if the fabry - perot is used in the pupil ( as is usually the case ) , a _ phase effect _ shows itself as a position - dependent change in wavelength across the field .
figure [ f : rawimg](@xmath5 ) illustrates this with data from the taurus tunable filter ( ttf ) system ( bland - hawthorn & jones 1998a , b ) at the anglo - australian telescope ( aat ) .
the centre of the interference pattern is coincident with the _ optical axis _ and the distance of an object from this point is its _
optical radius_. the phase change over the field gives rise to _ night - sky rings _ , broad diffuse circles of atmospheric oh emission lines on some images , centred on the optical axis ( fig .
[ f : rawimg]@xmath6 ) . they are circular because the phase change makes for a specific off - axis angle where the interference equation is satisfied for a specific wavelength . for cases in which this effect is present
, the flux needs to be re - normalised in radial bins , since it is not corrected through division by a flatfield .
if a tunable filter is used in the field ( _ i.e. _ after the field lens ) then the abovementioned effect does not occur .
meaburn ( 1976 ) shows examples of this arrangement for fabry - perot devices used in the near infrared .
the underlying principle of a tunable filter is the enlargement of the jacquinot spot at narrow plate spacings , and therefore , low spectral resolutions .
the _ jacquinot spot _ is defined as the field about the optical axis within which the peak wavelength variation with field angle does not exceed @xmath7 of the etalon band - pass ( jacquinot 1954 ; taylor & atherton 1980 ) .
this angular field can be used to perform close to monochromatic imaging .
the most direct route to the airy function , the instrumental response of the fabry - perot , is to use complex exponential notation .
an incoming plane wave with wavelength @xmath8 at an angle @xmath9 to the optical axis enters the etalon cavity and performs a series of internal reflections .
if the highly reflective inner surfaces have reflectivities of @xmath10 and @xmath11 , we can sum over the complex amplitudes of the outgoing plane waves such that @xmath12 in which @xmath13 is the phase difference between successive rays .
the total transmitted intensity is proportional to the squared modulus of the complex amplitude or @xmath14 where the refractive index and the plate separation of the cavity are @xmath15 and @xmath16 respectively , and @xmath17{{\cal r}_1{\cal r}_2}(1- \sqrt{{\cal r}_1{\cal r}_2})^{-1}$ ] . clearly , the airy function has a series of periodic maxima whenever @xmath18 which is the well - known equation of constructive interference in the @xmath19th order , with @xmath9 being the off - axis angle of the ray , ( as subtended at the etalon ) , and @xmath20 , since the fabry - perot is air - spaced .
the quantity @xmath21 is called the _ reflective finesse _ and depends only on the values of @xmath10 and @xmath11 .
it is normal procedure to manufacture an etalon with two identical coatings such that @xmath22 , but in app .
b we describe the extraneous etalon effect , where this condition is not met . in order to arrive at the correct calibration procedure , it is important to understand the nature of the airy function .
figure [ f : cyclic ] illustrates the three cyclic functions in table 1 and shows that the area of the airy function always exceeds the integral of the other functions for a given spectroscopic resolution . by analogy with the lorentzian profile , the coefficient of the @xmath23 factor in eqn .
( [ airy ] ) determines the width of the function .
the quantity @xmath24 is called the _ effective finesse _ , where @xmath25 is the spectral purity ( defined in eqn . ( [ purity ] ) below ) and @xmath26 is the periodic free spectral range . in fig .
[ f : finesse ] we illustrate how the airy function , when normalized to the gaussian and lorentzian functions , depends on @xmath21 . in practice
, there are factors other than coating reflectivities which contribute to the effective finesse @xmath27 aperture effects , imperfections within the plate coatings , etc .
@xmath27 some of which can serve to make the instrumental response more gaussian than lorentzian in form ( atherton 1981 ) . beyond a finesse of roughly 40 , the airy function
is highly lorentzian .
the reason for this is clear when looking at how the airy profile has been written in table [ t : cyclicfunctions ] . at high finesse ( @xmath28 ) , if we expand @xmath29 about the peak of the profile , the small angle formula reduces the airy function to the lorentzian form .
the normalized airy integral depends only weakly on finesse at large finesse values .
there are several approaches to deriving the jacquinot advantage ( roesler 1974 ; thorne 1988 ) , _ i.e. _ the throughput advantage of the fabry - perot interferometer . by considering the solid angle subtended by the innermost ring and using the small angle formula with eqn .
( [ famous ] ) , we arrive at the important relation @xmath30 which assumes that @xmath20 and that @xmath31 and @xmath32 are the solid angle and spectral resolving power of the ring , respectively .
this equation has been used to demonstrate that fabry - perot interferometers , at a given spectroscopic resolution , have a much higher throughput than more conventional techniques ( jacquinot 1954 ; 1960 ) .
but the jacquinot relation has another important consequence .
the solid angle of a ring defined by its fwhm intensity points can also be written @xmath33 for a fixed etalon spacing , the solid angle , and hence the spectroscopic resolution , of all rings is a constant .
this allows us to write down a simple equation valid for all rings for the signal - to - noise ratio in a monochromatic unresolved line , namely , @xmath34 where @xmath35 and @xmath6 are the source and background flux ( cts pix@xmath0 s@xmath0 ) , @xmath36 and @xmath37 are the efficiency and exposure times respectively , @xmath38 is the solid angle subtended by a pixel .
the quantity @xmath39 is the wavelength dispersion ( pix@xmath0 ) and @xmath40 is the number of ccd exposures combined to form the deep spectrum .
the factor @xmath41 is discussed below .
we normally choose to place the ring center at one corner of the field for two reasons .
first , it is often necessary to tilt the etalon in order to throw ghost light out of the field .
second , the factor @xmath42 in eqn . ( [ biggie ] ) now ensures that the spectroscopic sensitivity is constant over most of the field .
there will be an almost linear drop - off in sensitivity at large off - axis angles ( far corner from the optical axis ) where the rings become seriously incomplete .
an important characteristic of a spectrometer is its _ spectral purity _ , @xmath43 namely , the smallest measurable wavelength difference at a given wavelength ( thorne 1988 ) .
this is usually defined as the intensity fwhm of the instrumental profile .
when considering the amount of light transmitted by a spectrometer , we need to consider the total area under the instrumental function .
this issue is rarely mentioned in the context of long - slit spectrometers partly because their response is highly gaussian , response of slit - aperture devices is rarely achieved in practice . ] in which case the ` effective photometric bandpass ' ( total area divided by peak height ) is very close to the fwhm of the instrumental profile ( see table [ t : cyclicfunctions ] ) .
high spectral purity does not necessarily equate with high efficiency . at values of low finesse ,
the efficiency is high but spectral purity is compromised by the amount of power in the wings , thereby reducing spectral definition ( fig .
[ finesse_eff ] ) .
conversely , at high finesse a sharper instrumental profile is marred by a loss in efficiency .
the optimal trade - off lies with values of finesse in the range @xmath44 to 40 . at such moderate finesse ,
the airy function allows through 50% more light than a gaussian profile with equal spectroscopic resolution ( fig .
[ f : finesse ] ) .
thus , the _ effective photometric bandpass _
, @xmath45 , is almost 60% larger than the bandpass defined by the profile fwhm @xmath46 .
the factor @xmath41 in eqn .
( [ biggie ] ) corrects for a calculation based on the fwhm of a ring and is defined as @xmath47 .
technically , @xmath45 should be adopted as the spectral resolution of the airy instrumental profile , otherwise we are forced to an inconsistency when comparing fabry - perot spectrometers to other devices . using the fabry - perot as a tunable filter necessitates tuning the bandpass to several discrete wavelengths and obtaining an image at each .
figure [ figarrange ] shows such an arrangement for the _ ttf field galaxy survey _ , made with ttf at the aat ( jones 1999 ) .
also shown in this figure are the filters used to block light transmitted at unwanted neighbouring orders .
the ideal spacing of bandpasses for such detection / photometry applications is @xmath48 , which is half of the effective photometric bandpass .
however , for kinematic work , this sampling should be increased to @xmath49 .
it is hard to overstate the importance of the _ whitelight cube_. this is obtained by observing a whitelight source over the same range of etalon spacings used in the actual observations .
the whitelight cube maps the response of the filter as a function of position and etalon spacing .
there are three effects that we wish to divide out from the data .
firstly , the narrowband filter response , when convolved with the instrumental response , leads to a modulation in the observed spectrum . secondly , it is well known that filters have variable responses in both collimated and converging beams ( lissberger & wilcock 1959 ) .
thirdly , we seek to remove any inhomogeneities in the filter structure as a function of position .
when the whitelight cube is compressed in the spectral dimension , it provides a very high signal - to - noise flatfield for removing pixel - to - pixel sensitivity variations .
several approaches exist for the creation and use of whitelight flatfields , depending on the exact nature of the data .
dome flats do not mimic the sky well in cases where light scatters off the dome and irradiates the detector at extreme angles .
this can fill in the whitelight response structure with wavelength .
alternatively , twilight flatfields contain fraunhofer lines which generate fringe structure not present in science data .
the underlying point is that both the _ spatial _ and _ spectral _ distribution of the flatfield source must be well - understood for its proper use in removing pixel variations .
final flatfields can be constructed from raw flatfield frames in one of two ways , dependent upon whether interference fringes are present in the ccd used .
such fringes are due to interference within the substrate of the ccd at red wavelengths and are unrelated to the interference within the fabry - perot . in the absence of fringing the flatfield frames
can be normalised and retained as individual components of the scan , and used to divide images on an individual basis in the usual way .
if fringing is present in both the flatfields and science frames , division of the latter by the former can be tried , but its success relies on the stabilty of the fringe pattern . typically , fabry - perot plate spacings are not sufficiently stable to produce fringe structures constant over the hours that pass between flatfields and science frames .
alternatively , a fringe correction frame can be created by combining the science frames at a given etalon setting , using median filtering to remove the objects in the field , ( assuming that the object position has been dithered between exposures ) .
this produces a final frame of the fringe pattern alone , which can be subtracted .
however , the same technique can not be used for removing fringes from the flatfield .
more elaborate operations can be adopted , usually at the expense of added noise to the frame cleaned of fringing . in the end one has to determine whether it is better to use a noisier frame cleaned of fringes , or a less noisy frame on which the fringes remain .
this depends on the specific science requirements of the data .
while flatfields ( used in either manner ) successfully remove pixel - to - pixel variations , the instrumental response across the beam is modulated by the phase effect . in particular ,
care must be exercised in the use of flatfields near the edges of the blocking - filter transmission profile , as the effectiveness of the flatfield then becomes a function of optical radius . in the case of photometry of point - source objects ,
we can account for any residual radial changes in sensitivity by photometrically registering objects through a fit with optical radius at a later stage ( sect .
[ s : elcandextract ] ) . for detection of faint point - sources ,
night - sky rings need to be removed so that the detection software can successfully fit the sky background , ( which is typically done through the use of a two - dimensional sky mesh . )
although the rings look circular , they are in fact slightly elliptical ( 0.5% ) due to the imperfect alignment of the ccd plane and optical distortions within the focal reducer .
glazebrook & bland - hawthorn ( 2001 ) have demonstrated that the night sky lines can be removed from data to the level of 0.03% or better using nod and shuffle techniques . here
, the telescope is nodded between the object and sky positions in synchrony with charge being shuffled on the ccd with a cycle time of , say , 60 secs .
this results in a 50% loss of observing efficiency but does in fact lead to superb sky subtraction .
nod and shuffle with ttf produces almost perfect residuals largely because there is no entrance slit , and the monochromatic band is sampled by hundreds of ccd pixels along the optical radius vector . in those cases where nod and shuffle is not available or not feasible , alternative ways of removing the sky
can be followed . if the sky background and night sky lines can be taken as having a fixed pattern across the field of view
usually a safe assumption for the relatively small field and long exposure times of the ttf the problem of sky subtraction can be most readily approached by exploiting the azimuthal symmetry of the sky component relative to the optical axis .
a flat sky will be evident in the data as a radial variation in the background , with night sky lines appearing as broad rings . to parameterize the spectrum of the sky , the basic method is to azimuthally
bin the image about the optical axis .
bright stars and regions of evident emission are masked from the image prior to the binning procedure .
the sky spectrum is then calculated from the masked image using a median combination of the pixels remaining at each radius .
the median will eliminate fainter emission that the mask may have missed and additional cosmic rays .
the position of the optical axis is crucial to determining the accurate phase error over the field .
there are several ways to find its location : 1 .
if the tunable filter can be tuned to an interference order @xmath50 or so , the monochromatic ring will be narrow enough to be fit with least - squares techniques .
specific procedures for fitting the rings are discussed below and in app .
we can generate a well - defined ring by differencing two low order monochromatic scans where one is tuned slightly in @xmath51-space ( _ e.g. _ half a bandpass ) compared to the other .
the residual produces a p - cygni profile along the optical radius passing through zero counts which can then be fit with standard procedures discussed below and in app .
a. this residual image is typically very flat and not susceptible to ccd bias , flatfield or response variations .
diametric ghosts can be very useful for identifying the optical axis since these , by definition , reflect about the optical axis .
one technique is to generate the ghost map by imaging through a mask of pinholes arranged in a regular grid . at the aat
we use a matrix mask consisting of 100 holes drilled every 1 cm . the optical axis
can then be identified as the crossing point for all vectors which join the pinhole images with their respective ghosts ( fig .
[ matrix ] ) .
4 . a more involved approach is to derive the phase surface by scanning the ttf over a strong emission line ( bland & tully 1989 ) .
each spectrum within the monochromatic cube must now be fit with lorentzian profiles .
the resulting surface is highly paraboloidal and the optical axis along its centre is well defined . a quick method for finding the approximate centre of the night - sky ring pattern is to solve the equation of a circle from three points .
however , for the purpose of subtraction , circular fitting is usually insufficient and more sophisticated strategies must be adopted ( app .
[ ringfitting ] ) .
once the sky spectrum has been determined , a two - dimensional sky `` image '' can be constructed by sweeping the sky spectrum azimuthally around the position of the optical axis , in effect reversing the binning procedure .
this sky image may then be subtracted from the data image . as is illustrated in fig .
[ patfig ] , this procedure works quite well for sparse fields , where only a small percentage of the pixels need to be masked from the binning calculation .
we note that this method works best if one gives some consideration to the placement of the optical axis at the time of observation : within the field of view but away from the object of interest .
placement should also ensure that bright ghosts in the field are avoided . in those cases where the field is densely populated
, the empirical removal of a uniform azimuthal sky component may not be feasible . at this point
, the sky rings must be individually fit and parametrized .
the core of the fitting problem is that the sky - rings are typically incomplete , passing beyond the field - of - view for part of their length .
defining the problem as minimising the sum of squares , the so - called geometric distance , is computationally very expensive and requires careful weighting schemes and application ( _ e.g. _ joseph 1994 ) .
after years of experimentation with conventional methods ( _ e.g. _ bland - hawthorn , shopbell & cecil 1992 ) , we have settled on algebraic distance methods as the most economical and reliable approach to fitting incomplete arcs , as described in app . [ ringfitting ] .
one persistent problem has been the geometric distortion of cassegrain focal reducers exacerbated by slight tilts in the alignment of the ccd plane .
an alternative method for removal is feasible in the case where the objects of interest are much smaller than the ring structure ( _ e.g. _ at very low spectral resolution when the night - sky rings are large and diffuse ) .
the rings are removed by treating each science exposure individually .
a background map is created by median - filtering copies of the original , each one offset in a regular grid pattern from the other by a few pixels . the result
is then smoothed and subtracted from the original , leaving little or no night - sky residual .
the method depends upon the rings being of lower spatial frequency than the objects , and as such , would be less successful with extended sources .
bland - hawthorn & jones ( 1998b ) demonstrate this technique with an example .
the tunable filter is wavelength - calibrated through _
scans _ consisting of many images taken at a series of gap spacings .
spacing through the scan is controlled in software through a parameter , @xmath52 , of arbitrary units and proportional to physical plate spacing .
ordinarily , two scans are required : an _ emission - line scan _ of an arc lamp such as copper - argon or neon , and a _ continuum scan _ of a white - light source such as a tungsten lamp ( the white - light calibration cube ) , as mentioned earlier .
it is essential that both types of source are diffused by suitable optics .
past experience has shown that anomalous phase effects are produced by incorrect use of calibration lamps .
they must illuminate the instrument evenly and should not be placed between the telescope focus and the ccd .
an additional consideration in the choice of a suitable arc lamp is that a minimum of two lines , ( preferably more ) , should be visible within the restricted wavelength coverage of the blocking filter ( typically 25 to 30 nm ) .
lamps in use with ttf system at the aat are shown in table [ t : lamps ] .
around 500 nm , the astrophysical lines of and found in planetary nebula are a suitable alternative , and acker ( 1992 ) list many suitable objects . the continuum scan ( fig .
[ f : lampscans]@xmath5 ) is used to check where ( in @xmath52 ) the orders of interference occur and to ensure no overlap .
continuum scans can be sampled quite coarsely ( @xmath53 ) and should cover a range of more than one order , since the location of these is not known at the outset .
overlap of the resulting blocking filter traces indicates a plate spacing that is too large , causing the sampling of more than one order of interference . once
two orders have been located , a well - sampled scan ( @xmath54 ) through an arc lamp can follow ( fig .
[ f : lampscans]@xmath6 ) .
such a scan covers more than one order so that at least one arc line appears in adjacent orders .
this is necessary for a determination of physical plate spacing .
typically the number of arc lines available over the wavelength coverage of the blocking filter is few .
the two profiles in fig .
[ f : lampscans](@xmath6 ) show the wavelength offset between on and off - axis positions .
lines observed in the emission - line scan are identified both by the wavelength and @xmath52 at which they occur . across a single order , the @xmath55 relationship is linear , @xmath56 to determine the plate - spacing in physical units ( such as microns ) we need to first determine the order of interference .
if @xmath57 is the free spectral range ( in @xmath52-units ) separating an arc line ( wavelength @xmath58 ) in adjacent orders , then the order of interference @xmath19 is given by @xmath59 with @xmath19 known , the spectral resolving power @xmath32 and passband width @xmath46 can be determined through @xmath60 the large scan range of ttf sees resolving powers anywhere in the range of 150 to 1000 .
the physical plate - spacing calibration follows as an adaptation of eqn .
( [ e : lambdavsz ] ) , namely @xmath61 where @xmath62 is the effective plate spacing and @xmath63 for air . by
_ effective _ plate - spacing , we mean the gap as measured parallel to the optical axis , irrespective of etalon tilt .
the _ true _ gap between the plates is smaller than this , being @xmath64 , where @xmath4 is the tilt of the etalon , ( typically a few degrees to deflect ghosts out of the beam ) . at very narrow gaps ( @xmath65 )
, there is a wavelength - dependent phase - change in the reflections between optical coatings on the inner plate surfaces .
this introduces a degeneracy that requires an additional term to be introduced into eqn .
( [ famous ] ) .
the effect can be measured at such narrow plate - spacings , provided the phase - change behaviour of the inner coatings has been determined through independent means ( jones & bland - hawthorn 1998 ) .
the angle @xmath66 in eqn .
( [ e : spacing ] ) is that separating the rays on - axis and those sampled for calibration .
the wavelength offset between both points must be included in the calibration , through relations derived in bland - hawthorn & jones ( 1998a , b ) .
for example , suppose that the point used on the detector is offset by 500 pixels from the optical axis .
if the focal length of the camera is 130 mm and detector is a @xmath67 pixel ccd of @xmath68 mm , then @xmath69 if the observing conditions cause measurable drifts in wavelength of the etalon bandpass one can treat the affected frames as wavelength - averaged over the exposure time and bandpass from the time dependence .
temporal variations can be measured through rapid scans ( either small pixel - area or charge - shuffle ) and fit by a smoothly - varying function .
the offsets introduced for ghost - image removal necessitate spatial registering of scans .
there is distortion present in the field extremities of many focal reducers used to accommodate fabry - perot instruments , including the 2 focal reducer on the aat . to measure this
, we use a matrix mask consisting of 100 holes drilled every 1 cm .
figure [ radial ] shows a measurement of this by c. g. tinney through the residual of the distorted radial distance ( @xmath70 , as measured at the detector ) against that undistorted ( @xmath71 ) . at the largest radii
the maximum distortion is 2% , and relative astrometry to
better than 0.1 pix ( 0.03 at f/8 ) is possible .
the radial correction function requires terms up to @xmath72 .
scan images need not be corrected for this provided the dithering is small ( @xmath73 ) .
bacon , refregier & ellis ( 2000 ) give an elegant treatment of the general solution to image distortions across a field ( their sect . 5 ) , which can also be applied to tunable filter data in which high spatial precision is required .
typically it takes several hours to complete scans and as such , the observations are susceptible to changes in seeing between individual frames . in such cases
it is necessary to create a second version of the scan , in which all images have been degraded to the mean seeing @xmath74 of the worst frame in the original scan .
this is done by covolving each original frame with a gaussian kernal of fwhm @xmath75 where @xmath76 is the mean seeing fwhm of the given frame .
both the convolved and unconvolved ( original ) versions of the scans are used for the subsequent analysis . for the best detection and photometry of a field , there should be three or more scans ( as separate measurements ) , with slices @xmath77 , @xmath78 ,
, @xmath79 , where @xmath80 is the number of slices per scan and @xmath81 is the scan number ( up to the maximum @xmath82 scans taken ) .
each image corresponds to its central wavelength @xmath83 , ( @xmath84 ) .
there should also be matching frames @xmath85 ,
, @xmath86 , that have been smoothed to match the worst seeing image . a final co - added narrowband scan can be obtained by combining each set of narrowband frames of common wavelength . with two versions of all frames ( unconvolved and convolved ) , and two methods of combination ( straight summation or median - filtering ) , three versions of the co - added scan
are derived : 1 .
straight summation of the _ convolved _ frames at each wavelength , @xmath87 2
. straight summation of the _ unconvolved _ frames at each wavelength , @xmath88 3 .
median - filtering of the _ unconvolved _ frames at each wavelength .
median - filtering cleans the cosmic - rays and ghost images from scan ( iii ) .
however , it also lowers the signal - to - noise ratio by 30% or more on the faintest objects , thereby making scans ( i ) and ( ii ) more sensitive .
scans ( ii ) and ( iii ) have the better spatial resolution but are inferior to scan ( i ) for seeing stability , and therefore , photometric integrity .
optimally , scans ( ii ) and ( i ) can be used for the object detection and photometry , while ( ii ) and ( iii ) are used to build cosmic - ray and ghost image catalogues .
object detection and photometry on a stack of narrowband frames can be performed using any of the common software packages for detection and photometry of objects , such as ( bertin & arnouts 1996 ) or focas ( valdes 1993 ) .
the prescriptions that follow were developed using .
however , focas has also been used with similar tunable filter data ( _ e.g. _ baker 2001 ) .
the algorithm initially sees images background subtracted and filtered with a convolution mask to optimise detection sensitivity .
candidate objects are found as connected pixels above the prescribed threshold and passed through a deblending filter to separate into nearby overlapping objects if this is the case .
accepts input in the form of one ( or two ) fits images and several files .
we can make use of the dual - image capability to detect objects on the straight - summed / unconvolved frames and then photometer these positions on matching straight - summed / convolved frames .
basic shape and positional properties are measured from the detection image while photometric parameters are obtained from the second . in the analysis of a full scan
, the detection threshold should be set sufficiently low that all objects are recovered throughout the stack , including some noise peaks .
this ensures the faintest detections are not lost due to variations in background through the scan .
the object - finding software can be run on each scan frame in turn , the catalogues of which can be concatenated into a full _ raw catalogue_. each entry of this is a single detection ( or _ observation _ ) in @xmath89-space , where @xmath90 is the image containing the detection at spatial coordinate ( @xmath91 ) .
as with all imaging surveys , an inherent problem is not knowing the size of each object before its detection .
it is possible to use a series of apertures of increasing size to cover the range of expected object sizes .
if the innermost aperture is set to 1.35 times the seeing , the signal - to - noise ratio is maximised for a spatially unresolved source ( gaussian profile ) in a background - limited noise regime . for the purpose of maximising signal - to - noise ratio ,
it is preferable for a non - optimal aperture to be too big rather than small .
it is necessary to group multiple wavelength observations of the same object together . a straightforward algorithm for processing the raw catalogue is as follows . the first observation is written to the _ object buffer _ , an array to store all observables for a given object from each slice .
the remaining lines of the raw catalogue are then searched for @xmath92-matches to the object in the buffer .
if all observations for the object are found , or the limiting search depth is reached , the contents of the object buffer are summarised and written to file . the next observation is written to the cleared object buffer and the process repeats . depending on
the nature of the fields observed , the raw catalogue can contain @xmath93 observations for a 10 slice scan .
each time the object buffer is complete , the contents should be checked for at least two observations before being considered as an `` object '' .
such a double - detection criterion provides a robust way of removing noise - peak and isolated cosmic - ray detections in an automated fashion .
however , it also selects against emission - line detections in a single - frame . given the difficulty inherent in distinguishing such detections from cosmic - rays and noise peaks without manual inspection , single - frame detections are more easily extracted separately .
each double - detection object can be assigned an i d number and the distance of the object from the optical axis calculated and used to correct calibrated wavelengths for phase shift ( bland - hawthorn & jones 1998b ) .
the signal - to - noise ratio can be determined for the full rage of aperture sizes and those from the optimal aperture retained .
below - threshold detections can be discarded .
if the final catalogue contains mainly galaxies , then subset catalogues for bright stars should also be created .
these are needed for photometric registering during emission - line candidate extraction ( sect .
[ s : elcandextract ] ) . at this stage in the reduction ,
most of the cosmic - rays present on the images should have been caught by the detection software .
the minimum two - detection criterion also filters out those that the software misses .
however , two classes of spurious object are missed by both methods .
firstly , a cosmic - ray that occurs close enough to a real object will avoid removal if not detected by the software as a separate entity .
left uncaught , it will appear as a flux change in the nearby object , with very similar characteristics as genuine emission - line flux on continuum .
secondly , ghost images are usually admitted by detection algorithms as real objects because they show little change in flux or position between the frames of a single raw scan .
it is only through a frame - by - frame comparison with the _ cleaned _ images that they become apparent .
appendix [ ghostfamilies ] discusses the different types of ghost images and strategies for minimising their occurrence at the outset , before they become part of the data .
such spurious detections can be handled through the compilation of a catalogue of all cosmic - ray and ghost - affected pixels in @xmath89-space , and cross - checked with the @xmath89 object list .
any matches in the latter can then be removed .
the straight - summed and median - filtered scans ( scans ( ii ) and ( iii ) of sect .
[ s : coaddition ] ) can be searched by taking each matching pair of images and computing the ratio image @xmath94 here , @xmath95 and @xmath96 are a pair of straight - summed and median - filtered images , @xmath97 is an offset that governs search sensitivity and @xmath98 is the ratio of the two images measured from stellar fluxes .
spurious pixels are those for which @xmath99 exceeds @xmath100 , the _ cosmic - ray threshold _ , which is sensitive to @xmath97 . the parameter @xmath97 should be fixed 100 and 500 adu ) while @xmath98 and @xmath100 vary freely with each image @xmath90 .
trial divisions can be used to optimise these parameters .
the final cosmic - ray / ghost catalogue is used in the emission - line search described in sect .
[ s : elcandextract ] . in the case where only a pair images exists ( and
median - filtering is not possible ) , the approach of cianci ( 2002 , in prep ) can be used to identify affected pixels .
the dithered images are registered and pixel fluxes compared as a graph between the two images . in regions where they are identical , the fluxes follow a @xmath101 . pixels that differ , whether affected by ghosts or cosmic rays , deviate from this straight line and those beyond some threshold can be removed .
alternatively , rhoads ( 2000 ) describes a convolution filtering - algorithm for cosmic - rays , using the image point - spread function minus a delta function as its kernel .
this method is also useful when there are too few frames for median filtering .
appendix [ weakcosmics ] describes how to test the effectiveness of a given cosmic - ray removal technique .
this is particularly valuable in the case of data containing hundreds of faint events .
telluric absorption features ( stevenson 1994 ) , that give rise to variations in atmospheric transparency with wavelength , affect the near - infrared spectral regions sometimes scanned by a tunable filter . therefore , before any emission - line candidate extraction takes place , it is necessary to register all object fluxes on to a common scale at an unaffected wavelength .
the normalisation coefficients can be calculated using a sub - sample of stellar fluxes from the field .
a linear fit of flux ratio versus off - axis radius accounts for off - axis changes not removed by the original flat - fielding .
the extraction of emission - line candidates then proceeds in three stages . 1 .
an initial search is made for all objects detected in two adjacent bands and no others .
such objects have no detected continuum on the remaining frames .
a cross - check can be made between these and the cosmic - ray / ghost catalogue ( sect .
[ s : cosmicray ] ) and matches in @xmath89-space discarded .
the catalogue of non - continuum detections can then be set aside .
the second stage is the search for objects with an emission - line superposed on continuum .
each object with a sufficient number of observations ( _ e.g. _ five ) can have a straight line iteratively fit to its continuum , initially with no rejection .
the root - mean - square ( rms ) scatter @xmath102 about the line can then be compared to the mean of the flux measurement uncertainties , @xmath103 .
the larger of the two is taken as the dominant source of uncertainty , @xmath104 , against which all subsequent deviations should be measured .
if @xmath105 , successive fits should be weighted by the @xmath106 measurement errors ; otherwise , fits are best left unweighted .
points can then be rejected if they deviate by more than @xmath104 _ above _ the continuum .
a limit needs to be placed on the total number of points allowed to be deviant ( _ e.g. _ two ) , provided a sufficient number of points remain to constrain the continuum fit .
identical reasoning applies to the rejection of deviations _ below _
the line , ( as in searches for absorption features ) , which we do not go into here .
objects can be retained as emission - line objects if ( 1 ) the deviation is greater than the chosen threshold above the line , and ( 2 ) a cross - check with the cosmic - ray / ghost catalogue reveals no matches .
3 . the final stage is processing of the raw object - detection catalogues in a search for object detections on a single frame .
cross - checking between these detections and the cosmic - ray / ghost catalogue is made in the same way as for the double - detection objects .
however , the final catalogues can be too large ( several hundred such objects per field ) to inspect manually in the same fashion .
many of the single - frame detections are noise - peaks arising from the low detection threshold deliberately employed for object detection .
therefore , the flux threshold cut applied to the single - frame detections should be determined from the double - detections . upon completion ,
all three catalogues of non - continuum and continuum - fit objects can be combined .
figure [ f : inspect ] shows examples of distant emission - line galaxies detected for the _ ttf field galaxy survey _ ( jones & bland - hawthorn 2001 ) . both the strip - mosaic ( fig .
[ f : inspect]@xmath5 ) and spectral scan ( fig .
[ f : inspect]@xmath6 ) of each object are shown .
when we build up a data cube or take a series of observations , it is essential to think of the scan variable as the etalon gap @xmath107 rather than wavelength .
the wavelength range is moderated by the filter ; the etalon gap is not .
the physical plate scanning range is @xmath108 @xmath109 @xmath110 where @xmath111 is the zeropoint gap and @xmath112 is the free spectral range in physical gap units . with this important distinction in mind , for the flux in a standard star observation
, we are able to write @xmath113 where @xmath114 is the product of the stellar spectrum and the filter response
. the limits of the integral in eqn .
( [ integral1 ] ) are defined by the bandpass of the entrance filter.[multiblock footnote omitted ] bland - hawthorn ( 1995 ) gives a worked example of standard star flux measurement using a fabry - perot and the definitions above .
efficiency measurements are best made on a nightly basis and observed at a range of airmass encompassing the science observations where possible .
good sources of suitable standards are bessell ( 1999 ) and hamuy ( 1994 ) , although the latter have not removed the effects of atmospheric ( telluric ) absorption features from their measurements , with the partially saturated bands of h@xmath115o subject to variation with humidity and airmass ( stevenson 1994 ) .
these standards are sufficiently bright and isolated to permit measurement through large 10 20 apertures .
the distance of the star from the optical axis also needs to be measured and the appropriate wavelength correction applied ( bland - hawthorn & jones 1998b ) . since it is impractical to observe a flux standard at every pixel position , we can only flux calibrate the spectral response at each point in the field through the whitelight cube .
thus , we effectively calibrate the whitelight response at the position of the flux standard and thereafter the data cube .
the total efficiency @xmath116 of the telescope , optics and detector is the ratio @xmath117 of measured to published flux from the standard .
measured fluxes can be converted to true flux ( ) from adu through @xmath118 where @xmath119 is the ccd gain ( ) , @xmath120 is the energy of a photon ( @xmath121 ) of that wavelength ( ) , @xmath122 is the exposure time ( s ) , @xmath123 is the area of the telescope primary ( @xmath124 ) and @xmath125 is effective passband width ( ) as described in sect .
[ jacquinot ] .
@xmath126 is a correction for extinction , @xmath127 dependent upon both the extinction coefficient @xmath128 for the site and the mean airmass @xmath129 of the observation .
[ f : effic ] shows nightly efficiency measurements for the ttf at the aat with two types of ccd ( mit - ll and tektronix ) , over observations spread across four observing semesters ( two years ) .
the scatter about the mean is typically a few percent in each case .
this is most likely due to variations in the photometric quality between the nights . in this particular case ,
the design efficiencies of the individual system components are 65% for 2 , 90% for the aat , 70% for the ccd and 90% for ttf .
combined , these yield @xmath130% throughput , in broad agreement with our measurements . with a knowledge of @xmath116
we can convert directly from observed flux ( in adu ) to true flux ( in 2band ) . in a single observation ,
@xmath131 this is the _ total _ flux within the passband , not per unit wavelength .
implicit in it is the assumption of a linear ccd detector .
tunable filter object photometry typically consists of co - added observations at different airmass or differing exposure time .
if the co - added flux of an object ( in adu ) is the sum of individual frames , @xmath132 then that same object will have true flux given by @xmath133 where @xmath119 , @xmath134 and @xmath116 are constant for exposures of a given wavelength . at this point
, some additional flux calibrations may be necessary depending on the nature of the data . in point - source measurements for which the apertures have been chosen to optimise signal - to - noise ,
a correction is necessary to derive total fluxes .
one advantage of spectral imaging with a fabry - perot is that the object aperture can be optimised _
after _ the observation rather than set at the _ time _ of observation , as is the case with all slit spectrometers .
in this review we have outlined the main principles behind fabry - perot detection and spectrophotometry from narrowband scans of fabry - perot data , with particular reference to low resolution tunable filter data .
we have also discussed the most common limitations which need to be addressed for reliable photometric calibration .
the imaging fabry - perot interferometer has the capability to provide superior spectrophotometry , since slit - aperture devices suffer seeing losses and narrowband filters are tacitly assumed to have constant transmission properties as a function of both position and wavelength .
fabry - perot interferometers are still not common - user instruments at any observatory for a variety of reasons , most notably because of the restricted wavelength coverage . but for studying extended emission from a few bright lines , the capabilities of the fabry - perot are unmatched by any other technique , with the exception of the imaging fourier transform spectrometer .
figure [ f : qso ] shows tunable filter imaging of the faint h@xmath4 extended emission around the @xmath135 quasar mr2251 - 178 , obtained by shopbell , veilleux & bland - hawthorn ( 1999 ) using techniques described in this paper .
fabry - perot interferometers ( including ttf ) differ from an ideal tunable monochromator in three areas .
first , the triangular rather than square shape of the airy profile limits sampling efficiency and complicates flux calibration .
a potential solution is the development of a ` double cavity ' interferometer which squares the lorentzian instrumental profile while maintaining high throughput ( vand de stadt & muller 1985 ) .
second , the use of fixed blocking filters limits the true tunable potential of the device .
the use of two or more fabry - perot interferometers in series can bypass this restriction , thereby making a fully tunable system ( meaburn 1972 ) .
however , such a system comes at the cost of increased operating complexity .
an intermediate solution is the use of custom - built rugate multiband blocking filters , transmitting two widely - separated intervals simultaneously ( cianci , bland - hawthorn & obyrne 2000 ) .
finally , the phase change across the field presents a position - dependent passband that restricts the angular size of emission that can be imaged at any one time . however , there are many promising technologies that look set to take tunable imaging beyond the fabry - perot interferometer .
acousto - optic tunable filters are an alternative where the active optical element comprises piezoelectric transducers bonded to an anisotropic birefringent material ( harris & wallace 1969 ) . an ac signal through the transducers ( 10 - 250 mhz ) , creates sonic waves that vibrate the crystal where it acts like a diffraction grating .
such tunable filters are currently limited by the expense and difficulty at manufacturing suitably - sized crystals of good imaging quality .
the power to drive the vibrations is high ( several watts per square centimetre ) and there are difficulties concerning heat dissipation .
liquid crystal filters are an alternative technology that are now commercially produced , albeit with transmissions ( @xmath136% ) that remain too low for nighttime astronomical work .
solc filters ( solc 1959 ; evans 1958 ) are tunable devices utilising a pair of polarisers with a sequence of phase retarders set to differing position angles . with a solc filter it is possible to tune the _ shape _ of the spectral profile , in addition to its placement and dimensions .
the most advanced alternative tunable technique is that of the tunable lyot filter , ( lyot 1944 ) , consisting of many birefringent crystals in series .
birefringent filters cause incident light of two polarisation states to interfere with each other .
the resulting bandpass is tuned by rotating all elements and its width is set by the thickness of the thinnest crystal .
such systems are already in wide use at solar observatories and we believe it will not be long before such systems find their way into nighttime work .
the main drawback for astronomical use is that half the light is lost at the entrance polariser . the widefield tunable lyot design of bland - hawthorn ( 2001 ) places such an arrangement at the prime focus of the telescope .
not only does this combat the light - loss problem , but it gives the filter access to wide sky angles ( and short @xmath40-ratios ) that would never be feasible with a fabry - perot instrument . in the mean time ,
however , a low - order fabry - perot etalon such as ttf presents the best high - throughput system capable of tunable imaging .
we acknowledge c. tinney for the preparation of fig .
[ radial ] and g. cecil for fig .
[ f : ghosts ] .
we also thank s. serjeant who contributed to the derivation of the test in appendix [ weakcosmics ] .
anglo - australian observatory directors past and present , r. d. cannon and b. j. boyle , and the head of instrumentation , k. taylor , have strongly supported development of fabry - perot instrumentation at the aat .
not least , our thanks are also due to the technical and support staff at the aat for their assistance in implementing new and difficult techniques .
we would especially like to mention j. r. barton , e. j. penny , l. g. waller , t. j. farrell and c. mccowage for all the necessary upgrades to both hardware and software .
we would like to thank j. baker and d. rupke for their comments on earlier versions of this paper , and to the referee j. meaburn , whose comments improved the content of the final draft in several key areas
. finally , we acknowledge the australian time allocation committee ( 19961998 ) for their generous allocations of aat time upon which much of this work was fundamentally dependent .
= 0.5 in = 1 acker , a. , ochensenbein , f. , stenholm , b. , tylenda , r. marcout , j. , mischon , c. 1992 , the strasbourg - eso catalogue of galactic planetary nebulae ( eso : garching ) = 0.5 in = 1 atherton , p. d. , reay , n. k. , ring , j. , hicks , t. r. 1981 , opt eng 20 , 806 = 0.5 in = 1 atherton , p. d. , reay , n. k. , 1981 , mnras 197 , 507 = 0.5 in = 1 bacon , d. j. , refregier , a. r. , ellis , r. s. 2000 , mnras 318 , 625 = 0.5 in = 1 baker , j. c. , hunstead , r. w. , bremer , m. n. , bland - hawthorn , j. , athreya , r. a. and barr , j. 2001 , aj 121 , 1821 = 0.5 in = 1 bertin , e. , arnouts , s. 1996 , a&a 117 , 393 = 0.5 in = 1 bessell , m. s. 1999 , pasp 111 , 1426 = 0.5 in = 1 bland , j. , tully , r. b. 1989 , aj 98 , 723 = 0.5 in = 1 bland - hawthorn , j. , shopbell , p. l. , cecil , g. 1992 , in astronomical data analysis : software & systems @xmath27 i , eds . d. worrall , c. biemsderfer , j. barnes ( asp : san francisco ) , p. 393
= 0.5 in = 1 bland - hawthorn , j. 1995 , in tridimensional optical spectroscopic methods in astrophysics , asp conference series volume 71 , eds .
g. comte , m. marcelin ( asp : san francisco ) , p. 72 = 0.5 in = 1 bland - hawthorn , j. , jones , d. h. 1998a , pasa , 15 , 44 = 0.5 in = 1 bland - hawthorn , j. , jones , d. h. 1998b , in optical astronomical instrumentation , proc spie 3355 , ed .
s. dodorico , ( spie : bellingham , washington ) , 855 = 0.5 in = 1 bland - hawthorn , j. , van breugal , w. , gillingham , p. r. , baldry , i. k. , jones , d. h. 2001 , apj accepted = 0.5 in = 1 bookstein , f.l .
1979 , comp . graph . & image proc .
, 9 , 56 = 0.5 in = 1 cecil , g. , 1988 , apj 329 , 38 = 0.5 in = 1 cianci , s. 2002 , _ phd dissertation _ , university of sydney , in prep .
= 0.5 in = 1 cianci , s. , bland - hawthorn , j. , obyrne , j. w. 2000 , in optical and ir telescope instrumentation and detectors , proc spie 4008 , ( spie : bellingham , washington ) , 1368 = 0.5 in = 1 deutsch , e.w . , margon , b. , bland - hawthorn , j. 1998 , pasp 110 , 912 = 0.5 in = 1 evans , j. w .. 1958 , j opt soc amer 48 , 142 = 0.5 in = 1 gander , w. , von matt , u. 1993 , in scientific problems in scientific computing , eds . w. gander & j. hrebicek , ( springer - verlag ) , p 251 = 0.5 in = 1 glazebrook , k. , bland - hawthorn , j. , 2001 , pasp 113 , 197 = 0.5 in = 1 haffner , l. m. , reynolds , r. j. , tufte , s. l. 1999 , apj , 523 , 223 = 0.5 in = 1 hamuy , m. , suntzeff , n. b. , heathcote , s. r. , walker , a. r. , gigoux , p. , phillips , m. m. 1994 , pasp 106 , 566 = 0.5 in = 1 harris , s. e. , wallace , r. w. 1969 , j. opt . soc .
, 59 , 744 = 0.5 in = 1 jacquinot , p. 1954 , j opt soc amer 44 , 761 = 0.5 in = 1 jacquinot , p. 1960 , rep prog phys 23 , 267 = 0.5 in = 1 jones , d. h. , bland - hawthorn , j. 2001 , apj 550 , 593 = 0.5 in = 1 jones , d. h 1999 , _ phd dissertation _ , australian national university = 0.5 in = 1 jones , d. h. , bland - hawthorn , j. 1998 , pasp 110 , 1059 = 0.5 in = 1 joseph , s. h. 1994 , comp graph & image proc , 56 , 424 = 0.5 in = 1 laval , a. , boulesteix , j. , georgelin , y. p. , georgelin , y. m. , marcelin , m. 1987 , a&a , 175 , 199 = 0.5 in = 1 lissberger , p. h. , wilcock , w. l. 1959 , j opt soc amer 49 , 126 = 0.5 in = 1 lyot , b. 1944 , ann .
7 , 31 = 0.5 in = 1 meaburn , j. 1972 , a&a 17 , 106 = 0.5 in = 1 meaburn , j. 1976 , detection and spectrometry of faint light , dordrecht : reidel = 0.5 in = 1 rhoads , j. e. 2000 , pasp 112 , 703 = 0.5 in = 1 roesler , f. l. 1974 , meth expt phys , 12a , chap . 12 = 0.5 in = 1 shopbell , p. l. , bland - hawthorn , j. 1998 , apj 493 , 129 = 0.5 in = 1 shopbell , p. l. , veilleux , s. , bland - hawthorn , j. 1999 , apj 524 , l83 = 0.5 in = 1 stilburn , j. r. 2000 , in optical and ir telescope instrumentation and detectors , proc spie 4008 , eds .
m. iye , a. f. moorwood , ( spie : bellingham , washington ) , 1361 = 0.5 in = 1 solc , i. 1959 , czech j phys , 9 , 237 = 0.5 in = 1 stevenson , c. c. 1994 , mnras 267 , 904 = 0.5 in = 1 taylor , k. , atherton , p. d. 1980 , mnras 191 , 675 = 0.5 in = 1 thorne , a. p. 1988 ,
spectrophysics , chapman & hall , london = 0.5 in = 1 tinney , c. g. , tolley , a. j 1999 , mnras 304 , 119 = 0.5 in = 1 valdes , f. 1993 , _ focas user s guide _ , noao = 0.5 in = 1 van de stadt , h. , muller , j. m. 1985 , j. opt .
a 2 , 1363 = 0.5 in = 1 veilleux , s. , bland - hawthorn , j. , cecil , g. , 1999 , aj 118 , 2108
* circular fitting : * given three well - separated positions @xmath137 , @xmath138 and @xmath139 along the ring , the centre @xmath92 can be found by solving the determinant @xmath140 * elliptical fitting : * for an ellipse centred on the origin , we write @xmath141 for which @xmath142 is symmetric and positive definite . here , we have introduced a coordinate vector @xmath143 . in order to derive a form invariant under transformation ,
we replace @xmath144 with @xmath145 .
this leads to a similar form to eqn .
( [ ellip1 ] ) where @xmath142 is replaced by @xmath146 , @xmath147 becomes @xmath148 , and @xmath149 is replaced by @xmath150 .
the eigenvalues @xmath151 for @xmath142 are obtained from a suitable choice of @xmath152 ( @xmath153 ) , and @xmath154 can be chosen to ensure @xmath155 .
it follows that @xmath156 where @xmath149 is defined in the transformed frame .
this is the equation of an ellipse subject to @xmath157 since @xmath142 has the same real eigenvalues before and after transformation , all functions of the eigenvalues are invariant under transformation .
the eigenvalues share an important relationship with the coefficients of @xmath142 , i.e. @xmath158 moreover , the geometric properties of the ellipse before transformation are @xmath159 where @xmath160 are the semi - major axes ; @xmath154 defines the origin of the ellipse .
finally , once each of the @xmath161 data points are placed in eqn .
( [ ellip1 ] ) , we arrive at a highly overdetermined , nonlinear system of @xmath161 equations in the 4 unknown geometric parameters .
this is a standard problem which can be solved efficiently with any number of gradient search algorithms subject to a constraint .
the restriction needs to be invariant to the coordinate transformation : simply setting one of the unknowns to unity is not normally sufficient to ensure convergence .
two useful constraints ( everywhere non - zero ) are @xmath162 ( gander & von matt 1993 ) and @xmath163 ( bookstein 1979 ) .
it is important that the method , e.g. levenberg - marquardt , be efficient since @xmath164 typically in ttf analysis .
there are several possible weighting schemes which have varying merits for different data sets .
the most common is flux weighting to ensure the fitting establishes the major axes ( radius ) as accurately as possible .
internal reflections constitute a challenge for all optical systems , particularly imagers with dispersing optics at the pupil , such as tunable filters and wollaston prisms .
even a minimal arrangement can have eight or more optically flat surfaces ( _ e.g. _ blocking filter , ccd window ) . at some level ,
all of these surfaces interact separately to generate spurious reflections .
the periodic behaviour of the tunable filter always requires that we use an interference filter somewhere in the optical path , to block light from extraneous orders .
the filter introduces ghost reflections within the fabry - perot optics .
such a filter placed in the converging beam before the collimator or after the camera lens generates a distinct pattern of _ diametric _ ghosts .
a filter placed in the collimated beam has its own distinct family of _ exponential _ ghosts which are the most difficult to combat .
the pattern of ghosts imaged at the detector is different in both arrangements , as illustrated in fig . [
f : rays ] . examples of the appearance of such ghosts are shown in fig .
[ f : ghosts ] .
another ghost family arises from the down - stream optical element ( _ extraneous etalon _
the large optical gap of the outer plate surfaces produces a high - order airy pattern at the detector .
it is possible to suppress this signal by wedging the outer surfaces .
however , if we are forced to operate the blocking filters in the collimated beam , this compounds the problem further . to a large extent , we can minimize the impact of diametric and exponential ghosts through the use of well - designed anti - reflection coatings .
the reflectivity @xmath71 of an air - glass interface is @xmath165 where @xmath166 and @xmath167 .
this can be significantly reduced to 1% or better by the application of an ar coating ( half - optical band ) , or to below 0.1% for ar coatings customized to the particular filter in question .
coating performance is currently limited by the availability of pure transparent dielectrics with high refractive indices .
a promising alternative is the use of silica sol - gel coatings , that reduce the reflective losses inherent with traditional mgf@xmath115 coatings ( stilburn 2000 ) .
the durability of these coatings has been improved over recent years to the point where they are now being routinely used on instruments for 4 and 8 m - class telescopes with many air - glass surfaces .
a more difficult reflection problem arises from the optical blanks which form the basis of the etalon .
these can act as internally reflecting cavities since , from sect .
[ response ] , if we let @xmath168 and @xmath169 ( air - glass ) , it generates a ripple pattern with a finesse close to unity .
the large optical gap of the outer plates produces a high - order airy pattern at the detector ( fig .
[ f : ghosts ] ) . traditionally , the outer surfaces have been wedge - shaped to deflect this spurious signal out of the beam . however , if the surfaces are not coated with suitable anti - reflection coatings then the problem is compounded by an additional set of ghosts ( fig . [ wedge_ghosts ] ) .
these spurious reflections diverge as a result of the wedge angle .
furthermore , reflections from the front plate only occur for monochromatic sources , thereby complicating identification in fields containing both stars and emission - line sources .
even after we have paid full attention to minimizing _ air - glass _ reflections , there remains one fundamental ghost arising from the ccd .
as we see from the equation above , silicon ( @xmath170 ) leads to an air - silicon reflection of 30% . a modern prescription broadband ar coating for a ccd reduces this by a factor of 3 , meaning that ghosts reside at around 10% . since we can not customize the ccd response for a narrow wavelength range , bright ` in - focus ' diametric ghosts must always occur with a tunable filter imager .
a good way to track these down is to place a regular grid of holes in focus at the focal plane and illuminate the optical system with a whitelight source ( sect .
[ holes ] ) , tilting the etalon such that ghost images of the grid avoid the detector area .
if tilting is not possible , then dithering and median filtering are required , ( sect .
[ s : cosmicray ] ) .
here , we describe a simple test to evaluate the effectiveness of any software in removing weak cosmic - ray events . the method is to compare the probability distribution function ( pdf ) of cosmic ray events between an observation and a dark frame matched in exposure time .
we illustrate the basic idea using dark exposures from a tektronix @xmath67 ccd with varying exposure lengths and read - out times .
we used the tekronix ccd at the aat 3.9 m with varying exposure lengths ( 15 , 30 , 60 , 120 min ) and read - out times ( fast , slow , xtraslow ) .
the histogram of each frame shows the contribution from the bias , read and dark noise .
a millisecond exposure was used to remove the bias and read noise contribution to each histogram . the additional contribution from the dark noise
was well calibrated at 0.11 cts pix@xmath0 ksec@xmath0 .
it is assumed that the remaining events are related to cosmic rays .
we define @xmath171 to be the number of cosmic ray events with energies ( expressed in counts ) in the range @xmath172 .
the cumulative distribution is then @xmath173 when @xmath174 , the slope of the plot @xmath175 vs. @xmath176 is proportional to @xmath177 . in fig .
[ f : cosmics ] ( _ top _ ) , the noisy histogram is the bias / dark / read noise subtracted dark frame .
the monotonic curve is a plot of @xmath175 vs. @xmath176 which is found to be rather well defined and reproducible over the different exposures .
the pdf for the data is determined from all events identified by the deglitch algorithm .
since the energetic events are easier to find , the bright end of both pdfs will be well matched . in fig .
[ f : cosmics ] ( _ bottom _ ) , where the deglitch pdf turns over at low energy presumably but not necessarily at an energy greater than or equal to the turnover in the dark pdf gives some idea as to how effective the algorithm has been in removing the weaker events .
.cyclic functions which are periodic over @xmath26 with fwhm @xmath25 .
the ( g)aussian , ( l)orentzian and ( a)iry functions are illustrated in fig .
[ f : cyclic ] .
the _ mod _
function is the modulo function and @xmath178 .
note that for large @xmath4 , @xmath179 .
the gap scanning variable @xmath29 is offset by @xmath180 in practice .
[ cols= " < , < , < " , ] | the past five years have seen a rapid rise in the use of tunable filters in many diverse fields of astronomy , through taurus tunable filter ( ttf ) instruments at the anglo - australian and william herschel telescopes . over this time we have continually refined aspects of operation and developed a collection of special techniques to handle the data produced by these novel imaging instruments . in this paper , we review calibration procedures and summarize the theoretical basis for fabry - perot photometry that is central to effective tunable imaging .
specific mention is made of object detection and classification from deep narrowband surveys containing several hundred objects per field .
we also discuss methods for recognizing and dealing with artefacts ( scattered light , atmospheric effects , etc . ) which can seriously compromise the photometric integrity of the data if left untreated .
attention is paid to the different families of ghost reflections encountered , and strategies to minimise their presence . in our closing remarks ,
future directions for tunable imaging are outlined and contrasted with the fabry - perot technology employed in the current generation of tunable imagers .
2taurus-2 2bandergs@xmath0@xmath1band@xmath0 3mpc@xmath2 3mpc@xmath3 [ firstpage ] methods : data analysis , observational techniques : photometric instrumentation : fabry - perot interferometers |
recently inflation has become an essential part of our description of the universe .
not only does it solve the classical cosmological problems of flatness , horizon and relics , but also provides precise predictions for the primordial density inhomogeneities , predictions that are in good agreement with existing observations@xcite .
albeit these successes , the specific details of inflation are still unknown , since many different physical mechanisms and fields may generate a phase of accelerated cosmic expansion . for simplicity ,
it has become common practice to employ a scalar field , the inflaton , with a simple law for its potential , to generate inflation .
this picture is usually understood as the effective counterpart of a deeper and probably more complicated theory . in this context ,
the most generally employed inflationary theory is the so - called chaotic inflation , characterized by a simple quadratic potential with an initial high field value for the inflaton .
recently , due to advances in data quality and in anticipation of the data from the planck satellite @xcite and the large synoptic survey telescope ( lsst ) @xcite ) , for example , several groups started investigating more complicated forms for the inflaton potential to explain the present observational data .
@xcite used the cmb data to look for the presence of a general sinusoidal oscillation imprinted on the inflaton potential , for large field values ( i.e. , large spatial dimensions ) , placing strong limits on the amplitude of these osillations .
_ @xcite found with a 99,995% confidence level an oscillatory modulation for large spatial dimensions at @xmath3 , performing a monte - carlo markov - chain analysis using the cmb data , confirming similar results obained from the analysis of the cmb data using different techniques @xcite . from a theoretical perspective , features in the inflaton potential are well motivated .
@xcite showed that a class of models derived from supergravity theories gives rise to inflaton potentials with a large number of steps , each of these corresponding to a symmetry - breaking phase transition in a field coupled with the inflaton .
also , many transplanckian effects may generate oscillations in the primordial power spectrum of density inhomogeneities which could be described by an effective oscillating inflaton potential @xcite .
a present major problem in astrophysics is the large discrepancy between the predicted and the observed number of dark matter halos of mass @xmath4 .
n - body simulations designed to probe the formation and evolution of dark matter structures on small scales found @xmath5 dark matter halos with masses from @xmath6 to @xmath7 @xcite .
however , very much fewer small galaxies of comparable masses are observed in the local group @xcite .
several solutions to this number discrepancy have been proposed , ranging from selection effects in the observations @xcite to complex baryonic interactions that may have swept out the baryonic gas from the small halos @xcite .
the former explanation for the discrepancy can only dimish the problem but not resolve it : taking into account the limitations of present day observations , one may extrapolate the number of dwarf galaxies to at most a few hundred .
the latter type of solution can only be tested with semi - analytical models ( for a review see @xcite ) , which strongly rely on the details of the physics of star formation , which is not completely understood .
the ejection of gas would leave thousands of empty small dark matter halos essentially intact .
there is still no clear evidence of the presence of these objects in the dynamics of the local group . in this work
we propose an alternative approach .
bearing in mind the growing plausability of oscillating features in the inflaton potential , we examine to what extent a simple localized oscillating modification of the inflaton potential for small field values can change the number of small dark matter halos .
we restrict our analysis to perturbations of the inflaton potential which are still in the slow - roll regime . in sections [ secsr ] and [ secmf ] we review the formalism and the derivation of the important relations used . in section [ secpot ]
we discuss two simple modified oscillatory potentials : a saw - tooth modification and a sinusoidal modification . in section [ secresults ]
we present our results and in section [ secconcl ] our conclusions .
the primordial universe is assumed to be filled by an approximately homogeneous scalar field , the inflaton , governed by a klein - gordon equation of motion @xmath8 the friedmann equation becomes @xmath9\mbox { .}\label{fried}\ ] ] in order to have an inflationary period , the second derivative of the field and the kinetic term of the friedmann equation must both be small when compared with the other terms .
this can be obtained under the conditions of slow - roll , @xmath10 and @xmath11 where we used the notation @xmath12 it is possible to associate the comoving wavenumber , @xmath13 , of each mode of the density of perturbations with the inflaton value , @xmath14 , when this mode was leaving the hubble sphere .
we find this using the well known ( e.g. @xcite ) expression for the number of e - folds , @xmath15 , @xmath16\ ] ] with @xmath17 the approximation used in eq .
( [ efolds ] ) is justified by the fact that the assumed deviations of the chaotic inflaton potential ( @xmath18 ) is very small ( the interesting deviations , shown in section [ secresults ] are of order @xmath19 ) .
the adimensional curvature power spectrum , @xmath20 , is related to the inflaton potential , in the slow - roll approximation , by @xmath21 and the normalization of the power spectrum used is the one obtained from wmap5 @xcite . from @xmath20 it is possible to obtain @xmath22 , the power spectrum of the density perturbations using ( see e.g. @xcite ) @xmath23 ^ 2\left(\frac{k}{a(t)\,h(t)}\right)^4 t^2(k){\mathcal p_{\mathcal r}}\label{prpd}\ ] ] where @xmath24 is the equation of state parameter of the dominant component at instant @xmath25 . for the transfer function , @xmath26
, we used the analytical fit obtained by @xcite , which takes into account the presence of dark energy and baryons .
the mass function gives the number density , @xmath27 of dm halos of mass between @xmath28 and @xmath29 . in order to calculate it
, we need first to obtain the variance from the expression @xmath30 where we adopted a gaussian window function @xmath31 in order to evaluate the mass function at the present time , we set @xmath32 using the press - schechter@xcite formalism , we have @xmath33 where @xmath34 is the critical density contrast , @xmath35 is the average density of the universe and the mass , @xmath28 , is related with the length , @xmath36 , from the expression @xmath37
the modified potential to which we will refer to as saw - tooth is constructed by substituting the small field part of the quadratic potential by an oscillatory linear modification .
let @xmath38 be the wavelength of the oscillation and @xmath39 the field when the modification begins . for @xmath40 , @xmath41 . for @xmath42
we define @xmath43 . for @xmath44 ,
the potential has the form , @xmath45 for @xmath46 , @xmath47 for @xmath48 , @xmath49 and for @xmath50 , @xmath51 the potential is shown in fig .
[ figsaw ] .
and @xmath52 in fig .
[ figsaw ] we take @xmath53 in order to obtain changes in the variance for mass scales @xmath54 .
we are in the parameter space region of slow - roll inflation . the part of the parameter space which allows for this regime i.e. , where the slow - roll parameters ( eq . ( [ slowroll1 ] ) and eq .
( [ slowroll2 ] ) ) are smaller than 1 is shown in fig .
[ figparamsaw ] . , where the saw - tooth potential is compatible with slow - roll inflation .
] we also analyse the case when the modification of the potential is a simple sine wave , @xmath55
\,\text { , } \ , f \leq f_a\\ a f^2 \mbox { , } f >
f_a\\ \end{array}\label{vsin } \right . \quad \mbox{,}\ ] ] which is plotted in fig .
[ figsin ] . and
@xmath52 the parameter @xmath39 was chosen using the same criterion as in the saw - tooth case .
we tested again for the area of the parameter space which was compatible with the slow - roll regime which is shown in fig .
[ figparamsin ] . , where the sinusoidal potential is compatible with slow - roll inflation .
we calculated the dimensional density perturbation power spectrum ( @xmath56 , following eq . ( [ prpd ] ) , which is shown , for representative parameter values , in fig .
[ figpksaw ] . and @xmath38 .
the continuous ( _ green _ ) curve corresponds to @xmath57 and @xmath58 ; the dotted ( _ blue _ ) curve , to @xmath59 and @xmath60 , and the dashed ( _ black _ ) curve , to @xmath61 and @xmath62 . ] the same quantity was calculated for the sinusoidal potential , as shown in fig.[figpksin ] . and
the continuous ( _ green _ ) curve corresponds to @xmath57 and @xmath58 ; the dotted ( _ blue _ ) curve , to @xmath59 and @xmath60 , and the dashed ( _ black _ ) curve , to @xmath61 and @xmath62 . ] in fig .
[ figratiosaw ] we plot the ratio of the mass function obtained from the saw - tooth potential to the mass function of a featureless quadratic potential ( i.e. , @xmath63 ) , for different values of the parameters @xmath64 and @xmath38 , with @xmath64 varying from 1.5% to 5% .
when using parameters @xmath65 ( i.e. , a 5% modification of the chaotic potential ) and @xmath60 , we find a @xmath66 suppression in the number of halos for masses between @xmath67 and @xmath68 . .
the continuous ( _ green _ ) curve corresponds to @xmath57 and @xmath58 ; the dotted ( _ blue _ ) curve , to @xmath59 and @xmath60 , and the dashed ( _ black _ ) curve , to @xmath61 and @xmath62 . ] in fig .
[ figratiosin ] we plot the ratio of the mass function obtained from the sinusoidal potential to the mass function of a featureless quadratic potential , again for different values of the parameters @xmath64 and @xmath38 . using parameters @xmath65 and @xmath60
, we find a @xmath69 suppression in the number of halos for masses between @xmath67 and @xmath68 . .
the continuous ( _ green _ ) curve corresponds to @xmath57 and @xmath58 ; the dotted ( _ blue _ ) curve , to @xmath59 and @xmath60 , and the dashed ( _ black _ ) curve , to @xmath61 and @xmath62 . ]
we modified the inflationary potential introducing two kinds of oscillatory patterns superimposed , for small field values , on a quadratic potential .
our modifications are small enough to be compatible with the slow roll conditions and consequently do not alter the usual and successful inflationary predictions for the large scale regime .
the first modification studied ( eq.s ( [ vsaw1])-([vsaw4 ] ) ) has the form of a succession of linear segments that creates an oscillating saw - tooth pattern which deviates from the standard quadratic potential by a factor @xmath64 . using the press - schechter formalism
, we found that the number of small mass halos can be strongly reduced for many choices of the parameters @xmath64 ( which characterizes the amplitude ) and @xmath38 ( which characterizes the wavelength ) .
for example , we found that the number of halos with masses between @xmath67 and @xmath70 decreases by 47% for @xmath65 and @xmath60 .
the second modification studied ( eq . ( [ vsin ] ) ) is a simple sine function multiplying the quadratic potential .
this modification allows one to capture the effects of the previous one without the discontinuities in the second derivatives .
once again we calculated the mass function using the press - schechter formalism and found a strong suppression in the number of small halos for a wide area of the parameter space .
as a representative example , we found that the number of halos with masses between @xmath67 and @xmath70 decreases by 54% for @xmath65 and @xmath60 .
we conclude that small oscillatory patterns on the inflaton potential can cause large changes in the predicted halo mass function .
in particular , if the oscillations begin in the @xmath0 range , for example , the oscillations can appreciably suppress the number of dwarf galaxies in this mass range , as observed .
it also appreciably increases the number of halos of mass @xmath71 by factors @xmath72 .
this might be a solution to the reionization problem where a very large number of population iii stars in low mass halos are required .
although we found that a small amplitude ( @xmath73 ) saw - tooth oscillatory inflaton potential , with a wavelength " @xmath60 , causes a factor of @xmath74 decrease in the number of halos of masses @xmath75 , present observations indicate a much larger depression although the upper limit of the number of observed dwarf galaxies is not well defined due to observational difficulties . from our fig .
[ figparamsaw ] , however , @xmath38 could be much less than 0.75 with @xmath73 , in particular , @xmath38 could be as small as @xmath76 , and still be compatible with slow - roll inflation . a detailed analysis of the parameter space @xmath77 , as well as more general inflation models than the simple single scalar chaotic inflation model which was used here , is thus required ( which is now in progress ) .
only then can we know whether a small amplitude oscillatory inflaton potential , alone , can resolve the problem of the lack of observed dwarf galaxies in the range @xmath0 .
l.f.s.r . thanks the brazilian agency cnpq for financial support ( 142394/2006 - 8 ) . r.o.thanks the brazilian agency fapesp (
06/56213 - 9 ) and the brazilian agency cnpq ( 300414/82 - 0 ) for partial support .
this research has made use of nasa s astrophysics data system . | a smooth inflaton potential is generally assumed when calculating the primordial power spectrum , implicitly assuming that a very small oscillation in the inflaton potential creates a negligible change in the predicted halo mass function .
we show that this is not true .
we find that a small oscillating perturbation in the inflaton potential in the slow - roll regime can alter significantly the predicted number of small halos .
a class of models derived from supergravity theories gives rise to inflaton potentials with a large number of steps and many transplanckian effects may generate oscillations in the primordial power spectrum .
the potentials we study are the simple quadratic ( chaotic inflation ) potential with superimposed small oscillations for small field values . without leaving the slow - roll regime
, we find that for a wide choice of parameters , the predicted number of halos change appreciably .
for the oscillations beginning in the @xmath0 range , for example , we find that only a 5% change in the amplitude of the chaotic potential causes a 50% suppression of the number of halos for masses between @xmath0 and an increase in the number of halos for masses @xmath1 by factors @xmath2 .
we suggest that this might be a solution to the problem of the lack of observed dwarf galaxies in the range @xmath0 .
this might also be a solution to the reionization problem where a very large number of population iii stars in low mass halos are required . |
riemann zeta function @xmath16 ( where the product is taken over the set @xmath0 of all primes ) converges for @xmath17 .
it has a analytic continuation to the complex plane with a simple pole at @xmath18 .
the riemann hypothesis ( rh ) states that non - real zeros all lie on the critical line @xmath19 .
rh has equivalent formulations , many of them having to do with the distribution of prime numbers @xcite .
let @xmath20 be the divisor function .
there exists a remarkable parallel between the error term @xmath21 in the summatory function of @xmath20 ( dirichlet s divisor problem ) and the corresponding mean - square estimates @xmath22 of @xmath23 on the critical line , see @xcite for a review
. this may explain ramanujan s interest for highly composite numbers @xcite .
a highly composite number is a positive integer @xmath10 such that for any integer @xmath12 , @xmath24 , i.e. it has more divisors than any positive integer smaller than itself .
this landmark work eventually led to robin s formulation of rh in terms of the sum of divisor function @xmath25 @xcite .
more precisely , @xmath26 the numbers that do not satisfy ( [ robin ] ) are in the set @xmath27 @xmath28 .
if rh is false , the smallest value of @xmath29 that violates the inequality should be a superabundant number @xcite , i.e. a positive number @xmath10 satisfying @xmath30 no counterexample has been found so far . in the present paper ,
robin s criterion is refined by replacing the sum of divisor function by the dedekind psi function @xmath31 .
should not be confused with the second chebyshev function @xmath32 since @xmath33 , with equality when @xmath10 is free of square , we establish the refined criterion @xmath34 the numbers that do not satisfy ( [ dedekind ] ) are in the set @xmath35 . a number that would possibly violate ( [ dedekind ] ) should satisfy @xmath36 and be a primorial number @xmath37 ( the product of the first @xmath10 primes ) or one its multiples smaller than @xmath14 ( sloane sequence @xmath15 ) .
according to a mertens theorem @xcite , one has @xmath38 and we show that none of the numbers larger than @xmath39 in the sequence @xmath15 can violate ( [ dedekind ] ) . as a result , rh may only be true . in the next section ,
we provide the proof of ( [ dedekind ] ) and compare it with the robin s criterion ( [ robin ] ) .
furthermore , we investigate the structure of numbers satisfying ( [ newsuperab ] ) and justify why they fail to provide counterexamples to rh . originally , dedekind introduced @xmath5 as the index of the congruence subgroup @xmath40 in the modular group ( see @xcite , p. 79 ) . in our recent work ,
the dedekind psi function plays a role for understanding the commutation relations of quantum observables within the discrete heisenberg / pauli group @xcite .
in particular , it counts the cardinality of the projective line @xmath41 of the lattice @xmath42 .
the relevance of the dedekind psi function @xmath5 in the context of rh is novel .
for other works aiming at a refinement of robin s inequality , we mention @xcite , @xcite and @xcite .
let us compute the sequence @xmath43 of all positive numbers satisfying ( [ newsuperab ] ) . for @xmath44 ,
one obtains @xmath45 , which are the first terms of sloane sequence @xmath15 , consisting of the primorials @xmath13 and their multiples up to the next primorial @xmath14 .
it is straightforward to check that about half of the numbers in @xmath46 are not superabundant ( compare to sloane sequence @xmath47 ) .
based on calculations performed on the numbers in the finite sequence @xmath46 , we are led to three properties that the infinite sequence @xmath15 should satisfy * proposition 1 * : for any @xmath48 such that @xmath49 one has @xmath50 . _
proof _ : one may use @xmath51 instead of @xmath52 to simplify the proof . when @xmath53 is prime , one observes that @xmath54 for some @xmath55 .
the corresponding dedekind psi function is evaluated as @xmath56 then , with @xmath57 one concludes that @xmath58 when @xmath53 is not prime , a similar calculation is performed by decomposing @xmath53 into a product of primes and by using the multiplicative property of @xmath5 .
* proposition 2 * : given @xmath59 , for any @xmath60 such that @xmath61 one has @xmath62 .
_ proof _ : this proposition is proved by using inequality ( [ newsuperab ] ) at @xmath63 @xmath64}{(l+1)n_n}~\mbox{for}~\mbox{any}~ m<(l+1)n_n\ ] ] and the relation @xmath65=(l+1)\psi(n_n)$ ] . as a result @xmath66 * proposition 3 * : there exist _ infinitely many _
prime numbers @xmath67 such that @xmath68 . _
remarks on the proposition 3 : _ based on experimental evidence in table 1 , one would expect that @xmath69 and that the proposition 3 is false .
similarly , one would expect that @xmath70 for any @xmath10 .
for instance it is known @xcite that @xmath71 .an excerpt of values of @xmath72 and @xmath73 versus the number of primes in the primorial @xmath13 . [ cols=">,>,>,>,>",options="header " , ] [ table2 ] according to the calculations illustated in the table ii and the bound established in ( [ lbound ] ) , there are no exceptions to the refined robin s criterion . since robin s criterion has been shown to be equivalent to rh hypothesis , rh may only be true .
the author thanks p. sol for pointing out his paper @xcite , following the presentation of @xcite at qupa meeting in paris on 09/23/2010 . he acknowledges j. l. nicolas for helping him to establish the proposition 3 .
he also thanks fabio anselmi for his sustained feedback on this subject . | let @xmath0 be the set of all primes and @xmath1 be the dedekind psi function . we show that the riemann hypothesis is satisfied if and only if @xmath2 for all integers @xmath3 ( d ) , where @xmath4 is euler s constant .
this inequality is equivalent to robin s inequality that is recovered from ( d ) by replacing @xmath5 with the sum of divisor function @xmath6 and the lower bound by @xmath7 . for a square free number ,
both arithmetical functions @xmath8 and @xmath9 are the same .
we also prove that any exception to ( d ) may only occur at a positive integer @xmath10 satisfying @xmath11 , for any @xmath12 , hence at a primorial number @xmath13 or at one its multiples smaller than @xmath14 ( sloane sequence @xmath15 ) . according to a mertens theorem ,
all these candidate numbers are found to satisfy ( d ) : this implies that the riemann hypothesis is true . |
he internet traffic data , which is usually modeled by the superposition of diverse components @xcite@xcite@xcite , is of critical importance to network management .
the baseline traffic represents the most prominent traffic patterns @xcite , and is helpful to anomaly detection , capacity planning , and green computing .
most traditional studies focused on baselining the single - link traffic @xcite , however , as the network - wide traffic analysis is becoming increasingly popular , baselining the network - wide traffic turns into a most urgent problem .
the traffic matrix is an important kind of network - wide traffic data and has complex characteristics .
the baseline of a traffic matrix should capture the common patterns among traffic flows , and be stable against the disturbance of large anomalies .
principal component analysis ( pca ) was used for traffic matrix analysis in @xcite , and showed the low - rank nature of the baseline ( i.e. the deterministic ) component , but it performed poorly in the presence of contamination by large anomalies @xcite .
recently , the robust principal component analysis ( rpca ) theory @xcite has attracted wide attentions .
supposing a low - rank matrix is contaminated by a sparse matrix whose non - zeros entries may have large magnitudes , candes et al .
@xcite presented a robust matrix decomposition method named principal component pursuit ( pcp ) , and proved that under surprising board conditions , pcp could recover the low - rank matrix with high probability . following this exact `` low - rank and sparsity '' assumption , bandara et al .
@xcite proposed the robust baseline ( rbl ) method , and argued that rbl performs better than several existing traffic baseline methods . intuitively , the baseline traffic time - series of each flow should be `` smooth '' enough , i.e. it records the long - term and steady traffic trends ( such as the diurnal pattern ) and neglects the short - term fluctuations .
however , to the best of our knowledge , this problem has not been adequately studied when baselining the traffic matrix .
actually , the resulting baseline traffic time - series for rbl are not smooth enough , and it can be explained that the empirical traffic matrix does not exactly meet the `` low - rank and sparsity '' assumption , on the contrary , it also contains non - smooth noisy fluctuations which should be decomposed specially .
furthermore , the evaluation of traffic baseline methods was not sufficient in previous studies .
a key point is that obtaining the ground - truth baseline of real - world traffic data is usually impossible , consequently , one could neither measure the accuracy of a baseline method , nor compare the performances of different baseline methods . therefore
, baselining network - wide traffic is still a challenging task . in this letter , we first establish a refinement of the traffic matrix decomposition model in @xcite , which extends the assumption of deterministic traffic matrix to characterize the smoothness of baseline traffic .
secondly , we propose a novel baseline method for the traffic matrix data , which solves a constrained convex program named stable principal component pursuit with time - frequency constraints ( spcp - tfc ) . as an extension of the stable principal component pursuit ( spcp ) method @xcite
, spcp - tfc takes distinctive time - frequency constraints for our refined model .
thirdly , we design the accelerate proximal gradient ( apg ) algorithm for spcp - tfc which has a fast convergence rate . at
last , we evaluate our traffic baseline method by abundant simulations , and show it has a superior performance than rbl and pca .
suppose @xmath0 is a traffic matrix , and each column @xmath1 ( @xmath2 ) is a traffic flow in @xmath3 time intervals .
a traffic flow is the traffic of a original - destination ( od ) pair @xcite , or the traffic traversing a unidirectional backbone link @xcite . in @xcite , we proposed the simple traffic matrix decomposition model ( tmdm ) , assuming @xmath4 is the sum of a low - rank matrix , a sparse matrix , and a noise matrix .
this model is equivalent to the data model of the generalized rpca problem @xcite , and the low - rank deterministic traffic matrix corresponds to the baseline traffic .
however , tmdm gives no temporal characteristics of the baseline traffic .
since each baseline traffic flow records the long - term and steady traffic trends , it should display a smooth curve .
plenty of mathematical tools , such as wavelets and splines , can formulate smoothness .
as the most salient baseline traffic patterns are slow oscillation behaviors , this letter formulates the baseline traffic time - series as the sum of harmonics with low frequencies , and establishes a refined traffic matrix decomposition model ( r - tmdm ) : * definition 1 * ( r - tmdm ) _ the traffic matrix @xmath0 is the superposition of the deterministic ( baseline ) traffic matrix @xmath5 , the anomaly traffic matrix @xmath6 , and the noise traffic matrix @xmath7 .
@xmath5 is a low - rank matrix , and for each column time - series in @xmath5 , the fourier spectra whose frequencies exceed a critical value @xmath8 are zeros ; @xmath6 is a sparse matrix , but the non - zeros entries may have large magnitudes ; @xmath7 is a random noise matrix , and each column time - series is a zero - mean stationary random process . _ for simplicity , we assume each column @xmath9 ( @xmath2 ) is the white gaussian noise with variance @xmath10 in this letter
. stable principal component pursuit ( spcp ) for the generalized rpca problem solves this convex program @xcite : @xmath11 where @xmath12 , @xmath13 , @xmath14 denote the nuclear norm , the @xmath15 norm , and the frobenius norm , respectively ; @xmath16 is the balance parameter , and @xmath17 is the constraint parameter . to extract the baseline traffic more powerfully , we extend spcp by preserving the objective function and redesigning the constraint functions .
firstly , considering the r - tmdm model , it is necessary to add a constraint for the baseline traffic matrix based on its frequency - domain assumption .
@xmath18_{t \times t}$ ] denotes the discrete fourier basis matrix of length @xmath3 . for each @xmath19 ,
the fourier basis @xmath20 is defined as @xmath21 with frequency @xmath22 .
@xmath23 is made up of the high - frequency bases @xmath20 satisfying @xmath24 , thus @xmath25 is the projection matrix to high - frequency subspace .
we use the following constraint for the baseline traffic matrix : @xmath26 secondly , unlike the frobenius norm inequality in ( [ spcp ] ) , we use a different constraint strategy for the noise traffic . for each column vector @xmath9 ( @xmath2 ) , consider its periodogram @xmath27 for the fourier frequencies @xmath28 : @xmath29 @xmath30 denotes the chi - square ( chi ) distribution with @xmath31 freedom degree , and it is well - known @xcite that when @xmath9 are i.i.d .
standard gaussian variables , for each column vector @xmath32 ( @xmath2 ) , we divide it by @xmath33 .
this operation normalizes the gaussian variables in @xmath7 , and preserves all the hypotheses of @xmath5 and @xmath6 in the r - tmdm model . ] , @xmath34 are i.i.d @xmath35 variables , and @xmath36 is a @xmath37 variable .
therefore , @xmath38 are i.i.d @xmath39 variables , and @xmath40 is a @xmath41 variable .
we choose these time - frequency constraints for the noise traffic matrix : @xmath42^{\top}n_{j}\right\|_{\infty } \leq \delta_{1},\\ \left|w_{0}^{\top}n_{j}\right| \leq \delta_{2 } , \\
\left\|n_{j}\right\|_{\infty } \leq \delta_{3},\\ \end{cases } \ \
j=1 , ... , p.\ ] ] equivalently , @xmath7 lies in the convex subset @xmath43 : @xmath44 the main idea of ( [ noise_constraint1 ] ) is to eliminate the low - frequncy baseline traffic from the resulting noise traffic by utilizing the value distribution of its periodogram , while the frobenius norm constraint in ( [ spcp ] ) could not filter the baseline traffic so efficiently .
lastly , using the new constraints ( [ baseline_constraint ] ) and ( [ noise_constraint1 ] ) , we propose our traffic baseline method , named stable principal component pursuit with time - frequency constraints ( spcp - tfc ) : @xmath45 which outputs @xmath5 as the baseline traffic matrix . in this study , we set @xmath46 , @xmath47 , corresponding to the @xmath48 confidence intervals of the @xmath39 , @xmath41 , and gaussian variables , respectively .
spcp - tfc solves a constrained program which is computational expensive .
the accelerate proximal gradient ( apg ) algorithm considers this relaxed unconstrained program : [ relaxation ] & _ a , e , n f(a , e , n ) g(a , e , n ) + f(a , e , n ) + & f(a , e , n ) x - a - e - n_f^2 + w _ ( w_)^a_f^2 + & g(a , e , n ) a _ + e_1 + _
b(_1,_2,_3)(n ) + & the function @xmath49 , which is convex and differentiable , penalizes violations of the two equality constraints in ( [ spcp - tfc ] ) , and @xmath50 is a balancing parameter .
@xmath51 is a linear combination of three convex but non - differentiable functions .
@xmath52 is a relaxation parameter , as @xmath53 , the solution to ( [ relaxation ] ) approaches to the solution to ( [ spcp - tfc ] ) .
@xmath54 denotes the gradient of @xmath55 , which equals to @xmath56a + e+n - x \\ a+e+n - x \\ a+e+n - x \\
\end{array } \right]\ ] ] and is lipschitz continuous by the following proposition . * proposition 1 * : _ @xmath57 is lipschitz continuous , and the lipschitz constant is @xmath58 . _ for the proof , see the appendix section . instead of directly minimizing @xmath59
, the apg algorithm minimizes a sequence of quadratic approximations , denoted as @xmath60 , of @xmath59 at a specially chosen point @xmath61 ( renewed in each step ) : @xmath62 it can be derived that [ apg_subproblem ] & _ a , e , n q(a , e , n , y^a , y^e , y^n ) + = & _ a \ { a _ + a - g^a _ f^2 } + + & _ e \ { e_1 + e - g^e _ f^2 } + + & _ n \{_b(_1,_2,_3)(n ) + n - g^n _ f^2 } + + & where @xmath63y^{a } + y^{e } + y^{n } - x \right)\\ g^{e } = y^{e } - \frac{1}{l}\left(y^{a } + y^{e } + y^{n } - x\right)\\ g^{n } = y^{n } - \frac{1}{l}\left(y^{a } + y^{e } + y^{n } - x\right ) \end{cases}\ ] ] thus ( [ apg_subproblem ] ) splits into three subproblems , which are equivalent to the _ proximity operators _ associated with the convex functions @xmath64 , @xmath65 , and @xmath66 , respectively . using basic results in @xcite
, they have explicit solutions as follow : _ a \ { a _ + a - g^a_f^2 } & = u _ [ ] v^ , & _ e \ { e_1 + e - g^e _ f^2 } & = _ [ g^e ] , & & _ n\ { _ b(_1,_2,_3)(n ) + n - g^n _ f^2 } = + & _ b(_1,_2,_3)[g^n ] , & where @xmath67 is the singular value decomposition ( svd ) of @xmath68 , @xmath69 is the projection operator to @xmath43 , and @xmath70 is the soft - thresholding operator with parameter @xmath71 @xcite . for each @xmath0 ,
@xmath72 \in\mathbb{r}^{t\times p } $ ] and it satisifies @xmath73(i , j ) \triangleq \mathrm{sign}(x(i , j))\max\{|x(i , j)|-\epsilon,0\}.\ ] ] * input * : traffic matrix @xmath74 .
* normalization : * @xmath75 .
* initialization : * @xmath76 ; @xmath77 ; @xmath78 . @xmath79 ; @xmath80 ; @xmath81 . @xmath82 ; @xmath83 .
* while * not converged * do * @xmath84 ; @xmath85 ; @xmath86 ; @xmath87y_{k}^{a}+y_{k}^{e}+y_{k}^{n}-x\right)$ ] ; @xmath88 ; @xmath89 ; @xmath90 $ ] ; // singular value decomposition .
@xmath91v^{\top}$ ] ; @xmath92 $ ] ; @xmath93 $ ] ; @xmath94 ; @xmath95 ; @xmath96 .
* end while * * output * : @xmath97 is the baseline traffic matrix .
algorithm [ algo1 ] presents the apg algorithm for spcp - tfc , moreover , the following theorem proves that algorithm [ algo1 ] has the @xmath98 convergence rate .
since the proof of this theorem is quite close to theorem 4.4 in @xcite , we omit it and directly summarize this result : * theorem 1 * _ let @xmath99 .
then for all @xmath100 , we have @xmath101 where @xmath102 is defined in algorithm [ algo1 ] , and @xmath103 is a solution to ( [ relaxation ] ) when @xmath104 . _
we evaluate spcp - tfc by simulated traffic matrices , because the simulation approach could provide the ground - truth baseline data , perform abundant experiments independently , and change parameters to verify an algorithm s stability .
consequently , we first introduces the simulation techniques .
suppose the network has @xmath105 nodes , the measurement window is one week , and the minimal time interval is @xmath106 = @xmath107 minutes . under this setting , each traffic matrix @xmath0 has @xmath108 = @xmath109 = @xmath110 od flows in @xmath3 = @xmath111 = @xmath112 time intervals , and it is generated by the following three traffic components .
( 1 ) for each flow @xmath113 , the baseline traffic @xmath114 is the sum of a content @xmath115 and a few _ sin _ functions : @xmath116 we assume that in each simulation , @xmath117 satisfies the gravity model and @xmath118 = @xmath119 . since the period of the @xmath120-th _ sin _ is @xmath121 hours , we choose @xmath122 = @xmath123 , and the periods of oscillation traffic are @xmath124 hours .
moreover , the phase parameters @xmath125 are uniform distributed in @xmath126 $ ] , and the amplitudes of the _ sin _ functions decay quickly as @xmath120 increases by @xmath127 = @xmath128 .
( 2 ) the non - zero entries are randomly distributed in @xmath129 positions of the anomaly traffic matrix @xmath6 , and for each flow @xmath113 , the volume of each anomaly is @xmath130 .
( 3 ) for each flow @xmath113 , the noise traffic @xmath9 consists of @xmath3 independent gaussian variables @xmath131 , where @xmath132 , @xmath133 controls the noise rate .
we simulate @xmath110 baseline traffic matrices simultaneously , and generate one anomaly traffic matrix and two noise traffic matrices for each of them ( @xmath134 is chosen as @xmath135 and @xmath136 , which means a low - level noise and a high - level noise , respectively ) .
therefore , @xmath137 simulated traffic matrices are used in experiment .
it can be verified that these matrices satisfy the r - tmdm model , and the critical frequency @xmath8 = @xmath138 .
spcp - tfc is compared with other two network - wide traffic baseline methods : rbl @xcite and pca @xcite .
we apply these methods to the simulation dataset , and evaluate them by three metrics . [ 0.5 ] = @xmath135 ; right group : @xmath134 = @xmath136 .
, title="fig : " ] the first metric is the _
normalized root mean squared error _ ( nrmse ) between the ground - truth baseline traffic matrix @xmath5 and its estimation @xmath139 : @xmath140 fig .
[ nrmse ] illustrates the nrmses of different baseline methods under two noise levels . for spcp - tfc
, we test different values of parameter @xmath141 .
it can be observed that , spcp - tfc archives significant lower nrmses than rbl and pca , and this result is stable in a large range of @xmath142 values . and our traffic baseline method shows the best performance when @xmath142 = @xmath143 : under the low ( high ) noise level , the median of nrmses for spcp - tfc is as @xmath144 as that for rbl , and is as @xmath145 as that for pca .
hence we only consider spcp - tfc with this fixed value in the following discussions .
[ 0.43 ] = @xmath135 ; right panel : @xmath134 = @xmath136.,title="fig : " ] the second metric characterizes the temporal correlation between each pair of ground - truth and resulting baseline traffic flows using their _ pearson correlation coefficient_. as a traffic matrix contains 100 flows , under each noise level , we compute @xmath146 = @xmath147 correlation coefficients for each baseline method .
we display boxplots of these coefficients in fig .
[ correlation ] .
it is shown that the correlation coefficients for spcp - tfc are very close to @xmath148 , and have an obviously greater median than rbl and pca .
thus the spcp - tfc method extracts the temporal characteristics of the baseline traffic more precisely .
[ 0.43 ] [ 0.43 ] .a comparison between different baseline methods on the normalized mean smoothness values [ cols="<,<,<,<,<",options="header " , ] the third metric characterizes baseline traffic on smoothness , which adds up the _ total variations _ of the traffic flows in baseline traffic matrix .
a small value of this metric indicates the traffic is sufficient smooth . applying this metric to the baseline results aforementioned
, we display the curve of 100 smoothness values ( arranged by traffic matrix i d ) for each baseline method , and for the 100 ground - truth values , in fig .
[ smoothness ] .
we then compute the mean smoothness of each baseline method , and normalize it by dividing the mean smoothness of the ground - truth baseline traffic matrices . these normalized mean smoothness values
( under two noise levels ) are summarized in tab .
[ meansmoothness ] .
the estimated baseline traffic matrices by rbl and pca are significantly more coarse than the ground - truth baseline traffic , and their mean smoothness values are larger than three times and eight times of the ground - truth value , respectively .
instead , the spcp - tfc method leads to more accurate baseline estimations on smoothness .
this is because the high - frequency traffic , which has larger total variation , can be successfully eliminated from the baseline by spcp - tfc . from fig .
[ nrmse]-[smoothness ] and tab .
[ meansmoothness ] , we can also observe that when the noise rate @xmath134 grows from @xmath135 to @xmath136 , the performance of our baseline method shows a decline : the nrmses raise ,
the correlation coefficients drop down , and the smoothness values of resulting baseline traffic diverge from the ground - truth more evidently .
this is mainly because the low - frequency component of the noise traffic , which could not be directly distinguished from the baseline traffic , is proportional to @xmath134 .
in this letter , we propose a refined traffic matrix decomposition model , and introduce a novel baseline method for traffic matrix named spcp - tfc , which extends spcp by using new constraint functions for the baseline traffic and the noise traffic , respectively .
we design an apg algorithm for spcp - tfc , whose convergence rate is @xmath98 .
our baseline method is demonstrated by abundant simulations . using three distinct metrics to evaluate the baseline results , we show spcp - tfc has a superior performance than rbl and pca . 1 a. lakhina , k. papagiannaki , m. crovella , c. diot , e. d. kolaczyk , and n. taft . structural analysis of network traffic flows .
sigmetrics perform .
61 - 72 , june 2004 .
v. w. bandara and a. p. jayasumana .
extracting baseline patterns in internet traffic using robust principal components .
ieee lcn , 2011 .
z. wang , k. hu , k. xu , b. yin and x. dong .
structural analysis of network traffic matrix via relaxed principal component pursuit .
computer networks , vol .
7 , pp . 2049 - 2067 , 2012 .
h. hajji . baselining network traffic and online faults detection .
ieee icc , 2003 .
b. rubinstein , b. nelson , l. huang , a. joseph , s. lau , s. rao , n. taft and j. tygar .
antidote : understanding and defending against poisoning of anomaly detectors .
acm imc , 2009 .
e. candes , x. li , y. ma , and j. wright .
robust principal component analysis ?
journal of the acm . vol .
3 , pp . 1 - 37 , 2011 .
z. zhou , x. li , j. wright , e. candes , and y. ma .
stable principal component pursuit .
ieee isit , 2010 .
p. kokoszka and t. mikosch .
the periodogram at the fourier frequencies .
stochastic processes and their applications , vol .
49 - 79 , 2000 .
z. lin , a. ganesh , j. wright , l. wu , m. chen and y. ma .
fast convex optimization algorithms for exact recovery of a corrupted low - rank matrix .
ieee camsap , 2009 .
a. beck and m. teboulle .
a fast iterative shrinkage - thresholding algorithm for linear inverse problems .
siam journal on imaging sciences , vol .
183 - 202 , 2009 .
p. l. combettes and j .- c .
proximal splitting methods in signal processing .
fixed - point algorithms for inverse problems in science and engineering , pp .
185 - 212 .
springer , new york , 2011 .
proof : for any two points @xmath149 and @xmath149 in @xmath150 , we have @xmath151(a_{1}-a_{2 } ) + ( e_{1}-e_{2 } ) + ( n_{1}-n_{2})\right\|_{f}^{2 } \\ & + 2\left\|(a_{1}-a_{2 } ) + ( e_{1}-e_{2 } ) + ( n_{1}-n_{2})\right\|_{f}^{2 } \\
\leq & \big { ( } \beta\left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2 } + \|a_{1}-a_{2}\|_{f } + \|e_{1}-e_{2}\|_{f } \\ & \ + \|n_{1}-n_{2}\|_{f } \big{)}^{2 } + \\ & 2 \big{(}\|a_{1}-a_{2}\|_{f } + \|e_{1}-e_{2}\|_{f } + \|n_{1}-n_{2}\|_{f } \big{)}^{2}\\ = & \beta^{2}\left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2 } + \\ & 2\beta\big { ( } \left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2}\|a_{1}-a_{2}\|_{f } + \\ & \quad \ \
\left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2}\|e_{1}-e_{2}\|_{f } + \\ & \quad \ \
\left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2}\|n_{1}-n_{2}\|_{f } \big { ) } + \\ & 3 \big{(}\|a_{1}-a_{2}\|_{f}^{2 } + \|e_{1}-e_{2}\|_{f}^{2 } + \|n_{1}-n_{2}\|_{f}^{2 } \big { ) } + \\ & 6 \big{(}\|a_{1}-a_{2}\|_{f}\|e_{1}-e_{2}\|_{f } + \|a_{1}-a_{2}\|_{f}\|n_{1}-n_{2}\|_{f}\\ & \ \ \ + \|e_{1}-e_{2}\|_{f}\|n_{1}-n_{2}\|_{f } \big { ) } \\ \leq & ( 3\beta + \beta^{2})\left\|w_{\mathrm{h}}(w_{\mathrm{h}})^{\top}(a_{1}-a_{2})\right\|_{f}^{2 } + \\ & ( 9+\beta)\big{(}\|a_{1}-a_{2}\|_{f}^{2 } + \|e_{1}-e_{2}\|_{f}^{2 } + \|n_{1}-n_{2}\|_{f}^{2 } \big{)}.\\ \end{split}\ ] ] as @xmath152 is a projection operator in @xmath153 , for each @xmath154 , @xmath155 \right\| \leq \left\|a_{1}(p)-a_{2}(p)\right\|.\ ] ] therefore , we have @xmath156 \right\|^{2 } \\
\leq & \sum_{p=1}^{p}\left\|a_{1}(p)-a_{2}(p)\right\|^{2 } = \|a_{1}-a_{2}\|_{f}^{2}. \end{split}\ ] ] add ( [ projection_property ] ) to ( @xmath157 ) and let @xmath58 , we have @xmath158 this completes the proof . @xmath159 | in this letter , we develop a novel network - wide traffic analysis methodology , named stable principal component pursuit with time - frequency constraints ( spcp - tfc ) , for extracting the baseline of a traffic matrix . following a refined traffic matrix decomposition model , we present new time - frequency constraints to extend stable principal component pursuit ( spcp ) , and design an efficient numerical algorithm for spcp - tfc .
at last , we evaluate the spcp - tfc method by abundant simulations , and show it has superior performance than other traffic baseline methods such as rbl and pca .
baseline of traffic matrix , robust pca , time - frequency constraint , numerical algorithm , simulation . |
the mathematical modeling of biological networks has focused on the influence of the network structure on the functional properties of the system @xcite .
insights provided by these studies have shown , for example , that structural modules and hierarchical organization in the network are often related to compartmentalization of functional processes @xcite .
this is important since intracellular processes are rarely carried out by individual elements , and often involve the coordinated activity of multiple genes , proteins , and biochemical transformations . because different components may be recruited for different processes , the most fundamental aspect of the dynamics of complex intracellular networks concerns precisely the characterization of the specific parts of the network that are active under given conditions .
recent research focused on the modeling of metabolic networks has shown that typical metabolic states tend to recruit a much larger number of reactions than states that maximize growth rate @xcite .
this counterintuitive property is important in multiple contexts .
for example , it provides a partial explanation for the apparent dispensability of a large fraction of genes in single - cell organisms @xcite , since the genes associated with reactions that become inactive in growth - maximizing states are expected not to be essential .
this also explains why genetic and environmental perturbations that cause growth defect are accompanied by a burst in reaction activity @xcite .
these bursts can be attributed to the transient activation of otherwise inactive reactions that are recruited by the suboptimal states that follow the perturbation @xcite , since such states tend to have a larger number of active reactions .
another important implication concerns the possibility of synthetic rescues @xcite , where the inactivation of one gene can be compensated by the targeted inactivation of other genes .
the inactivation of such rescue genes can thus allow the recovery of lost biological function .
this is possible in part because the rescue genes correspond to genes that would be inactive in an optimal state , so that disabling them helps bring the state of the system closer to the desired optimal state @xcite .
understanding the root causes of the reduced reaction activity in optimal metabolic states is then of significant interest in the characterization and study of cellular metabolism .
focusing on steady - state dynamics , here we use flux - balance based analysis and linear programming techniques to establish rigorous results on the number of reactions that can be active in a given metabolic network .
we derive conditions for a specific reaction to be inactive in all ( sec .
[ sec3 ] ) or active in almost all ( sec .
[ sec4 ] ) feasible metabolic states .
we also establish bounds for the number of reactions that can be active in states that optimize an arbitrary linear function of reaction fluxes ( sec .
[ sec5 ] ) , which are derived based on the duality principle in linear programming , and we study the uniqueness of the optimal solution for typical linear objective functions ( sec . [ sec6 ] ) .
finally , we implement numerical simulations in reconstructed _ escherichia coli _ and human metabolic networks , both to compare with the rigorous bounds and to consider nonlinear objective functions ( sec .
[ sec7 ] ) . taken together ,
our results show that the reduced number of active reactions in the optimal states of a linear objective function , including growth rate , are mainly determined by the presence of irreversible reactions in the network .
the irreversibility constraints are shown to play a role also in nonlinear objective functions , such as the aggregated flux and mass flow activity , whose optimal solutions are shown to have a number of active reactions comparable to the corresponding number for linear functions .
we consider time - independent metabolic states , which serve as an appropriate representation of the state of single cells at time scales much shorter than the lifetime of the cells , as well as of the average behavior of a large population of cells at arbitrary time scales in time - invariant conditions . under this
_ steady - state _ assumption , a cellular metabolic state is a solution of a homogeneous linear equation that accounts for all stoichiometric constraints , @xmath0 where @xmath1 is the @xmath2 stoichiometric matrix and @xmath3 is the vector of metabolic fluxes . the components of @xmath4 include the fluxes of @xmath5 internal and transport reactions as well as @xmath6 exchange fluxes , which model the transport of metabolic species across the system boundary
. constraints of the form @xmath7 imposed on the exchange fluxes are used to limit the maximum uptake rates of substrates in the medium .
additional constraints of the form @xmath8 arise for the reactions that are irreversible .
assuming that the cell s operation is mainly limited by the availability of substrates in the medium , we impose no other constraints on the internal reaction fluxes , except for the atp maintenance flux , which is set to a fixed positive value .
these additional constraints can be organized in the form @xmath9 the set of all flux vectors @xmath10 satisfying eqs .
( [ eqn : mb ] ) and ( [ eqn : const ] ) defines the _ feasible solution space _ @xmath11 , representing the capability of the metabolic network as a system . because the number of fluxes @xmath12 is larger than the number of metabolic species @xmath13 , system ( [ eqn : mb ] ) is under - determined and @xmath14 is generally high dimensional .
our study is formulated in the context of flux balance analysis @xcite , which is based on the maximization of a metabolic objective function @xmath15 within the feasible solution space @xmath14 ( the superscript @xmath16 is used to denote transpose ) .
this reduces to a linear programming problem @xmath17 where we set @xmath18 if @xmath19 is not bounded from below and @xmath20 if @xmath19 is not bounded from above . for a given objective function
, we can numerically determine an optimal flux distribution for this problem .
this formulation is also appropriate for the derivation of the rigorous results presented below . in the particular case of growth maximization
, the objective vector @xmath21 is taken to be parallel to the biomass flux , which is modeled as an effective reaction that converts metabolic species into biomass .
the first question of interest is to determine the conditions under which a reaction will be inactive for any solution of eq .
( [ eqn : mb ] ) .
let us define the stoichiometric coefficient vector of reaction @xmath22 to be the @xmath22th column of the stoichiometric matrix @xmath23 .
we similarly define the stoichiometric coefficient vector of an exchange flux .
if the stoichiometric vector of reaction @xmath22 can be written as a linear combination of the stoichiometric vector of reactions / exchange fluxes @xmath24 , we say that @xmath22 is a linear combination of @xmath24 .
we use this linear relationship to completely characterize the set of all reactions that are always inactive due to the stoichiometric constraints , regardless of any additionally imposed constraints , such as the availability of substrates in the medium , reaction irreversibility , cell maintenance requirements , and optimum growth condition .
[ thm1 ] reaction @xmath22 is inactive for all @xmath10 satisfying @xmath25 if and only if it is not a linear combination of the other reactions and exchange fluxes .
we denote the stoichiometric coefficient vectors of reactions and exchange fluxes by @xmath26 .
the theorem is equivalent to saying that there exists @xmath10 satisfying both @xmath25 and @xmath27 if and only if @xmath28 is a linear combination of @xmath29 , @xmath30 , @xmath31 . to prove the forward direction in this statement ,
suppose that @xmath27 in a state @xmath10 satisfying @xmath25 . by writing out the components of the equation @xmath25 and rearranging , we get @xmath32 since @xmath27 , we can divide this equation by @xmath19 to see that @xmath28 is a linear combination of @xmath29 , @xmath31 with coefficients @xmath33 . to prove the backward direction , suppose that @xmath34 .
if we choose @xmath10 so that @xmath35 for @xmath36 and @xmath37 , then for each @xmath38 , we have @xmath39 so @xmath10 satisfies @xmath25 .
theorem [ thm1 ] holds true independently of other constraints because the proof does not involve eq .
( [ eqn : const ] ) .
in particular , the sufficient condition for inactivity applies to any nutrient medium condition and does not depend on the reversibility of the reactions under consideration . in the case of the reconstructed _
e. coli _ ( human ) metabolic network considered in this study ( described in sec .
[ sec7 ] ) , which includes 922 ( 3328 ) unique internal and transport reactions , a total of 141 ( 475 ) reactions are always inactive in steady states as a result of the condition in theorem [ thm1 ] .
the next question of interest concerns the number of reactions that will be active with probability one in typical metabolic states . the stoichiometric constraints @xmath25 define the linear subspace @xmath40 ( the null space of @xmath23 ) , which contains the feasible solution space @xmath14 .
however , the set @xmath14 can possibly be smaller than @xmath41 because of the additional constraints arising from environmental and physiochemical properties ( availability of substrates in the medium , reaction irreversibility , and cell maintenance requirements ) . therefore , @xmath14 may have smaller dimension than @xmath41 .
if we denote the dimension of @xmath14 by @xmath42 , there exists a unique @xmath42-dimensional linear submanifold of @xmath43 that contains @xmath14 , which we denote by @xmath44 .
we can then use the lebesgue measure naturally defined on @xmath44 @xcite to make probabilistic statements , since we can define the probability of a subset @xmath45 as the lebesgue measure of @xmath46 normalized by the lebesgue measure of @xmath14 .
in particular , we say that @xmath27 for _ almost all _
@xmath47 if the set @xmath48 has lebesgue measure zero on @xmath44 .
an interpretation of this is that @xmath27 with probability one for an organism in a random state under given environmental conditions , which can be used to prove the following theorem .
[ thm2 ] if @xmath27 for some @xmath47 , then @xmath27 for almost all @xmath47 .
suppose that @xmath27 for some @xmath47 .
the set @xmath49 is a linear submanifold of @xmath44 , so we have @xmath50 . if @xmath51 , then we have @xmath52 , implying that we have @xmath53 for any @xmath47 , which violates the assumption .
thus , we must have @xmath54 , implying that @xmath55 has zero lebesgue measure on @xmath44 .
since @xmath56 , we have @xmath57 , and thus @xmath58 also has lebesgue measure zero .
therefore , we have @xmath27 for almost all @xmath47 .
theorem [ thm2 ] implies that we can group the reactions and exchange fluxes into two categories : 1 .
_ always inactive _ : @xmath53 for all @xmath47 , and 2 .
_ almost always active _
: @xmath27 for almost all @xmath47 .
consequently , the number @xmath59 of active reactions satisfies @xmath60 where @xmath61 is the number of inactive reactions due to the stoichiometric constraints ( characterized by theorem [ thm1 ] ) and @xmath62 is the number of additional reactions in the category 1 above , which are due to the environmental and irreversibility conditions . combining this result with the finding that optimal states have fewer active reactions ( next section )
, it follows that a typical state @xmath47 is non - optimal .
equation ( [ eq : env ] ) will lead to a different number of active reactions for different nutrient medium conditions [ determined by eq .
( [ eqn : const ] ) ] , with the general trend that this number will be larger in richer medium conditions . in the case of the _ e. coli _ ( human ) reconstructed network simulated in glucose minimal medium , as considered here ( see sec . [ sec7 ] ) , the number @xmath62 of inactive internal and transport reactions is 182 ( 1274 ) , of which 158 ( 563 ) are due to environmental limitations and 24 ( 711 ) are due to reaction irreversibility .
the latter includes the cascading - induced inactivation of some reactions due to the inactivation of different , irreversible reactions .
we now turn to the central part of our study , which concerns the number of reactions that can be active in steady states that optimize a linear function of the metabolic fluxes .
the linear programming problem for finding the flux distribution maximizing a linear objective function can be written in the matrix form : @xmath63 where @xmath64 and @xmath65 are defined as follows . if the @xmath22th constraint is @xmath66 , the @xmath22th row of @xmath64 consists of all zeros except for the @xmath38th entry that is @xmath67 , and @xmath68 . if the @xmath22th constraint is @xmath69 , the @xmath22th row of @xmath64 consists of all zeros except for the @xmath38th entry that is @xmath70 , and @xmath71 .
a constraint of the type @xmath72 is broken into two separate constraints and represented in @xmath64 and @xmath65 as above .
the inequality between vectors is interpreted as inequalities between the corresponding components , so if the rows of @xmath64 are denoted by @xmath73 , the inequality @xmath74 represents the set of @xmath75 constraints @xmath76 , @xmath77 . by defining the feasible solution space @xmath78 the problem
can be compactly expressed as maximizing @xmath15 in @xmath14 .
the duality principle @xcite expresses that any linear programming problem ( primal problem ) is associated with a complementary linear programming problem ( dual problem ) , and the solutions of the two problems are intimately related .
the dual problem associated with problem is @xmath79 where @xmath80 is the dual variable .
a consequence of the strong duality theorem @xcite is that the primal and dual solutions are related via a well - known optimality condition : @xmath10 is optimal for problem if and only if there exists @xmath80 such that @xmath81 note that each component of @xmath82 can be positive or zero , and we can use this information to find a set of reactions that are forced to be inactive under optimization , as follows . for any given optimal solution @xmath83 , eq .
is equivalent to @xmath84 , @xmath85 where @xmath86 is the @xmath22th component of @xmath82 .
thus , if @xmath87 for a given @xmath22 , we have @xmath88 , and we say that the constraint @xmath89 is _ binding _ at @xmath83 . in particular , if an irreversible reaction ( @xmath8 ) is associated with a positive dual variable ( @xmath90 ) , then the irreversibility constraint is binding , and the reaction is inactive ( @xmath53 ) at @xmath83 .
in fact , we can say much more : we prove the following theorem stating that such a reaction is actually _ required to be inactive for all possible optimal solutions _ for a given objective function @xmath15 .
[ thm : optimal ] suppose @xmath80 is a dual solution corresponding to an optimal solution of problem .
then , the set @xmath91 of all optimal solutions of problem can be written as @xmath92 and hence every reaction associated with a positive dual component is binding for all optimal solutions in @xmath91 .
let @xmath83 be the optimal solution associated with @xmath80 and let @xmath93 denote the right hand side of .
any @xmath94 is an optimal solution of problem , since straightforward verification shows that it satisfies ( [ opt1]-[opt3 ] ) with the same dual solution @xmath80 .
thus , we have @xmath95 .
conversely , suppose that @xmath10 is an optimal solution of problem .
then , @xmath10 can be shown to belong to @xmath96 , which we define to be the hyperplane that is orthogonal to @xmath21 and contains @xmath83 , i.e. , @xmath97 this , together with the fact that @xmath10 satisfies @xmath25 and @xmath74 , from , can be used to show that @xmath94 . therefore , any optimal solution must belong to @xmath93 . putting both directions together
, we have @xmath98 . as an example , consider the five - reaction network shown in fig .
[ fig : example](a ) where the flux @xmath99 is maximized .
the problem can be written in the form of problem with @xmath100 1 & 0 & 0 & \,\,\,\ , 0 & 0\\ -1 & 0 & 0 & 0 & 0\\ 0 & -1 & 0 & 0 & 0\\ 0 & 0 & -1 & 0 & 0\\ 0 & 0 & 0 & 0 & -1 \end{pmatrix*}\!,\,\,\ , { \mathbf{b}}= \begin{pmatrix*}[r ] 1\\ -1\\ 0\\ 0\\ 0 \end{pmatrix*}\!,\,\text{and } { \mathbf{s}}=\begin{pmatrix*}[r ] 1 & -1 & 0 & -1 & 0\\ 0 & 1 & -1 & 0 & -1 \end{pmatrix*}\!. \nonumber\ ] ] note that the equality constraint @xmath101 is split into two inequality constraints @xmath102 and @xmath103 for convenience and corresponds to the first two rows of @xmath64 and @xmath65 .
the optimal solution space @xmath91 consists of the single point @xmath104 and a possible choice of corresponding dual solution @xmath105 is given by @xmath106 for this dual solution , we have @xmath107 ( corresponding to the constraint @xmath108 ) and @xmath109 ( corresponding to the constraint @xmath110 ) , and eq .
in theorem [ thm : optimal ] becomes @xmath111 note that the constraint @xmath101 can be omitted since it is satisfied by any @xmath47 in this example .
once we solve problem numerically and obtain a _
single _ pair of primal and dual solutions ( @xmath83 and @xmath80 ) , we can use the characterization of @xmath91 given in eq . to identify all reactions that are required to be inactive ( or active ) for any optimal solutions . to do this
we solve the following auxiliary linear optimization problems for each @xmath112 : @xmath113 if the maximum and minimum of @xmath19 are both zero , then the corresponding reaction is required to be inactive for all @xmath114 . if the minimum is positive or maximum is negative , then the reaction is required to be active .
otherwise , the reaction may be active or inactive , depending on the choice of an optimal solution .
thus , we obtain the numbers @xmath115 and @xmath116 of internal and transport reactions that are required to be active and inactive , respectively , for all @xmath114 .
the number of active reactions for any @xmath114 is then bounded as @xmath117 the distribution of @xmath59 within the bounds is singular : the upper bound in eq .
is attained for almost all @xmath114 . to see this , we apply theorem [ thm2 ] with @xmath14 replaced by @xmath91 .
this is justified since we can obtain @xmath91 from @xmath14 by simply imposing additional equality constraints .
therefore , if we set aside the @xmath116 reactions that are required to be inactive ( including @xmath61 and @xmath62 reactions that are inactive for all @xmath47 ) , all the other reactions are active for almost all @xmath114 . consequently , @xmath118 we can also use theorem [ thm : optimal ] to further classify those inactive reactions caused by the optimization as due to two specific mechanisms : 1 . _ irreversibility . _
the irreversibility constraint ( @xmath8 ) on a reaction can be binding ( @xmath53 ) , which directly forces the reaction to be inactive for all optimal solutions .
such inactive reactions are identified by checking the positivity of dual components ( @xmath86 ) .
_ cascading . _
all other reactions that are required to be inactive for all @xmath114 are due to a cascade of inactivity triggered by the first mechanism , which propagates over the metabolic network via the stoichiometric constraints .
these inactive reactions occur in addition to the irreversibility / cascading - induced reaction inactivation identified in sec .
[ sec4 ] for typical ( in fact all ) steady states . in general , a given solution of problem can be associated with multiple dual solutions .
the set and the number of positive components in @xmath82 can depend on the choice of a dual solution , and therefore the categorization according to these specific mechanisms is generally not unique . for the example problem of fig .
[ fig : example ] , it is clear from @xmath119 that three reactions @xmath120 , @xmath121 , and @xmath122 are required to be inactive in the optimal state .
the dual solution given above categorizes @xmath120 under irreversibility , and @xmath121 and @xmath122 under cascading .
this reflects the fact that making @xmath123 under the stoichiometric constraint @xmath124 , along with the irreversibility constraints @xmath125 and @xmath126 , forces @xmath127 [ fig .
[ fig : example](c ) ] .
another possible dual solution is given by @xmath128 which leads to an alternative characterization of the same @xmath91 : @xmath129 categorizing @xmath121 and @xmath122 under irreversibility , and @xmath120 under cascading .
one can indeed see graphically in the projection onto @xmath130-coordinates in fig .
[ fig : example](b ) that the maximization of @xmath99 under the stoichiometric constraint @xmath131 ( which follows from @xmath101 and @xmath25 ) forces the irreversible fluxes @xmath121 and @xmath122 to be zero , which in turn forces @xmath132 [ fig .
[ fig : example](c ) ] .
clearly , one can also characterize the optimal solution space as @xmath133 which corresponds for example to the choice of dual solution given by @xmath134 this leads to the categorization of all three inactive reactions under irreversibility .
thus , we can interpret the non - uniqueness of the categorization as the fact that different sets of triggering inactive reactions can create the same cascading effect on the reaction activity .
the results above are important both because they can be applied to any linear objective function and because a significant fraction of real metabolic reactions are irreversible . in the case of the _ e. coli _ ( human ) reconstructed network ,
a total of 73.4% ( 65.6% ) of all internal and transport reactions are irreversible .
moreover , a number of other reactions are effectively irreversible because the irreversibility of different reactions in the same pathway constrains them not to run in one of the two directions ; this leads to a total of 94.3% ( 74.9% ) of the reactions whose fluxes are either necessarily nonnegative or necessarily nonpositive in all steady - state solutions . in the case of growth - maximizing states for the conditions considered in our numerical experiments , out of all 922 ( 3328 ) internal and transport reactions in the reconstructed network , a total of 146 ( 106 ) reactions are inactive due to irreversibility constraints , and a total of 114 ( 293 ) other reactions are inactive due to a cascade of reaction inactivation ; some of the irreversible reactions can be assigned to either the first or the second of these two groups .
the bounds provided by eq .
( [ naopt - s ] ) depend on the objective function . in the case of growth - maximizing states ,
the lower and upper bounds are 273 ( 113 ) and 339 ( 1180 ) , respectively .
these numbers should be compared with the number 599 ( 1579 ) of reactions that are active in typical , suboptimal states .
this clearly shows that optimal states are necessarily constrained to have a smaller number of active reactions , and that this is due to the presence of irreversible reactions in the network .
another problem of interest concerns the uniqueness of the optimal solutions .
while a number of necessary and/or sufficient conditions for this uniqueness are known @xcite , we are not aware of any probabilistic statements in the literature addressing this issue .
since the feasible solution space @xmath14 is convex , its `` corners '' can be mathematically formulated as _
extreme points _ , defined as points @xmath47 that can not be written as @xmath135 with @xmath136 , @xmath137 and @xmath138 such that @xmath139 .
intuition from the two - dimensional case ( fig .
[ fig : extreme ] ) suggests that for a typical choice of the objective vector @xmath21 such that problem has a solution , the solution is unique and located at an extreme point of @xmath14 .
we prove here that this is indeed true in general , as long as the objective function is bounded on @xmath14 , and hence an optimal solution exists .
[ thm : corner ] suppose that the set of objective vectors @xmath140 has positive lebesgue measure .
then , for almost all @xmath21 in @xmath141 , there is a unique solution of problem , and it is located at an extreme point of @xmath14 . for a given @xmath142 ,
the function @xmath15 is bounded on @xmath14 , so the solution set @xmath143 of problem consists of either a single point or multiple points .
suppose @xmath91 consists of a single point @xmath10 and it is not an extreme point . by definition , it can be written as @xmath144 with @xmath136 , @xmath137 and @xmath138 such that @xmath139 .
since @xmath10 is the only solution of problem , @xmath145 and @xmath146 must be suboptimal , and hence we have @xmath147 and @xmath148 .
then , @xmath149 and we have a contradiction with the fact that @xmath10 is optimal . therefore ,
if @xmath91 consists of a single point , it must be an extreme point of @xmath14 .
we are left to show that the set of @xmath142 for which @xmath150 consists of multiple points has lebesgue measure zero . by theorem [ thm : optimal ] , for a given @xmath21 , there exists a set of indices @xmath151 such that @xmath152 , so @xmath153 where the union is taken over all @xmath151 for which @xmath154 contains multiple points .
if @xmath21 is in one of the sets in the union in eq . , the set @xmath154 , being the set of all optimal solutions , is orthogonal to @xmath21 .
hence , @xmath21 is in @xmath155 , the orthogonal complement of @xmath154 defined as the set of all vectors orthogonal to @xmath154 .
therefore , @xmath156 because @xmath154 is convex , it contains multiple points if and only if its dimension is at least one , implying that each @xmath155 in the union in eq .
has dimension at most @xmath157 , and hence has zero lebesgue measure in @xmath43 .
since there are only a finite number of possible choices for @xmath151 , the right hand side of eq .
is a finite union of sets of lebesgue measure zero .
therefore , the left hand side also has lebesgue measure zero .
note that growth rate is not a typical objective function .
because this objective function has nonrandom coefficients and involves only a fraction of all metabolic fluxes , the objective vector * c * is generally perpendicular to a surface limiting the space of feasible solutions .
for this reason , the growth - maximizing states are generally not unique . in the case of the _ e. coli _ reconstructed network simulated in glucose minimal medium , our numerical calculations indicate that the growth - maximizing solutions form a space that is @xmath158-dimensional . in the case of the human reconstructed network ,
the corresponding dimension is 494 .
two questions follow from the results above . first , given the _ a priori _ surprising finding that metabolic activity as measured by the number of active reactions decreases in optimal states , what happens if we use other measures of metabolic activity such as total reaction flux in the network ?
second , given that these results were derived for linear objective functions , to what extent does the observed reduction in the number of active reactions manifest itself in nonlinear objective functions of biological significance ? these two questions are best examined using numerical experiments .
both are addressed below by considering the following objective functions : 1 .
_ total flux in the network _
: defined as @xmath159 , where @xmath22 runs through all internal and transport reactions ( excluding the biomass flux ) , it measures the overall metabolic activity while accounting for the differences in the fluxes of different reactions .
this objective function is nonlinear because of the absolute value used to properly measure the flux of the reversible reactions .
2 . _ total mass flow in the network _ : defined as @xmath160 , where @xmath161 is the mass of the reactants ( or products ) involved in reaction @xmath22 , it measures the overall metabolic activity weighted by the mass flow of each reaction @xcite .
the sum is over the same reaction set considered in the definition of @xmath162 . for other nonlinear objective functions of biological significance in cellular metabolism ,
we refer to ref .
@xcite .
figure [ fig : numerical ] shows the results of our numerical experiments for @xmath162 and @xmath163 on the _ e. coli _ reconstructed network . in both cases , the maximum ( minimum ) of the objective function for a given growth rate decreases ( increases ) as the growth rate increases , and it converges to essentially a single intermediate value for all states that maximize growth rate [ fig .
[ fig : numerical](a , b ) ] .
the average activity , which we determined by randomly sampling the solution space , is essentially constant ( it decreases very slowly as the growth rate increases ) .
the sampling of the solution space was performed using the hit - and - run method @xcite , which is an efficient algorithm to sample high - dimensional convex regions .
our implementation of this method is as described in our previous study @xcite and involves artificial centering @xcite .
the observed behavior of the total flux activity and total mass flow activity should be contrasted with metabolic activity as measured in terms of the number of active reactions , which is significantly smaller at growth - maximizing states .
the average number of active reactions in the @xmath162-maximizing states across different growth rates is just 302 out of a total of 571 that would be active in typical states ( the latter too is an average over different growth rates , and is smaller than the number 599 anticipated in sec . [ sec5 ] because of additional constraints set to the diverging cycles throughout this section see below ) .
a similar result holds true for @xmath163-maximizing states ( table [ tabl ] ) .
therefore , the maximizations of total flux and total mass flow in the network also lead to a reduced ( rather than increased ) number of active reactions compared to typical states [ such as those determined by the hit - and - run method in fig .
[ fig : numerical](a , b ) ] .
this number varies very little with growth rate and is essentially undistinguishable from the number of active reactions in growth - maximizing states ( see standard deviations in table [ tabl ] ) . while these results concern _
e. coli _ , we note that similar trends are observed for the human metabolic network .
if the irreversible reactions are made reversible [ fig .
[ fig : numerical](c , d ) ] , then the number of active reactions increases . for states that minimize @xmath162 and @xmath163
, the number of active reactions jumps to a large number when @xmath164 of the irreversible reactions is assigned to be just slightly negative , and then decreases as the irreversibility constraints are further relaxed ( table [ tabl ] ) . for states that maximize
@xmath162 and @xmath163 , the increase in the number of active reactions is by a factor of nearly @xmath165 ( table [ tabl ] ) .
this number is comparable to the number of active reactions in typical suboptimal states of the original network .
therefore , like in the case of linear objective functions , the reduced number of active reactions found in states that maximize the total flux or the total mass flow is due to the presence of irreversible reactions in the metabolic network .
.number of active reactions in states maximizing or minimizing the total flux , @xmath162 , and the total mass flow , @xmath163 .
the relaxation of the irreversibility constraints is implemented by allowing @xmath164 to be negative for all irreversible reactions , as indicated in the leftmost column .
each column shows the average and standard deviation calculated over the growth rates considered in fig .
[ fig : numerical ] . [ cols="<,^,^,^,^",options="header " , ] [ tabl ]
all simulations presented in this paper are based on a reconstructed metabolic network of _
e. coli _
k-12 , which represents a further curated version of the ijr904 model @xcite in which duplicated reactions have been removed , and on the most complete reconstructed human metabolic network , generated by applying the same curation to the _ homo sapiens _ recon 1 model @xcite .
the _ e. coli _ ( human ) network used consists of 922 ( 3328 ) reactions , 901 ( 1491 ) enzyme- and transport protein - coding genes , 618 ( 2766 ) metabolites , 143 ( 404 ) exchange fluxes , and the biomass flux . for the _ e. coli _
network , the simulated medium had limited amount of glucose ( 10 mmol / g dw - h ) and oxygen ( 20 mmol / g dw - h ) , and unlimited amount of sodium , potassium , carbon dioxide , iron ( ii ) , protons , water , ammonia , phosphate , and sulfate ; the flux through the atp maintenance reaction was set to 7.6 mmol / g dw - h . for the human network ,
we used a medium with limited amount of glucose ( 1 mmol / g dw - h ) and unlimited amount of oxygen , sodium , potassium , calcium , iron ( ii and iii ) , protons , water , ammonia , chlorine , phosphate , and sulfate ; for the biomass composition , we followed ref . @xcite . in our simulations , @xmath166 mmol / g dw - h was used as a flux threshold to define the set of reactions considered active .
a few cycles whose flux or mass flow would diverge in the optimization of the corresponding objective function were assigned the minimum feasible flux in the optimization of @xmath162 and the minimum feasible mass flow in the optimization of @xmath163 ( under the constraint of not altering the fluxes of the other reactions ) .
these minimum flux values were also adopted as bounds in our hit - and - run sampling .
all numerical calculations were implemented using the cobra toolbox @xcite and the cplex optimization software @xcite .
this study was supported by the national science foundation under grant dms-1057128 , the national cancer institute under grant 1u54ca143869 - 01 , and a sloan research fellowship to a.e.m .
fong , a.r .
joyce and b. .
palsson , _ parallel adaptive evolution cultures of _ escherichia coli _ lead to convergent growth phenotypes with different gene expression states _ , genome . res . *
15 * ( 2005 ) , 13651372 .
fong , a. nanchen , b. .
palsson and u. sauer , _ latent pathway activation and increased pathway capacity enable _ escherichia coli _ adaptation to loss of key metabolic enzymes _ , j. biol .
* 281 * ( 2006 ) , 80248033 .
t. shlomi , t. benyamini , e. gottlieb , r. sharan and e. ruppin , _ genome - scale metabolic modeling elucidates the role of proliferative adaptation in causing the warburg effect _ , plos comput . biol .
* 7 * ( 2011 ) , e1002018 . | the metabolic network of a living cell involves several hundreds or thousands of interconnected biochemical reactions .
previous research has shown that under realistic conditions only a fraction of these reactions is concurrently active in any given cell .
this is partially determined by nutrient availability , but is also strongly dependent on the metabolic function and network structure . here ,
we establish rigorous bounds showing that the fraction of active reactions is smaller ( rather than larger ) in metabolic networks evolved or engineered to optimize a specific metabolic task , and we show that this is largely determined by the presence of thermodynamically irreversible reactions in the network .
we also show that the inactivation of a certain number of reactions determined by irreversibility can generate a cascade of secondary reaction inactivations that propagates through the network .
the mathematical results are complemented with numerical simulations of the metabolic networks of the bacterium _ escherichia coli _ and of human cells , which show , counterintuitively , that even the maximization of the total reaction flux in the network leads to a reduced number of active reactions .
joo sang lee takashi nishikawa adilson e. motter |
for geometric frustrated magnetic materials it is impossible to choose one conguration with minimal energy by the individual minimization of the magnetic moments interactions .
consequently , their ground states are degenerated .
the most studied cases are systems with crystalline structures composed by triangles and tetrahedrons with nearest - neighbor antiferromagnetic interaction of their magnetic moments @xcite . recently frustration has been observed in jarosites and pyrochlore oxides as a result of ferromagnetic interaction @xcite . in these structures
the spins are arranged in kagom and pyrochlore lattices and forced to point in towards the center of the triangles and tetrahedrons respectively .
for this reason , there is only the two possible states , `` in '' or `` out '' , and therefore can be approximated through the classical ising model . since the angle between the spins the ferro - antiferromagnetic rule is changed so ferromagnetic systems can be mapped by the antiferromagnetic ising model and vice - versa @xcite .
these systems are called spin ice because of the similarity between their ground state conguration and the pauli model for the water ice . due to the simplicity of the ising model in comparison with heisenberg they became a model for the study of geometric frustration . in some cases it is even possible an analytic treatment in order to compare to experimental and computational results @xcite .
the recent artificial production of the first spin ices in the square and honeycomb lattices had attracted a lot of attention @xcite .
they are constructed with bi - dimensional sequences of nanoislands with ferromagnetic magnetic moments considered as effective ising spins .
recently it was possible to produce artificial jarosite and volborthite crystals with practically no distortions or impurities , something not possible for natural materials found in nature @xcite .
another recent example of the ising model is the study of magnetic materials in the triangular kagom lattice , artificially produced , as an alternative for adiabatic demagnetization @xcite .
they consist of a kagom lattice with an inserted site between two nearest - neighbors .
additionally it can be used in the study of the exchange coupling constants @xcite .
they are composed of sequences kagom lattices with weak interaction between the layers so they are considered bidimensional structures .
in this paper it is presented an alternative to the conventional representation of the kagom @xcite and pyrochlore @xcite lattices for monte carlo study of spins systems .
furthermore , the critical temperatures and exponents using the standard ferromagnetic ising model is shown . as suggested ,
there are the studies of the xy and heisenberg models for these lattices or the addition of the dipolar term in the ising model @xcite .
there is also the possibility of the study of critical quantities and phase diagram for diluted cases .
the kagom lattice is a two - dimensional arrangement of sites with four nearest - neighbors as shown in figure [ fig : kagome ] .
the unitary cell in gray has three sites with bonds forming equilateral triangles , each site in an indenpendent rhombic sublattices labeled a , b and c. for a site with label @xmath0 and coordinates @xmath1 and @xmath2 its nearest - neighbors are defined using the two versors * * and * * as shown in table [ tab : kag_neighbors ] ( @xmath3 , @xmath4 and @xmath5 for a , b and c respectivelly ) .
the sites are indexed by @xmath6 , the space between the sites are set to 1 and @xmath7 .
.nearest - neighbors for the kagom lattice . [ cols="^,^,^",options="header " , ]
a. l. passos is grateful for the partial support provided by fapitec / se . | this work aims to present a representation for the kagom and pyrochlore lattices for monte carlo simulation and some results of the critical properties .
these geometric frustrated lattices are composed of corner sharing triangles and tetrahedrons respectively .
the simulation was performed with the cluster wollf algorithm for the spins updates using the standard ferromagnetic ising model .
the determination of the critical temperature and exponents was done through the histogram technique and the finite - size scalling theory .
kagom , pyrochlore , ising model , monte carlo |
strong coupling in solid state microcavities ( mcs ) , has been exhaustively studied in inorganic quantum well devices @xcite . in these structures
wannier excitons and photon cavity modes are mixed in coherent excitations , called polaritons . a number of interesting phenomena , due to the mixed nature of such eigenstates , such as the parametric processes @xcite and non - equilibrium polariton condensation @xcite , were revealed . on the other hand , organic strongly coupled mcs
have been developed since 1998 @xcite , by using different kinds of optically active organic layers , among which the cyanine dye j - aggregate films are probably the most typical .
organic materials possess frenkel excitons , instead of wannier ones , with large binding energy and oscillator strength . because of this ,
organic mcs allow obtaining values of the rabi splitting up to @xmath0 mev @xcite and offer the possibility of easily observing polaritons at room temperature @xcite .
electroluminescence was demonstrated in a j - aggregate strongly coupled microcavity led @xcite at room temperature .
peculiar molecular phenomena , like the observation of strongly coupled vibronic replicas have also been demonstrated @xcite .
a detailed review about organic mcs can be found in refs.@xcite the nature of polaritons in disordered organic mcs has been addressed from the theoretical point of view @xcite , showing the coexistence of delocalized and partially localized polaritons .
the possible interaction with molecular vibrations was discussed @xcite , the effect of anisotropy in organic crystal was characterized @xcite , as well as the possible occurrence of non - linear phenomena @xcite for high density of polaritons .
we believe that , as in the case of inorganic mcs , theoretical efforts towards a qualitative , but also quantitative , understanding of the relaxation processes in organic mcs dynamics are an essential support to the work of experimental teams involved in the study of organic microcavities , in principle able to give rise to a rich variety of phenomena , similar to what obtained with their inorganic counterpart and more , thanks to the infinite possibilities of organic chemistry . a recent model described the dynamics of a j - aggregate mc due to the strong coupling of polariton with optical molecular vibrations @xcite thorough the application of a quantum kinetic theory .
our approach @xcite , even though restricted to a rate equation description of the relaxation processes , allows for a more flexible and complete description of the physical system at hand , accounting for the photoluminescence of a 2d mc starting from an accurate description of the bare active material ( the j - aggregate film ) .
besides we are also able to perform time dependent simulations of j - aggregate microcavity photoluminescence @xcite and estimate their radiative decay time in linear pumping regime . in particular , we here extend our model , which includes the presence of a large number of weakly coupled excitons , considering the occurrence of a radiative relaxation channel from such excitons to the cavity polaritons , as first suggested in ref.@xcite .
we also include the effects on the polariton relaxation due to the presence of a discrete spectrum of molecular vibrations , as in @xcite , which however are here supposed to scatter weakly with the electronic excitations without taking part directly in the cavity - polariton formation .
a j - aggregate can be thought as a chain of @xmath1 dye molecules linked by electrostatic interactions .
its excitations are described by the frenkel hamiltonian : @xmath2 where @xmath3 are the monomer exciton creation operators , corresponding to gaussian distributed energies @xmath4 , with standard deviation @xmath5 , to account for static disorder .
the hopping term is given by @xmath6 where @xmath7 is the nearest neighbor coupling strength .
the eigenstates are found by direct diagonalization of the frenkel hamiltonian and are described by the operators @xmath8 , where @xmath9 is the coefficient of the @xmath10-th exciton on the i - th monomer .
the oscillator strength of each state is given by @xmath11 , with @xmath12 , having the molecules equal transition dipole and being the wavelength much greater than the aggregate length .
the excitations are delocalized frenkel excitons with an energy dispersion curve , called the j band , where most part of the aggregate oscillator strength is concentrated on the superradiant bottom states @xcite .
higher energy excitons very weakly couple with light . while the hopping term determines the formation of the j band , static disorder gives rise to a low energy tail in the exciton dos , corresponding to partially localized , inhomogeneously broadened , superradiant eigenstates , responsible of the film luminescence and , in mcs , of the strong light - matter coupling .
delocalized polaritons are the result of the strong light - matter interaction between such superradiant excitons , localized on each aggregate , and the photon cavity modes , ideally extended over the whole structure . to treat the mc case we include a model polariton wavefunction of wavevector @xmath13 , linking the j aggregate excitons into the polariton states @xcite .
the total exciton and photon fraction , @xmath14 , and the polariton dispersion curve @xmath15 are given by the usual two coupled oscillator model .
the modulus of the exciton coefficients @xmath16 , describing how each exciton @xmath10 of @xmath17-th aggregate participates in the polariton states , are the only relevant quantities entering our model .
we consider the polaritons to be extended with equal degree on each one of the @xmath18 aggregates and make the assumption that the weight of each exciton in the polariton states is proportional to its oscillator strength , so that @xmath19 . in organic microcavities the strong coupling region , where polaritons do form ,
is typically extended till about @xmath20 @xmath21 , where the bragg mirrors stop - band starts to fail , or is also more restricted by inhomogeneous disorder @xcite .
therefore the excitation spectrum is composed , besides polaritons , of a number of uncoupled molecular excitons that form an excitonic reservoir ( er ) @xcite as shown in fig.[fig : schema ] . the excitonic part of the polaritons are formed by a bloch sum , with @xmath22 , made from the superradiant molecular excitons of the j - aggregates , while the er is composed by uncoupled superradiant excitons from the remaining part of the brillouin zone , plus a number of higher energy `` dark '' excitons . we note that the ratio between polaritons and uncoupled excitons of the er can be of the order of @xmath23 or less .
the dynamics of the system is described by rate equations for the population @xmath24 of the state @xmath25 , including the scattering rates , due to a weak linear exciton - phonon coupling , among all the kinds of particles present in the system which have a non - null exciton component . dealing with a 2d microcavity , we assume isotropic conditions , in which the polariton population only depends on the modulus of the wavevector , in order to reduce the dimensionality of the problem .
we consider localized vibrations , that can be due to the molecules itself or to the local environment , located on each chain site with a continuous spectrum , choice that follows literature reports @xcite on j - aggregate simulations .
we will address this model as the continuous spectum of vibration ( csv ) model .
we also extend the model to account for a series of discrete optical vibrations in what we will refer as the discrete vibrations model ( dv model ) , such specific modes can be of particular relevance when their energy is nearly resonant with the rabi splitting @xcite .
in particular the dv model includes a continuum of vibrations for energy below @xmath26 mev and two discrete molecular vibrations of @xmath27 mev and @xmath28 mev , with an exciton - phonon coupling factor @xmath29 and @xmath30 , respectively .
the values of the molecular vibrations are chosen following the raman experiments on j - aggregate films , as reported in @xcite .
the scattering rates from the state @xmath25 to the @xmath31 , which can either be an exciton @xmath32 , the @xmath10-th exciton of the @xmath17-th aggregate , or a delocalized polariton of wavevector @xmath13 , can be expressed in an unique compact notation . if we consider the interaction with a continuous spectrum of vibrations , for which the vibrational dos is non - zero for any energy difference @xmath33 between final and initial state , we adopt the following expression @xmath34 where @xmath35 and @xmath36 are parameters related to the exciton - photon interaction energy and to the shape of the vibration spectrum , @xmath37 is the bose - einstein occupation factor , @xmath38 the step function .
if we instead consider the interaction with a molecular vibration of discrete energy @xmath39 we obtain @xmath40 where @xmath41 is the dirac delta and @xmath42 the exciton - phonon coupling parameter for the vibration of energy @xmath39 .
what distinguish the different processes is the overlap factor @xmath43 .
in fact the emission or absorption of localized vibrations promotes the scattering among states that have an exciton participation on the same molecular site .
therefore the factor @xmath43 take in account of the partial exciton nature of polaritons , of their delocalized nature and are proportional to an excitonic overlap factor @xmath17 obtained by carrying out the sum of the excitonic participation of the initial and final state on each molecular site .
the overlap factor is one of the following : @xmath44 @xmath45 respectively , in the case of scattering from an exciton to another exciton ( @xmath46 , @xmath47 ) , where @xmath48 is the excitonic overlap factor ; or for the scattering from an exciton @xmath49 to a polariton @xmath13 , with @xmath50 their excitonic overlap factor , and @xmath51 the number of states corresponding , in the inverse space grid , to the polariton variable of modulus wavevector @xmath13 ; or when describing the opposite process ( @xmath52 , @xmath47 ) ; or in the latter case of scattering between two polaritons , where @xmath53 is an excitonic overlap factor between polaritons ( the derivation of the rates can be found in ref .
@xcite ) .
we also add a radiative channel of decay from the er to the polaritons , accounting for the presence in the mc of a fraction of excitons weakly coupled with light .
in fact , inside a strongly coupled cavity a fraction of weakly coupled excitons can decay radiatively pumping the photonic part of polaritons @xcite .
we model the spontaneous emission of polaritons from the er similar to the bare film photoluminescence times a switching factor @xmath54 : @xmath55 the contribution of the @xmath10 excitons is proportional to the bare exciton luminescence ratio @xmath56 , the transfer matches energy conservation , while the normalization is chosen to obtain a net exciton radiative decay towards polaritons of @xmath54 times the spontaneous emission ( @xmath57 ) .
@xmath58 is a lorentzian shape broadening function , where the energy uncertainty is given by the finite particle lifetime .
the damping rate of polaritons , due to photon escape through the cavity mirrors , is given by @xmath59 , where typically @xmath60 fs .
for the upper and lower polaritons we consider the reciprocal space up to @xmath61 @xmath62 , discretized in a grid of @xmath63 points .
the effect of the inhomogeneous disorder are taken in account averaging the scattering rates over an ensemble of @xmath64 j - aggregates each having @xmath65 monomers .
the dyes hopping energy is fixed at @xmath66 mev , and the dye spontaneous emission lifetime is @xmath67 ns@xmath68 ; while the parameters @xmath69 , @xmath70 and @xmath71 , have been fixed in order to obtain the best fit the bare j - aggregate film absorption and emission features @xcite .
solving the rate equations for the j - aggregate microcavity , with the same procedure described in ref.@xcite , we obtain the polariton steady state population , from which we calculate the angle resolved photoluminescence , by imposing the conservation of energy and in - plane momentum between cavity - polaritons and external photons . for both the csv and the dv models , we draw the indication that the major part of the population is frozen on the er . the origin of this bottleneck is due essentially to three reasons : the small number of polariton states with respect to the er ones , second the interaction between a localized vibration and a polariton is reduced by the delocalized , and mixed , nature of exciton - polaritons , in fact the scattering rates scale as @xmath72 , where @xmath73 is the number molecular sites over which the polariton is delocalized .
third is the small photon lifetime , due to the low quality factor of organic mcs .
the photoluminescence can be thought of as the result of the escape through the cavity mirrors of a particle , following the previous scattering on the polariton branches from the er , with the absorption or emission of a vibrational quantum . in fig.[fig :
schema ] we draw a sketch of the excited states on a j - aggregate microcavity and of the possible scattering processes following non resonant pumping . in particular we stress that the relaxation inside a j - aggregate is faster than er to polariton scattering and leads to a steady state population ( shown in the right figure in continuous line ) , from which polariton pumping occurs .
the er dos is inhomogeneously broadened by static disorder and er steady state population is similar to that of a non - cavity sample .
we note here that the average difference between the energy of the steady state population and the film absorption maximum results to be about @xmath74 mev .
an experimentally accessible parameter , used to analyze the effectiveness of relaxation in organic microcavities , is the photoluminescence intensity ratio @xmath75 recorded at resonance angle , where polariton branches are half matter and half light , and their energy distance from the exciton resonance is half the rabi splitting ( @xmath76 ) .
we focus on the temperature dependence of the @xmath77 ratio between @xmath78 and @xmath0 k for a mc with a rabi splitting of @xmath79 mev , shown in fig.[fig : temp ] , and compare it with the experimental result reported by d.g .
lidzey s group @xcite . with @xmath80
the @xmath77 ratio shows a smooth thermal activation , but it is difficult to get any up pl at low temperature .
this is consistent with the fact that up pl follow the absorption of a vibrational quantum from a particle at the bottom of the er and therefore is proportional to number of vibrations ( varying as a boltzmann factor ) .
turning on the radiative channel ( @xmath81 ) , being it independent on temperature , we obtain a low temperature saturation of the up pl and the @xmath77 ratio exhibits a plateau as experimentally found @xcite .
the experimental data are well reproduced and stay between the simulations with @xmath82 and @xmath83 .
we also show in fig.[fig : rabi ] the simulation of the @xmath77 ratio as a function of the rabi splitting in comparison with experimental data from ref.@xcite . increasing the rabi splitting value
reduces the probability for a particle in the er to be scattered on the up branch , because such process would need the absorption of a vibration of greater energy from the thermal bath .
the data therefore roughly follows a boltzmann factor @xmath84 , shown for comparison , where @xmath76 corresponds , at resonance angle , to the energy separation from the er bottom and the up and lp states . with a continuous vibrational spectrum we observe the smooth boltzmann behavior , while the inclusion of the radiative channel rigidly shifts the @xmath77 ratio toward higher values .
the experimental data ( blue x symbols ) stay in between the simulation performed without the radiative channel @xmath85 and the simulation with the radiative process turned on with @xmath86 .
the temperature dependence of the @xmath77 ratio , shown in fig.[fig : temp ] is largely unaffected by the introduction of the discrete spectrum of vibration and the same considerations as for the csv case still apply . if we instead consider the rabi splitting dependence , the inclusion of discrete energy vibrations leads to the fluctuation of the @xmath77 ratio and to a deviation from the smooth boltzmann - like behavior .
in particular the dv model reproduces the drop in the @xmath77 factor between @xmath87 to @xmath88 mev , associated with @xmath89 mev and the activation of the up depletion towards the er bottom with emission of a vibrational quantum and permits to obtain a better fit of experimental data .
we believe that the agreement we obtain with the available data as show in fig.[fig : rabi ] is of comparable quality to that obtained by chovan et al.@xcite . of course , this is not to say that the phonon strong coupling approach they consider is immaterial .
it is plausible that also in j - aggregate microcavities exciton and molecular vibrations may interact strongly leading to non perturbative effects @xcite such as , in particular , the formation of cavity polaritons with the concomitant participation of optical phonon , exciton and cavity photon modes , a phenomenon which has also been considered in bulk semiconductors ( see the concept of `` phonoriton '' @xcite ) .
it remains to be seen , however , whether there is as yet a compelling experimental evidence for the occurrence of such a phonon strong coupling in j - aggregate microcavities , the photoluminescence of which is here considered .
apart for the particular choice of the definite energy vibrations we made in the dv model ( @xmath90 mev , @xmath27 mev , @xmath28 mev ) , we also tested different sets of optical vibrations in order to understand their effects on the @xmath77 ratio as a function of the rabi splitting .
in general it is not possible to completely distinguish the individual effects of each one of @xmath91 , @xmath92 , @xmath93 , however we are able to identify some trends .
firstly we analyze the role of optical vibration @xmath93 by increasing it from @xmath94 mev to @xmath95 mev . as shown in fig.[fig : rabic ] a minimum is observed at different @xmath96 for different values of @xmath93 .
the motivation is that the er to lp scattering rate is in resonance when @xmath97 , with @xmath98 averaged on the steady state population of the exciton reservoir .
when the process is in full resonance we expect a minimum of the @xmath77 factor , as it is evidenced in fig.[fig : rabic ] . following this rule we expect the minimum to be located in @xmath99 .
because of the different position of @xmath100 , when er to lp scattering is maximum , for small rabi splitting the @xmath77 factor decrease with @xmath93 , while we find the opposite behavior for @xmath96 greater than @xmath88 mev . as we will see , the occurrence of the @xmath77 minimum depends on the presence of higher energy vibrations active in relaxation processes , here not included .
we now focus on the role of the low energy vibrations .
initially we modify the dv model including the single vibration @xmath101 mev instead of the continuous spectrum of low energy vibrations .
the result , in fig.[fig : rabia ] , is a significant reduction of the @xmath77 factor for @xmath96 smaller than @xmath102 mev
. we can understand this if we consider that up population is mainly due to the er to up scattering , that takes place from steady state er population with absorption of a molecular vibration , while the direct relaxation in up from high energy excitons gives a minor contribution
. this process involving absorption of a vibrations is active at room temperature till around @xmath103 ( if the vibrational dos is not zero ) , therefore it is significant for up energies equal to the mean er population energy ( @xmath104 ) plus @xmath105 , that corresponds to rabi splitting values @xmath106 mev .
therefore a reduction in the active low energy dos means a significant reduction of the er to up scattering and of the @xmath77 ratio for small @xmath96 .
if we modify dv choosing @xmath107 mev we note that after @xmath108 mev we have a rapid decrease of the @xmath77 factor and a minimum , as predicted around @xmath102 mev .
so it is clear that to recover the experimental data we need a series of active vibrations between @xmath109 and @xmath26 mev .
finally let s consider the effects of a third optical vibration @xmath110 , in particular we set @xmath111 mev or @xmath112 mev with @xmath113 . as can be seen in fig.[fig : rabid ] the presence of @xmath114 has effects on @xmath77 only for large enough rabi splitting , for which @xmath115 approach @xmath114 .
for @xmath111 mev , with @xmath96 in the range considered , the non - monotonic behavior of @xmath77 vanishes correspondent to the fact that the lp is pumped from the er almost thermalized population with an emission of a vibration @xmath114 .
if we increase @xmath114 to @xmath116 mev the er to lp scattering become too much out of resonance to contribute and numerical data are nearer to the dv case .
it must be noted that the possible presence of a @xmath77 minimum is conditioned to the absence of optical vibrations of higher enery @xmath114 active in relaxation processes .
in conclusion our model , based on semiclassical rate equations , explains sufficiently well the presently available experimental data on strongly coupled j - aggregate microcavities .
in particular , the photoluminescence under non - resonant pumping follows the scattering of an electronic excitation from the er bottom , where the population is accumulated , towards the polariton branches .
such scattering can either happen with the absorption / emission of vibrational quanta ( with a continuous as well as discrete spectrum ) , process which is sensitive to temperature , or be due to pumping of polaritons by the spontaneous radiative decay of weakly coupled excitons .
the interplay between the two processes determines both the specific temperature dependence of the polariton branches photoluminescence intensity ratio and its rabi splitting dependence .
j. kasprzak , m. richard , s. kundermann , a. baas , p. jeambrun , j.m.j .
keeling , f.m .
marchetti , m.h .
szymanska , r. andre , j.l .
staehli , v. savona , p.b .
littlewood , b. deveaud , le si dang , _ nature _ * 443 * , 409 ( 2006 ) . | we have developed a model accounting for the photo - excitation dynamics and the photoluminescence of strongly coupled j - aggregate microcavities . our model is based on a description of the j - aggregate film as a disordered frenkel exciton system in which relaxation occurs due to the presence of a thermal bath of molecular vibrations . in a strongly coupled microcavity exciton - polaritons
are formed , mixing superradiant excitons and cavity photons .
the calculation of the microcavity steady - state photoluminescence , following a cw non resonant pumping , is carried out . the experimental photoluminescence intensity ratio between upper and lower polariton branches is accurately reproduced . in particular both thermal activation of the photoluminescence intensity ratio and its rabi splitting dependence
are a consequence of the bottleneck in the relaxation , occurring at the bottom of the excitonic reservoir .
the effects due to radiative channels of decay of excitons and to the presence of a paritticular set of discrete optical molecular vibrations active in relaxation processes are investigared . |
in this article , we provide a detailed theory for the temperature - dependent electronic compressibility ( @xmath1 ) of an interacting two dimensional ( 2d ) electron ( or hole ) system within the hartree - fock theory .
we consider three distinct 2d systems of active current interest in condensed matter physics : 2d semiconductor systems ( e.g. , quantum wells , heterostructures , inversion layers , etc . ) ; monolayer graphene ; bilayer graphene .
for the purpose of comparison ( and also for the sake of completeness ) , we also provide as an appendix the corresponding finite temperature hartree - fock compressibility for a standard three dimensional electron gas ( 3deg ) since this result does not appear to be available in the theoretical literature in spite of the sixty - year long history of studying many - body interaction effects in 3deg . ever since the pioneering measurement of the 2d compressibility by eisenstein _
et al . _ , @xcite it has been extensively studied in 2d systems because the compressibility provides fundamental insight into quantum ground state properties which are not readily obtained from transport measurements .
some experimental studies @xcite of the compressibility have shed light on understanding the 2d metal - insulator transition ( mit)@xcite . the inverse compressibility ( @xmath2 ) is positive when the kinetic energy dominates over interactions at high densities . as the 2d density is reduced @xmath2 changes sign and becomes negative due to the increase of exchange energy associated with electron - electron interaction@xcite .
the negative @xmath2 reaches a minimum value at a certain low density @xmath3 , and then increases dramatically with further decreasing @xmath3 as the 2d mit sets in . the minimum point in @xmath2
has occasionally been loosely identified as the critical density for the 2d mit and is closely related to the transition from a homogeneous system to an inhomogeneous nonlinear screening regime , @xcite where the system is dominated by the formation of electron ( or hole ) puddles . whether this observed low - density behavior of @xmath1 is a cause for or an effect of 2d
mit is unclear .
the low - density compressibility of 2d electronic ( in this paper we use the term electron " to imply either electron or hole as should be obvious from the context ) systems has been of interest for almost 20 years now with the early experiments@xcite studying semiconductor - based 2d systems extensively and very recent experiments studying monolayer and bilayer graphene .
the reason for the focus on the density dependence of compressibility is that , in general , quantum interaction effects increase monotonically with decreasing density in a coulomb system ( with monolayer graphene being an odd exception where the interaction parameter , the so - called graphene fine structure constant , is independent of carrier density@xcite ) , as discussed above , leading eventually to the electronic compressibility becoming negative at low , but easily accessible , densities due to exchange effects .
it is curious , however , that in spite of this intense interest in the strongly interacting low - density 2d compressibility , there has been little research on the theoretical functional dependence of 2d compressibility on temperature .
this is strange because the scale for the temperature dependence of any electronic property is the fermi temperature @xmath4 , which invariably decreases with decreasing density .
thus , temperature effects on 2d compressibility become progressively more important _ even at a fixed temperature _ as the electron density goes down , since the important dimensionless temperature @xmath5 increases with decreasing density even if @xmath6 is kept fixed .
for example ,
2d gaas holes have @xmath7 k for a hole density of @xmath8 @xmath9 , which means that @xmath10 k is effectively a high - temperature regime for 2d hole densities @xmath11 @xmath9 !
our work in this paper takes a first step in correcting this omission in the literature .
our results are important for interpreting low - density 2d compressibility ( even at relatively low temperatures ) existing in the literature .
the thermodynamic isothermal compressibility ( or simply compressibility ) of a system is defined as the change of pressure with volume , @xmath12 , where @xmath13 is the particle number , @xmath6 is the temperature , @xmath14 is the system volume , @xmath15 is the pressure , and @xmath16 is the total energy of the system .
thus , @xmath17 . in the non - degenerate or classical systems the velocity of thermodynamic sound , @xmath18 , often provides a convenient experimental measure of compressibility , @xmath19 , where @xmath20 and @xmath21 are the mass and the density of the system , respectively . in the quantum limit
the compressibility can be obtained theoretically by using the theorem of seitz @xcite , ( i.e. , by connecting @xmath1 with the change in chemical potential with density , @xmath22 , where @xmath23 is the chemical potential and @xmath3 is the carrier density of the system ) .
another method of evaluating @xmath1 is through the compressibility sum rule .
this is an exact relationship between the compressibility and the long wavelength limit of the static dielectric function @xcite . in 2d quantum systems
the compressibility is often measured from the quantum capacitance @xcite which is proportional to @xmath24 .
a scanning single electron transistor has also been used to directly measure the density dependent chemical potential @xcite , and hence @xmath25 or @xmath26 .
we will mostly discuss in this work the behavior of @xmath27 , where @xmath28 is the non - interacting compressibility , and @xmath29 .
we note that sometimes @xmath26 is referred to as the thermodynamic density of states since at @xmath30 and for noninteracting system @xmath31 , the noninteracting density of states . in this paper
, we refer to @xmath1 and @xmath2 as compressibility and incompressibility ( i.e. inverse compressibility ) respectively , and also discuss results for @xmath32 , which is sometimes directly measured experimentally , remembering that this quantity is directly proportional to the incompressibility .
a part of our theoretical motivation for exploring the temperature dependence of compressibility in 2d semiconductor systems arises from recent experiments conducted on gaas - based 2d systems @xcite .
in addition , there has been substantial experimental interest in the temperature dependence of graphene compressibility.@xcite we find that the calculated temperature dependent compressibility including exchange energy is non - monotonic in temperature in both graphene and 2d semiconductor systems .
however , their temperature behaviors appear to be reversed , i.e. , the inverse compressibility in graphene @xmath33 increases first with increasing temperature and then decreases at high temperatures while the inverse compressibility in 2d semiconductor systems decreases with temperature first and then increases at higher temperatures .
we find that the leading order temperature correction to 2d @xmath2 within hartree - fock approximation ( hfa ) is @xmath34 at low temperatures for both systems ( @xmath5 ) .
we also find that the blg compressibility has very weak temperature dependence because the non - interacting kinetic energy is independent of temperature . to calculate the compressibility we use seitz s theorem which is given by @xcite @xmath35 where @xmath36 is the finite temperature chemical potential and @xmath3 the free carrier density .
in general , the direct measurement of @xmath22 provides information on the thermodynamic many body renormalization ( for example , fermi velocity ) arising from electron - electron interaction effects@xcite .
our goal here is to theoretically calculate the renormalized @xmath22 in graphene and 2deg including exchange interaction effects , or equivalently in the hfa , which should be an excellent quantitative approximation for compressibility in 2d systems .
our calculated carrier density and temperature dependence of compressibility ( or @xmath37 ) can be directly compared to experimental measurements in 2d systems including graphene .
the rest of the paper is organized as follows . in sec .
[ sec : semi ] , we introduce the formalism that will be used in calculating compressibility in 2d semiconductor systems , which includes the model describing the coulomb system , the procedure of calculation , and the numerical and analytical results of compressibility in 2d semiconductor systems . in secs .
[ sec : graphene ] and [ sec : blg ] we present the theoretical formalism and our analytical and numerical results of compressibility in mlg and blg , respectively .
[ sec : discussion ] contains discussions and conclusions .
in this section , we present the compressibility of 2d semiconductor systems ( 2ds ) such as si inversion layers in mosfets , 2d gaas heterostructures or quantum wells . in some of our numerical calculations we use the parameters corresponding to the n - gaas or p - gaas system , since these are the most - studied 2d system in the literature .
when ever possible , we provide our results in dimensionless units for universal applicability . in the absence of interaction , the non - interacting finite temperature chemical potential , @xmath38 , is calculated through the conservation of the total electron density , i.e. , @xmath39 where @xmath40 , @xmath41 is the density of states with @xmath42 being a degeneracy factor including both spin and possible valley degeneracy , and @xmath43^{-1}$ ] is the fermi distribution function .
we take @xmath44 throughout . by solving eq .
( [ eq : nt ] ) self - consistently we have the 2d chemical potential at finite temperatures @xmath45 where @xmath46 is the fermi energy at @xmath30 and @xmath47 with @xmath4 . to include interaction effects in the chemical potential , we calculate the exchange self - energy contribution within hfa .
we then have the interacting chemical potential as , @xmath48 where @xmath49 is the hf exchange self - energy calculated at fermi momentum @xmath50 .
the hf exchange energy with the electron - electron coulomb interaction is given by@xcite @xmath51 with @xmath52 is the 2d bare coulomb interaction in the momentum space ( @xmath53 is the background lattice dielectric constant ) , @xmath54 is the single particle energy ( @xmath20 is the effective mass of the particle ) . in the following ,
we calculate the compressibility of 2d semiconductor systems for both zero temperature and finite temperatures by finding the total chemical potential including exchange effects as a function of density and temperature . at zero temperature ( @xmath30 ) , the non - interacting part of the chemical potential is just the fermi energy @xmath55 . then @xmath56 becomes the inverse of the noninteracting single - particle density of states at the fermi level : @xmath57 , which is a density - independent constant for 2d . from eq .
( [ eq : def ] ) we have the zero temperature compressibility of non - interacting 2d systems , @xmath28 , @xmath58 the noninteracting compressibility , @xmath28 , is a positive quantity and is inversely proportional to the square of particle density , @xmath59 . for interacting 2d systems
the exchange self - energy can be calculated at @xmath60 from eq .
( [ eq : semiex ] ) , @xmath61 where @xmath62 is the the complete elliptic integral of the second kind .
the total chemical potential within hfa can be calculated by putting @xmath63 , i.e. , @xmath64 where @xmath65 ( @xmath66 , @xmath67 are the spin and valley degeneracy , respectively ) and the dimensionless coupling constant @xmath68 is the ratio of the average coulomb potential energy @xmath69 to the average @xmath30 kinetic energy .
the @xmath70 parameter as defined above is the dimensionless interaction parameter for a 2d electron liquid .
we show most ( except for sec .
[ sec : semi]d ) of our results in the dimensionless units of @xmath5 and @xmath70 , which would apply to any 2d semiconductor system .
differentiating eq .
( [ eq : chemi ] ) with respect to @xmath3 and making use of the relation @xmath71 , we get the zero temperature inverse compressibility within hfa @xmath72 the interacting inverse compressibility @xmath27 monotonically decreases as @xmath70 increases ( or density decreases ) and changes its sign from positive to negative at a coupling strength @xmath73 .
this behavior has been observed in experiments@xcite and much discussed in the literature .
generally , negative compressibility leads to a thermodynamic instability of a system .
however , the compressibility we have discussed in this paper applies only for the electronic part , i.e. , the quantity is not actually the compressibility of the whole system including the positive background charge which is necessary for neutrality .
the interaction between the electrons and its associated positive neutralizing background has been ignored .
in fact , a system with the negative electronic compressibility can be stabilized by the positive background , which gives rise to a positive compressibility for the whole system .
the negative electronic compressibility of the 2deg has been directly measured in experiments@xcite which probe only the electronic part of the compressibility , obviously the total compressibility of such a system would still be positive in order to maintain thermodynamic stability of the whole system .
results given in eqs .
( [ eq : k0])([eq : k0k ] ) are of course well - known@xcite , and are given here only for the sake of completeness and for the sake of comparison with our finite-@xmath6 results discussed below . at finite temperatures
the chemical potential with exchange self - energy can be calculated from eq .
( [ eq : mu2d ] ) and the normalized chemical potential , @xmath74 is expressed as @xmath75 \label{eq : chemire}\end{aligned}\ ] ] where @xmath76 is the non - interacting chemical potential given in eq .
( [ eq : noninter ] ) , @xmath77 , and @xmath78 is the complete elliptic integral of the first kind . in the absence of exchange interaction , we obtain the non - interacting inverse compressibility at finite temperatures @xmath79 where @xmath80 and @xmath81 is the finite temperature non - interacting compressibility . for @xmath82 , @xmath83 .
thus we have @xmath84 .
the leading order correction to @xmath85 at low temperatures is exponentially suppressed . for @xmath86 , @xmath87 .
thus the noninteracting inverse compressibility increases linearly in the high temperature limit , @xmath88 . including exchange effects
, the finite temperature inverse compressibility @xmath89 can be obtained from eq .
( [ eq : chemire ] ) @xmath90 \nonumber \\ & + & \frac{\sqrt{g}r_s}{2\pi t}\int _ 0 ^ 1\text{d}x k_1(x)\big[\frac{x ( x^2-h(t))}{\cosh ^2\frac{(x^2-\tilde{\mu}_0)}{2 t } } + \frac{\frac{1}{x^2}(\frac{1}{x^2}-h(t))}{\cosh ^2\frac{(1/x^2-\tilde{\mu}_0)}{2t}}\big ] .
\label{eq : comre}\end{aligned}\ ] ] the full numerical results of the compressibility are given in sec .
[ results_t ] .
we first consider the asymptotic behavior of the temperature dependence both at high temperatures ( @xmath91 ) and at low temperatures ( @xmath92 ) . in the low temperature limit ( @xmath93 ) ,
the integrand in the temperature - dependent term is only appreciable near the divergent regime of the first elliptic function @xmath78 ( @xmath94 ) .
we could get the asymptotic form by expanding the elliptic integrals about that point , i.e. , @xmath95 .
\label{eq : k1}\ ] ] replacing @xmath78 with eq .
( [ eq : k1 ] ) in the integrand of eq .
( [ eq : chemire ] ) , the chemical potential in the low temperature limit @xmath96 is given by @xmath97 \label{eq:2dnormchem}\ ] ] where @xmath98 , @xmath99 \approx 0.449 $ ] and @xmath100 is the derivative of zeta function . differentiating the asymptotic formula of the many - body chemical potential , we obtain the inverse compressibility for @xmath101 , @xmath102 .
\end{array } \label{eq : lhfsemi}\end{aligned}\ ] ] since @xmath103 for @xmath93 , the leading order correction to the inverse compressibility @xmath104 comes from the exchange energy term , i.e. , @xmath34 , which gives rise to the decrease of @xmath27 at low @xmath6 with increasing temperature as shown in fig .
[ fig : semi](a ) . the asymptotic formula given in eq .
( [ eq : lhfsemi ] ) agrees well with our numerical calculation at low temperatures as shown in fig .
[ fig : semi](a ) .
the low - temperature behavior of @xmath104 is dominated by the @xmath34 term which produces the shallow minimum in @xmath27 as a function of @xmath105 with the size of the minimum increasing with increasing ( decreasing ) @xmath70 ( density ) .
this interesting low - temperature non - monotonicity in the temperature - dependent 2d compressibility is entirely an exchange effect .
we emphasize , however , that this exchange - induced minimum is very shallow in @xmath89 , and never exceeds @xmath106 of its @xmath30 value ( often it is much less ) .
the asymptotic behavior of high temperature compressibility is obtained by approximating the fermi - dirac distribution by the classical boltzmann distribution . for @xmath107 , the normalized chemical potential with exchange energy
is given by @xmath108 the corresponding asymptotic formula for the high temperature inverse compressibility is given by @xmath109 for @xmath110 , the exchange energy contribution to the inverse compressibility decreases as @xmath111 while the kinetic energy contribution increases lineally since @xmath87 .
thus , the kinetic term dominates in the high temperature limit and the combined compressibility approaches the non - interacting result as it should at very high temperatures . at high temperatures the role of the exchange - correlation effects is diminished due to the increase of the thermal kinetic energy . as a consequence , the negative compressibility at low densities becomes positive at a high enough temperature , @xmath112 .
this behavior ( reversing sign due to increasing temperature ) was observed in high mobility p - gaas systems@xcite .
we have explicitly verified that our numerical hfa results agree precisely with the asymptotic high - temperature result of eq .
[ eq : semiht ] for @xmath113 .
we note that the exchange correction to @xmath27 falls off very slowly only as @xmath114 for @xmath113 , and as such quantum effects are quite large at large @xmath70 even for @xmath86 .
in fact , the classical regime in compressibility is approached only for @xmath115 . in 2deg as a function of rescaled temperature @xmath116 for different values of @xmath70 .
the inset presents @xmath27 in the low temperature regime .
the dashed and solid lines are corresponding to the analytical results ( given in eq .
[ eq : lhfsemi ] ) and the numerical results , respectively .
( b ) the inverse compressibility @xmath27 in 2deg as a function of @xmath70 for different values of rescaled temperature @xmath116 .
( c ) calculated @xmath117 , where the minimum of @xmath27 occurs , as a function of @xmath70 . ] in fig .
[ fig : semi](a ) , we show the calculated inverse compressibility @xmath27 as a function of rescaled temperature @xmath116 for six different values of @xmath70 .
the red solid line corresponds to the non - interacting case ( @xmath81 ) without the exchange interaction . as shown in sec .
[ results_a ] @xmath118 for @xmath107 increases linearly in the high temperature classical limit . for finite @xmath70 the zero temperature inverse compressibility decreases with @xmath70 and eventually becomes negative ( if @xmath119 ) due to the exchange energy . as shown in fig .
[ fig : semi](a ) @xmath27 is not a monotonic function of rescaled temperature @xmath116 .
as @xmath116 increases from zero @xmath120 initially decreases since the kinetic energy term is exponentially suppressed due to the fermi surface restriction , but the exchange term keeps decreasing as @xmath34 . after reaching the minimum value at an @xmath70-dependent characteristic temperature @xmath117 , @xmath2 increases linearly at high temperatures . this non - monotonic behavior becomes stronger at higher @xmath70 .
more interestingly the negative value of @xmath2 at low temperature and high @xmath70 reverses its sign as the temperature increases . at high enough temperatures , where the kinetic energy dominates over the interaction energy
, the system always has a positive compressibility , but typically strongly suppressed in magnitude from @xmath121 except for @xmath107 .
thus , many - body effects manifest strongly in the compressibility even for @xmath122 in most situations . as a function of @xmath116 for @xmath123 comparing the hf result with the non - interacting result @xmath124 as in eq .
[ eq : semiclass ] and the pure classical result @xmath125 .
inset shows the hf result and the asymptotic form for the high temperature @xmath27 as in eq .
[ eq : semiht ] with @xmath126 .
( b ) and ( c ) show @xmath127 for @xmath128 , @xmath129 , @xmath130 , @xmath131 for @xmath132 . ] in fig .
[ fig : semi](b ) , we show our numerically calculated inverse compressibility @xmath27 as a function of @xmath70 for different values of the rescaled temperature @xmath116 .
it is clear that the interacting @xmath27 manifests monotonically stronger many - body effects with increasing @xmath70 .
the curves for @xmath133 and @xmath134 cross each other at finite value of @xmath70 , which corresponds to the non - monotonic temperature dependence of inverse compressibility @xmath27 in the low temperature regime . as an inset in fig .
[ fig : semi](a ) we explicitly show a quantitative comparison between our derived low-@xmath6 asymptotic formula ( eq . [ eq : lhfsemi ] ) and the exact numerical hfa results verifying the presence of the @xmath34 term , which leads to the low-@xmath116 minimum , in the hf @xmath27 . in fig . [
fig : semi](c ) , we show as a function of @xmath70 our numerically calculated value of @xmath135 where the low-@xmath116 minimum of @xmath27 occurs .
we note that although @xmath117 increases monotonically with increasing ( decreasing ) @xmath70 ( density ) , this increase is sublinear implying that the actual temperature ( in kelvin ) @xmath136 , where the minimum occurs , decreases ( increases ) with decreasing ( increasing ) carrier density @xmath3 since @xmath137 and @xmath138 .
this means that the experimental observation of this non - monotonicity of @xmath27 as a function of temperature may be extremely difficult , if not impossible , with the experimental @xmath27 manifesting only a monotonic increase with increasing temperature at all densities .
the experimental observation is further hampered by two additional complications : ( 1 ) the actual decrease in the magnitude of @xmath27 associated with the shallow minimum is rather small ( @xmath139 ) ; ( 2 ) the low - density , large-@xmath70 regime of the 2d system often develops strong disorder - driven density inhomogeneity . in fig .
[ fig : semi2 ] we compare our numerical results with our high - temperature analytic theory as given in eq .
[ eq : semiht ] .
the purely classical compressibility @xmath140 is given by @xmath141 , which is the @xmath142 limit of @xmath143 , as given in eq .
[ eq : semiclass ] .
we note that the exchange correction to the non - interacting result , as given in eq .
[ eq : semiht ] falls off very slowly as @xmath144 . in fig .
[ fig : semi2](a ) we show calculated @xmath27 in the @xmath145 range for @xmath123 comparing the hf result with the non - interacting result @xmath124 as in eq .
[ eq : semiclass ] , and the pure classical result @xmath125 . for the sake of comparison ,
we also show ( for @xmath146 ) the corresponding high - temperature hfa result as in eq .
[ eq : semiht ] .
the interesting point to note here is that , as can be seen from eq .
[ eq : semiht ] , the quantum exchange correction is quantitatively substantial even at a temperature @xmath86 for large @xmath70 with the hfa results being quantitatively well below ( by @xmath147 ) the classical result even for @xmath148 .
although this appears somewhat counter - intuitive , the importance of quantum interaction persisting to high temperatures ( @xmath86 ) for large @xmath70 can be understood by considering the relative magnitudes of the three dimensionless energy parameters @xmath149 , @xmath5 , and @xmath150 that control the physics of an interacting quantum system we note that @xmath70 , @xmath116 , @xmath151 are not independent parameters since @xmath152 .
the classical noninteracting limit requires both @xmath116 , @xmath153 which necessitates @xmath154 as well .
thus one condition for the classical limit is @xmath154 when @xmath155 ( and @xmath156 for @xmath157 ) , which is much stronger than @xmath113 when @xmath70 is large ! in fig .
[ fig : semi2](b ) and ( c ) we show @xmath127 for @xmath128 , @xmath129 , @xmath130 , @xmath131 for @xmath132 to emphasize that the exchange correction to the inverse compressibility is substantial in magnitude even for @xmath86 .
since there has been considerable experimental activity@xcite in measuring the 2d compressibility of both electrons@xcite and holes@xcite in gaas - based two - dimensional semiconductor systems , we provide in this section a set of numerical results for the hf compressibility of 2d gaas systems ( electrons and holes ) as a function of carrier density ( @xmath3 in the unit of @xmath9 ) and temperature ( @xmath6 in k ) .
the parameters used in these numerical calculations are for electrons ( holes ) : @xmath158 ; @xmath159 ( i.e. , @xmath160 , @xmath161 ) ; @xmath162 , where @xmath163 is the free electron mass in vacuum and @xmath53 is the background lattice dielectric constant of gaas - algaas heterostructure .
there is no valley degeneracy ( @xmath164 ) in gaas , and we consider the spin - degenerate ( @xmath160 ) zero - magnetic field situation . in the hfa for 2d gaas electrons ( @xmath165 ) and holes ( @xmath166 ) respectively for @xmath167 k ( from bottom to top ) .
the solid horizontal line corresponds to the non - interacting @xmath30 result , @xmath168^{-1}$ ] .
insets of ( a ) and ( b ) : calculated @xmath169 , where the minimum of @xmath25 occurs , as a function of @xmath6 for electrons and holes respectively .
( c ) and ( d ) zoom - in figures for the low - density parts of @xmath170 versus @xmath3 for 2d gaas electrons and holes respectively . ] in fig .
[ fig : semi3](a ) and ( b ) we show our calculated @xmath171 in the hfa for 2d gaas electrons and holes respectively for @xmath172 k. the corresponding non - interacting @xmath30 result , @xmath173^{-1}=0.24(m_e / m)10^{-11}$ ] mev@xmath174 , is also shown as a constant horizontal line in each figure .
the calculated temperature dependence is stronger for the holes than the electrons in fig .
[ fig : semi3 ] since the scale for the @xmath6-dependence , @xmath175 , is much smaller ( almost by a factor of six for the same density ) for the holes compared with the electrons because of the large difference in the two effective masses ( @xmath176 versus @xmath177 ) . for @xmath178 @xmath9 : @xmath179 k , @xmath180 k. of course the qualitative behavior of @xmath181 is the same for both electrons and holes , it is only that the temperature scale for the holes is lower . the qualitative behavior of @xmath170 as a function density ( @xmath3 ) and temperature ( @xmath6 ) as shown in fig .
[ fig : semi3 ] is in very good agreement with experimental results@xcite , showing that the temperature dependence of compressibility can indeed be very important in samples with low densities ( and consequently with low @xmath175 ) , particularly for holes because of their large effective mass .
we note that @xmath182 k ( electrons ) and @xmath183 k ( holes ) for a density of @xmath184 @xmath9 ( and @xmath137 ) .
a particular qualitative feature of fig . [ fig : semi3 ] deserves special attention : @xmath185 as a function of @xmath3 for a fixed temperature shows a very sharp upward turn with well - defined minimum at low densities , particularly for the low-@xmath6 results .
this striking low-@xmath3 ( and low-@xmath6 ) non - monotonicity arises from a quantum - classical crossover effect which turns on when @xmath186 from being @xmath187 at some density and can therefore only be seen for low-@xmath6 and low-@xmath3 results .
the minimum occurs at a density @xmath169 which we plot as a function of temperature @xmath6 in the inset showing that @xmath169 increases with increasing @xmath6 . to further emphasize this interesting behavior , we show in fig .
[ fig : semi3](c ) and ( d ) , the low - density parts of @xmath170 versus @xmath3 for a few low @xmath6 values where the minimum is clearly visible .
this sharp increase of @xmath2 or @xmath170 for low density ( and low temperature ) has been experimentally observed in 2d gaas systems , and has often been associated with the 2d metal - insulator - transition ( mit ) driven by disorder . by contrast , our theory does not include disorder effects , we only include finite - temperature effects which are , of course , very strong at low densities where @xmath186 .
the intuition based on the @xmath30 theory of compressibility clearly must fail at some low density ( i.e. , for @xmath188 ) since the @xmath30 limit of @xmath2 becomes large ( in magnitude ) and negative at very low @xmath3 whereas the finite-@xmath6 theory ( _ for any finite @xmath6 _ ) predicts that at the lowest densities , where @xmath186 , @xmath2 must become large and positive ( i.e. , the classical behavior for @xmath86 ) , in fact eventually diverging as @xmath189 for @xmath86 ! we believe , based on the results presented in fig .
[ fig : semi3 ] , that experimentalists should re - investigate the older data for @xmath2 at low densities where finite temperature effects may be playing a significant role . there are some recent experimental results ref .
[ ] supporting our finding , but more experimental data are necessary in the interacting low density and low temperature regime to settle this question definitively . as a function of carrier density in the hfa for electrons ( @xmath165 ) and holes ( @xmath166 ) ,
respectively , over a wide range of temperature @xmath190 k. ( c ) and ( d ) show the high-@xmath6 behavior ( @xmath191 k ) of @xmath27 for electrons and holes respectively .
the temperature increases from bottom to top in each figure . ]
given the possible qualitative importance of finite temperature corrections to the 2d compressibility as discussed above ( and shown in fig .
[ fig : semi3 ] ) , we provide in fig . [
fig : semi4 ] the calculated @xmath27 in the hfa for both electrons ( fig .
[ fig : semi4](a ) ) and holes ( fig .
[ fig : semi4](b ) ) over a wide range of density and temperature .
the non - monotonicity apparent for higher-@xmath6 results in figs .
[ fig : semi4](a ) and ( b ) particularly for the hole data , is actually present in all the curves except that the non - monotonicity manifests itself at much lower density than the range covered in fig .
[ fig : semi4 ] for the curves that simply look like that @xmath27 keeps on decreasing with density monotonically . in figs .
[ fig : semi4](c ) and ( d ) we show the high-@xmath6 behavior ( @xmath191 k ) of @xmath27 . in fig .
[ fig : semi5 ] , we show our calculated compressibility for 2d gaas electrons and holes as a function of temperature for a few densities .
[ fig : semi5](a ) and ( b ) correspond to electrons showing @xmath27 as a function of temperature respectively over a wide @xmath192 k range [ [ fig : semi5](a ) ] and a narrow low temperature range @xmath193 k [ [ fig : semi5](b ) ] whereas figs .
[ fig : semi5](c ) and ( d ) show the same for 2d holes in gaas . in figs .
[ fig : semi5](b ) and ( d ) , we show a comparison between our low-@xmath6 analytical and numerical results also . the important point to note in fig .
[ fig : semi5 ] is that the low - temperature minimum should be observable in 2d semiconductor systems in careful measurements of @xmath27 as a function of temperature provided disorder effects are unimportant , i.e. , highest - mobility samples are used for the study .
lower density samples would typically manifest deeper minima as can be seen in fig .
[ fig : semi5 ] . in the hfa as a function of temperature for various densities .
( a ) and ( b ) are for electrons(@xmath165 ) , ( c ) and ( d ) are for holes(@xmath166 ) .
( a ) and ( c ) show @xmath27 in the hfa for a wide range of temperature ( @xmath194 k ) for various densities @xmath195 @xmath9 ( from top to bottom ) .
( b ) and ( d ) show @xmath27 at low temperatures ( @xmath193 k ) for various densities @xmath195 @xmath9 ( from bottom to top ) and compare the low - t analytical results ( dashed lines ) and numerical results ( solid lines ) . ]
in this section , we theoretically calculate the finite temperature compressibility of monolayer graphene including exchange interaction effects .
it has been shown that the hfa is an excellent quantitative approximation due to the small contribution of the correlation energy in graphene@xcite .
we focus on extrinsic graphene , i.e. , gated or doped graphene with a tunable 2d free carrier density @xmath3 of electrons ( holes ) in the conduction ( valence ) band , i.e. , the chemical potential @xmath23 being positive ( negative ) . for undoped ( intrinsic ) graphene
the chemical potential is zero even at finite temperatures , which gives rise to a logarithmically divergent compressibility if disorder effects are neglected @xcite .
the basic feature of graphene compressibility and its functional dependence on density have been well studied in the literature at @xmath30@xcite .
we use the same definition for the compressibility given in sec . [
sec : semi ] . however , there are two main differences between the usual 2d semiconductor and graphene .
the first one is the difference between their energy dispersion relation : 2d semiconductor has parabolic dispersion relation while graphene has linear dispersion .
the kinetic energy in graphene is given by @xmath196 , where @xmath197 is the wave vector with respect to the dirac point , @xmath198 is the fermi velocity with the value of @xmath199 cm / s and @xmath200 for the conduction and valence band respectively .
the second difference is that we need to consider the contribution from valence band electrons in graphene because graphene is a gapless semiconductor so that the valence band and the conduction band touch each other at the dirac point . in graphene the total degeneracy @xmath201 ( from spin degeneracy @xmath202 and valley degeneracy @xmath203 ) .
since the non - interacting chemical potential of graphene is @xmath204 at zero temperature the corresponding non - interacting inverse compressibility is given by @xmath205 where we use @xmath201 and the electron density is related to the fermi wave vector as @xmath206 .
unlike the regular 2d systems , noninteracting @xmath207 in graphene is density dependent ( i.e. @xmath208 ) due to the linear energy dispersion of graphene .
the chemical potential within hfa is the sum of the non - interacting kinetic energy part and the exchange self - energy part .
the exchange self - energy for graphene is given as@xcite : @xmath209 where @xmath210 indicates the band indices , @xmath211 is the bare coulomb interaction ( @xmath53 is the background dielectric constant ) , and @xmath212 arises from the wavefunction overlap factor where @xmath213 is the angle between @xmath214 and @xmath215 . in regular ( i.e. , non - chiral ) 2d systems , @xmath216 , but in graphene , which is a chiral material , the chiral factor @xmath217 is important due to the underlying pseudospin dynamics . with eq .
( [ eq : gex ] ) we calculate the interacting compressibility of graphene at finite temperatures .
since the zero temperature exchange self energy can be calculated analytically@xcite we first consider the zero temperature compressibility as the starting point of our discussion . at @xmath30 , the fermi distribution function in eq .
[ eq : gex ] becomes @xmath218 .
we divide the exchange self - energy into two parts@xcite : one from the intraband transition @xmath219 , and the other from the interband transition @xmath220 .
that is , @xmath221 , where @xmath222 where @xmath223 is the difference in the electron occupation from the intrinsic @xmath30 case . after some algebra ( the detail of the derivation is given in ref .
[ ] ) , we find the density dependent exchange contribution to the inverse compressibility @xmath27 at zero temperature @xmath224\ ] ] where @xmath225 with a momentum cut - off @xmath226 where @xmath227 is a lattice constant of graphene , @xmath228 is catalan s constant , and @xmath229 is the graphene coupling constant ( or fine structure constant ) .
we note that the @xmath230 term takes care of the divergent compressibility of intrinsic graphene which does not enter our discussion in any significant manner . unlike in 2d systems ,
where the interaction parameter @xmath231 , for graphene @xmath70 is a constant in density due to its linear energy dispersion .
however , by adjusting the background dielectric constant ( @xmath53 ) we can vary the @xmath70 value , from @xmath232 ( for @xmath233 , graphene suspended in vacuum ) to very small @xmath70 by making @xmath53 very large . for graphene on sio@xmath234 , @xmath235 . note that unlike ordinary 2d systems the calculated graphene compressibility with the exchange correction is always positive , which has recently been measured by several different techniques recently @xcite . in this subsection
, we present the theoretical formalism of the finite temperature @xmath27 and its asymptotic analytical formula at low temperatures ( @xmath236 ) .
we consider extrinsic graphene with @xmath237 and concentrate on the situation with the chemical potential lying in the conduction band with no loss of generality . the finite temperature chemical potential ( without exchange energy ) @xmath238
must be calculated by the conservation of the total electron density , i.e. @xmath239 , where @xmath240 and @xmath241 are the electron and hole density at @xmath6 , respectively .
they are given by @xmath242 + 1}\ ] ] and @xmath243 + 1},\ ] ] where @xmath42 is the total degeneracy and @xmath244 .
thus , we have the self consistent equation for @xmath23 as @xmath245 where @xmath246 is given by @xmath247 then we obtain the non - interacting chemical potential for both low and high temperature limits for graphene as @xmath248 \;\ ; { \rm for } \ ; t / t_f \ll 1 \label{eq : chel } \\
\mu_0(t ) & \approx & \frac{e_f}{4 \ln 2 } \frac{t_f}{t } \hspace{2.2 cm } { \rm for } \ ; t / t_f \gg 1 .
\label{eq : chet}\end{aligned}\ ] ] with eqs .
( [ eq : chel ] ) and ( [ eq : chet ] ) we can find the noninteracting graphene compressibility at low temperatures and at high temperatures , respectively .
let @xmath249 and @xmath77 , then the non - interacting part of the inverse compressibility can be written as @xmath250 thus the leading order correction to the inverse compressibility @xmath251 increases quadratically at low temperatures and then it decreases inverse linearly at high temperatures .
this fact indicates that the graphene compressibility shows the opposite behavior to the ordinary 2d systems as shown in sec .
[ sec : semi ] , where the inverse compressibility of 2d systems decreases first and then increases as the temperature increases .
we emphasize that this qualitative difference in the noninteracting finite-@xmath6 compressibility between 2d systems and monolayer graphene arises both from the linear , chiral energy dispersion of graphene and from its gaplessness allowing for thermal excitations of conduction valence band electron - hole pairs at finite @xmath6 .
now we calculate the full compressibility at low temperatures , including the exchange energy .
as done for the zero temperature case , the exchange self energy of graphene can be separated into contributions from the intrinsic part @xmath219 and the extrinsic part @xmath220@xcite .
we can see that the first term in eq .
( [ eq : e ] ) is @xmath219 while the second term is @xmath220 . for @xmath252
the hartree - fock self energy in graphene at finite temperature becomes @xmath253 the intrinsic exchange self - energy @xmath219 in eq .
( [ eq : e ] ) can be derived by using the fact that the exponential term in @xmath219 is exponentially suppressed and almost equal to @xmath254 in the low temperature limit .
then , we have @xcite @xmath255 + o(e^{-\beta } ) , \end{array}\ ] ] where @xmath256 , @xmath78 and @xmath62 are the complete elliptic integral of the first and second kinds , respectively .
note that the leading order temperature correction of the intrinsic exchange self - energy at low temperatures is exponentially suppressed . on the other hand ,
the extrinsic part of the exchange self - energy becomes @xmath257 \label{eq:2}\ ] ] where @xmath228 is catalan s constant , @xmath258 where @xmath259 is glaisher s constant and @xmath260 is euler s constant .
differentiating the total chemical potential with respect to the density , we have the asymptotic form of inverse compressibility in the low temperature limit @xmath261 , \label{eq : comreg}\end{aligned}\ ] ] where @xmath262 is the non - interacting part of the inverse compressibility given in eq .
( [ eq : grani1 ] ) . in the low temperature limit
the leading order temperature correction to the total inverse compressibility is the same as that in the 2d semiconductor system due to the exchange energy , i.e. , @xmath263 .
however , the logarithmic correction can be detected only at very low temperatures ( @xmath264 ) . in general
, the @xmath265 correction from the non - interacting part dominates at low temperatures ( @xmath266 ) and the positive coefficient of @xmath265 term gives rise to increasing behavior of @xmath2 with temperature .
our asymptotic results at low temperatures agree very well with the full numerical results shown in fig .
[ fig : gra](a ) . the results provided in this section
generalize the existing graphene literature on compressibility@xcite to finite temperatures . in graphene as a function of temperature @xmath116 for three different values of density in the units of @xmath267@xmath9 .
( b ) calculated @xmath27 in graphene as a function of carrier density @xmath3 for six different values of rescaled temperature @xmath116 . in this calculation the momentum cut - off @xmath268 ( @xmath269 ) and @xmath270 are used .
( c ) the semi - log plot of @xmath27 versus carrier density with different temperatures @xmath6 for @xmath271 and ( d ) for @xmath270 . ] in this subsection , we present our numerical results of finite temperature compressibility in graphene . throughout this section
, we use @xmath272 cm / s and the wave vector cut - off @xmath273 ( @xmath269 ) .
we show results for suspended graphene ( @xmath274 ) and for graphene on sio@xmath234 ( @xmath275 ) . in fig .
[ fig : gra](a ) we show the temperature dependence of the total inverse compressibility @xmath27 for three different values of carrier density @xmath276 , @xmath277 @xmath9 , @xmath278 @xmath9 with dielectric constant @xmath270 ( corresponding to graphene on the si / sio@xmath234 substrate ) .
we can clearly see that the temperature dependence is non - monotonic and this non - monotonic behavior mostly comes from the non - interacting compressibility , which is different from the 2d semiconductor case .
unlike the 2d parabolic - band case , the compressibility in graphene does not change sign in the range of experimentally relevant parameters and typically exchange corrections are always quantitatively small ( @xmath279 ) .
we show the carrier density dependence of @xmath27 in fig .
[ fig : gra](b ) , where the @xmath27 is a monotonically decreasing function of carrier density for small values of @xmath116 ( @xmath280 for @xmath281 and @xmath282 ) . in fig .
[ fig : gra](c ) and ( d ) , we present the calculated @xmath27 as a function of carrier density for different temperatures . for fixed temperatures ,
@xmath27 is a decreasing function of @xmath3 at higher carrier density , which corresponds to the low temperature limit manifesting @xmath283 density dependence . on the other hand , at lower carrier density , @xmath27 has @xmath284 density dependence , which corresponds to the high temperature limit and the main contribution to @xmath27 comes from the non - interacting part ( see eq .
[ eq : grani2 ] ) . in monolayer graphene as a function of temperature @xmath116 for three different values of background dielectric constant within hfa for carrier density @xmath285@xmath9 .
the dot - dashed ( orange ) line presents the temperature dependence of the non - interacting @xmath27 .
( b ) calculated @xmath27 in monolayer graphene as a function of the background dielectric constant @xmath53 for a few fixed values of @xmath116 .
@xmath268 ( @xmath269 ) . ] in fig .
[ fig : gracom ] , we compare the temperature dependence for different values of background dielectric constant @xmath271 ( i.e. graphene in vacuum ) , @xmath270 ( graphene on si / sio@xmath234 ) and @xmath286 ( graphene on hafnium oxide hfo@xmath234 ) , respectively .
the value of @xmath70 represents the fine structure constant of graphene which depends only on @xmath53 and does not depend on the carrier density because of the linear energy dispersion .
larger dielectric constant corresponds to smaller values of @xmath70 , which indicates a weak - coupling system in terms of electron - electron interaction@xcite .
note that the trends of @xmath27 among different values of dielectric constant are similar .
the larger the value of dielectric constant ( the smaller the value of @xmath70 ) , the less strong is the density dependence of @xmath27 . we also show in this figure our calculated @xmath27 in hfa compared with the noninteracting @xmath27 as a function of the background dielectric constant @xmath53 for a few fixed values of @xmath116 .
we could see that the inverse compressibility @xmath27 first increases as the temperature @xmath116 increases , reaches a maximum , and then decreases as the temperature @xmath116 further goes up .
this non - monotonic behavior of @xmath27 dependence on the temperature is mainly due to the non - interacting part of the chemical potential , which is consistent with the calculation of the non - interacting part of the chemical potential ( eq.[eq : chet ] ) . from eq ( [ eq : chet ] ) we know that the temperature dependence of the chemical potential is @xmath265 ( with a positive slope ) at low temperatures while the temperature dependence of the chemical potential approximates as @xmath287 ( with a negative slope ) at higher temperatures .
there must be an intermediate point at which the chemical potential changes its sign of slope and so does the compressibility @xmath1 . since @xmath2 is the derivative of chemical potential with respect to @xmath3 , it has similar temperature dependence as the chemical potential
. therefore , the non - interacting part of @xmath27 behaves as @xmath265 at low temperatures and then changes to @xmath287 behavior .
the domination of graphene compressibility by noninteracting effects is just a direct manifestation of the weakness of electron - electron interaction in graphene compared with the 2d semiconductor case .
in this section , we calculate the blg compressibility within the hf approximation . there is some related theoretical work in the literature dealing with the problem of compressibility in blg , @xcite but the specific and detailed temperature dependent compressibility presented in the current work is not available in the literature . to calculate blg compressibility at finite temperatures we use the two - band approximation , which is valid for the low density limit . at high density ,
blg dispersion approaches the linear energy dispersion of monolayer graphene as the fermi energy becomes large@xcite , and therefore the high - density blg compressibility behaves similar to the mlg compressibility studied in the last section .
it has been shown that the two band model presents the most important qualitative signatures of the blg compressibility .
@xcite for low density bilayer graphene within two - band model , the quadratic approximation is commonly used , i.e. , @xmath288 , where @xmath289 is the interlayer hopping parameter from the tight - binding approximation and @xmath290 is the effective mass of the electrons .
this dispersion comes from a low - energy effective theory of bilayer graphene@xcite , which essentially discards the two split bands and confines electrons to those lattice sites not involved in the interlayer coupling .
while the quadratic dispersion is relevant at low energies , the actual dispersion of the bands is hyperbolic . at large wave vectors , relevant at large densities ,
the dispersion is effectively linear and the system should behave like single layer graphene . thus our results obtained in the previous section should be applicable to the high density blg .
the exchange self energy of blg is given by @xmath291 where @xmath292 is the coulomb potential , @xmath293 , @xmath294 is the fermi distribution function , and @xmath295 is the wave function overlap factor and @xmath296 is the angle between @xmath297 and @xmath298 . to get the explicit temperature dependence we rewrite eq .
( [ eq : blg ] ) as @xmath299 [ 1-\cos(2\theta ) ] , \label{eq : blg2}\end{aligned}\ ] ] where @xmath300 and @xmath301 , @xmath302 is the ordinary 2d self energy given in eq . ( 5 ) , and @xmath303 is given by @xmath304 \nonumber\end{aligned}\ ] ] where @xmath305 and @xmath306 here @xmath307 , @xmath308 is the momentum cut - off ( similar to the mlg situation ) with a lattice constant @xmath227 . @xmath78 and @xmath62 are the complete elliptic integral of the first and the second kind , respectively . since we have studied @xmath309 in sec .
ii and @xmath303 is independent of temperature , the new feature of temperature dependent compressibility of blg arises from the third term in eq . ( [ eq : blg2 ] ) , i.e. , @xmath310 [ 1-\cos(2\theta ) ] .
\label{eq : blg3term}\ ] ] then the total temperature dependent hfa chemical potential of blg can be calculated to be @xmath311 at @xmath30 , using @xmath312 and @xmath313 in eq .
( [ eq : blg3term ] ) , the exchange self - energy can be calculated as : @xmath314.\end{aligned}\ ] ] then , the exchange self - energy at @xmath63 is given by @xmath315.\ ] ] where @xmath225 .
the total chemical potential for blg within hfa is then written as @xmath316\right)\right ] \label{eq : chblg}\ ] ] where @xmath317 is the dimensionless blg coupling . in the following numerical calculation
, we use the the value of dielectric constant @xmath318 and effective mass @xmath319@xcite .
differentiating eq .
[ eq : chblg ] with respect to @xmath3 and making use of the relation @xmath320 , we get the zero temperature blg inverse compressibility within hfa @xmath321\right)+\frac{r_s}{\pi } w(y ) \label{eq : blgcomzero}\ ] ] where @xmath28 is the non - interacting compressibility of blg , given by @xmath322 and @xmath323 at low enough densities the zero temperature compressibility within hf approximation becomes negative similar to the corresponding 2d semiconductor system .
the corresponding density is @xmath324 @xmath325 or @xmath326 , for which the negative compressibility manifests itself .
this is a very small ( large ) value of density ( @xmath70 ) compared with the ordinary 2d case where the negative compressibility shows up for @xmath327 .
however the negative compressibility obtained in hfa has not been experimentally observed in bilayer graphene systems@xcite .
the non - negative compressibility may be attributed to correlation effects .
it is shown that the contributions from exchange and correlation almost cancel each other out in blg at low density , leaving a small positive compressibility at all densities in contrast to the corresponding 2deg systems @xcite where the electronic compressibility can become negative at low densities .
@xcite the non - quadratic band structure also gives rise to significant affects on the compressibility as has recently been discussed in the literature .
@xcite in blg as a function of temperature @xmath77 for different values of @xmath70 .
the inset presents @xmath27 for @xmath328 showing the non - monotonic behavior .
@xmath268 ( @xmath269 ) and @xmath270 are used .
( b ) the inverse compressibility @xmath27 in blg as a function of @xmath70 for different values of temperature @xmath116 .
( c ) @xmath329 in blg versus carrier density @xmath3 for various temperatures .
inset shows the details in the low density limit in the semi - log scale . ] at finite temperatures the non - interacting chemical potential is temperature independent in blg , i.e. , @xmath330 then the non - interacting compressibility of blg is also temperature independent and is the same as the zero temperature compressibility given in eq .
( [ eq : blgzero ] ) @xmath331 at finite temperatures , the temperature dependence of blg compressibility comes from the exchange contribution .
this is very intriguing because the non - interacting kinetic energy dominates exchange self energy at high temperatures for both ordinary 2d systems and mlg .
but in blg the entire temperature dependence arises from the exchange energy ! thus ,
any experimentally observed temperature dependence in the blg compressibility must entirely be a many - body effect , at least within the 2-band approximation ( i.e. , at not - too - high carrier densities ) .
let us consider the exchange self energy , eq .
( [ eq : blg2 ] ) , at finite temperatures . the first term in eq .
( [ eq : blg2 ] ) has been calculated in sec .
[ sec : semi ] [ see eq .
( [ eq:2dnormchem ] ) ] and the second term is temperature independent . the third term [ eq .
( [ eq : blg3term ] ) ] becomes in the low temperature limit @xmath332 :
@xmath333 thus , we have the total hfa chemical potential in the low temperature limit @xmath334\right)\right ] \nonumber \\ & -&e_f\frac{r_s}{\pi}\left[\frac{b t^2}{4}+\frac{5 \pi ^2 t^2}{72}-\frac{1}{48 } \pi ^2 t^2 \text{ln}\frac{16}{t}\right ] .
\label{eq : blgnormchem}\end{aligned}\ ] ] where @xmath99 \approx 0.449 $ ] and @xmath100 is the derivative of the zeta function . differentiating the asymptotic formula of chemical potential with respect to the density
, we have the blg @xmath27 for @xmath101 @xmath335 .
\end{array } \label{eq : lhfblg}\end{aligned}\ ] ] where @xmath336 has been derived in eq .
[ eq : blgcomzero ] .
we have the same leading order term @xmath337 arising from the exchange self - energy as the ordinary 2d system .
thus , at low temperatures the calculated inverse compressibility decreases as the temperature increases . in the high temperature limit @xmath91 , the chemical potential can be calculated to be : @xmath338\right)\right]\ ] ] where @xmath339/64 \approx 0.4 $ ] and @xmath340 is the zeta function .
then the inverse compressibility is given by @xmath341 in the high temperature limit ( @xmath113 ) the inverse compressibility of the ordinary 2d systems increases linearly with temperature and approaches the non - interacting value .
however , since the non - interacting compressibility of blg is temperature independent and the most dominant exchange term decreases as @xmath111 , the high temperature inverse compressibility of blg approaches the following high - temperature limit : @xmath342 where @xmath343 is the blg effective bohr radius .
since @xmath344 the negative inverse compressibility reverses its sign to the positive values at high temperatures and asymptotically approaches a smaller value than the zero temperature inverse compressibility .
this intriguing result is purely a high - temperature manifestation of exchange effect within the 2-band blg approximation . in fig .
[ fig : blg ] , we present our numerical results of temperature dependent inverse compressibility in blg systems . in the calculation we use @xmath268 ( @xmath269 ) and @xmath270 . as shown in fig .
[ fig : blg ] the calculated blg inverse compressibility shows very weak temperature dependence .
the weak non - monotonic behavior of temperature - dependent inverse compressibility is shown in the inset of fig .
[ fig : blg](a ) . in fig .
[ fig : blg](b ) , we present the inverse compressibility versus @xmath70 .
@xmath2 decreases monotonically with @xmath70 and becomes negative at @xmath326 . in fig .
[ fig : blg](c ) ) we show @xmath22 as a function of density and the inset shows the same figure in the semi - logarithm scale to show clearly the behavior at low densities . at densities @xmath345
the zero temperature @xmath346 is negative , but as temperature increases it reverses its sign at @xmath347 ^ 2 $ ] .
as temperature increases further it approaches @xmath348 regardless of density . in fig .
[ fig : blgf ] , we show , for the purpose of comparison , the calculated @xmath27 for the 2d , mlg , and blg systems as functions of @xmath70 and @xmath116 in order to demonstrate the qualitative difference between the results . for the 2d , mlg , and blg systems as a function of @xmath116 for a few values of @xmath70 .
in this work , we have calculated the finite temperature compressibility of graphene and ordinary 2d semiconductor systems within the hartree - fock approximation .
we present both analytical and numerical results of @xmath27 as a function of temperature , the dimensionless interaction parameter @xmath70 , and density .
we find that the calculated temperature dependent compressibility is non - monotonic in both graphene and 2d semiconductor systems . in monolayer graphene ,
the inverse compressibility @xmath2 increases with temperature at low temperatures and decreases at high temperatures , reaching a maximum value at an intermediate temperature .
this non - monotonicity arises entirely from the behavior of the non - interacting graphene compressibility .
the temperature dependent inverse compressibility in 2d semiconductor systems decreases with temperature at low temperatures and increases at high temperatures . in blg
the inverse compressibility decreases as temperature increases at low temperatures , but it approaches a value which is less than the non - interacting value at high temperatures .
the leading order temperature correction to the inverse compressibility @xmath2 in hfa for both graphene and 2d semiconductor systems is @xmath263 .
our analytic results are in agreement with our numerical results both at low and at high temperatures .
our use of the hartree - fock approximation is not a particularly restrictive approximation for the theory of the electronic compressibility since it is well - known that hartree - fock theory works well for the calculation of the compressibility and its density dependence , at least at @xmath30 .
the hartree - fock approximation gives results numerically very close to the full dynamical rpa for the 2d compressibility at @xmath30 @xcite , and the hartree - fock compressibility results are in good agreement with the compressibility measurements for 2d semiconductors@xcite and graphene@xcite .
in fact , the hartree - fock approximation for 2d semiconductors is in better quantitative agreement with the density dependence of the measured compressibility@xcite than the corresponding rpa theory as long as the strict 2d approximation is used since the finite width corrections for the realistic 2d semiconductors tend to cancel out the correlation corrections neglected in the hfa . in any case
, we consider the hfa as the first step necessary for understanding the temperature dependent 2d compressibility in interacting electron systems , and only future work , particularly experiments comparing with our predictions , can establish the necessity of improved theoretical treatments involving dynamical correlations neglected in the hfa .
our main finding that the temperature dependence of compressibility is much stronger in 2d systems than in graphene is consistent with existing experimental results .
in particular , our results of sec .
[ sec : semi ] agree well with the experimental study of shapira _
et al._,@xcite on 2d gaas holes who discovered substantial temperature dependence in the 2d hole compressibility in the @xmath349 k temperature regime with the experimental data showing reasonable qualitative agreement with our theoretical hfa results presented in sec .
[ sec : semi]d .
detailed quantitative comparison with the experimental data of shapira _
et al.,_@xcite is not particularly meaningful ( and is not attempted ) because of a number of reasons including our use of the exchange - only hartree - fock approximation , our neglect of the finite thickness of the 2d semiconductor structures which is often important for quantitative considerations , our neglect of disorder effects which are certainly important at lower carrier densities , and the lack of information about some essential experimental parameters ( e.g. , depletion charge density ) .
in addition to the experimental study of the temperature dependent 2d hole compressibility by shapira _
et al . , _
@xcite discussed above , there have been several low - density studies of the temperature dependence of the compressibility in both electron and hole 2d systems@xcite carried out in the context of 2d mit ( i.e. , disorder - induced density - tuned localization of 2d semiconductor systems ) . as we discussed in sec .
[ sec : semi ] , these low - density experimental investigations should be revisited in light of our current work demonstrating the importance of the temperature dependence of @xmath27 in the low carrier density regime by virtue of the dimensionless temperature @xmath189 being large precisely in the low - density regime ( where disorder effects are also strong ) since @xmath137 in 2d systems .
unfortunately , temperature , disorder , and interaction effects are all strong in the low density regime , making any quantitatively reliable theoretical work essentially impossible in the low carrier density regime , and our work establishes the importance of finite temperature effects on the compressibility in the low carrier density regime .
in particular , the density inhomogeneity and puddle formation that happens at low carrier density @xcite due to the failure of screening would very much complicate the observation of the low - density ( and low - temperature ) features in the compressibility predicted in our theory unless one uses extremely high - quality samples with very low disorder .
recently , there has been a theoretical investigation @xcite of the effect of disorder on low - density graphene compressibility clearly establishing the importance of disorder in low - density compressibility .
since the fermi temperature @xmath175 goes as @xmath175(in k)@xmath350 ( 2d gaas electrons ) ; @xmath351 ( 2d gaas holes ) ; @xmath352 ( monolayer graphene ) ; @xmath353 ( bilayer graphene ) , where @xmath354 is the carrier density measured in the units of @xmath8 @xmath9 , it is obvious that the quantitative effect of finite temperature , even at low carrier densities , is by far the strongest ( weakest ) in 2d gaas holes ( monolayer graphene ) , which is consistent with experimental observations . since disorder effects become important for @xmath355 , where @xmath356 is the background random charged impurity density in the environment , the low density regime associated with large @xmath189 values becomes even more challenging to achieve in the laboratory experiments on graphene .
we expect our predicted temperature dependence to manifest in graphene compressibility in room temperature experiments in very high - mobility samples where @xmath357 values may be achievable .
an important relevant question for our theory is : what is the most suitable system for the experimental observation of our theoretical predictions ? obviously ,
our predicted high - temperature behavior , where the theory is in very firm ground since the exchange energy is likely to be the exact leading - order many - body correction to the non - interacting compressibility in the high - temperature limit , should be valid for all systems and should be observable in 2d semiconductors ( both electrons and holes ) rather easily by measuring the compressibility for @xmath358 k for electron ( hole ) densities@xmath359 @xmath9 so that @xmath360 condition is satisfied .
given that @xmath361 at the same density , the high - temperature behavior is much more easily observable in 2d gaas holes , and has , in fact , already been observed at least in one experimental study in 2d holes@xcite . while 2d gaas holes are the obvious ideal candidates for observing our predicted high - temperature compressibility behavior , monolayer graphene may turn out to be not particularly well - suited for the temperature - dependent compressibility studies because of its very high relative fermi - temperature ( @xmath362 k for a doping density of @xmath267 @xmath9 ) and its relatively weak temperature dependence . obviously , the condition for observing the high - temperature behavior of compressibility is the ability to reach @xmath363 which necessitates lower fermi temperatures .
by contrast , the low-@xmath6 behavior associated with the @xmath0 term in the compressibility may be better observed in a system with a relatively high value of @xmath175 so that the regime @xmath364 can be explored over a fairly broad range of temperature with the temperature @xmath117 not being too low .
thus , very high mobility n - gaas heterostructures or quantum wells with @xmath365 @xmath9 so that @xmath366 k ( and @xmath367 k ) may be the ideal system to search for our predicted shallow minimum in @xmath27 .
one serious problem is that the minimum may be too shallow to be uniquely determined experimentally .
finally , we discuss our very interesting analytical finding of the non - monotonicity in graphene inverse compressibility associated with the @xmath368 correction we obtain analytically ( and verify numerically ) .
although this would not be an easy effect to detect experimentally because of its quantitative weakness ( and became the subleading correction goes as @xmath369 and thus the ratio of the two , @xmath370 , is always challenging to observe in experiments even under the best of circumstances ) , it is nevertheless interesting to discuss its origin and its robustness beyond our approximation scheme .
our results are exact within the hartree - fock approximation scheme , and there is no doubt that the exchange self - energy contributes a @xmath0 leading - order contribution to the inverse compressibility .
the first question is whether this is intrinsically a dimensionality effect occurring only in two dimensions .
we have therefore carried out the corresponding analytical calculations for the 3d temperature - dependent compressibility ( presented in the appendix [ sec : appen ] of this paper ) , finding that @xmath27 has a @xmath0 contribution in three dimensions also .
thus , it appears that the interesting non - monotonicity associated with the @xmath0 correction is a results of the hartree - fock approximation arising from the exchange self - energy diagram , and is not intrinsically a two - dimensional effect .
we have investigated this question by calculating the interacting compressibility in the screened hartree - fock approximation ( sometimes also called static rpa ) for the 2d system , where the bare coulomb interaction ` @xmath14 ' appearing in the exchange self - energy is screened by the static rpa dielectric function .
( these results are shown in appendix [ sec : appenb ] . )
the screened hfa ( appendix [ sec : appenb ] ) or static rpa does not contain the @xmath0 term in the finite temperature inverse compressibility , but is not a reliable approximation at all since at @xmath30 it predicts that @xmath27 should be always positive for all values of @xmath70 in clear disagreement with experimental finding@xcite .
thus the absence of the @xmath0 term in the screened hf approximation can not be taken seriously since the corresponding @xmath30 result is in qualitative disagreement with experimental results .
it is interesting to speculate whether the @xmath0 term survives higher - order diagrams associated with dynamical correlation effects , and this remains a challenge for the future . although it is fairly straightforward to calculate the compressibility including correlation effects at @xmath30
for example , the ring - diagram contributions to the compressibility can be exactly calculated at @xmath30 @xcite it is a formidable challenge , both numerically and analytically , to extend to the corresponding ring diagram calculations of interacting compressibility to finite temperatures .
we have studied this question carefully and have been able to show that there is definitely a @xmath369 term in @xmath27 arising from the infinite series of the ring diagrams , but we still do not know whether the @xmath0 term of the exchange self - energy is exactly canceled by the dynamical effects arising from the infinite ring diagram series . although the effect of the @xmath371 term matters only at low temperatures and low densities , where disorder effects dominate making it difficult to observe the log term experimentally , it is nevertheless important to establish whether this non - monotonicity associated with @xmath0 survives higher - order correlation terms in the theory .
we leave this as an unanswered theoretical question for the future .
a direct experimental observation of the @xmath0 term in the low - temperature inverse compressibility of 2d electrons or holes will go a long way in settling this important question .
to summarize , our goal of this paper is to understand the temperature dependence of compressibility in a high density homogeneous system where the interaction effect is not too strong and hfa is valid . at very low density ,
disorder effects are of particular importance and the system may be highly inhomogeneous due to the formation of electron - hole puddles@xcite , which are not included in our theory .
our main result is that temperature effects in the compressibility could be quite important at higher temperatures .
the work is supported by us - onr - muri and nri- nsf - swan .
in this appendix , we provide the corresponding finite temperature hartree - fock compressibility for a standard three dimensional electron gas ( 3deg ) . for 3deg , the bare coulomb interaction @xmath372 , where @xmath53 is the background dielectric constant .
we assume the total degeneracy @xmath159 in the calculation .
the finite temperature exchange energy of 3deg is given by @xmath373 \end{array}\end{aligned}\ ] ] where @xmath374 , @xmath375 , @xmath20 is the effective mass , and @xmath376 is the finite temperature chemical potential without exchange energy . the non - interacting chemical potential , @xmath377 , is obtained by solving the following self - consistent equation , @xmath378 where @xmath379 , @xmath77 and @xmath380 for 3deg .
there is no explicit analytical formula for the finite temperature chemical potential in 3deg .
but at very low temperatures ( @xmath93 ) , the asymptotic form of the non - interacting chemical potential is given by @xmath381 the hfa chemical potential , @xmath23 , can be calculated by including the exchange energy for @xmath63 , i.e. , @xmath382 then , the normalized chemical potential @xmath74 can be expressed @xmath383 , \label{eq : chemire_a}\end{aligned}\ ] ] where @xmath384 with the dimensionless parameter @xmath385 , and @xmath386 by differentiating eq .
( [ eq : chemi_a ] ) with respect to @xmath3 and using @xmath387 , the corresponding non - interacting compressibility of 3deg is given by @xmath388 where @xmath28 and @xmath81 are the zero temperature and the finite temperature non - interacting compressibility , respectively , and @xmath389 .
the finite temperature inverse compressibility @xmath27 within hfa is given by @xmath390 \nonumber \\
& + & \dfrac{\alpha}{2\pi t } \int _ 0 ^ 1\text{d}x \ln \left ( \dfrac{1+x}{1-x } \right ) \left [ \frac{x ( x^2-h(t))}{\cosh ^2\frac{(x^2-\tilde{\mu _ 0})}{2 t } } + \frac{\frac{1}{x^3}(\frac{1}{x^2}-h(t))}{\cosh ^2\frac{(1/x^2-\tilde{\mu _ 0})}{2 t } } \right ] .
\label{eq : comre_a}\end{aligned}\ ] ] at zero temperature @xmath30 , we have @xmath391{\dfrac{4}{9\pi}}r_s.\ ] ] the hf chemical potential and the compressibility in the low temperature limit @xmath93 can be calculated by expanding the logarithmic function near @xmath392 .
then the chemical potential in eq .
( [ eq : chemire_a ] ) becomes @xmath393{\dfrac{4}{9\pi}}r_s \left [ 1+\frac{\pi^2 t^2}{24 } ( a_3 + \text{ln}t ) \right ] , \ ] ] where @xmath394 ( where @xmath395 is euler s constant , with numerical value @xmath396 . ) .
finally we find that the low temperature 3d inverse compressibility within hfa as @xmath397{\dfrac{4}{9\pi^4}}r_s \left [ 1 \right .
\nonumber \\ & + & \left .
0.136\pi ^2 t^2 + \dfrac{1}{8 } \pi ^2 t^2\text { } \ln ( \dfrac{\pi t}{8 } ) \right ] .\end{aligned}\ ] ] the asymptotic behavior of high temperature compressibility is obtained by approximating the fermi - dirac distribution function by the classical boltzmann function . in the high temperature regime @xmath398 ,
the normalized chemical potential within hfa becomes @xmath399{\frac{4}{9\pi^4}}r_s,\ ] ] and the corresponding high temperature inverse compressibility is calculated as @xmath400{\frac{4}{9\pi^4}}r_s,\ ] ] which agrees with the non - interacting high - temperature classical result .
we provide the screened hartree - fock compressibility ( sometimes also called static rpa ) for the 2d system , where the bare coulomb interaction ` @xmath14 ' appearing in the exchange self - energy is screened by the static rpa dielectric function .
the self - energy with the finite temperature static rpa dielectric function is given by @xmath401 where @xmath402 , @xmath403 is the fermi distribution function .
the static rpa dielectric function @xmath404 is given by @xcite @xmath405 where @xmath406 is the finite temperature static polarization function and @xmath407 is the momentum dependent screening wave vector @xcite . at low temperatures ( @xmath92 )
the 2d polarizability becomes @xcite , @xmath408 and its asymptotic form at high temperatures ( @xmath86 ) becomes@xcite @xmath409 the finite temperature chemical potential within static rpa , @xmath36 , can be calculated by including the self - energy of eq .
( [ eq : selfb ] ) , i.e. , @xmath410 at zero temperature ( @xmath30 ) , the non - interacting part of chemical potential is the fermi energy @xmath411 . for @xmath412 ,
the chemical potential approaches the non - interacting value , which has been discussed in sec .
[ sec : semi ] . for interacting 2d systems the self - energy within screened hfa has the asymptotic form ( with @xmath202 and @xmath161 ) for @xmath413
, @xmath414 differentiating eq . [ eq : self0 ] with respect to @xmath3 and using the relation @xmath415 , we get the zero temperature inverse compressibility within the screened hfa @xmath416 since @xmath28 is a positive value we have the positive compressibility even for @xmath417 , which clearly disagrees with both the hf results and experiments@xcite . with eqs .
( [ eq : selfb])([eq : apblowt ] ) , we find the asymptotic form of the self - energy at low temperatures ( @xmath82 ) and for @xmath413 @xmath418 we find that the leading order temperature dependent term in the self energy is exponentially suppressed .
consequently , the temperature dependent inverse compressibility at low temperatures becomes @xmath419 at high temperatures ( @xmath86 ) , the inverse compressibility calculated within the static rpa approaches the hfa results given in eq .
( [ eq : semiht ] ) because the non - interacting kinetic energy dominates . | we calculate the finite temperature compressibility for two - dimensional semiconductor systems , monolayer graphene , and bilayer graphene within the hartree - fock approximation .
we find that the calculated temperature dependent compressibility including exchange energy is non - monotonic . in 2d systems at low temperatures
the inverse compressibility decreases first with increasing temperature , but after reaching a minimum it increases as temperature is raised further . at high enough temperatures the negative compressibility of low density systems induced by the exchange energy becomes positive due to the dominance of the finite temperature kinetic energy .
the inverse compressibility in monolayer graphene is always positive and its temperature dependence appears to be reverse of the 2d semiconductor systems , i.e. , it increases first with temperature and then decreases at high temperatures .
the inverse compressibility of bilayer graphene shows the same non - monotonic behavior as ordinary 2d systems , but at high temperatures it approaches a constant which is smaller than the value of the non - interacting bilayer graphene .
we find the leading order temperature correction to the compressibility within hartree - fock approximation to be @xmath0 at low temperatures for all three systems . |
hermes ( * her*a * me*asurement of * s*pin ) is an experiment which is comprehensively studying the spin structure of the nucleon by deep inelastic scattering ( dis ) of polarised positrons from polarised protons and neutrons @xcite .
both inclusive and semi - inclusive spin dependent scattering are simultaneously measured with good particle identification . by measuring the longitudinal polarization asymmetry of the cross section ,
hermes determines the nucleon spin structure functions in a wide range of @xmath1 and @xmath2 ( 0.02 @xmath3 0.8 , 0.2 @xmath4 20 ) , to more precisely test fundamental sum rules such as those of bjrken and ellis - jaffe .
a central aspect of the physics program is ` flavor - tagging ' the struck quark via detection of the leading hadron in semi - inclusive channels , which enables hermes to disentangle the spin contributions of different quark flavors and of gluons , in an effort to solve the nucleon spin puzzle @xcite .
the hermes spectrometer @xcite is installed in the east hall of the hera storage ring at desy .
it consists of two identical halves above and below the positron ring plane .
this provides two independent measurements of spin observables and thus a cross check on systematic uncertainties .
the spectrometer is configured around a large dipole magnet with a bending strength of 1.3 t@xmath5 m and scattering angle acceptance 40
220 mrad , a tracking system with chambers before , in and behind the magnet , and a particle identification detector ( pid ) system .
the pid system consists of four detectors : a lead - glass calorimeter , two plastic scintillator hodoscopes , a transition radiation detector , and a threshold erenkov detector .
the hodoscope immediately in front of the calorimeter is preceded by two radiation lengths of lead and acts as a pre - shower detector .
a more detailed description of the spectrometer and of its performances is given in refs .
@xcite .
the hermes spectrometer has been in operation for about three years , for measurements on polarised and unpolarised targets of @xmath6h , @xmath7h , @xmath8he and n. this paper reports on the performance of the hermes calorimeter during this running period .
the electromagnetic calorimeter is one of the four detectors of the hermes pid system .
its function is : _
i ) _ to provide a first - level trigger for scattered positrons , based on energy deposition in a localized spatial region ; _ ii ) _ to separate positrons from pions with a rejection factor of more than 10 at the first - level trigger and an additional factor of more than 100 in event reconstruction analysis ; _ iii ) _ to provide a measurement of the energy of dis positrons ; _ iv ) _ to measure the energy of photons from radiative processes or from @xmath9 and @xmath10 decays and _ v ) _ to give a coarse position measurement of scattered electrons and photons .
[ ht ] the solution chosen to meet these requirements consists of 840 radiation resistant f101 lead - glass ( lg ) blocks @xcite arranged in a configuration with one wall above and one below the beam , and with photomultipliers ( pmts ) viewing from the rear , as shown in fig . [ fig : calorimeter ] .
each wall is composed of 420 identical lead - glass blocks , stacked in a 42@xmath1110 array .
each block has an area 9@xmath119 @xmath12 and a length of 50 cm ( about 18 radiation lengths ) .
this cell size meets the requirement that @xmath13 90@xmath14 of the shower is contained in the cell for an axially - incident positron .
the blocks were polished , wrapped with 50 @xmath15 m thick aluminized mylar foil and covered with a 125 @xmath15 m thick tedlar foil to provide light isolation . each block
is coupled to a 7.5 cm photomultiplier philips xp3461 with a silicone glue ( silgard 184 ) with refraction index 1.41 .
a @xmath15-metal magnetic shield of 1.5 mm thickness surrounds the pmt .
the light seal is provided by an aluminium enclosure , which is mounted on a flange that is glued to the surface of the lead - glass .
this flange is made of titanium to match the thermal expansion coefficient of f101 .
it carries the light fiber for monitoring of the counter response .
the characteristics of the f101 blocks were measured at cern and desy test beams using 3x3 arrays of counters @xcite . before the installation in the hera east hall
, all lead - glass counters were equalised at desy with a 3 gev electron beam .
an array of forty - two blocks at a time was placed on a platform that could be moved in both the horizontal and vertical directions to vary the impact point of the beam on the counters . the response equalisation procedure consisted in adjusting the pmt high voltages so that the mean charge measured by the adc was @xmath16 ( pc ) = 22.22 @xmath17 ( gev ) , where @xmath17 is the mean energy deposition in the cell .
2@xmath18 and 2@xmath19 show the distributions of the means and variances ( in adc values ) of the spectra of the 840 blocks in response to a 3 gev electron beam incident at the center of the block .
the mean adc channels m of all f101 counters were adjusted to be between 580 and 620 .
the resulting distribution of the means has a average value 601 and width ( @xmath20 ) 6 .
this means an overall equalisation within 1% .
the standard deviations @xmath21 of the responses of the 840 lead - glass blocks are distributed around a central value 62 with @xmath20=3 : this implies a uniformity of the single - block resolutions to within 5% .
c + [ fig : equalization ] the block size was chosen in order to provide containment in a 3x3 matrix of more than 99@xmath14 of electromagnetic showers up to 30 gev energy .
hence shower leakage has negligible influence on the energy resolution . on the other hand
, the length of the lead - glass module does not lead to excessive absorption of erenkov light .
measurements with 130 gev electron beams have been performed at cern and desy with a 3x3 array of counters : all data , apart from that at 1 gev , are reproduced to better than 1@xmath14 by a linear fit @xcite . in the off - line analysis of hermes data ,
the comparison of the energy @xmath17 to the independently measured momentum @xmath22 determined by tracking @xcite provides a good identification of scattered positrons over the whole energy range which constitutes a powerful tool for calibration .
in fact , after correction for radiative effects in front of the calorimeter , the ratio @xmath23 is expected to be close to unity , independently of the positron energy .
3 shows the calorimeter response for scattered positrons in comparison to the reconstructed momentum .
good linearity is observed over the full energy range . during the data taking period ( 1995 - 1997 ) the @xmath23 distribution for scattered positrons
was regularly observed for each individual counter . fig .
[ fig : ep ] shows a distribution of the means of such @xmath23 spectra , measured over about a one - year running period .
the ratio is distributed around the central value 1.00 with a width ( @xmath20 ) 0.01 , demonstrating a uniformity of response of the counters around @xmath24 .
[ fig : evsp ]
electromagnetic showers typically spread their energy over the eight modules surrounding the hit counter .
such a group of nine modules is called in the following a cluster .
while the energy distribution over the single blocks of the cluster strongly depends on the hit position relative to the module boundaries and on the angle of incidence , the cluster energy is found to be independent of them to better than 1% @xcite .
the energy resolution for scattered positrons obtained during normal operation is shown in fig .
[ fig : resolution ] .
the data are well described by the following parameterization : @xmath25 = \frac{(5.1 \pm 1.1)}{\sqrt{e ( \mathrm{gev } ) } } + ( 2.0 \pm 0.5 ) + \frac{(10.0 \pm 2.0)}{e ( \mathrm{gev})}\ ] ] which is slightly degraded compared to the test beam results ( @xmath26 = ( 5.1 \pm 1.1 ) / \sqrt{e ( \mathrm{gev } ) } + ( 1.5 \pm 0.5 ) $ ] ) @xcite .
this because of pre - showering of the positrons in the material before the calorimeter , which improves the discrimination between positrons and hadrons , but produces the @xmath27 term , and of imperfections in the gain matching among modules , which slightly enhances the constant term . note that these values are similar to those obtained for other large lead - glass calorimeters [ 9 - 15 ] in spite of the use here of a less transparent material .
the segmentation of the calorimeter allows to obtain the hit position from the energy distribution inside a cluster with an accuracy better than the cell size .
the hit position is calculated by using the following energy - weighted average position of the nine blocks of a cluster : @xmath28 and @xmath29 where @xmath30 and @xmath31 are the central horizontal and vertical coordinates of the @xmath32-module and @xmath33 is the corresponding measured amplitude .
[ fig : position ] shows the distribution of the differences @xmath34 between the estimated hit positions of scattered positrons in the calorimeter @xmath35 , and the extrapolations of the charged particle tracks , @xmath36 @xcite .
it is seen that the resolution of the reconstructed hit position amounts to : @xmath37 and is significantly better than the cell size .
these resolutions , which are the same for the @xmath1 and @xmath38 directions , were found to be almost independent of the energy @xmath17 of the incident positron .
the energy of the electromagnetic shower measured as the sum of two adjacent calorimeter columns is used to provide the first - level trigger for positrons in a deep inelastic process . during the 1995 data taking period
the trigger consisted of a coincidence of both hodoscopes and the calorimeter .
the trigger threshold was set to a deposited energy of 3.5 gev in 1995 and of 1.5 gev in late 1996 and 1997 .
this already provided a suppression of hadronic background of about one order of magnitude .
a forward trigger scintillator system was introduced in 1996 in front of the hermes spectrometer magnet .
it reduced the trigger rate from background generated by the hera proton beam by distinguishing forward and backward going particles by using the time of flight between forward and rear scintillators .
a gain monitoring system ( gms ) is used to monitor the possible gain variations of the photomultipliers during normal running .
the system is based on a dye laser light source at 500 nm , which sends light pulses of varied intensities through glass fibers to every pmt of the calorimeter , and additionally to a reference counter photodiode .
the different intensities are achieved by a rotating wheel with several attenuation plates .
the light is split in several stages and fed into glass fibers @xcite .
the ratios of multiplier signals to that of the reference photodiode can be used to monitor relative gain changes in the multipliers . the long - term stability of the calorimeter has been evaluated by observing changes of the pedestal and gain value .
these values have been found to be stable within the accuracy of the measurement during the entire time of operation .
[ fig : piedistalli ] shows the values of pedestals observed over a several months running period for two typical modules . from the known conversion gain of 5 mev / ch , it can be seen that the data are consistent to about 10 mev .
[ fig : gain96 ] shows the relative gain variations of two typical counters as a function of accumulated events for several months running period .
the values are the ratios between the actual and reference gains .
straight line fits to the data result in slopes of 1.2@xmath1110@xmath39 year@xmath40 and 1.1@xmath1110@xmath41 year@xmath40 , confirming the above stated long - term stability of the response to within 1% per year .
the long - term stability of the response can also be monitored by observing the mean value of the @xmath23 distribution , measured for each run .
[ fig : epmeanrun ] shows the distribution of the averages over all blocks of their @xmath23 centroids , accumulated over a one year running period .
it is seen that the response was stable within 0.5@xmath14 ( corresponding to the @xmath20 of the distribution ) .
degradation of the optical properties of the lead - glass by radiation is a danger in the hera environment .
the choice of f101 material was motivated by its radiation hardness .
in fact , previous measurements on a 45 cm long block with @xmath42 rays @xcite and high - energy hadrons @xcite have shown that an accumulated dose of more than 10@xmath7 gy produces a degradation of f101 transmittance less than @xmath43 over the lead - glass length .
after irradiation by 10@xmath44 gy the f101 turned visibly rust - brown with a tint of red and did not recover .
thus f101 is expected to be 1050 times less sensitive to radiation damage than other types of lead - glass , like sf2 @xcite , depending on wavelength .
this is due to the addition of cerium , which has the disadvantage that it worsens the optical transmission characteristics . to prevent radiation damage of the lead - glass ,
both calorimeter walls are vertically displaced away from the beam pipe by 50 cm during beam injection .
therefore , to monitor the potential radiation damage , particular attention was devoted to those blocks positioned at lower scattering angles , which should suffer a stronger gain reduction due to their proximity to the beam .
[ fig : gain96.top.bottom ] shows the distributions of the relative gains for a few of these blocks measured during one year of operation .
the central values are at 1.003 and 0.9999 for the top and bottom walls , respectively , with a @xmath20 @xmath13 1@xmath14 .
this result is also confirmed by the long - term stability of the response to the gms pulses shown in fig .
[ fig : gain96 ] . in conclusion , over three years of operations , there has been no observed degradation of performance that could suggest ageing effects .
radiation damage to the lead - glass is also monitored by using dedicated tf1 blocks placed behind the calorimeter .
this material is about 20 times more sensitive to radiation damage than f101 @xcite .
therefore , gain reduction would be seen sooner in these monitor detectors if there had been a large radiation dose incident on the back of the calorimeter caused by showers produced by beam loss in the hera proton storage ring . within the reproducibility of the measurements ( 1% ) , no variation has yet been observed in their response , indicating that the effect of radiation damage is negligible .
the hermes pid system has been designed to provide at least an order of magnitude in hadron suppression at the trigger level to keep data acquisition rates reasonable , and to provide a hadron rejection factor ( hrf ) of 10@xmath44 to keep the contamination of the positron sample by hadrons below 1% over the entire kinematic range .
the hrf is defined as the ratio of the total number of incident hadrons to the number of hadrons that are misidentified as electrons .
the calorimeter and the hodoscopes are used to select dis events .
this selection is more critical near the energy threshold where the ratio between the fluxes of pions and positrons is high .
it is accomplished with a passive radiator composed of 2 radiation lengths of lead sandwiched between two 1.3 mm stainless steel sheets and installed immediately before the second hodoscope .
this passive radiator acts like a preshower and initiates electromagnetic showers that deposit significantly more energy in the scintillator than minimum ionizing particles .
measurements with test beams @xcite have shown that such a configuration yields a hadron rejection factor of @xmath13 5@xmath1110@xmath45 in an event reconstruction analysis combining a lead - glass cut retaining 95% electron efficiency with a preshower cut keeping 98% .
specifically , the pion rejection provided by a single lead - glass block is about 100 and this is improved by the preshower by a factor of about 40 . during data acquisition
the hadron contamination at the trigger level was suppressed by the calorimeter threshold by a factor 10100 , depending on positron energy and threshold setting . in fig .
[ fig : rej_eff ] are shown the additional hrf and the efficiency for the combined calorimeter+preshower system obtainable in off - line analysis : the hrf ( efficiency ) values increase from @xmath4650 ( 0.94 ) at 4.5 gev up to @xmath46160 ( 0.98 ) at 13.5 gev .
it s worth noticing that in the event reconstruction the responses of the four pid detectors ( electromagnetic calorimeter , pre - shower , erenkov counter , and transition radiator detector ( trd ) @xcite ) are combined to further improve the hadron rejection to the required value .
more detailed studies on the particle identification system can be found in ref .
hermes provides detailed information on the hadronic final states in semi - inclusive deep inelastic scattering measurements .
this yields information on various flavor contributions to the nucleon spin .
the calorimeter plays an essential role in the identification of @xmath9 and @xmath10 , because they mainly decay into two photons ( branching ratios : ( [email protected] ) % and ( [email protected] ) % respectively @xcite ) , which are identified as pairs of energetic clusters in the calorimeter with no corresponding charged tracks in the spectrometer . from the energy measurement of the two photons and
the opening angle between them , it is possible to reconstruct the invariant mass of the corresponding meson .
12@xmath18 shows an invariant mass distribution for events with two neutral clusters in the calorimeter in coincidence with a scattered positron .
both the @xmath9 and @xmath10 peaks are clearly visible . fig .
12@xmath19 and fig .
12@xmath48 show the @xmath9 and @xmath10 invariant masses distributions obtained after applying kinematical cuts and background subtraction .
the centroids of the peaks are @xmath49=0.135 gev with @xmath20=0.011 gev , and @xmath50=0.549 gev with @xmath20=0.030 gev , which are in good agreement with the particle data group values @xcite .
the resolution of the estimate @xmath51 for the meson invariant mass can be expressed as follows : @xmath52^{1/2}\ ] ] where @xmath53 is the opening angle between the two photons . using this equation
we can examine whether the energy and position resolutions derived from calibration data still apply in the experimental environment .
[ fig : resolpi0 ] displays the @xmath9 invariant mass resolution obtained from dis events , and a monte carlo calculation based on the measured energy and position resolutions given in section 3 .
there is good agreement between the measured and calculated values . at low energies the energy resolution dominates the invariant mass resolution , while at high energies
the angular resolution is more important .
the decrease of slope for energies @xmath54 9 gev is due to a cut on minimum inter - cluster distance related to the cell size .
c + [ fig : invmass ]
the electromagnetic calorimeter is an important component of the hermes spectrometer .
it provides the dis trigger of the experiment in conjunction with scintillator hodoscopes and plays a major role in the particle identification .
in addition , it is essential for the identification of neutral particles in semi - inclusive measurements .
the performance and the stability of the calorimeter response were continuously measured during the past three years of data taking and the data are in good agreement with the design values and expectations .
they can be summarized as follows : * uniformity of the response of all counters within 1@xmath14 ; * linearity of the response to positrons within 1@xmath14 over the energy range 130 gev ; * resolution @xmath25 = \frac{(5.1 \pm 1.1)}{\sqrt{e ( \mathrm{gev } ) } } + ( 1.5 \pm 0.5)\ ] ] for a 3x3 array of counters and @xmath25 = \frac{(5.1 \pm 1.1)}{\sqrt{e ( \mathrm{gev } ) } } + ( 2.0 \pm 0.5 ) + \frac{(10.0 \pm 2.0)}{e ( \mathrm{gev})}\ ] ] for the whole calorimeter operating in the spectrometer , including the effect of pre - showering of the positrons in the material before the calorimeter ; * position reconstruction with resolution about 0.7 cm ; * stability in time of the response within 1@xmath14 ; * no observed degradation of performance due to radiation damage , within the accuracy of the measurements ; * a hadron rejection factor exceeding 10 at the trigger level , and a further off - line rejection factor of about 100 ; * reconstruction of @xmath9 and @xmath10 masses in agreement with the pdg values .
we gratefully acknowledge the frascati hermes technical staff m. albicocco , a. orlandi , w. pesci , g. serafini and a. viticchi for the continuous and effective assistance .
we are also grateful to c. a. miller for many stimulating discussions .
this work was in part supported by the intas ( contract number 93 - 1827 ) and the tmr research network ( contract number fmrx - ct96 - 0008 ) contributions from the european community .
15 hermes collaboration , proposal desy - prc * 90/01 * ( 1990 ) ; + k. ackerstaff _
et al . _ ,
b404 * ( 1997 ) 383 . v.
hughes and c. cavata , _ symposium on the internal spin structure of the nucleon _ - yale university 1994 , world scientific ( 1995 ) .
k. ackerstaff _
et al . _ , desy - hermes * 97 - 09 * ( 1997 ) , nucl .
instr . and meth . submitted ;
hermes collaboration , technical design report , desy - prc * 93/06 * ( 1993 ) .
m. dren , habilitation thesis , desy - hermes * 95 - 02 * ( 1995 ) . m. kobayashi _
_ , _ kek internal report _ * 93 - 178 * ( 1993 ) .
a. fantoni , ph.d .
thesis , february 1996 , universit degli studi della calabria , _ hermes internal note _ * 96 - 015*. h. avakian
_ et al . _ ,
instr . and meth .
* a378 * ( 1996 ) 155 .
w. wander , ph.d .
thesis , april 1996 , friedrich alexander university ( erlangen - nuernberg ) , _ hermes internal note _ * 96 - 023*. k. ogawa _
et al . _ ,
instr . and meth .
* a243 * ( 1986 ) 58 .
t. sumiyoshi _
et al . _ ,
instr . and meth .
* a271 * ( 1988 ) 432 .
h. stroher _
et al . _ ,
instr . and meth .
* a269 * ( 1988 ) 568 .
instr . and meth .
* a275 * ( 1989 ) 59 .
h. baumeister _
et al . _ ,
instr . and meth .
* a292 * ( 1990 ) 81 .
l. bartoszek _
et al . _ ,
instr . and meth .
* a301 * ( 1991 ) 47 .
w. brckner _
instr . and meth .
* a313 * ( 1992 ) 345 .
a. fantoni _
et al . _ , proceeding of calor96 , frascati june 1996 , frascati physics series , vol .
vi , ( 1997 ) 659 - 668 , a. antonelli _ et al . _
et al . _ ,
instr . and meth .
* 215 * ( 1983 ) 103 .
m. holder _
et al . _ ,
instr . and meth .
* 108 * ( 1973 ) 541 .
r. kaiser , ph.d .
thesis , april 1997 , simon fraser university , burnaby , _ hermes internal note _ * 97 - 017*. r. kaiser , _ hermes internal note _ * 97 - 025*. particle data group , _ review of particle physics _ , phys . rev . *
d54 * ( 1996 ) . | the performance of the electromagnetic calorimeter of the hermes experiment is described .
the calorimeter consists of 840 radiation resistant f101 lead - glass counters .
the response to positrons up to 27.5 gev , the comparison between the measured energy and the momentum reconstructed from tracking , long - term stability , hadron rejection and neutral meson invariant mass reconstruction are shown .
@xmath0 corresponding author : e - mail [email protected] |
study of form factors has been important in the field of hadron physics as they provide important information about internal structure of the nucleon @xcite .
recently , experiments have measured and produced remarkably precise data for the electromagnetic form factors @xcite .
the nucleon electromagnetic form factors have been used to study the distribution of charge and magnetization densities of quarks inside the nucleon @xcite . to get a more clear picture of the nucleon structure , one can fully parameterize the structure of the nucleon in terms of generalized parton distributions ( gpds ) @xcite which are functions of longitudinal momentum @xmath0 of the quark , the invariant momentum transfer @xmath1 and the skewness parameter @xmath2 which gives the fraction of the longitudinal momentum transfer to the nucleon .
the fourier transform of the gpds w.r.t transverse momentum transfer has not only been used to study the nucleon structure in transverse impact parameter space @xcite but it also gives a tomographic picture of the distribution of quark charge densities inside the nucleon through the impact parameter dependent parton distribution function ( ipdpdf ) @xmath3 for a quark of momentum fraction @xmath0 located at a transverse position @xmath4 .
it has been shown that ipdpdf of the transversely localized states of the same helicity is the two - dimensional fourier transform of the spin non - flip gpd @xmath5 @xcite .
the role of transverse momentum of parton has been significant to understand the nucleon structure as it yields information on the fragmentation functions and the single spin asymmetries ( ssas ) @xcite .
some efforts have been made to understand this phenomena by taking one gluon exchange in the final state interactions when the distribution functions depend on the longitudinal as well as transverse momenta of the partons @xcite .
transverse charge density defines a way to analyze electromagnetic form factors of hadrons moving at infinite momentum with transverse distance @xmath4 from the transverse center of momentum @xcite .
charge densities have been discussed in impact parameter space for the polarized and transversely polarized nucleons @xcite , in finite radius approximation @xcite , in scalar diquark model using transverse coordinate space @xcite as well as using sachs form factors @xcite .
transverse densities can also be useful in the study of the spatial distribution of the momentum @xmath6 and can be related with the fourier transforms of the gravitational form factors @xmath7 and @xmath8 @xcite .
in addition to this , magnetization density can be evaluated to get information on the anomalous magnetic moment of the nucleon @xcite .
recently , transverse charge and magnetization densities have also been studied in the holographic qcd @xcite .
the above discussion motivate us to study the charge and magnetization densities obtained from the helicity non flip and helicity flip form factors .
one of the most successful model which finds its application for the quantities discussed above consists of spin-@xmath9 system as a composite of spin-@xmath9 fermion and spin-1 vector boson .
we have generalized the framework of qed by assigning a mass @xmath10 to external electrons in the compton scattering process , but a different mass @xmath11 to the internal electron line and a mass @xmath12 to the internal photon line .
the idea behind this is to model the structure of a composite fermion state with a mass @xmath10 by a fermion and a vector constituent with respective masses @xmath11 and @xmath12 @xcite . in our case
we take @xmath13 @xcite which represents the momentum transfer exclusively in transverse direction leading to the study of ipdpdfs in transverse impact parameter space . this model
serve as a template to study the internal structure of nucleon and is self consistent since it has correct correlation of different fock components of the state as given by light front eigen value equation .
this model has been used to calculate the spin and orbital angular momentum of a composite relativistic system as well as the gpds in impact parameter space and it also gives the schwinger anomalous magnetic moment and the corresponding dirac s and pauli form factor including the vanishing of the anomalous gravitomagnetic moment .
the purpose of the present communication is to determine the distribution of nucleon charge density in transverse coordinate space and in impact parameter space using light front wave functions ( lfwfs ) of simple spin-@xmath9 objects like dressed quark or a dressed electron in theory .
we begin by computing the electromagnetic dirac and pauli form factors as well as the sachs form factors @xmath14 and @xmath15 from the overlap of lfwfs . the anomalous magnetization density of the nucleon obtained from the spin flip pauli form factor has also been discussed .
further , it would be significant to develop a relation between the distribution of quark charge densities inside nucleon in the transverse coordinate space as well as in the impact parameter space particularly for the @xmath16 and @xmath17 quark distribution by using a fitted set of parameters
. this will give a deeper understanding of both the spatial and transverse distribution of the quark densities .
furthermore , it would be interesting to extend the calculations to present the predict the results of the qcd transverse ads charge density inspired from the holographic qcd model .
the nucleon form factors are defined in terms of the electromagnetic current operator @xmath18 as @xcite p , | j^(0)|p,= |u(p,)(^f_1(q^2 ) + i q_f_2(q^2))u(p , ) , where @xmath19 is the momentum transfer and in the present case it is taken as space - like ( @xmath20 ) leading to @xmath21 .
the @xmath22 is the spin operator , @xmath23 denote the dirac matrices and @xmath24 for the dirac spinor with mass @xmath10 .
the initial ( final ) momentum and helicity of the nucleon are taken as @xmath25 @xmath26 respectively .
the dirac and pauli form factors ( @xmath27 and @xmath28 ) are normalized so that @xmath29 and @xmath30 are the nucleon charge and the anomalous magnetic moment at zero momentum transfer respectively .
the sachs form factors @xcite can be defined in terms of the dirac and pauli form factors as follows g_e(q^2)=f_1(q^2)-f_2(q^2 ) , [ ge ] and g_m(q^2)=f_1(q^2)+f_2(q^2 ) .
[ gm ] transverse charge density for a unpolarized nucleon @xmath31 gives the charge density of the nucleon at a transverse position @xmath4 for any value of the longitudinal momentum or position .
it can can be obtained by integrating the ipdpdfs over the longitudinal momentum fraction @xmath0 and can be expressed in terms of the dirac form factor @xcite as follows ( b_)= f_1(q^2 ) e^-i _ _ .
[ rho_density ] in terms of @xmath32 and @xmath33 the transverse charge density for unpolarized nucleon can be expressed as ( b_)=_0^ j_0(q b _ ) , [ tran_charge_density ] where @xmath34 is a cylindrical bessel function and @xmath35 .
the main advantage of charge density in impact parameter space is that it gives the density interpretation when one compares it with the fourier transform of the sachs form factor .
the density in impact parameter space is calculated from the dirac form factor which is further obtained by integrating the gpds over the longitudinal momentum fraction @xmath0 .
the gpds have been studied @xcite in impact parameter space where the parton distributions are expressed in terms of impact parameter dependent wave functions using lorentz invariance to describe the transverse structure of a fast moving hadron .
these distributions are the matrix elements of the quark operators between nucleon states with different momenta .
the gpds are of high interest as they can be related to the total angular momentum carried by the quarks in the nucleon and can be directly found from the deep virtual compton scattering ( dvcs ) experiments @xcite .
the dirac and pauli form factors can be expressed as overlaps of the light front wave functions ( lfwfs ) for the two - particle fock state .
we consider here a spin-@xmath9 system as a composite of spin-@xmath9 fermion and spin-1 vector boson .
the details of the model have been presented in ref .
@xcite , however , for the sake of completeness we present here the essential two - particle wave functions for spin up and spin down electron expressed as @xmath36 and @xmath37 with the energy denominator defined as @xmath38 here the formalism has been generalized by assigning a mass @xmath10 to external fermions in the compton scattering process but a different mass @xmath11 to the internal fermion line and a mass @xmath12 to the internal boson line .
the energy denominator can be further generalized by adjusting its power behaviour _ p _ @xcite as follows @xmath39 the term @xmath40 has been added for dimensionality and @xmath41 to satisfy the polynomiality condition . in the present work we have taken @xmath42 .
the dirac and pauli form factor defined in terms of the spin non - flip and spin flip matrix elements of the @xmath43 current @xcite are expressed as follows @xmath44\ , \label{diracff}\end{aligned}\ ] ] f_2(q^2)&= & -2m(q^1-iq^2 ) + & = & -2m(q^1-iq^2 ) .
[ pauliff ] the spin non - flip and spin flip gpds can similarly be expressed in terms of the spin up and spin down electron wave functions as h(x,0,_)&= & , [ h2 ] e(x,0,_)= .
[ e2 ] the dirac and pauli form factors obtained from eqs .
( [ spinup ] ) , ( [ spindown ] ) , ( [ diracff ] ) and ( [ pauliff ] ) are given as f_1(q^2)&= & m^4 dx ( 2 ( 1+x^2)(1-x ) i_1 + i_2 ) , f_2(q^2)&= & - 4 m^5 dx ( m x - m)x(1-x)^3 i_2 , where i_1&= & _ 0 ^ 1 d , + i_2&= & _ 0 ^ 1 d , + d&= & ( 1- ) ( 1-x)^2 q_^2 - m^2 x(1-x)+m^2 ( 1-x ) + ^2 x . [ integrals ] using eq .
( [ ge ] ) and ( [ gm ] ) , we can also solve for the sachs form factors and obtain @xmath45 g_m(q^2)&= & ( m^4 dx(2(1+x^2)(1-x)i_1 + i_2)- + & & 4 m^5 ( m x - m ) x ( 1-x)^3 dx i_2 ) . the transverse charge density from eq .
( [ tran_charge_density ] ) in terms of sachs form factors can now be written as ( b_)&=&_0^ j_0(q b _ ) _ 0 ^ 1 dx + + & & . in fig .
[ nucleon_density ] we have presented the result for nucleon charge density as a function of impact parameter .
the impact parameter space is phenomenologically important as it can be obtained from the gpds ( @xmath5 and @xmath46 ) which satisfy the lorentz invariance principle .
even the constraint due to the invariance of the light cone formulation under transverse boost @xmath47 identifies @xmath4 as transverse center of momentum of the partons in fock state .
in addition to this , impact parameter representation of the gpds also satisfies the positivity constraints @xcite . for numerical calculations
we have taken the value of masses of external fermion , internal fermion and internal boson as @xmath48mev and @xmath49mev respectively after imposing the condition @xmath50 to prevent nucleon decay @xcite . from the figure
, we find that the charge density is peaked near @xmath51 and decreases as the value of @xmath4 increases which implies a maximum distribution near the center of momentum .
similar technique has recently been used to extract the transverse pion and proton charge density from the pion form factor data @xcite where it has been discussed that the data from jlab 12 gev and eic @xcite will increase the dynamic extent of the form factor data to higher values of @xmath52 and thus reduce the uncertainties in the extracted pion transverse charge density . ) as a function of @xmath4 ( in @xmath53).,width=226 ]
the lfwfs in transverse coordinate space @xmath54 can be obtained by fourier transforming @xmath55 in the momentum space ( x,_)&= & e^i _ _ ( x , _ ) , [ fttrans ] where @xmath56 is the transverse coordinate .
the lfwfs for the spin up state in the transverse coordinate space can now be expressed as ^_++1(x , _ ) & = & i m^2 ( 1-x)^ r_k_0 ( r _ ) , + ^_+-1(x , _ ) & = & -i m^2x r_k_0 ( r _ ) , + ^_-+1(x , _ ) & = & - m^2 ( m x - m)(1-x)^ r_k_1 ( r _ ) , + ^_--1(x,_)&= & 0 , whereas for spin down state we have ^_++1(x,_)&= & 0 , + ^_+-1(x,_)&= & - m^2(m x - m)(1-x)^ k_1 ( r _ ) , + ^_-+1(x,_)&= & i m^2 x ( 1-x)^ r_k_0 ( r _ ) , + ^_--1(x,_)&= & - i m^2 r_(1-x)^ k_0 ( r _ ) , where @xmath57 is the bessel function of second kind . even though the nucleon charge density in impact parameter space ( @xmath31 ) is the more correctly defined quantity with proper invariance properties to be compared directly with the observables , we define the density distribution in the transverse coordinate space for the sake of comparison as follows p(_)&= & dx [ ^_++1(x , _ ) ^_++1(x , _ ) + ^_+-1(x , _ ) ^_+-1(x , _ ) + + & & ^_-+1(x , _ ) ^_-+1(x , _ ) ] , [ transcoor ] leading to p(_)&= & dx .
a precise relation between the nucleon density distribution @xmath58 and charge distribution @xmath31 can be obtained using elementary convolution theorems and fourier transforms ( details have been presented in appendix a ) and is given as ( x,_)= p(x , ) . [ relation ] in order to obtain the explicit contributions of the @xmath16 and @xmath17 quarks in the nucleon density distribution @xmath58 and charge distribution @xmath31 , we have followed the formalism used in ref .
@xcite where the isospin symmetry @xmath59 , @xmath60 has been used with similar type of relations for charge density in the impact parameter space .
a fitting can be carried out with the experimental data of the dirac form factors of the nucleon to obtain the fitted values of the masses of the internal fermions and bosons .
for the present work we have used the same set of fitted values of @xmath11 , @xmath61 and @xmath62 as in ref.@xcite . in fig .
[ p_and_rho_up_down ] , we have presented the result for density distribution and charge distribution for _ u _ and _ d _ quarks respectively .
these plots give the complete spatial information about the nucleon .
it is clear from the plots that charge and density distributions increase with the increasing value of the impact parameter @xmath4 and the transverse coordinate @xmath63 respectively , reach a maxima and then start decreasing .
the peak for the charge distribution in impact parameter space is around @xmath64 whereas the peak for the density distribution in transverse coordinate space is around @xmath65 for both the @xmath16 and @xmath17 quarks .
the magnitudes of @xmath66 and @xmath67 are however different from @xmath68 and @xmath69
. a closer look at the figure reveals that the density distribution in transverse coordinate space falls off slowly as compared to the charge distribution in impact parameter space .
this is due to the factor @xmath70 appearing in eq .
( [ relation ] ) . to get a more clear picture of the distribution
, we can plot the charge and density distribution at different values @xmath0 .
we have presented the results in fig .
[ p_and_rho_up_down_x ] and as expected , the peak for different @xmath0 values occur the same value of @xmath4 and @xmath63 but the magnitude of the peak decreases with the increasing value of @xmath0 .
it is important to mention here that since @xmath0 is the momentum fraction of the active quark , at @xmath71 , the active quark carries all the momentum and the contribution from other partons is expected to be zero at this limit .
the phenomenological analysis of existing data to determine the charge density of the proton and neutron has been discussed in ref .
@xcite where a model independent analysis of the charge density of partons has been done in the transverse plane for the nucleon .
the charge density of the @xmath17 quarks is found to be larger than that of the @xmath16 quarks by 30% .
we also observe a large value of the @xmath17 quark density as compared to the @xmath16 quark density in both the spaces .
the difference between the distribution in the impact parameter and transverse coordinate space can be explained as follows .
the gpds @xmath5 and @xmath46 have dependence on @xmath0 , @xmath2 and @xmath1 with @xmath0 being the longitudinal momentum fraction , @xmath2 the longitudinal momentum transfer and @xmath1 the invariant momentum transfer .
we have fixed the longitudinal momentum transfer as @xmath13 in the present calculations which in result fixes the longitudinal momentum fraction @xmath0 in the impact parameter space @xmath4 .
thus , one can get the exact information about the charge distribution in impact parameter space @xcite .
however , in the case of transverse coordinate space , @xmath63 is a simple coordinate parameter directly obtained by taking fourier transformation w.r.t .
momentum @xmath72 and there is no parameter like @xmath2 .
since the longitudinal momentum fraction @xmath0 can not be fixed in the transverse coordinate space , one can get some contribution from longitudinal momentum as well and this basically leads to the slight difference of charge distribution when compared with the distribution in the impact parameter space .
( in @xmath73 ) and @xmath74 ( in @xmath73 ) for the up ( down ) quark in @xmath4 ( in @xmath53 ) and @xmath63 ( in @xmath53 ) space respectively . , title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) for the up ( down ) quark in @xmath4 ( in @xmath53 ) and @xmath63 ( in @xmath53 ) space respectively . , title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) for the up ( down ) quark in @xmath4 ( in @xmath53 ) and @xmath63 ( in @xmath53 ) space respectively . ,
title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) for the up ( down ) quark in @xmath4 ( in @xmath53 ) and @xmath63 ( in @xmath53 ) space respectively . , title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) at @xmath75 , @xmath76 and @xmath77 . the units of @xmath4 and @xmath63 are in @xmath53.,title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) at @xmath75 , @xmath76 and @xmath77
. the units of @xmath4 and @xmath63 are in @xmath53.,title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) at @xmath75 , @xmath76 and @xmath77 .
the units of @xmath4 and @xmath63 are in @xmath53.,title="fig:",width=226 ] ( in @xmath73 ) and @xmath74 ( in @xmath73 ) at @xmath75 , @xmath76 and @xmath77 .
the units of @xmath4 and @xmath63 are in @xmath53.,title="fig:",width=226 ]
another important quantity which has a form similar to the transverse charge density of a polarized nucleon is the nucleon magnetization density .
following ref .
@xcite , we can define the anomalous magnetic moment @xmath78 as _ a = y| d^2 b b_y q^_+ q_+ | y , [ ano_mag_moment1 ] where @xmath79 is a quark field operator and @xmath80 .
the state for the transversely polarized target @xmath81 is |y = [ |p^+ , = , + + |p^+ , = , - ] , [ ano_mag_moment ] with @xmath82 representing a transversely localized state . using the impact parameter distribution @xcite
, the magnetization density can be expressed as _
m(b_)=f_2(q^2 ) e^-i . , [ rho_mag_moment ] which leads to the the anomalous magnetic moment _
d^2b _ _ m(b _ ) . using eq .
( [ ano_mag_moment1 ] ) we can also express @xmath78 in terms of the anomalous magnetization density @xmath83 as _
d^2 b b_y .
[ ano_mag_moment2 ] substituting the value of @xmath28 , we can write _ m(b_)&= & ^2 ( ) b__0^ j_1(q b _ )
f_2(q^2 ) + & = & ^2 ( ) b__0^ j_1(q b_)(- 4 m^5 dx ( m x - m ) x ( 1-x)^3 i_2 ) , [ rhomag ] with @xmath84 represents the angle between the direction of nucleon polarization ( or the transverse magnetic field ) and @xmath4 .
( in @xmath73 ) as a function of @xmath4 ( in @xmath53).,width=302 ] in fig .
[ mag3d ] we present the result for anomalous magnetization density as a function of @xmath4 .
since the magnetization density also depends on @xmath84 , we have taken @xmath85 in the present calculations where @xmath86 will have the largest value .
the anomalous magnetization density vanishes for @xmath87 .
the anomalous magnetization density also vanishes for @xmath51 and shows a maxima around @xmath88 . for higher values of @xmath4
it falls off very rapidly leading to negative @xmath86 values .
in this section we have studied the charge density in ads space .
the lfwfs encode all the properties of hadron like bound state quark and gluon properties .
a string amplitude @xmath89 can be defined on the fifth dimension in ads@xmath90 space which can easily map the lfwfs of the hadrons @xcite .
it has been shown that a relation exists between the fifth dimension holographic variable @xmath91 and impact variable @xmath92 which represents the measure of the transverse distance between the constituent within the hadrons .
this relation provides an exact mapping at all energy scales between string modes in ads and boundary states with a well defined number of partons .
the string mode @xmath89 can be taken as probability amplitude for n - partons at @xmath93 .
the form factor in ads is the overlap of the normalizable and the non - renormalizable modes @xciteand can be expressed as f(q^2)=2 _ 0 ^ 1 dx dj_0(q ) ( x , ) .
the impact parameter space , obtained by taking the two dimensional fourier conjugate with respect to transverse momentum transfer , gives us the information about the transverse structure of hadrons .
on the other hand , due to correspondence between the fifth dimensional variable @xmath91 and the impact variable @xmath92 for each n - parton fock state in ads space , it becomes very interesting to study the charge density in such a holographic model which explains the hadron spectrum remarkably .
since our aim is to study the charge density , the impact parameters in both the spaces can be related to each other by the relation @xmath94 for two partons . in the light of present work
, the string modes can be further related to the qcd transverse charge density @xmath95 as follows ( x,)= e^3 a ( ) . since the variable @xmath92 represent the transverse distance between the constituent and holographic variable @xmath91 in ads , we can take @xmath93 leading to _ n=2(x,)= , where _
l , k(x,)=b_l , k j_l(_l , k _
qcd)(z ) , and b_l , k=_qcd ^- . for two partons we can write the ads density for ground state @xmath96 and excited state @xmath97 as follows _ n=2(x,)(l=0,k=1)&= & , + _ n=2(x,)(l=1,k=1)&=&- . and @xmath92 ( in gev@xmath98 ) with @xmath99gev for ground state @xmath100 ( lhs ) and orbital excited state @xmath101 ( rhs ) . , title="fig:",width=226 ] and @xmath92 ( in gev@xmath98 ) with @xmath99gev for ground state @xmath100 ( lhs ) and orbital excited state @xmath101 ( rhs ) . , title="fig:",width=226 ] in fig .
[ ads_density ] we have plotted the ads density with @xmath99gev and @xmath102 .
we observe that for ground state @xmath96 the magnitude of charge density decreases as the value of @xmath92 increases which is primarily due to the presence of the bessel function in the lfwfs .
the magnitude is maximum when @xmath103 which implies that when the transverse distance between the constituent within the hadrons is minimum the density is maximum . in the first excited state @xmath97
, we have obtained a peak for charge density in ads space for a comparatively higher value of @xmath92 .
the density is also dependent on the variable @xmath0 and it is observed that for the lower values of @xmath0 , the density is zero .
the peak occurs only at higher values of @xmath0 .
more secondary peaks will be observed for the charge density as the value of @xmath104 and @xmath105 increase further on this scale .
in the present work , we have considered the light - cone fock state representation of the composite system consisting of external and internal fermion state with respective masses @xmath10 and @xmath11 . the mass of the vector constituent is taken to be @xmath12 .
the electromagnetic dirac and pauli form factors as well as the sachs form factors @xmath14 and @xmath15 have been calculated from the overlap of lfwfs .
the nucleon charge density has been obtained for the case of unpolarized nucleon in transverse coordinate space as well as in the impact parameter space .
further , a relation has been developed between the charge distribution of the quarks inside nucleon in the transverse coordinate space as well as in the impact parameter space particularly for the @xmath16 and @xmath17 quark distribution by using a fitted set of parameters .
it is found that density distribution in transverse coordinate space falls off slowly as compared to the charge distribution in impact parameter space but the magnitude of the peak decreases with the increasing value of the momentum fraction of the active quark .
this is primarily because of some contribution of the longitudinal momentum fraction @xmath0 in the transverse coordinate space whereas in the impact parameter space , the longitudinal momentum fraction @xmath0 can be fixed .
further as @xmath106 , the contribution from other partons is expected to be zero and the active quark carries all the momentum .
this investigation gives a deeper understanding of both the spatial and transverse distribution of the quark densities inside the nucleon .
we have also studied the anomalous magnetization density of the nucleon obtained from the spin - flip matrix element of the @xmath43 current .
the anomalous magnetization density vanishes when the angle between the direction of nucleon polarization and @xmath4 is zero as well as when @xmath51 .
it maxima occurs around @xmath88 and for higher values @xmath86 falls off very rapidly .
further , the calculations were extended to predict the results of the qcd transverse ads charge density inspired from the holographic qcd model and it is observed that for the ground state when the transverse distance between the constituent within the hadrons is minimum the density is maximum . for the first excited state , a peak for charge density is observed at a comparatively higher value of transverse distance .
even more secondary peaks will be observed for higher excited states .
authors acknowledge helpful discussions with s.j . brodsky .
hd would like to thank department of science and technology , government of india for financial support . if we have any two arbitrary functions @xmath107 and @xmath108 , their fourier transforms can be given as _
c(_)= d^2 _
e^- i _ _ ( _ ) , _
d(_)= d^2 _
e^- i _ _ ( _ ) .
the form factor can then be expressed as f(_)= _ c^*(_- _ ) _ d ( _ ) , which , in fourier conjugate space , becomes diagonal e^i _ _ f(_)= ^ * ( _ ) ( _ ) .
similarly , for g(_)= ^*(_- y _ ) ( _ ) , we have e^i _ _ g(_)= ^ * ( ) ( ) . 99 m. k. jones _
et al . _ ,
lett . * 84 * , 1398 ( 2000 ) .
et al . _ ,
lett . * 88 * , 092301 ( 2002 ) . v. p. druzhinin , arxiv:1311.7517 .
d. t. spayde _
b * 583 * , 79 ( 2004 ) .
f. e. maas _
et al . _ ,
lett * 93 * , 022002 ( 2004 ) .
k. a. aniol _
et al . _ ,
c * 69 * , 065501 ( 2004 ) . v. punjabi _
( jefferson lab hall a collaboration ) , phys .
c * 71 * , 055202 ( 2005 ) .
a. j. r. puckett _ et al .
* 104 * , 242301 ( 2010 ) .
j. c. bernauer _
( a1 collaboration ) , phys .
lett . * 105 * , 242001 ( 2010 ) .
( jefferson lab hall a collaboration ) , phys .
c * 84 * , 055204 ( 2011 ) .
x. zhan , k. allada , d. s. armstrong , j. arrington , w. bertozzi , w. boeglin , j. -p .
chen and k. chirapatpimol _
et al . _ ,
b * 705 * , 59 ( 2011 ) .
g. a. miller , annu .
nucl . part .
sci . * 60 * , 1 ( 2010 ) .
d. s. hwang , d. s. kim and j. kim , phy .
b * 669 * , 345 ( 2008 ) .
o. v. selyugin , o. v. teryaev , phys .
d * 79 * , 033003 ( 2009 ) .
m. diehl , phys .
rept . * 388 * , 41 ( 2003 ) .
x. ji , annu .
nucl . part .
sci * 54 * , 413 ( 2004 ) .
a. v. belitsky , a. v. radyushkin , phys . rept . * 418 * , 1 ( 2005 ) . m. diehl , eur .
c * 25 * , 223 ( 2002 ) .
m. burkardt , phys .
d * 62 * , 071503 ( 2000 ) .
r. manohar , a. mukherjee and d. chakrabarti , phys .
d * 83 * , 014004 ( 2011 ) .
m. burkardt , int . j. mod
a * 18 * , 173 ( 2003 ) .
d. chakrabarti , a. mukherjee , phys . rev .
d * 72 * 034013 ( 2005 ) .
s. j. brodsky , d. s. hwang and i. scmidt , phy .
b * 530 * , 99 ( 2002 ) .
m. burkardt , nucl .
phys . a * 735 * , 185 ( 2004 ) .
m. burkardt , d. s. hwang , phys . rev .
d * 69 * , 074032 ( 2004 ) .
m. burkardt , phys .
d * 66 * , 114005 ( 2002 ) .
d. chakrabarti , a. mukherjee , phys .
d * 71 * , 014038 ( 2005 ) .
d. e. soper , phys .
d * 15 * , 1141 ( 1977 ) .
g. a. miller , j. arrington , phys .
c , * 78 * , 032201 ( 2008 ) .
g. a. miller , phys .
lett . * 99 * , 112001 ( 2007 ) . c. e. carlson , m. vanderhaeghen , phys .
* 100 * , 032004 ( 2008 ) .
s. venkat , j. arrington , g. a. miller and x. zhan , phys .
c * 83 * , 015203 ( 2011 ) .
m. diehl , phys .
rev . * 126 * , 2256 ( 1962 ) .
j. j. kelly , phys . rev .
c * 70 * , 068202 ( 2004 ) .
z. abidin , c. e. carlson , phys .
d * 78 * , 071502 ( 2008 ) .
g. a. miller , e. piasetzky and g. ron , phys .
* 101 * , 082002 ( 2008 ) .
d. chakrabarti , c. mondal , eur .
phys . j. c * 74 * , 2962 ( 2014 ) .
s. j. brodsky , g. f. de teramond , phys .
lett . * 96 * , 201601 ( 2006 ) .
s. j. brodsky , s. d. drell , phys .
d * 22 * , 2236 ( 1980 ) ; g. p. lepage , s. j. brodsky , phys .
d * 22 * , 2157 ( 1980 ) .
s. j. brodsky , d. s. hwang , b. q. ma and i. schmidt , nucl .
b * 593 * , 311 ( 2001 ) .
d. chakrabarti , a. mukherjee , phys .
d * 71 * , 014038 ( 2005 ) ; d. chakrabarti , a. mukherjee , phys . rev .
d * 72 * , 034013 ( 2005 ) . m. guidal , m. v. polyakov , a. v. radyushkin and m. vanderhaeghen , phys .
d * 72 * , 054013 ( 2005 ) .
j. arrington , c. d. roberts and j. m. zanotti , j. phys .
g * 34 * , 523 ( 2007 ) . c. f. perdrisat , v. punjabi and m. vanderhaeghen , prog . part .
nucl . phys .
* 59 * , 694 ( 2007 ) . c. adlloff
_ et al . _
( h1 collaboration ) , phys .
b * 517 * , 47 ( 2001 ) ; s. chekanov _
( zeus collaboration ) , phys .
b * 573 * , 46 ( 2003 ) ; a. airpetian _
( hermes collaboration ) , phys .
* 87 * , 182001 ( 2001 ) ; s. stepanyan _
( clas collaboration ) , phys .
* 87 * , 182002 ( 2001 ) ; n. dhose , e. burtin , p.a.m .
guichon , j. marroncle ( compass collaboration ) , eur .
j. a * 19 * , 47 ( 2004 ) .
r. jakob , p. j. mulders and j. rodrigues , nucl .
a * 626 * , 937 ( 1997 ) .
p. v. pobylitsa , phys .
d * 66 * , 094002 ( 2002 ) .
s. j. brodsky , d. chakrabarti , a. harindranath , a. mukherjee and j. p. vary , phy . lett .
b * 641 * , 440 ( 2006 ) ; s. j. brodsky , d. chakrabarti , a. harindranath , a. mukherjee and j. p. vary , phys . rev .
d * 75 * , 014003 ( 2007 ) .
m. carmignotto , t. horn and g. a. miller , phys .
c * 90 * , 025211 ( 2014 ) .
et al . _ , approved jefferson lab experiment e12 - 06 - 101 .
et al . _ , approved jefferson lab experiment e12 - 07 - 105 .
j. j. kelly , phys .
c * 66 * , 065203 ( 2002 ) .
g. f. de teramond , s. j. brodsky , phys .
lett . * 102 * , 081601 ( 2009 ) .
g. f. de teramond , s. j. brodsky , phys .
* 94 * , 201601 ( 2005 ) .
s. j. brodsky , g. f. de teramond , h. g. dosch and j. erlich , arxiv:1407.8131 . | electromagnetic form factors obtained from the overlap of light front wave functions ( lfwfs ) have been used to study the transverse densities of charge and magnetization .
the calculations have been carried out to develop a relation between the charge distribution of the quarks inside nucleon in the transverse coordinate space as well as in the impact parameter space .
when a comparison is carried out , it is found that the transverse distribution in the impact parameter space , where the longitudinal momentum fraction @xmath0 can be fixed , falls off faster than the spatial distribution in the transverse coordinate space where there is some contribution from the longitudinal momentum as well .
the anomalous magnetization density of the nucleon has also been discussed .
further , we have also presented the results of the qcd transverse ads charge density inspired from the holographic qcd model . |
optical modes with complex structures have attracted increasing attention in recent years .
the subject comprises very different issues due to the variety and versatility of higher - order modes , i.e. modes with complex intensity and phase patterns in their cross - sections .
very prominent examples are light beams carrying orbital angular momentum ( oam ) , which is a direct consequence of a helical phase front of the beam @xcite
. a set of solutions to the paraxial wave equation exhibiting oam are so - called laguerre - gaussian ( lg ) beams .
oam has also been found to be an interesting additional degree of freedom for communication tasks that could help to meet increasing capacity demands @xcite .
complex structures may not only be found in intensity and phase , but also in the polarization structure of optical beams .
such vector beams have interesting properties that can be utilized in diverse applications @xcite .
they may exhibit extraordinary structures that are non - separable and show intriguing properties similar to entangled quantum systems @xcite that can be exploited for practical tasks , such as high - speed kinematic sensing @xcite .
they also have non - trivial focusing properties , allowing , for instance , to focus light to a tighter spot compared to homogeneously polarized modes @xcite .
tightly focused and polarization - tailored light beams have also proven to be versatile tools in nano - optics and plasmonics for studying the optical properties of individual nanostructures @xcite .
moreover , very efficient light - atom coupling can be realised by shaping the input field into a radially polarized mode @xcite .
when quadrature squeezed , vector beams represent a natural system for the study of hybrid quantum entanglement between the polarization and transverse spatial degrees of freedom @xcite .
higher - order vector beams may also serve as a building - block to perform interesting quantum computational tasks .
an implementation of a cluster state generation using vector beams that may play essential roles in one - way quantum computation protocols is presented in @xcite . in general
, it is surprising how little quantum features of vector beams have been studied up to now .
fundamental properties and aspects of the theoretical description in terms of quantum mechanics of these beams are still being discovered .
considering the vast range of possibilities for applications in this domain , it becomes clear that it is a worthwhile topic to explore .
recently , we reported on a scheme for the experimental generation of squeezed spatial modes @xcite .
that work was concerned with scalar laguerre - gaussian and bessel beams only . here
, we want to go a step further and prepare amplitude squeezed vector beams . in a practical setting ,
several specific requirements have to be considered for the preparation of these modes . the option to switch flexibly between different kinds of high - order modes , or different orders of one type of mode ,
is highly desirable .
an efficient technique using q - plates for the generation of vector beams @xcite unfortunately can not offer this flexibility .
the fiber - based technique used in @xcite is also not applicable as the squeezing was generated in a nonlinear medium supporting only a first - order subset of vector beams .
a viable strategy is to perform the mode conversion after the nonlinear process .
this technique has been realized previously @xcite , but a mode - converter was utilized which entails similar limitations as in @xcite . converting the mode after generating the squeezing , makes us face another serious restriction regarding losses which must be kept low . a commonly used tools for performing arbitrary mode conversions are spatial light modulators ( slm ) . given the currently still high intrinsic losses of such devices , restricting the losses to a low level is not a simple task . in @xcite , an experimental setup is presented with which higher - order vector beams of a very high quality have been generated interferometrically .
but even without considering the technical losses caused by the slm , the system - inherent losses amount to 75% .
also in the scheme presented in @xcite , a good mode quality comes at the price of losses .
these high losses make those interferometric methods unsuitable for preserving squeezing during the mode conversion process .
furthermore , for the sake of stability and simplicity of alignment , a setup is preferable in which an interferometer interfering spatially separated beams is avoided . in this paper
, we present a scheme that meets all these requirements .
it allows almost arbitrary vector beams to be generated from one setup .
apart from this flexibility , the particular double reflection technique involving a spatial light modulator avoids intrinsic losses as well as stability concerns as associated with a phase locking technique .
in the first step of the experimental procedure , the amplitude of a linearly polarized fundamental gaussian mode is squeezed . in our experiment , this is achieved by exploiting the optical kerr effect in a standard fiber @xcite . in a second step , the mode conversion from the fundamental to higher - order modes is performed via a double reflection technique involving a spatial light modulator .
the resulting higher - order vector beams are analyzed in terms of their intensity profiles , the spatial distribution of the stokes parameters and their level of amplitude squeezing .
the corresponding experimental setup is sketched in fig .
[ setup ] . as a light source , an ultrashort pulsed laser (
origami onefive ) producing 220fs pulses with a central wavelength of 1560 nm is utilized .
a sagnac interferometer @xcite employing a polarization - maintaining single - moder fiber ( fs pm 7811 by 3 m ) is used to amplitude squeeze the beam . here
, a 90:10 beam splitter is used to couple a strong and a weak light pulse into the two opposite ends of the fiber . due to the third - order nonlinear kerr effect , the strong pulse is quadrature squeezed .
the two counter - propagating pulses interfere , and the quadrature squeezing is transformed into squeezing of the amplitude quadrature , if the input powers are adjusted in a suitable way . at the output of the sagnac loop
, we measure a quantum noise reduction in the amplitude squeezed beam of up to @xmath1 by direct detection at a radio frequency sideband at 10mhz ( resolution bandwidth : 300khz , video bandwidth : 3khz ) .
the 461 data points of one trace measured by the spectrum analyzer have been divided in chunks of 15 data points .
the error given for the squeezing is the standard deviation of the squeezing determined for these individual blocks .
the prepared amplitude - squeezed fundamental gaussian mode is then converted into the desired higher - order mode . for this mode conversion ,
the input beam undergoes two separate modulations which are performed subsequently .
we use a reflective , phase - only liquid - crystal - based slm ( pluto holoeye , lcos , 1920x1080 pixel ) and split its display into two halves . on each half , we display individually tailored phase masks .
the incoming beam is reflected from and modulated by the first pattern , before impinging on the second one ( see fig.[setup ] ) . due to the specific properties of the slm ,
only the horizontally polarized part of the light is modulated .
this polarization dependence is due to the elongated shape of the liquid crystal molecules and their birefringence .
we choose the state of polarization of the incoming light beam to be 45@xmath2 relative to the slm .
the vertically polarized component of the light remains unaffected while the horizontally polarized component is transformed .
after the first reflection , a half - wave plate rotates the states of polarization by 90@xmath2 , the modulated part is now vertically polarized , the unmodulated part horizontally polarized .
the second reflection on the other half of the slm thus causes the conversion of the so far unmodulated part of the light .
the horizontally and vertically polarized components of the beam have now been reflected on different phase patterns which are tailored to produce the two different modes ( see fig.[setup ] ) .
we have chosen to work with lg modes .
thus , two copropagating lg - modes with opposite azimuthal indices ( opposite phase charge ) and opposite handedness ( circular polarization ) are generated .
after passing these copropagating modes through a quarter wave - plate their interference results in the desired vector beams ( see @xcite for details ) .
this method utilizing an lg - mode instead of hermite - gaussian ( hg ) mode basis turned out to be preferable .
we achieved a better mode quality with a simpler phase pattern .
this collinear common - path interferometer does not require phase - locking , as the prepared modes are co - propagating on the same beam path , they are naturally superimposed .
the interference of these two modes leads to the desired polarization patterns within the beam profile , depending on the phase patterns displayed on the slm . to improve the mode quality
, we additionally added the phase pattern of a kinoform lens @xcite , and used combined patterns as illustrated in fig.[setup ] .
the kinoform lens is a phase hologram that introduces a phase modulation equivalent to the one that would be caused by a real lens .
fig.[figlg01 ] and fig.[figccd ] show the spatial images of the obtained modes .
consistent with the expected losses in the conversion setup , the amount of squeezing is reduced .
we measure a quantum noise reduction of up to @xmath3 in the converted modes . as in @xcite , no excess noise
is added by the slm at the measured sideband frequency .
the overall losses in the setup amount to 64% .
it should be noted that these losses are not approach - inherent but solely due to the inefficiencies of some components , in particular of the slm .
we analyze the generated higher - order vector beams with regard to their mode structure and their quantum noise reduction in the amplitude quadrature . to this end
, we measure the amplitude squeezing by direct detection with the help of a spectrum analyzer and a coherent beam serving as a reference level .
the mode structure is studied by measuring the spatial intensity and polarization distributions in the beam cross - section .
the total intensity profile is observed directly by using a ccd camera . a quick , yet quite precise check of the polarization structure is realized by inserting a rotatably mounted polarizer and taking images for different orientations of the polarization axis .
a much more detailed analysis is the measurement of the spatial distribution of the stokes parameters that we performed using a technique based on the method described in @xcite . with this technique ,
the beam profile after a rotatable quarter - wave plate and a polarizing beam splitter is determined by a ccd camera . from the images taken for different orientations of the quarter - wave plate ,
the stokes parameters may be calculated using a fourier algorithm described in @xcite . in this way
, we obtain the polarization state for each pixel of the camera picture and , therefore , the spatial vectorial structure of the generated modes .
as a proof - of - principle , we report in this manuscript the generation of laguerre - gauss modes with radial indices up to 1 and azimuthal indices up to 3 , which have sophisticated polarization structures and show a quantum noise reduction in the amplitude of up to -0.9db@xmath4db . in fig.[figlg01 ]
, we illustrate our measured results and the theoretical expectation for a radially polarized lg@xmath5 mode . in our nomenclature
a lg@xmath6 mode is a laguerre gauss mode with the radial index @xmath7 and the azimuthal index @xmath8 .
thus , here , the radial index is equal to 0 , and the azimuthal index is 1 .
the first image in fig.[figlg01]a ) is the intensity pattern of the experimentally generated mode .
the following images depict the intensity patterns for a polarizer being inserted in front of the camera , orientated in the direction indicated by the white arrows .
fig.[figlg01]b ) shows the spatially resolved stokes parameters of the experimental mode , fig.[figlg01]c ) depicts the theoretical expectations .
we verified @xmath3 of amplitude squeezing in this mode .
we encounter only small deviations , e.g. slightly imbalanced lobes for the stokes parameter @xmath9 and a nonzero @xmath10 ( for discussion see sec.[sec_discussion ] ) .
fig.[figstokes ] shows our experimental results for lg@xmath11 , lg@xmath12 , lg@xmath13 , lg@xmath14 and lg@xmath15 modes with complex polarization patterns , the achieved quantum noise reduction is written directly below the images .
the mode quality is high and the amplitude squeezing is significant for all of them . for the modes with two rings ,
i.e. for lg@xmath13 , lg@xmath14 and lg@xmath15 , one can see that the polarization in the two rings is slightly twisted with respect to each other .
this is in accordance with the spiral structures that can be noticed in the distribution of the stokes observables ( sec.4 ) .
the effect increases with the complexity of the mode .
its cause is discussed in sec.[sec_discussion ] . apart from the spatial distributions of the stokes parameters in fig.[figstokes ]
, we also present the images from the direct observation of the modes with a ccd camera behind a rotatable polarizer in fig.[figccd ] .
they provide a depiction in a different manner that may give a more intuitive impression about some aspects of the mode quality .
both , fig.[figstokes ] and fig[figccd ] demonstrate that our setup permits the flexible generation of different amplitude squeezed higher - order laguerre - gauss modes of good quality .
in the practical implementation of the mode preparation , special care has to be devoted to the overlap of the two orthogonally polarized modes stemming from the reflections on the two individual segments of the slm .
since each polarization component interacts with the slm in a different plane , the phase patterns imposed on each half must be carefully tweaked to ensure that both components emerge from the setup having the same width and divergence .
the generation of the modes becomes more difficult the higher their order , for the simple reason that the higher - orders possess more complicated radial structures that need to overlap properly .
since the two polarization components originate from two modes propagating within one beam , the necessity of a phase locking technique is avoided altogether .
we want to note that this advantage comes at a price : the mode quality is reduced by the imperfect diffraction efficiency of the slm .
the slm modulates only a part of the impinging horizontally polarized light , the unmodulated part of the light is just reflected .
when a polarizer is inserted for the mode analysis , a particular linear polarization is chosen . in this way , one particular higher - order mode and the equally polarized part of the zeroth - order gaussian mode is picked .
the interference of these modes causes characteristic spiral structures .
the appearance of these patterns can not be avoided as the separation of the zeroth - order is not possible in this low - loss setup ( see sec.[modeprep ] ) and remain an undesired side effect especially for higher orders .
the characteristic spirally curved patterns are particularly visible in fig.[figccd]b ) .
studying the stokes parameters shows that circularly polarized components are present , resulting in non - zero @xmath10 components ( see fig.[figlg01 ] and fig.[figstokes ] ) .
our technique requires the presence of the unmodulated vertically polarized light and thus the remaining fundamental gaussian mode can not be separated by adding a blazed grating to the phase picture , for example , as it is very common in other experimental schemes . in @xcite , this procedure can be applied and the mode quality is enhanced in this way . however , the system inherent losses render this approach not applicable for our purposes as this would considerably degrade the amount of squeezing . in other cases , superimposing a fundamental gaussian mode with a higher - order mode
is intentionally exploited to analyze the phase structure of the higher - order mode in question , see @xcite .
in this work , we present a flexible technique for the experimental generation of squeezed higher - order modes with nonuniform polarization patterns . we show that it is possible to minimize losses in a setup employing spatial light modulators .
the amplitude squeezing initially generated by a nonlinear fiber - sagnac interferometer was reduced during the mode conversion process due to inevitable losses , but remained significant in the higher - order laguerre - gauss modes that we studied , reaching values of up to @xmath3 .
the detailed analysis of our generated modes shows that we achieve a high mode quality .
we discuss degradations of intensity profile of the modes , especially important at higher orders , due to a remaining fundamental mode . a conceivable possibility to further improve the mode generation could be the combination of our method with the mode conversion technique described in @xcite .
the repeated reflection on a deformable mirror allows to generate modes with an extremely high mode quality .
unfortunately , the multiple reflections also cause a significant increase of losses such that a trade - off between mode quality and acceptable losses has to be found .
another strategy to enhance the mode quality could be to maintain our present scheme using an slm , but further optimize the phase patterns displayed on the slm @xcite . as a proof - of - principle
, the results presented in this article can serve as a basis to study continuous - variable quantum states of arbitrary vector modes and their use in quantum processing and metrology . | we present an experimental method for the generation of amplitude squeezed high - order vector beams .
the light is modified twice by a spatial light modulator such that the vector beam is created by means of a collinear interferometric technique .
a major advantage of this approach is that it avoids systematic losses , which are detrimental as they cause decoherence in continuous - variable quantum systems .
the utilisation of a spatial light modulator ( slm ) gives the flexibility to switch between arbitrary mode orders . the conversion efficiency with our setup is only limited by the efficiency of the slm .
we show the experimental generation of laguerre - gauss ( lg ) modes with radial indices up to 1 and azimuthal indices up to 3 with complex polarization structures and a quantum noise reduction up to [email protected] .
the corresponding polarization structures are studied in detail by measuring the spatial distribution of the stokes parameters .
99 l. allen , m. w. beijersbergen , r. j. c. spreeuw , and j. p. woerdman , orbital angular momentum of light and the transformation of laguerre - gaussian laser modes , phys .
rev . a * 45 , * 8185 ( 1992 ) .
j. wang , j. yang , i. fazal , n. ahmed , y. yan , h. huang , y. ren , y , yue , s. dolinar , m. tur , a. willner , terabit free - space data transmission employing orbital angular momentum multiplexing , nature photonics * 6 , * 488496 ( 2012 ) .
q. zhan , cylindrical vector beams : from mathematical concepts to applications , adv .
. photon .
* 1 , * 157 ( 2009 ) .
a. aiello , f. tppel , ch .
marquardt , e. giacobino , g. leuchs , quantum - like nonseparable structures in optical beams , new j. phys .
* 17 , * 043024 ( 2015 ) .
f. tppel , a. aiello , ch .
marquardt , e. giacobino and g. leuchs , classical entanglement in polarization metrology , new j. phys .
* 16 , * 073019 ( 2014 ) .
s. berg - johansen , f. tppel , b. stiller , p. banzer , m. ornigotti , e. giacobino , g. leuchs , a. aiello , and ch .
marquardt , classically entangled optical beams for high - speed kinematic sensing , optica * 2*(10 ) , 864 ( 2015 ) .
s. quabis , r. dorn , m. eberler , o. glckl , g. leuchs , focusing light to a tighter spot , opt . commun .
* 179 , * 17 ( 2000 ) .
r. dorn , s. quabis , g. leuchs , sharper focus for a radially polarized light beam , phys .
rev .
lett .
* 91 , * 233901 ( 2003 ) .
k. youngworth , t. brown , focusing of high numerical aperture cylindrical - vector beams , opt .
express * 7*(2 ) , 7787 ( 2000 ) .
j. mller , p. banzer , s. quabis , u. peschel and g. leuchs , waveguide properties of single subwavelength holes demonstrated with radially and azimuthally polarized light , appl .
phys .
b * 89*(4 ) , 517 - 520 ( 2007 ) . t. zchner , a. v. failla , a. hartschuh , and a. j. meixner , a novel approach to detect and characterize the scattering patterns of single au nanoparticles using confocal microscopy , j. microscopy * 229 * , 337343 ( 2008 ) .
p. wozniak , p. banzer and g. leuchs , selective switching of individual multipole resonances in single dielectric nanoparticles , laser photon . rev .
* 9*(2 ) , 231 - 240 ( 2015 ) .
m. sondermann , r. maiwald , h. kronermann , n. lindlein , u. peschel , g. leuchs , design of a mode converter for efficient light - atom coupling in free space , appl .
phys .
b * 89 , * 489492 ( 2007 ) . c. gabriel , a. aiello , w. zhong , t.g .
euser , n.y .
joly , p. banzer , m. frtsch , d. elser , u.l .
andersen , ch .
marquardt , p.st.j .
russell , g. leuchs , entangling different degrees of freedom by quadrature squeezing cylindrically polarized modes , phys .
rev .
lett .
* 106 , * 060502 ( 2011 ) .
i. rigas , c. gabriel , s. berg - johansen , a. aiello , p. van loock , u.l .
andersen , ch .
marquardt , g.leuchs , compact generation of easy - to - access continuous - variable cluster states , arxiv:1210.5188v3 ( 2012 ) .
m. semmler , s. berg - johansen , c. gabriel , p. banzer , a. aiello , ch .
marquardt , g. leuchs , experimental generation of squeezing in arbitrary spatial modes using a spatial light modulator , arxiv:1601.03961 ( 2016 ) .
f. cardano , e. karimi , s. slussarenko , l. marrucci , c. de lisio , e. santamato , polarization pattern of vector vortex beams generated by q - plates with different topological charges , appl .
opt . *
51 * , c1c6 ( 2012 ) . c. gabriel , a. aiello , s. berg - johansen , ch .
marquardt , g. leuchs , tools for detecting entanglement between different degrees of freedom in quadrature squeezed cylindrically polarized modes , eur .
phys .
j. d * 66 , * 172 ( 2012 ) . c. maurer , a. jesacher , s. frhapter , s. bernet , m. ritsch - marte , tailoring of arbitrary optical vector beams ,
new j. phys .
* 9 , * 78 ( 2007 ) . w. han , y. yang , w. cheng , q. zhan , vectorial optical field generator for the creation of arbitrarily complex fields , opt .
express * 21*(18 ) , 2069220706 ( 2013 ) .
m. lassen , g. leuchs , u.l .
andersen , continuous variable entanglement and squeezing of orbital angular momentum states , phys .
rev .
lett .
* 102 , * 163602 ( 2009 ) .
s. schmitt , j. ficker , m. wolff , f. knig , a. sizmann , g. leuchs , photon - number squeezed solitons from an asymmetric fiber - optic sagnac interferometer , phys .
rev .
lett .
* 81 , * 2446 ( 1998 ) .
l. b. lesem , p. m. hirsch , j. a. jordan , the kinoform : a new wavefront reconstruction device , ibm journal of research and development * 13*(2 ) , 150155 ( 1969 ) . b. schaefer , e. collett , r. smyth , d. barrett , b. fraher , `` measuring the stokes polarization parameters , '' am . j. phys . * 75 , * 163168 ( 2007 ) .
h. xum , l. yang , conversion of orbital angular momentum of light in chiral fiber gratings , opt . lett . * 38*(11 ) , 19781980 ( 2013 ) . j. morizur , l. nicholls , p. jian , s. armstrong , n. treps , b. hage , m. hsu , w. bowen , j. janousek , h. bachor , programmable unitary spatial mode manipulation , j. opt
. soc
. am . *
27*(11 ) , 25242531 ( 2010 ) .
e. bolduc , n. bent , e. santamato , e. karimi , and r. w. boyd , exact solution to simultaneous intensity and phase encryption with a single phase - only hologram , opt .
lett .
* 38 , * 35463549 ( 2013 ) . |
the cosmic baryon abundance was first inferred by applying the theory of primordial nucleosynthesis to observations of light element abundances ( e.g. @xcite ) and has been confirmed by observations of the cosmic microwave background radiation ( cmb ) .
a recent measurement from the wilkinson microwave anisotropy probe reveals that @xmath3 , where the hubble parameter is @xmath4 @xcite . at @xmath5
most of the baryons are thought to reside in the diffuse , photoionised intergalactic medium ( igm ) with @xmath6 k , traced by the ly@xmath7 forest forest observations is inversely proportional to the assumed ionisation rate .
since independent constraints on the uv intensity are only accurate to a factor of few at best , we can not prove that most of the baryons reside in the forest . ]
( e.g. @xcite ; @xcite ; @xcite ) .
the gas traced by the ly@xmath7 forest , together with the mass in galaxies , fully accounts for the cosmic baryon abundance at @xmath8 and confirms the prediction of nucleosynthesis theories and the cmb measurements .
however , when the mass of stars , interstellar gas and plasma in clusters of galaxies at redshift @xmath0 is considered , only a small fraction of the mass budget can be accounted for ( e.g. @xcite ; @xcite ; @xcite ) .
while the gas traced by ly@xmath7 forest may account for about a third of the low-@xmath9 baryons ( e.g. @xcite ) , the majority of the baryons remain invisible . unless the baryon budget at high redshift has been overestimated by two different measurements
, a large amount of the low redshift baryons must be `` missing '' .
one of the basic predictions of cosmological , gas - dynamical simulations is the distribution of baryons in the universe .
how many baryons are locked in stars ? how many reside in clusters and filaments ? what is the state of the diffuse gas ?
simulations of structure formation predict that the missing baryons problem is measure of our technological limitations rather than a real problem .
in other words , the baryons are out there , but we can not see them because they reside in gas that is too dilute to be detected in emission and too hot to be traced by ly@xmath7 absorption .
the idea of gravitational heating of the intergalactic gas was first suggested by @xcite and subsequently developed ( e.g. @xcite ; @xcite ; @xcite ) . while the largest structures in the universe form , the igm is heated by gravitational shocks that efficiently propagate from the collapsing regions to the surrounding medium .
simulations predict that gas compressed and heated by shocks can reach temperatures of @xmath10 k in rich clusters of galaxies , while filaments and mildly overdense regions are heated to temperatures in the range @xmath11 to @xmath12 k ( e.g. @xcite ; @xcite ; @xcite ) .
the latter are commonly known as the warm - hot intergalactic medium ( whim ) , which represents about half the total baryonic mass in the universe at @xmath0 . at @xmath13
k , the igm is collisionally ionised and becomes transparent to ly@xmath7 radiation . as a consequence , the h ly@xmath7 forest does not trace the bulk of the gas mass at low redshift . although the shock - heated igm emits radiation in the uv and in the soft x - ray bands , the total energy radiated away by mildly overdense gas is orders of magnitude too small to be detected by current instruments .
similarly , the absorption from such a gas along the line of sight to a bright x - ray source is too weak to be resolved by current spectrographs ( @xcite - chapter 3 , this volume ) . in this paper
, we review the predictions of numerical simulations for the properties and the observability of the whim . in sect .
[ origin ] we briefly mention the numerical techniques used to simulate the igm and we discuss the mechanisms that produce the whim , namely heating by gravitational accretion shocks and other , non - gravitational , heating processes . a more detailed description of the numerical techniques themselves can be found in @xcite - chapter 12 , this volume .
[ background ] describes the contribution of the whim emission to the soft x - ray background .
the current predictions for the detectability of the whim in emission and absorption are discussed in sect .
[ emission ] and in sect . [ absorption ] respectively .
[ sz ] briefly reports on the effect of the whim on the cmb radiation and sect .
[ bfield ] discusses the magnetisation of the igm .
we draw our conclusions in sect .
[ conclusion ] .
in this subsection , we briefly discuss the numerical techniques that have been used to investigate the igm and the whim in particular . however , we refer the reader to the review of @xcite - chapter 12 , this volume for a thorough description of specific simulation techniques .
the most common approach to model the whim is through cosmological hydrodynamical simulations .
the strength of hydrodynamical simulations is that the basic gas dynamical processes that determine the evolution of the diffuse gas are directly simulated and that it is possible to make virtual observations of the whim with existing or future instruments .
however , _
ad hoc _ empirical prescriptions are needed to include processes such as galactic winds and feedback from active galactic nuclei ( agn ) , whose physics is still poorly understood and in any case would require a resolution far beyond the current standards to be included self - consistently .
hydrodynamical simulations of the whim have typically used either smoothed particle hydrodynamics ( sph , e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , a uniform grid ( e.g. @xcite ; @xcite ) , or adaptive mesh refinement ( amr , e.g. @xcite ) to solve the gas dynamics .
several properties of the whim , including the impact of gravitational and non - gravitational heating processes on the igm , and the observability of emission lines and absorption systems in uv and x - ray spectra , have been successfully investigated with analytical and semi - analytical techniques ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . 3 , 2 , 1 , and 0 .
the gas mostly resides in three phases : a cool , photoionised , diffuse igm ( tail to lower left ) ; the warm - hot shocked igm ; and gas condensed into stars and star forming clouds ( tail to lower right ) .
the solid lines indicate an approximate separation between these phases . from @xcite.,scaledwidth=90.0% ] theoretical predictions and numerical simulations agree that the heating of the igm is fuelled by the gravitational accretion shocks emerging during the collapse of cosmic structures ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite - chapter 7 , this volume ) .
simulations suggest that the distribution of the warm - hot gas closely reflects the distribution of the shock waves , and therefore the underlying density field .
the role of shock waves in heating the igm has been investigated using cosmological hydrodynamical simulations by @xcite , who find that the fastest shocks can be found around clusters and groups of galaxies , while the slowest shocks usually propagate across low density regions like sheets and filaments .
this reflects the fact that the infall velocity is a measure of the depth of the potential well .
@xcite reach similar conclusions using analytical techniques and find that the shocks with the highest temperatures are associated with the most massive virialised objects , up to distances larger than their virial radii .
the whim gas is organised in a complex structure of filaments and sheets , that generally reflects the dynamics of structure formation . according to @xcite ,
the gas density in filaments is typically @xmath14 times the mean density and varies more smoothly than in clusters .
filaments have a coherence length of about 5 mpc , although some can be as long as 25 mpc , and a diameter of about @xmath15 mpc .
the detailed thermal structure of the whim strongly depends on the direction of propagation of the accretion shocks , while the velocity field appears to be orthogonal to the filaments at large distances from galaxy clusters and aligned to the filaments at small distances .
most simulations predict that between 30 and 50 per cent of all the baryons in the low redshift universe have temperatures in the interval @xmath16 k ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
[ dave ] shows the distribution of the gas in the temperature - density plane at @xmath173 , 2 , 1 , and 0 .
hot gas with @xmath18 k resides in the hottest cores of groups and in large clusters of galaxies .
most of the gas , however , is at temperatures @xmath19 k and at @xmath20 is about evenly spread between the whim and the cooler , mostly photo - ionised igm . at higher redshifts ,
a larger fraction of the baryons resides in the photo - ionised igm .
@xcite find that , besides the gas shock - heated to @xmath21 k , a significant amount of mass , distributed mostly along sheet - like structures , is shock - heated to @xmath22 k by shock waves propagating with velocities @xmath23 km s@xmath24 .
this mostly photo - ionised gas is responsible for the majority of the o and o absorption lines , but contributes little to the total o and o emission , which is instead produced by warmer , collisionally ionised ions .
simulations that include non - equilibrium processes have been performed by @xcite , @xcite and @xcite . in most hydrodynamical simulations , it is routinely assumed that diffuse gas is in ionisation equilibrium ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . in general ,
ionisation equilibrium is nearly achieved in the centres of galaxy clusters and in photo - ionised regions with @xmath25 k. however , this may not happen in the outer regions of clusters , in groups , and in the whim @xcite .
similar conclusions are reached by @xcite and @xcite .
@xcite claim that observational results for the abundance of o absorption lines are better reproduced by models including non - equilibrium processes .
@xcite find that a significant fraction of the whim at @xmath20 is not in ionisation equilibrium , as do @xcite .
however , @xcite calculate that although this strongly affects the relative abundance of ions such as o , o and o , it does not have a significant effect on the observable signatures of the whim in emission or absorption .
when gas is infalling onto a collapsing structure and is heated by gravitational accretion shocks , most of its energy is carried by ions , while electrons can only gain thermal energy through collisions with ions .
the temperatures of ions and electrons can therefore be very different and converge only if energy can be efficiently transferred from ions to electrons .
if only coulomb interactions are considered , then the relaxation timescale can be shorter than the age of the universe for the whim ( @xcite ; @xcite ) .
since the relative abundance of metal ions and the plasma emissivity are sensitive to the electron temperature , this would have consequences for the observability of the whim .
k. from @xcite.,width=377 ] @xcite introduce a two - temperature model for the thermal evolution of an ionised gas .
their model includes an explicit treatment for collisional relaxation processes between electrons and ions and allows one to follow the temperature of free electrons separately from the temperature of ions .
they find that at @xmath20 most whim gas shows a two - temperature structure .
[ yoshida2005 ] shows the electron temperature in the two - temperature model compared to the mean temperature of the gas at @xmath20 for the simulations of @xcite . clearly , the electron temperature of the whim is significantly lower than the mean gas temperature , when non - equilibrium processes are taken into account .
@xcite calculate that the relative abundance of o , o and o predicted by a two - temperature model can be about an order of magnitude different from that predicted by a one - temperature model .
this yields similarly different values for the line emission of ions .
however , from supernova remnants we know that plasma waves are typically much more efficient than collisional processes in transferring energy from ions to electrons ( e.g. @xcite ) and these effects have not yet been included in simulations of the whim .
@xcite may therefore have overestimated the importance of the two - temperature structure of the post - shock gas .
several studies have investigated whether sources of non - gravitational heating might significantly affect the thermal state of the igm .
simulations by @xcite predict that non - gravitational processes , such as galactic winds and x - ray emission from galaxies and quasars , contribute no more than 20 per cent of the energy required to heat the igm , in agreement with previous results from @xcite and @xcite . a powerful test to constrain the contribution of non - gravitational processes is the comparison of the observed intensity of the soft x - ray background to the predictions of numerical simulations .
since most of the observed background emission has been resolved into point sources ( @xcite ; @xcite ) , the comparison of the predictions of hydrodynamical simulations with observations can put tight constraints on the amount of the observed background emission that can be attributed to the diffuse intergalactic gas .
in general , the intensity of the soft x - ray background predicted by simulations that include only gravitational heating processes strongly exceeds the observed emission from high density , collapsed regions ( e.g. @xcite ; @xcite ; @xcite ; @xcite ) .
the emission is primarily due to bremsstrahlung and reflects the cooling of gas in groups and clusters of galaxies .
analytical calculations by @xcite and @xcite find an order of magnitude more background light than predicted by simulations , owing to the importance of gravitational pre - heating .
they suggest that additional heating from non - gravitational processes might be necessary to `` puff up '' and hence decrease the emission from collapsed regions and reconcile the theoretical predictions with the observations . however , they note that non - gravitational heating does not seem to be relevant for uncollapsed regions like filaments and voids .
@xcite argue that non - gravitational processes can help to establish an entropy floor in high density regions .
the lowest entropy material will drop out and form stars unless it is heated through feedback from star formation .
they suggest that an entropy floor of @xmath26 kev @xmath27 is required to reconcile the theoretical results with observations . @xcite
demonstrate further that radiative cooling and additional heating from feedback can efficiently suppress the x - ray background , while substantially modifying the characteristic spectral shape of the background itself .
the extended spatial distribution over cosmological scales makes the whim an essential contributor to the diffuse x - ray background radiation .
the most intense x - ray emission comes from relatively compact sources such as clusters , active galactic nuclei and galaxies .
the gas in these objects has high emissivity power , but accounts for only a small fraction of the baryon mass in the universe .
the emissivity of the whim is usually a few orders of magnitude lower , because of its lower density and temperature .
however , the whim accounts for up to half the baryonic mass at @xmath0 and its integrated emission over cosmological scales might be roughly comparable to that of dense regions and compact objects .
analytical calculations @xcite and numerical simulations predict that the whim contributes between a few per cent @xcite and 40 per cent ( @xcite ; @xcite ) of the total x - ray background emission , consistent with the observational upper limit of @xmath28 ergs@xmath24@xmath27deg@xmath29 @xcite .
because of the softness of the spectrum , the whim produces only between 4 per cent @xcite and 10 per cent @xcite of the total hard x - ray emission . according to the simulations of @xcite ,
about 90 per cent of the x - ray background emission is produced at @xmath30 .
[ roncarelli ] shows simulated maps of the soft ( @xmath31 kev , upper panels ) and hard ( @xmath32 kev , lower panels ) x - ray emission from a region 3.78 degree on a side .
the left panels show the total emission from the gas , while the right panels show the contribution of the whim . assumed in the simulations or to the underestimate of the observed agn formation rate used to compute the spectrum . from @xcite.,scaledwidth=90.0% ] @xcite estimate that in hydrodynamical simulations the whim gas at @xmath20 is responsible for about @xmath33 per cent of the total extragalactic surface brightness in the energy range @xmath31 kev ( fig .
[ phillips ] ) , corresponding to a flux of @xmath34 ergs@xmath24@xmath27deg@xmath29 .
the whim line emission peaks in the @xmath35 kev energy range , where observations by _ asca _ and _ rosat _ show a spectral bump , as shown in fig .
[ phillips ] .
@xcite argue that the bump is due to whim emission , and in particular to metal line emission . similarly , @xcite produce simulations that faithfully reconstruct the local universe and find that about @xmath36 per cent of the x - ray background at energies of about 1 kev may be contributed by the whim .
@xcite find a contribution of up to 20 per cent in the energy range @xmath37 kev , most of which is emitted by filaments at redshifts between 0.1 and 0.6 .
dense and bright filaments are present in about 10 per cent of a generic field of view , and can account for more than 20 per cent of the total emission .
in such cases , redshifted emission lines from highly ionised atoms should be clearly detectable .
soft x - ray excess radiation in clusters has been detected in observations by the extreme ultraviolet explorer ( _ euve _ ) , _ rosat _ and _ xmm - newton _ ( @xcite ; @xcite ; @xcite ; see @xcite - chapter 4 , this volume for an overview ) .
the temperature of the icm is usually estimated by assuming a one - temperature plasma model for the diffuse gas . the soft x - ray flux calculated from the observed x - ray emission of clusters
, however , often appears to be above the expected thermal contribution .
it has been suggested that the excess radiation may be due to foreground emission from the whim , or from warm diffuse gas in the clusters .
@xcite investigate the origin of this excess with a set of cluster simulations .
they find that the presence of whim is not necessary to explain the observations , because the excess emission is produced by extremely dense gas associated with merging substructures within the cluster . in clusters where this gas is present ,
the temperature of the inner regions is usually @xmath38 per cent above that predicted by the one - temperature plasma model .
@xcite find that , since the whim surface brightness in x - rays reaches at most @xmath39 ergs@xmath24@xmath27arcmin@xmath24 , projection effects of whim filaments in front of galaxy clusters can account for no more than 10 per cent of the cluster emission .
this may partly explain the soft x - ray excess , but clearly does not fully account for it .
similar conclusions have been reached by @xcite .
most of the energy radiated by the whim in the uv and soft x - ray bands is emitted through emission lines .
the detection of emission lines from the whim is key to understanding the properties of the gas itself .
the relative intensities of emission lines from highly ionised metals in the whim gives information about the temperature of the emitting gas , while the absolute intensities reflect the density and the degree of metal enrichment of the gas .
the relative abundances of ionised atoms is determined by the temperature and , if photo - ionisation plays a role , the density of the whim .
ions whose abundance peaks at the highest temperatures , such as fe and fe , are concentrated in the hottest , highest density regions , while ions that peak at lower temperatures , such as o , o and fe , are more widespread and originate in groups and filaments @xcite .
the detection of whim emission from the intracluster medium of filaments , groups and cluster with different masses helps to understand the impact of feedback in these structures @xcite .
@xcite and @xcite find that the temperature of the igm can be estimated if both the o and o ions are simultaneously observed .
however , if only one line is detectable , then the line can not be identified unambiguously .
[ yoshikawa2003 ] shows template spectra of a collisionally ionised plasma with temperature @xmath40 k , @xmath41 k , and @xmath42 k , and with metallicity @xmath43 and @xmath44 , which display the contribution of oxygen ions as a function of temperature .
clearly , the detectability of the line emission depends strongly on the metallicity .
k ( lower panels ) , @xmath41 k ( middle panels ) , and @xmath12 k ( upper panels ) .
spectra for metallicity @xmath43 and @xmath44 are shown in the left and right panels , respectively . from @xcite.,scaledwidth=90.0% ] it is still controversial if the whim can actually be detected in emission by current and future x - ray telescopes .
@xcite calculate that the virialised regions of groups and clusters cover about a third of the sky .
this is a potential problem in the identification of x - ray emission from the whim in filaments and distant groups , because it creates source - confusion and strongly contaminates the observations .
@xcite predict that _ xmm - newton _ should be able to detect diffuse emission from strong filaments at @xmath45 , although no such detections have been confirmed so far . as @xcite calculate , x - ray emission lines from the whim are hard to detect in high density regions and in voids .
the emission from clusters and groups is dominated by the hot intracluster medium , while in voids the x - ray spectrum is completely dominated by radiation from background agn and by the galactic foreground . in both cases
, the emission of the whim would be too low to be distinguished from the dominant component .
on the other hand , it might be possible to resolve at least a few strong emission lines from the whim in filamentary regions , despite the fact that most of the emission is contributed by agn and by the galactic foreground .
@xcite argue that planned telescopes such as _ constellation - x _ and the x - ray evolving universe spectroscopy mission ( _ xeus _ ) have little chance to see a significant signal for the whim emission , because their effective collecting area is too small to detect diffuse line emission from an extended area .
ideally , a mission to detect line emission from the whim should have a large field of view , coupled with a high resolving power and spectral resolution . @xcite and @xcite
use large scale simulations and simulations that reproduce the local universe to produce mock x - ray spectra of the whim emission .
the simulated maps of the local universe @xcite demonstrate that whim gas in filamentary structures can be successfully imaged in targeted observations by the proposed diffuse intergalactic oxygen surveyor ( _ dios _ ) , an x - ray telescope of new generation which combines a large field of view with high spatial and spectral resolution .
_ dios _ and several other proposed missions are discussed in more detail in @xcite - chapter 19 , this volume . [ [ uv ] ] uv ~~ while o and o lines are emitted by gas with @xmath46 k , line emission from the o @xmath47 and c @xmath48 doublets traces the coolest fraction of the whim , with @xmath49 k. for the bulk of the whim mass , the line strengths are predicted to be a few orders of magnitude lower than the background @xcite .
@xcite estimate the intensity of the o emission line at @xmath50 and column densities @xmath51 @xmath27 to be @xmath52 photons
@xmath27 s@xmath24 sr@xmath24 , which is an order of magnitude below the sensitivity of past and current telescopes such as the far ultraviolet spectroscopic explorer ( _ fuse _ ) and _
spear_. however , @xcite predict that o and c line emission could be detected above the background from enriched dense regions with sizes of about @xmath53 kpc .
@xcite argue that the connection between the line emission and the enrichment by galactic winds may help understand the chemical enrichment history of the igm and the physics of winds .
the x - ray forest is the high - energy counterpart of the ly@xmath7 forest .
while the ly@xmath7 forest is produced by cool gas absorbing ly@xmath7 photons in the optical spectra of high redshift quasars , the x - ray forest is the result of absorption by highly ionised atoms in the spectra of bright x - ray sources ( @xcite - chapter 3 , this volume ) .
the detection of the x - ray forest in high resolution spectra of quasars and gamma ray bursts ( grb ) is possibly one of the most powerful tools to investigate the properties of the whim in the low redshift universe .
for example , the number density of absorbers gives an estimate of the baryon density times the metallicity of the igm , while the ratio between the o and the o line strengths gives information on the distribution of the gas temperature and density .
a thorough knowledge of the physical state of the whim , as @xcite point out , helps to understand the mechanisms that have determined its evolution in a cosmological context .
@xcite use analytic techniques to investigate how the accretion shocks that heat up the igm during the formation of cosmic structures can give rise to absorption systems such as o and o. they find that the observed column densities of absorbers are predicted by post - shock cooling models , when both slow and fast cooling channels are taken into account
. the observed number density of absorbers can be realistically reproduced if each virialised structure is surrounded by a network of shocks with a total cross section a few times the size of the virialised region . in this scenario ,
fast cooling would produce stronger o absorbers associated to o systems than expected from models of collisional ionisation equilibrium .
o absorption systems should be the most common because the abundance of the o ion dominates over a large temperature range .
the observability of the whim in the x - ray forest by current and future facilities has been a very popular topic in recent years ( @xcite - chapter 3 , this volume ) .
a variety of predictions has been provided by hydrodynamical simulations ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) and semi - analytic models ( e.g. @xcite ; @xcite ; @xcite ) .
@xcite find that the regions of the highest column densities of o and o correspond to the high density regions in and around galaxy groups .
the gas in these regions , with @xmath40 k and @xmath54 , should produce the strongest o and o absorption lines , with equivalent width @xmath55 km s@xmath24 @xcite .
these lines should occur on average once per sight line in the redshift range @xmath56 ( @xcite ; @xcite ) .
@xcite predict abundances for o , o , and o lines in the range @xmath53 per unit redshift at @xmath57 km s@xmath24 , decreasing to @xmath58 per unit redshift at @xmath59 km s@xmath24 .
@xcite suggest that the most promising strategy to find o and o absorbers in the x - ray forest would be to search at the redshifts of known o absorbers .
however , since o absorption traces gas at somewhat lower temperatures than o and o , only future missions like _ constellation - x _ and _ xeus _ will be able to find absorption systems at higher temperatures .
most predictions agree that _ chandra _ and _ xmm - newton _
do not have enough sensitivity to detect the whim in absorption .
@xcite claim that a few strong o and o absorption systems might be within reach of _ chandra _ and _ xmm - newton _ , but the probability to detect such lines in random lines of sight is less than 5 per cent @xcite . future telescopes such as _ constellation - x _ and _ xeus _ might have a better chance to detect the whim in absorption , thanks to higher spectral resolution and larger collecting areas . _
constellation - x _ should have a 50 per cent probability to detect the strongest o and o absorption lines @xcite . because of its high sensitivity at high energies , it should also be able to detect absorption from ions such as ne and fe , whose intensity peaks at about the same temperature as o , and fe lines , which appear for @xmath60 k @xcite .
@xcite predict that one day of integration time will allow _ constellation - x _ and _ xeus _ to detect a few o absorption lines per unit redshift in the spectra of x - ray bright quasars . according to @xcite ,
_ constellation - x _ will detect about 6 absorption lines per unit redshift with @xmath61 km s@xmath24 , while _ xeus _ could detect up to 30 lines with @xmath62 km s@xmath24 per unit redshift . under the assumption of constant igm metallicity , @xcite estimate that _ xeus
_ will be able to detect on average two o absorption systems with @xmath63 along any line of sight to bright quasars at @xmath64 in an 8 hours exposure .
it should be noted , however , that the characteristics of the proposed facilities have been evolving and that some of the above predictions are therefore likely too optimistic .
[ [ uv-1 ] ] uv ~~ @xcite find that about 2030 per cent of the whim at @xmath20 , or equivalently , about 10 per cent of the baryonic mass , is traced by o absorption lines with equivalent width @xmath65 m .
they expect about five o absorption lines per unit redshift with @xmath66 m and about 0.5 per unit redshift with @xmath67 m .
@xcite estimate that most o absorption lines are correlated to filamentary regions with overdensities of 5100 and temperatures of a few times @xmath11 k. the equivalent width of absorption systems in the ultraviolet region can be useful to distinguish a prevalently photoionised from a prevalently collisionally ionised igm .
collisional ionisation dominates in the denser and warmer regions of the igm and is responsible for most lines with @xmath68 m . on the other hand , photoionisation is predominant in the cooler and less dense regions and usually produces narrow absorption lines with @xmath69 m .
essentially all lines with equivalent width @xmath70 m are due to collisionally ionised gas @xcite
. @xcite investigate the origin of the broad h ly@xmath7 absorbers ( bla ) with doppler parameters @xmath71 km s@xmath24 seen in uv spectra from the _ hubble space telescope _ and _ fuse _ ( @xcite ; @xcite - chapter 4 , this volume ) .
they find that such absorbers are associated with whim gas with temperatures in the range @xmath72 k , that represents about a quarter of the gas mass in the simulation at @xmath20 . less than one third of the broad h absorbers are associated with cooler gas with @xmath73 k , whose broadening is produced by non - thermal processes such as turbulence and velocity structures in the igm .
it has been suggested that ionised whim gas could possibly be detected through the thermal sunyaev - zeldovich effect it produces in the cmb ( e.g. @xcite ; @xcite ; @xcite ; @xcite ) .
when the cmb radiation crosses a hot ionised medium such as the whim , the energy of the passing photons can be increased by collisions with the free electrons in the whim .
this produces a local increase of the cmb temperature , with an increase proportional to the line integral of the pressure of the hot gas .
although the signal arising from the hot gas in large clusters has been clearly identified , a weaker signal associated with whim gas has not yet been reliably detected in current cmb data ( @xcite ; @xcite ) .
@xcite estimate that the whim could contribute on average between 4 and 12 per cent of the integrated sunyaev - zeldovich effect ( sze ) signal @xmath74 for individual sources identified in upcoming surveys .
the whim is a particularly relevant source of contamination for observations of clusters of galaxies , because it would strongly affect the calibration of the cluster @xmath75 relationship .
unless the whim signal can be unambiguously modelled and separated from the cluster signal , it introduces a bias and a source of scatter difficult to quantify in the @xmath75 relationship .
ultimately , this would limit the reliability of the @xmath75 relationship for estimating cosmological parameters .
it has been suggested that cosmic shear and gravitational shocks might be able to amplify the magnetic fields that permeate the large - scale structure ( e.g. @xcite ; @xcite ; @xcite ) . given the strong interconnection between the origin of the whim and gravitational shocks , it is possible for the whim to be magnetised , although it might be difficult to predict to what level .
while magnetic fields have been unambiguously detected in the cores of galaxy clusters , few observations have attempted to establish the presence of magnetised plasma in low density regions such as groups and filaments . however , there is now compelling evidence that magnetic fields might exist beyond clusters , and in particular in the whim .
@xcite claim the detection of an extended magnetic field in the region of the coma supercluster .
they observe radio - synchrotron radiation from a region with an enhanced number density of galaxies , seeming to indicate the presence of a group of galaxies , perhaps in the process of merging with the coma cluster .
evidence for intergalactic magnetic fields in the region zwcl 2341 + 0000 at @xmath76 , where galaxies seems to lie along a filament , has been reported by @xcite . in both cases ,
the inferred strength of the magnetic field is of order @xmath77 @xmath78 g .
several mechanisms have been proposed to explain the injection of seed magnetic fields into the intracluster and intergalactic media , later to be amplified by cosmic shear and gravitational shocks . among them , primordial magnetic fields , jets and radio lobes emerging from radio galaxies ( e.g. @xcite ; @xcite ; @xcite ) and galactic winds ( e.g. @xcite ; @xcite ) . simulations ( @xcite ; @xcite ; @xcite ) show that magnetic fields can be efficiently transported into the igm and fill most of the universe by @xmath20 .
@xcite predict that galactic winds alone could be responsible for igm magnetic fields of order @xmath79 @xmath78 g , in agreement with observations .
the search for the `` missing baryons '' has opened the way to our understanding of the most elusive baryonic component in the universe : the hot diffuse gas that traces the cosmic web . given the paucity of observational findings , most of what we know about the whim is based on numerical simulations .
these predict that at @xmath80 up to 50 per cent of the igm has been heated to temperatures in the range @xmath81 k. while most of the heating is provided by gravitational accretion shocks propagating out of regions undergoing gravitational collapse , non - gravitational processes such as galactic winds and agn feedback may play an important role in high density regions , mostly by preventing cooling and regulating the intensity of the x - ray background .
the observability of the whim by future facilities has received a lot of attention and a wealth of predictions have been made based on numerical results .
the detection of the whim seems to be extremely challenging , if not impossible , with current instruments and a new generation of x - ray telescopes with higher sensitivity and spectral resolution , such as _ constellation - x _ and _ xeus _ , is needed to study the whim in absorption .
in addition to high sensitivity and spectral resolution , high spatial resolution on a large field of view , as proposed for the explorer of diffuse emission and gamma - ray burst explosions ( _ edge _ ) , would be ideal for mapping the whim emission on the sky .
although most theoretical predictions claim optimistically that we will be able to detect the whim in absorption with _ constellation - x _ and _ xeus _ , and maybe in emission with some luck , there is a long way to go before observations can confirm the simulation predictions , or show us a different picture altogether of what and where the whim is .
we are grateful to frits paerels and philipp richter for a careful reading of the manuscript .
we thank issi ( bern ) for support of the team `` non - virialized x - ray components in clusters of galaxies '' .
this work was supported by a marie curie excellence grant mext - ct-2004 - 014112 .
sb acknowledges support from stfc and from a nsf grant ast-0507117 . | in this paper we review the current predictions of numerical simulations for the origin and observability of the warm hot intergalactic medium ( whim ) , the diffuse gas that contains up to 50 per cent of the baryons at @xmath0 . during structure formation ,
gravitational accretion shocks emerging from collapsing regions gradually heat the intergalactic medium ( igm ) to temperatures in the range @xmath1 k. the whim is predicted to radiate most of its energy in the ultraviolet ( uv ) and x - ray bands and to contribute a significant fraction of the soft x - ray background emission . while o and c absorption systems arising in the cooler fraction of the whim with @xmath2 k are seen in _ fuse _ and _ hubble space telescope _
observations , models agree that current x - ray telescopes such as _ chandra _ and _ xmm - newton _ do not have enough sensitivity to detect the hotter whim .
however , future missions such as _ constellation - x _ and _ xeus _ might be able to detect both emission lines and absorption systems from highly ionised atoms such as o , o and fe . |
the study of geometrical and topological properties of classical non - abelian gauge theories turned out to be very important for our understanding of non - perturbative aspects of the corresponding quantum field theories .
the configuration space of the theory is the gauge orbit space , which is obtained by factorizing the space of connections with respect to the action of the group of local gauge transformations .
this space has the structure of a stratified set , because , usually , besides the principal orbit type also non - generic orbit types occur . these may give rise to singularities of the configuration space .
first , the generic , or principal , stratum was investigated leading to a deeper understanding of the gribov - ambiguity @xcite and of anomalies in terms of index theorems @xcite .
in particular , one gets anomalies of purely topological type , which can not be seen by perturbative quantum field theory @xcite .
next , in a paper by kondracki and rogulski @xcite , a systematic study of the structure of the full gauge orbit space was presented . in particular , it was shown that the gauge orbit space is a stratified topological space in the ordinary sense , cf .
@xcite and references therein .
there are partial results and conjectures concerning the physical relevance of nongeneric strata .
first of all , nongeneric gauge orbits affect the classical motion on the orbit space due to boundary conditions and , in this way , may produce nontrivial contributions to the path integral .
they may also lead to localization of certain quantum states , as it was suggested by finite - dimensional examples @xcite .
further , the gauge field configurations belonging to nongeneric orbits can possess a magnetic charge , i.e. they can be considered as a kind of magnetic monopole configurations , which seem to be related to the quark confinement problem in chern - simons theory @xcite . finally , it was suggested in @xcite that non - generic strata may lead to additional anomalies . most of the problems mentioned here are still awaiting a systematic investigation . in a series of papers we are going to make a new step in this direction . in @xcite we have presented a complete solution to the problem of determining the strata that are present in the gauge orbit space for @xmath2 gauge theories in compact euclidean space - time of dimension @xmath3 .
the basic idea behind is the 1 - 1-correspondence between orbit types and equivalence classes of so - called holonomy - induced howe subbundles of the principal @xmath4bundle , where the gauge connections of the theory under consideration live on .
it turns out that howe subgroups of @xmath4 as well as ( holonomy - induced ) howe subbundles can be classified , leading to a classification of orbit types in terms of certain algebraical and topological data . as a first application , we have shown in @xcite that within the context of chern - simons theory in @xmath5 dimensions the property of a configuration to be nodal in the sense of asorey ,
see @xcite , is a property of strata . for a given model of this type , the nodal strata can be easily determined . in @xcite
one basic problem was left open : the determination of the natural partial ordering in the set of orbit types . in the present paper
we solve this problem .
first , in section [ sprev ] we recall the classification of gauge orbit types from @xcite . in section [ parord ]
we prove that the natural partial ordering is characterized by a system of algebraic equations relating the classifying data via a matrix with non - negative integer entries ( inclusion matrix ) .
the inclusion matrix can be visualized by a bratteli diagram , as explained in section [ bratteli ] . in sections
[ dirsuc ] and [ dirpred ] direct successors and direct predecessors are characterized . in particular , operations which generate the direct successors ( splitting and merging ) and the direct predecessors ( inverse splitting and inverse merging ) are defined .
finally , an example is discussed : for gauge group @xmath6 and some space - time manifolds the complete hasse diagram of the set of orbit types is derived .
let @xmath7 be a principal @xmath2-bundle over a compact , connected , orientable riemannian manifold @xmath8 of dimension @xmath9 .
let @xmath10 and @xmath11 denote the sets of connection forms and gauge transformations , respectively , of sobolev class @xmath12 .
provided @xmath13 , @xmath10 is an affine hilbert space and @xmath14 is a hilbert lie group acting smoothly from the right on @xmath10 @xcite .
if we view gauge transformations as equivariant maps @xmath15 then for @xmath16 and @xmath17 the action is given by @xmath18 let @xmath19 denote the quotient topological space @xmath20 .
this is the gauge orbit space , i.e. , the configuration space of our gauge theory .
let @xmath21 denote the set of orbit types of the action of @xmath14 on @xmath10 . recall that orbit types are given by conjugacy classes in @xmath14 of stabilizer , or isotropy , subgroups of connections .
the set @xmath21 carries a natural partial ordering : let @xmath22 .
then @xmath23 iff there exist representatives @xmath24 of @xmath25 , @xmath26 , respectively , such that @xmath27 .
note that this definition is consistent with @xcite but not with @xcite and several other authors who define the partial ordering inversely .
in @xcite it was shown that the family @xmath28 , where @xmath29 denotes the subset of @xmath19 of orbits of type @xmath25 , is a stratification of @xmath19 into smooth hilbert manifolds . for the notion of stratification ,
see @xcite or @xcite .
moreover , for any @xmath30 , @xmath29 is open and dense in the union @xmath31 . in this sense ,
the partially ordered set @xmath21 encodes the stratification structure of the gauge orbit space . in @xcite , we have derived a description of the elements of @xmath21 in terms of certain cohomology elements of @xmath8 . in the present article
, we are going to discuss the partial ordering .
for the convenience of the reader , we begin with briefly recalling the basic results of @xcite . a _ howe subgroup _ of a group @xmath32 is a subgroup @xmath33 that is the centralizer @xmath34 of some subset @xmath35 . a _
howe subbundle _ of a @xmath32-bundle @xmath7 is a reduction of @xmath7 to a howe subgroup .
a howe subbundle is called _ holonomy - induced _ iff it admits a connected reduction @xmath36 to a subgroup @xmath37 such that @xmath38 let @xmath39 denote the set of isomorphism classes of holonomy - induced howe subbundles of @xmath7 factorized by the natural action of the structure group @xmath32 .
note that here an isomorphism of principal bundles is assumed to commute with the structure group action and to project to the identical mapping on the base space .
the set @xmath39 carries a natural partial ordering defined by the relation of inclusion up to isomorphy and up to the action of @xmath32 .
[ potgau ] @xmath39 is isomorphic , as a partially ordered set , to @xmath21 . _
proof : _ see ( * ? ? ?
3.3 ) . ' '' ''
+ we note that in the case @xmath40 , any howe subbundle is holonomy - induced , see ( * ? ? ?
hence , this condition is redundant here .
the following description of @xmath39 has been derived in @xcite .
first , the howe subgroups of @xmath2 were determined .
let @xmath41 denote the set of pairs of sequences of strictly positive integers @xmath42 obeying @xmath43 .
let @xmath44 denote the greatest common divisor of the members of @xmath45 and let @xmath46 be defined by @xmath47 @xmath48 .
we shall always view @xmath49 as an @xmath50-matrix ( row vector ) and @xmath45 as a @xmath51-matrix ( column vector ) .
this turns out to be their natural character .
any @xmath52 defines a decomposition @xmath53 and an embedding @xmath54 here @xmath55 stands for the algebra of complex @xmath56-matrices . identifying @xmath57 , @xmath58
, the tensor product @xmath59 corresponds to the @xmath60 block matrix @xmath61 we denote the image of the embedding ( [ gdefmj ] ) by @xmath62 and its intersections with @xmath63 and @xmath2 by @xmath64 and @xmath65 , respectively .
note that @xmath64 is the image of the restriction of ( [ gdefmj ] ) to @xmath66 . by construction , @xmath62 is a unital @xmath67-subalgebra of @xmath68 .
[ phsg ] up to conjugacy , the howe subgroups of @xmath2 are given by @xmath65 , @xmath52 .
_ proof : _ see ( * ? ? ?
* lemma 4.1 ) . ' '' '' + in order to classify principal @xmath65-bundles over @xmath8 , the homotopy classes of maps from @xmath8 to the classifying space @xmath69 have to be determined .
through building the postnikov tower of @xmath69 up to the @xmath70th stage the following was shown .
[ ppost ] let @xmath8 be a manifold , @xmath9 and let @xmath71 , @xmath72 be principal @xmath65-bundles over @xmath8 .
assume that for any characteristic class @xmath73 defined by an element of @xmath74 , @xmath75 , or @xmath76 there holds @xmath77 .
then @xmath71 and @xmath72 are isomorphic . _
proof : _ see ( * ? ? ?
* cor . 5.5 ) .
+ a generating set for the characteristic classes mentioned in the proposition can be constructed as follows .
consider the natural homomorphisms @xmath78 for any positive integer @xmath79 , let @xmath80 denote the sum of generators of the cohomology algebra @xmath81 .
we assume that the generators are chosen in such a way that for the canonical blockwise embedding @xmath82 there holds @xmath83 @xmath84 .
( recall that @xmath85 is the map between classifying spaces associated to @xmath86 . ) then , in particular , the characteristic classes defined by the generators @xmath87 are the @xmath88th chern classes .
the cohomology elements @xmath89 of @xmath90 define characteristic classes @xmath91 where @xmath92 . here
@xmath93 is the classifying map of @xmath71 and @xmath94 stands for the set of isomorphism classes of principal @xmath65-bundles over @xmath8 .
we denote @xmath95 .
next , for any positive integer @xmath79 , let @xmath96 denote the canonical embedding and let @xmath97 denote the endomorphism @xmath98 of @xmath99 .
we define a homomorphism @xmath100 one can check that the diagram @xmath101 \begin{array}{c } \square[{{{{\rm u}}j}}`{{\rm u}}n`{{\rm u}}1`{{\rm u}}1;j_j`{\lambda^{{\rm u}}_j}`\det\nolimits_{{{\rm u}}n}`p_{g } ] \end{array}\ ] ] commutes .
moreover , we notice that the image of @xmath65 under @xmath102 is the subgroup @xmath103 of @xmath99 .
thus , @xmath102 induces a homomorphism @xmath104 by requiring the diagram @xmath105 \begin{array}{c } \square[{{{{\rm su}}j}}`{{{{\rm u}}j}}`{{\mathbb{z}}}_g`{{\rm u}}1;j_j`{\lambda^{{\rm s}}_j}`{\lambda^{{\rm u}}_j}`j_g ] \end{array}\ ] ] to commute .
( in fact , one can show that @xmath106 projects to an isomorphism of the group of connected components of @xmath65 onto @xmath107 . )
one can show that the bockstein homomorphism @xmath108 , induced by the short exact sequence @xmath109 , is an isomorphism , see the proof of lemma 5.9 in @xcite .
thus , we can consider the cohomology element @xmath110 of @xmath111 .
it defines a characteristic class @xmath112 by construction , the characteristic classes @xmath113 and @xmath114 are subject to a relation . to formulate it ,
let us introduce the following notation .
let @xmath115 be positive integers .
for any @xmath116 ( set of @xmath117-matrices with nonnegative integer entries ) , we define a map @xmath118 here powers are taken w.r.t . the cup product and @xmath119 denotes the subset of @xmath120 of elements of the form @xmath121 .
note that @xmath122 is a semigroup w.r.t . the cup product .
let @xmath123 denote the component of degree @xmath124 of the @xmath125-th member of @xmath126 .
[ prel ] the characteristic classes @xmath113 , @xmath114 are subject to the relation @xmath127 ( recall that @xmath128 is viewed as a @xmath51-matrix . )
+ _ proof : _ see ( * ? ? ?
* theorem 5.13 ) . ' '' '' + we introduce the notation @xmath129 consider the following two equations in the variables @xmath130 , @xmath131 : @xmath132 here @xmath133 denotes the total chern class of @xmath7 . [ psujbun ] if @xmath9 , the characteristic classes @xmath113 and @xmath114 define a bijection from @xmath94 onto the set of solutions of equation ( [ gkmj ] ) . by restriction
, they define a bijection from the subset of @xmath94 of reductions of @xmath7 onto the set of solutions of the equations ( [ gkmj ] ) and ( [ gkpj ] ) . _
proof : _ see ( * ? ? ?
* thms . 5.14 , 5.17 ) . ' '' '' + note that the content of eq .
( [ gkpj ] ) in degree @xmath134 is a consequence of eq .
( [ gkmj ] ) .
let @xmath135 denote the disjoint union of the solution sets of eqs .
( [ gkmj ] ) , ( [ gkpj ] ) over all @xmath52 .
we write the elements of @xmath135 as triples @xmath136 , where @xmath52 and @xmath137 is a solution of the corresponding equations . according to proposition [ psujbun ] , the set @xmath135 classifies the howe subbundles of @xmath7 up to isomorphy .
+ finally , the action of the structure group @xmath2 on howe subbundles of @xmath7 was factored out by passing to the set @xmath138 that is obtained from @xmath135 by identifying @xmath136 with @xmath139 for all permutations @xmath140 of @xmath141 . here
@xmath142 stands for @xmath143 [ thsb ] the collection of characteristic classes @xmath144 , defines , by passing to quotients , a bijection from @xmath39 onto @xmath138 . _
proof : _ see ( * ? ? ?
7.2 ) . ' '' ''
+ in the sequel , it is convenient to work with the inverse of this bijection . to construct it , for
any @xmath145 , @xmath146 , let @xmath147 denote the isomorphism class of @xmath65-subbundles of @xmath7 defined by @xmath148 then the pre - image of the element of @xmath138 represented by @xmath149 is given by the conjugacy class of @xmath147 under @xmath2-action .
the ( isomorphy classes of ) subbundles @xmath147 may be viewed as some kind of standard representatives of the elements of @xmath39 .
+ to conclude this section , for later use , let us collect some formulae involving the function @xmath150 . for any @xmath125 , one has @xmath151 see ( * ? ? ?
* lemma 5.11 ) .
in particular , for any nonnegative integer @xmath79 , @xmath152 taking into account that the cup product is commutative in even degree , one can also check that for any @xmath116 and @xmath153 there holds @xmath154
in this section we are going to determine the natural partial ordering of @xmath39 on the level of the classifying set @xmath138 .
let @xmath162 such that @xmath163
. then there also holds @xmath164 and @xmath165 .
we have an associated homomorphism @xmath166 and , derived from that , homomorphisms @xmath167 and @xmath168 . due to @xmath62 and @xmath169 being finite - dimensional unital @xmath170-algebras ,
the embedding @xmath171 is characterized by an @xmath117-matrix @xmath172 ( nonnegative integer entries ) , called _
inclusion matrix_. the matrix @xmath173 can be constructed as follows : for @xmath174 and @xmath175 , consider the homomorphism @xmath176 where the first map is canonical embedding to the @xmath177th factor of @xmath62 . define @xmath178 to be the number of fundamental irreps contained in the representation of @xmath179 defined by ( [ grepiip ] ) .
[ leqkm ] let @xmath180 .
let @xmath162 such that @xmath163 .
then @xmath181 conversely , let @xmath116 be a solution of ( [ geqkd ] ) and ( [ geqmd ] ) .
then there exists @xmath162 such that @xmath163 and @xmath182 .
_ proof : _ first , let @xmath183 be given as proposed . consider the representations @xmath184 the numbers of fundamental irreps contained in ( [ gleqkm1 ] ) and ( [ gleqkm2 ] ) are @xmath185 and @xmath186 , respectively .
since ( [ gleqkm1 ] ) and ( [ gleqkm2 ] ) are isomorphic a bijective intertwiner being given by @xmath183 we obtain ( [ geqmd ] ) . moreover , inserting this equation into @xmath187 yields @xmath188 .
by construction , the members of the sequence @xmath189 are nonnegative . since the members of @xmath45 are strictly positive , eq . ( [ geqkd ] ) follows .
conversely , let @xmath190 be a solution of ( [ geqkd ] ) and ( [ geqmd ] ) .
consider the decompositions @xmath191 defined by @xmath192 and @xmath193 , respectively .
due to ( [ geqmd ] ) and ( [ geqkd ] ) , ( [ gdecompj ] ) and ( [ gdecompjp ] ) admit subdecompositions @xmath194 respectively .
there exists @xmath162 transforming ( [ gdecompjp1 ] ) into ( [ gdecompj1 ] ) by a suitable permutation of the subspaces @xmath195 .
one can check that @xmath196 leaves the decomposition ( [ gdecompjp ] ) invariant .
it follows @xmath197 , hence @xmath163 .
moreover , from ( [ gdecompj1 ] ) and ( [ gdecompjp1 ] ) one can read off that @xmath182 . ' '' '' + we remark that for general inclusions of @xmath198 , inclusion matrices only have to obey @xmath199 , where the inclusion is unital iff there holds equality for all @xmath125 .
let us denote the set of solutions of the system of equations ( [ geqkd ] ) and ( [ geqmd ] ) by @xmath200 .
we note that if @xmath201 , then ( [ geqmd ] ) implies that @xmath202 divides @xmath44 .
hence , reduction @xmath203 @xmath204 is defined and is a ring homomorphism .
again , let @xmath162 such that @xmath163 .
let @xmath205}}=q_l\times_{{{{\rm su}}j}}{{{{\rm su}}j}}'$ ] denote the @xmath206-subbundle of @xmath7 associated to @xmath158 by virtue of the homomorphism @xmath207 .
[ lqd ] the characteristic classes of @xmath205}}$ ] are @xmath208}}\right ) & = & e_{\delta(d)}\left(\alpha\right)\ , , \\
\label{gxqd } \xi_{j'}\left({q_{l } \,\!^{[\,{h^{{\rm s}}_d}\,]}}\right ) & = & \varrho_{gg'}\left(\xi\right)\,.\end{aligned}\ ] ] _ proof : _ the classifying map of @xmath205}}$ ] is @xmath209 } } } } = { { \rm b}}{h^{{\rm s}}_d}\circ{f_{q_l}}\,.\ ] ] hence , according to ( [ gdefaji ] ) , @xmath210}}\right ) & = & \left({f_{{q_{l } \,\!^{[\,{h^{{\rm s}}_d}\,]}}}}\right)^\ast \left ( \left({{\rm b}}j_j\right)^\ast \left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j',i'}}\right)^\ast \gamma_{{{\rm u}}k'_{i ' } } \right ) \\
\nonumber & = & \left({f_{q_l}}\right)^\ast \left({{\rm b}}{h^{{\rm s}}_d}\right)^\ast \left ( \left({{\rm b}}j_j\right)^\ast \left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j',i'}}\right)^\ast \gamma_{{{\rm u}}k'_{i ' } } \right ) \\ \label{gajpql } & = & \left({f_{q_{l}}}\right)^\ast \left({{\rm b}}j_j\right)^\ast \left({{\rm b}}{h^{{\rm u}}_d}\right)^\ast \left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j',i'}}\right)^\ast \gamma_{{{\rm u}}k'_{i'}}\,.\end{aligned}\ ] ] in order to calculate @xmath211 , consider the homomorphisms @xmath212 since the image of ( [ ghommj ] ) is a unital @xmath67-subalgebra of @xmath213 , the image of ( [ ghomuj ] ) is a howe subgroup of @xmath214 .
hence , the latter is conjugate to @xmath215 for some @xmath216 .
one can check that @xmath217 is obtained from the pair of sequences @xmath218 by deleting all pairs of entries @xmath219 , @xmath220 for which @xmath221 . on the other hand
, @xmath215 is the image of the homomorphism @xmath222 here @xmath223 denotes diagonal embedding into the @xmath79-fold product , where for @xmath224 this product is assumed to reduce to @xmath225 , and @xmath226 is a standard blockwise embedding .
having conjugate images , the homomorphisms ( [ ghomuj ] ) and ( [ ghomfip ] ) are conjugate themselves @xcite , i.e. , there exists an inner automorphism @xmath227 of @xmath214 such that the following diagram commutes : @xmath228 \end{array}\ ] ] since @xmath214 is connected , @xmath229 is null - homotopic .
thus , on the level of cohomology , @xmath230 from the decomposition ( [ ghomfip ] ) one derives @xmath231 see the proof of ( * ? ? ?
* l. 5.12 ) for details .
we remark that ( [ gfipguk ] ) is an analogue of the whitney sum formula . using ( [ gdefe ] ) , from ( [ gbfip ] ) and ( [ gfipguk ] ) we deduce @xmath232 inserting ( [ gbhud ] ) into ( [ gajpql ] ) and using ( [ gdefaji ] ) and ( [ gaql ] ) we find @xmath233}}\right ) &
= & \left({f_{q_{l}}}\right)^\ast \left({{\rm b}}j_j\right)^\ast e_{\delta(d),i'}\left ( \left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j,1}}\right)^\ast\gamma_{{{\rm u}}k_1 } , \stackrel{\phantom{j^j}}{\dots } , \left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j , r}}\right)^\ast\gamma_{{{\rm u}}k_r } \right ) \\ & = & \left({f_{q_{l}}}\right)^\ast e_{\delta(d),i'}\left ( \left({{\rm b}}j_j\right)^\ast\left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j,1}}\right)^\ast\gamma_{{{\rm u}}k_1 } , \stackrel{\phantom{j^j}}{\dots } , \left({{\rm b}}j_j\right)^\ast\left({{\rm b}}{{{\rm pr}}^{{{\rm u}}}_{j , r}}\right)^\ast\gamma_{{{\rm u}}k_r } \right ) \\ & = & e_{\delta(d),i'}\left ( \alpha_j\left(q_{l}\right ) \right ) \\ & = & e_{\delta(d),i'}\left(\alpha\right)\ , .
\end{array}\ ] ] this proves ( [ gaqd ] ) .
now consider ( [ gxqd ] ) . using ( [ gdefxj ] ) and ( [ gclfmapassbun ] ) , we compute @xmath234}}\right)\right ) & = & \beta_{g ' } \left({f_{{q_{l } \,\!^{[\,{h^{{\rm s}}_d}\,]}}}}\right)^\ast \left ( \left({{\rm b}}{\lambda^{{\rm s}}_{j'}}\right)^\ast \beta_{g'}^{-1 } \left({{\rm b}}j_{g'}\right)^\ast \gamma_{{{\rm u}}1}^{(2 ) } \right ) \\ \nonumber & = & \left({f_{{q_{l } \,\!^{[\,{h^{{\rm s}}_d}\,]}}}}\right)^\ast \left({{\rm b}}{\lambda^{{\rm s}}_{j'}}\right)^\ast \left({{\rm b}}j_{g'}\right)^\ast \gamma_{{{\rm u}}1}^{(2 ) } \\ \label{gxjql } & = & \left({f_{q_{l}}}\right)^\ast \left({{\rm b}}{h^{{\rm s}}_d}\right)^\ast \left({{\rm b}}{\lambda^{{\rm s}}_{j'}}\right)^\ast \left({{\rm b}}j_{g'}\right)^\ast \gamma_{{{\rm u}}1}^{(2)}\,.\end{aligned}\ ] ] let @xmath79 be such that @xmath235 .
the following relation will be proved afterwards : @xmath236 inserting ( [ glqd1 ] ) into ( [ gxjql ] ) yields @xmath237}}\right)\right ) = \left({f_{q_{l}}}\right)^\ast \left({{\rm b}}{\lambda^{{\rm s}}_j}\right)^\ast \left({{\rm b}}j_g\right)^\ast \left({{\rm b}}p_l\right)^\ast \gamma_{{{\rm u}}1}^{(2)}\,.\ ] ] it is easily seen that @xmath238 is multiplication by @xmath79 .
therefore , @xmath239 then ( [ gxjql1 ] ) becomes @xmath234}}\right)\right ) & = & l\,\left({f_{q_{l}}}\right)^\ast \left({{\rm b}}{\lambda^{{\rm s}}_j}\right)^\ast \left({{\rm
b}}j_g\right)^\ast \gamma_{{{\rm u}}1}^{(2 ) } \\ \nonumber & = & l\,\beta_{g}\left({f_{q_{l}}}\right)^\ast \left ( \left({{\rm b}}{\lambda^{{\rm s}}_j}\right)^\ast \beta_{g}^{-1 } \left({{\rm b}}j_g\right)^\ast \gamma_{{{\rm u}}1}^{(2 ) } \right ) \\ \nonumber & = & l\,\beta_{g}\left(\xi_j\left(q_{l}\right)\right ) \\
\label{gxjql2 } & = & l\,\beta_{g}\left(\xi\right)\,,\end{aligned}\ ] ] where for the last two equalities we have used ( [ gdefxj ] ) and ( [ gxql ] ) , respectively . as a direct consequence of the definition of the bockstein homomorphism
, one has @xmath240 thus , ( [ gxjql2 ] ) yields @xmath241}}\right)\right ) = \beta_{{g'}}\varrho_{{g}{g'}}\left(\xi\right)\,.\ ] ] consider the following portion of the long exact sequence of coefficient homomorphisms which is induced by the short exact sequence @xmath109 , see ( * ? ?
iv,5 ) : @xmath242 since @xmath243 , see ( * ? ? ?
5.8 ) , @xmath244 is injective here .
hence , ( [ gxjql3 ] ) implies ( [ gxqd ] ) . it remains to prove the relation ( [ glqd1 ] ) .
according to ( [ gdefluj ] ) and ( [ gctvdgrlsj ] ) , for any @xmath245 , @xmath246 using ( [ gpsipfip ] ) to replace @xmath247 and taking into account that an inner automorphism does not change the determinant , ( [ gxqd1 ] ) yields @xmath248 by construction of @xmath249 , see ( [ ghomfip ] ) , for any @xmath250 , @xmath251 thus , ( [ gxqd2 ] ) becomes @xmath252 due to ( [ geqmd ] ) , @xmath253 , hence @xmath254 consequently , ( [ gxqd3 ] ) implies @xmath255 where the last two equalities are due to ( [ gdefluj ] ) and ( [ gctvdgrlsj ] ) , respectively .
this proves ( [ glqd1 ] ) and , therefore , concludes the proof of the lemma . ' '' '' [ lsubbun ] let @xmath162 such that @xmath163 .
then @xmath256 is a reduction of @xmath205}}$ ] to the structure group @xmath257 . _
proof : _ define a map @xmath258}}$ ] , @xmath259 $ ] .
this map is obviously smooth . to check equivariance ,
let @xmath260 . then @xmath261 \\ & = & [ \left(q,{h^{{\rm s}}_d}(c)\right ) ] \\ & = & \left[\left(q,{{\mathbbm{1}}}\right)\right]\cdot{h^{{\rm s}}_d}(c ) \\
& = & \left[\left(q,{{\mathbbm{1}}}\right)\right]\cdot d^{-1}cd\ , .
\end{array}\ ] ] this proves the lemma . ' '' '' [ tpo ] let @xmath146 , @xmath155 be elements of @xmath135 . then @xmath262\leq[q_{l'}]$ ] if and only if \(a ) @xmath202 divides @xmath44 and there holds @xmath263 , \(b ) there exists @xmath116 such that @xmath264 _ proof : _ to begin with , assume @xmath262\leq[q_{l'}]$ ]
. then there exists @xmath162 such that @xmath265 . since @xmath266 has structure group @xmath257 , @xmath163 . as a consequence ,
the homomorphism @xmath267 and the inclusion matrix @xmath173 exist . due to lemma [ leqkm ] , @xmath268 ,
hence it obeys ( [ geqk ] ) and ( [ geqm ] ) .
the latter equation implies , in particular , that @xmath202 divides @xmath269 .
moreover , by construction , @xmath159 can be reduced to @xmath266 .
according to lemma [ lsubbun ] , so can the @xmath206-bundle @xmath205}}$ ] .
since @xmath159 and @xmath205}}$ ] have the same structure group , it follows @xmath270}}$ ] .
then lemma [ lqd ] yields @xmath271}}\right ) = e_{\delta(d)}(\alpha)\,.\ ] ] thus , @xmath173 satisfies ( [ geqa ] ) . by an analogous argument , we finally find @xmath263 .
conversely , assume that assertions ( a ) and ( b ) hold . then
, due to lemma [ leqkm ] , there exists @xmath162 such that @xmath163 and @xmath182 .
consider the @xmath206-bundle @xmath205}}$ ] associated to @xmath158 .
due to lemma [ lqd ] and ( [ geqa ] ) , @xmath272}}\right ) = e_\delta(\alpha ) = \alpha ' = \alpha_{j'}\left(q_{l'}\right)\,.\ ] ] analogously , we obtain @xmath273}}\right ) = \xi_{j'}\left(q_{l'}\right)$ ] . hence , @xmath159 and @xmath205}}$ ] are isomorphic
. then lemma [ lsubbun ] implies @xmath265 , up to isomorphy ( which is sufficient ) .
it follows @xmath262\leq[q_{l'}]$ ] . ' '' '' + let @xmath274 . if condition ( a ) of theorem [ tpo ] holds , we define @xmath275 to be the set of solutions of the system of equations ( [ geqk])([geqa ] ) .
if this condition does not hold , we define @xmath276 . in order to be able to argue entirely on the level of @xmath138
, we define a partial ordering on
@xmath138 as the image of the natural partial ordering of @xmath39 under the bijection defined by the collection of characteristic classes @xmath113 , @xmath114 , @xmath52 . according to theorem [ tpo ] , the partial ordering so defined can be characterized as follows .
[ cpo ] let @xmath277 .
then the following assertions are equivalent : \(a ) @xmath278 .
\(b ) there exist representatives @xmath149 , @xmath161 of @xmath279 , @xmath280 , respectively , such that @xmath275 is nonempty .
\(c ) for any two representatives @xmath149 , @xmath161 of @xmath279 , @xmath280 , respectively , @xmath275 is nonempty .
_ proof : _
\(a ) @xmath281 ( c ) : let @xmath149 , @xmath161 be given . by assumption , @xmath156\leq[q_{l'}]$ ]
. then theorem [ tpo ] implies that @xmath275 is nonempty .
\(c ) @xmath281 ( b ) : obvious .
\(b ) @xmath281 ( a ) : let @xmath149 , @xmath161 be the representatives provided by assertion ( b )
. since @xmath275 is nonempty , assertions ( a ) and ( b ) of theorem [ tpo ] hold .
it follows that the subbundles @xmath147 and @xmath159 obey @xmath156\leq[q_{l'}]$ ] .
hence , @xmath278 . ' '' '' let @xmath282 .
consider elements @xmath146 , @xmath155 of @xmath135 , where @xmath283 and @xmath284 .
we remark that the subgroup @xmath285 has connected components @xmath286 hence is isomorphic to the direct product @xmath287 .
the subgroup @xmath206 can be parametrized as follows : @xmath288 thus , it is isomorphic to the direct product @xmath289 . in order to find out whether @xmath156\leq[q_{l'}]$ ] , we are going to determine @xmath275 .
condition ( a ) of theorem [ tpo ] is obviously satisfied .
thus , we can proceed as follows : first , we solve eqs .
( [ geqk ] ) and ( [ geqm ] ) , i.e. , we derive @xmath200 .
then , for all @xmath290 , we compute @xmath291 and compare the result with @xmath292 . eqs .
( [ geqk ] ) and ( [ geqm ] ) read @xmath293 we extract the equations @xmath294 the solutions are @xmath295 for @xmath296 , they yield @xmath297 thus , @xmath298 , i.e. , @xmath156\leq[q_{l'}]$ ] , if and only if @xmath292 coincides with one of the elements @xmath299 , @xmath300 , or @xmath301 , listed above .
any @xmath116 can be visualized by a diagram consisting of a series of upper vertices , labelled by @xmath92 , and a series of lower vertices , labelled by @xmath302 . for each combination of @xmath177 and @xmath125
the corresponding vertices are connected by @xmath220 edges .
for example , the matrices @xmath303 , @xmath304 , and @xmath305 in ( [ gdadbdc ] ) give rise to the following diagrams : 1.8 cm ( 1,2.6 ) ( 0,1.3)(0,0)[cr]@xmath303 : ( 0.5,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath306(0.05,-0.9)(-0.1,-0.2)[tr]@xmath307 ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath308(0.05,-0.9)(-0.1,-0.2)[tc]@xmath308 ( 0,0)(0,0)(0.0447,-0.0894)(1,-2)0.4103(0.5,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath309(0.55,-0.9)(-0.1,-0.2)[tc]@xmath134 ( 0.5,0)(0,0)(-0.0447,-0.0894)(-1,-2)0.4103(-0.5,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath134(-0.45,-0.9)(-0.1,-0.2)[]@xmath309 ( 0.5,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath309(0.05,-0.9)(-0.1,-0.2)[]@xmath309 ( 1,2.6 ) ( 0,1.3)(0,0)[cr]@xmath304 : ( 0.5,1.8 ) ( 0,0)(0,0)(-0.05,-0.1)(0,-1)0.8(0.05,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath306(0.05,-0.9)(-0.1,-0.2)[tr]@xmath307 ( 0,0)(0,0)(-0.05,-0.1)(0,-1)0.8(0.05,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath308(0.05,-0.9)(-0.1,-0.2)[tc]@xmath308 ( 0.5,0)(0,0)(-0.05,-0.1)(0,-1)0.8(0.05,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath134(0.05,-0.9)(-0.1,-0.2)[tc]@xmath134 ( 1,2.6 ) ( 0,1.3)(0,0)[cr]@xmath305 : ( 0.5,1.8 ) ( 0,0)(0,0)(0,-0.1118)(1,-2)0.4106(0.0894,-0.0671)(1,-2)0.4106(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath306(0.55,-0.9)(-0.1,-0.2)[]@xmath309 ( 0.5,0)(0,0)(-0.0894,-0.0671)(-1,-2)0.4106(0,-0.1118)(-1,-2)0.4106(0,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath309(-0.45,-0.9)(-0.1,-0.2)[tr]@xmath307 ( 0,0)(0,0)(0,-0.1118)(1,-2)0.4106(0.0894,-0.0671)(1,-2)0.4106(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath308(0.55,-0.9)(-0.1,-0.2)[tc]@xmath134 ( 0.5,0)(0,0)(-0.0894,-0.0671)(-1,-2)0.4106(0,-0.1118)(-1,-2)0.4106(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath134(-0.45,-0.9)(-0.1,-0.2)[tc]@xmath308 the diagrams associated in this way to the elements of @xmath200 , @xmath180 , are special cases of so - called _ bratteli diagrams _ @xcite .
the latter have , in general , several stages picturing the subsequent inclusion matrices associated to an ascending sequence of finite dimensional von - neumann algebras @xmath310 .
for this reason , we refer to the diagram associated to @xmath311 as the bratteli diagram of @xmath190 .
we remark that , due to eq .
( [ geqk ] ) , @xmath190 can not have a zero row .
due to ( [ geqm ] ) , it can not have a zero column either .
accordingly , each vertex of the bratteli diagram of @xmath190 is cut by at least one edge .
let @xmath146 and @xmath155 be elements of @xmath135 . in terms of the bratteli diagram of the variable @xmath190 , eqs .
( [ geqk ] ) , ( [ geqm ] ) , and ( [ geqa ] ) can be rewritten as follows : @xmath312 the main use of bratteli diagrams is to simplify calculations as , for instance , solving the equations determining @xmath275 .
furthermore , some of the arguments in the sequel are easier to formulate on the level of these diagrams than on the level of the corresponding matrices .
in this section , we are going to derive a characterization of direct successor relations in @xmath138 and to formulate operations that generate the direct successors of any given element of @xmath138 .
let @xmath180 .
for any @xmath290 , we define the _ level _ of @xmath190 to be the integer @xmath313 using the quantities @xmath314 we can write @xmath315 due to ( [ geqk ] ) and ( [ geqm ] ) , each row and each column of @xmath190 contain at least one non - zero entry .
it follows that @xmath316 , @xmath317 . hence , due to ( [ gll+l- ] ) , @xmath318 . as for the interpretation , @xmath319 measures , in a sense ,
how much @xmath193 deviates from @xmath192 ( up to permutations ) . on the level of the bratteli diagram of @xmath190
, @xmath319 is twice the number of edges minus the number of vertices , whereas @xmath316 and @xmath320 count the edges at the vertices @xmath177 and @xmath125 , respectively , minus the obligatory one edge per vertex . for later use , we note the following formulae , which follow immediately from ( [ gll+l- ] ) : @xmath321 [ ll1 ] let @xmath322 and let @xmath323 , @xmath324
. then @xmath325 and @xmath326 moreover , @xmath327 or @xmath328 imply equality in ( [ glineq ] ) . _
proof : _ let @xmath146 , @xmath155 , and @xmath329 . by the assumption that @xmath275 and @xmath330 be nonempty
, @xmath202 divides @xmath44 and @xmath331 divides @xmath202 , hence @xmath331 divides @xmath44 . also by this assumption , @xmath332 and @xmath333 , hence @xmath334 . moreover
, one can check that @xmath335 obeys eqs .
( [ geqk ] ) , ( [ geqm ] ) , and ( [ geqa ] ) , where for the last one , ( [ gedd ] ) has to be used . to prove ( [ glineq ] ) , using ( [ gdefl+ ] ) , ( [ gdefl- ] ) , and ( [ gdefl ] ) , we compute @xmath336{rcl } 2\,\sum_{i'=1}^{r ' } { \ell}^+_{i'}\left(\delta'\right ) { \ell}^-_{i'}\left(\delta\right ) & = & 2\,\sum_{i'=1 } \big(\big(\sum_{i''=1}^{r''}\delta'_{i''i'}\big ) - 1 \big ) \big(\big(\sum_{i=1}^r\delta_{i'i}\big )
- 1 \big ) \\
& = & 2\,\big ( \sum_{i''=1}^{r''}\sum_{i'=1}^{r'}\sum_{i=1}^r\delta'_{i''i'}\delta_{i'i } - \sum_{i'=1}^{r'}\sum_{i=1}^r\delta_{i'i } \\ & & - \sum_{i''=1}^{r''}\sum_{i'=1}^{r'}\delta'_{i''i ' } + r'\big ) \\ & = & { \ell}\left(\delta'\delta\right ) - { \ell}\left(\delta\right ) - { \ell}\left(\delta'\right)\ , .
\end{array}\ ] ] since the lhs . of ( [ gldd ] ) is nonnegative , this yields ( [ glineq ] ) .
moreover , if @xmath328 or @xmath327 then , due to ( [ gll+l- ] ) , @xmath337 or @xmath338 , respectively , for all @xmath125 .
hence , the lhs . of ( [ gldd ] ) vanishes , so that equality holds in ( [ glineq ] ) . ' '' '' [ ll2 ] let @xmath274 and let @xmath224 or @xmath308 .
if @xmath275 contains an element of level @xmath79 then all its elements have level @xmath79 . _
proof : _ let @xmath146 , @xmath155 and let @xmath323 .
due to ( [ geqk ] ) and ( [ gdefl+ ] ) , @xmath339 since @xmath340 and @xmath341 for all @xmath177 , ( [ gkl+ ] ) implies @xmath342 by a similar argument we find @xmath343 now assume that @xmath344 , where @xmath224 or @xmath308 .
then at most one of the integers @xmath316 or @xmath320 can be nonzero .
thus , ( [ gl+0 ] ) or ( [ gl-0 ] ) holds . in either case
, the assertion holds for any @xmath345
. then ( [ gll+l-2 ] ) implies @xmath346 . ' '' '' + _ remarks : _ + 1 .
the proof of lemma [ ll2 ] shows that the lemma still holds if one replaces @xmath275 by @xmath200 , for any @xmath180 .
in general , the level function @xmath347 may not be constant on the sets @xmath275 .
for example , let @xmath7 be the trivial @xmath348-bundle over @xmath8 and let @xmath146 , @xmath155 be given by @xmath349 , @xmath350 , @xmath351 and @xmath352 , @xmath353 , @xmath354 .
obviously , @xmath355 and @xmath356 .
one can check that @xmath275 contains the following two inclusion matrices : @xmath357 one has @xmath358 and @xmath359 .
[ ll3 ] let @xmath274 .
the following assertions are equivalent : + ( a ) @xmath149 and @xmath161 are equivalent .
+ ( b ) @xmath275 contains an element of level @xmath360 .
+ ( c ) @xmath275 is nonempty and all of its elements have level @xmath360 . _
proof : _ due to lemma [ ll2 ] , ( b ) @xmath361 ( c ) .
hence , it suffices to prove ( a ) @xmath361 ( b ) .
let @xmath146 , @xmath155 .
first , assume that there exist @xmath323 such that @xmath328
. then @xmath362 for all @xmath177 and @xmath337 for all @xmath125 .
that means , each row and each column contains exactly one nonzero entry and this entry has value @xmath308 .
it follows that @xmath190 is square , i.e. , @xmath363 , and that there exists a permutation @xmath140 of @xmath141 such that @xmath364 as an immediate consequence , @xmath365 since @xmath323 , ( [ gdperm2 ] ) implies @xmath366 , @xmath367 , and @xmath263 .
in particular , @xmath368 , hence @xmath369 .
it follows @xmath370 .
thus , @xmath149 and @xmath161 are equivalent .
conversely , assume that @xmath370 and that there exist a permutation @xmath140 of @xmath141 such that @xmath366 and @xmath367 .
since , in particular , @xmath368 , @xmath202 and @xmath44 coincide .
thus , trivially , @xmath202 divides @xmath44 and @xmath263 .
hence , if we find a solution @xmath190 of eqs .
( [ geqk ] ) , ( [ geqm ] ) , and ( [ geqa ] ) then @xmath323 . due to ( [ gdperm2 ] ) ,
such a solution is given by the matrix ( [ gdperm ] ) . by construction
, it has level @xmath360 . ' '' '' let @xmath371 . in this subsection
, we are going to formulate operations that create new elements of @xmath135 out of @xmath149 .
these operations will be used to prove a decomposition lemma in subsection [ ssdecomp ] and , later on , to generate direct successors .
+ _ splitting :
_ choose @xmath372 such that @xmath373 .
choose a decomposition @xmath374 with strictly positive integers @xmath375 .
define sequences of length @xmath376 @xmath377 since the greatest common divisor @xmath378 of @xmath379 divides @xmath269 , we can furthermore define @xmath380 denote @xmath381 and @xmath382 .
we claim that @xmath383 .
it is easily seen that @xmath384 and @xmath385 .
consequently , it suffices to check that @xmath386 and @xmath387 obey eqs .
( [ gkmj ] ) and ( [ gkpj ] ) .
first , consider ( [ gkmj ] ) .
let the integer @xmath79 be such that @xmath388 . using ( [ glqd3 ] ) and ( [ gle2 ] ) as well as taking into account that ( [ gkmj ] ) holds for @xmath389 and @xmath390 , we compute @xmath391 expanding the rhs . according to ( [ ge2 ] ) yields @xmath392 where the penultimate equality is due to the fact that @xmath378 divides both @xmath393 and @xmath394 .
now consider ( [ gkpj ] ) .
using commutativity of the cup product in even degree , one can check that @xmath395 .
since ( [ gkpj ] ) holds for @xmath389 , it holds for @xmath386 .
this proves @xmath383 .
we say that @xmath396 arises from @xmath149 by a splitting of the @xmath397th member .
+ _ merging :
_ choose @xmath398 such that @xmath399 .
define sequences of length @xmath400 @xmath401 where @xmath402 indicates that the entry is omitted , as well as @xmath403 denote @xmath404 and @xmath405 .
let us show @xmath383 .
as in the case of splitting , one can immediately verify that @xmath384 , @xmath406 , and @xmath407 .
it follows that @xmath408 obeys eq .
( [ gkpj ] ) . due to @xmath409
, a similar calculation shows @xmath410 . since also @xmath411
, we obtain @xmath412 .
thus , @xmath383 .
we say that @xmath396 arises from @xmath149 by merging the @xmath413th and the @xmath414th member .
+ _ remark : _ it may happen that for certain elements of @xmath135 no splittings or no mergings can be applied . amongst these elements
are , for example , those with @xmath415 ( no splitting ) and those having pairwise distinct @xmath185 ( no merging ) .
[ lopn ] let @xmath416 .
@xmath396 can be obtained from @xmath149 by a splitting of the @xmath397th member if and only if @xmath417 contains an element with bratteli diagram @xmath418{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[]{\scriptsize$$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 + 1~~$ } } } } { \put(2,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 + 1$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$~~i_0 + 2 $ } } } } \put(2.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(3,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r+1 $ } } } } } \end{picture } \end{array } \ ] ] @xmath396 can be obtained from @xmath149 by merging the @xmath413th and the @xmath414th member if and only if @xmath417 contains an element with bratteli diagram @xmath419{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 $ } } } } { \put(2,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 + 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 + 1 $ } } } } \put(2.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(2.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 - 1 $ } } } } { \put(3.5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0894,-0.0447){\line(-2,-1){1.821}}}{\put(-2,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2$}}\put(-1.95,-0.9){\makebox(-0.1,-0.2)[]{\scriptsize$$ } } } } { \put(4,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 + 1$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 $ } } } } \put(4.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(4,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r-1 $ } } } } } \end{picture } \end{array } \ ] ] _ proof : _ assume @xmath420 , @xmath421 . since the proofs for the cases of splitting and merging are completely analogous , we only give the first one .
first , assume that @xmath396 arises from @xmath149 by a splitting of the @xmath397th member .
then , by construction , @xmath378 divides @xmath44 and @xmath422 .
hence the matrix given by the bratteli diagram ( [ gspli0 ] ) belongs to @xmath417 iff it satisfies eqs .
( [ geqk ] ) , ( [ geqm ] ) , and ( [ geqa ] ) . by the help of eqs .
( [ gdgrk])([gdgra ] ) , this can be easily checked on diagram level .
conversely , assume that @xmath417 contains an element with bratteli diagram ( [ gspli0 ] )
. then , in particular , condition ( a ) of theorem [ tpo ] holds , i.e. , @xmath378 divides @xmath44 and @xmath422 .
an inspection of ( [ gdgrk])([gdgra ] ) shows that @xmath423 , @xmath424 , and @xmath425 , whereas @xmath426 , @xmath427 , @xmath428 for @xmath429 and @xmath430 , @xmath431 , @xmath432 for @xmath433 .
thus , @xmath396 is obtained from @xmath149 by a splitting of the @xmath397th member according to the decomposition @xmath434 . ' '' '' [ ldecomp ] let @xmath274 and let @xmath323 .
if @xmath435 then there exist @xmath383 and @xmath436 , @xmath437 such that @xmath438 and @xmath439 .
_ proof : _ to begin with , assume that there exist @xmath397 such that @xmath440 .
choose @xmath441 such that @xmath442 .
we have the following estimate : @xmath443 this shows that @xmath444 is a decomposition into strictly positive integers .
we define @xmath396 to be the element of @xmath135 obtained from @xmath149 by the corresponding splitting operation .
furthermore , we define @xmath445 to be the @xmath446-matrix @xmath447 and @xmath448 to be the @xmath449-matrix @xmath450 we notice that @xmath445 has bratteli diagram ( [ gspli0 ] ) . hence , due to lemma [ lopn ] , @xmath436 . from the diagram we read off that @xmath439 . moreover , by means of a direct computation using ( [ gdecompdo ] ) and ( [ gdecompdop ] )
one can check that @xmath451 .
thus , it remains to prove that @xmath437 .
this amounts to the following items : \(a ) @xmath202 divides @xmath378 : we recall from ( [ gsplm ] ) that @xmath452 by assumption , @xmath202 divides @xmath269 , hence all the @xmath185 . by definition
, it also divides @xmath453 .
\(b ) @xmath454 : according to ( [ gsplx ] ) , @xmath455 . here
the last equality holds by assumption .
\(c ) @xmath456 : using that @xmath436 and @xmath323
we compute @xmath457 .
\(d ) @xmath458 : this has to be checked by a direct computation using ( [ gdecompdop ] ) and ( [ gdecompmo ] ) .
\(e ) @xmath459 : using the same arguments as for ( c ) , as well as ( [ gedd ] ) , we obtain @xmath460 .
this proves @xmath437 .
now assume that @xmath362 for all @xmath177 .
then in each column of @xmath190 there exists exactly one nonzero entry , and this entry has value @xmath308 . on the other hand , since @xmath461 , there exists @xmath441 such that @xmath462 .
this means , the row labelled by @xmath463 has at least two entries of value @xmath308 .
therefore , we find two columns , labelled by @xmath464 , such that @xmath465 then @xmath466 , @xmath467 , hence @xmath399 .
thus , we can define @xmath396 to be the element of @xmath135 obtained by merging the @xmath413th and the @xmath414th member of @xmath149 .
moreover , we define @xmath445 to be the @xmath468-matrix @xmath469 and @xmath448 to be the @xmath470-matrix @xmath471 @xmath445 now having bratteli diagram ( [ gmrgi1i2 ] ) , @xmath436 by lemma [ lopn ] .
analogously to the first case one can check that @xmath472 , @xmath451 , and @xmath437 .
this proves the lemma . ' '' '' [ tds ] let @xmath277 .
the following assertions are equivalent .
\(a ) @xmath280 is a direct successor of @xmath279 .
\(b ) there exist representatives @xmath149 and @xmath161 of @xmath279 and @xmath280 , respectively , such that @xmath275 contains an element of level @xmath308 .
\(c ) for any representatives @xmath149 , @xmath161 of @xmath279 , @xmath280 , respectively , @xmath275 is nonempty and its elements have level @xmath308 .
_ proof : _
\(a ) @xmath281 ( c ) : let @xmath149 and @xmath161 be representatives of @xmath279 and @xmath280 , respectively . since @xmath278 , due to corollary [ cpo ] , there exists @xmath323 . since @xmath473 ,
due to lemma [ ll3 ] , @xmath474 .
then lemma [ ldecomp ] implies that there exist @xmath383 and @xmath436 , @xmath475 such that @xmath476 and @xmath477 .
let @xmath478 denote the equivalence class of @xmath396 .
we have @xmath479 . according to lemma [ ll3 ] , @xmath480 .
it follows that @xmath481 .
hence , again due to lemma [ ll3 ] , @xmath482 .
then the sharpened version of ( [ glineq ] ) implies @xmath483 .
\(c ) @xmath281 ( b ) : obvious .
\(b ) @xmath281 ( a ) : let @xmath274 be given as assumed .
let @xmath484 such that @xmath485 .
for any representative @xmath396 of @xmath478 , there exist @xmath486 and @xmath475 . due to lemma [ ll1 ] , @xmath487 . since , by assumption , this set contains an element of level @xmath308 , lemma [ ll2 ] yields @xmath488 .
then ( [ glineq ] ) requires that either @xmath489 or @xmath482 . according to lemma [ ll3 ] , in the first case , @xmath490 , whereas in the second case , @xmath481 .
this shows that @xmath280 is a direct successor of @xmath279 . ' '' '' + _ the bratteli diagram of an inclusion matrix of level @xmath308 : _ let @xmath274 and let @xmath323 .
assume that @xmath472 .
then either there exists @xmath397 such that @xmath491 and @xmath362 for all @xmath492 and @xmath337 for all @xmath125 , or there exists @xmath463 such that @xmath493 and @xmath337 for all @xmath494 and @xmath362 for all @xmath177 .
accordingly , the bratteli diagram of @xmath190 is given by @xmath495{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 + 1 $ } } } } \put(2,0){\makebox(0,0)[cc]{$\cdots$ } } \put(2.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(2.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 - 1$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 $ } } } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0832,-0.0554){\line(-3,-2){1.3336}}}{\put(-1.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0$}}\put(-1.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 $ } } } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(0.0894,-0.0447){\line(2,-1){1.821}}}{\put(2,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[]{\scriptsize$$}}\put(2.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 $ } } } } { \put(3.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 + 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 + 1
$ } } } } \put(4,0){\makebox(0,0)[cc]{$\cdots$ } } \put(4,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(4.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 - 1 $ } } } } { \put(5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 + 1 $ } } } } \put(5.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(6,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(6,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r+1 $ } } } } } \end{picture } \end{array } \ ] ] for some @xmath496 , or by @xmath497{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0832,-0.0554){\line(3,-2){1.3336}}}{\put(1.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1$}}\put(1.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 $ } } } } { \put(2,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 + 1$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 $ } } } } \put(2.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(2,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 - 1 $ } } } } { \put(3.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 + 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 + 1 $ } } } } \put(4,0){\makebox(0,0)[cc]{$\cdots$ } } \put(4,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(4.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 - 1 $ } } } } { \put(5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0894,-0.0447){\line(-2,-1){1.821}}}{\put(-2,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2$}}\put(-1.95,-0.9){\makebox(-0.1,-0.2)[]{\scriptsize$$ } } } } { \put(5.5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 + 1$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 $ } } } } \put(6,0){\makebox(0,0)[cc]{$\cdots$ } } \put(5.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(6.5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r-1 $ } } } } } \end{picture } \end{array } \ ] ] for some @xmath398 , respectively .
in particular , in the first case , @xmath498 and in the second case , @xmath499 .
[ tdsgen ] let @xmath500 and let @xmath149 be a representative of @xmath279 .
then the direct successors of @xmath279 are obtained by applying all possible splittings and mergings to @xmath149 and passing to equivalence classes . _
proof : _ as an immediate consequence of lemma [ lopn ] and theorem [ tds ] , any element of @xmath138 generated in the way proposed is a direct successor of @xmath279 .
conversely , let @xmath280 be a direct successor of @xmath279 . choose a representative @xmath161 of @xmath280 . due to theorem [ tds ]
, @xmath275 contains an element of level @xmath308 . as noted above , the bratteli diagram of such an element is of the form ( [ gdgrspl ] ) or ( [ gdgrmrg ] ) . by a permutation of the lower vertices we can turn this diagram into ( [ gspli0 ] ) or ( [ gmrgi1i2 ] ) , respectively .
this corresponds to the passage from @xmath161 to another representative @xmath396 of @xmath280 .
it is immediately seen that the matrix given by the diagram with permuted lower vertices belongs to @xmath417 .
then lemma [ lopn ] implies that @xmath396 can be obtained from @xmath149 by a splitting or a merging , respectively .
this proves the theorem . ' '' '' let @xmath7 be a principal @xmath501-bundle .
let @xmath145 , @xmath146 , where @xmath502 .
then @xmath73 has components @xmath503 , @xmath504 .
( one may wish to recall from the example of section [ parord ] that @xmath65 has isomorphism type @xmath287 . )
we are going to determine the direct successors of the equivalence class of @xmath149 .
let us begin with splitting operations . for @xmath505 , the only possible
splitting is given by the decomposition @xmath506 .
it yields @xmath507 , where @xmath508 , @xmath509 , and @xmath510 .
the passage from @xmath149 to @xmath511 can be represented conveniently in a bratteli diagram whose vertices are labelled by the respective quantities @xmath512 and @xmath513 ( rather than by the mere number @xmath177 ) : 1.8 cm ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br ] @xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr ] @xmath515{c}l_a^\circ\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 0,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath514{c}\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 1,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_2\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,2\\#7\end{array}$ ] ( 1,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] for @xmath516 , a similar splitting operation creates @xmath517 , given by the labelled bratteli diagram 1.8 cm ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br ] @xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr ] @xmath515{c}l_b^\circ\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,2\\#7\end{array}$ ] ( 1,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_2\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 1,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath514{c}\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 1,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] as for merging operations , the only choice for @xmath413 , @xmath414 is @xmath518 , @xmath519 .
this yields @xmath520 : 1.8 cm ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br ] @xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr ] @xmath515{c}l_c^\circ\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}2,2\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_2\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2)[]@xmath515{c}\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] next , we have to pass to equivalence classes .
generically , @xmath511 , @xmath517 , and @xmath520 generate their own classes .
however , while @xmath520 can never be equivalent to @xmath511 or @xmath517 , the latter are equivalent iff @xmath521 . in order to see for which bundles @xmath7 this can happen , consider eqs .
( [ gkmj ] ) and ( [ gkpj ] ) .
the first one requires @xmath522 to be a torsion element .
then , due to @xmath523 , the second one implies @xmath524 .
thus , @xmath511 and @xmath517 can be ( occasionally ) equivalent only if @xmath7 is trivial .
in this section , we formulate operations to generate the direct predecessors of any given element of @xmath138 .
direct predecessors are , for our purposes , more interesting than direct successors for at least two reasons .
first , they allow one to reconstruct the set @xmath138 together with its partial ordering from the unique maximal element ( which , in terms of howe subbundles , is given by @xmath7 itself ) .
second , on the level of the stratification of the gauge orbit space , predecessors correspond to strata of higher symmetry . in the preceding section we have been able to create all direct successors of a given element of @xmath138 from one and the same representative .
this was achieved by using the freedom in the choice of the representatives of the direct successors .
since here we wish to proceed likewise , we have to carry this freedom from the level of successors to that of predecessors .
for this reason , the inverted operations are not just splitting and merging read backwards .
they rather take the following form .
let @xmath145 , @xmath146 .
+ _ inverse splitting : _ choose @xmath398 such that @xmath525 and @xmath526 . define sequences of length @xmath400 @xmath527 we note that @xmath44 divides the greatest common divisor @xmath378 of @xmath379 , so that @xmath528 is well - defined
choose @xmath529 such that @xmath530 and @xmath531 denote @xmath381 and @xmath382 .
we check that @xmath383 : by construction , @xmath384 and @xmath406 .
due to ( [ ginvsplx ] ) , @xmath386 and @xmath387 obey eq .
( [ gkmj ] ) . using @xmath526
one can check that @xmath532 .
hence , since @xmath389 obeys eq .
( [ gkpj ] ) , so does @xmath386 .
this proves @xmath383 .
we say that @xmath396 arises from @xmath149 by an inverse splitting of the @xmath413th and the @xmath414th member .
+ _ inverse merging _ choose @xmath372 such that @xmath533 . choose a decomposition @xmath534 with strictly positive integers @xmath535 .
choose cohomology elements @xmath536 such that @xmath537 for @xmath538 , @xmath539 , and @xmath540 define sequences of length @xmath376 @xmath541 as well as @xmath542 denote @xmath381 and @xmath382 . to see that @xmath383 , we check @xmath384 and @xmath543 . using ( [ ginvmrga ] )
one can verify that @xmath544 .
consequently , @xmath386 obeys eq .
( [ gkpj ] ) . a similar calculation , using , in addition , @xmath409 , shows that @xmath545 . since also @xmath546 , @xmath386 and @xmath387 obey eq .
( [ gkmj ] ) .
we say that @xmath396 arises from @xmath149 by an inverse merging of the @xmath397th member .
+ _ remark : _ like for the operations of splitting and merging , for some of the elements of @xmath135 , inverse splitting or inverse merging may not be applicable .
in particular , it may happen that there does not exist a solution @xmath387 of eq .
( [ ginvsplx ] ) .
[ linvopn ] let @xmath416 .
@xmath396 arises from @xmath149 by an inverse splitting of the @xmath413th and the @xmath414th member if and only if @xmath547 contains an element with bratteli diagram @xmath548{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0894,-0.0447){\line(2,-1){1.821}}}{\put(2,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[]{\scriptsize$$}}\put(2.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 $ } } } } { \put(2,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_1 + 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_1 + 1 $ } } } } \put(2.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(2.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(3,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 - 1 $ } } } } { \put(3.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_2$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_2 + 1 $ } } } } \put(4,0){\makebox(0,0)[cc]{$\cdots$ } } \put(4.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(4.5,0){{\put(0,0){\circle*{0.06}}}{\put(0.0447,-0.0894){\line(1,-2){0.4103}}}{\put(0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(0.55,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r+1 $ } } } } } \end{picture } \end{array } \ ] ] @xmath396 arises from @xmath149 by an inverse merging of the @xmath397th member if and only if @xmath547 contains an element with bratteli diagram @xmath549{\scriptsize$1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$1 $ } } } } \put(0.5,0){\makebox(0,0)[cc]{$\cdots$ } } \put(0.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(1,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 - 1$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 - 1 $ } } } } { \put(1.5,0){{\put(0,0){\circle*{0.06}}}{\put(0,-0.1){\line(0,-1){0.8}}}{\put(0,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0$}}\put(0.05,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 $ } } } } { \put(2,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$i_0 + 1~~$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[]{\scriptsize$$ } } } } { \put(2.5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$~~i_0 + 2$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$i_0 + 1 $ } } } } \put(3,0){\makebox(0,0)[cc]{$\cdots$ } } \put(2.5,-1){\makebox(0,0)[cc]{$\cdots$ } } { \put(3.5,0){{\put(0,0){\circle*{0.06}}}{\put(-0.0447,-0.0894){\line(-1,-2){0.4103}}}{\put(-0.5,-1){\circle*{0.06}}}\put(0.05,0.1){\makebox(-0.1,-0.2)[bc]{\scriptsize$r$}}\put(-0.45,-0.9){\makebox(-0.1,-0.2)[tc]{\scriptsize$r-1 $ } } } } } \end{picture } \end{array } \ ] ] _ proof : _ the proof is completely analogous to that of lemma [ lopn ] and shall be omitted . ' '' '' [ tdpgen ] let @xmath500 and let @xmath149 be a representative of @xmath279 .
then the direct predecessors of @xmath279 are obtained by applying all possible inverse splittings and inverse mergings to @xmath149 and passing to equivalence classes . _
proof : _ the proof is completely analogous to that of theorem [ tdsgen ] . the only difference is that here we are allowed to pass to another representative of the _ predecessor _ , i.e. , to permute the _ upper _ vertices in the diagrams ( [ gdgrspl ] ) and ( [ gdgrmrg ] ) , thus arriving at ( [ ginvspli1i2 ] ) and ( [ ginvmrgi0 ] ) . ' '' '' as in subsection [ sdssex ] , let @xmath7 be a principal @xmath501-bundle and let @xmath145 , @xmath146 , where @xmath283 .
we are going to determine the direct predecessors of the equivalence class of @xmath149 .
inverse splittings can be applied only if @xmath521 . in this case , for any solution @xmath550 of the system of equations @xmath551 we obtain an element @xmath552 , where @xmath553 and @xmath554 .
the passage from @xmath149 to @xmath396 can be summarized in the labelled bratteli diagram 1.8 cm ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br ] @xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr ] @xmath515{c}l\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,2\\#7\end{array}$ ] ( 0,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath514{c}\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,2\\#7\end{array}$ ] ( 0,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi^\circ\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] that has to be read upwards .
each @xmath396 generates its own equivalence class .
due to @xmath555 , inverse mergings can not be applied to @xmath149 .
thus , in the case @xmath521 the direct predecessors of the equivalence class of @xmath149 are labelled by the solutions of eqs .
( [ gexinvmrg1 ] ) and ( [ gexinvmrg2 ] ) , whereas in the case @xmath556 direct predecessors do not exist .
recall from subsection [ sdssex ] that the first case can only occur if @xmath7 is trivial .
as another example , consider an element @xmath161 of @xmath135 , @xmath155 , where @xmath557 .
inverse mergings can be applied and yield elements @xmath558 as follows : 1.8 cm ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr]@xmath515{c } l'\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}{\alpha_1'}^\circ\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}2,2\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}{\alpha_2'}^\circ\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2)[]@xmath515{c}\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}{\xi'}^\circ=\xi'\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] here @xmath559 , @xmath504 , such that @xmath560 .
when passing to equivalence classes , elements @xmath558 with @xmath561 and @xmath562 have to be identified .
since @xmath161 does not allow inverse splittings , there are no more direct predecessors .
the gauge orbit types for @xmath563 , i.e. , the set @xmath138 for a principal @xmath563-bundle @xmath7 over @xmath8 , was calculated in @xcite by solving eqs .
( [ gkmj ] ) and ( [ gkpj ] ) for all @xmath192 .
here we are going to recover this result using a different technique that will also yield the partial ordering of orbit types .
a partially ordered set can be reconstructed either ( a ) from its minimal elements by successively determining direct successors , or ( b ) from its maximal elements by successively determining direct predecessors . in the case of @xmath138
, there exists a unique maximal element , namely the class corresponding to the bundle @xmath7 itself .
minimal elements are , in general , not unique . in fact
, their number can be infinite .
thus , the preferred algorithm is ( b ) .
( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr]@xmath515{c } l_{{\rm max}}\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1^\circ\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}2,1\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_2^\circ\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2)[]@xmath515{c}\\#7\end{array}$ ] ( 1,0)(0,0)(-0.0707,-0.0707)(-1,-1)0.8586(-1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi^\circ=0\\#3\end{array}$](-0.95,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] where @xmath568 such that @xmath569 . sorting by degree yields the equations @xmath570 and @xmath571 .
we parametrize @xmath572 where @xmath573 has to obey @xmath574 the passage to equivalence classes leads to an identification of solutions @xmath575 and @xmath576 .
we note that the howe subgroup labelled by @xmath577 is the toral subgroup @xmath99 of @xmath563 and that the parameter @xmath575 is just the first chern class of the corresponding reduction of @xmath7 . by virtue of this transliteration , eq .
( [ gex1 ] ) coincides with the result given in @xcite .
next , consider the direct predecessors of the classes generated by @xmath396 .
inverse mergings can not be applied .
inverse splittings can be applied provided @xmath578 , i.e. , @xmath579 .
then for any solution @xmath580 of the equation @xmath581 we obtain an element @xmath582 : ( 2.7,2.6 ) ( 0,1.8 ) ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[br]@xmath514{c}\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tr]@xmath515{c } l^\circ\hspace{1.1cm}\\#7\end{array}$ ] ( 0,0)(0,0)(0,-0.1)(0,-1)0.8(0,-1)(0.05,0.1)(-0.1,-0.2)[bc]@xmath514{c}\alpha_1^\circ\\#3\end{array}$](0.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 0,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[]@xmath514{c}\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2)[tc]@xmath515{c}1,1\\#7\end{array}$ ] ( 0,0)(0,0)(0.0707,-0.0707)(1,-1)0.8586(1,-1)(0.05,0.1)(-0.1,-0.2)[bl]@xmath514{c}\hspace*{1cm}\xi^{\circ\circ}\\#3\end{array}$](1.05,-0.9)(-0.1,-0.2 ) [ tl]@xmath515{c}\phantom{1,2}\\#7\end{array}$ ] now let us draw hasse diagrams of @xmath138 for specific space - time manifolds @xmath8 . in the following ,
vertices stand for the elements of @xmath138 and edges indicate the relation left vertex @xmath586 right vertex. when viewing the elements of @xmath138 as howe subbundles , the vertex on the rhs
. represents the class corresponding to @xmath7 itself , the vertices in the middle and on the lhs
. represent reductions of @xmath7 to the howe subgroups @xmath99 and @xmath584 , respectively . when viewing the elements of @xmath138 as orbit types , or strata of the gauge orbit space , the vertex on the rhs
. represents the generic stratum , whereas the vertices in the middle and on the lhs .
represent @xmath99-strata and @xmath563-strata . here the names @xmath99-stratum and @xmath563-stratum mean that the stratum consists of ( orbits of ) connections whose stabilizers are isomorphic to @xmath99 or @xmath563 , respectively . since @xmath588 , eq .
( [ gex1 ] ) can be solved iff @xmath524 , i.e. , iff @xmath7 is trivial .
the solution is @xmath589 . then eq . ( [ gex2 ] )
is trivially satisfied by @xmath590 . due to @xmath591
, there are no more solutions .
thus , in the case @xmath524 , the hasse diagram of @xmath138 is on the level of gauge orbit types , the result means that in the sector of vanishing topological charge the gauge orbit space decomposes into the generic stratum , a @xmath99-stratum , and a @xmath563-stratum .
if , on the other hand , a topological charge is present , only the generic stratum survives . to perform the first step in the reconstruction procedure , let @xmath594 and @xmath595 be generators of @xmath596 and @xmath597 , respectively .
due to the knneth theorem , @xmath598 is generated by @xmath599 and @xmath600 , whereas @xmath601 is generated by @xmath602 .
here @xmath603 denotes the cohomology cross product . writing @xmath604 with @xmath605 , eq .
( [ gex1 ] ) becomes @xmath606 if @xmath524 , there are two series of solutions : @xmath607 and @xmath608 as well as @xmath609 and @xmath610 . due to
@xmath591 , eq . ( [ gex2 ] )
tells us that out of the elements just obtained only that labelled by @xmath611 has a direct predecessor .
thus , in the case @xmath524 the hasse diagram of @xmath138 is ( 2,2.55 ) ( 0,1.3 ) ( 0,0)(0,0)(0.1,0)(1,0)0.8 ( 0.05,0.1)(-0.1,-0.2)[bc]@xmath309 ( 1,0.75)(0.08,-0.06)(0.1,-0.075)9 ( 1,0.75)(0,-0.1)(0,0.1)3 ( 1,0.5)(0,0)(0.0894,-0.0447)(2,-1 ) 0.821(0.05,0.1)(-0.1,-0.2)[]@xmath309 ( 1.01,0.46)(0.05,0.1)(-0.1,-0.2)[br ] @xmath612 ( 1,0.25)(0,0)(0.0968,-0.0242)(4,-1 ) 0.8064(0.05,0.1)(-0.1,-0.2)[]@xmath309 ( 1.01,0.21)(0.05,0.1)(-0.1,-0.2)[br ] @xmath613 ( 1,0)(0,0)(0.1,0)(1,0)0.8 ( 0.05,0.1)(-0.1,-0.2)[]@xmath309 ( 1.01,-0.04)(0.05,0.1)(-0.1,-0.2)[br ] @xmath614 ( 1,-0.25)(0,0)(0.0968,0.0242)(4,1)0.8064 ( 0.05,0.1)(-0.1,-0.2)[]@xmath309 ( 1.01,-0.29)(0.05,0.1)(-0.1,-0.2)[br ] @xmath615 ( 1,-0.5)(0,0)(0.0894,0.0447)(2,1)0.821 ( 0.05,0.1)(-0.1,-0.2)[]@xmath309 ( 1.01,-0.54)(0.05,0.1)(-0.1,-0.2)[br ] @xmath616 ( 1,-0.75)(0,-0.1)(0,0.1)3 ( 1,-0.75)(0.08,0.06)(0.1,0.075)9 ( 2,0)(0,0)(0.05,0.1)(-0.1 , -0.2)[bc]@xmath309 if @xmath619 , @xmath620 , then the solutions of ( [ gex1a ] ) are @xmath621 and @xmath622 , where @xmath623 runs through the ( positive and negative ) divisors of @xmath79 . for none of these solutions , ( [ gex2 ] ) is solvable .
hence , here the hasse diagram is ( 2,2.05 ) ( 0,1.05 ) ( 1,0.5)(0,0)(0.0894,-0.0447)(2,-1 ) 0.821(0.05,0.1)(-0.1,-0.2)[cr]@xmath624 ( 1,0.25)(0.0968,-0.0242)(0.1,-0.025)9 ( 1,0.25)(0,-0.1)(0,0.1)3 ( 1,0)(0,0)(0.1,0)(1,0)0.8 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath625 ( 1,-0.25)(0,-0.1)(0,0.1)3 ( 1,-0.25)(0.0968,0.0242)(0.1,0.025)9 ( 1,-0.5)(0,0)(0.0894,0.0447)(2,1)0.821 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath626 ( 2,0)(0,0)(0.05,0.1)(-0.1 , -0.2)[bc]@xmath309 recall that @xmath630 and @xmath631 . let @xmath632 be a generator of @xmath633 and let @xmath634 be a generator of @xmath635 . due to the knneth theorem ,
@xmath598 is generated by @xmath636 .
we write @xmath637 due to @xmath638 , @xmath639 . hence , eq . ( [ gex1 ] ) is solvable iff @xmath524 , in which case the solutions are given by @xmath640 . since
when passing to equivalence classes we have to identify @xmath575 and @xmath576 , i.e. , @xmath641 and @xmath642 , the direct predecessors are labelled by elements of @xmath643 . next ,
consider the second step of the reconstruction procedure .
let @xmath644 , @xmath645 , and @xmath646 be generators of @xmath647 , @xmath648 , and @xmath649 , respectively .
then , again due to the knneth theorem , @xmath650 is generated by @xmath651 and @xmath652 .
moreover , one can check that @xmath653 decomposing @xmath654 and using ( [ gexdecompa ] ) and ( [ gexbock ] ) , ( [ gex2 ] ) becomes @xmath655 thus , only the elements labelled by @xmath607 and @xmath656 have direct predecessors .
these are given by the values @xmath657 , @xmath658 and @xmath659 , @xmath658 , respectively .
( 2,2.05 ) ( 0,1.05 ) ( 0,0.5)(0,0)(0.1,0)(1,0)0.8 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath660 ( 0,0.25)(0,0)(0.0968,0.0242)(4,1)0.8064 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath661 ( 0,-0.25)(0,0)(0.0968,-0.0242)(4,-1 ) 0.8064(0.05,0.1)(-0.1,-0.2)[cr]@xmath662 ( 0,-0.5)(0,0)(0.1,0)(1,0)0.8 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath663 ( 1,0.5)(0,0)(0.0894,-0.0447)(2,-1 ) 0.821(0.05,0.1)(-0.1,-0.2)[bc]@xmath360 ( 1,0.25)(0,0)(0.0968,-0.0242)(4,-1 ) 0.8064(0.05,0.1)(-0.1,-0.2)[cr]@xmath664 ( 1,0)(0,-0.1)(0,0.1)3 ( 1,0)(0.1,0)(0.1,0)9 ( 1,-0.25)(0,0)(0.0968,0.0242)(4,1)0.8064 ( 0.05,0.1)(-0.1,-0.2)[cr]@xmath665 ( 1,-0.5)(0,0)(0.0894,0.0447)(2,1)0.821 ( 0.05,0.1)(-0.1,-0.2)[tc]@xmath666 ( 2,0)(0,0)(0.05,0.1)(-0.1 , -0.2)[bc]@xmath309 here the vertices on the lhs
. are labelled by @xmath667 , whereas those in the middle are labelled by @xmath641 . in the case
@xmath592 , @xmath138 is trivial .
again , the interpretation in terms of strata of the gauge orbit space goes along the lines of the case @xmath587 above .
+ to conclude , let us remark that , while for @xmath563 the picture is relatively simple , already for @xmath668 the partial ordering becomes rather involved , and the hasse diagrams representing it are very complex .
99 asorey , m. , falceto , f. , lpez , j.l . ,
luzn , g. : nodes , monopoles , and confinement in @xmath5-dimensional gauge theories .
b * 345 * , 125130 ( 1995 ) atiyah , m.f . ; singer , i.m .
: dirac operators coupled to vector fields .
usa * 81 * , 25972600 ( 1984 ) bredon , g.e . : _
introduction to compact transformation groups_. new york : academic press , 1972 bratteli , o. : inductive limits of finite dimensional @xmath669-algebras .
math . soc .
* 171 * , 195234 ( 1972 ) bredon , g.e . : _ topology and geometry_. berlin , heidelberg , new york : springer , 1993 emmrich , c. , rmer , h. : orbifolds as configuration spaces of systems with gauge symmetries . commun .
phys . * 129 * , 6994 ( 1990 ) gelbrich , g. : on pointwise conjugate homomorphisms of compact lie groups .
seminar sophus lie * 1 * 217223 , ( 1991 ) , heldermann , 1991 heil , a. , kersch , a. , papadopoulos , n.a . ,
reifenhuser , b. , scheck , f. : anomalies from nonfree action of the gauge group .
* 200 * , 206215 ( 1990 ) isham , c.j .
: space - time topology and spontaneous symmetry breaking . j. phys .
a * 14 * , 2943 - 2956 ( 1981 ) kobayashi , s. , nomizu , k. : _ foundations of differential geometry_. vol . i. new york : wiley interscience , 1963 kondracki , w. , rogulski , j. : on the notion of stratification .
institute of mathematics , polish academy of sciences , preprint 281 , warszawa , 1983 kondracki , w. , rogulski , j. : on the stratification of the orbit space for the action of automorphisms on connections .
dissertationes mathematicae * 250*. warszawa : panstwowe wydawnictwo naukowe , 1986 mitter , p.k .
, viallet , c .- m . : on the bundle of connections and the gauge orbit manifold in yang - mills theory .
phys . * 79 * , 457472 ( 1981 ) rudolph , g. , schmidt , m. , volobuev , i.p . :
classification of gauge orbit types for @xmath670-gauge theories .
preprint math - ph/0003044 , submitted to commun .
singer , i.m . : some remarks on the gribov ambiguity .
phys . * 60 * , 712 ( 1978 ) steenrod , n. : _ the topology of fibre bundles_. princeton , nj : princeton university press , 1951 witten , e. : an @xmath671-anomaly .
lett . * 117 b
* , 324328 ( 1982 ) | the natural partial ordering of the orbit types of the action of the group of local gauge transformations on the space of connections in space - time dimension @xmath0 is investigated . for that purpose , a description of orbit types in terms of cohomology elements of space - time , derived earlier , is used .
it is shown that , on the level of these cohomology elements , the partial ordering relation is characterized by a system of algebraic equations .
moreover , operations to generate direct successors and direct predecessors are formulated . the latter allow to successively reconstruct the set of orbit types , starting from the principal type .
+ + _ subj . class . : _ differential geometry + _ 2000 msc : _ 53c05 ; 53c80 + _ keywords : _ gauge orbit space ; gauge orbit types ; stratification 1 cm institute for theoretical physics + university of leipzig + augustusplatz 10 + 04109 leipzig & germany + e - mail : [email protected] + fax : + 49 - 341 - 97/32548 + @xmath1nuclear physics institute , moscow state university + 119899 moscow & russia |
in recent years , an increasing number of objects have been discovered that occupy asteroid - like orbits in the main asteroid belt but have shown evidence of comet - like activity , typically in the form of transient comet - like dust emission .
the suspected sources of this dust emission vary .
some instances of activity are believed to result from comet - like sublimation of volatile sub - surface ice ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the objects exhibiting this type of activity have come to be known as main - belt comets ( mbcs ; * ? ? ?
* ) . for most mbcs ,
the presence of gas is only inferred by the presence and behavior of visible dust emission and is not directly detected . however , a direct detection of water vapor outgassing has recently been made for main - belt object ( 1 ) ceres by the @xmath6 @xcite , marking the first time that sublimation on a main - belt object has been unambiguously detected . in other instances ,
apparent comet - like dust emission is found to be the result of impacts , rotational destabilization , or a combination of several of these types of effects ( e.g. , * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in these cases
, the objects can be referred to as disrupted asteroids ( cf .
instances where a combination of both sublimation and disruptive effects may be responsible for activity are also possible , such as the cases of 133p / elst - pizarro , for which rapid nucleus rotation may enhance the strength of its repeated dust emission events @xcite , and p/2013 r3 ( catalina - panstarrs ) , for which rapid nucleus rotation may have induced its distintegration , but sublimation may have been responsible for ongoing post - disintegration activity @xcite . in such cases ,
the inferred presence of ice is the defining characteristic , and as such , we still consider these objects as mbcs .
mbcs have attracted interest in astrobiology for their potential to constrain theoretical studies indicating that material from the asteroid belt region could have been a significant primordial source of the water and other volatiles on earth ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the existence of water in the asteroid belt in the past has long been inferred from the existence of hydrated minerals in ci and cm carbonaceous chondrite meteorites believed to originate from main - belt asteroids ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , as well as from spectroscopic observations of asteroids themselves ( cf .
if ice is still present today in the asteroid belt , as the mbcs and other recent work ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) suggests , it would represent a opportunity to probe a potential primordial water source through compositional and isotopic studies using either in situ measurements by a visiting spacecraft or some of the next - generation extremely large telescopes now in development .
icy asteroids also contain some of the least altered material from the inner part of the protosolar disk still in existence today , and could give insights into the early stages of the formation of our solar system .
the added bonus of their close proximity in the main asteroid belt means that in situ spacecraft studies are feasible given present - day technical capabilities .
collectively , mbcs and disrupted asteroids comprise the class of objects known as active asteroids ( e.g. , * ? ? ?
orbits of small solar system bodies are typically classified as asteroidal or cometary using the tisserand parameter , or tisserand invariant , with respect to jupiter , @xmath7 , as the primary dynamical discriminant , where asteroids have @xmath8 and comets have @xmath9 @xcite . as such ,
a full accounting of active asteroids includes not only objects found in the main asteroid belt , but also other comet - like objects such as ( 2201 ) oljato , ( 3200 ) phaethon , and 107p/(4015 ) wilson - harrington @xcite , which have @xmath8 but whose orbits carry them well outside the asteroid belt . in this work here , however , we are primarily interested in objects in the main asteroid belt .
the currently known active asteroids found in the main asteroid belt , along with their orbital elements and absolute magnitudes , are listed in table [ table : knownaas ] .
the most likely classification of each object as either a mbc or disrupted asteroid based on the available evidence ( e.g. , numerical modeling indicating whether dust production is impulsive or ongoing , photometric measurements showing steady , increasing , or decreasing dust cross - sections , or observations of repeated activity ; cf . * ? ? ?
* ) is also indicated , except for 233p / la sagra , for which no physical analysis is available at this time .
we also plot the orbital elements of the known active asteroids in the main asteroid belt in figure [ figure : mbcs_aei ] .
.known active main - belt asteroids@xmath10 [ cols="<,^,>,>,>,>,>,>,^ " , ] [ table : rubblemantles ] for freshly exposed ice , the subsolar approximation ( @xmath11 ) may be appropriate .
however , once a diurnally insulating rubble mantle has formed , material located below the diurnal thermal skin depth effectively assumes the diurnally averaged temperature at the object s current heliocentric distance , making the isothermal case ( @xmath12 ) the better approximation .
as the mantle grows thicker , subsurface ice experiences lower peak temperatures , the diffusion barrier becomes thicker , and the reduced gas flow is less able to eject large particles , which leads to a more fine - grained and thus less permeable mantle .
all of these factors decrease the water emission rate . for our analysis here where we are concerned with observable activity , we are interested in the timescale for the formation of a mantle that is thinner than the seasonal skin depth , but may be several diurnal skin depths thick . assuming @xmath13 m and @xmath14 m , we compute water sublimation rates ( for surface ice ) and critical dust particle radii for objects at typical perihelion , semimajor axis , and aphelion distances for outer - belt mbcs ( 2.5 au , 3.1 au , and 3.8 au , respectively ) for these two cases , i.e. , no mantle and @xmath11 , and a mantle with a thickness on the order of the diurnal skin depth and @xmath12 ( table [ table : rubblemantles ] ) .
we also compute sublimation efficiencies , effective sublimation rates ( i.e. , the product of the base sublimation rate and the sublimation efficiency ) , and mantle growth rates in these two situations .
our results show that in the absence of a mantle ( i.e. , for fresh surface ice ) , where we assume @xmath11 , the effective sublimation rate and mantle formation rate are initially high at any position in the orbit of a typical mbc ( from @xmath15 au to @xmath16 au ) . as the subsolar regime transitions into the isothermal regime , the effective sublimation rate and the mantling rate decrease , and also become more strongly dependent on heliocentric distance , decreasing by about four orders of magnitude from perihelion to aphelion for @xmath12 ( cf
. figure [ figure : sublim_rate_heliodist ] ; table [ table : rubblemantles ] ) . during this transition ,
strong activity , and therefore significant mantle growth ( since the mantle growth rate is directly related to activity strength ) , should increasingly become confined to the region of the mbc s orbit near perihelion .
a detailed consideration of the transition between the subsolar and isothermal regimes is beyond the needs of our analysis here , as it is sufficient to simply note that the formation timescale of a diurnally insulating mantle should lie between the formation timescales calculated under each approximation .
a simple iterative calculation indicates that under the subsolar approximation , the mantle should reach one diurnal skin depth in thickness in a timescale of @xmath17 yr ( i.e. , several weeks ) from the time of the initiation of sublimation , while using the isothermal approximation , we calculate that the mantle should reach one diurnal skin depth in thickness in @xmath18 yr .
therefore , the true formation time of a diurnally insulating mantle should lie in the range of @xmath19 yr , where we note that the longer formation timescales calculated using the isothermal approximation may actually correspond to the duration of multiple orbit periods since they only include the total time spent close to perihelion .
eventually , the mantle should become so thick that it reduces even the peak sublimation - driven dust production rate to below detectable levels . at this point ,
the object effectively becomes permanently inactive , that is until another impact event occurs and strips away enough surface material for sublimation to begin again .
given that the timescale for another such impact to occur on the same body is far larger than a single orbit period for a typical mbc , however ( cf . * ? ? ?
* ) , the crude estimates in table [ table : rubblemantles ] provide a plausibility check that a mbc exhibiting cometary mass loss due to the sublimation of recently collisionally excavated ice at small heliocentric distances can be repeatedly active over multiple orbit passages ( e.g. , 133p and 238p ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite have carried out detailed model calculations of activity and mantle formation for comet p/2008 r1 ( garradd ) , with several sets of example parameters , and found that the emission of water decreases from one orbital period to another , but in some cases only slowly . @xcite and @xcite also computed comparable mass loss rates for 133p for active episodes separated by more than a decade ( after accounting for different values for grain size and density used in each analysis ) .
our crude estimates in the first row of table [ table : rubblemantles ] ( i.e. , for @xmath15 au ) , where ice initially retreats rapidly but slows significantly after a mantle of diurnal skin depth has formed , are consistent with these results . ) for a typical mbc as a function of time spent near perihelion ( @xmath20 au ) ( blue line ) .
we begin our calculations at 0.1 yr after the start of sublimation , which is the estimated time required for a diurnally insulating rubble mantle to form under the subsolar approximation ( vertical dashed line ) , assume an initial mantle depth of @xmath21 , and also assume @xmath12 for the duration of the modeled mantle growth .
, width=412 ] intriguingly , this analysis indicates that there could be a physical basis for expecting widespread low - level activity in the asteroid belt , as suggested by @xcite .
we schematically illustrate the isothermal phase of mantle growth in figure [ figure : mantle_growth ] where we numerically calculate and plot the effective sublimation rate ( @xmath22 ) at the perihelion distance of a typical mbc ( @xmath15 au ) as a function of time spent near perihelion .
the calculations start at 0.1 yr ( i.e. , the time required for a diurnally insulating rubble mantle to form under the subsolar approximation ; vertical dashed line in figure [ figure : mantle_growth ] ) after the start of sublimation , assume an initial mantle depth of @xmath23 , and assume @xmath12 for the duration of the modeled mantle growth .
activity is seen to decline asymptotically , becoming suppressed to half of its initial strength on a timescale of @xmath24 years of mantle formation time , and then taking another @xmath25 years to halve again .
this asymptotic decline in activity strength means that a mbc should spend a long period of time in a weakly active state , where we assume that the currently known mbcs exhibit comparatively strong activity ( since strong activity is easier to discover than weaker activity ) .
this means that there could be a large population of mbcs that formerly exhibited mass loss rates comparable to that of the currently known objects ( @xmath26 kg s@xmath27 ; e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but now exhibit long - lived but weaker activity that is just beyond the detection limits of current surveys .
more sensitive future surveys capable of detecting weaker cometary activity than current surveys could be able to access this segment of the active mbc population , and we might therefore expect many more mbc discoveries to result from future search efforts .
re - computation of mass loss rates from observations of 133p during its many observed active episodes @xcite using consistent values for relevant physical parameters and identical analysis techniques could also help test this hypothesis and set constraints on the rate of mbc activity attenuation .
for simplicity , we have assumed that the water sublimation rate is directly related to the dust mass loss rate .
the dust to gas ratio in mbcs is unconstrained by observations , however , and so a direct relation between the water sublimation rate and dust emission rate may not accurately represent reality .
we also note that rapid rotation or jet - like emission could give rise to larger critical radii than those estimated in table [ table : rubblemantles ] ( cf . * ? ? ?
* ; * ? ? ?
* ) , leading to slower mantle growth
. nonetheless , the timescale for the growth of a seasonal insulating rubble mantle , even in the extremely limiting case of a typical mbc at aphelion in the isothermal approximation , is still extremely short compared to the estimated dynamical lifetimes of these objects ( from @xmath420 myr to @xmath28 gyr ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . as such
, they can not have been continuously active over their entire residence times in the main asteroid belt , and therefore a recent event , such as an impact that excavated subsurface ice , must have occurred on each of these objects to enable observable present - day sublimation to take place .
the mechanism that we explore here , where the thermal wave associated with the perihelion passage of a mbc penetrates through a dry mantle until it reaches the ice and activates the comet via sublimation , is consistent with the pattern of nearly all mbcs exhibiting activity close to or shortly after perihelion .
since the heat wave penetrates only slowly through the layer , there is a phase delay between the surface temperature and the temperature experienced by the ice .
the thickness of the refractory mantle , relative to the thermal skin depth , can be expected to be associated with a delay in the activation ( e.g. , * ? ? ?
without a mantle , activity may be expected to be nearly symmetric around perihelion . with a thick mantle
, activation can be delayed even until after perihelion is reached .
bodies in the middle main belt are typically warmer than in the outer main belt , and thus less likely to retain ice near the surface
. this may not be the only reason why active mbcs are predominantly observed in the outer main belt and rarely in the middle main belt though .
a population of middle - belt mbcs could exist , and upon being collisionally activated , the activity of these objects could be much more vigorous than for outer belt mbcs .
however , this activity would also be expected to be shorter - lived due to more rapid mantle formation , and therefore less likely to be discovered than activity in outer - belt mbcs .
the aim of this discussion is merely to outline a possible framework for understanding the observed pattern of mbc behavior , where more work is needed to examine other aspects of the activity modulation mechanism described here .
for example , @xcite showed that an object s orbital obliquity can significantly reduce the overall average temperature of its surface over its lifetime , meaning that for objects with non - zero obliquity , the temperatures and sublimation rates discussed above represent upper limits to realistic cases .
moreover , mantles of very low thermal conductivity cause additional cooling due to the amplitude dependence of @xmath29 thermal emission .
a full treatment of this problem via detailed numerical modeling , which is beyond the scope of this work , should also include a more systematic exploration of the ranges of possible rotation rates , densities , porosities , heat capacities , thermal diffusivities , thermal inertias , ice - to - dust ratios , total volatile contents , and particle size distributions than are considered here .
the analysis here is intended only to illustrate a possible mechanism for mbc activity modulation using physically plausible parameters , while a more complete exploration of the available parameter space would better determine the range of material properties and environmental conditions under which mbc activity can plausibly arise . in this
regard , the discovery of even more mbcs as well as continued observational studies of known mbcs to determine their physical properties ( e.g. , nucleus sizes and rotation rates ) will also help constrain various aspects of the model framework outlined here .
we report the following key points and make the following conclusions : 1 .
we analyze a set of 760475 observations with @xmath30 of 333026 unique main - belt objects obtained by
the ps1 survey telescope between 2012 may 20 and 2013 november 9 using @xmath31 , @xmath32 , @xmath33 , or @xmath34 filters .
these observations include the discoveries of two main - belt comets , one disrupted asteroid , one active centaur , eighteen jupiter - family comets , and six long - period comets , as well as several observations of known main - belt comets , disrupted asteroids , and other comets
the comet detection procedures currently in use for the ps1 survey consist of the comparison of the point - spread - functions of moving objects to those of reference stars in the same field , and the flagging of objects ( on the order of several hundred to several thousand per night ) that show significant psf excesses for human inspection ( details in [ appendix : comet_screening ] ) . the majority of these flagged detections can be immediately rejected as false detections , while a small number are selected for manual psf analysis , and an even smaller number are selected for observational follow - up . known
issues that can result in missed comet discoveries include the misidentification of real objects as false detections , the subjectivity of the human review process , and the dependence of the discovery process on successful follow - up observations .
morphology parameter measurements are also known to be less reliable for low s / n detections or detections in close proximity to bright field stars or galaxies , chip gaps , other detector defects , or masked regions , causing a further decrease in our comet discovery efficiency .
based on the number of missed discovery opportunities by ps1 among comets discovered by other observers , we estimate an upper limit discovery efficiency rate of @xmath470% for ps1 .
3 . based on the statistics of the ps1-observed sample of asteroids observed shortly after perihelion ( @xmath35 ) where we expect activity detectable by ps1 to be present
, we find an expected fraction of @xmath36 mbcs per @xmath1 outer main - belt asteroids and a 95% confidence limit upper limit of @xmath37 mbcs per @xmath1 outer main - belt asteroids , assuming a detection efficiency rate of @xmath38 , corresponding to a total expected population of @xmath2 mbcs and an upper limit population of @xmath39 mbcs with absolute magnitudes of @xmath3 and activity levels detectable by ps1 ( i.e. , equivalent to mass loss rates on the order of @xmath26 kg s@xmath27 ) .
the known mbc population in the outer main asteroid belt has significantly higher eccentricities than the background asteroid population . at these eccentricities ,
the theoretical sublimation rate at perihelion is orders of magnitude larger than at aphelion , consistent with observations showing that mbcs peak in activity strength near perihelion .
this implies that mbc activity is predominantly modulated by variations in heliocentric distance , rather than seasonal variations in solar illumination of isolated active sites related to the orientation of the rotation axis .
these results indicate that the overall mantle growth rate should be slow , as significant mantling should only occur near perihelion , pointing to a process by which mbc activity can be sustained over multiple orbit passages .
we review lessons learned from the ps1 survey that may help to improve the comet detection capabilities of future surveys ( [ appendix : futurelessons ] ) . at the current time , ps1 is unable to detect or is poor at detecting extremely bright objects exhibiting relatively faint activity ( i.e. , scheila - type objects ) , low - level activity with little or no effect on an object s psf ( i.e. , 133p - type objects ) , unresolved activity of the type that could be detected photometrically ( i.e. , chiron - type objects ) , and outbursts on extremely bright and large known comets ( i.e. , 17p / holmes - type objects ) .
implementation of multiple screening methods , including radial psf comparison , linear psf comparison , modeling of trailed psfs , detection of azimuthally localized activity , comparison of photometry optimized for point sources and photometry optimized for extended sources , detection of photometric deviations from expected brightnesses , and crowd - sourcing , should help to fill these blind spots in future surveys .
regardless of what method or methods are used , however , we emphasize that false detection minimization , balance between screening sensitivity and human and computational overheads , and robust follow - up observation plans should also be important considerations in the formulation of future comet detection systems .
ps1 s exceptional success at discovering mbcs is likely largely due to its sensitivity to faint , low - activity comets .
future surveys that will be sensitive to even fainter and lower - activity comets should find many more mbcs , perhaps even in excess of the expected population estimated in this paper , and therefore the development of robust comet detection algorithms for these surveys should be a high priority .
sensitivity to faint comets will also aid studies of active centaurs , and long - term studies of earth - approaching comets .
the recent detection of water vapor emission from ceres @xcite and probable detections of water ice on main - belt object ( 24 ) themis @xcite have provided additional evidence of present day water in the asteroid belt .
these findings support the notion that water ice has been able to survive in a number of main - belt asteroids until the present day , and that mbc activity is being driven by sublimation .
two priorities for future mbc research are the discovery of more mbcs , and the continued physical and dynamical characterization of individual mbcs and the population in general .
while ps1 has been exceptionally successful at discovering mbcs ( and disrupted asteroids ) compared to previous and ongoing surveys , a substantial increase in the discovery rate is still needed to enable statistically meaningful analyses to be conducted of the spatial distribution of these objects . in turn
, such analyses will enable the use of mbcs to trace the current water content of the asteroid belt , and therefore infer its primordial water content and distribution , subject of course to various dynamical disturbances that very likely occurred between the formation of the solar system and the present day ( e.g. , * ? ? ?
* ; * ? ? ?
fortunately , the discovery power to accomplish such a leap should soon be available with wide - field high - spatial - resolution surveys using large aperture telescopes like the hsc survey and lsst . by going deeper than any previous surveys before them
, they should be able to discover many more new mbcs , shedding new light on these still - enigmatic objects , and perhaps on our own origins as well .
we thank pedro lacerda and bin yang and two anonymous referees for helpful comments on this manuscript .
acknowledges support for this work by the national aeronautics and space administration ( nasa ) through hubble fellowship grant hf-51274.01 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy ( aura ) for nasa , under contract nas 5 - 26555 .
n.s . acknowledges support by nasa through the nasa astrobiology institute under cooperative agreement no .
nna09da77a issued through the office of space science . j.k . acknowledges support through nsf grant ast 1010059 .
the pan - starrs1 surveys ( ps1 ) have been made possible through contributions of the institute for astronomy , the university of hawaii , the pan - starrs project office , the max - planck society and its participating institutes , the max planck institute for astronomy , heidelberg and the max planck institute for extraterrestrial physics , garching , the johns hopkins university , durham university , the university of edinburgh , queen s university belfast , the harvard - smithsonian center for astrophysics , the las cumbres observatory global telescope network incorporated , the national central university of taiwan , the space telescope science institute , the national aeronautics and space administration under grant no .
nnx08ar22 g issued through the planetary science division of the nasa science mission directorate , the national science foundation under grant no .
ast-1238877 , the university of maryland , and eotvos lorand university ( elte ) .
we thank the ps1 builders and ps1 operations staff for construction and operation of the ps1 system and access to the data products provided .
in terms of the individual steps required for an object observed by ps1 to be discovered as a comet , first , a minimum of two detections of an object must be made , identified as transients by ipp , and linked as a tracklet by mops .
a common situation in which this process may not occur successfully is when observations of a particular object in an observing sequence are lost to chip gaps or other unusable regions of the ps1 camera due to the the camera s @xmath475% effective fill factor ( which includes the effects of chip gaps , guide star cells , and masking of bad pixels , detector artifacts , and diffraction spikes ) .
tracklet linking can also fail if variable sky conditions hinder successful subtraction of background sources , or cause some images of faint objects to fall below ipp s detectability threshold within an observing sequence even if the object is detectable in other exposures .
if tracklet linking is completed successfully , the tracklet must then be flagged by the mops comet screening system , but as discussed above , reliable morphology parameter measurements ( on which our flagging system relies ) from ipp are not guaranteed , particularly for low @xmath40 detections or detections in close proximity to bright field stars or galaxies , chip gaps , or masked regions .
a notable example of a comet being missed due to issues with tracklet linking is mbc p / la sagra . following its discovery on 2010 september 14 @xcite
, it was found to have been clearly cometary in two sets of ps1 observations obtained in the month prior to its official discovery @xcite . on one night , it was observed twice , and in another , it was observed four times .
however , the high false detection rate at the time meant that pairs of detections were considered unreliable , and so the first set of observations was simply ignored by the system .
in the set of four detections , the object passed near a bright field star in one image , skewing its measured magnitude . at the time , mops required detections within a tracklet to have similar magnitudes to ward against mis - linked detections ( this is no longer true , largely due to this case ) , and so this tracklet was rejected and also never displayed for human review .
the automated portion of the ps1 comet screening process identifies several hundred to several thousand comet candidates each night . of these , most are data artifacts ( e.g. , internal reflections , imperfectly - subtracted stationary sources , and diffraction spikes ; figure [ figure : falsedetections ] ) , low @xmath40 detections , or saturated detections , giving a relatively large false positive rate on the order of a few hundred to one .
it is however at a level that can be managed by a secondary level of human screening , where real objects can be rapidly identified by eye by a mops team member who then makes a visual assessment and , if warranted , a quantitative assessment of the cometary nature of the object .
this initial evaluation process is subjective and can depend on the experience of the human screener on a given day , the amount of time that can be devoted to each day s screening given other operational priorities , and the number of objects flagged for evaluation on any particular night .
this process can also be affected by other candidates in the same data set , since the discovery of several good candidates in a single night could necessitate the prioritization of follow - up efforts at the expense of lower - probability candidates , while a data set of generally poor quality may not be reviewed as carefully as a better quality data set .
comet candidates that pass all the screening stages up to this point must then be scheduled for follow - up observations to confirm the activity , and also extend the orbital arc if the candidate is an unknown object .
follow - up observations by ps1 team members are typically conducted with the university of hawaii 2.2 m telescope or cfht , both on mauna kea in hawaii , or the 2.0 m faulkes telescopes ( north and south ) on haleakala in hawaii and siding spring in australia .
however , scheduling , weather , instrument availability , or other technical issues can sometimes prevent us from using these facilities .
external observers are sometimes able to provide follow - up confirmation and astrometry , since we report all of our high - probability comet candidates to the mpc upon identification . prior to october 2013
, the mpc would place comet candidates on their near - earth object confirmation page where experienced amateur observers and professional astronomers can view a list of high - priority targets needing follow - up observations .
starting in october 2013 , the mpc introduced a new possible comet confirmation page , where comet candidates needing confirmation are now listed .
however , the quality of observing facilities available to these external observers varies widely , and for very difficult ( i.e. , faint or marginal ) comet candidates , sometimes follow - up on 2-m - class professional telescopes or larger is required to obtain the observations that are needed . if an object is already known and confirmation of cometary activity is all that is needed , delays in obtaining follow - up observations are often tolerable as cometary activity typically remains observable for at least several days , or even several weeks or months .
since the time baseline spanned by the typical tracklet is usually only about one hour though , positional uncertainties based on initial orbit determinations of previously unknown objects are often large and increase rapidly . as such
, follow - up typically has to be performed within several days of the original ps1 observations before the positional uncertainties simply grow too large for the object to be recovered , even by as large a camera as cfht s megacam and its @xmath41 field of view , in which case , the object is effectively lost .
occasionally , precoveries or self - recoveries of objects are found among other ps1 observations , but due to survey constraints , we can not perform targeted follow - up with the ps1 telescope itself .
ps1 observations of moving objects are not currently automatically stacked to create higher @xmath40 composite images ( although they are often manually stacked during the human - screening stage of the comet identification process ) . while such composite images are normally useful for searching for low - surface - brightness activity , the current level of inconsistency between ps1 observations of moving objects is sufficiently high that this technique would have limited usefulness as part of the automatic screening process .
the small sizes of the individual ccds in the ps1 detector array means that one or more detections in a tracklet may be obstructed by a chip gap , thus contaminating the remaining gap - free images if these obstructed detections are indiscriminately included when stacking .
chip defects that are automatically masked by ipp present a similar issue . in most cases , median stacking is unhelpful for compensating for this problem due to the small number of detections per tracklet .
stacking data also hinders visual identification of poorly subtracted background objects that could be mistakenly interpreted as faint cometary features .
given these considerations , we currently prefer to consider individual detections , looking for consistency of morphology measurements and visual appearance of detections within each tracklet .
promising detections flagged during visual screening can then be manually stacked , allowing any undesirable detections within a tracklet to be identified and omitted .
given the added sensitivity to low - level cometary activity that would be gained , though , automated stacking of all tracklets may be worthwhile to perform during post - processing of ps1 data , as well as in real - time for future surveys , if the complications discussed here can be mitigated .
despite ps1 s success to date at discovering a wide range of comet - like objects , from classical comets to mbcs to das , it is important to note what types of comet - like objects ps1 is currently unable to detect , i.e. , the system s `` blind spots '' , so that future surveys can attempt to fill similar gaps in their comet discovery procedures . for example , comparison of asteroidal and stellar radial psfs is effective for detecting coma in minimally trailed objects , as we have demonstrated over the course of the ps1 survey , but is less effective for objects that are significantly trailed ( i.e. , for objects with large non - sidereal velocities , long exposure times , or both ) .
this complication can be handled by measuring linear psfs perpendicular to the direction of trailing for both asteroids and stars instead of measuring radial psfs .
this method will be sensitive to spherical coma or directed emission perpendicular to the direction of trailing , and will be insensitive to directed emission parallel to the direction of trailing , but at least permits trailed objects to be searched for activity at all .
this type of analysis is already often manually performed on candidate comet images in ps1 data as part of the human review process to determine whether detected psf excesses are due to cometary activity or trailing effects , but automating the process for future surveys could prove productive .
ps1 is currently unable to detect scheila - type objects , i.e. , large , bright objects that undergo substantial outbursts due to impacts or cometary activity , because such objects saturate the ps1 camera .
this will be an even more severe problem with upcoming wide - field surveys using even larger aperture telescopes like lsst .
this problem could be alleviated by the development of algorithms to detect excess sky flux far from the photocenter of any saturated moving objects , i.e. , in the presumably unsaturated wings of those objects psfs .
meanwhile , relatively shallow surveys such as atlas @xcite could also potentially allow bright asteroids to be productively searched for activity .
the potential for confusion of background objects or noise artifacts with comet - like activity by automated routines will be even higher for activity searches at large sky - plane separations from saturated objects , however , compared to searches for close - in activity of unsaturated objects similar to what we describe here .
thus , even more care will need to be taken to mitigate the effect of false detections in this type of search .
ps1 s automated flagging system relies entirely on the measurement of deviations in an object s psf as compared to psfs of nearby field stars ( section [ section : ps1comets ] ) , and so another current blind spot for ps1 is the inability to detect comet - like activity ( e.g. , a faint dust tail ) that has minimal effect on an object s psf .
a notable example of such an object is 133p @xcite . to account for objects like this ,
we had originally intended to employ a tail - finding algorithm similar to that described in @xcite where a circular annulus is placed around each candidate object and divided into slices .
the flux inside each slice can then be measured , with any excess flux along a particular azimuthal direction noted as an indication of possible directed dust emission such as a dust tail .
this technique unfortunately was only minimally effective for finding comets in ps1 data due to the large number of data artifacts which led to an unmanageably large number of false positive detections , but may be more effective in analyzing higher quality data from future surveys . searching for unresolved activity by looking for asteroids exhibiting photometric excesses compared to their expected brightnesses is another method for detecting cometary activity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) that is currently unimplemented as part of ps1 s comet screening procedures
. work is ongoing to develop mechanisms for reliably detecting comets from photometry alone , but challenges are numerous .
detecting photometric enhancements in asteroids relies on having accurate knowledge of their photometric brightnesses at the time of observation as well as their expected brightnesses based on their absolute magnitudes and phase functions .
ps1 data are now fully photometrically calibrated across the entire sky @xcite , but the lack of accurate photometric references across the entire sky during the early stages of the survey impeded efforts to implement photometric activity detection until just recently .
furthermore , accurate measurements of asteroid phase functions are lacking for a large segment of the main - belt population , meaning that predictions of the brightnesses of these objects must rely on relatively low - precision mpc data .
photometric activity searches are still possible given these circumstances , but would only be sensitive to activity causing photometric enhancements in excess of the uncertainties due to imprecisely known predicted magnitudes as well as those due to imprecisely known observed magnitudes .
however , given the large sizes of these uncertainties ( at least until recently ) , any objects showing such large photometric enhancements were probably exhibiting bright enough activity to be discovered via other means .
this situation is greatly improved now , however , as ps1 data can now be considered photometrically reliable as discussed above , and thanks to ps1 , we have also amassed baseline photometry for a much larger segment of the main - belt population than was available before and so have been able to calculate phase functions for these asteroids ( vere et al . , 2014 , in prep ) .
this method for searching for photometric indications of activity should therefore henceforth be much easier to conduct for both ps1 and other surveys , but will still remain limited to detecting activity that causes photometric enhancements in excess of the brightness uncertainties due to an object s unknown rotational state at the time of observations . finally , there is the problem of automating the detection of comets that are extremely bright ( @xmath42 mag ; cf .
table [ table : sampledist ] ) , have large angular sizes , or both ( e.g. , figure [ figure : ps1brightcomet]a ) , because of the difficulty of even identifying them as real sources using ps1 s point - source detection software .
most bright comets are already known and so this problem likely does not result in a large number of missed comet discoveries by ps1 .
previously unknown comets have occasionally been discovered due to undergoing sudden outbursts ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , however , and the inability of the ps1 system to automatically detect extremely bright objects means that we also risk missing the discovery of sudden cometary outbursts from known comets such as the one exhibited by 17p / holmes in 2007 @xcite .
comets with unusual morphologies , without distinct central condensations , or that are otherwise extremely non - point - source - like present similar problems .
p/2012 f5 ( gibbs ) was discovered on 2012 march 22 @xcite but appeared visibly cometary in ps1 data as early as 2011 december 6 , and again on 2012 february 28 ( figure [ figure : ps1brightcomet]b ) .
however , because its central condensation was so indistinct compared to the rest of the comet , and the comet itself was dominated by an extremely long and narrow tail , ipp and mops filters designed to screen out image artifacts like diffraction spikes rejected these detections , preventing them from ever entering the mops comet screening pipeline .
it is difficult to conceive of what reasonable modifications or additions to the ps1 comet detection pipeline would allow us to deal with such unusual objects , and so for finding such comets in future surveys , perhaps the solution lies not in automation , but rather in the outsourcing of comet detection to the public via citizen science projects ( e.g. , * ? ? ?
* ; * ? ? ?
such projects have been used to address problems in astronomy as diverse as classifying galaxy morphologies ( e.g. , * ? ? ?
* ) , identifying supernovae @xcite , searching for infrared bubbles in the inner galactic plane @xcite , measuring structural properties of galaxies @xcite , searching for planetary transits in kepler data ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and locating precoveries of near - earth asteroids in sdss data @xcite . without fixed preconceived notions of what types of comet morphologies to search for that are inherent in any automated detection algorithm , citizen scientists could help ensure that comets with morphologies that are unexpected or simply difficult for automated routines to detect are not missed , while also adding the potential to make serendipitous discoveries of unexpected phenomena ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . while final evaluation of candidate objects would almost certainly still have to be done by professional astronomers , implementing citizen science comet detection programs could be a relatively low - effort way for future surveys to supplement automated detection pipelines with a large - scale human visual screening effort to minimize the chances of missing unusual comets .
image of c/2012 k5 ( linear ) obtained by ps1 on 2013 january 7 .
( b ) composite @xmath43-band image of p / gibbs constructed from data obtained by ps1 on 2011 december 06 .
the area of sky shown is @xmath44 in size with north at the top and east to the left . in this image ,
the dust trail is observed to extend @xmath45 to the northwest from the nucleus ( lower left corner ) .
the grid - like pattern in the sky is due to chip gaps in the ps1 ccd mosaic .
, width=480 ] among several other practical lessons we can take from our experience with ps1 that are applicable to future surveys , we find that automated comet detection remains difficult to perform reliably , with major challenges including false detection minimization and coping with the wide range of comet morphologies that are possible . unfortunately , at least for ps1 , many common data artifacts have very comet - like morphologies ( cf .
figure [ figure : falsedetections ] ) , and so are routinely flagged as comet candidates by our automated detection system . minimizing the number of these obviously false detections was a critical requirement in the development of our comet screening procedures because while a rate of several hundred false detections for every real detection is manageable for a human screener , a rate of several thousand or tens of thousands to one ( as we originally had ) was simply intractable .
even then , more subtle false detections frequently still require expert human review to identify , a process that has become comparatively more streamlined over time as we gained experience in recognizing the most common types of subtle false detections , but one that still represents a significant burden on human resources over the course of the survey .
we therefore suggest that development of automated false detection algorithms that are as thorough and reliable as possible should be considered a high priority for any comet discovery efforts as part of future surveys . in terms of dealing with comets with a wide range of morphologies , the best way to handle this issue
may simply be the use of multiple screening methods .
many possible screening methods that future surveys could implement are discussed above , including radial psf comparison , comparison of linear psfs measured perpendicular to the direction of an object s trailing , modeling of trailed psfs , searching for azimuthally localized excess flux around an object , comparison of an object s total flux to its peak flux as a measure of degree of condensation , searching for photometric enhancements of asteroids of known brightnesses , and crowd - sourcing .
in addition to these screening methods , future surveys could also compare psf magnitudes ( which are measured by fitting a gaussian point - spread function model to an object , and are optimized for photometry of point - like sources ) and so - called model magnitudes ( which are optimized for photometry of extended objects such as galaxies , and provide more accurate photometry of resolved sources ) , where extended objects should show the largest deviations between the two measurements .
this method was employed by @xcite to find comets in sdss data , and is also used by sdss for star - galaxy separation as well @xcite .
similar screening methods including comparisons of aperture magnitudes and kron magnitudes to psf magnitudes were also explored for ps1 comet detection but were ultimately rejected due to decreased reliability of these parameters at faint magnitudes or near chip artifacts such as detector gaps or masks .
the suitability of any of these methods for finding comets will depend on the specific characteristics of a given survey , but also on the cost in either human resources , computational power , or time needed to implement them . for initial screening , it may only be feasible to perform simple magnitude comparisons or basic psf comparisons using parameters produced automatically by a survey s general reduction pipeline , rather than more detailed custom analyses requiring access to image data that could represent unacceptable burdens on available computational resources , depending on the amount of data involved and the speed at which the data are processed .
similarly , analyses that are excessively time - consuming because of the amount of computing resources or human intervention required may risk not being performed at all , and could also adversely affect the timeliness with which observational follow - up of candidates can be scheduled , putting their recoveries at risk .
as such , when choosing the most suitable method or set of methods to search for comets , consideration of overheads is perhaps just as important as consideration of the absolute effectiveness of any particular comet detection scheme .
finally , we must emphasize that a robust follow - up plan is an essential complement to any survey aiming to find comets .
follow - up confirmation and astrometry of newly discovered comets has frequently been provided in the past by amateur astronomers using small telescopes ( @xmath46 m ) at sub - optimal observing sites . as comets discovered by advanced surveys include increasingly fainter and weaker comets , either by utilizing increasingly sensitive algorithms ( as ps1 does ) or increasingly larger telescopes ( as the hsc survey and lsst will do ) , follow - up observations will become increasingly out of the reach of amateur observers , necessitating the use of professional facilities .
such facilities will generally not need to have comparable aperture sizes to the telescopes conducting the surveys in question , since they can simply use longer integration times .
ideal requirements for any follow - up facilities , however , do include being located at sites with reliable observing - quality weather , availability of a wide - field imager most or all of the time , and the ability to rapidly respond to observation requests on short notice ( e.g. , through a queue - scheduled system or a rapid - response target - of - opportunity policy ) , where we note that self - follow - up by the original survey facility would also be suitable if it can be accommodated as part of ongoing survey operations .
bauer , j. m. , mainzer , a. k. , grav , t. , et al .
wise / neowise observations of active bodies in the main belt .
astrophys .
j. 747 , 49 .
birtwhistle , p. , ryan , w. h. , sato , h. , beshore , e. c. , kadota , k. 2010 .
a2 ( linear ) .
bodewits , d. , kelley , m. s. , li , j .- y . , landsman , w. b. , besse , s. , & ahearn , m. f. 2011 .
collisional excavation of asteroid ( 596 ) scheila .
astrophys .
j. 733 , l3 .
bolin , b. , denneau , l. , wainscoat , r. j. , et al .
s4 ( panstarrs ) .
bureau electr .
telegrams 3245 , 1 .
bolin , b. , vere , p. , wainscoat , r. j. , micheli , m. , draginda , a. , hormuth , f. , ries , j. g. , sato , h. 2013a .
g8 ( panstarrs ) .
bureau electr .
telegrams 3477 , 1 .
bolin , b. , denneau , l. , vere , p. , et al .
bm_80 = ( 323137 ) 2003 bm_80 .
bureau electr .
telegrams 3559 , 1 .
bolin , b. , denneau , l. , wainscoat , r. j. , et al .
n3 ( panstarrs ) .
bureau electr .
telegrams 3583 , 1 .
bolin , b. , hsieh , h. h. , wainscoat , r. j. , et al .
o2 ( panstarrs ) .
bureau electr .
telegrams 3604 , 1 .
bolin , b. , wainscoat , r. j. , denneau , l. , micheli , m. , forshay , p. , woodworth , d. , williams , g. v. 2013e .
n5 ( panstarrs ) .
bureau electr .
telegrams 3617 , 1 .
bolin , b. , micheli , m. , hsieh , h. h. , et al . 2013f . comet p/2013 p1 ( panstarrs ) . cent . bureau electr .
telegrams 3618 , 1 .
bolin , b. , denneau , l. , micheli , m. , wainscoat , r. j. , tholen , d. j. , lister , t. 2013 g .
p5 ( panstarrs ) .
bureau electr .
telegrams 3639 , 1 .
bowell , e. , west , r. m. , heyer , h .- h . , et al
( 4015 ) 1979 va = comet wilson - harrington ( 1949 iii ) .
burbine , t. h. 1998 .
could g - class asteroids be the parent bodies of the cm chondrites ?
meteoritics & planet .
33 , 253 - 258 .
bus , s. j. , bowell , e. , french , l. m. 1988 .
( 2060 ) chiron .
buzzi , l. , muler , g. , kidger , m. , et al . 2007 .
comet 17p / holmes .
campins , h. , hargrove , k. , pinilla - alonso , n. , et al . 2010 . water ice and organics on the surface of the asteroid 24 themis . nature .
464 , 1320 - 1321 .
capria , m. t. , marchi , s. , de sanctis , m. c. , coradini , a. , ammannito , e. 2012 .
the activity of main belt comets .
537 , a71 .
cardamone , c. , schawinski , k. , sarzi , m. , et al .
galaxy zoo green peas : discovery of a class of compact extremely star - forming galaxies .
mon . not .
r. astron .
399 , 1191 - 1205 .
castillo - rogez , j. c. 2011 .
ceres - neither a porous nor salty ball .
215 , 599 - 602 .
cikota , s. , ortiz , j. l. , cikota , a. , morales , n. , tancredi , g. 2014 . a photometric search for active main belt asteroids .
562 , a94 .
delsemme , a. h. , miller , d. c. 1971 .
physico - chemical phenomena in comets iii : the continuum of comet burnham ( 1960 ii ) . planet .
space sci .
19 , 1229 - 1257 .
demeo , f. e. , carry , b. 2013 . the taxonomic distribution of asteroids from multi - filter all - sky photometric surveys . icarus .
226 , 723 - 741 .
demeo , f. e. , carry , b. 2014 .
solar system evolution from compositional mapping of the asteroid belt .
505 , 629 - 634 .
denneau , l. , wainscoat , r. j. , hsieh , h. h. , et al .
v1 ( panstarrs ) .
bureau electr .
telegrams 3289 , 1 .
denneau , l. , jedicke , r. j. , grav , t. , et al .
the pan - starrs moving object processing system .
125 , 357 - 395 .
elst , e. w. , pizarro , o. , pollas , c. , ticha , j. , tichy , m. , moravec , z. , offutt , w. , & marsden , b. g. 1996 . comet p/1996 n2 ( elst - pizarro ) .
iau circ .
fanale , f. p. , salvail , j. r. 1989 .
the water regime of asteroid ( 1 ) ceres .
82 , 97 - 110 .
farinella , p. , davis , d. r. 1992 .
collision rates and impact velocities in the main asteroid belt .
97 , 111 - 123 .
fernndez , j. a. , gallardo , t. , brunini , a. 2002 .
are there many inactive jupiter - family comets among the near - earth asteroid population ?
159 , 358 - 368 .
fischer , d. a. , schwamb , m. e. , schawinski , k. , et al .
planet hunters : the first two planet candidates identified by the public using the _ kepler _ public archive data . mon . not .
r. astron .
419 , 2900 - 2911 .
garradd , g. j. , sostero , g. , camilleri , p. , guido , e. , jacques , c. , & pimentel , e. 2008 . comet c/2008 r1 ( garradd ) .
iau circ . 8969 , 1 .
gibbs , a. r. , sato , h. , ryan , w. h. , et al .
f5 ( gibbs ) .
bureau electr .
telegrams 3069 , 1 .
gilbert , a. m. , & wiegert , p. a. 2009 . searching for main - belt comets using the canada - france - hawaii telescope legacy survey .
201 , 714 - 718 .
gilbert , a. m. , & wiegert , p. a. 2010 . updated results of a search for main - belt comets using the canada - france - hawaii telescope legacy survey .
210 , 998 - 999 .
haghighipour , n. 2009 .
dynamical constraints on the origin of main belt comets . meteor . planet .
44 , 1863 - 1869 .
hainaut , o. r. , kleyna , j. , sarid , g. , et al . 2012 .
p/2010 a2 linear , i. an impact in the asteroid main belt .
537 , a69 .
hainaut , o. r. , boehnhardt , h. , snodgrass , c. , et al .
2014 . continued activity in p/2013 p5 panstarrs - the comet that should not be .
astron . astrophys . in press .
hardorp , j. 1980 .
the sun among the stars : ii . solar color , hyades
metal content , and distance .
88 , 334 - 344 .
heiken , g. h. , vaniman , d. t. , french , b. m. 1991 .
lunar sourcebook : a user s guide to the moon .
cambridge univ . press .
hill , r. e. , bolin , b. , kleyna , j. , et al .
r3 ( catalina - panstarrs ) .
bureau electr .
telegrams 3658 , 1 .
hiroi , t. , zolensky , m. e. , pieters , c. m. , lipschutz , m. e. 1996 .
thermal metamorphism of the c , g , b , and f asteroids seen from the 0.7 @xmath47 m , 3 @xmath47 m , and uv absorption strengths in comparison with carbonaceous chondrites .
meteoritics & planet .
31 , 321 - 327 .
hsieh , h. h. 2009 . the hawaii trails project : comet - hunting in the main asteroid belt .
505 , 1297 - 1310 .
hsieh , h. h. 2014a , iau symposium , 293 , 212 - 218 .
hsieh , h. h. 2014b .
the nucleus of main - belt comet p/2010 r2 ( la sagra ) .
icarus . in press .
hsieh , h. h. , & jewitt , d. 2006 . a population of comets in the main asteroid belt .
312 , 561 - 563 .
hsieh , h. h. , jewitt , d. , & fernndez , y. r. 2004 .
the strange case of 133p / elst - pizarro : a comet among the asteroids .
j. 127 , 2997 - 3017 .
hsieh , h. h. , jewitt , d. , & pittichova , j. 2006 .
re_70 = ( 118401 ) .
hsieh , h. h. , jewitt , d. , & ishiguro , m. 2009a .
physical properties of main - belt comet p/2005 u1 ( read ) . astron .
j. 137 , 157 - 168 .
hsieh , h. h. , jewitt , d. , & fernndez , y. r. 2009b .
albedos of main - belt comets 133p / elst - pizarro and 176p / linear .
astrophys .
j. 694 , l111-l114 .
hsieh , h. h. , jewitt , d. , lacerda , p. , lowry , s. c. , & snodgrass , c. 2010 .
the return of activity in main - belt comet 133p / elst - pizarro .
mon . not .
r. astron .
403 , 363 - 377 .
hsieh , h. h. , ishiguro , m. , lacerda , p. , & jewitt , d. 2011a .
physical properties of main - belt comet 176p / linear .
j. 142 , 29 .
hsieh , h. h. , meech , k. j. , & pittichov , j. 2011b .
main - belt comet 238p / read revisited .
astrophys .
j. 736 , l18 .
hsieh , h. h. , denneau , l. , wainscoat , r. j. , fitzsimmons , a. , armstrong , j. d. , yang , b. , hergenrother , c. 2011c .
vw_139 = ( 300163 ) 2006 vw_139 .
bureau electr .
telegrams 2920 , 1 .
hsieh , h. h. , yang , b. , haghighipour , n. 2012a . optical and dynamical characterization of comet - like main - belt asteroid ( 596 ) scheila .
astrophys .
j. 744 , 9 .
hsieh , h. h. , yang , b. , haghighipour , n. , et al . 2012b .
discovery of main - belt comet p/2006 vw@xmath48 by pan - starrs1 .
astrophys .
j. 748 , l15 .
hsieh , h. h. , yang , b. , haghighipour , n. , et al .
observational and dynamical characterization of main - belt comet p/2010 r2 ( la sagra ) .
astron . journal .
143 , 104 .
hsieh , h. h. , denneau , l. , micheli , m. , et al .
x2 ( panstarrs ) .
bureau electr .
telegrams 3343 , 1 .
hsieh , h. h. , kaluna , h. m. , novakovi , b. , et al .
2013a . main - belt comet p/2012 t1 ( panstarrs ) .
astrophys .
j. 771 , l1 .
hsieh , h. h. , pohlen , m. , matulonis , a. 2013 .
comet 133p / elst - pizarro .
bureau electr .
telegrams 3564 , 1 .
hsieh , h. h. , denneau , l. , fitzsimmons , a. , et al . 2014 .
search for the return of activity in active asteroid 176p / linear .
j. 147 , 89 .
ishiguro , m. , hanayama , h. , hasegawa , s. , et al .
observational evidence for an impact on the main - belt asteroid ( 596 ) scheila .
astrophys .
j. 740 , l11 .
ishiguro , m. , hanayama , h. , hasegawa , s. , et al .
interpretation of ( 596 ) scheila s triple dust tails .
astrophys .
j. 741 , l24 .
ishiguro , m. , jewitt , d. , hanayama , h. , et al .
2014 . outbursting comet p/2010 v1 ( ikeya - murakami ) : a miniature comet holmes .
astrophys .
j. in press .
jewitt , d. 1996 . from comets to asteroids
: when hairy stars go bald .
earth , moon , and planets .
72 , 185 - 201 .
jewitt , d. 2002 . from kuiper belt object to cometary nucleus : the missing ultrared matter .
j. 123 , 1039 - 1049 .
jewitt , d. 2009 .
the active centaurs .
j. 137 , 4296 - 4312 .
jewitt , d. 2012 .
the active asteroids . astronomical journal .
jewitt , d. , lacerda , p. , peixinho , n. 2007 .
comet 133p / elst - pizarro .
iau circ . 8847 , 1 .
jewitt , d. , yang , b. , & haghighipour , n. 2009 .
main - belt comet p/2008 r1 ( garradd ) .
j. 137 , 4313 - 4321 .
jewitt , d. , weaver , h. , agarwal , j. , mutchler , m. , & drahus , m. 2010 . a recent disruption of the main - belt asteroid p/2010 a2 . nature .
467 , 817 - 819 .
jewitt , d. , stuart , j. s. , li , j. 2011a .
pre - discovery observations of disrupting asteroid p/2010 a2 .
j. 142 , 28 .
jewitt , d. , weaver , h. , mutchler , m. , larson , s. , & agarwal , j. 2011b .
hubble space telescope observations of main - belt comet ( 596 ) scheila .
astrophys .
j. 733 , l4 .
jewitt , d. , ishiguro , m. , agarwal , j. 2013a .
large particles in active asteroid p/2010 a2 .
astrophys .
j. 764 , l5 .
jewitt , d. , li , j. , agarwal , j. 2013b .
the dust tail of asteroid ( 3200 ) phaethon .
astrophys .
j. 771 , l36 .
jewitt , d. , agarwal , j. , weaver , h. , mutchler , m. , & larson , s. 2013c . the extraordinary multi - tailed main - belt comet p/2013 p5 . astrophys .
j. 778 , l21 .
jewitt , d. , agarwal , j. , li , j. , weaver , h. a. , mutchler , m. , larson , s. 2014a . disintegrating asteroid p/2013 r3 .
astrophys .
j. 784 , l8 .
jewitt , d. , ishiguro , m. , weaver , h. a. , agarwal , j. , mutchler , m. , larson , s. 2014b .
hubble space telescope investigation of main - belt comet 133p / elst - pizarro .
j. 147 , 117 .
kaluna , h. , & meech , k. 2011 .
analysis of the activity on main belt comet 133p / elst - pizarro .
epsc - dps joint meeting 2011 .
keil , k. 2000 .
thermal alteration of asteroids : evidence from meteorites .
planet . and
space sci .
48 , 887 - 903 .
kendrew , s. , simpson , r. , bressert , e. , et al .
the milky way project : a statistical study of massive star formation associated with infrared bubbles .
astrophys .
j. 755 , 71 .
kossacki , k. j. , szutowicz , s. 2012 .
main belt comet p/2008 r1 garradd : duration of activity .
217 , 66 - 76 .
kowalski , r. a. , hsieh , h. h. , tholen , d. j. , micheli , m. , elliott , g. t. , williams , g. v. 2012 .
sb_6 ( lemmon ) .
kresk , l. 1972 . on the dividing line between cometary and asteroidal orbits . in chebotarev g.
a. , kazimirchak - polonskaia e. i. , marsden b. g. , eds , proc .
iau symp .
45 , the motion , evolution of orbits , and origin of comets .
reidel , dordrecht .
503 - 514 .
kppers , m. , orourke , l. , bockele - morvan , d. , et al .
localized sources of water vapor on the dwarf planet ( 1 ) ceres . nature .
505 , 525 - 527 .
larson , s. m. 2010 .
( 596 ) scheila .
larson , s. m. , micheli , m. , wainscoat , r. j. , elliott , g. t. , williams , g. v. 2012 .
tb_36 ( lemmon ) .
bureau electr .
telegrams 3269 , 1 .
li , j .- y . , kelley , m. s. p. , knight , m. m. , et al .
2013 . characterizing the dust coma of comet
c/2012 s1 ( ison ) at 4.15 au from the sun .
astrophys .
j. 779 , l3 .
licandro , j. , moreno , f. , de len , j. , tozzi , g. p. , lara , l. m. , cabrera - lavers , a. 2013a . exploring the nature of new main - belt comets with the 10.4 m gtc telescope : ( 300163 )
2006 vw@xmath48 .
550 , a17 licandro , j. , moreno , f. , cabrera - lavers , a. , et al .
r3 ( catalina - panstarrs ) .
bureau electr .
telegrams 3679 , 1 .
lin , h. w. , chen , y. t. , lacerda , p. , et al .
pan - starrs 1 observations of the unusual active centaur p/2011 s1 ( gibbs ) .
j. in press .
lintott , c. j. , schawinski , k. , ane , s. , et al .
galaxy zoo : morphologies derived from visual inspection of galaxies from the sloan digital sky survey .
mon . not .
r. astron .
389 , 1179 - 1189 .
lintott , c. j. , schawinski , k. , keel , w. , et al .
galaxy zoo : ` hanny s voorwerp ' , a quasar light echo ?
mon . not .
r. astron .
399 , 129 - 140 .
lowry , s. c. , fitzsimmons , a. 2005 .
william herschel telescope observations of distant comets .
mon . not .
r. astron .
358 , 641 - 650 .
maclennan , e. m. , & hsieh , h. h. 2012 .
the nucleus of main - belt comet 259p / garradd .
astrophys .
j. 758 , l3 .
magnier , e. 2006 . the pan - starrs ps1 image processing pipeline .
proceedings of the advanced maui optical and space surveillance technologies conference , ed . : s. ryan ( kihei : the maui economic development board ) .
magnier , e. a. , schlafly , e. f. , finkbeiner , d. p. , et al .
2013 . the pan - starrs1 photometric reference ladder , release 12.01 .
astrophys .
j. suppl . series .
mainzer , a. , read , m. t. , scotti , j. v. , ryan , w. h. , ryan , e. v. 2010 .
iau circ . 9117 , 1 .
masiero , j. r. , jedicke , r. , durech , j. , gwyn , s. , denneau , l. , larsen , j. 2009 .
the thousand asteroid light curve survey .
204 , 145 - 171 .
mccord , t. b. , sotin , c. 2005 .
ceres : evolution and current state .
journal of geophysical research ( planets ) , 110 , e05009 mccord , t. b. , castillo - rogez , j. , rivkin , a. 2011 .
ceres : its origin , evolution and structure and dawn s potential contribution .
space sci .
163 , 63 - 76 .
meech , k. j. , yang , b. , kleyna , j. , et al .
outgassing behavior of c/2012 s1 ( ison ) from 2011 september to 2013 june . astrophs .
j. 776 , l20 .
morbidelli , a. , chambers , j. , lunine , j. i. , petit , j. m. , robert , f. , valsecchi , g. b. , & cyr , k. e. 2000 . source regions and timescales for the delivery of water to the earth .
meteoritics & planet .
35 , 1309 - 1320 .
moreno , f. , lara , l. m. , licandro , j. , ortiz , j. l. , de len , j. , al - lagoa , v. , ags - gonzlez , b. , molina , a. 2011a .
the dust environment of main - belt comet p/2010 r2 ( la sagra ) .
astrophys .
j. 738 , l16 .
moreno , f. , licandro , j. , ortiz , j. l. , et al .
( 596 ) scheila in outburst : a probable collision event in the main asteroid belt .
astrophys .
j. 738 , 130 .
moreno , f. , licandro , j. , cabrera - lavers , a. 2012 .
a short - duration event as the cause of dust ejection from main - belt comet p/2012 f5 ( gibbs ) .
astrophys .
j. 761 , l12 .
moreno , f. , cabrera - lavers , a. , vaduvescu , o. , licandro , j. , pozuelos , f. 2013 .
the dust environment of main - belt comet p/2012 t1 ( panstarrs ) .
astrophys .
j. 770 , l30 .
moreno , f. , licandro , j. , lvarez - iglesias , c. , cabrera - lavers , a. , & pozuelos , f. 2014 .
intermittent dust mass loss from activated asteroid p/2013 p5 ( panstarrs ) .
astrophys .
j. 781 , 118 nakano , s. , ikeya , k. , murakami , s. itagaki , k. 2010 .
comet 2010 v1 .
nevski , v. , novichonok , a. , burhonov , o. , et al .
s1 ( ison ) .
bureau electr .
telegrams 3238 , 1 .
nomen , j. , marsden , b. g. , birtwhistle , p. , et al .
p/2010 r2 ( la sagra ) .
novakovi , b. , hsieh , h. h. , cellino , a. 2012 .
p/2006 vw@xmath48 : a main - belt comet born in an asteroid collision ?
mon . not .
r. astron .
424 , 1432 - 1441 .
novakovi , b. , hsieh , h. h. , cellino , a. , micheli , m. , pedani , m. 2014 .
discovery of a young asteroid cluster associated with p/2012 f5 ( gibbs ) .
231 , 300 - 309 .
obrien , d. p. , morbidelli , a. , levison , h. f. 2006 .
terrestrial planet formation with strong dynamical friction .
184 , 39 - 58 .
orourke , l. , bockele - morvan , d. , biver , n. , et al .
herschel and iram-30 m observations of comet c/2012 s1 ( ison ) at 4.5 au from the sun .
560 , a101 .
prialnik , d. , rosenberg , e. d. 2009 .
can ice survive in main - belt comets ?
long - term evolution models of comet 133p / elst - pizarro . mon . not .
r. astron .
399 , l79-l83 .
raymond , s. n. , quinn , t. , lunine , j. i. 2004 .
making other earths : dynamical simulations of terrestrial planet formation and water delivery .
168 , 1 - 17 . read , m. t. , bressi , t. h. , gehrels , t. , scotti , j. v. , & christensen , e. j. 2005 . comet p/2005
u1 ( read ) .
rinner , c. , wainscoat , r. j. , denneau , l. , et al .
ce_31 ( moss ) .
bureau electr .
telegrams 3418 rivkin , a. s. , howell , e. s. , vilas , f. , lebofsky , l. a. 2002 .
hydrated minerals on asteroids : the astronomical record .
asteroids iii , 235 - 253 .
rivkin , a. s. , emery , j. p. 2010 .
detection of ice and organics on an asteroidal surface .
464 , 1322 - 1323 .
russell , c. t. , aroian , r. , arghavani , m. , & nock , k. 1984 .
interplanetary magnetic field enhancements and their association with the asteroid 2201 oljato . science .
226 , 43 - 45 .
schlafly , e. f. , finkbeiner , d. p. , juri , m. , et al .
photometric calibration of the first 1.5 years of the pan - starrs1 survey .
astrophys .
j. 756 , 158 .
schrghofer , n. 2008 . the lifetime of ice on main belt asteroids .
astrophys .
j. 682 , 697 - 705 .
schwamb , m. e. , lintott , c. j. , fischer , d. a. , et al .
planet hunters : assessing the _ kepler _ inventory of short - period planets . astrophys .
j. 754 , 129 .
schwartz , m. , holvorcem , p. r. , wainscoat , r. j. , micheli , m. , elliott , g. t. , williams , g. v. 2012 .
tk_8 ( tenagra ) .
bureau electr .
telegrams 3262 , 1 .
smith , a. m. , lynn , s. , sullivan , m. , et al .
galaxy zoo supernovae .
mon . not .
r. astron .
412 , 1309 - 1319 .
snodgrass , c. , tubiana , c. , vincent , j .- b .
, et al . 2010 . a collision in 2009 as the origin of the debris trail of asteroid p/2010 a2 .
467 , 814 - 816 .
solano , e. , rodrigo , c. , pulido , r. , carry , b. 2014 .
precovery of near - earth asteroids by a citizen - science project of the spanish virtual observatory .
335 , 142 - 149 .
solontoi , m. , ivezi , . , west , a. a. , et al . 2010 . detecting active comets in the sdss .
205 , 605 - 618 .
sonnett , s. , kleyna , j. , jedicke , r. , masiero , j. limits on the size and orbit distribution of main belt comets .
215 , 534 - 546 stevenson , r. , kramer , e. a. , bauer , j. m. , masiero , j. r. , mainzer , a. k. 2012 .
characterization of active main belt object p/2012 f5 ( gibbs ) : a possible impacted asteroid .
j. 759 , 142 .
strauss , m. a. , weinberg , d. h. , lupton , r. h. , et al .
spectroscopic target selection in the sloan digital sky survey : the main galaxy sample .
j. 124 , 1810 - 1824 .
takir , d. , emery , j. p. 2012 .
outer main belt asteroids : identification and distribution of four 3-@xmath47 m spectral groups .
219 , 641 - 654 .
tedesco , e. f. , noah , p. v. , noah , m. , price , s. d. 2004 .
iras minor planet survey .
nasa planetary data system , iras - a - fpa-3-rdr - imps - v6.0 .
tholen , d. j. , hartmann , w. k. , cruikshank , d. p. , lilly , s. , bowell , e. , hewitt , a. 1988 .
( 2060 ) chiron .
tonry , j. l. , stubbs , c. w. , lykke , k. r. , et al .
the pan - starrs1 photometric system .
j. 750 , 99 .
tonry , j. l. 2013 .
pan - starrs , atlas , and optical transient searches . royal society of london philosophical transactions series a. 371 , 201220269 .
tth , i. 2006 .
search for comet - like activity in asteroid 7968 elst - pizarro and limitation of its rotational pole orientation .
446 , 333 - 343 .
vere , p. , micheli , m. , wainscoat , r. j. , et al .
wa_34 ( lemmon - panstarrs ) .
bureau electr .
telegrams 3379 , 1 .
vere , p. , denneau , l. , hsieh , h. h. , et al .
g4 ( panstarrs ) .
bureau electr .
telegrams 3473 , 1 .
vere , p. , wainscoat , r. j. , hsieh , h. h. , micheli , m. , forshay , p. , williams , g. v. 2013c .
cu_129 ( panstarrs ) .
bureau electr .
telegrams 3558 , 1 .
vinson , k. , thilker , d. 2013 . a boinc based citizen - science project for pixel spectral energy distribution fitting of resolved galaxies in multi - wavelength surveys . astronomy and computing . 3 , 1 - 12 .
walsh , k. j. , morbidelli , a. , raymond , s. n. , obrien , d. p. , mandell , a. m. 2011 . a low mass for mars from jupiter s early gas - driven migration .
475 , 206 - 209 .
wainscoat , r. j. , micheli , m. , wells , l. , et al .
l4 ( panstarrs ) .
wainscoat , r. j. , vere , p. , bolin , b. , birtwhistle , p. , armstrong , j. d. , micheli , m , sato , h. , williams , g. v. 2012 .
s3 ( panstarrs ) .
bureau electr .
telegrams 3244 , 1 .
wainscoat , r. j. , hsieh , h. h. , denneau , l. , et al .
t1 ( panstarrs ) .
bureau electr .
telegrams 3252 , 1 .
wainscoat , r. j. , hsieh , h. h. , sonnett , s. , et al .
t2 ( panstarrs ) .
bureau electr .
telegrams 3254 , 1 .
wainscoat , r. j. , denneau , l. , hsieh , h. h. , kaluna , h. m. , sheppard , s. s. , trujillo , c. a. , williams , g. v. 2012 .
t3 ( panstarrs ) .
bureau electr .
telegrams 3255 , 1 .
wainscoat , r. j. , micheli , m. , sato , h. , et al .
u1 ( panstarrs ) .
bureau electr .
telegrams 3264 , 1 .
wainscoat , r. j. , hsieh , h. h. , bolin , b. , buzzi , l. , sato , h. , williams , g. v. 2012 .
u2 ( panstarrs ) .
bureau electr .
telegrams 3268 , 1 .
wainscoat , r. , micheli , m. , guido , e. , howes , n. , williams , g. v. 2013a .
g3 ( panstarrs ) .
bureau electr .
telegrams 3472 , 1 .
wainscoat , r. , denneau , l. , micheli , m. , woodworth , d. , wells , l. , williams , g. v. 2013b .
j4 ( panstarrs ) .
bureau electr .
telegrams 3521 , 1 .
wainscoat , r. , vere , p. , micheli , m. , et al .
p4 ( panstarrs ) .
bureau electr .
telegrams 3638 , 1 .
wainscoat , r. j. , denneau , l. , vere , p. , et al .
t1 ( panstarrs ) .
bureau electr .
telegrams 3668 , 1 .
waszczak , a. , ofek , e. o. , aharonson , o. , et al .
main - belt comets in the palomar transient factory survey - i. the search for extendedness . mon . not .
r. astron .
433 , 3115 - 3132 .
yang , b. , keane , j. , meech , k. j. , owen , t. , wainscoat , r. j. 2014 .
multi - wavelength observations of comet c/2011 l4 ( panstarrs ) .
astrophys .
j. 784 , l23 . | we analyze a set of 760475 observations of 333026 unique main - belt objects obtained by the pan - starrs1 ( ps1 ) survey telescope between 2012 may 20 and 2013 november 9 , a period during which ps1 discovered two main - belt comets , p/2012 t1 ( panstarrs ) and p/2013 r3 ( catalina - panstarrs ) .
ps1 comet detection procedures currently consist of the comparison of the point spread functions ( psfs ) of moving objects to those of reference stars , and the flagging of objects that show anomalously large radial psf widths for human evaluation and possible observational follow - up . based on the number of missed discovery opportunities among comets discovered by other observers
, we estimate an upper limit comet discovery efficiency rate of @xmath0% for ps1 .
additional analyses that could improve comet discovery yields in future surveys include linear psf analysis , modeling of trailed stellar psfs for comparison to trailed moving object psfs , searches for azimuthally localized activity , comparison of point - source - optimized photometry to extended - source - optimized photometry , searches for photometric excesses in objects with known absolute magnitudes , and crowd - sourcing .
analysis of the discovery statistics of the ps1 survey indicates an expected fraction of 59 mbcs per @xmath1 outer main - belt asteroids ( corresponding to a total expected population of @xmath2 mbcs among the outer main - belt asteroid population with absolute magnitudes of @xmath3 ) , and a 95% confidence upper limit of 96 mbcs per @xmath1 outer main - belt asteroids ( corresponding to a total of @xmath4230 mbcs ) , assuming a detection efficiency of 50% .
we note however that significantly more sensitive future surveys ( particularly those utilizing larger aperture telescopes ) could detect many more mbcs than estimated here .
examination of the orbital element distribution of all known mbcs reveals an excess of high eccentricities ( @xmath5 ) relative to the background asteroid population .
theoretical calculations show that , given these eccentricities , the sublimation rate for a typical mbc is orders of magnitude larger at perihelion than at aphelion , providing a plausible physical explanation for the observed behavior of mbcs peaking in observed activity strength near perihelion .
these results indicate that the overall rate of mantle growth should be slow , consistent with observational evidence that mbc activity can be sustained over multiple orbit passages .
[ accepted for publication in icarus , 2014 oct 19 ] asteroids ; comets ; astrobiology ; asteroids , composition ; comets , origin |
there has been considerable discussion recently about strange matrix elements of the nucleon @xcite .
much impetus for this was generated by the deep - inelastic scattering experiments with polarized targets at cern and more recently at slac @xcite , which seemed to imply a large polarized strange quark distribution in the proton . at about the same time
a measurement of the elastic neutrino proton scattering cross section @xcite at lower values of @xmath0 had also suggested a non - zero value for the strange axial vector form factor of the proton @xcite .
these observations spurred many investigations of other processes in which traces of strangeness in the nucleon could be detected .
it was argued @xcite for example that semi - leptonic neutral current scattering experiments could be used to extract the strange vector as well as axial vector form factors .
parity - violating electron scattering experiments were proposed as a means of probing neutral current form factors @xcite . approved parity - violating experiments at mit - bates @xcite and
jefferson lab @xcite will provide information on the strangeness form factors at low and intermediate @xmath0 values , and more precisely determine the strange radius and magnetic moment of the nucleon . at higher @xmath0
, the ccfr collaboration @xcite has recently investigated a possible asymmetry between the strange and antistrange quark distributions in neutrino deep - inelastic scattering .
perturbative qcd alone would be expected to produce identical @xmath2 and @xmath3 distributions , while any asymmetry would imply the presence of non - perturbative effects in the nucleon at deep - inelastic scales .
of course the strangeness content of the nucleon is not a scale - invariant concept strange matrix elements are not renormalization group invariant . the evolution with @xmath0 of the moments of the polarized deep - inelastic strange quark distribution can be studied using the altarelli - parisi equations , so that a zero value for @xmath4 at a low scale is not necessarily incompatible with a non - zero value at a higher scale @xcite .
naturally , extending such analyses to very low values of @xmath0 is problematic at some stage non - perturbative techniques or models of qcd need to be invoked when describing data .
recently tang and ji @xcite attempted to link the strangeness radius of the nucleon with the densities of strange and antistrange deep - inelastic quark distributions , arguing on the basis of quark `` locality '' that the distributions in coordinate space and momentum space should be correlated . in order to investigate this proposal more quantitatively , we will explore this relationship within a simple model , namely the meson cloud model . in the meson cloud model ,
the strangeness of the nucleon is assumed to be carried by the kaon
hyperon components of the physical nucleon .
this model has in the past been utilized to calculate corrections to low energy nucleon properties @xcite , and has also been invoked to describe flavor symmetry breaking in nucleon sea quark distributions @xcite . however , direct comparison of results for the strange matrix elements in meson cloud models has thus far not been possible due to the fact that form factors and structure functions are usually calculated within different theoretical frameworks . while the natural framework for analyzing strange quark distributions is the light - cone @xcite , strange form factor calculations
have usually been performed within covariant perturbation theory in instant - form quantization @xcite .
( a light - cone constituent quark model was used in ref.@xcite to obtain the coupling of the photon to the kaon
hyperon cloud of the nucleon , however the kaon dressing itself was not described in terms of light - cone dynamics . )
meaningful comparison of meson cloud corrections to light - cone quark distributions and low-@xmath0 form factors obviously requires a single framework to be used for both . in the present work we shall present for the first time a light - cone analysis of the strange nucleon form factors , using the same framework in which structure functions are calculated @xcite .
we believe this is the only systematic way to test the hypothesis of ref.@xcite that the properties of the nucleon sea at low and high @xmath0 might somehow be correlated .
in sec.ii we briefly outline the pertinent features of the meson cloud model on the light - cone .
more detailed accounts can be found in ref.@xcite .
we explain that the advantage of this framework lies in that it allows for a more consistent treatment of the hadronic meson baryon vertex functions . unlike previous calculations within covariant perturbation theory , which have usually had to make _ ad hoc _ assumptions about the dependence of the vertex functions on various loop momenta
, one can formally avoid this on the light - cone , while simultaneously satisfying charge and momentum conservation .
the kaon cloud model is applied in sec.iii to the calculation of the @xmath0 dependence of the strange vector and axial vector form factors of the nucleon , from which the strangeness radius , strange magnetic moment and strange axial charge are extracted .
sec.iv discusses the contribution from the kaon cloud to the deep - inelastic strange quark distribution , and to a possible strange antistrange asymmetry in the nucleon .
we find that the asymmetry is consistent with the latest neutrino deep - inelastic scattering data , within experimental errors .
we also estimate the contribution to the polarized strange quark distribution from kaon loops . finally our findings are summarized in sec.v .
the basic hypothesis of the meson cloud model of the nucleon is that the nucleon on the light - cone has internal meson and baryon degrees of freedom .
the physical nucleon state ( momentum @xmath5 ) can then be expanded ( in the one - meson approximation ) as a series involving bare nucleon and two - particle meson baryon states : @xmath6 the function @xmath7 is the probability amplitude for the physical nucleon @xmath8 to be in a state consisting of a baryon @xmath9 and meson @xmath10 , having transverse momenta @xmath11 and @xmath12 , and carrying longitudinal momentum fractions @xmath13 and 1@xmath14 , respectively .
the bare nucleon probability is denoted by @xmath15 , and @xmath16 is the @xmath17 coupling constant .
the one - meson approximation in eq.([fock ] ) is valid as long as the meson cloud is relatively soft ( @xmath18 ) .
it will progressively break down for harder @xmath17 vertex functions , at which point one will need to include two - meson and higher order fock state components in eq.([fock ] ) .
the strangeness of the nucleon in this model is carried by the @xmath19 , @xmath20 , etc .
fock state components , and we shall consider only these henceforth . therefore the properties of the bare nucleon state ( such as its intrinsic form factor , or its structure function ) will not be relevant for considerations of the nucleon strangeness content .
for the sake of presentation we will consider the @xmath21 hyperon as representative in our discussions , although contributions from the @xmath22 will be included in the numerical results in secs.iii & iv .
all expressions for the @xmath23 distributions can be obtained from the @xmath24 by appropriate replacement of masses , coupling constants and isospin factors .
contributions from heavier mesons and hyperons , such as @xmath25 , @xmath26 , etc . can easily be added , although their role is insignificant as long as the @xmath17 vertex functions are not very hard @xcite . according to eq.([fock ] ) , the probability to find a @xmath21 inside a nucleon with light - cone momentum fraction @xmath27 is given by the @xmath28 splitting function ( to leading order in the coupling constant ) : @xmath29 charge and momentum conservation require that this must also be the probability to find a kaon inside a nucleon with light - cone momentum fraction @xmath30 .
the kaon distribution function , @xmath31 , should therefore be related to the baryon distribution function by @xmath32 for all @xmath27 , if the above interpretation is valid .
this guarantees equal numbers of mesons emitted by the nucleon , @xmath33 , and virtual baryons accompanying them , @xmath34 : [ cons ] @xmath35 this is just the statement that the nucleon has zero net strangeness .
momentum conservation imposes the further requirement that @xmath36 where @xmath37 and @xmath38 are the average momentum fractions carried by kaon and @xmath21 , respectively .
equations ( [ ncons ] ) and ( [ mcons ] ) , and in fact similar relations for all higher moments of @xmath39 , follow automatically from eq.([sym ] ) .
any consistent treatment of the meson cloud model must therefore reproduce this symmetry relation .
the advantage of a light - cone formulation @xcite is that one can construct a covariant framework in which eq.([sym ] ) is manifestly preserved , regardless of the dynamics of the meson baryon system . if the dynamics are parametrized in terms of an effective hadronic meson baryon vertex function , the symmetry relations ( [ sym])([cons ] ) can be satisfied if the vertex functions are functions of the invariant mass squared , @xmath40 , of the @xmath24 system : @xmath41 where @xmath42 and @xmath43 are the masses of @xmath21 and kaon , respectively , and @xmath44 and @xmath45 their four - momenta .
note that the variable @xmath40 is related to the virtualities of the kaon ( @xmath46 ) and @xmath21 ( @xmath47 ) by : @xmath48 where @xmath49 and @xmath50 . a hadronic vertex function which is a function of @xmath40
will automatically have the correct crossing symmetry structure in the @xmath51 and @xmath52 channels @xcite ( see eqs.([ff ] ) , ( [ fft ] ) below ) . in instant - form calculations , on the other hand , the functions @xmath39 are usually evaluated in terms of @xmath53 vertex functions taken from fits to @xmath54 scattering data @xcite , where the vertex functions are assumed to depend only on the virtuality @xmath51 of the off - shell kaon .
in this case the @xmath55 symmetry is lost , and the relation ( [ sym ] ) can not be maintained , making verification of the charge and momentum conservation relations , eqs.([cons ] ) , very difficult @xcite . actually , the non - preservation of the relation ( [ sym ] ) is not surprising since the splitting functions @xmath39 refer to probability distributions of specific _ particles _ , and a probabilistic interpretation does not hold in all reference frames ( since particle number is not preserved by lorentz boosts ) . indeed , for structured particles
the factorization of the @xmath56 cross section itself into @xmath57 and @xmath58 ( or @xmath59 and @xmath60 ) cross sections is not valid in all frames of reference @xcite .
such factorization , or convolution , can in fact only be achieved by eliminating antiparticle degrees of freedom , which can formally be done only in the infinite momentum frame @xcite , or on the light - cone .
we will be interested in the strange matrix elements @xmath61 of operators @xmath62 , where @xmath63 or @xmath64 .
while these have been investigated in earlier studies within the meson cloud model @xcite , we will present here the first analysis of the strange form factors on the light - cone , in which the distribution functions of kaons and hyperons in the nucleon have the correct symmetry properties as specified in eq.([sym ] ) .
this will be our main contribution to the form factor discussion .
other aspects of the problem , such as the modeling of the intrinsic @xmath65 and @xmath21 form factors , or the truncation of the fock state expansion in eq.([fock ] ) , are not developed here beyond what exists already in the literature .
the momentum dependence of the matrix elements of the strange vector current @xmath66 is conventionally written : @xmath67 where @xmath5 and @xmath68 are the initial and final nucleon momenta , and @xmath69 is the momentum transferred to the nucleon .
the form factors @xmath70 and @xmath71 are the dirac and pauli strange form factors , and @xmath72 is the free dirac spinor of the nucleon . for convenience we work in a lorentz frame ( viz . the breit frame ) where the photon momentum @xmath73 is purely transverse , so that @xmath74 .
the various particle momenta are parameterized as in ref.@xcite .
zero net strangeness in the nucleon requires the vanishing of the dirac form factor at the on - shell photon point , @xmath75 . on the light - cone
the form factors can be evaluated by choosing the @xmath76 ( or `` good '' ) component of the current @xmath77 @xcite .
usually one introduces more convenient combinations of @xmath78 , in the form of the sachs electric ( @xmath79 ) and magnetic ( @xmath80 ) form factors : @xmath81 the conventional definition of the strangeness radius of the nucleon is in terms of @xmath79 : @xmath82 an alternative definition of the strangeness radius is in terms of the dirac form factor : @xmath83 the strange magnetic moment of the nucleon is defined by : @xmath84 the contributions of the kaon cloud to the strange form factors are represented in fig.1 . to maintain the analogy with the structure function calculations in sec.iv
, we will present the results for @xmath78 , and evaluate @xmath85 from these .
therefore we define : @xmath86 where @xmath87 and @xmath88 refer to the @xmath21 and @xmath65 interaction diagrams in figs.1(a ) and ( b ) , respectively . the @xmath21 contribution to @xmath89
can be written : @xmath90 where the @xmath21 light - cone distribution functions @xmath91 are : [ f12l ] @xmath92 for the @xmath53 vertex we assume a pseudoscalar @xmath93 interaction ( the same results are obtained with a pseudovector coupling ) , with @xmath94 the coupling constant and @xmath95 the hadronic vertex function .
note that identical expressions to these can be obtained using time - ordered perturbation theory in the infinite momentum frame @xcite .
for the sign of the strangeness we adopt the convention of jaffe @xcite , so that @xmath96 is the strangeness charge of the @xmath21 . the intrinsic @xmath21 form factor @xmath97 contains possible @xmath0 dependence in the @xmath59 interaction vertex . the squared center of mass energies in eqs.([f12l ] )
are : @xmath98 with @xmath40 defined in eq.([m2 ] ) .
the contributions to @xmath78 from the coupling to the kaon in fig.1(b ) are written : @xmath99 where @xmath100 is defined to be the strangeness charge of the kaon , and where the kaon light - cone distribution functions @xmath101 are : [ f12k ] @xmath102 the @xmath24 squared center of mass energies are : @xmath103 as discussed in sec.ii , on the light - cone it is natural to take the @xmath53 vertex function to be a function of @xmath104 , which guarantees local gauge invariance @xcite as well as energy - momentum conservation , as embodied in eq.([sym ] ) . in
instant - form approaches gauge invariance has sometimes been enforced through introduction of seagull diagrams @xcite , which can be generated according to various prescriptions @xcite in order to satisfy the ward - takashi identity .
however , as observed by musolf and burkardt @xcite , such prescriptions are not unique , since additional seagull terms which individually satisfy the ward - takahashi identity are also allowed .
furthermore , inconsistencies with some of the prescriptions , when applied to loop calculations , have also recently been pointed out by wang and banerjee @xcite . with @xmath40 dependent vertex functions the functions @xmath105 and @xmath106 in eqs.([f1k ] ) and ( [ f1l ] ) at @xmath107
satisfy the relation ( [ sym ] ) explicitly , which they must because of strangeness conservation .
the functions @xmath108 and @xmath109 at @xmath107 , on the other hand , do not need to satisfy such a relation , because they have their origin in the different spin couplings of the @xmath110 to the spin-0 @xmath65 and spin-1/2 @xmath21 .
indeed , the contribution to the strange magnetic moment of the nucleon would be zero if @xmath111 exhibited such a symmetry . from eqs.([f2l ] ) and ( [ f2k ] ) one can see in fact that @xmath112 in the kaon cloud model is always @xmath113 .
although it is clear that in order to preserve the relation ( [ sym ] ) the vertex function should be a function of the @xmath24 invariant mass , its specific functional dependence is not prescribed , and several forms have been suggested in the literature @xcite . in our numerical studies
we will use a simple monopole - type function : @xmath114 where @xmath115 is the cut - off mass parameter . in the light - cone formulation
the cut - off directly determines the probability of finding the physical nucleon in a @xmath24 configuration , which provides a constraint on the range of @xmath115 for which the average kaon number density can be viewed as reliable .
( in covariant instant - form calculations the probability for each individual fock state component is frame - dependent , since particle number is not invariant under lorentz boosts . ) for values larger than @xmath116 gev the average probability to find kaons in the nucleon would be @xmath117 , so that without including higher fock state components , the one - meson approximation in eq.([fock ] ) could not be considered trustworthy .
therefore cut - off masses between @xmath118 and @xmath119 gev , for which @xmath120 37% , can be considered as representative of a reasonable range for which the model ( [ fock ] ) is a valid approximation ( for these values the corresponding probability to find pions in the nucleon is @xmath121 30%50% ) . on the other hand
, one can also try to constrain @xmath115 phenomenologically by fitting , within the kaon cloud model , the available strange quark distribution data @xcite , as well as the inclusive @xmath122 production data @xcite . for the monopole parametrization ( [ ff ] )
one typically finds values of @xmath123 gev , with larger cut - offs difficult to accommodate @xcite ( see also sec.iv ) . in instant - form calculations @xcite
one has often tried to connect the vertex function cut - offs with those obtained from @xmath124 potential model fits @xcite , where the vertex function is assumed to depend on the virtuality @xmath51 of the kaon , @xmath125 here the vertex function cut - offs are found to be typically @xmath126 1.21.5 gev .
note that the form ( [ fft ] ) is an approximation to the @xmath40 dependent light - cone vertex function in eq.([ff ] ) , obtained by taking @xmath127 and @xmath128 , see eq.([stu ] ) .
a typical value of @xmath129 gev would therefore correspond to @xmath130 gev .
the function @xmath131 in eq.([fik ] ) , like @xmath97 in eq.([filam ] ) , reflects the possible @xmath0 dependence in the interaction vertex in fig.1(b ) .
the simplest approximation is to assume point - like @xmath57 and @xmath59 couplings , @xmath132 @xcite .
of course the virtual @xmath65 and @xmath21 do have finite size , and their coupling to @xmath110 should in reality exhibit some @xmath0 dependence . the most natural way to model this @xmath0 dependence is through the vector meson dominance model , in which the virtual photon at low @xmath0 couples to the kaon or @xmath21 through its fluctuations into correlated @xmath133 pairs @xcite . since the photon has @xmath134 , the states with the correct quantum numbers to which the @xmath110 can couple are the vector mesons , @xmath135 , @xmath136 and @xmath137 .
indeed , the @xmath136 and @xmath137 mesons can act as an extra source of strangeness in the nucleon through this mechanism .
a detailed treatment of the vector meson dominance model in strange form factors was presented in refs.@xcite , where it was found that @xmath138 increased by a factor @xmath139 with respect to the kaon loop only result ( @xmath112 is of course not affected by any changes in the @xmath0 dependence ) .
other , quark - type models have also been used to estimate the intrinsic form factors of the bare particles @xcite . however , as outlined in sec.i , our aim here is more to explore the relationship between strange quark distributions in coordinate and momentum space within a single specific model , rather than present a comprehensive analysis of the form factors and structure functions themselves .
therefore in order to avoid diluting the result from the kaon cloud model by introducing extra degrees of freedom into the calculation , for our purposes we will begin by investigating the contributions to the @xmath0 dependence of the strange form factors arising from kaon loops alone . the @xmath0 dependence of the strange sachs electric and magnetic @xmath79 and @xmath80 form factors is shown in figs.2 and 3 for cut - off masses @xmath140 and @xmath119 gev .
the values of the masses and couplings used in the numerical calculations are @xmath141 mev , @xmath142 mev , @xmath143 mev , @xmath144 mev and @xmath145 , @xmath146 @xcite .
the results in figs.2 & 3 contain contributions from both @xmath24 and @xmath23 components , with the latter contributing @xmath147 of the total , which reflects the ratio @xmath148 expected from su(3 ) symmetry .
contributions from @xmath149 loops to the strange form factors , where @xmath26 represents the @xmath150 decouplet hyperons , are suppressed by more than two orders of magnitude relative to the @xmath24 loops .
the sign of @xmath79 is positive , while that of @xmath80 negative , in agreement with the instant - form kaon cloud model results of refs.@xcite .
the magnitude depends strongly on @xmath115 , although even for the largest cut - off of @xmath119 gev it is still somewhat smaller than in refs.@xcite .
the strangeness ( sachs ) radius is found to be small and negative , ranging from @xmath151 @xmath152 for @xmath153 gev to @xmath154 @xmath152 for @xmath155 gev .
this is slightly smaller than that obtained in previous kaon cloud model calculations @xcite , although still within the same order of magnitude .
the strange dirac radius is somewhat smaller still , @xmath156 @xmath152 for the two cut - off masses .
the strange magnetic moment is also found to be negative as in earlier studies @xcite although somewhat smaller , @xmath157 . to obtain the same numerical values for
the form factors and radii as in refs.@xcite would require values for the vertex function cut - off of @xmath158 gev , as seen in fig.4 , where the @xmath115 dependence of @xmath138 , @xmath159 and @xmath112 is plotted ( note that @xmath112 is scaled down by a factor 10 ) .
for such hard vertex functions , however , one would find it difficult to be consistent with the data on the @xmath2@xmath3 asymmetry , as we shall discuss in the next section .
furthermore , the average number of kaons in the nucleon for such a cut - off mass would be @xmath160 , which is clearly beyond the limits of a perturbative treatment of the meson cloud .
the matrix elements of the strange axial vector current @xmath161 are parametrized in terms of the axial form factors @xmath162 : and @xmath163 : @xmath164 the strange pseudoscalar form factor @xmath165 is not observable in semi - leptonic neutral current reactions @xcite , and is therefore not considered here .
because the kaon has spin 0 , the strange axial vector form factor @xmath166 receives contributions only from the @xmath59 coupling in fig.1(a ) . in this case one has : @xmath167 where the light - cone axial distribution function @xmath168 is given by : @xmath169 the @xmath0 dependence of the axial @xmath166 form factor is shown in fig.5 for three values of the @xmath53 vertex function cut - off , @xmath140 and @xmath119 gev . at zero transferred momentum the ratio @xmath170 where @xmath171 from nucleon @xmath172-decays , measures the strange isovector axial charge of the proton . for @xmath53 vertex function factor cut - offs between @xmath118 and @xmath119 gev , @xmath173
is found to be very small , ranging between @xmath174 and 0.002 .
note that the sign of @xmath173 ( and @xmath166 ) is _ positive _ , becoming negative only for @xmath175 gev , as seen in fig.5 . as we discuss in more detail in sec.iv b
, this fact illustrates that kaon loops alone can not explain the apparent large strange component of the proton spin .
the meson cloud model of the nucleon has been successful in providing understanding of the origin of some of the symmetry breaking amongst the proton s sea quark distributions observed in recent experiments .
a pion cloud , for example , naturally allows one to account for the excess of @xmath176 quarks over @xmath177 in the proton @xcite . in a similar vein ,
the kaon cloud of the nucleon gives rise to the observed su(3 ) flavor symmetry breaking in the proton sea @xcite .
furthermore , it also leads to different strange and antistrange quark distributions in the nucleon , as first pointed out by signal and thomas @xcite .
this question has recently come to prominence again with the availability of new neutrino and antineutrino deep - inelastic scattering data , which were analyzed for a possible non - zero @xmath2@xmath3 difference @xcite .
such an asymmetry arises naturally in a kaon cloud picture of the nucleon , since the @xmath2 and @xmath3 quarks have quite different origins in this model .
indeed , it has been argued @xcite that because the @xmath2 quark comes from the @xmath21 , its distribution should be valence - like , while the @xmath3 , originating in the lighter kaon , should be much softer and resemble a typical sea distribution .
on the other hand , the experimental @xmath178 ratio was found to be consistent , within large errors , with unity , @xmath179 , prompting suggestions @xcite that the meson cloud model is ruled out by these data . in this section
we discuss how the above argument is modified when one takes into account the different @xmath65 and @xmath21 distribution functions in the nucleon .
the difference @xmath2@xmath3 turns out to be very sensitive to the details of the hadronic vertex functions used in calculating these distributions .
when calculated consistently in terms of the light - cone vertex functions discussed in the previous sections , the kaon cloud model predicts a very small excess of @xmath3 over @xmath2 at large @xmath1 , which is not in contradiction with the neutrino deep - inelastic data @xcite . within the same impulse approximation in which the form factors in sec.iii are calculated ,
the deep - inelastic quark distribution ( at a scale @xmath180 ) in the meson cloud model can be written as a one - dimensional convolution of the meson or hyperon light - cone distribution function and the intrinsic quark distribution in the meson or hyperon .
the @xmath2 and @xmath3 quark distributions , generated from figs.1(a ) and ( b ) , respectively , can be written @xcite : [ conv ] @xmath181 where the @xmath28 splitting functions are related to the @xmath21 and @xmath65 distribution functions in eqs.([f1l ] ) and ( [ f1k ] ) by : [ fdis ] @xmath182 the net strangeness in the nucleon being zero implies that the distributions must be normalized such that : @xmath183 as can be explicitly verified from eqs.([conv ] ) .
as for the intrinsic form factors of the virtual @xmath65 and @xmath21 , the intrinsic @xmath65 and @xmath21 structure functions are essentially unknown .
however , the advantage of the light - cone approach is that the intermediate state particles are on - mass - shell , thus allowing the on - mass - shell structure functions of the kaon and @xmath21 to be used @xcite . in a covariant instant - form formulation where the @xmath65 and @xmath21 are off their mass shells
, one needs to make additional assumptions about the extrapolation of the structure functions into the off - shell region , and indeed about the definition of the structure function of an off - shell @xmath65 or @xmath21 itself @xcite . according to the usual prescription adopted in the literature , in the instant - form calculations
it is usually simply assumed that the off - mass - shell structure function is the same as that on - shell . while possibly a reasonable approximation for heavy baryons , the justification for such an ansatz is certainly not clear for the kaon , which is typically much further off - mass - shell .
the @xmath3 distribution in kaons is obtained from measurements of final states in inclusive @xmath184 target @xmath185 reactions , where @xmath186 in drell - yan production @xcite , or @xmath187 in inclusive meson production @xcite .
one finds that the ratio of the @xmath65 structure function to the much better determined @xmath188 structure function @xcite is consistent with unity over most of the range of @xmath1 , dropping slightly at large @xmath1 , @xmath189 @xcite . for the @xmath2 quark distributions in @xmath21 and @xmath22
one can expect the su(3 ) symmetric relations @xmath190 and @xmath191 to be reasonable approximations . with these distribution functions , and the splitting functions in eqs.([f12l ] ) , ( [ f12k ] ) and ( [ fdis ] ) ,
the resulting @xmath2@xmath3 asymmetry is plotted in fig.6 at @xmath192 gev@xmath193 , for different values of the @xmath53 vertex function cut - off , @xmath194 and @xmath119 gev .
( note that it would not be meaningful to compare the calculated @xmath2 and @xmath3 distributions in eq.([conv ] ) separately with the structure function data , since the distributions calculated in the kaon cloud model do not contain any contributions from the perturbative process @xmath195 . )
the asymmetry in the kaon cloud model turns out to be very small , and for not too large values of @xmath115 , broadly consistent with the ccfr experiment within the given errors @xcite .
to obtain the difference @xmath2@xmath3 we have used the absolute values of @xmath196 from the parametrizations of refs.@xcite .
we should point out , however , that there exists some controversy regarding the overall normalization of the deep - inelastic neutrino data from which the strange quark distribution was extracted , resulting from an apparent inconsistency between the neutrino data and data on inclusive charm production @xcite .
in addition , the ccfr data were collected with fe nuclei targets , so that one needs to consider possible nuclear emc corrections in the data analysis @xcite before making any definitive conclusions about the @xmath2 and @xmath3 distributions . in view of these uncertainties in the data themselves
, one can certainly not conclude that the meson cloud model is inconsistent with the data from deep - inelastic scattering experiments .
the sign of the difference @xmath2@xmath3 is very sensitive to the vertex function used at the @xmath53 vertex . with a vertex function that depends only on the virtuality of the off - mass - shell kaon , such as in eq.([fft ] ) ,
the covariant perturbation theory calculation @xcite leads to @xmath197 at large @xmath1 .
the origin of this difference can be traced back to the fact that the light - cone @xmath65 distribution function peaks at larger values of @xmath27 compared with the instant - form distribution with the vertex function ( [ fft ] ) , which is somewhat less symmetric about @xmath198 , fig.7 . upon convoluting the more symmetric light - cone distribution with the @xmath199 and @xmath200 distributions , the @xmath201 behavior of @xmath199 and the harder @xmath202 behavior of @xmath200
are translated into a harder overall @xmath3 distribution compared with the @xmath2 . on the other hand , since at small @xmath27 the instant - form distribution @xmath203 , the original @xmath1 dependence in @xmath199 and @xmath200 is skewed to such an extent that the resulting @xmath2 distribution actually becomes dominant at large @xmath1 . as discussed in sec.ii , however , the instant - form function @xmath204 , evaluated with the @xmath51-dependent vertex function ( [ fft ] ) , does not satisfy the probability conservation relation in eq.([sym ] ) .
hence we believe that the results for the @xmath2@xmath3 difference in fig.6 are more realistic for the light - cone distribution functions . given the simplicity with which one can describe symmetry breaking in the nucleon s sea quark distributions with the meson cloud model , it is tempting to attribute the large violation of the ellis - jaffe sum rule @xcite to a kaon cloud of the nucleon . by carrying away some orbital angular momentum of the nucleon , a kaon cloud could naively be expected to reduce the intrinsic spin carried by quarks , and give rise to a negatively polarized strange quark distribution , @xmath205 .
the effect of the meson cloud on the polarized nucleon quark distributions has been addressed by several authors @xcite .
although the kaon cloud picture can provide a simple framework within which the non - perturbative sea could be understood qualitatively , it is clearly important to determined whether such a picture can provide , or even be consistent with , a more quantitative description . in view of the sensitivity of the sign and magnitude of the strange form factors and the unpolarized @xmath2@xmath3 difference to the hadronic vertex functions , we shall use the light - cone framework to re - examine the question of what role kaons play in the proton spin problem .
the contribution to the polarized strange quark distribution in the nucleon from kaon loops can be written : [ dconv ] @xmath206 where @xmath207 is zero for the same reason that the @xmath166 form factor receives contributions only from the @xmath59 interaction diagram in fig.1(a ) .
the helicity - dependent @xmath208 splitting function is given by : @xmath209 with @xmath210 defined in eq.([dfl ] ) .
if one had available information about the spin - dependent @xmath1-distribution of the strange quark in the bare @xmath21 , one could predict the resulting @xmath1-dependence of @xmath205 arising from kaon loops .
unfortunately , there is no information about polarized quark distributions in any hadron other than the nucleon .
nevertheless , one can still get an estimate of the size of the kaon loop contribution by considering the first moment of @xmath205 . in the su(6 ) symmetric model the spin of the @xmath21 is carried entirely by the @xmath2 quark .
it may seem reasonable therefore to expect that @xmath211 at some low scale @xmath180 , even taking into account su(6 ) symmetry - breaking effects , or some of the spin residing on gluons or in the form of angular momentum .
evolution to larger values of @xmath180 would imply that @xmath212 @xcite . in this case
the total strange quark contribution to the proton spin would be : @xmath213 for cut - offs @xmath1181.3 gev this would give @xmath214 .
this is to be compared with values @xmath215 quoted in connection with determinations of @xmath4 from the spin - dependent structure functions in ref.@xcite .
these results therefore suggest that a kaon cloud alone can not be expected to reproduce the observed deviation from the ellis - jaffe sum rule .
other mechanisms for generating a non - zero @xmath4 , such as ones based on the gluon axial u(1 ) anomaly , need to be invoked to account for the data .
we have presented a framework for studying strange form factors and quark distributions of the nucleon consistently in terms of the meson cloud model .
working on the light - cone , one avoids many of the problems and ambiguities associated with the choice of momentum dependent hadronic vertex functions encountered in instant - form calculations . within the uncertainty range of the input parameters , the strangeness ( sachs ) radius
is found to be very small and negative , in the vicinity @xmath216 @xmath152 for @xmath53 vertex function cut - offs of @xmath153@xmath119 gev .
the strange dirac radius is somewhat smaller , @xmath217 @xmath152 .
the strange magnetic moment is found to be @xmath157 , which is two to three times smaller than in previous estimates .
the strange axial vector charge is also small , @xmath218 , but in addition comes with the opposite sign compared with instant - form calculations with the meson cloud model .
combined , these results suggest that the strangeness content of the nucleon at low @xmath0 is indeed very small , and may represent a challenge to experimentalists seeking to verify its presence . using the same light - cone framework , we have estimated the asymmetry between the @xmath2 and @xmath3 quark distributions in the nucleon , which has been the subject of recent experimental investigation @xcite .
we find the magnitude of the @xmath2@xmath3 difference to be very small , with the @xmath3 distribution slightly harder than the @xmath2 , but somewhat sensitive to the shape of the hadronic @xmath53 vertex function . within the current experimental errors
this is consistent with the recent experimental determination of the asymmetry from neutrino deep - inelastic scattering , if moderately soft @xmath53 vertex functions are used .
for very hard vertex functions , the predicted asymmetry seems to be somewhat large in comparison with the data of ref.@xcite .
however , to be more definitive , more statistics on the charm production data are needed , and the apparent discrepancy between the inclusive deep - inelastic muon and neutrino data and those on @xmath219 production must be resolved @xcite . finally , the contribution of the kaon cloud to the polarized @xmath4 quark distribution is very small and _ positive _ , @xmath220 , and therefore is unlikely to feature prominently in any final explanation of the proton spin puzzle . aside from the experimental uncertainties , the small values for both
the slope of the strangeness form factor and the @xmath2@xmath3 difference within the simple model considered here gives some support to the suggestion of ref.@xcite that the coordinate and momentum space strange quark distributions are correlated .
this relationship will be clarified when new , more precise data from deep - inelastic neutrino scattering become available .
measurement of the @xmath0 dependence of strange matrix elements at low and intermediate values of @xmath0 at mit - bates , jefferson lab and elsewhere will also be very valuable @xcite .
the dynamical source of intrinsic strangeness can also be explored in semi - inclusive experiments involving coincidence measurement of scattered leptons and specific hadronic final states @xcite .
measurement of the spin transfer from target protons to recoiling @xmath21 hyperons in the target fragmentation region could discriminate between kaon cloud models , in which the spins are highly correlated , and parton fragmentation models , for which the correlations are very weak @xcite .
similar effects have also been discussed in hadronic reactions , @xmath221 @xcite . while the upcoming experiments will be difficult , if successful they should provide quite important information on the role of strangeness in the nucleon , as well as its dynamical origin .
we would like to thank m.k .
banerjee for useful remarks and suggestions .
helpful discussions with m. burkardt , t.d .
cohen , h. ito , x. ji , b.q .
ma , m. nielsen , j. speth , f.m .
steffens , a.w .
thomas and s. wang are gratefully acknowledged .
mm would like to thank the tqhn group at the university of maryland for their hospitality during his extended visit , and the brazilian agency capes for the financial support which made this visit possible .
this work was supported by the department of energy grant de - fg02 - 93er-40762 .
g. charlton _
et al_. , phys .
* 30 * , 574 ( 1973 ) ; v. blobel _ et al_. , nucl . phys . * b135 * , 379 ( 1978 ) ; h. kichimi _ et al_. , phys .
b * 72 * b , 411 ( 1978 ) ; t. kasahara , prog .
phys . * 51 * , 1836 ( 1974 ) ; t. aziz _
et al_. , z. phys .
c * 29 * , 339 ( 1985 ) ; m. asai _
et al_. , z. phys .
c * 27 * , 11 ( 1985 ) . | strange matrix elements of the nucleon are calculated within the light - cone formulation of the meson cloud model .
the @xmath0 dependence of the strange vector and axial vector form factors is computed , and the strangeness radius and magnetic moment extracted , both of which are found to be very small and slightly negative . within the same framework one finds a small but non - zero excess of the antistrange distribution over the strange at large @xmath1 .
kaon loops are unlikely , however , to be the source of a large polarized strange quark distribution . |
topological and geometric effects are fundamental quantum - mechanical phenomena .
they appear in various physical contexts , as the aharonov - bohm and aharonov - casher effects @xcite , berry s phase @xcite and other models .
such effects are manifested by accumulated topological or geometrical path - dependent phases , which are observed in interference experiments .
such phases have been experimentally detected several times over the years , and are recently one of the interests of quantum simulations @xcite ; for example , several proposals and experiments probing the effects of an external vector - potential have been suggested with ultracold atoms in optical lattices @xcite . in quantum field theory and particle physics ,
a topological phase similar to the aharonov - bohm effect appears in the context of the wilson - loop operator @xcite , which is an order parameter manifesting the disorder in the confining phase of a gauge theory .
the wilson - loop operator along a curved spacetime path @xmath0 is @xmath1 where @xmath2 stands for path ordering ( see , for example , @xcite ) .
wilson - loops manifest confinement through the area dependence of their expectation value , and thus are an important test for confinement .
they are extremely useful in euclidean spacetime , for numerical ( monte - carlo ) simulations . in minkowski spacetime , besides the fact they involve a product of operators along a loop and thus are non - local , they can be interpreted as transition amplitudes , and thus their phases do not contribute to the related probabilities . to gain information from these phases ,
one has to use interference effects as we propose here . in this paper
, we present a method to observe the area - law manifested in confining theories using superposition and interference of mesons , which unlike in the wilson - loop approach , contains the relevant phase as a relative one .
we draw the connection between our method and the well - known wilson - loop .
note that previous works have already discussed the properties of the inter - quark potential using multiple wilson - loops @xcite .
measurements of wilson - loop operators were discussed in @xcite .
the paper is organized as follows : first , in section ii , we consider the interference effect of two mesons , consisting of static quarks , in a superposition , and show how to gain the string tension from it .
the relation to the wilson - loop operator is drawn . in section iii , we allow one of the quarks to be dynamic , modeling it as a particle in a moving harmonic potential , and show how to obtain the string tension in that case , using an exact solution of schrodinger equation .
we discuss the relation of our approach to ordinary wilson - loops in section iv .
finally , in section v , we discuss the possibility of realizing the proposed idea using a quantum simulator @xcite .
free quarks can not be found in nature , but rather form hadrons , due to the phenomenon of quark confinement @xcite . a quark and an anti - quark , attached to each other by a confining flux - tube , form a meson , which is the simplest hadron in qcd .
the static potential between the quarks , as a function of their distance @xmath3 , takes the form should also include a @xmath4 term .
however , we neglect it , assuming that @xmath3 is large enough .
this is sufficient for the interference experiment proposed hereby as it is only affected by an area difference . ]
@xmath5 for large values of @xmath3 , where @xmath6 is called the string tension , and @xmath7 is the coupling constant @xcite . , @xmath8 form a meson of length @xmath9 at time @xmath10 .
while @xmath11 remains static , @xmath8 is externally moved ( with no dynamics of its own ) in a superposition of two possible trajectories , corresponding to its internal states @xmath12 and @xmath13 , creating a superposition of mesons . the path of the @xmath14 fermion is drawn in red , and the path of the @xmath15 one - in blue . at time @xmath16 , both the trajectories arrive again at @xmath9 , and the state is then mixed using a unitary transformation ( equation ( [ unit ] ) ) , causing an interference effect . in case of confinement , the interference phase depends on the area @xmath17 ( defined in equation ( [ adefinition ] ) ) in case of confinement - as does the wilson - loop .
the phase is accessible from the probabilities to find @xmath8 in either of its internal states ( equation ( [ staticprobs ] ) ) . ]
consider two _ static _ quarks , initially separated by distance @xmath9 . by static
we mean that we treat them as external sources of electric field , which do not have their own dynamics , but can be externally moved .
taking confinement into account , we know that the two quarks are connected by a long flux - tube , with length @xmath3 , forming a `` meson '' . if we allow one quark to move along some trajectory @xmath18 , we get the hamiltonian @xmath19 where we generally assume that @xmath20 is an unknown quantity . next ,
assume that the moving quark has some two internal energy levels @xmath21 , which we denote by @xmath22 - eigenstates of the pauli matrix @xmath23 .
we define the projection operators to the subspaces of internal levels as @xmath24 , and introduce an internal level dependent paths @xmath25 , i.e. , @xmath26 the energy levels are used to generate level - dependent spatial positions .
the choice of these levels depends on the separation method .
for example , if one considers an abelian theory , where @xmath21 corresponds to the spin ( @xmath27 ) , the separation could be achieved using an external magnetic field , @xmath28 , then @xmath29 .
a concrete example of generating such a separation is found in @xcite .
we wish to consider the interference effect of two mesons in superposition , varying their length .
we shall consider a superposition of a meson with a @xmath14 fermion and a meson with a @xmath15 fermion , initially with the same length ( i.e. , @xmath30 for @xmath31 ) , which are stretched by moving the right fermion in two opposite directions , and then , at time @xmath32 , brought back to the same length ( i.e. , @xmath30 also for @xmath33 .
the opposite directions impose more conditions on the paths : @xmath34 , @xmath35 , and since the @xmath15 fermion should not `` go through '' the static fermion at @xmath36 , we also demand @xmath37 .
the initial state is , of course , @xmath38 solving schrodinger equation , one gets that @xmath39 and so in order to measure @xmath20 , one has merely to cause an interference between the two states .
we perform ramsey interference , by applying the rotation @xmath40 on the state at @xmath16 , and one obtains , up to a global phase , @xmath41 where @xmath42 is the area enclosed between the two trajectories ( as in figure [ fig1 ] ) . thus , the probabilities to find the system on each of the internal levels are @xmath43 by performing such an interference experiment , and measuring the phase , knowing the area difference one can calculate the string tension @xmath20 . moreover , if the phase does not exhibit such an area law , it means that the system is not within a confining phase , and hence this measurement can be used for probing the confining phase as well .
next we wish to introduce a simple model of dynamical charges , in order to examine the corrections to the latter static case .
in particular , we wish to examine decoherence and destructive interference due to excitations of the mesons . for the sake of simplicity
, we assume that one of the quarks is static ( the one placed in @xmath36 ) , and that the other one , having a mass @xmath44 , is trapped in a harmonic potential , centered around @xmath45 .
this harmonic trap is merely an external trapping potential .
the hamiltonian of the system takes the form @xmath46 assume that the dynamic quark has two internal energy levels , denoted and treated as before .
we introduce an interaction between the internal and external degrees of freedom , of the form @xmath47 where @xmath48 are opening functions which are zero for times @xmath49 , and are assumed to be smooth enough , i.e. at least their first and second time derivatives vanish for @xmath50 .
define @xmath51 ( the same conditions on @xmath25 apply as in the static case , of course , and that poses conditions on @xmath52 ) , @xmath53 .
then one gets , after completing the square , that the total hamiltonian is ( neglecting constants ) @xmath54 where the first part is just a harmonic oscillator , with the center of its potential following classically the trajectory @xmath55 : @xmath56 and @xmath57 .
the next step is the solution of schrodinger equation for our hamiltonian : @xmath58 noting that the internal levels are not changed by the hamiltonian , we can first solve for a given internal level @xmath21 and consider only the harmonic part .
thus if we set for a given @xmath21 , @xmath59 we get that @xmath60 is the solution of a schrodinger equation for a classically moving harmonic potential : @xmath61 using the solution of this equation in @xmath62-space , we get that the solutions ( not eigenstates ) are @xmath63 where @xmath64 are the energies and eigenstates of a `` regular '' , fixed - potential harmonic oscillator , @xmath65 and @xmath66 .
our initial condition is @xmath67 . for @xmath31 ,
the oscillator is supposed to be in its non moving ground state , and thus we expect that @xmath68 for these times .
thus @xmath69 . using the continuity of the equation of motion of @xmath70
, we get that @xmath71 as well . from the smoothness of the opening functions we know that @xmath72 .
thus we conclude that in order to start from a local ground state , the solution must be @xmath73 next , we wish to interpret this solution in terms of local instantaneous eigenvalues .
that is , the states @xmath74 , defined as eigenvalues of @xmath75 in the `` frozen '' time @xmath76 .
we already know that the system starts at @xmath10 with the eigenstate @xmath77 , but what s later ? in order to do that , we define @xmath78 .
writing @xmath75 in terms of @xmath79 at a fixed @xmath76 yields this diagonalization : it is merely a transformation to a frame which moves with the potential , which is its instantaneous rest frame ( irf ) .
consider the position - dependent part of @xmath80 , in terms of the irf .
call it @xmath81 : @xmath82 in the irf basis , this is a coherent state - poissonian distribution of @xmath74 states : @xmath83 where @xmath84 suppose we start , at @xmath10 , with an initial state @xmath85 then , defining @xmath86 and using the solutions from the previous section , we get @xmath87 let us understand the meaning of this state .
we start at @xmath10 with a superposition of two states with two different values of @xmath21 .
both of them are in the ground state of an oscillator , centered around the same position ( @xmath88 ) .
then we `` move '' the wavefunctions together with the potential : the opening functions are translated to the trajectories @xmath89 .
each element of the superposition `` goes '' through another path , since the moving of the potential depends on the internal level .
a superposition of coherent states in the terms of the local irfs is created , and eventually , at @xmath16 , both the interaction functions are closed , i.e. the two potentials experienced by internal levels overlap again , and @xmath90 . next , trace out the oscillator degrees of freedom at @xmath16 , to obtain an internal - level density matrix .
the density matrix at @xmath16 is @xmath91 where @xmath92 , @xmath93 and @xmath94 .
one can see that the phases depend on two parts : one , @xmath95 , is totally calculable .
the other is @xmath20 dependent and ca nt be calculated unless @xmath20 is known .
the @xmath20 dependent phases are only global ( the probabilities - the diagonal terms in @xmath96 do not depend on them ) . in order to observe the phases , we wish to cause an interference , i.e. to rotate the state
: we act on the system at @xmath16 with the rotation operator @xmath40 .
one can use the diagonal terms in the new density matrix @xmath97 to determine the probabilities to measure each of the internal levels .
the probabilities are @xmath98 where @xmath99 we see that now the probabilities depend on the phase difference , and that it became a relative phase indeed .
however , how do we use it in order to measure the string tension ?
first , one must note that @xmath100 is governs the visibility of the interference : if @xmath101 , one gets equal probabilities to both measurement outcomes , and hence no information can be gained and the interference is lost .
we shall consider the effect of @xmath100 in detail .
however , let us first focus on the role of the phase difference , assuming it is not screened by @xmath100 .
the phase difference consists of two parts . the first one , @xmath102 , is @xmath20 independent , and once the phase is obtained from the probabilities , it can be subtracted .
hence we are left with @xmath103 where @xmath17 is the area enclosed between the two paths in spacetime !
exactly the area dependence which is expected in any abelian and non - abelian gauge theory within the confining phase .. thus , the interference effect measures the string tension @xmath20 in case of confinement ; otherwise , the area law will not be manifested and thus being outside the confining phase can be probed this way as well .
let us discuss the effect of @xmath100 .
in order to understand it , we calculate it explicitly ; using the definitions of @xmath104 , we get that @xmath105 , where @xmath106 the interference is maximal when @xmath107 ; that corresponds to @xmath108 , which means that the final state is an eigenstate rather than a poissonian superposition . in that case ,
. then we get @xmath110 if we wish to consider the cases in which there is some disturbance to the interference , but it is negligible , we should consider the limit @xmath111 .
this is obtained when the conditions @xmath112 are met .
this can be understood in terms of uncertainty principle : the final coherent state is displaced in phase space .
the displacement in @xmath62 is @xmath113 , and the first condition is met if we require it to be much smaller than the ground state s @xmath114 ; the displacement in @xmath115 is @xmath116 , and the second condition is met if we require it to be much smaller than the ground state s @xmath117 . let us see what is the limitation on the trajectories @xmath118 , if one wishes to get a good interference according to this criterion . in order to do that ,
let us write the explicit solution for @xmath119 .
we wish to solve the differential equation @xmath66 .
previously , we have obtained the initial conditions @xmath120 , and thus the homogenous solution is zero , and we are left only with the particular solution , which can be found using green s function : @xmath121 demanding @xmath122 , we get a condition on the maximal acceleration @xmath123 : @xmath124 and the very same condition is obtained from demanding @xmath125 .
thus we conclude that the interference is not ruined if the charges accelerations are small enough all along the paths .
next , let us examine the relation of the proposed method with the wilson - loop operator approach .
the area @xmath17 , enclosed between the two paths of the two quarks in superposition , is the same area on which the wilson - loop of a single quark , moved in spacetime along the union of the paths , would depend , and with the same string tension .
thus , the interference phase and this wilson - loop s phase are the same .
we shall now describe how to derive a quantitative relation to the wilson - loop operator in the case of static quarks and strong coupling limit .
denote the state of heavy ( _ static _ ) @xmath126 , separated by distance @xmath3 by @xmath127 , and write it in terms of the gauge field degrees of freedom .
assuming confinement , this corresponds to a meson state , where the two quarks are connected by a flux tube , and thus @xmath128 where @xmath129 is the hamiltonian of the gauge field ; in particular , in the strong coupling limit one gets @xmath130 where @xmath2 stands for path ordering .
we wish to calculate , within the strong coupling limit , the expectation value ( in minkowski space ) of the wilson - loop operator , corresponding to the loop depicted in figure ( [ fig2]a ) .
this can be decomposed to four different parts : @xmath131 , @xmath132 , @xmath133 and @xmath134 .
the @xmath134 part contribution is zero if we work in the temporal gauge .
let us now see the contribution of the other three parts of the loop . in order to do that ,
let us discretize the function @xmath135 as in figure ( [ fig2]b ) .
define @xmath136 acting on the vacuum , this operator creates ( in the strong limit ) a flux tube from @xmath137 to @xmath138 ; i.e. , the static charge in @xmath137 is raised and the static charge in @xmath138 is lowered . using these terms ,
the wilson - loop operator we wish to calculate becomes @xmath139 since @xmath140 , this expression simplifies to @xmath141 finally , let us calculate the expectation of the wilson - loop in the vacuum state .
taking the vacuum energy as zero , the left @xmath142 contributes 1 .
the right @xmath143 , acting on the vacuum , creates a flux tube between @xmath36 and @xmath144 , and the @xmath145 on the left of it contributes a phase of @xmath146 since it is an eigenstate .
then , each of the @xmath147 s in the product shortens or stretches the flux tube , and the @xmath145 to the left of it adds up a phase of @xmath148 .
this is a process of creating a flux tube with length @xmath9 at @xmath10 , changing its length according to @xmath135 , until @xmath16 where its original length is regained and it is destroyed .
the amplitude for this process is thus , according to the given explanation , the expectation value of the wilson - loop , and it is @xmath149 taking back the continuum limit , we take @xmath150 , or @xmath151 . this transforms the sum to an integral , and this integral is equal to @xmath17 , the area enclosed by the curve - the area law of confinement : @xmath152 thus we can conclude , that in the strong coupling limit , taking the solution from equation ( [ statschro ] ) , @xmath153 ( where @xmath154 denotes the curve enclosed by the motion of the @xmath21th element of the superposition ) - so we see that indeed , in this limit , the transition amplitude is the corresponding wilson - loop , and it is a global phase , so in order to obtain knowledge about it one has to transform it to a relative one , using the above interference prescription .
quantum simulations @xcite are a rapidly growing field , based on the idea that quantum systems can simulate each other . thus , quantum systems which are inaccessible for measurement , can be simulated using other quantum systems , which are controllable , accessible and measurable in the laboratory , such as cold atoms in optical lattices @xcite , trapped ions @xcite and other systems .
these systems serve as an `` analog quantum simulators '' .
recently , several methods have been proposed for quantum simulation of high energy physics , for example , simulations of dynamic scalar @xcite ( vaccum entanglement ) and fermionic fields @xcite ( thirring and gross - neveu models ) , and fermions in lattice qft @xcite . simulations for dynamic gauge fields have been proposed as well .
simulations of pure - gauge @xmath155 theories ( simulating the abelian kogut - susskind @xcite hamiltonian or a truncated version of it ) , using becs @xcite or single atoms @xcite in optical lattices have been proposed , as well as simulations of other pure - gauge @xmath155 theories with ultracold atoms @xcite . simulations of @xmath155 theories with dynamic matter have been proposed as well @xcite . recently , several proposals for quantum simulations of non - abelian theories have been suggested as well @xcite . besides these lattice works , a proposal for continuous qed simulation @xcite has been suggested as well .
a recent detailed description of a simplified simulation approach is found in @xcite . in quantum simulations of lattice gauge theories using cold atoms in optical lattices , such as @xcite
, one could realize the experiment proposed in this paper .
a first proposal for area law probing in the suggested method has been proposed for an abelian @xmath156 gauge theory , in which lasers are used to create a superposition of fermions , which results in a superposition of mesons , and to perform the ramsey interference required for the measurement .
a detailed discussion of the proposal can be found in @xcite . as the above proposal is general ,
we believe it may be useful in future simulations of gauge theories for probing confinement and its area law .
we have presented a method to measure the string tension of a confining flux - tube using superposition and interference of mesons .
interestingly , the two states in superposition experience _ different _ electric fields .
this is a reminiscent of the concept of `` private potential '' @xcite . by exploiting an area - dependent phase due to the linearity of the static quark potential in the confining phase of gauge theories , we have observed confinement using an interference of the mesons
this allows to convert the global phase ( which appears , for example , in wilson - loops which are related to transition amplitudes ) , to a relative phase , observable in probabilities . in the coulomb phase ( as in 3 + 1 qed ) or in any other @xmath157 phase , with @xmath158 ,
the appropriate gauge field state does not include a flux tube as in the confining phase .
hence , the final probabilities will not manifest a simple area dependence .
this can be used to probe a transition between confining and non - confining phases .
although it is only a gedanken experiment in the context of particle physics , such an experiment , in its lattice version , may be realized and used to observe confinement effects and phase transitions within a quantum simulation of confining gauge theories @xcite .
the authors would like to thank r. ber , m. karliner , s. nussinov and b. svetitsky for helpful discussions .
br acknowledges the support of the israel science foundation , the german - israeli foundation , and the european commission ( picc ) .
ez acknowledges the support of the adams fellowship , of the israel academy of sciences and humanities .
10 y. aharonov and d. bohm .
significance of electromagnetic potentials in the quantum theory .
, 115:485491 , aug 1959 .
l. mazza , a. bermudez , n. goldman , m. rizzi , m. angel .
martin - delgado , and m. lewenstein .
an optical - lattice - based quantum simulator for relativistic field theories and topological insulators .
, 14(1):015007 , 2012 .
d. banerjee , m. dalmonte , m. mller , e. rico , p. stebler , u - j .
wiese , and p. zoller .
atomic quantum simulation of dynamical gauge fields coupled to fermionic matter : from string breaking to evolution after a quench .
, 109:175302 , oct 2012 . | we introduce a new topological effect involving interference of two meson loops , manifesting a path - independent topological area dependence .
the effect also draws a connection between quark confinement , wilson - loops and topological interference effects . although this is only a gedanken experiment in the context of particle physics , such an experiment may be realized and used as a tool to test confinement effects and phase transitions in quantum simulation of dynamic gauge theories . |
in typical s - wave superconductors , nonmagnetic impurities are not pair - breaking , and therefore have little effect on either the transition temperature or the superfluid density@xcite .
by contrast , the substitution of a nonmagnetic impurity into a d - wave superconductor is expected to be pair - breaking , and , therefore , should have a significant effect on both . since the cuo@xmath0-based high - temperature superconductors are believed to possess an order parameter with @xmath19 symmetry@xcite
, such an effect should be observable in these materials .
indeed , the substitution of a nonmagnetic impurity such as zn ( or even an insulating impurity such as he , introduced by ion implantation ) in a high - t@xmath20 material such as yba@xmath0cu@xmath1o@xmath2 is now well known to dramatically suppress the zero - temperature superfluid density@xcite and also to have a significant , though lesser , effect on the critical temperature .
this striking behavior has provoked numerous theoretical studies of zn impurities and other nonmagnetic disorder in the cuprate superconductors . for example ,
annett _ et al_@xcite and hirschfeld and goldenfeld@xcite , have shown that the impurity scattering can convert the @xmath21-linear behavior of the low - temperature superfluid density to a @xmath22 behavior , in agreement with experiment .
other workers@xcite have refined the theory so that it now agrees quite well with experiment .
choi@xcite has used a t - matrix formalism to calculate the zn impurity dependence of the superfluid density in yba@xmath0cu@xmath1o@xmath2 for various temperatures .
et al_@xcite used a model of dirty d - wave superconductivity to fit the impurity - dependence of the superfluid density in a ndba - based cuprate superconductor .
hettler and hirschfeld@xcite have shown how inhomogeneities in a d - wave order parameter can lead to both strong suppression of superfluid density and enhanced microwave conductivity .
several authors have treated the impurity problem by solving the bogoliubov - de gennes equations@xcite for the pairing amplitude , which is position - dependent in a dirty d - wave or s - wave superconductor with a short coherence length .
et al_@xcite used this approach to show that disorder in a 2d s - wave superconductor reduces the effective superfluid density , makes the superfluid inhomogeneous , and even breaks up the 2d layer into superconducting islands .
more recently@xcite , the same authors generalized this approach to d - wave superconductors .
zhitomirsky and walker@xcite showed how spatial fluctuations in the energy gap can reduce t@xmath20 in a d - wave superconductor .
et al_@xcite solved the bogoliubov - de gennes equations with a momentum - dependent pairing interaction to obtain an extended gapless region in the overdoped cuprates , which they connected to a rapid doping - induced decrease of both @xmath23 and superfluid density .
et al_@xcite , using the bogoliubov - de gennes approach for a spatially varying energy gap , found that the superfluid density goes rapidly to zero with increasing zn concentration , while t@xmath20 decreases much more slowly with concentration , in reasonable agreement with experiment .
neto@xcite has treated the zn - doped cuprates as a collection of fluctuating stripes modeled as an array of coupled overdamped josephson junctions pinned by impurities , and has used this model to estimate the doping - dependence of the transition temperature .
finally , uemura@xcite has discussed the low - temperature superfluid density in zn - doped cuprates from the point of view of the `` universal correlation '' between @xmath23 and zero - temperature superfluid density in the underdoped cuprates .
as noted by emery and kivelson@xcite , such a connection would exist if @xmath23 were controlled by phase fluctuations rather than by the vanishing of the gap amplitude .
most of these models require a rather elaborate calculation to find the influence of non - magnetic impurities on the superfluid density .
by contrast , the goal of this paper is to analyze the behavior to be expected of a very simple model of the zn - doped cuprate superconductors which can be worked out by intuitively appealing calculations .
although this model is difficult to derive from first principles , it may still be of value , because it can be used to make simple predictions for specific experiments , and to understand trends in those experiments .
the model we study was first described by nachumi _
et al_@xcite , and is sometimes known as the `` swiss cheese '' model . in this model , the zn atoms are assumed to replace the cu atoms substitutionally within the cuo@xmath0 layer , which is believed to be the locus of superconductivity in the cuprate superconductors , and to disrupt the @xmath19 order parameter , producing a region approximately of radius @xmath5 within which the layer is normal .
( here @xmath5 is the zero - temperature in - plane coherence length of yba@xmath0cu@xmath1o@xmath2 . )
this disruption is plausible , because @xmath5 is very small in the cuprate superconductors ( probably on the order of 10 - 20 @xmath10 at low temperatures , as discussed further below ) .
indeed , some dramatic experimental evidence supporting both the swiss cheese model and the d - wave order parameter picture has recently been presented by pan _
. these workers showed , using scanning tunneling microscopy , that the gap was significantly suppressed near the zn atoms , and that the density of states of single - particle excitations near the zn atoms exhibited a fourfold symmetry . in the present paper ,
we extend the approach of nachumi _ et al _ by analyzing the swiss cheese model within a simple percolation picture .
our approach differs from that of nachumi _ et al _ by accounting for the connectivity of the superconducting part of the cuo@xmath0 layer .
by contrast , in their approach , the superfluid density is directly proportional to the superconducting area .
the consequences of our approach are easily worked out if we use percolation theory and interpret the cuo@xmath0 layer as an inhomogeneous superconductor with spatially distinct superconducting and normal region .
the resulting variation of superfluid density with zn concentration seems to agree somewhat better with experiment than does the simplest version of the swiss - cheese model .
the remainder of this paper is organized as follows . in the next section ,
we describe our percolation model in detail .
the numerical results following from this model are presented in section iii .
finally , in section iv , we discuss our numerical results , briefly compare them to experiment , and analyze the limitations and predictions of the model .
fig . 1 shows our picture of a single cuo@xmath0 layer in zn - doped yba@xmath0cu@xmath1o@xmath2 , in which a fraction @xmath3 of the cu atoms are replaced at random by zn . within the `` swiss cheese '' model@xcite
, each zn atom completely suppresses superconductivity in a region of area @xmath24 around it .
thus , as @xmath3 increases , a smaller and smaller areal fraction of the cuo@xmath0 plane is superconducting . eventually , for a large enough @xmath3 , the remaining superconducting region is disconnected and the effective superfluid density must vanish .
there are two parameters in our geometrical model : the fraction @xmath3 of in - plane cu atoms which are replaced by zn , and the quantity @xmath25 , which is the ratio of the zero - temperature in - plane coherence length @xmath5 ( evaluated at @xmath6 ) to the cu - cu distance @xmath26 . neither parameter is trivial to determine experimentally . for example , @xmath3 is not necessarily equal to @xmath27 , the _ overall _ fraction of cu atoms which are replaced by zn .
in fact , it is believed that @xmath28 , i. e. , that all the zn dopant atoms go into the cuo@xmath0 planes @xcite . obviously , the connection between @xmath3 and @xmath27 is important in making quantitative comparisons between theory and experiment .
also , the magnitude of @xmath14 is not universally agreed upon for any value of @xmath29 .
et al_@xcite , for example , quote @xmath30 at @xmath31 and @xmath8 , extrapolating from @xmath32 . since @xmath33 for this value of @xmath29@xcite , this gives @xmath34 .
we shall later discuss the effects of assumptions about @xmath14 on the predictions obtained from this model . .
to estimate the superfluid density as a function of @xmath3 , we assume that the region more than a distance @xmath5 from any zn atom has a complex conductivity @xmath37 characteristic of a superconductor .
we obtain this conductivity by using the first london equation , as applied to a cuo@xmath0 plane at @xmath6 : @xmath38 where @xmath39 is the supercurrent density and @xmath40 is the electric field . in gaussian units , @xmath41 where @xmath42 is the in - plane penetration depth ( evaluated at @xmath6 ) , and
@xmath43 is the mass and @xmath44 the charge of a cooper pair .
fourier transforming gives @xmath45 , where @xmath46 with @xmath47 we assume that the region _ within _ a distance @xmath5 of a zn atom is simply a normal metal with frequency - independent conductivity @xmath36 .
clearly , this simple model can only be appropriate for @xmath48 , where @xmath49 . at higher frequencies
, there will be effects arising from the superconducting energy gap , which will produce additional structure in @xmath37 .
also , this inductive form also implies no absorption at low frequencies in the limit @xmath50 . in reality , many of the cuprate superconductors do exhibit low - frequency absorption , even at low temperatures@xcite .
this extra absorption can readily be included within the present model by adding to @xmath35 a parallel normal conductivity .
we have carried out a few calculations including this extra term ; as discussed below , they do not affect the @xmath3-dependent superfluid density .
we may infer the effective superfluid density @xmath18 from an effective london equation @xmath51 which relates the space - averaged current density @xmath52 to the space - averaged electric field @xmath53 .
@xmath54 is related to @xmath18 through a concentration - dependent generalization of eq .
( 2 ) : @xmath55 in the frequency domain , eqs .
[ eq : london1 ] and [ eq : london2 ] may be written @xmath56 where @xmath15 is the effective complex conductivity .
therefore , @xmath57 and , in order to compute @xmath18 , we need to calculate @xmath58 at sufficiently low frequencies .
the ratio @xmath59 can also be calculated more directly by using a homogeneity property of @xmath60@xcite . in the present context , this property states that @xmath61 here @xmath62 is the effective complex conductivity of a two - component system made up of constituents with conductivities @xmath35 and @xmath36 and in - plane zn concentration @xmath3 _ in a particular geometry _ , and @xmath63 is an arbitrary constant . in effect , homogeneity means that ( for fixed geometry and fixed @xmath3 ) if the conductivities of each constituent are multiplied by a certain factor @xmath64 , the _ effective _ conductivity is multiplied by that same factor @xmath64 .
in particular , if @xmath65 , we get @xmath66 hence , for any @xmath3 , @xmath67 here we have used the fact that @xmath68 , the superfluid density at @xmath6 , is given by @xmath69 , and also the fact that lim@xmath70 .
we have evaluated @xmath18 [ or equivalently , @xmath58 ] by two complementary methods .
the first method is a finite - element technique , and should become very accurate for a sufficiently large numerical sample , given the geometrical assumptions of the model .
we consider an l @xmath72 l region of a cuo@xmath0 plane , assuming that the cu ions sit on a square lattice of lattice constant @xmath26 , and choosing @xmath73 where @xmath74 is an integer .
then , using a random number generator , we replace a fraction @xmath3 of the cu ions by zn .
we then assign conductances @xmath37 or @xmath36 to each bond connecting nearest neighbor cu sites , by the following rule .
if the entire length of the bond is at least a distance @xmath14 from any zn site , it is assigned a conductance @xmath37 ; otherwise , it is assigned @xmath36 .
next , we calculate @xmath58 using a standard technique known as the y-@xmath75 transformation@xcite , which is very efficient and accurate for two - dimensional impedance networks .
this technique is approximate only insofar as it is limited to finite networks .
we call this method the random conductance approach ( rca ) .
our second method for calculating @xmath58 is a simple analytical approximation , the effective medium approximation ( ema)@xcite . for the assumed geometry , this approximation is implemented as follows .
first , we calculate @xmath76 , the areal fraction occupied by the superconducting material , as a function of @xmath3 , using the relation @xmath77 where @xmath78 is the number of zn ions in a plane of area @xmath79 .
this limit is readily evaluated using the relation @xmath80 , with the result @xmath81 given @xmath76 , one can calculate @xmath60 using the standard ema relation for this two - dimensional system , namely , @xmath82 the physically correct solution to this quadratic equation is the one which satisfies im@xmath83 . of course , ideally , the ema is suited for application to a geometry in which the @xmath84 and @xmath79 components are distributed in a symmetrical fashion , whereas here , because of the assumption of excluded area , the @xmath84 and @xmath79 are not distributed symmetrically .
nonetheless , it is a reasonable approximation even for this nonsymmetric geometry .
once @xmath85 has been calculated at a given @xmath3 by either of the methods described above , the effective superfluid density can be inferred from the relation ( [ eq : super ] ) .
we can also compute the ratio @xmath59 directly , using eq .
( [ eq : ratio ] ) .
that is , for a given @xmath3 , we decide which are the superconducting bonds , using the method described earlier in this section , and assign these bonds conductances of unity .
the remaining normal bonds are assigned conductances of zero .
the effective conductance is then evaluated using either the ema or the rca , and the ratio @xmath59 is then obtained using eq .
( [ eq : ratio ] ) .
in carrying out the rca calculations , we have generally used a square networks of size ranging from @xmath86 to @xmath87 cu sites . in order to reduce statistical fluctuations ,
the rca results quoted below are averaged over a number of different realizations of the disorder for each @xmath88 and @xmath3 .
these realizations are `` correlated '' in the sense that , for each realization , we increase @xmath3 by adding a few impurities to the configuration of the previous @xmath3 value . in evaluating the ema conductivities ,
no such averaging is necessary , since the approximation can be evaluated analytically in the limit of a very large sample .
our results for @xmath59 are shown in fig . 2 for several choices of the parameter @xmath25 . the full , dashed , dot - dashed , and dotted curves are the ema results .
the diamonds are the results of the rca calculations at @xmath89 , using eq .
( 12 ) , for two correlated realizations of a @xmath87 lattice
. there are several striking features .
first of all , in both methods @xmath59 monotonically decreases with increasing @xmath3 , reaching zero at a percolation threshold @xmath13 which depends on @xmath25 .
moreover , @xmath13 increases with decreasing @xmath25 .
this relationship is not surprising : a larger @xmath25 implies that a larger area is converted from superconducting to normal by a single zn impurity , within this model . also , the ema and rca agree quite well over most of the concentration range , except near the percolation threshold .
the ema percolation threshold always corresponds to @xmath90@xcite .
the rca threshold is higher , because the superconducting fraction remains connected to a somewhat higher fraction of zn than is predicted by the ema . in figs . 3 and 4
, we show im@xmath91 and re@xmath91 as a function of x at @xmath92 , where @xmath49 and @xmath93 is defined in eq .
( [ eq : defa ] ) .
we use using parameters thought to be appropriate for the case @xmath8 , namely , @xmath89@xcite . in both figures ,
the dashed line represents the ema ; the diamonds denote the rca ( average over 20 correlated realizations of a @xmath86 lattice ) and the the open squares are the rca , averaged over 2 realizations of a @xmath87 lattice .
3 shows that im@xmath91 closely mirrors the behavior of @xmath59 shown in fig . 2 , as expected from the homogeneity relations ( 11 ) and ( 12 ) .
the apparent percolation threshold is again somewhat larger ( @xmath94 ) in the rca than in the ema ( @xmath95 ) . for values of @xmath3 well below @xmath13 ,
im@xmath96 , like @xmath59 , decreases linearly with @xmath3 .
this linear behavior is well known in the ema@xcite .
4 shows that re@xmath58 has a strong peak occurring near the percolation threshold ( @xmath95 in the ema and @xmath97 within the rca for our choice of @xmath98 ) .
the physical origin of this peak is discussed in the next section .
we believe that the greater half - width of the rca results arises because the calculation is carried out for a finite sample , and also because the average is taken over a number of different realizations , each of which has a slightly different percolation threshold for these finite samples . in support of this picture ,
note that the rca half - width is smaller for the larger ( @xmath87 ) samples , for which these fluctuations should be smaller .
the slight random fluctuations in these curves as a function of @xmath3 are , we believe , also due to these finite - size effects . by contrast , the ema results are obtained analytically for an effectively infinite sample ; so these finite - size fluctuations play no role . note that , even for a @xmath99 sample , there are still only 375 impurities at @xmath100 ; so these fluctuations are still significant in the rca . in fig . 5 , we show re@xmath91 for @xmath101 , once again using @xmath34 , as calculated in the ema and in the rca for two different sample sizes .
the ema peak is clearly stronger for @xmath101 than @xmath92 .
the rca peak is also stronger , especially for the larger sample where the peak height is not washed out by finite size effects .
the apparent double peak for the @xmath87 sample is , we believe , just the result of finite - size fluctuations , which are magnified at such low frequencies where the conductivity contrast @xmath102 is larger . finally , we have calculated @xmath91 within the ema for the same model as before , again with @xmath34 , but a slightly different choice for @xmath37 .
namely , we write @xmath103 where we arbitrarily choose @xmath104 .
the motivation behind this choice is that , in a material such as , which is believed to have an order parameter with @xmath19 symmetry , the in - plane conductivity is expected to have a quasiparticle contribution , even at very low temperatures .
the additional term @xmath105 is a crude way of modeling this contribution .
the resulting re@xmath58 is plotted versus @xmath3 at @xmath106 , both for the model ( [ eq : sigs1 ] ) and for a purely inductive @xmath107 .
clearly , the peak near @xmath13 is little affected by the change , but there is now , in addition , some extra contribution to re@xmath58 for @xmath108 .
although this extra contribution to @xmath35 does slightly change re@xmath58 at finite @xmath88 , the homogeneity relations ( 11 ) and ( 12 ) imply that it has no effect on @xmath18 , either in the ema or in the rca .
it is of interest to compare our calculated values of @xmath13 with the experimentally observed critical concentration for yba@xmath0(cu@xmath109zn@xmath20)@xmath1o@xmath2 at @xmath8 .
the critical value of @xmath27 is between @xmath110 and @xmath111@xcite ( probably closer to @xmath111 , judging from in - plane resistivity measurements@xcite ) , as compared to our calculated critical value @xmath94 for this @xmath29 [ cf .
fig . 2 ] . but a direct comparison of these values is difficult : the in - plane zn concentration may differ from @xmath27 .
if in fact @xmath28 , as mentioned above , then our model seems to underestimate @xmath13 ; the underestimate is less in the rca than in the more approximate ema .
but this apparent discrepancy in @xmath13 may not be very significant , for two reasons .
first , @xmath13 is quite sensitive to the parameter @xmath98 : a change of @xmath14 only from @xmath112 to @xmath113 increases @xmath13 by more than a factor of 2 in the ema , and presumably by a similar amount in the rca .
as already mentioned , this parameter is not known experimentally with great precision .
secondly , our model assumes that the cuo@xmath0 layer undergoes an _ discontinuous _ change from s to n character at a distance @xmath14 from a zn impurity .
it would be more realistic to assume a gradual change .
although such a transition region might be difficult to include in the present model , its qualitative effect would probably be to reduce the effective radius of the normal region to a value smaller than @xmath14 .
such a reduction would , according to our results , _ increase _ @xmath13 , making it closer to the reported value@xcite .
it is of interest to compare the present model to several others in the literature . in the original model of nachumi _
et al_@xcite , the superfluid density is proportional to the superconducting areal fraction , i. e. , to the fraction of the cuo@xmath0 plane which is more than a distance @xmath5 from any zn impurity .
the present model improves on the implementation of this swiss cheese model by taking into account effects of connectivity of the superconducting parts .
this treatment is consistent with the generalized london equation [ eq . ( 5 ) ] . in another class of models ( see , e. g. , refs .
@xcite ) @xmath18 is calculated using the bogoliubov - de gennes equation with some form of static disorder .
this model , like the swiss - cheese model , also leads to a spatially varying gap function or pairing potential .
although this model seems quite different from ours , we speculate that both actually have much of the same physical content .
the difference is that in the models of refs .
@xcite , the disorder is on an atomic scale and treated microscopically , whereas in the present case the disorder is on a greater length scale and treated within continuum electrodynamics . nonetheless , both lead to surprisingly similar behavior ( superfluid density diminishing rapidly with increasing @xmath3 and vanishing at a critical concentration ) .
we do not know how to connect the two on a more rigorous basis , however .
a nontrivial prediction of the present model is the existence of a peak in re@xmath85 near @xmath13 .
if such a peak were observed in experiments , it could be viewed as strong evidence in favor of the present model .
this `` percolation peak '' is reminiscent of the well - known fluctuation peak in re@xmath114 near t@xmath20 in a homogeneous superconductor , but there are significant differences in the underlying physics .
the fluctuation peak results from slow motions of superconducting fluctuations with size of order @xmath115 .
the spatial extent of these fluctuations diverges near @xmath23 , as does the time scale on which they move ; this diverging time scale leads to a strong low - frequency peak in the conductivity .
by contrast the percolation peak involves motion of charge carriers through a _
static _ disordered structure ; the frequency - dependence arises from the slow motion of these charges through the weakly connected structure ( but _ static _ structure of superconductor that exists , in our model , near the @xmath116 ( or @xmath13 ) .
the characteristic length for this structure is the so - called percolation correlation length @xmath117 , which diverges near @xmath116 on both sides of the percolation transition@xcite .
the existence of this percolation peak can be understood from the following crude argument . on the normal metal side of the percolation threshold , the static d. c.
conductivity is finite , but should diverge as @xmath118 .
this divergence is due to the ever - larger superconducting regions , which come closer and closer to shorting out the voltage across the normal metal as @xmath116 is approached .
this divergence should disappear as the frequency is increased , because the contrast in conductivities between superconducting and normal regions will become much smaller in absolute value ( that is , the ratio @xmath119 should approach unity ) .
therefore , at a fixed @xmath76 just above @xmath116 ( that is , on the normal metal side ) , re@xmath120 , should show a strong maximum at @xmath121 , and this maximum should diverge as @xmath118 .
since @xmath122 should vary continuously across the percolation threshold , except at exactly @xmath121 , we expect the same behavior on the superconducting side of the threshold .
this is qualitatively the behavior we see in both our ema and rca results .
the behavior of this `` critical peak '' can be estimated from the standard scaling theory of the percolation threshold ( see , for example , ref .
this scaling approach is based on the assumption that the percolation threshold is a critical point described by a single diverging length @xmath117 ; if that assumption is correct , then the scaling approach is believed to be exact . as applied to the present model , and using the homogeneity relations ( 11 ) and ( 12 ) .
this scaling approach dictates that @xmath60 have the form @xmath123 where @xmath124 , and @xmath125 and @xmath126 are standard percolation exponents .
the function @xmath127 of the complex variable @xmath128 has the limiting behaviors @xcite @xmath129 @xmath130 @xmath131 where @xmath132 , @xmath133 , and @xmath134 are constants .
this form implies that precisely at @xmath135 both the real and imaginary parts of @xmath60 satisfy @xmath136 in conventional 2d percolation theory , @xmath137 ; so both re@xmath138 and im@xmath138 would be proportional to @xmath139 .
the present `` swiss cheese '' model may not , however , be in the same universality class as percolation on a lattice . instead
, it may fall into the class of certain continuum percolation models@xcite . in the present case
, the exponent @xmath126 and @xmath125 should have the values corresponding to another `` swiss cheese '' model in 2d , in which the area between the holes is conducting while the holes are insulating .
according to the discussion of ref .
@xcite , the swiss cheese @xmath126 and @xmath125 are , in fact , unchanged from their lattice values of 1.3 in two dimensions .
therefore , one still expects that both re@xmath138 and i m @xmath138 should have a @xmath139 frequency dependence in this model .
it is also of interest to consider the predictions of the present model regarding the _ critical current density _
@xmath140 in zn - doped . in this case
, we use the results of continuum percolation theory as applied to @xmath140@xcite , or equivalently , to @xmath141 . in two dimensions , for the present swiss - cheese model
, it is found@xcite that @xmath142 .
the critical exponent @xmath143 is estimated as @xmath144 for swiss cheese in @xmath145 dimensions , where @xmath146 is the percolation correlation length exponent defined by @xmath147 .
thus , in two dimensions @xmath148 .
the conductivity exponent @xmath126 is unchanged from its lattice value in @xmath149 . using the approximations @xmath150 in @xmath151@xcite
, we obtain @xmath152 .
equivalently , this may be written ( for @xmath76 near the percolation threshold ) @xmath153^{7/4}$ ] . in terms of @xmath3 , @xmath154^{7/4}$ ] . in summary , @xmath140 in zn - doped yba@xmath0cu@xmath1o@xmath2
is predicted to go to zero near the percolation threshold even more rapidly than does @xmath18 .
it would be of interest if this prediction could be tested experimentally .
the present model does not give any information directly about the superconducting transition temperature @xmath155 .
some experiments on zn - doped yba@xmath0cu@xmath1o@xmath2 indicate that @xmath155 decreases much more slowly with @xmath3 than does @xmath156 ( see , for example , ref.@xcite ) .
although we have no quantitative model for this behavior , a simple qualitative argument does suggest a explanation .
let us suppose that @xmath108 , so that the superconducting portion of the material is connected throughout the plane . in that case ,
if _ thermal fluctuation effects _ can be neglected , then @xmath155 ( as measured by a vanishing d. c. resistivity ) should be _ independent _ of @xmath3 for @xmath108 , and zero for @xmath157 .
in this limit , clearly @xmath158 . in reality , thermal fluctuations should reduce @xmath155 below its expected value in the absence of fluctuations , but this inequality should still hold , as observed experimentally . in summary , we have presented a simple model for the variation of superfluid density with in - plane zn concentration @xmath3 in zn - doped .
the model is based on the `` swiss cheese '' picture suggested by nachumi _
et al_@xcite . using this model
, we calculate the ratio @xmath59 for zn - doped yba@xmath0cu@xmath1o@xmath2 as a function of the ratio @xmath25 , using two different approximations .
the model predicts a critical zn concentration above which @xmath18 is zero , in rough agreement with experiment for zn - doped yba@xmath0cu@xmath1o@xmath2 at @xmath8 .
the model also predicts a low - frequency peak in re@xmath85 near @xmath13 , the existence of which has apparently not been checked experimentally .
finally , the model suggests that the low - temperature critical current ratio @xmath159^{7/4}$ ] near @xmath135 .
the observation of these features would give additional experimental support to the present simple model .
we would like to thank professor thomas r. lemberger for helpful discussions .
jeng - da chai thanks l. j. schradin for valuable discussions regarding computation and m. e. howard for useful comments on the manuscript .
this work has been supported by nsf grant no .
dmr97 - 31511 .
d. a. wollman , d. j. van harlingen , w. c. lee , d. m. ginsberg , and a. j. leggett , phys .
lett.*71 * , 2134 ( 1993 ) ; c. c. tsuei , j. r. kirtley , c. c. chi , l. s. yujahnes , a. gupta , t. shaw , j. z. sun , and m. b. ketchen , phys .
lett . * 73 * , 593 ( 1994 ) . | we describe a simple model for calculating the zero - temperature superfluid density of zn - doped yba@xmath0cu@xmath1o@xmath2 as a function of the fraction @xmath3 of in - plane cu atoms which are replaced by zn .
the basis of the calculation is a `` swiss cheese '' picture of a single cuo@xmath0 layer , in which a substitutional zn impurity creates a normal region of area @xmath4 around it as originally suggested by nachumi _
et al_. here @xmath5 is the zero - temperature in - plane coherence length at @xmath6 .
we use this picture to calculate the variation of the in - plane superfluid density with @xmath3 at temperature @xmath7 , using both a numerical approach and an analytical approximation . for @xmath8 , if we use the value @xmath9 @xmath10 , we find that the in - plane superfluid decreases with increasing @xmath3 and vanishes near @xmath11 in the analytical approximation , and near @xmath12 in the numerical approach .
@xmath13 is quite sensitive to @xmath14 , whose value is not widely agreed upon .
the model also predicts a peak in the real part of the conductivity , re@xmath15 , at concentrations @xmath16 , and low frequencies , and a variation of critical current density with @xmath3 of the form @xmath17 near percolation , where @xmath18 is the in - plane superfluid density . |
damped lyman - alpha systems ( dla ) provide an effective abundance probe of the early universe .
ranging in redshift from 1.7 - 4 , they allow us to study the elemental composition of the universe back to within roughly one billion years of the big bang .
since dlas are absorption systems with neutral h column densities @xmath1 @xmath2 seen against light from background qsos , their optical depths are large enough to offer a good chance for detecting faint metal lines .
converting these line profiles into ion abundances is relatively straightforward and corrections for unseen ions appear to be minimal , so that inferred gas - phase elemental abundances are reasonably dependable .
abundances of numerous elements including c , n , o , si , s , fe , zn , and cr are routinely studied and measured in dlas . extended discussions about abundances in dlas
are available in prochaska et al .
( 2001 ) , prochaska & wolfe ( 2002 ) , prochaska et al .
( 2002 ; hereafter pr02 ) , pettini et al . ( 2002 ) , and molaro s review article in this conference . the use of dlas to probe the early production of nitrogen is of particular interest .
because temperatures required for its synthesis are relatively low ( @xmath315 million k ) , nitrogen may be produced within the h - burning cores and shells of stars representing a broad mass range .
there is strong empirical ( henry , kwitter , & bates 2000 ) and theoretical ( van den hoek & groenewegen 1997 ) evidence that intermediate mass stars ( ims ; 1@xmath4m@xmath48 m@xmath5 ) produce prodigious quantities of nitrogen .
in addition , henry et al .
( 2000 ; hereafter hek ) and chiappini et al .
( 2003 ) have analyzed extant data and concluded that ims are the principal sources of n in the universe . using the same data izotov & thuan ( 1998 ) have argued that massive stars produce the bulk of the nitrogen when metallicity is low , while pilyugin et al .
( 2003 ) have recently suggested that the data do not allow us to say for certain whether ims or massive stars are the principal n production sites .
so we really do nt know for sure where n comes from .
if izotov & thuan are correct , then n and other metals produced in massive stars are released into the ism simultaneously . however , if ims produce the bulk of the n , then its production lags behind most other elements produced in massive stars because of the finite evolutionary lifetimes of ims , a time which hek have estimated to be about 250 myr .
this delay could be longer if rotation or low metallicity lower the effective mass range for n production .
all of this means that nitrogen is a particularly interesting element to track over time , since it s evolution with respect to elements from massive stars can help us gauge the true lag time of n production , and establish the stellar source , i.e. ims or massive stars .
because dlas probe early universe chemical evolution , studies of their n abundances can help us gauge n production in systems of low metallicity which are experiencing their first episodes of star formation .
recently , pettini et al .
( 2002 ) and pr02 have published n abundances for over 20 dlas , along with abundances for si and s , which , because of their massive star origins , serve as oxygen surrogates and track total stellar chemical processing of gas .
1 is a plot of [ n/@xmath0 ] versus [ @xmath0/h ] , where the bracketed values refer to logarithmic quantities normalized to their respective solar values ( grevesse et al .
1996 ) . the filled circles ( established values ) and triangles ( upper / lower limits ) represent a combined sample from pr02 and pettini et al .
the letter symbols show h ii region and dwarf irregular galaxy data , while stars show stellar data , all compiled and explained in hek .
note that o , si , and s are collectively referred to as alpha elements , since their abundances appear to scale with each other .
1 can be interpreted as follows . the abcissa gauges the extent of chemical processing of a system , so objects located farther to the right have experienced a greater amount of star formation and alpha element production over their lifetimes .
ordinate values provide a measure of n nucleosynthesis relative to that of the alpha elements . the loose pattern formed by objects
other than dlas suggests that n/@xmath0 is roughly constant at low metallicity , but increases at higher metallicity values as metallicity climbs .
this is consistent with the idea that nitrogen follows a primary - like behavior at low metallicities but a secondary - like behavior at higher values .
conclusions relating to the dla distribution are : 1 ) most dlas fall along the primary plateau at [ n/@xmath0]@xmath7 - 0.75 ; but 2 ) a small number fall significantly below the plateau , albeit at similar metallicity values as those in the first group , with a centroid value of [ n/@xmath0]@xmath7 - 1.4 .
this apparent bimodality in the dla distribution can not be explained by dust or unseen ions of nitrogen .
clearly a group of five objects such as the low nitrogen dlas ( ln - dla ) represents a statistically small sample , and it is possible that the distinction between the two dla groups will become more blurred as more n abundances are measured in dlas .
but the empirical result is nevertheless intriguing .
so , in what follows we will assume that the bimodal dla distribution is real and that ims produce most of the nitrogen in the universe .
we now explore four possibilities for explaining the dla bimodality .
the first explanation is that we are witnessing the effects of time delay in n production ( hek ; pettini et al .
following initial star formation , massive stars die quickly and expel alpha elements along with a small amount of n into the system s interstellar medium . then after a delay , primary n from ims is expelled and n/@xmath0 rises to the primary plateau .
the ln - dlas , therefore , may be objects in which ims have not yet released their n , so their n/@xmath0 levels are determined only by massive star n production .
2 shows the temporal evolution of [ n/@xmath0 ] after a star burst , as predicted by a chemical evolution model similar to those discussed in hek .
notice the abrupt rise in [ n/@xmath0 ] to a small plateau value of around -1.8 .
the plateau arises because massive stars have ceased their n production but ims have not yet begun to make their contribution to the interstellar medium .
however , as ims begin to expel their n at around 60 myr , [ n/@xmath0 ] resumes it upward trend until it reaches the primary plateau value near -0.7 .
notice that the rise to the plateau value is gradual rather than abrupt , forecasting a smooth distribution of objects between the ln - dla location and the primary plateau , contrary to what is observed . on the other hand ,
a bimodal distribution in [ n/@xmath0 ] is more consistent with a sudden rise in n production , so that there would be only a small probability of observing dlas between the primary plateau and the observed region of ln - dlas .
an example of such behavior is shown with a dashed line in fig . 2 .
if the gap becomes filled in with additional objects in the future , then the time delay scenario would become feasible .
for now , it appears to be inconsistent with a bimodal distribution . a second possibility for explaining the observed dla distribution relates to the metallicity sensitivity of stellar yields .
a simple closed - box model of chemical evolution can be used to show that the effective yield , is the stellar yield as a function of mass integrated over a salpeter imf .
see hek for details .
] ratio of n / o needed to explain the observed [ n/@xmath0 ] of -1.4 of the ln - dlas is @xmath8 henceforth called the `` empirical '' value .
the left panel of fig . 3 shows a comparison of this empirical ratio with predictions of two yield sets .
the first set combines the yields of van den hoek & groenewegen ( 1997 ) for ims and woosley & weaver ( 1995 ) for massive stars ( hg+ww ) , while the second set links marigo s ( 2001 ) ims yields and portinari et al.s ( 1998 ) massive star yields ( mp ) .
the first set results in a value which is 1.5 dex above the empirical level , while the second set predicts a number which is 0.9 dex above the empirical value .
so current ratios of effective yields appear significantly larger than what is required to explain the ln - dlas .
additionally , it would be difficult to understand how a stellar population characterized by a single metallicity could produce a bimodal distribution in [ n/@xmath0 ] from a single set of yields unless one introduces another parameter such as rotation into the picture ( langer et al .
1997 ; maeder & meynet 2001 ) . a third explanation for the ln - dlas may come from nucleosynthesis occurring at zero metallicity , i.e. from population iii stars .
yields for such star groups have been calculated by woosley & weaver ( 1995 ; 13 - 40 m@xmath5 ; ww ) , umeda et al .
( 2000 ; 13 - 25 m@xmath5 ; u00 ) , and heger & woosley ( 2002 ; 140 - 260 m@xmath5 ; hw ) for the indicated mass ranges .
the results from these yields are plotted in the right panel of fig .
the predictions of woosley & weaver and heger & woosley fall far below the empirical value , while the predictions of umeda et al .
exceed it .
currently , the population iii yield predictions are inconsistent with the observed [ n/@xmath0 ] level , although investigations of these stars are in an early stage .
thus , future work may still prove that this mechanism is feasible .
the fourth hypothesis , and the one that we currently favor , involves altering the imf .
we experimented with both top - heavy ( flatter ) imfs and imfs which are truncated below a certain mass threshold .
in both cases this reduces the ims : massive star ratio and causes the n / o ratio to fall noticeably , since we are assuming that ims produce the bulk of the n. we calculated simple numerical chemical evolution models of closed systems using the program described in hek along with the mp yield set and experimented with various imf slopes and mass thresholds .
the solid lines in fig . 1 correspond to models which assumed instantaneous recycling , i.e. no time lag between ims and ms evolution , while the dot - dashed lines show the effect of relaxing this assumption .
the top curves refer to an imf close to the salpeter relation ( slope=1.35 ) .
meanwhile , the bottom curves show what happens when the imf slope is flattened to 0.10 .
in addition , the dashed line refers to a model in which the imf is truncated below 4.8m@xmath5 .
similar results are obtained if these experiments are repeated using the hg+ww yield set .
clearly , a flat or truncated imf is consistent with the ln - dlas .
it is interesting to note that the popular view of population iii stars comprising only massive objects is consistent with the idea of a truncated imf .
what we are proposing , then , is that star formation at low metallicity produces a relatively top - heavy population of stars in some systems , resulting in a n - deficient enrichment compared to the standard primary ratio .
such a scenario offers a straightforward mechanism for producing the low [ n/@xmath0 ] value observed for the ln - dlas .
if correct , these new observations may offer the first evidence that the mass spectra of some early stellar populations were much different than what we observe today .
grevesse , n. , noels , a. , & sauval , a.j .
1996 , grevesse , n. , noels , a. , & sauval , a.j .
1996 , in asp conf .
99 , `` cosmic abundances , '' ed .
holt & g. sonneborn ( san francisco : asp ) , 117 in asp conf .
99 , cosmic abundances , ed .
holt & g. sonneborn ( san francisco : asp ) , 117 | recent measurements of nitrogen and alpha elements in over 20 damped ly-@xmath0 systems ( dla ) are compared with similar measurements for numerous emission line objects and stars .
it is found that the dla distribution in the n/@xmath0-@xmath0/h plane is bimodal , where most sample dlas fall along the n/@xmath0 plateau defined at low z by dwarf irregulars , while a small group possesses n/@xmath0 values roughly 0.7 dex less than those on the plateau at similar @xmath0/h values .
we demonstrate with chemical evolution models that a top - heavy or truncated imf can account for the low n in this second group .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
with a riemannian symmetric space @xmath4 of the non - compact type comes a natural complexification @xmath2 , the so - called _ complex crown _ of @xmath5 .
one can define @xmath2 in various ways . to begin with
let @xmath6 be an iwasawa decomposition and let @xmath7 be the lie algebra of @xmath8 .
studying proper @xmath9-actions on @xmath10 akhiezer and gindikin were led to the following definition ( cf .
@xcite ) @xmath11 where @xmath12 is a certain bounded convex set
section 2 ) . observe that @xmath2 is a @xmath9-invariant domain in @xmath13 containing @xmath5 .
moreover , the definition of @xmath2 is independent from the choice of @xmath14 and hence @xmath2 is generically defined through @xmath5 .
equivalently , one can define @xmath2 by @xmath15 where @xmath16 refers to the connected component of @xmath17 containing @xmath5 ( cf .
@xcite , @xcite , @xcite and @xcite ) . using the _ complex convexity theorem _ from @xcite one
can significantly improve on [ b ] , namely @xmath18 the various definitions of @xmath2 are useful . in @xcite , where the foundations for the @xmath9-invariant complex geometry of @xmath2 were laid out ( construction of natural @xmath9-invariant psh exhaustion functions , khler metrics etc .
) , one mainly used the description ( [ a ] ) . for other purposes such as analytic continuation of eigenfunctions or the heat kernel on @xmath5 to holomorphic functions on @xmath2 ( cf .
@xcite , @xcite ) the characterizations in ( [ b ] ) and in particular ( [ c ] ) are more appropriate .
one of the main objectives in the study of complex crowns is to achieve a better understanding of harmonic analyis on symmetric spaces .
for example it was shown in @xcite that all non - compactly causal symmetric spaces @xmath19 associated to @xmath9 appear in the so - called _ distinguished boundary _ of @xmath2 .
subsequently this was used in @xcite to construct a hardy - space for non - compactly causal symmetric spaces @xmath19 yielding first progress towards a geometric realization of some of the continuous series in @xmath20 ( gelfand - gindikin program ) .
the motivation for this article stems from our interest to obtain a first understanding of the various kinds of weighted bergman - spaces which can be associated with @xmath2 , most notably fock - spaces which play a prominent role in the context of the heat kernel transform on @xmath5 ( cf .
@xcite ) . intimately related to this circle of problems is the growth behavior of analytically continued spherical functions .
a spherical function @xmath21 on @xmath5 is left @xmath22-invariant , hence its analytic continuation @xmath23 to @xmath2 is necessarily ( locally ) @xmath24-invariant .
this suggests that one should study the relation of @xmath2 with the complexified polar decomposition @xmath25 in @xmath13 . whereas the complex crown @xmath2 is contained in the complexified iwasawa decomposition @xmath26 ( cf .
( [ b ] ) ) , the same does not hold for the complexified polar decomposition , i.e. , @xmath27 .
furthermore , the polar domain @xmath28 is not even open in @xmath13 .
these are disappointing facts . but before giving up , we at least wanted to understand what actually goes `` wrong '' in the most important case of @xmath29 .
it turned out that the above mentioned difficulties can be overcome , and , moreover , by settling the technical problems we could make new observations regarding the structure theory of complex crowns .
this paper focuses mostly on @xmath29 , but we made an effort to present results in a fashion which allows immediate generalization to all semisimple lie groups .
an approach in full generality will be given elsewhere .
let us now describe some of our results in more detail .
first observe that the polar domain @xmath28 contains a zariski open subset of @xmath13 ; in particular it is dense in @xmath13 .
thus almost all elements in @xmath2 are contained in @xmath28 .
our main observation is the following non - linear convexity theorem : * theorem 4.1 . *
_ let @xmath30 and @xmath31 with @xmath32 and @xmath33 ( cf .
( [ a ] ) ) . then @xmath34 with @xmath35 the convex hull of the weyl group orbit of @xmath36 .
_ we conjecture that theorem 4.1 holds true for all semisimple lie groups and we provide additional evidence with a discussion of the lorentz groups @xmath37 .
using general results of @xcite one obtains from theorem 4.1 the following estimate : * theorem 5.2 . *
_ let @xmath30 .
let @xmath21 be a positive definite spherical function on @xmath5 and @xmath23 its analytic continuation to @xmath2
. then @xmath23 is bounded .
_ finally , as an application of our methods we give an estimate for the @xmath38-projection in @xmath28 for elements in @xmath2 .
it is our pleasure to thank the referee for his very careful reading and his useful suggestions .
in particular the easy proof of theorem [ union ] is due to him .
let @xmath39 be a real semisimple lie algebra and @xmath40 a cartan decomposition .
for a maximal abelian subspace @xmath14 of @xmath41 let @xmath42 be the corresponding restricted root system .
then @xmath43 admits a root space decomposition @xmath44 where @xmath45 and @xmath46 = \alpha(h)x\ } \ $ ] .
+ for a fixed positive system @xmath47 define @xmath48 .
then we have the iwasawa decomposition on the lie algebra level : @xmath49 we write @xmath50 for the corresponding _
weyl group_. for an element @xmath51 we denote by @xmath52 the convex hull of the weyl group orbit of @xmath5 .
+ + for any real lie algebra @xmath53 we write @xmath54 for its complexification .
+ in the sequel @xmath55 will denote a simply connected lie group with lie algebra @xmath56 .
we write @xmath57 and @xmath58 the analytic subgroups of @xmath55 corresponding to subalgebras @xmath59 and @xmath60 , respectively .
+ + the following bounded and convex subset of @xmath14 plays a central role : @xmath61 + with @xmath62 we define a left @xmath9 and right @xmath24-invariant domain in @xmath55 by @xmath63 also we write @xmath64 for the union of right @xmath24-cosets of @xmath65 in the complex symmetric space @xmath66 .
we refer to @xmath2 as the _ complex crown _ of the symmetric space @xmath67 .
notice that @xmath2 is independent of the choice of @xmath68 and hence generically defined through @xmath67 .
as every root @xmath69 is analytically integral , it gives rise to a character of @xmath38 by @xmath70 + let us define the set of _ regular elements _ in @xmath38 by @xmath71 notice that @xmath72 is an algebraic variety ( principal open set in @xmath38 ) .
+ we also define @xmath73 and call it the _ singular set _ in @xmath38 .
the following assertions hold : 1 .
the multiplication mapping @xmath74 is submersive . in particular
, @xmath75 is open .
@xmath76 is dense in @xmath55 .
\(1 ) is a standard computation which will not be repeated here .
\(2 ) notice that @xmath77 is a constructible set as the image under the regular mapping @xmath78 .
thus it follows from ( 1 ) that @xmath79 contains a zariski - open subset of @xmath55 .
this proves ( 2 ) .
notice that the @xmath24-bi - invariant domain @xmath80 is not open in @xmath55 ( cf .
our discussion of @xmath29 in section 3 ) .
in this section we will give a detailed analysis of the double @xmath24-coset decomposition for @xmath81 .
further we will start the investigation of the intersections of @xmath65 with the various cells in the double @xmath24-coset decomposition .
let us introduce the necessary notation .
our choice of a maximal compact subgroup of @xmath9 is @xmath82 and our choice of @xmath68 will be @xmath83 we choose @xmath84 such that @xmath85 notice that @xmath86 is simply connected and @xmath87 . in order to study @xmath24-double cosets
the use of spherical functions is useful .
on @xmath86 we define the _ elementary spherical function _ @xmath88 by @xmath89 ( for @xmath90 we have @xmath91 . ) note that the function @xmath92 is holomorphic and @xmath24-bi - invariant .
the set of singular elements in @xmath38 is given by @xmath93 let @xmath94 according to lemma 2.1 we can find sequences @xmath95 [ limits ] let @xmath96 be the limit of the sequence @xmath97 then @xmath98 is bounded and for every convergent subsequence @xmath99 we have @xmath100 write @xmath101 . by continuity of the map
@xmath92 we get @xmath102 it follows that there exist @xmath103 such that @xmath104 passing to an appropriate subsequence we can assume that @xmath105 + + suppose @xmath106 has a convergent subsequence .
we may assume that @xmath107 but then @xmath108 also converges .
+ we have now @xmath109 but this would imply @xmath110 contradicting the assumption .
+ by the same argument we see that @xmath111 can not have a convergent subsequence .
+ + we want to know more about the limits @xmath112 .
+ notice that @xmath113 converges in @xmath55 .
for simplicity we omit the indices writing @xmath114 and @xmath115 .
we look at the elements @xmath116 + due to the @xmath24-bi - invariance of @xmath117 and the compactness of @xmath22 we can assume that @xmath118 where @xmath119 and @xmath120 .
then with @xmath121 we have @xmath122 we know that @xmath123 and @xmath124 + the upper left entry in ( [ matrix ] ) must converge , therefore @xmath125 but this implies @xmath126 , hence @xmath127 the vector space @xmath128 admits a decomposition @xmath129 into irreducible @xmath130-modules with @xmath131 the corresponding analytic subgroups of @xmath55 are given by @xmath132 we introduce punctured discs in @xmath133 and @xmath134 by @xmath135 and similarily we define @xmath136 .
+ finally , we set @xmath137 + and define an element @xmath138 we can now describe how @xmath55 decomposes into a disjoint union of @xmath24-bi - invariant subsets .
furthermore , we will give a complete description of the intersection of @xmath139 with the lower dimensional cells .
the intersection of @xmath140 with the big cell @xmath141 will be subject of the next section .
+ + [ union ] 1 . the @xmath24-double coset decomposition of @xmath55 is given by : @xmath142 2 .
the following equality holds : @xmath143 in order to prove theorem [ union ] it is useful to adopt a more geometric point of view .
we closely follow the suggestions of the referee . in the sequel
we realize @xmath144 as the upper sheet of the two - sheeted hyperboloid @xmath145 and @xmath146 as the complex quadric @xmath147 let @xmath148 .
then the polar decomposition in @xmath5 is given by @xmath149 , @xmath150 with @xmath151 .
@xmath152 complexifying the real polar decomposition of @xmath5 it follows that a point @xmath153 belongs to @xmath154 if and only if there exist @xmath155 with @xmath156 such that @xmath157 therefore @xmath158 decomposes into four @xmath159-orbits @xmath160 where @xmath161 .
+ the orbits @xmath162 correspond to the @xmath159-bi - invariant sets @xmath163 , @xmath164 , @xmath165 , @xmath166 , respectively .
+ this proves part @xmath167 of the theorem .
+ + the points @xmath168 are characterized by the property @xmath169 for an element @xmath170 in the orbit @xmath167 this translates to @xmath171
. therefore , @xmath172 and this subset of @xmath2 corresponds to @xmath173 .
+ similarly one relates orbit @xmath174 with @xmath175 .
+ the intersections of both orbits @xmath176 and @xmath177 with @xmath65 are empty .
\(a ) theorem [ union](1 ) can be deduced from general results of matsuki ( cf .
@xcite ) .
\(b ) the cells in the decomposition of @xmath55 in theorem [ union ] are characterized by the values of @xmath92 .
the function @xmath92 attains the value @xmath178 on @xmath24 and on @xmath179 , the value @xmath180 on @xmath181 , and @xmath92 can take any value except @xmath178 on @xmath182 .
in this section we will investigate the intersection of @xmath140 with the big cell @xmath183 .
an element @xmath184 can be written as @xmath185 where @xmath186 and @xmath187 we want to describe the dependence of @xmath188 from @xmath189 and @xmath190 .
+ + recall that for @xmath191 the weyl group consists of only two elements and for @xmath192 we have @xmath193 \cdot x \ $ ] .
[ conv ] let @xmath194 and @xmath195 .
suppose that @xmath196 .
then @xmath197 using the iwasawa decomposition @xmath198 we may assume that @xmath199 if we set @xmath200 and @xmath201 , then @xmath202 yielding @xmath203 in figure 1 the set of points of the hyperbola is @xmath204 .
the complex number @xmath205 lies within the shaded region .
therefore , there exists @xmath206 such that @xmath207 for some real number @xmath208 .
we now obtain the assertion of the theorem with the following lemma [ convlem ] .
[ convlem ] let @xmath209 and @xmath210 .
define complex numbers by @xmath211 and @xmath212 .
assume that @xmath213 , with @xmath214 .
then , @xmath215 .
clearly we can assume @xmath216 .
comparing real and imaginary parts of @xmath217 , followed by squaring , yields @xmath218 and @xmath219 adding these two equations we get @xmath220 combining ( [ eq1 ] ) and ( [ eq2 ] ) gives @xmath221 if we assume @xmath222 , then @xmath223 and @xmath224 a contradiction to ( [ eq3 ] ) .
theorem [ conv ] holds for any real connected semisimple lie group @xmath9 . to give more evidence for our conjecture
we will now discuss the case of @xmath225 , @xmath226 .
on the lie algebra level we have the cartan decomposition @xmath227 , where @xmath228 choosing @xmath229 as maximal subalgebra of @xmath41 yields the root system @xmath230 with @xmath231 therefore , @xmath232 we choose @xmath233 as the positive root
. then the subgroups in the iwasawa decomposition @xmath198 are @xmath234 we consider the standard representation @xmath235 of @xmath9 on @xmath236 .
+ clearly , @xmath237 is a @xmath22-fixed vector .
the elementary spherical function @xmath92 on @xmath238 is given by @xmath239 note that @xmath240 is just the upper left entry @xmath241 in the matrix @xmath242 .
+ now let @xmath243 .
we can write @xmath244 for some @xmath245 and @xmath246 applying @xmath92 yields @xmath247 or @xmath248 equation ( [ so(1,n ) ] ) , figure 1 ( with a slight modification in the notation ) and lemma [ convlem ] show again that @xmath249 , as claimed .
let us briefly introduce the necessary notation .
write @xmath6 for the iwasawa decomposition of @xmath29 with @xmath250 , @xmath8 and @xmath22 as in section 3 .
for @xmath32 we define @xmath251 by @xmath252 . for @xmath253 and @xmath254
we set @xmath255 .
following harish - chandra we define the _ spherical function on @xmath67 with parameter @xmath253 _ by @xmath256 where @xmath257 denotes the normalized haar - measure on @xmath22 and @xmath258 with @xmath259 .
define @xmath260 and recall that @xmath261 .
we have to recall some facts from @xcite 4 on the analytic continuation of the spherical functions ( actually valid for all semisimple lie groups @xmath9 ) .
[ ks ] let @xmath253 .
then the following assertions hold : 1 .
the spherical function @xmath21 has a unique holomorphic extension to @xmath2 .
2 . the spherical function
@xmath21 has a unique continuation to a @xmath24-bi - invariant function on @xmath262 such that the restriction to @xmath263 is holomorphic .
moreover , if @xmath264 and @xmath265 is a compact subset , then the restriction of @xmath21 to @xmath266 is bounded . in the sequel
we also denote by @xmath21 the holomorphic extension of @xmath21 to @xmath2 .
we denote by @xmath267 the center of @xmath9 .
our result in this section then is : [ sph ] let @xmath268 .
then the spherical function @xmath21 is bounded on @xmath2 , i.e. , @xmath269 by theorem [ conv ] we have @xmath270 thus proposition [ ks ] implies that @xmath21 is bounded on @xmath271 .
let now @xmath272 such that @xmath273 .
by the density of @xmath183 in @xmath55 we can find sequences @xmath274 , @xmath275 such that @xmath276 for all @xmath277 and @xmath278 . by theorem [ conv ]
we have @xmath279 and thus @xmath280 by proposition [ limits ] . by the continuity of @xmath21
we hence obtain that @xmath281 concluding the proof of the theorem .
\(a ) note that proposition [ ks ] holds for all semisimple lie groups .
thus in order to generalize theorem [ sph ] to all semisimple lie groups one needs proposition [ limits ] and theorem [ conv ] in full generality ( cf . our conjecture in section 4 ) .
\(b ) fix a @xmath253 and consider the following function on @xmath282 @xmath283 identifying @xmath67 with the unit disc @xmath284 and @xmath22 with its boundary , this function is easily seen to be a power of the poisson kernel on @xmath284 . according to @xcite the function @xmath285 admits an analytic continuation to @xmath286 which is holomorphic in the first variable . also by @xcite
we know that this function is unbounded for @xmath287 .
observe that the spherical function on @xmath2 is given by @xmath288 now notice that theorem [ sph ] says that @xmath21 stays bounded despite the fact that the integrands become singular towards the boundary of @xmath2 .
in this section we will investigate the growth of the middle projection in the @xmath183-decomposition for elements in @xmath2 .
using @xmath304 we can write @xmath305 we may assume that @xmath306 and write @xmath307 .
then @xmath308 hence @xmath309 is a convex combination of @xmath310 and @xmath311 and therefore lies on the vertical segment shown in figure 2 .
+ + if @xmath323 , then ( [ hirsch ] ) implies @xmath324 since @xmath325 this means @xmath326 but now we end up with @xmath327 which contradicts the assumption .
thus @xmath328 proving our first assertion .
the second assertion is now immediate from ( [ hirsch ] ) . | we make a fine study of the @xmath0-double coset decomposition in @xmath1 and give a full description of the intersection of the various cells with the complex crown @xmath2 of @xmath3 .
a non - linear convexity theorem is proved and applications to analytically continued spherical functions are given . |
numerical simulation of the nonlinear evolution of collisionless dark matter , usually by following a set of point masses as they move under their mutual gravitational influence , has , over the last few decades , played an important role in increasing our understanding of cosmological structure formation .
rather than attempt to make a summary of the extensive literature available on this subject here , we direct the reader to the recent reviews by @xcite and @xcite . a search of the internet will reveal a diverse set of n - body codes now available to be downloaded by those interested in carrying out cosmological simulations .
one of the first to be developed is the direct code of @xcite .
the operation count for implementing any such particle particle code scales as @xmath1 , where @xmath2 is the number of particles .
one can reach larger @xmath2 and thus higher mass resolution ( but spatial resolution limited by the cell size on a grid ) in the same amount of cpu time using the pm ( particle mesh ) method , for example the code by @xcite .
the operation count here scales as @xmath2log@xmath2 , with a small prefactor ; pm is implemented on regular grids , where efficient fast fourier transform ( fft ) algorithms are available . to attain higher , subgrid resolution one can add to a pm code particle particle interactions and refined subgrids , such as in the hydra code @xcite .
a gridless approach is used in tree codes , for example the zeno package of @xcite . the operation count in the barnes
hut tree algorithm @xcite also scales as @xmath2log@xmath2 but the prefactor is much larger by a factor of 1050 @xcite than for a pm code .
the tree code of @xcite has been made parallel @xcite .
a tree code with individual particle time steps called gadget has recently been released in both serial and distributed memory parallel versions @xcite .
one promising new development is the use of multigrid methods involving mesh refinements of arbitrary shape @xcite , such as in mlapm @xcite .
another cell - based approach has recently been developed which shows improved performance over standard tree codes @xcite and is included as part of the nemo package @xcite . for
a given choice of algorithm , a related problem is making it run efficiently on parallel computers @xcite .
gravity is long - range , so to compute the force on a given particle , one needs some information from every other particle , and communication is thus required if particles are distributed among many processors .
@xcite presented a new algorithm , called tree particle
mesh , based on using domain decomposition in a manner designed to overcome this difficulty .
tpm uses the efficient pm method for long - range forces and a tree code for sub - grid resolution .
isolated , overdense regions are each treated with a separate tree , thus ensuring coarse grained parallelism .
@xcite , hereafter box , made several improvements in the code of @xcite . in this paper
we present further refinements which improve the accuracy and efficiency of the algorithm . these include the following : allowing individual particle time steps within trees , an improved treatment of tidal forces , new criteria for higher force resolution and for the choice of time step , and parallel treatment of large trees . in
[ sec : overview ] we present an overview of latest version of tpm , including time stepping ; [ sec : implementation ] discusses implementation of domain decomposition and the particle time step criterion , with an overview of how the code works in practice .
the accuracy of the code is explored in [ sec : perfandacc ] by comparing results with two other algorithms : p@xmath0 m and the gadget tree code .
concluding remarks and details of the public release of the tpm code are given in
[ sec : discussion ] .
@xcite discuss the distinction between verifying a simulation ( that is , being sure that the equations are being solved correctly ) and validating it ( having confidence that the equations and their solution actually resemble a real world problem ) .
obviously , in this paper only the former is done .
various aspects of the validity of cosmological simulations are addressed in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and references therein .
when using an n - body code , the simulator will need to be cognizant of these issues .
tpm begins with a standard pm code ( in fact , if there are no particles selected to be in trees , then tpm defaults to a pm code ) ; this portion is similar to the pm code described by @xcite .
the density on a regular grid is found by interpolating particle positions using the cloud in cell ( cic ) method , and poisson s equation is solved on this grid using fast fourier transforms . as noted in the introduction ,
the fft technique is highly efficient and scales as @xmath2log@xmath2 @xcite . in order to obtain subgrid resolution a tree code
one possible manner of doing this is to compute shorter range forces for every particle with a tree code @xcite , in a manner analogous to the p@xmath0 m algorithm .
instead , tpm uses domain decomposition , taking advantage of the fact that potentials due to matter outside and inside a given domain can be added linearly , and it uses different solvers to find these two potentials .
this section begins with a very broad description of the tpm algorithm , and then considers various aspects in more detail .
a basic outline of tpm can be summarized as follows : 1 .
identify tree regions and the particles in each such region ; particles outside of tree regions are evolved with pm only .
push pm particles to midstep 3 .
find the tidal potential in each tree region the potential due to all mass outside of that region .
4 . integrate each tree region forward to the middle of the pm step .
5 . find the pm potential and update the acceleration for pm particles .
integrate each tree region forward to the end of the pm step .
update pm particle velocities and positions to the end of the time step .
the motivation behind this algorithm is that the motions of particles within a given tree region can be integrated independently using high temporal and spatial resolution : knowledge of the rest of the simulation is not required , except for the time averaged , externally produced tidal forces .
the stepping in time of positions @xmath3 and velocities @xmath4 is accomplished with a standard second order leapfrog integration in comoving coordinates , the cosmological model determining the scale factor @xmath5 : [ eqn : leapfrog ] @xmath6 here @xmath7 , @xmath8 , and the gravitational potential @xmath9 are determined at time @xmath10 .
one advantage of the tpm approach is that it allows the use of multiple time steps .
tree particles are required to take at least two steps per pm step , but each particle has an individual time step so that finer time resolution can be used if required .
particles are arranged in an hierarchy of time bins differing by a factor of two , in the manner of @xcite : @xmath11 where @xmath12 is the pm time step and the integer @xmath13 . in a sense tpm operates along the same lines as a tree code , except that the particles in the @xmath14 time step bin are handled differently from the rest .
a diagrammatic representation of the time stepping is shown in fig@xmath15[fig : tstep ] for the case when all tree particles are in the longest time step bin , @xmath16 .
beginning at time @xmath17 , the pm particle positions are moved forward to midstep ( eq . [ [ eqn : leapfrog]a ] with @xmath18 ) .
the tidal potential is then found , and the tree particles are evolved forward one full step with @xmath19 .
the pm potential is updated ( eq .
[ [ eqn : leapfrog]b ] ) , and for a second time the tree particles are evolved for one full step , to time @xmath20 .
finally , the pm particle velocities and positions are updated to the end of the step ( eq .
[ [ eqn : leapfrog]c - d ] ) . to repeat the tpm algorithm in more detail : [ [ section ] ] 1 .
+ + identify tree regions and the particles in each region .
this is done by identifying pm cells above a given density threshold , described in
[ sec : domdec ] .
adjoining cells are then grouped into isolated tree regions , as described in box .
[ [ section-1 ] ] 2 .
+ + push pm particles to midstep ( eq . [ [ eqn : leapfrog]a ] ) .
[ [ section-2 ] ] 3 .
+ + find the tidal potential @xmath21 for each tree region .
first the total potential is computed on the grid in the standard pm manner .
then for each of the tree regions , a small portion of the grid containing the pm potential is saved , and the contribution to this potential made by the tree itself is subtracted out , leaving the tidal potential due to all external mass .
see box for details .
since @xmath21 can only be calculated once per pm step , it will be calculated at midstep .
pm particles are thus already in the proper location for this , but tree particles , with positions at the beginning of the pm time step , are not . however , since tree regions are spatially separated and the halo density profiles are evolving on slower timescales than any individual particle s orbital period , an approximate position is sufficiently accurate .
( note that since a given tree s own contribution to @xmath21 is exactly subtracted back out , only the effect on other trees needs to be considered ) .
thus , an approximation of the tree particles positions at the middle of the pm step is taken to be the position one full particle time step ahead ; since tree particle time steps are half the pm time step or less , this is no more than the pm midstep . with these advanced positions , the potential on the grid
is found in the standard pm manner . for each tree
the pm potential in a cubical subvolume is saved .
this volume is slightly larger than the active cell region ( in order to allow for finding the gradient of the potential through finite differencing ) plus one extra cell on a side in case particles migrate out of the active cell region during the time integration .
since the pm time step is limited by a courant condition ( see [ sec : timestep ] ) such that pm particles can not move more than a fraction of a cell per step , one extra cell is sufficient particles near the edge of a tree region are at only a slightly higher overdensity than the densest pm regions , and hence have similar velocities . at the beginning of a tree step the portion of the pm potential due to the tree particles themselves is subtracted out , leaving the tidal potential . with @xmath21 saved in this manner , every time a particle s acceleration is updated , the tidal force is calculated from the grid in the same manner that pm forces are found .
this is an improvement over the method used in box .
[ [ section-3 ] ] 4 .
+ + integrate each tree region forward to the middle of the pm step .
this is done with a tree code , adding in the tidal forces .
the tree code we use was written by lars hernquist @xcite .
any other potential solver could be used , but a tree code is well suited for this type of problem , in that it can efficiently handle a wide variety of particle distributions , can include individual particle time steps , and scales as @xmath2log@xmath2 .
note that for each tree this portion of the code is self contained ; that is , once the particle data and potential mesh for a given tree have been collected together no further information is required to evolve that tree forward .
this makes tpm well suited for parallel processing on distributed memory systems .
once the tree data is received by a given processor this step which is the most computationally expensive part of the algorithm can be performed without any further communication . given this coarse grained parallelism ,
one can use widely distributed and heterogeneously configured processors , with more capable processors ( or groups of processors ) reserved for the largest trees .
[ [ section-4 ] ] 5 .
+ + find the density and potential on the pm grid again , and update the acceleration for pm particles ( eq . [ [ eqn : leapfrog]b ] ) .
all tree particles are at the pm midstep , the proper location for computing the force on pm particles .
[ [ section-5 ] ] 6 .
+ + integrate each tree region forward to the end of the pm step .
[ [ section-6 ] ] 7
. + + update pm particle velocities and positions to the end of the time step ( eqs .
[ [ eqn : leapfrog]c - d ] ) .
tpm exploits the fact that gravitational instability in an expanding universe creates spatially isolated high density peaks ; for typical cdm models these peaks will contain a significant fraction of the mass ( @xmath22 ) but occupy a tiny fraction of the volume ( @xmath23 ; see for example fig@xmath152 of @xcite ) . as described in detail in box , this is accomplished by tagging active mesh cells based on the cell density ( which has already been calculated in order to perform the pm step ) .
adjacent cells are linked together , dividing the cells into groups in a friends of friends manner .
thus isolated high density regions of space are identified , separated by at least one cell width . in box ,
the criterion used to decide if a cell @xmath24 was active was if its density @xmath25 exceeded a global threshold density @xmath26 where @xmath27 is the mean density and @xmath28 is the dispersion of cell densities . while temporally adaptive ( in the sense that the threshold is lower at early times when peaks are rare ) , this criterion has two main drawbacks .
first , small peaks in isolated regions or voids will not be picked up , especially when @xmath28 is large , because the total mass in such a halo would not put a pm cell above @xmath29 .
secondly , in the case of a region with peak density near @xmath29 , whether or not a given pm cell is above @xmath29 can depend on the offset of the grid with respect to the particles the halo mass may be divided into two or more cells .
we have developed an improved criterion for selecting tree regions involving the contrast of a cell with its surroundings ( rather than a single global threshold value ) found by comparing the density smoothed on two scales .
for each cell three densities are computed .
first @xmath30 , found by boxcar smoothing over a length of 5 cells along each dimension .
( another type of smoothing , _ e.g. _ with a gaussian , would be possible , but the boxcar can be done with minimal interprocessor communication ) .
second , @xmath31 , the mean density in the eight cell cube of which the cell under consideration is the lower octant ( when using cic , a particle in this cell would also assign mass to the other cells in this cube ) .
third , @xmath32 , a measure of the density surrounding this cube , defined as @xmath33 all eight cells used to find @xmath31 are marked as active if @xmath34 the reasons for such a choice of form can be seen by examining the behavior of eq .
[ [ eqn : rhocrit ] ] in various limits . at early times in a cosmological simulation @xmath35 and @xmath36 , so @xmath37 .
thus with a choice of @xmath38 those regions which are only slightly overdense will be selected . at late times
@xmath39 . in void regions @xmath40 , which means @xmath41 ; thus an isolated cell will be chosen if it is a fixed multiple above its surroundings .
if on the other hand @xmath42 , then @xmath43 , meaning that at each step there is a given density above which all cells will be chosen ; this limit will increase with time as @xmath28 increases .
keep in mind that the distribution of cell densities will be approximately lognormal ( @xcite and references therein ) when interpreting the value of @xmath28 . after cell selection ,
the domain decomposition proceeds in the manner described in box .
cells are linked by friends of friends , yielding regions of space separated by at least one cell length . any particle which contributes some mass to an active cell when finding the pm density
is assigned to the corresponding tree . because the volume of all tree regions is less than one percent of the total volume , the amount of memory allocated to hold the tidal potential data is generally not significant compared to that already needed for the pm mesh . however
, one complication of this scheme is that at early times , when @xmath28 is small , the resulting trees can be very long filaments . while not containing a large amount of matter , being not very overdense and nearly one dimensional
, the size of a cubical volume enclosing such a tree will sometimes be a significant fraction of the entire simulation volume .
this can cause difficulties because of the amount of computer memory required to compute and save the tidal potential in such subvolumes .
the value of @xmath44 in eq .
[ [ eqn : rhocrit ] ] is thus allowed to increase at early times when the spatial extent of trees tends to become larger , and decrease at later times when trees are more compact .
we have settled on the following method , which has worked well in a variety of simulations .
space is allocated for a maximum subvolume length one quarter of the pm mesh size .
the size of the largest sub - box used is checked at the end of each pm step . at early times ( when @xmath45 ) @xmath44 is increased by 0.5% if this box has a length more than a third of the allocated value .
once @xmath28 exceeds 1.6 in a typical simulation , the spatial extent of trees has stopped growing , so in this case @xmath44 is decreased by 0.25% if the largest sub - box has a length less than half the maximum allocated .
a minimum value is set for @xmath44 to prevent too many tree particles being chosen ; a choice of @xmath46 will place roughly half of the particles in trees at @xmath47 .
an example of this in practice is discussed in
[ sec : overv ] .
the number of pm grid cells can be more or less than the number of particles , though of course the finer the grid , the greater the accuracy of the pm and tidal forces . generally ,
the number of cells should at least equal the number of particles ; we have found that eight grid cells per particle works well .
a variety of methods for determining the particle time step have been proposed .
this diversity of criteria reflects the fact that the appropriate time step can depend on a number of quite different considerations .
these include the local density or the local dynamical time , nearby substructure , the nearest neighbor , and the softening length @xcite .
in addition , for any single criterion , one can imagine a special circumstance where it breaks down .
this suggests the possibility of basing @xmath48 on more than one criterion .
thus in the tpm code we have decided to combine a number of possible criteria , drawn from variables which do not require any significant computation to find , in the following manner : [ eqn : deltat ] @xmath49 here @xmath50 is an adjustable dimensionless parameter , @xmath51 is the spline kernel softening length , and @xmath52 is the distance to the nearest neighboring particle .
the latter is found as a byproduct of the force calculation ; if at the beginning of a tree step @xmath52 has not been calculated , one can as a proxy find the minimum @xmath52 that is consistent with the particle s current @xmath48 .
the time derivative of @xmath53 is approximated by @xmath54 .
this criterion for @xmath48 is quite conservative , and is similar in principle to that adopted by @xcite . a comparison of the time step returned by eq .
[ [ eqn : deltat ] ] and the often used criterion @xmath55 is given in fig@xmath15[fig : dtcomp ] , which shows how these two criteria compare at @xmath47 for particles in the largest tree of the simulation discussed in
[ sec : overv ] and [ sec : codecomp ] .
this tree is made up of over @xmath56 particles , and contains two large halos plus a number of smaller satellites and infalling matter . each contour level in fig@xmath15[fig : dtcomp ] encloses a tenth of the particles , and the remaining tenth are plotted as points .
it can be seen that eq .
[ [ eqn : deltat ] ] tends to yield a smaller timestep than @xmath55 ; this is the case for 74% of the particles .
generally the differences are not large less than a factor of two for 60% of the particles .
another trend seen in the figure is that as @xmath55 becomes smaller it is more likely for eq . [
[ eqn : deltat ] ] to give an even smaller value , and conversely eq .
[ [ eqn : deltat ] ] tends to give a longer time step than @xmath55 as the latter becomes larger . in most cases the value of @xmath48
is set by the factor @xmath57 .
in other words , the distance to the nearest neighbor is not allowed to change by a large factor , unless this distance is small compared to the softening length . even in cases
when this factor is larger than @xmath58 , it is still smaller than @xmath59 and @xmath60 ; since the equations being integrated have the form @xmath61 and @xmath62 this limits integration errors .
the pm time step is limited by a courant condition , such that no pm particle moves more than one quarter of a cell size per step .
also , @xmath7 is not allowed to change by more than 1% per step .
initially this latter criterion is the most restrictive , but over time it allows a longer @xmath12 and becomes unimportant .
the @xmath12 allowed by the courant condition also tends to increase over time , although more slowly .
this is because particles falling into dense regions are placed into trees , and the velocities of the remaining particles are redshifted ( eq . [ [ eqn : leapfrog]c ] ) .
the largest time step for any particle , @xmath12 , is kept constant until it can safely be increased by a factor of 2 , and it is then doubled .
the time step for tree particles is unchanged , which means they take twice as many steps per pm step as before ; during tree evolution particles will move into a lower time step bin if allowed by eq . [ [ eqn : deltat ] ] . whenever any particle changes time step , its position is updated as described in @xcite to preserve second order accuracy .
various aspects of a typical tpm run can be demonstrated with a standard cosmological simulation .
this test case contains @xmath2=128@xmath0 particles in a box @xmath63=40@xmath64mpc on a side .
the number of grid points for the pm and domain decomposition portions of the code is eight times the number of particles , or @xmath65 .
the initial conditions were generated using the publicly available codes grafic1 and lingers @xcite .
a spatially flat lcdm model was chosen , with cosmological parameters close to the concordance model of @xcite : @xmath66 , and @xmath67 .
the particle mass is thus @xmath68 . within the tree portion of the code , the opening parameter @xmath69 in the standard barnes
hut algorithm @xcite is set to @xmath70 , and the time step parameter @xmath50=0.3 ( see eq . [ [ eqn : deltat ] ] ) .
the cubic spline softening length was chosen to be 3.2@xmath64kpc so that the spatial dynamic range @xmath71 .
with such a small softening length , halos with fewer than 150 particles are likely to undergo some 2-body relaxation in their cores over a hubble time ( assuming the particle distribution follows an nfw profile with concentration @xmath72=12 ) .
fig@xmath15[fig : overv ] displays the evolution of several important quantities during this run as a function of expansion parameter @xmath7 .
the top panel shows the dispersion of pm cell densities @xmath28 ( where the mean cell density is unity ) , and the second panel shows the value of @xmath44 ; both of these factor in eq . [ [ eqn : rhocrit ] ] , which in turn plays a major part in determining the tree distribution .
various aspects of this distribution are shown in the remaining curves .
the third panel shows the size of the largest cubical subvolume required ( in units of pm grid cells ) .
initially small , this grows extremely rapidly as @xmath28 rises from its initial value of 0.3 to @xmath73 ; in response @xmath44 is increased .
when this size is at its greatest ( at @xmath74 ) , the percentages of total volume and mass in trees ( shown in the next two panels ) are still quite small .
the trees at this time tend to follow caustics they are only slightly overdense and not very massive , but because of their filamentary nature they can have a large spatial extent .
these caustics then fragment and collapse , so even while the total mass in trees and the number of trees ( shown in the third panel from the bottom ) increase the maximum subvolume size decreases .
the changes in @xmath44 affect mainly the maximum subvolume size ; the number of trees and volume contained in tree regions both increase at a steady rate up to @xmath75=1 .
the number of tree particles increases monotonically throughout the simulation .
after @xmath76 ( by far the bulk of the computational time ) , the characteristics of the simulation change much more slowly .
roughly half the particles are in @xmath772700 trees ranging from a few to @xmath78 particles but occupying only 0.4% of the simulation volume .
the penultimate panel shows the number of particles in the first and third largest trees ; by @xmath76 the growth in mass of these objects has slowed to a fairly constant , low rate .
trees are distributed in mass roughly as a power law , with 67% of the trees having fewer than 100 particles and 95% less than 1000 .
most tree particles take two or four steps per pm step , but some are taking up to 64 ; there were 1075 pm steps total in this run . as @xmath28 increases ,
so does the limiting density above which all cells are treated at full resolution .
it is important to keep track of this limit when interpreting the results of a tpm run .
the bottom panel shows the highest value of the density among those cells evolved only with pm . by @xmath76
this density is 120 times the mean , corresponding to 15 particles inside a cell . by the end of the computation , the densest pm only cell contains 26 particles ; so trying to make statements about objects smaller than this will be complicated by the varying spatial resolution of tpm .
note that most objects with @xmath79 will in fact be followed at full resolution substructures inside larger objects because they are in regions of higher density , and small isolated halos because they present a density contrast to their surroundings . to be cautious
we will limit our analysis to objects 50% larger than this limit , _ i.e. _ 40 particles or @xmath80 .
as seen in the previous section , during later epochs which take up most of the computational time needed for a run the mass distribution of trees is generally well fit by a power law ranging from a few particles up to of order 0.01@xmath2 .
the actual mass of the largest tree will depend on the ratio of the largest non - linear scale in the box to the box size ; as this ratio becomes larger so does the mass of the largest tree .
furthermore , the amount of computation required for a tree with @xmath81 particles will scale as @xmath81log@xmath81 .
thus , there can be a substantial variation in the computational time required between different trees , and evolving the largest tree can comprise a significant fraction of the total computational load .
an efficient parallel code must handle this situation well when dividing work up among ncpu processors .
load balancing of trees is achieved in three ways .
first , trees are sorted by @xmath81 and then divided into `` copses '' of roughly equal amounts of work using the `` greedy '' algorithm .
that is , starting with the largest tree , each one in turn is given to the copse which up to that point has been assigned the least amount of total work .
usually there is one copse per cpu , but there can be two or more per cpu if required by space constraints .
there is a communication step , when all data associated with a copse is sent to one cpu .
a cpu then evolves each tree in its local copse in turn , starting with the most massive .
additionally , the largest trees can be done in parallel .
if a few large trees dominate the work load , then it is impossible for all copses to contain equal amounts of work ideally , one would want ncpu copses , each comprising 1/ncpu of the total amount of work , but this clearly ca nt happen if the largest tree takes a larger fraction just by itself . in this case , the force calculation for these large trees is done in parallel by a small number of cpus .
this is currently done quite crudely , with only the tree walk and force calculation actually carried out in parallel ; in theory a fully parallel tree code could be used for every tree , allocating more processors to those copses containing the most work . only a few nodes are usually required to reduce the time spent on the largest tree to the level required for load balancing . as a final means of balancing the load ,
when a cpu finishes evolving all the trees in its local copse , it then sends a signal to the other cpus .
a cpu with work still left will send an unevolved tree to the idle cpu for it to carry out the evolution .
the scaling of the current implementation of tpm with number of processors is shown in fig@xmath15[fig : scale ] .
the test case for these timing runs is a standard lcdm model at redshift @xmath75=0.16 with @xmath2=256@xmath0 particles in a 320@xmath64mpc cube .
this particular set of runs was carried out on an ia-64 linux cluster at ncsa named `` titan '' .
this machine consists of 128 nodes , each with dual intel 800mhz itanium processors , and a myrinet network interconnect . while the total time required depends on processor performance , similar scaling with ncpu has been found on a number of machines with various types of processors and interconnects .
the topmost line in fig@xmath15[fig : scale ] is total time , calculated as the number of seconds wallclock time per pm step multiplied by ncpu ; perfect scaling would be a horizontal line .
tpm performs quite well at ncpu=64 the efficiency is still 91% as compared to ncpu=4 . beyond this point scaling begins to degrade , with the efficiency dropping to 75% for ncpu=128 .
the reason tpm scales well can be seen in the second line from the top , which shows the total time spent in tree evolution .
this part of the code takes up most of the cpu time , but it requires no communication and there are enough trees so that just coarse grained parallelization works reasonably well .
the next two curves shown indicate the amount of time in the pm portion of the code and overhead related to trees ( identifying tree regions and particles , etc . ) . these take a small fraction of the total time and scale well since the grid is distributed across all processors .
the final curve in fig@xmath15[fig : scale ] shows the time related to communication and imbalance in the tree part of the code .
as ncpu increases it becomes more difficult to divide the tree work evenly and processors spend more time either recruiting work from others or waiting for them to finish .
this overhead could likely be reduced by incorporating a fully parallel tree code and also by computing the nonperiodic fft ( required to find the tidal potential see box ) in parallel .
a run with a larger number of particles and grid points , and thus a larger number of trees , would show efficient scaling beyond ncpu=128 .
as a test of new code we have carried out a simulation of the secondary infall and accretion onto an initially uniform overdensity . this can be compared both to other codes and to the analytic , self - similar solution of @xcite . to create the initial conditions ,
@xmath82 particles were placed on a uniform grid with zero velocity .
the eight particles on the grid corners were then removed and placed at the center of the volume with a spacing one half that of the regular grid .
thus , this is actually more of a cubic overdensity than spherical , but it quickly collapses and the subsequent infall is independent of the details of the initial state . also , there is a void located half the box distance away from the overdensity , but the evolution is not carried out for a long enough time for this to be significant .
the initial condition is integrated from expansion factor @xmath83 to @xmath84 .
the values @xmath38 and @xmath46 were used for the constants in eq .
[ [ eqn : rhocrit ] ] . since
only one halo is forming in the box , @xmath28 remains small , rising to only 0.5 by the end of the run . at the end , there are roughly 500 particles within the turnaround radius .
this test can be seen as exploring how well the initial collapse of small objects is followed , which is the beginning stage of halo formation in an hierarchical scenario .
the final state of this run is shown in fig@xmath15[fig : sphod ] .
the top panel shows density as a function of radius .
filled points are the tpm model ; error bars are the square root of the number of particles in a given radial bin .
the inner edge of the innermost bin is twice the softening length .
also shown is the same run carried out with a p@xmath0 m code ( as open circles ; details of this code are discussed in [ sec : codecomp ] ) , and the solution of @xcite .
the agreement is quite good the main limitation of this run is likely the mass resolution . as a measure of the phase space density
, the lower panel shows @xmath85 , where @xmath86 is the velocity dispersion of the particles in each radial bin .
following @xcite , we compare this to the @xcite solution for a @xmath87 gas . again
the agreement is quite reasonable .
differences between tpm and p@xmath0 m arise from different types of softening ( p@xmath0 m uses plummer softening ; here @xmath51 was set to half the tpm value ) and from time stepping ( for p@xmath0 m , @xmath48 for all particles was set by the minimum @xmath88 ) . as a test of tpm in a less idealized situation
, the initial conditions described in
[ sec : overv ] were again evolved , but using two other n - body codes .
the first is the p@xmath0 m code of @xcite in a version made parallel by @xcite this code was also used in [ sec : sphodtest ] . as in the tpm run ,
a mesh of @xmath65 grid points was used .
this code uses plummer softening ; the softening length was set to half the value used in the tpm run .
the time variable in this code is @xmath89 , defined by @xmath90 ; all particles have the same time step , set by the minimum @xmath88 .
the other n - body code used is the tree code named gadget of @xcite . in this code ,
periodic boundary conditions are handled with ewald summation , the time variable is the expansion factor @xmath7 , and each particle has an individual time step . following @xcite , the conservative tree node opening criterion flagged by -dbmax was used , and the time step was set by @xmath91 .
in addition , the maximum allowed time step was set to 1% of the initial expansion factor ; whenever @xmath7 doubled this maximum was also increased by a factor of two , by checkpointing the run and restarting with the new value .
the softening length was set to be the same as the tpm run ( allowing for different notational conventions ) . as a first comparison of the codes
, the two point correlation function @xmath92 , found by counting the number of particle pairs in bins of separation @xmath52 , was calculated . to compute @xmath92 ,
51 logarithmically spaced bins were used , with the minimum pair separation considered being @xmath93kpc and the maximum one third of the box size .
the results at various redshifts are shown for all three runs in fig@xmath15[fig : cfcmp ] .
clearly there is little difference to be seen in the three codes . at larger scales
this is to be expected ; at smaller scales force approximations , smoothing , and relaxation may become important .
the fact that p@xmath0 m uses plummer rather than spline softening explains why it shows a lower @xmath92 at scales less than a few times the softening length .
if high values of @xmath92 provide a measure of accuracy , the tpm and gadget are slightly more accurate than p@xmath0 m .
the statistics of mass peaks the dark halos surrounding galaxies , clusters , and so on are an important product of n - body codes @xcite .
one commonly used halo finder is friends of friends , or fof @xcite .
the cumulative mass function , found with fof using a linking length of 0.2 times the mean interparticle separation , is shown in fig@xmath15[fig : fofcmp ] . at higher redshifts
there is little discernible difference between the three codes . at later times
, tpm seems to have fewer halos containing @xmath9450 particles .
it is unclear how to interpret this , because halos with less than 150 particles ( i.e. mass @xmath95 ) are affected by two - body relaxation .
@xcite found that an adaptive p@xmath0 m and gadget both produced more small fof halos than the adaptive multigrid mlapm and art codes , so in this case tpm may agree more closely with the latter . to investigate
further a different halo finder was used , namely the bound density maxima ( bdm ) algorithm @xcite .
this algorithm is significantly different from fof , so the resulting mass function is probing different qualities of the dark matter distribution
. density maxima inside spheres of radius 100 @xmath64kpc were found , and halo centers were then calculated using spheres of radius 40 @xmath64kpc .
the mass for each halo is taken to be that inside a sphere containing the overdensity expected for a virialized halo just collapsed from a spherical top
hat ; to calculate this overdensity the fit of @xcite was used . in bdm ( unlike fof )
particles moving faster than the escape velocity are removed from the halo .
the bdm cumulative mass functions for the three n - body codes are shown in fig@xmath15[fig : mfcmp ] .
as with fof , the agreement between codes is quite good . in this case it appears to be tpm which produces more small halos , but again this effect is for low mass halos likely affected by relaxation . using the bdm halos with more than 40 particles , the halo
halo correlation function is shown in fig@xmath15[fig : cfhal ] .
the number of pairs of halos in 21 logarithmically spaced radial bins , with the smallest separation being 200 @xmath64kpc and the largest one third of the box size , were tallied and compared with the expectation for a random distribution .
all three codes give the same result . from the results presented so far in this section
it is clear that tpm yields a quite similar evolved matter distribution as compared to other codes .
finally we turn to the internal properties of halos .
one probe of the mass distribution in a halo is the circular velocity @xmath96 . for each simulation
we divide halos up into six mass bins . to avoid relaxation effects ,
the lowest mass considered is @xmath97 .
the width of each bin is a factor of @xmath98 , so the largest mass bin contains two halos with mass above @xmath99 . for each set of halos
the average circular velocity as a function of radius was calculated ; these average velocity curves are shown in fig@xmath15[fig : vcirc ] .
it can be seen that these curves are quite similar for all three codes .
two main trends are noticeable , the main one being that p@xmath0 m shows lower circular velocities at small radii .
this is likely due to the use of plummer softening and less strict time step criterion in the p@xmath0 m run , both of which would lead to lower densities in the innermost parts of halos .
the other main difference is that for the lowest masses it appears that tpm yields higher circular velocities than the other two codes . instead of averaging over a number of halos of similar mass , it is also possible to compare halos on a one - by - one basis .
to find the appropriate pairs of halos , the list of halos with more than 150 particles found by bdm was sorted by mass , and for each target gadget or p@xmath0
m halo a tpm halo was selected as a match if its mass was within 25% and its position within 1 @xmath64mpc of the target halo ; if more than one halo passed this test then the nearest in position was selected .
the selected tpm halo was removed from further consideration , and the process repeated for next target halo . in this manner a match was found for 588 out of 609 gadget halos and 573 out of 582 p@xmath0 m halos . for various measured halo properties , the percentage difference of the tpm halo from the gadget or p@xmath0 m halo was calculated . since the dispersion of these differences increases as less massive halos are considered , halo pairs
are split into two groups , with the higher mass group containing all target halos with 394 or more particles ( i.e. mass @xmath100 ) .
results are shown in table [ tab : hdiff ] , which for each property gives the mean percentage difference and one standard deviation , as well as the first quartile , median percentage difference , and third quartile .
the first property shown is halo mass @xmath101 .
the difference here is constrained to be less than 25% , but in most cases the tpm value is within 10% of the other code s .
it appears that p@xmath0 m yields slightly lower masses than the other two codes , but the latter two agree quite well .
the agreement in masses shows that the simple scheme used to find matching halo pairs works well , as does the fact that the difference in position is less than 165@xmath64kpc in 95% of the cases .
this is reinforced by the agreement between the halo center of mass velocities @xmath102 shown in table [ tab : hdiff ] ; the velocity vectors are closely aligned , differing by less than @xmath103 in 95% of the pairs .
thus only a few percent of the halo pairs are mismatches ; since these pairs still have similar masses and are likely in similar environments , they are kept in the comparisons . in terms of the 3-d root mean square velocity @xmath104 and maximum circular velocity @xmath105 , the more massive halos are very similar in all three codes , with no offset between codes . however , at the lower mass end tpm gives values that tend to be systematically higher by a few percent
this can also be seen for the average circular velocity in the bottom curve of fig@xmath15[fig : vcirc ] ; this is curve is the average for halos below roughly @xmath106 .
the difference between tpm and the other two codes is most pronounced near halo centers .
the last property shown in table [ tab : hdiff ] is a comparison of central density @xmath107 , which is computed by measuring the amount of mass within @xmath108=12.8@xmath64kpc of the halo center . as would be expected from fig@xmath15[fig : vcirc ] , tpm yields higher central densities than gadget , with p@xmath0 m giving lower @xmath107 than either of these .
one possible source of the differences seen in tpm halos as compared to a pure tree code could simply be the choice of time step .
the p@xmath0 m run took @xmath109 steps and the gadget run @xmath110 ; tpm on the other hand took @xmath111 , or 60% more than gadget .
of course this comparison is not entirely straightforward because different particles determine the smallest time step at different times .
part of the difference between gadget and tpm may be due to the fact that tpm bins time steps by factors of two , whereas gadget allows time steps to vary more gradually .
a second tpm run was carried out to separate other code differences from the time step criterion .
this run was identical to the first except that the time step was set not by eq .
[ [ eqn : deltat ] ] , but rather by @xmath112 ( similar to the gadget and p@xmath0 m codes ) ; as a result it took fewer steps than any of the other runs .
there is no significant difference in the mass functions and halo halo correlation functions between this run and the original tpm run .
the particle - particle correlation function is different , however .
this can be seen in fig@xmath15[fig : xicmp ] , which shows the ratio of the original tpm run s @xmath92 to that of the new run , at redshift @xmath75=0 .
the new tpm run has a lower @xmath92 for @xmath113kpc , roughly 5 - 10% lower than the original tpm and similar to the p@xmath0 m run .
this indicates the internal structure of halos has been affected by the longer time steps , which is confirmed by repeating the comparison of individual halos .
the difference between halos in the new tpm run and the gadget run are given in table [ tab : hdiff ] .
for the higher mass halos the new run , unlike the original one , tends to give lower maximum circular velocities and central densities than the tree code .
this indicates that the longer time step is in fact leading to inaccuracies .
the differences between tpm and the tree code for low mass halos seen in the original run persist in the new run , though they are not quite as pronounced .
another point of comparison between codes is efficiency with which they use computing resources .
fig@xmath15[fig : times ] shows the wall - clock time consumed by each of the codes , as a function of expansion parameter , in carrying out the test simulation .
all the runs used four 300 mhz ip27 processors of an sgi origin 2000 ; the codes were compiled with the sgi compilers and mpi library .
all three codes used roughly the same amount of memory ( tpm requires at least 20 reals per particle plus 3 reals per mesh point , divided evenly among processors ) . at the earliest times both p@xmath0 m and tpm spend most of their time in the pm fft , and so they behave similarly . however , as objects begin to collapse p@xmath0 m begins to consume more time computing particle - particle interactions .
the fact that the accelerations of all particles are updated every step also makes p@xmath0 m use more time than do the two multiple time step codes .
the tree code takes more time when the particle distribution is nearly homogeneous , demonstrating that a pm code ( which is what the gridded codes basically are in this situation ) is very efficient .
however , the tree code timing is roughly independent of the particle distribution , and once inhomogeneity develops it does not require more time per update , whereas the other codes do .
tpm does well compared to the tree code for a couple of reasons .
by @xmath7=1 roughly half the particles in the tpm run are still being handled solely by pm .
also , imagine breaking @xmath2 particles up into @xmath17 trees each with @xmath114 particles ; the time required to solve all the trees will scale as @xmath115 , lower by a logarithmic factor than using the same tree solver on all the particles .
overall , despite taking more timesteps , in this test case tpm required less cpu time than the other codes , by a factor of @xmath116 relative to the tree code and @xmath117 relative to p@xmath0 m .
this paper has presented a parallel implementation of the tpm algorithm .
several improvements over the implementation of box have been made .
particles in tree regions each have an individual time step , half of the pm time step or less , making the tree integration more efficient .
the treatment of tidal forces on trees is also improved by saving the tidal potential on a grid and evaluating the force as a function of particle position at each smaller particle time step ; thus a greater ratio of tree to pm time steps is allowed . a new , more stringent , time step criterion has been implemented .
a new criterion for locating regions for treatment with higher resolution is given ; by finding cells with higher density than their surroundings , small halos in lower density regions are located and followed at full resolution .
these changes significantly increase the speed and accuracy of tpm : a computation of the test case discussed in [ sec : codecomp ] using the code described in box required more cpu time ( by a factor of over three ) than the current version , but was only accurate for halos with more than 315 particles .
comparisons with other widely used algorithms were made for a typical cosmological structure formation simulation .
these show excellent agreement .
the particle - particle correlation functions , the halo mass functions , and the halo - halo correlation functions from tpm agree quite well with those from p@xmath0 m and tree codes .
the internal properties of halos also agree ; the main difference being that , for lower mass halos , tpm yields higher @xmath118 and maximum circular velocities ( by a few percent over a tree code ) .
tpm halos also show higher central densities than those of the other two codes , though the mean difference is smaller than the dispersion .
this difference disappeared , at least among the more massive halos , when the time step criterion of eq .
[ [ eqn : deltat ] ] was replaced with one similar to that employed by the other two codes .
thus we conclude that a choice of a relatively conservative time step criterion contributed to a slightly improved accuracy .
tpm yielded results of similar accuracy to the other codes used here while using significantly less computational time ( about a quarter of that needed by a tree code and an eighth of that needed by p@xmath0 m ) .
it also scales well on distributed memory parallel machines , such as networked pcs , because this parallelism is built in as part of the design of tpm .
however , in committing to an algorithm which accentuates the coarse - grained parallelism inherent in a typical cosmological simulation , a large degree of flexibility is sacrificed .
a basic presumption of tpm is that the largest nonlinear structure inside the simulation box is a small fraction of the total mass and volume . to simulate a situation where this is not the case ( e.g. two colliding galaxies ) another code would be preferred .
the tpm source code can be obtained at http://astro.princeton.edu/$\sim$bode/tpm or by contacting the authors .
the code is written in fortran 77 and uses mpi for message passing ; thus it is very portable and can be used on clustered pc s or other distributed memory systems .
many thanks are due to lars hernquist for generously supplying a copy of his tree code ; also edmund bertschinger for use of his p@xmath0 m code , joe henawi for help with fof , and scott tremaine for useful discussions .
this research was supported by the national computational science alliance under nsf cooperative agreement asc97 - 40300 , paci subaward 766 .
computer time was provided by ncsa and the pittsburgh supercomputing center .
aarseth , s. 1999 , , 111 , 1333 baertschiger , t. , joyce , m. & labini , f.s .
2002 , , 581 , 63 ( astro - ph/0203087 ) bagla , j.s . 1999 ,
preprint ( astro - ph/9911025 ) barnes , j.e .
1998 , galaxies : interactions and induced star formation , r.c .
kennicutt jr .
, f. schweizer and j.e .
barnes , berlin : springer , 275 barnes , j. & hut , p. 1986 , nature , 324 , 446 becciani , u. & antonuccio - delogu , v. 2001 , comp .
comm . , 136 , 54 bertschinger , e. 1985 , , 58 , 39 bertschinger , e. 1998 , , 36 , 599 bertschinger , e. 2001 , , 137 , 1 bryan , g.l . & norman , m.l .
1998 , , 495 , 80 bode , p. , ostriker , j.p .
, & xu , g. 2000 , , 128 , 561 ( box ) calder , a.c .
2002 , , in press couchman , h.m.p . ,
thomas , p.a . & pearce , f.r .
1995 , , 452 , 797 davis , m. , efstathiou g. , frenk , c. & white , s.d.m .
1985 , , 292 , 371 dehnen , w. 2000 , , 536 , l39 dehnen , w. 2002 , j. comp .
, 179 , 27 dorband , e.n . , hemsendorf , m. & merritt , d. 2002 , j. comp .
, in press ( astro - ph/0112092 ) efstathiou g. , davis , m. , frenk , c. & white , s. 1985 , , 57 , 241 ferrell , r. & bertschinger , e. 1994 , int .
c , 5 , 933 frederic , j.j .
1997 , ph.d .
thesis , mit hamana , t. , yoshida , n. & suto , y. 2002 , , 568 , 455 hernquist , l. 1987 , , 64 , 715 hernquist , l. 1990 , j. comp .
phys . , 87 , 137 hernquist , l. & katz , n. 1989 , , 70 , 419 hockney , r.w . & eastwood , j.w .
1981 , computer simulation using particles , new york : mcgraw hill kayo , i. , taruya , a. & suto , y. 2001 , , 561 , 22 jing , y.p .
& suto , y. 2002 , , 574 , in press klypin , a. 2000 , preprint ( astro - ph/0005502 ) klypin , a. , gottlber , s. , kravtsov , a.v . & khokhlov , a.m. 1999 , , 516 , 530 klypin , a. , & holtzman , j. 1997 , preprint ( astro - ph/9712217 ) knebe , a. 2002 , , submitted ( astro - ph/0201490 ) knebe , a. , green , a. & binney , j. 2001 , , 325 , 845 knebe , a. , kravtsov , a.v .
, gottlber , s. & klypin , a.a .
2000 , , 317 , 630 kravtsov , a.v .
, klypin , a.a . & khokhlov a.m. 1997 , , 111 , 73 lia , c. & carraro , g. 2001 , , 276 , 1049 ma , c .- p . , & bertschinger , e. 1995 , , 455 , 7 miocchi , p. & capuzzo - dolcetta , r. 2002 , , 382 , 758 mo , h.j . & white , s.d.m .
2002 , , 336 , 112 ( astro - ph/0202393 ) ostriker , j.p . & steinhardt , p.j .
1995 , nature , 377 , 600 power , c. , navarro , j.f . ,
jenkins , a. , frenk , c.s . , white , s.d.m .
, springel , v. stadel , j. & quinn , t. 2002 , , submitted ( astro - ph/0201544 ) ricker , p.m. , dodelson , s. & lamb , d.q .
2000 , , 536 , 122 springel , v. , yoshida , n. & white , s.d.m .
2001 , new astronomy , 6 , 79 stadel , j.g .
2002 , ph.d .
thesis , university of washington , seattle taylor , j.e . &
navarro , j.f .
2001 , , 563 , 483 teuben , p. 1995
, astronomical data analysis software and systems iv , r.a .
shaw , h.e .
payne & j.j.e .
hayes , san francisco : astronomical society of the pacific , 398 van kampen , e. 2000 , , submitted ( astro - ph/0002027 ) viturro , h.r .
& carpintero , d.d .
2000 , a&as , 142 , 157 xu , g. 1995 , , 98 , 355 white , m. 2002 , , in press yahagi , h. , mori , m. & yoshii , y. 1999 , , 124 , 1 yahagi , h. & yoshii , y. 2001 , , 558 , 463 lccccccccccc & & & + & mean & s.d . & @xmath119 & @xmath120 & @xmath121 & & mean & s.d . & @xmath119 & @xmath120 & @xmath121 + + @xmath101 & -0.03 & 8.21 & -5.07 & 0.08 & 4.80 & & -0.37 & 5.33 & -2.33 & -0.12 & 2.08 + @xmath102 & 0.06 & 9.38 & -1.08 & 0.52 & 1.74 & & 0.49 & 4.28 & -0.54 & 0.59 & 1.74 + @xmath104 & 1.59 & 7.94 & -2.20 & 1.99 & 5.97 & & -0.05 & 4.38 & -2.12 & 0.18 & 2.23 + @xmath105 & 2.69 & 6.98 & -1.14 & 2.89 & 6.68 & & 0.35 & 3.85 & -1.48 & 0.25 & 2.01 + @xmath107 & 16.2 & 38.1 & -2.22 & 8.33 & 26.0 & & 4.03 & 22.1 & -8.82 & 2.45 & 13.7 + + @xmath101 & 1.84 & 8.65 & -3.40 & 1.75 & 6.74 & & 0.38 & 5.14 & -1.51 & 0.39 & 2.37 + @xmath102 & 0.22 & 7.44 & -1.19 & -0.10 & 1.20 & & -0.38 & 4.26 & -0.93 & -0.07 & 0.84 + @xmath104 & 3.84 & 7.13 & -0.80 & 3.59 & 8.94 & & 0.23 & 3.82 & -1.83 & -0.03 & 2.16 + @xmath105 & 6.20 & 8.93 & 1.02 & 5.11 & 10.7 & & 0.32 & 4.06 & -1.94 & -0.24 & 2.00 + @xmath107 & 29.4 & 39.2 & 3.45 & 20.0 & 47.2 & & 13.7 & 35.0 & -2.46 & 6.82 & 19.0 + + @xmath101 & -0.25 & 8.11 & -4.70 & -0.08 & 4.26 & & -0.63 & 5.82 & -2.73 & -0.33 & 2.18 + @xmath102 & -0.01 & 8.19 & -0.97 & 0.34 & 1.65 & & 0.76 & 7.10 & -0.68 & 0.45 & 1.76 + @xmath104 & 1.55 & 7.83 & -3.07 & 1.99 & 5.36 & & -1.00 & 3.86 & -2.65 & -0.62 & 1.28 + @xmath105 & 1.74 & 6.80 & -2.61 & 1.50 & 5.69 & & -0.77 & 3.85 & -2.44 & -0.49 & 1.03 + @xmath107 & 12.7 & 30.5 & -6.24 & 6.89 & 26.7 & & -0.75 & 24.3 & -12.9 & -4.16 & 6.49 + | an improved implementation of an n - body code for simulating collisionless cosmological dynamics is presented .
tpm ( tree particle mesh ) combines the pm method on large scales with a tree code to handle particle - particle interactions at small separations .
after the global pm forces are calculated , spatially distinct regions above a given density contrast are located ; the tree code calculates the gravitational interactions inside these denser objects at higher spatial and temporal resolution .
the new implementation includes individual particle time steps within trees , an improved treatment of tidal forces on trees , new criteria for higher force resolution and choice of time step , and parallel treatment of large trees .
tpm is compared to p@xmath0 m and a tree code ( gadget ) and is found to give equivalent results in significantly less time .
the implementation is highly portable ( requiring a fortran compiler and mpi ) and efficient on parallel machines
. the source code can be found at http://astro.princeton.edu/$\sim$bode/tpm . |
the discovery of ultra - high energy cosmic neutrinos @xcite opens a unique window to test neutrino properties at very high energies @xcite . in these experiments
the lepton flavor composition of the neutrino flux is essential for understanding mechanisms of generation , propagation and interactions of the neutrinos .
there are many running experiments such as anita @xcite , ara @xcite , antares @xcite and icecube @xcite , as well as the next generation deep water neutrino telescopes km3net
@xcite and nt1000 @xcite ( lake baikal ) which will provide results for cosmic neutrinos . of primary interest is the flavor content of the cosmic flux , because even though the initial mechanisms produce mostly muon and electron neutrinos , oscillations over cosmological distances generate a substantial @xmath0 component .
the detection of @xmath0 s in the telescopes has attracted special interest @xcite and the various event topologies have been classified .
an analysis of the icecube data using the double pulse algorithm did not establish any @xmath0 component at energies 214 tev to 72 pev @xcite . a second possibility to identify tau neutrinos uses the decay mode @xmath9 and separates the events into showers and tracks @xcite . in this approach
the observable quantities are @xmath10 and @xmath11 .
the icecube collaboration classified their data into showers and tracks and uses a likelihood function in order to obtain the best fit of the events as a function of the sum @xmath12
they combined the results with the astronomical flux and deduced its flavor content with a large uncertainty .
a more detailed analysis is possible when the energies of the muons , @xmath10 , and of the hadronic showers , @xmath11 , are measured separately in the same event .
the distributions in the two variables are sensitive to the @xmath0 component .
the purpose of this article is to provide explicit distributions in ratios of two variables and point out quantities which are less dependent on the shape of the incident fluxes and on the quark distribution functions .
reactions and differ because the values for the energy @xmath10 are lower for reaction and the two reactions have different angular distributions .
our results are computed analytically and it is possible to adopt them to the experimental situations .
we wish to distinguish the two cc reactions : @xmath13 @xmath14 & \\ & & \hskip1.cm\mu+\nu_{\tau}+\bar\nu_{\mu},}\label{eq : rectau}\ ] ] where @xmath15 denotes the hadronic final state . the hadrons will initiate showers in the detector and the muons will produce single tracks , as illustrated schematically in fig .
[ fig1 ] . in both cases
the energy distributions in the hadronic showers are the same , but the @xmath10 distributions differ because of the intervening @xmath16-decay .
the leptonic @xmath16-decay produces two invisible neutrinos .
it remains to quantify how much lower is the muon energy in the second reaction .
for this purpose we study distributions in the two dimensionless variables @xmath17 with @xmath18 denoting @xmath19 or @xmath16 leptons and the visible inelasticity , @xmath20 in this investigation the neutrino energies are very high ; they range above 100 gev to many 10.0 tev , where the momentum transfer squared , @xmath21 , is small relative to @xmath22 and we use the four - point interaction . in addition at this energy range
the cc cross section behaves as @xmath23 which follows from bjorken scaling . in the following we use the parton model cross sections where both @xmath24 and @xmath25 processes are calculated with the same quark distribution functions .
we shall show later that many results are independent of the quark distribution functions and also independent of the neutrino / antineutrino fluxes . in the above approach the differential cross section for muon neutrinos is @xmath26 , \label{eq : sigmup5}\ ] ] where @xmath27 are the integrals over quark or antiquark distribution functions .
reaction is a two step process .
first the @xmath16-lepton is produced and then it decays .
the cross section differential on the muon energy is a convolution over the energies of the intermediate @xmath16-lepton @xmath28 where @xmath29 is the cross section for reaction differential to the @xmath16-lepton energy , @xmath30 is the total decay width of the tau , @xmath31 is the differential decay width of the tau with respect to the muon energy in the reference frame where @xmath16 moves with energy @xmath32 .
this two step process is analogous to the parton model calculations @xcite .
the initial @xmath0nucleon interaction is the source of a beam of @xmath16-leptons with a known `` lepton distribution function '' .
it is followed by the known fragmentation of the @xmath16 s into muons .
the result of a straightforward calculation is @xmath33\ln ( 1+r_\mu ) \right\}\label{eq : distrmm},\end{aligned}\ ] ] where br is the branching ratio for the decay @xmath34 .
equations and are for neutrino - induced reactions ; for antineutrinos we interchange the quark and antiquarks distribution functions @xmath35 .
we also notice that @xmath36 and @xmath37 are independent variables with the neutrino energy appearing as an overall factor .
thus when we multiply the cross sections with the flux factors the integrals over neutrino fluxes factorize .
let @xmath38 be the flux factor of flavor @xmath18 measured in units @xmath39 .
the fluxes of the extraterrestrial neutrinos and antineutrinos are expected to be same : @xmath40 they may be different for various flavors and we present later figures for various compositions
. then the yield averaged over the fluxes is @xmath41de_{\nu},\label{eq : events3}\end{aligned}\ ] ] where @xmath42 is the time of exposure , @xmath43 is the solid angle of coverage , @xmath44 is the effective number of nucleons in the target , @xmath45 and @xmath46 define the range of neutrino energies investigated . for illustration
we consider the case where the @xmath8 and @xmath0 fluxes are equal .
we sum over neutrino and antineutrino interactions and normalize to the total number of events .
note that a molecule of ice and/or water neutrino detectors consists of 10 protons and 8 neutrons being close to isoscalar , and we do not treat the protons and neutrons separately @xcite .
the final result reads @xmath47.\label{eq : events34}\end{aligned}\ ] ] it is remarkable that the @xmath36-distribution depends neither on the incoming ( anti)neutrino energy nor on the quark distribution functions .
the formalism presented above can be easily adopted to cases where the flavor compositions of the neutrino flux are different .
figure [ fig3 ] shows the @xmath36-distributions for the @xmath48 and @xmath49 compositions . for the curves and the rest of the article we use @xmath50 @xcite . as seen in fig .
[ fig3 ] , the @xmath36-distribution is not dramatically affected by the presence of equal fraction of tau ( anti)neutrinos in the flux . there is only @xmath51 change at relatively large values of @xmath36 . at the same time , in this region the relative number of expected neutrino events drops down by orders of magnitude and it would be difficult to resolve experimentally the difference in the @xmath36-distributions .
a reason for such a behavior is that the energy spectrum of muons coming from taus decaying in flight has no longer the distinct peak at @xmath52 as in the tau rest frame , but becomes a relatively smooth function of @xmath10 without peculiarities ( see fig .
[ fig2 ] ) .
the tau neutrinos may more clearly manifest themselves if we concentrate all events in an interval which does not stretch as @xmath36 from zero to infinity but is much smaller .
this may be achieved , for example , by using the visible inelasticity @xmath5 which varies from zero to unity .
proceeding in the same way as we described above , one can calculate the @xmath5-distribution for the case @xmath49 by adding the events originating from @xmath53 and @xmath54 interactions and obtain @xmath55.\label{eq : events34f}\end{aligned}\ ] ] for other flavor compositions of the neutrino flux we can also calculate the visible inelasticity distribution .
we computed the cases @xmath48 , @xmath49 and @xmath56 shown in fig .
one can see that in the case of equal fluxes of muon and tau neutrinos there is a dramatic increase in the number of events as @xmath7 .
we also present the situation of a purely @xmath57 flux which is an idealized model of the so - called upward - going neutrinos .
this occurs when the neutrinos travel through the earth and the @xmath58 and @xmath53-components are absorbed thus only tau neutrinos survive at the detector .
details of the calculations are given in reference @xcite .
the fact that many models predict a substantial component of tau neutrinos in the extraterrestrial neutrino flux raises the challenge of obtaining direct or indirect evidence for their presence . the large volume neutrino telescopes may have the capability to provide such evidence , but this will require considerable experimental effort .
the challenge is to determine which distributions , accessible to experiments , are sensitive to this effect .
the analysis of this work concerns only those neutrino events which contain simultaneously a hadronic shower and a muon illustrated in fig .
we computed the corresponding event distributions for reactions and as functions of the dimensionless variables @xmath3 and @xmath4 .
we adopted the quark - parton model and the four - fermion interaction ( no @xmath59-propagator ) working well for @xmath60 tev .
however we expect our results not to change significantly at higher neutrino energies , since we consider ratios of the cross sections in which case the @xmath59-propagator corrections cancel out to an extent .
we emphasize an advantage of the approximations adopted in this article that the quark distribution functions factorize and drop out in ratios as well .
the same is true for the neutrino flux which makes possible to collect events at lower energies .
we find that more @xmath0-induced events populate the region @xmath61 as shown in fig .
for example , the presence of equal fraction of tau and muon neutrinos in the flux produces events with muons which have a minimum at @xmath6 and an excess of events as @xmath7 ; in contrast the muon neutrino induced events are decreasing monotonically as @xmath7 . for the @xmath49 flavor composition ,
the ratio of the number of events above @xmath62 to that below is 0.55 , while for muon neutrinos only this ratio equals to 0.46 . thus there is a difference between the ratios of the order of @xmath63 .
it will therefore be necessary and efficient to rely on the difference in the shapes of the two curves .
in other words , this @xmath0-component should lead to an increase of the number of slow or stopping muons in the detector in comparison to the case of a purely @xmath8-flux . to make a significant distinction between these curves we estimated that @xmath64 neutrino events with final muons are needed . assuming the best fit for the diffuse astrophysical neutrino flux @xcite , this number of events will be collected in a few years in the icecube and other large volume neutrino telescopes . in addition the region @xmath65 ( @xmath66 ) must be excluded from the analysis because the background from cc interactions of @xmath58 s and @xmath0 s and nc interactions is very large .
nevertheless , one can approach this point as close as it is allowed by the detector muon energy threshold .
likewise , the case @xmath67 with no visible cascade can be imitated by atmospheric muons penetrating into the detector . the contribution from the atmospheric muons and neutrinos
can be rejected by defining an outer veto volume in the detector and by observing extensive air showers on coincidence with neutrino events .
a full detector simulation including these backgrounds with monte carlo generators can describe in detail the behavior of the proposed distributions for each experiment .
one of us ( i.a . ) acknowledges daad support through the funding program _
research stays for university academics and scientists _ and hospitality at tu dortmund where this work was initiated .
i.a . also thanks h. ps for an invitation to attend weekly meetings of his research group .
i.a . was supported in part by the program for basic research of the presidium of the russian academy of sciences _ fundamental properties of matter and astrophysics_. * fig . 1 : * schematic illustration of neutrino events with muons in the final state : ( a ) charged current muon neutrino scattering ; ( b ) charged current tau neutrino scattering with the subsequent decay @xmath68 . the final hadron state ( shower )
is denoted by @xmath15 .
* fig . 4 : * production spectra as a function of the visible inelasticity ( @xmath71 ) .
we show three flavor compositions of the neutrino flux : @xmath48 ( dashed ) , @xmath49 ( solid ) and @xmath56 ( dash - dotted ) . | we investigate the problem of identifying @xmath0-induced reactions in large neutrino telescopes .
we concentrate on events with tracks and showers where the respective energies @xmath1 and @xmath2 are measured separately .
then we compute analytically and numerically event distributions in two variables @xmath3 and @xmath4 .
we find that the @xmath5-distribution is especially useful because in the @xmath0-induced reactions the distribution has a minimum at @xmath6 and then increases as @xmath7 .
this is different in @xmath8-induced reactions where the @xmath5-distribution decreases monotonically .
the results are demonstrated with figures where one can estimate the required sensitivity of the experiments .
another attractive property is that in several ratios the neutrino flux factorizes and drops out .
we hope the results of this article will be useful for searches of tau ( anti)neutrinos of energies above 100 gev to several tev in the present and planned large volume neutrino telescopes .
tau neutrino ; neutrino detection 95.85 ry , 95.55.vj , 98.70.sa , 13.35.dx |
numerous infrared space observatories , first of all , the spitzer space telescope , as well as msx and iso satellites , opened up a new era in studies of the star formation both in nearby and distant galaxies . the so - called polycyclic aromatic hydrocarbons ( pah ) @xcite macromolecules consisting of several tens or several hundred atoms , mostly carbon and hydrogen
are of special interest in relation to these observations .
the absorption of an ultraviolet ( uv ) photon by such a molecule excites bending and vibrational modes , and , as a result , near ir photons are emitted .
pah emission bands may account for a substantial fraction ( up to several dozen percent ) of the entire infrared luminosity of the galaxy @xcite .
polycyclic aromatic hydrocarbons attract considerable interest at least for two reasons .
first , their emission is related to the overall uv radiation field of a galaxy , making them a natural indicator of the star formation rate .
second , pah molecules not only trace the state of the interstellar medium , but also play an important role in its physical and chemical evolution .
the former aspect is interesting both for interpretation of available ir observations and for planning new near - ir space missions ( jwst , sofia , spica , etc . ) .
the latter aspect is of great importance for development of models for various objects ranging from protoplanetary disks to the interstellar medium of an entire galaxy .
unfortunately , pah formation and destruction mechanisms in the interstellar medium still are not understood .
such possible scenarios as the synthesis of pahs in carbon - rich atmospheres of agb and post - agb stars or in dense molecular clouds as well as the destruction of pahs by shocks and uv radiation are discussed extensively in the literature ( see sandstrom et al .
@xcite and references therein ) .
the observed deficit of the emission of these macromolecules in metal - poor galaxies may be an important indication of the nature of the pah evolutionary cycle .
note that , as shown by draine et al .
@xcite , this deficit is related to the real lack of pahs and not to the low efficiency of the excitation of their ir transitions . in galaxies with oxygen abundance
@xmath3 the typical mass fraction of pahs @xmath4 ( the fraction of the total dust mass in particles consisting of at most one thousand atoms ) is equal to about 4% , i.e. , about the same as in the milky way . at metallicities @xmath5 the average @xmath4 decreases quite sharply down to 1% and even lower . in order to carify the cause of this transition and to identify the pah formation and destruction mechanisms as well as their relation to the physical parameters and the metallicity of a galaxy , sandstrom et al .
@xcite analyzed in detail spitzer observations of the dust component in the nearest irregular dwarf galaxy the small magellanic cloud ( smc ) .
these authors found weak correlation or no correlation between @xmath4 and such smc parameters as the location of carbon - rich asymptotic giant branch stars , supergiant hi shells and young supernova remnants , and the turbulent mach number .
they showed that @xmath4 correlates with co intensity and increases in regions of high dust and molecular gas surface density .
sandstrom et al .
@xcite concluded that pah mass fraction is high in regions of active star formation , but suppressed in bright hii regions .
the irregular dwarf galaxy ic 10 is analogous to the smc in terms of a number of parameters .
the average gas metallicity in ic 10 is @xmath6 , varying from 7.6 to 8.5 in different hii regions ( @xcite and references therein ) , i.e. , it covers the very range where the transition from high to low pah abundance occurs .
the interstellar medium of this galaxy is characterized by a filamentary , multi - shell structure . in h@xmath0 and [ sii ] images ic 10 appears as a giant complex of multiple shells and supershells , arc- and ring - shaped features with sizes ranging from 50 to 8001000 pc ( see @xcite and references therein ) .
the hi distribution also shows numerous `` holes '' , supershells , and extended irregular features with rudiments of a spiral pattern @xcite .
the ic 10 galaxy is especially attractive for the analysis of the dust component because , unlike the smc , it is a starburst galaxy .
it is often classified as a bcd - type object because of its high h@xmath0 and ir luminosity @xcite .
the stellar population of ic 10 shows evidence of two star formation bursts .
the first burst is at least 350 myr old , while the second one has occurred 410 million years ago ( see @xcite and references therein ) . for
the purpose of identifying the influence of shocks and/or uv radiation on dust the anomalously large population of wolf rayet ( wr ) stars in ic 10 is of special interest . here
the highest density of wr stars is observed among the known dwarf galaxies , comparable to the density of these stars in massive spiral galaxies @xcite .
high h@xmath0 and ir luminosity of ic 10 , combined with the large number of wr stars , indicates that the last burst of star formation in this galaxy must have been short , but engulfed most of the galaxy .
the anomalously high number of wr stars means that we are actually witnessing a short period immediately after the last episode of star formation .
the central , brightest region , associated with this last star formation episode , is located in the south - eastern part of the galaxy and includes the largest and densest hi cloud , a molecular cloud seen in co lines , a conspicuous dust lane , and a complex of large emission nebulae , reaching 300400 pc in size , with two shell nebulae hl111 and hl106 ( according to the catalog of hodge & lee @xcite ) , as well as young star clusters and about a dozen wr stars ( see @xcite and references therein ) .
an h@xmath0 image of this central region of the galaxy is shown in figure [ general ] .
image of the central region of the ic 10 galaxy with the contours showing regions of high h@xmath0 intensity .
the asterisks and circles show wr stars and clusters from the list of tikhonov & galazutdinova @xcite , respectively . hl111 and hl106 regions , mentioned in the text ,
are also indicated along with stars wr m24 , r2 , and r10.,scaledwidth=90.0% ] according to vacca et al .
@xcite , the center of the last star formation episode is located near the object that was earlier classified as the wr star m24 .
( hereafter letters r and m , followed by a number , refer to wr stars from the lists of royer et al . @xcite and massey & holmes @xcite , respectively ) .
vacca et al .
@xcite showed that m24 is actually a close group consisting of at least six blue stars , four of these stars being possible wr candidates .
lopez - sanchez et al .
@xcite have conclusively identified two wr stars in this region .
the hl111c nebula , surrounding m24 , is one of the brightest hii regions in ic 10 and the brightest part of the hl111 shell .
the neighborhood of m24 and the inner cavern of this shell host youngest ( 24 myr old ) star clusters in the galaxy @xcite .
the shell nebula hl106 is located in the densest southern part of a complex , consisting of hi , co , and dust clouds mentioned above .
the ionizing radiation in this region must be generated by wr stars r2 and r10 and clusters 4 - 3 and 4 - 4 from the list of hunter @xcite . according to the above author ,
these clusters are a few times older than young clusters 4 - 1 and 4 - 2 in the hl111 region . from the south , adjacent to the hi and
co clouds and the dust lane is a unique object , the so - called synchrotron supershell @xcite . until recently , it was believed to have been formed as a result of multiple explosions of about a dozen supernovae @xcite .
lozinskaya & moiseev @xcite for the first time explained the formation of this synchrotron supershell by a hypernova explosion .
the above features of ic 10 offer great possibilities for the study of the structure and physical characteristics of the dust component of a dwarf galaxy and the role that shocks play in its evolution . in this paper
we analyze the connection between our data on ic 10 , obtained earlier , and observations of this galaxy with the spitzer telescope . in the following sections we describe the technique used to analyze these observations , present the obtained results and discuss them . in conclusions we summarize the main findings of this work .
in this paper we use spitzer archive observations of ic 10 obtained as a part of the program `` a mid - ir hubble atlas of galaxies '' @xcite .
these data were downloaded from the spitzer heritage archive .
the mopex software was utilized to compose image mosaics and custom idl procedures were used to analyze them .
one of the most complicated issues in the analysis of such images is the choice of the background level .
sandstrom et al .
@xcite used a rather sophisticated procedure for this purpose , because the small magellanic cloud occupies a large area on the sky .
the angular size of ic 10 is small and one may therefore expect that background ( mostly intragalactic and zodiacal ) variations are small toward this galaxy . in this paper
we set the background level in all the ir bands considered to the average brightness in areas located far from the star forming regions in ic 10 .
the adopted background values are listed table [ bkgr ] .
such a simple procedure for background estimation is acceptable for our purposes .
it is interesting that background values estimated using spot ( spitzer planning observations tool ) , which are also shown in table [ bkgr ] ( these are values written in fits file headers ) , in some cases differ appreciably from the adopted values .
this further supports the use of the background estimate extracted from real data .
.ir background levels toward ic 10 adopted in this paper and background levels according to spot . [ cols="^,^,^ " , ] to analyze possible effects of shocks on the dust component , we use results of h@xmath0 and [ sii]@xmath7 observations , made with the scorpio focal reducer and the scanning fabry - perot interferometer at the 6-m telescope of the special astrophysical observatory of the russian academy of sciences and described in detail by lozinskaya et al .
@xcite and egorov et al .
@xcite . to compare the dust component distribution with the large - scale structure and kinematics of hi in ic 10 ,
we used 21-cm vla data obtained by wilcots and miller @xcite .
egorov et al .
@xcite reanalyzed the data cube of these observations , provided to us by the authors , in order to study the `` local '' structure and kinematics of hi in the neighborhood of the star forming complex and the brightest nebulae hl111 and hl106 .
we used the data with an angular resolution of @xmath8 ( corresponding to a linear resolution of about 20 pc for the adopted distance of 800 pc to the galaxy ) .
the ir and h@xmath0 maps of the central region of ic 10 are shown in figure [ irha ] .
the interest to near - ir observations of galaxies is related to the fact that uv - excited pah bands can be used as an indicator of the number of hot stars and hence as an indirect indicator of the star formation rate .
draine & li @xcite proposed to parameterize the uv radiation field of the galaxy as the sum of the `` minimum '' diffuse uv field @xmath9 ( the lower cutoff of the starlight intensity distribution ) , filling up most of the galaxy s volume , and a more intense uv field with a power - law distribution , which illuminates only the mass fraction @xmath10 of all the dust in the galaxy .
the @xmath9 quantity , expressed in units of the average uv radiation field in our galaxy , characterizes the overall rate of star formation in the system studied , whereas @xmath10 allows one to estimate the mass fraction of the galaxy involved in the ongoing star formation .
other parameters introduced are the 24-to-70 @xmath11 m flux ratio , which characterizes the fraction of `` hot '' dust , and @xmath12 , the dust luminosity fraction contributed by regions with the uv radiation intensity @xmath13 , i.e. , the dust luminosity coming from photodissociation regions .
m ( the upper panel ) ; 8 @xmath11 m ( the middle panel ) , and 24 @xmath11 m ( the lower panel ) . superimposed on the maps
are h@xmath0 contours corresponding to the levels of @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 , and @xmath26 in arbitrary units.,scaledwidth=50.0% ] draine & li @xcite proposed a general algorithm for estimating the parameters of a galaxy from ir observations at 8 @xmath11 m , 24 @xmath11 m , 70 @xmath11 m , and 160 @xmath11 m ( 3.6 @xmath11 m data are used to remove the starlight contribution ) . unfortunately , this algorithm can be applied to ic 10 only partially , because 160 @xmath11 m observations are not available for a substantial part of the galaxy .
results of long - wavelength observations are shown in figure [ mips ] . as is evident from the figure ,
160 @xmath11 m data are mostly available for outskirts of the galaxy and cover the star forming regions only partially ( taking into account the angular resolution , which is equal to 40@xmath27 at 160 @xmath11 m ) .
m ( left panel ) and 70 @xmath11 m ( the right panel ) maps of the central part of the star forming region in ic 10 with 160 @xmath11 m emission contours.,scaledwidth=80.0% ] nevertheless , we used the available data and the technique described by draine & li @xcite to determine @xmath9 , @xmath10 , and @xmath4 , the mass fraction of dust contained in pah ( or , more precisely , in particles with less than 1000 carbon atoms ) .
we then averaged the data of ir observations over the region of ic 10 , covered by 160 @xmath11 m data , and inferred the following parameters for this region : @xmath28 , @xmath29 , and @xmath30 .
a comparison of these values with results of draine et al .
@xcite for 65 galaxies of different types shows that parameters of ic 10 differ appreciably from the typical values of the corresponding quantities .
comparable @xmath9 values are only found for two other irr galaxies ngc 2915 ( @xmath31 ) and ngc 5408 ( @xmath28 ) .
note that ngc 5408 was also shown to contain starburst regions @xcite .
the @xmath10 parameter of ic 10 is also close to that of ngc 5408 , but is much lower than the corresponding value in mrk 33 , another irregular galaxy from the list of draine & li @xcite , where it exceeds 10% .
( draine & li @xcite also report high @xmath10 value for the seyfert galaxy ngc 5195 , however , the result for this object is more dependent on the adopted radiation field parameters than in the case of mrk 33 . ) in the sample of draine & li @xcite mrk 33 is also the galaxy with the highest 24-to-70 @xmath11 m flux ratio ( @xmath32 ) and the largest value of @xmath33 .
the corresponding values for ic 10 are equal to about 0.7 and 23% , respectively . on the whole , as far as radiation - field parameters are concerned
, ic 10 is quite a typical irr starburst galaxy .
the ic 10 galaxy has unusually high pah mass fraction @xmath4 .
its value , inferred using the technique of draine & li @xcite , significantly exceeds the corresponding parameters for all the galaxies mentioned above ( 1.3% in mrk 33 , 2.4% in ngc 5195 , 1.4% in ngc 2915 , and 0.4% in ngc 5408 ) . in the algorithm of draine & li @xcite this parameter is inferred from the sole quantity the ratio of the average 8 @xmath11 m flux to the sum of the 70 @xmath11 m and 160 @xmath11 m fluxes .
our estimate for this parameter is 0.19 , which , according to draine & li @xcite , corresponds to @xmath30 . to check whether such an unusually high @xmath4 is obtained due to the lack of 160 @xmath11 m data
, we performed a more detailed fit of the observed 5.8 @xmath11 m , 8 @xmath11 m , 24 @xmath11 m , and 70 @xmath11 m fluxes based on the models of draine & li @xcite using local rather than average fluxes .
this technique allowed us to obtain individual estimates for different regions of the galaxy .
our modeling showed that the final average @xmath4 and its distribution across the galaxy do depend appreciably on the choice of the passbands used in the fit .
however , all the considered cases still yield a high 8 @xmath11 m flux - averaged pah fraction ranging from 2.9% to 4.5% .
note that the distribution of @xmath4 in the galaxy is quite irregular .
along with regions of high @xmath4 there are vast areas , where @xmath4 is less than 1% .
as we mentioned above , the particular features of the distribution of @xmath4 depend on the choice of passbands used in the fit of photometric data .
hereafter we use the 8 @xmath11 m to 24 @xmath11 m ( @xmath34 ) flux ratio as the local indicator of the pah fraction . a number of authors and , in particular , sandstrom et al .
@xcite , pointed out the possibility of using the above flux ratio for this purpose ( note , however , that the correlation between @xmath34 and @xmath35 found by sandstrom et al .
@xcite is rather weak ) .
the halftones in figure [ rat824 ] ( left panel ) show the distribution of this flux ratio in the central star forming region of ic 10 , and contours correspond to the distribution of h@xmath0 intensity .
lighter tones indicate low @xmath34 ratios and , correspondingly , low @xmath4 , whereas darker tones indicate higher @xmath4 .
a wide semi - ring near the hl111 and hl106 regions is immediately apparent , which can be traced by low @xmath34 ratio , weak h@xmath0 intensity , and the locations of wr stars ( we connected them by lines to emphasize the location of the semi - ring ) .
it might be supposed that the low pah abundance in this region is caused by the destruction of these particles by the ultraviolet radiation of wr stars , however , further studies are needed for a more definite conclusion .
flux ratio in the central star forming region of ic 10 .
lighter tones correspond to lower flux ratios . also shown
are h@xmath0 contours .
asterisks indicate positions of wr stars .
right panel : the correlation between the @xmath34 flux ratio and h@xmath0 intensity for the region shown in the left panel ; only data points where 8 @xmath11 m flux exceeds 5 mjy / sr are included in the plot.,title="fig:",scaledwidth=50.0% ] flux ratio in the central star forming region of ic 10 .
lighter tones correspond to lower flux ratios . also shown
are h@xmath0 contours .
asterisks indicate positions of wr stars .
right panel : the correlation between the @xmath34 flux ratio and h@xmath0 intensity for the region shown in the left panel ; only data points where 8 @xmath11 m flux exceeds 5 mjy / sr are included in the plot.,title="fig:",scaledwidth=40.0% ] the relation between the pah abundance and star formation tracers is apparent not only in this semi - ring , but in the entire considered region . in the right panel of figure [ rat824 ]
we show the correlation between the @xmath34 flux ratio and h@xmath0 intensity .
it is evident from the figure that @xmath34 ( and hence @xmath4 ) decreases with increasing h@xmath0 intensity .
this may indicate that factors operating in the vicinity of the region of ongoing star formation , e.g. , the uv radiation , have a destructive effect on pah particles .
the @xmath34 flux ratio approaches unity in regions with less intense h@xmath0 flux , and this corresponds to @xmath4 values of about 23% @xcite .
pah particles may also be destroyed by shocks .
therefore , generally speaking , the above - mentioned low @xmath34-ratio `` semi - ring '' located near hl111 and hl106 might have formed due to the destructive effect not only from the uv radiation of the wr stars located within it , but also from shocks produced by winds of these stars .
the primary shock indicators are high - velocity gas motions .
a detailed study of the ionized gas kinematics in ic 10 by egorov et al .
@xcite indeed revealed weak high - velocity features in wings of h@xmath0 and [ sii]@xmath7 lines in the inner cavern of the hl111 nebula and in other regions of the complex of violent star formation .
in particular , such features were found in the vicinity of two wr stars located in the `` semi - ring '' mentioned above .
we reanalyzed the results of observations of the galaxy in both lines made with the fabry
perot interferometer at the 6-m telescope of the special astrophyscial observatory of the russian academy of sciences in order to reveal possible anticorrelation between high - velocity gas motions and @xmath4 .
we computed h@xmath0 and [ sii]@xmath7 line profiles for several regions of high and low @xmath34 ratio .
weak high - velocity features at a level of about 26% of the peak intensity are found in wings of both lines , and this coincidence confirms the reality of corresponding motions
. however , these high - velocity features show up both in regions with high and low @xmath34 ratios .
to obtain more definitive results , we mapped the distributions of velocities and intensities of high - velocity features in blue and red wings of h@xmath0 line in the entire available field of the galaxy and in the central star forming region .
the resulting maps indicate that high - velocity features in blue and red wings of the line show up in ranges from 5060 to 100110 km / s and 50 to 100 km / s , respectively , relative to the velocity of the line peak . in figure [ hivel ]
we compare the @xmath34 flux ratio to the intensities of high - velocity features in blue ( left panel ) and red ( right panel ) wings of h@xmath0 line .
if the pah abundance depends on the presence of shocks , one would expect the intensity of high - velocity features to anticorrelate with @xmath34 .
no such anticorrelation can be seen in the figure , albeit a certain pattern does emerge : higher intensities of high - velocity features in both the blue and red wings tend to `` avoid '' regions with the highest @xmath34 ratios , although they are observed in nearby , slightly offset , locations .
nonetheless , the results reported here do not allow us to conclusively associate the destruction of pah particles with shocks produced by stellar winds and/or supernova explosions .
line with the @xmath34 flux ratio contours superimposed.,scaledwidth=50.0% ] arkhipova et al .
@xcite determined metallicities for a number of hii regions in ic 10 .
it is interesting to relate these metallicities to pah content in order to see whether @xmath4 decreases with decreasing metallicity within the galaxy in the same way as it does when we compare different irr galaxies .
figure [ oh824 ] shows the @xmath34 ratios as a function of oxygen abundance for hii regions from the list of arkhipova et al .
in some cases two data points in this plot correspond to the same hii region .
metallicities of these hii regions inferred from long - slit and mpfs observations differ slightly , possibly due to different integration areas .
we show only the data points with metallicity errors smaller than or equal to 0.05 dex .
flux ratio toward hii regions from the list of arkhipova et al .
@xcite as a function of oxygen abundance.,scaledwidth=50.0% ] it is evident from figure [ oh824 ] that the @xmath34 ratio indeed decreases with decreasing oxygen abundance , although the turn - off value of @xmath36 is about @xmath37 rather than 8.08.1 as found in earlier works .
this result indicates that the metallicity dependence of the pah abundance shows up not only globally ( at the level of entire galaxies ) , but also locally ( at least at the level of individual hii regions ) . as we pointed out in the introduction , the only important difference between the two nearest irr galaxies ic 10 and smc is that in ic 10 we observe the interstellar medium immediately after a violent burst of star formation that has encompassed most of the galaxy .
it is therefore of interest to compare the data obtained in this work with results of a detailed study of the dust component in the smc performed by sandstrom et al .
( recall that we use the @xmath34 flux ratio to measure the pah mass fraction and draw our conclusions based on this ratio . )
ic 10 , like the smc , shows strong @xmath4 variations from one region to another with pah avoiding bright hii regions .
the lower spatial resolution prevents us from concluding that pahs are located in the shells of bright nebulae , however , the large - scale map ( figure [ rat824 ] ) shows clearly that the h@xmath38 brightness anticorrelates with the @xmath34 ratio .
sandstrom et al .
@xcite found weak or no correlation between @xmath4 and the location of hi supershells in the smc .
the ic 10 galaxy , on the contrary , shows well - correlated ( nearly coincident ) extended shell - like structures in maps of @xmath34 flux ratio and 21-cm hi emission ( figure [ rat8 - 24-hi - map ] ) . a correlation between the 8 @xmath11 m ir emission and
the extended hi shell is also apparent in the middle panel of figure [ irha ] .
flux ratio with hi emission contours superimposed.,scaledwidth=50.0% ] the large - scale correlation between 8 @xmath11 m brightness and extended arcs and hii and hi shells in ic 10 and in a number of other starburst irr galaxies has been known since long ( hunter et al .
the above authors attributed all the 8 @xmath11 m flux solely to pah emission and concluded that @xmath35 correlates with brightness of giant shells and supershells .
the correlation between the @xmath34 flux ratio and the 21-cm line emission leads us to the same conclusion in a somewhat more straightforward way .
we believe this is a real correlation , as the stellar wind from numerous young star clusters and wr stars located inside extended shells shapes observed shell - like structure of both the gas and dust in ic 10 .
another implication is that pah molecules do not undergo significant destruction during the sweep - up of giant shells ( see also @xcite ) .
the brightest extended co cloud in the galaxy and the dust lane that coincides with it are located just to the south of the complex of ongoing star formation and are immediately adjacent to the hl106 nebula .
figure [ comap ] shows the structure of this cloud according to data of leroy et al .
@xcite superimposed on the distribution of @xmath34 flux ratio .
( we obtained a composite map of the entire co cloud by combining maps of its individual components : clouds b11a , b11b , b11c , and b11d in figure 7 from @xcite . ) the total gas column density @xmath39(hi ) toward the dense cloud , discussed here , amounts to @xmath40 @xmath41 @xcite . according to co emission observations , the column density of neutral and molecular hydrogen @xmath39(h ) in this direction is about @xmath42 @xmath41 @xcite .
flux ratio with the contours of molecular gas distribution superimposed.,scaledwidth=50.0% ] it follows from figure [ comap ] that three regions with highest @xmath34 ratios have exactly the same locations and sizes as the co clouds b11a , b11c , and b11d from @xcite
. the fourth co cloud b11b coincides with the bright shell nebula hl106 .
egorov et al .
@xcite showed that the optical nebula hl106 is not located behind a dense cloud layer , but is partly embedded in it .
egorov et al .
@xcite also concluded that b11b is physically associated with the optical nebula .
first , the radial velocity of the b11b cloud ( @xmath43 km / s @xcite ) coincides with the velocity of ionized gas in hl106 determined by egorov et al .
but most importantly , the brightest southern arc hl106 exactly outlines the boundaries of the b11b cloud .
such an ideal coincidence can not be accidental and hints to a physical relation between the thin ionized shell and the molecular cloud b11b .
the hl106 shell , which exactly bounds the b11b cloud , formed due to photodissociation of molecular gas at the boundary of this cloud as well as due to ionization by the uv radiation of wr stars r2 and r10 and clusters 4 - 3 and 4 - 4 .
hunter @xcite estimates these clusters to be about 20 - 30 myr old .
the presence of a bright ionizing nebula that surrounds b11b explains low @xmath34 flux ratio toward this cloud .
we can thus conclude with certainty that highest @xmath34 ratios and , consequently , highest pah fractions @xmath4 are indeed found toward dense co clouds .
the only exception is the region of the brightest nebula hl45 .
the evident drop of the @xmath34 flux ratio observed in this area may be due to destruction of pahs by strong uv radiation .
in the introduction we have already emphasized the importance of understanding the evolution of pahs and their relation to other galaxy components . in this work
we compare results of infrared observations of ic 10 with other available observations in order to identify possible indications to the origin of pah .
one of the key pah properties to be explained by their evolutionary model is the low content of these particles in metal - poor galaxies .
two hypotheses are mainly discussed in the literature less efficient formation and more efficient destruction of pahs in metal - poor systems .
galliano et al .
@xcite argue that the dependence of @xmath4 on the metallicity of a host galaxy can be naturally explained if we assume that pah particles are synthesized in the atmospheres of long - lived agb stars . in this case
low metal and pah abundances are due to the slower stellar evolution . however ,
if this assumption were true , the pah fraction @xmath4 in the ic 10 galaxy would be , first , low , and second , uniformly distributed throughout the galaxy .
we show the pattern to be exactly the opposite @xmath4 ( given by the @xmath34 flux ratio ) varies appreciably across the galaxy and amounts almost to 4% in some areas . from the viewpoint of the spatial localization ,
pahs correlate with both dense - gas indicators studied ( hi and co clouds ) .
the pah mass fraction decreases only in the neighborhood of hii regions and wr stars , which is consistent with the hypothesis that these particles are destroyed by uv radiation and shocks ( although we failed to find convincing evidence for pah destruction by shocks ) . on the whole ,
the pattern observed in ic 10 is qualitatively consistent with the assumption that pah particles in molecular clouds form _ in situ_. in this case the current high @xmath4 value in ic 10 may be related to a recent burst of star formation during which pah particles have formed in dense gas , and did not have enough time to be destroyed anywhere except for the immediate neighborhood of the uv radiation sources .
if this interpretation is correct , the metallicity dependence of @xmath4 should show up until pah particles begin to be destroyed by ultraviolet radiation , and reflects the peculiarities of their formation rather than their subsequent evolution .
we further plan to verify our conclusions by analyzing observational results on other dwarf galaxies .
this work was supported by the russian foundation for basic research ( grants nos .
10 - 02 - 00091 and 10 - 02 - 00231 ) and the russian federal agency on science and innovation ( contract no .
02.740.11.0247 ) .
egorov thanks the dynasty foundation of noncommercial programs for financial support .
authors are grateful to suzanne madden and tara parkin for useful discussions .
99 tielens a.g.g.m .
astron . astroph . *
46 * , 289 ( 2008 ) smith j.d.t .
, draine b.t . ,
dale d.a . ,
moustakas j. et al .
j. * 656 * , 770 ( 2007 ) sandstrom k.m .
, bolatto a.d .
, draine b. , bot c. , stanimirovic s. astrophys .
j. * 715 * , 701 ( 2010 ) draine b.t . ,
, bendo g. et al .
j. * 663 * , 866 ( 2007 ) lozinskaya t.a .
, egorov o.v . ,
moiseev a.v .
, bizyaev d.v .
. lett . * 35 * , 730 ( 2009 ) magrini l. , gonalves d.r . mon . not .
astron . soc .
* 398 * , 280 ( 2009 ) arkhipova v.p .
, egorov o.v . ,
lozinskaya t.a .
, moiseev a.v .
pisma astron .
zh . * 37 * , 83 ( 2011 ) wilcots e.m .
, miller b.w .
astron.j . * 116 * , 2363 ( 1998 ) gil de paz a. , madore b.f . , pevunova o. astrophys .
* 147 * , 29 ( 2003 ) leroy a. , bolatto a. , walter f. , blitz l. astrophys .
j. * 643 * , 825 ( 2006 ) chyzy k.t .
, knapik j. , bomans d.j . , klein u. , beck r. , soida m. , urbanik m. astron .
astrophys . * 405 * , 513 ( 2003 ) lozinskaya t.a .
, moiseev a.v . , podorvanyuk n.yu .
, burenkov a.n .
lett . * 34 * , 217 ( 2008 ) richer m.g .
, bullejos a. , borissova j. et al .
. astrophys . * 370 * , 34 ( 2001 ) hunter d. astrophys .
j. * 559 * , 225 ( 2001 ) massey p. , olsen k. , hodge p. , jacoby g. , mcneill r. , smith r. , strong sh .
j. * 133 * , 2393 ( 2007 ) vacca w.d . , sheehy c.d .
, graham j.r .
j. * 662 * , 272 ( 2007 ) massey p. , armandroff t.e .
, conti p.s .
j. * 103 * , 1159 ( 1992 ) massey p. , holmes s. astrophys .
j. * 580 * , l35 ( 2002 ) crowther p.a , drissen l. , abbott j.b . ,
royer p. , smartt s.j .
astrophys . *
404 * , 483 ( 2003 ) hodge p. , lee m.g .
. pacif . * 102 * , 26 ( 1990 ) tikhonov n. a. , galazutdinova o. a. astron . lett . *
35 * , 748 ( 2009 ) royer p. , smartt s.j . , manfroid j. , vreux j. astron
. astrophys . *
366 * , l1 ( 2001 ) lopez - sanchez a.r .
, mesa - delgado a. , lopez - martin l. , esteban c. mon . not .
astron . soc . in press ( arxiv:1010.1806 ) yang h. , skillman e.d .
j. * 106 * , 1448 ( 1993 ) bullejos a. , rozado m. rev .
( ser de conf . )
* 12 * , 254 ( 2002 ) rosado m. , valdez - guti ' errez m. , bullejos a. , arias l. , georgiev l. , ambrocio - cruz p. , borissova j. , kurtev r. asp conf
. ser . * 282 * , 50 ( 2002 ) thurow j.c . , wilcots e.m .
j. * 129 * , 745 ( 2005 ) lozinskaya t.a .
, moiseev a.v .
mon . not .
. soc . * 381 * , l26 ( 2007 ) fazio g. , pahre m. spitzer proposal i d 69 ( 2004 ) egorov o.v . ,
lozinskaya t.a .
, moiseev a.v .
astron . rep . *
87 * , 277 ( 2010 ) draine , b.t .
, li a. astrophys .
j. * 657 * , 810 ( 2007 ) i. d. karachentsev , m. e. sharina , a. e. dolphin , e. k. grebel , d. geisler , p. guhathakurta , p. w. hodge , v. e. karachentseva , a. sarajedini , p. seitzer . astron
. astrophys . *
385 * , 21 ( 2002 ) hunter d.h .
, elmegreen b.g .
, martin e. astron .
j. * 132 * , 801 ( 2006 ) bolatto a.d .
, jackson j.m . , wilson c.d . , moriarty - schieven g. astrophys .
j. * 532 * , 909 ( 2000 ) galliano f. , dwek e. , chanial p. astrophys .
j. * 672 * , 214 ( 2008 ) | infrared observation from the spitzer space telescope archive are used to study the dust component of the interstellar medium in the ic 10 irregular galaxy .
dust distribution in the galaxy is compared to the distributions of h@xmath0 and [ sii ] emission , neutral hydrogen and co clouds , and ionizing radiation sources .
the distribution of polycyclic aromatic hydrocarbons ( pah ) in the galaxy is shown to be highly non - uniform with the mass fraction of these particles in the total dust mass reaching 4% .
pahs tend to avoid bright hii regions and correlate well with atomic and molecular gas .
this pattern suggests that pahs form in the dense interstellar gas .
we propose that the significant decrease of the pah abundance at low metallicity is observed not only globally ( at the level of entire galaxies ) , but also locally ( at least , at the level of individual hii regions ) .
we compare the distribution of the pah mass fraction to the distribution of high - velocity features , that we have detected earlier in wings of h@xmath0 and sii lines , over the entire available galaxy area .
no conclusive evidence for shock destruction of pahs in the ic 10 galaxy could be found .
@xmath1__institute of astronomy , russian academy of sciences , ul .
pyatnitskaya 48 , moscow , 119017 russia _ _
@xmath2__sternberg astronomical institute , universitetskii pr .
13 , moscow , 119992 russia _ _ |
we hope that supersymmetric ( susy ) gauge theories in 2 + 1 dimensions will help us learn about general features of 2 + 1 dimensional gauge theories which in turn might shed light on certain condensed matter systems with emergent gauge symmetry at low temperatures . for many years , there has been strong evidence that , similar to the four dimensional case , three dimensional gauge theories are related to each other by a large number of non - perturbative dualities . in four dimensions
, anomalies provided quantitative tests of these conjectured dualities , while in three dimensions , because of the absence of anomalies , the situation was more difficult . recently , with the introduction of methods for calculating a superconformal index on @xmath6 @xcite and the euclidean partition function on @xmath2 @xcite , the situation has changed dramatically . in this paper , we continue an investigation , started in @xcite , of the the large @xmath3 limit of the @xmath2 partition function of supersymmetric ( susy ) chern - simons ( cs ) theories .
the partition function @xmath7 is calculated using the matrix model derived in ref .
@xcite by localization ( later improved by @xcite to allow matter fields to acquire anomalous dimensions ) . for the cs theory at its superconformal fixed point , the matrix model of @xcite computes exactly the partition function and certain supersymmetric wilson loop expectation values .
the theories we examine here have @xmath0 susy , a product @xmath1 gauge group structure , and field content summarized by a quiver diagram .
while in our previous work @xcite we examined cs theories that had a @xmath8 gauge group , in this work we relax the constraint that the ranks of the @xmath1 groups all be equal to each other .
we examine quantitatively some of these three dimensional dualities .
we will study the generalized seiberg duality of @xcite .
modulo convergence issues , refs .
@xcite demonstrated that @xmath9 is invariant under this duality . in this paper , we investigate what this invariance looks like in the large @xmath3 limit .
we will also examine the ads / cft duality between these gauge theories and freund - rubin compactifications of eleven dimensional supergravity .
a big motivation for this paper and our previous work @xcite is the hope that the matrix model can shed light on the microscopic origin of the mysterious @xmath10 scaling of the free energy @xcite indicated by ads / cft .
an early investigation of this matrix model @xcite demonstrated that the free energy of maximally supersymmetric @xmath11 yang - mills theory at its infrared fixed point scales as @xmath10 .
by `` free energy '' @xmath12 , we mean minus the logarithm of the partition function on @xmath2 , f = - |z_s^3| . while one could declare victory , the localization procedure of @xcite leaves the microscopic physics of this scaling obscure .
here we study a larger class of gauge theories using different methods in the hopes of shedding further light on this large @xmath3 scaling . in the context of the ads / cft correspondence ,
the free energy @xmath12 can be computed from a classical gravity model dual to the strongly interacting , large @xmath3 field theory . for a cft
dual to @xmath13 of radius @xmath14 and effective four - dimensional newton constant @xmath15 , @xmath12 is given by @xcite f = .
[ emparanresult ] these types of @xmath13 backgrounds arise as freund - rubin compactifications of @xmath16 of m - theory , where @xmath17 is a seven - dimensional sasaki - einstein space supported by @xmath3 units of four - form flux .
the quantization of @xmath14 in planck units implies that at large @xmath3 eq .
( [ emparanresult ] ) becomes @xcite f = n^3/2 .
[ fvolrel ] ( here , the volume of @xmath17 is computed with an einstein metric that satisfies the normalization condition @xmath18 . ) these types of freund - rubin solutions arise as the near horizon limit of a stack of @xmath3 m2-branes placed at the tip of the calabi - yau cone @xmath19 over @xmath17 .
the cft in question is then the low energy field theory description of the m2-branes .
the matrix model can be used to compute not only the three - halves scaling but also the volume factor and thus allows for quantitative tests of proposed ads / cft dual pairs .
note that the conjecture of ref .
@xcite preceded the discovery of ads / cft and also concerned the scaling not of @xmath12 but of the thermal free energy .
ads / cft makes clear that the @xmath10 scaling comes from the over - all normalization of the gravitational action and thus is a universal property of many quantities in the dual field theory . in trying to elucidate the connection between matrix model quantities and field theory quantities in @xcite
, we found an interesting relationship between the saddle point eigenvalue distribution and the distribution of gauge invariant operators in the chiral ring .
we now review this relationship .
the localization procedure of @xcite reduces the partition function to an integral over @xmath20 constant @xmath21 matrices @xmath22 where @xmath22 is the real scalar that belongs to the same @xmath23 multiplet as the gauge connection .
@xcite presented a procedure ( later improved in @xcite ) to evaluate the matrix model by saddle point integration in the large @xmath3 limit . at the saddle point ,
the real parts of the eigenvalues @xmath24 of @xmath22 grow as a positive power of @xmath3 while their imaginary parts stay of order one as @xmath3 is taken to infinity .
additionally , the real parts of the eigenvalues are the same for each gauge group . therefore , in order to find the saddle point , one can consider the large @xmath3 ansatz _
j^(a ) = n^ x_j + i y_a , j + as one takes @xmath25 , the @xmath26 and @xmath27 become dense ; one can introduce an eigenvalue density ( x ) = _ n _ j=1^n ( x - x_j ) .
the chiral ring of these quiver gauge theories consists of gauge invariant products of the bifundamental and fundamental matter fields and monopole operators modulo superpotential and monopole relations . at large @xmath3 ,
only the so - called `` diagonal monopole operators '' are important .
these operators turn on the same number of units of flux through the diagonal @xmath28 subgroup of each @xmath1 gauge group .
operators in the chiral ring therefore have an associated r - charge @xmath29 and a ( diagonal ) monopole charge @xmath30 . to characterize these operators , we introduce the function @xmath31 which counts the number of operators in the chiral ring with r - charge less than @xmath29 and monopole charge less than @xmath30 . in @xcite , we found evidence for the following relation between the saddle point eigenvalue distribution and @xmath31 : .
|_m = rx / = ( x ) , [ evchiralrel ] where @xmath32 is a lagrange multiplier that enforces the constraint @xmath33 . defining a function @xmath34 that counts the number of operators without the field @xmath19 , there is a similar conjectured relation @xcite between @xmath35 and the @xmath27 which we shall review in section [ sec : d4counting ] . in the current paper ,
we begin in section [ sec : ade ] by examining cs theories with @xmath0 susy with matter fields only in bifundamental representations of the product gauge group .
however , we relax the constraint that the ranks of the @xmath36 gauge groups all be equal .
we find that the saddle point procedure described in @xcite for evaluating the matrix model works if the ranks of the gauge groups satisfy the following condition : 2 n_a = _
b | ( a , b ) e n_b , [ adecond ] where @xmath37 is the set of bifundamental fields . ( in other words , the number of flavors for a given gauge group should be twice the number of colors . )
if this condition is not satisfied , there will in general be long range forces between the eigenvalues that render the method in @xcite inapplicable .
interestingly , this condition implies that the gauge groups and matter content are described by a simply laced , affine dynkin diagram ( see figure [ fig : dynkin ] ) . for a @xmath38 dimensional theory , the condition ( [ adecond ] ) would seem very natural because it implies the vanishing of the nsvz beta function and the existence of an ir superconformal fixed point . in this @xmath39 dimensional context , it is less clear where such a condition comes from .
gaiotto and witten @xcite call such quivers balanced . in their study of @xmath40 yang - mills theories in 2 + 1
dimensions ( without cs terms ) , they find that the coulomb branch of the moduli space for balanced quivers has an enhanced global symmetry . when the ranks @xmath41 of the gauge groups are all equal , the dynkin diagram is of @xmath4 type and was studied in detail in our earlier papers @xcite .
however , one may also try to solve the matrix model for the @xmath42 and @xmath43 quivers . in section [ sec : seiberg ] , we discuss how generalized seiberg duality acts on the @xmath4 , @xmath42 , and @xmath43 quivers in the large @xmath3 limit .
in particular we investigate how the saddle point eigenvalue distribution transforms under seiberg duality . in section [ sec : d4 ] , we investigate the @xmath5 case in detail .
we find that the relation ( [ evchiralrel ] ) between @xmath44 and the chiral ring continues to hold .
a discussion section contains some partial results and conjectures . in appendix
[ app : simple ] , we find a simple class of solutions for any ade quiver .
0.1 in * note added : * after this paper appeared , we became aware of @xcite .
section 4 of @xcite has some overlap with section [ sec : seiberg ] of this paper .
we are interested in chern - simons theories with matter and at least @xmath0 supersymmetry in 2 + 1 dimensions .
we assume the gauge group to have the product structure @xmath45 .
in other words , for each @xmath36 factor , there exists a vector multiplet transforming in the corresponding adjoint representation .
we denote the chern - simons level for each @xmath36 factor @xmath46 .
in addition to the vector multiplets , we allow the theory to contain hypermultiplets in bifundamental representations of the gauge groups .
more precisely , in @xmath23 language , the hypermultiplet consists of a pair of chiral multiplets @xmath47 transforming in conjugate bifundamental representations , @xmath48 and @xmath49 .
the field content of such a theory is conveniently described with a quiver . for each @xmath36
, we draw a vertex . for each pair
@xmath47 , we join nodes @xmath50 and @xmath51 with an edge .
let @xmath37 be the set of edges .
recall that an @xmath0 vector multiplet contains an @xmath23 vector multiplet @xmath52 and a @xmath23 chiral multiplet @xmath53 transforming in the adjoint representation of the gauge group .
the scalar component @xmath54 of @xmath53 combines with the auxiliary scalar @xmath55 in @xmath52 to form a triplet under the @xmath56 symmetry .
similarly , the pair @xmath47 form a doublet .
the @xmath23 formalism leaves only a @xmath57 of the r - symmetry manifest . under this @xmath58
, @xmath53 has charge one and @xmath47 has charge 1/2 .
we would like to compute the euclidean partition function of such an @xmath0 chern - simons theory on an @xmath2 . as explained by kapustin , willett , and yaakov @xcite , this path integral localizes to configurations where the scalars @xmath22 in the vector multiplets are constant hermitian matrices . denoting the eigenvalues of @xmath22 by @xmath59 , @xmath60 ,
the partition function takes the form of the eigenvalue integral z
= ( _ a , i d _ a , i ) l_v(\{_a , i } ) l_m(\{_a , i } ) = ( _ a , i d _ a , i ) , [ generalz ] where the vector multiplets contribute l_v = _ a=1^d ( _ i > j 2 ) ^2 ( i _ a , j k_a _ a , j^2 ) and the bifundamental matter fields contribute l_m = _ ( a , b ) e _ i , j .
we wish to approximate this integral using the saddle point method in the limit where all of the @xmath41 are large .
in particular , we will assume that @xmath61 where the @xmath62 are relatively prime integers and @xmath63 .
the saddle point equation for the eigenvalue @xmath59 is - = 2 i k_a _ a , i + _ j i 2 - _ b | ( a , b ) e _ j = 0 .
[ saddle point ] guided by numerical experiments , we will assume that we can pass to a continuum description where each set of @xmath41 eigenvalues lie along @xmath62 curves , @xmath3 eigenvalues per curve : _ a , i(x ) = n^x + i y_a , i(x ) , [ evalansatz ] where @xmath64 and @xmath65 .
the density of eigenvalues @xmath44 is assumed to be independent of @xmath50 and @xmath66 , and we normalize the density such that @xmath67 . at large @xmath3 , we may use the approximations ( x_i - x_j ) . assembling these large @xmath3 approximations , the leading order term in the saddle point equation ( [ saddle point ] )
is ( assuming @xmath68 ) ( 2 n_a - _ b | ( a , b ) e n_b ) n dx ( x - x ) ( x ) .
for this term to vanish , a sufficient condition is that 2 n_a = _ b | ( a , b ) e n_b . [ confcond ] to find saddle points for more general theories , one obvious thing to try is a more general ansatz for the eigenvalue distribution , in particular one that does not assume @xmath44 is the same for each gauge group .
the condition ( [ confcond ] ) is surprisingly stringent .
the only allowed quivers that satisfy this condition are the extended ade dynkin diagrams ! in this context , the @xmath62 are called the marks . to each node of the dynkin diagram , we associate a simple root @xmath69 . for an affine dynkin diagram
, we have one additional root @xmath70 associated to the extra node .
this extra root , which is the highest root in the adjoint representation , satisfies the condition @xmath71 .
the mark associated with @xmath70 is defined to be equal to one . in the rest of the paper we will assume that ( [ confcond ] ) holds .
next we examine the large @xmath3 limit of @xmath12 itself , assuming ( [ confcond ] ) .
it is convenient to separate the free energy into pieces : @xmath72 where f_tree = -i _ a , i k_a _ a , i^2 .
the @xmath73 contribution will turn out to be subleading in @xmath3 , and we hereby drop it . in the large @xmath3 limit , we find that f_tree = n _ a=1^d .
[ ftree ] the loop contribution to @xmath12 is @xmath74 \nonumber \\ & & + \sum_{(a , b ) \in e } \sum_{i=1}^{n_a } \sum_{j=1}^{n_b } \ln 2 \cosh [ \pi n^\alpha ( x - x ' ) + i \pi ( y_{a , i}(x ) - y_{b , j}(x ' ) ) ] \biggr ) \ . \ ] ] this integral is dominated by the region @xmath75 which suggests we change variables from @xmath76 to @xmath77 so that to a first approximation at large @xmath3 we may write @xmath78 .
the integral reduces to @xmath79 \bigr ) \ .\end{aligned}\ ] ] given ( [ confcond ] ) , the integral over @xmath80 can be performed and is convergent .
the following two intermediate results are useful : @xmath81 \ , dx- m^2 \right ] & = & \frac{\pi^2}{12 } - \frac{1}{4 } \arg \left ( e^{2 i ( a - \pi/2 ) } \right)^2 \ , \\
\lim_{m \to \infty } \left [ \int_{-m}^m \ln 2 \cosh(x + i a)\ , dx - m^2 \right ] & = & \frac{\pi^2}{12 } - \frac{1}{4 } \arg \left ( e^{2 i a } \right)^2 \ .\end{aligned}\ ] ] the end result is that @xmath82 \ .\end{aligned}\ ] ] to find a nontrivial saddle point , we have two choices .
we can balance @xmath83 against the first term in ( [ ftree ] ) , in which case @xmath84 .
alternately , we can assume that @xmath85 and balance @xmath83 against the second term of ( [ ftree ] ) in which case @xmath86 .
the first case has a massive type iia supergravity dual @xcite . )
needs to be slightly modified in this case to allow both the real and imaginary parts of @xmath87 to scale with @xmath88 .
see @xcite . ]
we shall make the second choice , leading to an eleven dimensional supergravity dual .
the free energy that we must maximize is then @xmath89 dx - 2 \pi \mu n^{3/2 } \left ( \int \rho(x ) \ , dx- 1 \right ) \end{aligned}\ ] ] where we have added a lagrange multipler @xmath32 to enforce the constraint that the eigenvalue density integrates to one .
we believe that the free energy can always be written in a way such that it depends only on the average values y_a = _
i y_a , i rather than on the @xmath90 individually .
in particular , we conjecture that @xmath12 will take the form @xmath91 dx \nonumber \\ & & - 2 \pi \mu n^{3/2 } \left ( \int \rho(x ) \ , dx -1 \right ) \ .
\label{conjf } \end{aligned}\ ] ] we have verified this conjecture in the case of the @xmath4 and @xmath5 quivers .
( indeed , in the @xmath4 case , this result was given in @xcite . )
note we are not claiming that the differences @xmath92 remain bounded between @xmath93 . indeed , for the @xmath5 case
, the differences are not bounded .
we are arguing that a more subtle cancellation happens among the integer shifts introduced in evaluating the arg functions such that ( [ finalf ] ) reduces to ( [ conjf ] ) . ) and the second term of ( [ conjf ] ) do not have the same periodicity properties as functions of the @xmath90 . in the limit @xmath94 , we do not expect ( [ finalf ] ) and ( [ conjf ] ) to be equal .
what we are claiming is a region in @xmath90 space where ( [ finalf ] ) and ( [ conjf ] ) are equal and which contains the saddle point eigenvalue distribution . ]
modulo some subtle convergence issues , refs .
@xcite demonstrated that the absolute value @xmath9 of the partition function ( [ generalz ] ) is invariant under a generalized seiberg duality @xcite .
for a @xmath95 theory with @xmath96 flavors and cs level @xmath97 , this generalized seiberg duality acts by sending @xmath98 and @xmath99 . in the context of our quiver theories , seiberg duality can be performed on any individual node of the quiver .
the bifundamental fields look like flavors from the point of view of the node to be dualized .
we have the long range force cancellation condition ( [ confcond ] ) which implies that @xmath100 .
we also work in the large @xmath3 limit which implies that @xmath101 .
thus , under seiberg duality @xmath102 is invariant to leading order in @xmath3 . in the quiver context ,
in addition to changing the cs level of the dualized node , the duality also changes the cs levels of the neighboring nodes . if we dualize node @xmath50 , then for @xmath103 , @xmath104 .
note that the sum @xmath105 is invariant under the duality given ( [ confcond ] ) . to motivate why the cs levels change in the way they do
, we recall the brane construction of the @xmath4 quivers @xcite . we start with type iib string theory with the 6 direction periodically identified .
we place @xmath3 d3-branes in the 0126 directions .
intersecting these d3-branes at intervals in the 6 direction , we insert @xmath106 5-branes that span the 012 directions as well as lines in the 37 , 48 , and 59 planes that make angles @xmath107 with the 3 , 4 , and 5 , axes respectively . the choice of @xmath108 guarantees that the construction has the right number of supercharges for the 2 + 1 dimensional gauge theory living on the d3-branes to have @xmath0 supersymmetry .
each unbroken d3-brane interval in the 6 direction corresponds to a @xmath1 gauge group in the @xmath4 quiver .
the cs level of the gauge group is given by the difference @xmath109 .
seiberg duality corresponds to interchanging neighboring @xmath110 5-branes .
quivers @xcite . ] for the simply laced affine dynkin diagrams @xmath4 , @xmath42 , and @xmath43 , seiberg duality can be reinterpreted as the action of the weyl group .
( there is a similar story for 3 + 1 dimensional gauge theories with @xmath111 susy @xcite . ) to @xmath112 of the gauge groups we associate the corresponding simple root @xmath113 of the dynkin diagram .
the @xmath114st gauge group is assigned the root @xmath115 where @xmath70 is the highest root .
the cs level is @xmath116 where @xmath117 is an arbitrary weight vector .
note that _ a n_a k_a = - p+ _
n_a _ a p = 0 , since @xmath118 .
( the mark associated with @xmath70 is defined to be one . ) returning to the brane construction for the @xmath4 quivers , the weyl group is the permutation group acting on the @xmath106 branes .
we start by assigning to the @xmath114 gauge groups the @xmath112 simple roots @xmath119 with @xmath120 a unit coordinate vector in @xmath121 and the highest root @xmath122 .
note that these @xmath114 roots live in a @xmath123 subspace of @xmath121 .
let @xmath124 be a weight vector corresponding to the charges of the 5-branes .
as claimed , the cs levels are then @xmath116 . the results of refs .
@xcite imply that @xmath9 is invariant under this generalization of seiberg duality for any @xmath3 .
the large @xmath3 limit ( [ finalf ] ) must also be invariant , but it is still interesting to see how the invariance arises in our case .
if we shift @xmath125 , @xmath104 for @xmath103 , then the first term of can be made invariant if we take _
i y_a , i -_i y_a , i + _ b|(a , b ) e _
j=1^n_b y_b , j .
[ imprecisesd ] in the case when @xmath126 , it is clear that not just the first term of ( [ finalf ] ) but all of ( [ finalf ] ) is invariant under such a transformation .
for example , for the @xmath4 quivers , the transformation on @xmath90 reduces to y_a - y_a + y_a-1 + y_a+1 . to understand seiberg duality on the internal nodes of the @xmath42 and @xmath43 quivers
, we should describe how each @xmath90 transforms rather than how the average transforms .
however , it is straightforward to demonstrate that the conjectured form of the free energy ( [ conjf ] ) is invariant under ( [ imprecisesd ] ) .
thus , if we believe ( [ conjf ] ) , i.e. that the free energy depends only on the average values @xmath127 , then @xmath12 is invariant under our generalized seiberg duality .
we should note that our large-@xmath3 approximation is insensitive to shifting all of the @xmath128 s by the same amount .
the true seiberg duality is actually a combination of with such a shift .
( see section [ sec : numerics ] for an example of this phenomenon . )
let us examine the @xmath5 type quiver .
let the external vertices have cs levels @xmath46 , @xmath129 and the internal vertex have cs level @xmath97 where @xmath130 .
we label the imaginary part of the eigenvalues of the internal vertex @xmath131 and @xmath132 and define @xmath133 . based on numerical simulation and intuition from the @xmath134-type quivers @xcite
, we expect the saddle point eigenvalue distributions are piecewise smooth functions of @xmath135 .
the action of generalized seiberg duality separates the space of cs levels into different weyl chambers .
it is enough to analyze the theory in a given weyl chamber .
instead we will make the choice @xmath136 which includes three different weyl chambers . at the end
, we will be able to see explicitly how seiberg duality acts on the solution .
in the central region @xmath137 , the equations of motion yield ( x ) = ; y_a , i ( x ) = .
once @xmath138 , the solution gets more complicated . in general one of the @xmath139 , which for convenience we choose to be @xmath140 ,
will saturate at @xmath141 and the other will continue to grow such that @xmath142 .
there are three sub - cases to consider : 1 ) @xmath143 , 2 ) @xmath144 and 3 ) @xmath145 .
let @xmath146 .
the next three piecewise linear regions of the solution are @xmath147 where in cases 1 and 2 , @xmath148 and in case 3 , @xmath149 .
the fifth and final piecewise linear region of the solution is different in each of the three cases . in cases 1 and 2
, we find @xmath150 where @xmath151 in case 1 and @xmath152 at the endpoint . in constrast , in case 2 , @xmath153 and @xmath154 at the endpoint . in case 3
, we have instead @xmath155 in the three cases , the value of the lagrange multiplier is @xmath156 a virial theorem proven in @xcite asserts that @xmath157 , and hence from ( [ fvolrel ] ) it follows that @xmath158 .
we can be more precise about what exactly the seven dimensional manifold @xmath17 is .
it is a tri - sasaki einstein space .
a tri - sasaki einstein space can be defined as the level surface of a conical manifold with hyperkhler structure @xcite .
the eight dimensional hyperkhler cone in turn can be thought of as a hyperkhler quotient @xcite of the the quaternionic manifold @xmath159 by a subgroup @xmath160 .
we can define the two special @xmath28 s that do not appear in the quotient using the gauge connections @xmath161 and @xmath162 of the @xmath163 and @xmath164 groups respectively .
one is the diagonal subgroup @xmath165 .
the other is picked out by the cs levels , @xmath166 .
see for example @xcite for more details .
these three cases are related by the seiberg duality described in the previous section .
consider a seiberg duality on the central node , then on node 1 , and then again on the central node , @xmath167 .
under such a transformation the cs levels change such that ( k_1 , k_2 , k_3 , k_4 ) ( k_1 + k_2 + k_3 + k_4 ,
k_1 + k_2 -k_3 - k_4 , k_1 - k_2 + k_3 - k_4 , k_1 -k_2 -k_3 + k_4 ) . given these relations , it is straightforward to verify that case one is mapped to case two and vice versa .
the map @xmath168 takes case three into case one with the rule ( k_1 , k_2 , k_3 , k_4 ) ( k_1 + k_2 + k_3 + k_4 , k_1 + k_2 -k_3 - k_4 , k_1 - k_2 + k_3 - k_4 , k_1 -k_2 -k_3 + k_4 ) . to express the action of seiberg duality in terms of the weyl group , we need to introduce simple roots for @xmath5 : @xmath169 , @xmath170 , @xmath171 , and @xmath172 .
note that @xmath173 corresponds to the central node .
the highest root is then @xmath174 and the cs levels can be expressed in terms of auxiliary variables @xmath175 such that k_1 = p_1 + p_2 , k_2 = -p_1 + p_2 , k_3 = -p_3 + p_4 , k_4 = -p_3 - p_4 .
the weyl group acts by permuting the @xmath175 and by changing the sign of an even number of the @xmath175 . also has an outer automorphism group isomorphic to @xmath176 , corresponding to permutations of three of the @xmath97 s .
these automorphisms can not be expressed in terms of seiberg dualities .
swapping @xmath177 and @xmath178 negates @xmath179 ; other permutations act less nicely on the @xmath180 .
naively , there is an @xmath181 , but the four dimensional subgroup generated by the swaps @xmath182 , @xmath183 , and @xmath184 are seiberg dualities . ]
we can write a more general expression for @xmath32 , assuming @xmath185 : @xmath186 to check our large @xmath3 approximation , we can use a computer to calculate the eigenvalue distributions numerically .
we use the same relaxation algorithm described in @xcite . for the simple seiberg dual cases @xmath187 and
@xmath188 , we plot the answer in figure [ fig : comptonum ] .
note that these cases are somewhat degenerate . instead of having five distinct piecewise linear regions ( for @xmath189 )
, we have only two for the @xmath190 theory and one for the @xmath188 theory . for @xmath190
, the last three regions collapse to the point @xmath191 , while for @xmath188 the last four regions collapse to the point @xmath192 .
the thin black lines are the analytic , large @xmath3 prediction .
the agreement is reasonably good and should improve as @xmath3 is increased .
note however that we predict only the differences between the @xmath193 . to make the plot , we first fit @xmath194 to a straight line . given the value of the fit ( @xmath195 for @xmath190 and @xmath196 for @xmath188 ) , the transformation ( [ imprecisesd ] ) is obeyed only up to an overall additive shift of all the @xmath193 .
we would like to count the number of gauge invariant chiral operators of the @xmath5 quiver when @xmath197 .
to that end , we should first construct words out of the bifundamental fields @xmath198 and @xmath199 that are invariant under the @xmath200 gauge group .
the indices @xmath201 correspond to the four @xmath28 gauge groups while @xmath202 index the @xmath164 group .
then we quotient by the superpotential ( or moment map ) relations : @xmath203 we use the convention that we sum over repeated @xmath164 indices , but we do not sum over repeated @xmath28 indices . finally , we restrict to operators with @xmath28 charges that can be canceled by the corresponding diagonal monopole operators which have charge @xmath204 under the @xmath163 , where @xmath30 is an integer .
( there is a sense in which what we do here is a generalization of the analysis performed in @xcite . )
our counting problem is equivalent to a much simpler counting problem where we ignore the @xmath205 and @xmath206 fields .
let @xmath207 denote the ring of operators .
for any @xmath207-module @xmath208 , let @xmath209 denote the hilbert series of @xmath208 .
for any ideal @xmath66 , we have @xmath210 if we let @xmath211 , then @xmath207 and @xmath66 are isomorphic @xmath207-modules and @xmath212 .
so @xmath213 in terms of the function @xmath31 defined in the introduction , now with a subscript to indicate the choice of ring , we have from ( [ evchiralrel ] ) @xmath214 when @xmath215 , eqs .
( [ firstabrel ] ) and ( [ secabrel ] ) become @xmath216 all @xmath217 invariant operators may be written as polynomials in @xmath218 , @xmath219 , and @xmath220 .
these polynomials satisfy various quadratic relations .
some of these relations can be obtained by contracting with bifundamental fields ( either @xmath221 or @xmath222 for upper indices , and @xmath223 or @xmath224 for lower indices ) .
we can obtain other relations by contracting bifundamental fields with the identity @xmath225 once we apply , each of these relations will have zero , two , or three nonzero terms .
we first mod out by the relations that have two terms .
we get two - term relations by contracting with two fields with different gauge indices , and by contracting with four fields with two or three different gauge indices .
modding out by relations with two terms gives us a seven dimensional toric variety .
the two - term relations are all invariant under a @xmath226 action .
we will let the first four charges @xmath227 be the charges under the @xmath28 gauge groups .
we choose @xmath228 so that each contraction of an @xmath229 or @xmath230 with an @xmath231 or @xmath232 and each contraction of an @xmath233 or @xmath234 with an @xmath235 or @xmath236 has @xmath237 and all other contractions have @xmath238 . likewise , we choose @xmath239 to count the number of @xmath240-@xmath241 and @xmath242-@xmath243 contractions , and @xmath244 to count @xmath240-@xmath243 and @xmath242-@xmath241 contractions .
note that @xmath245 . also , the following quantitites are even : q_1+q_2+q_3+q_4 , q_1 + q_2 + q_6+q_7 , q_1 + q_3 + q_5 + q _ 7 , q_1 + q_4 + q_5 + q_6
. we can check that all contractions of a pair of @xmath134 s and @xmath246 s can be expressed in terms of @xmath220 and @xmath247 .
we obtain an expression for @xmath248 by contracting with two @xmath229 s , a @xmath232 , and a @xmath249 .
we can then obtain @xmath250 by contracting with @xmath249 and @xmath251 , where @xmath252 are all different . finally , contracting with one
each of @xmath253 gives us @xmath219 .
therefore , the variety must be seven dimensional . for simplicity and to make contact with `` case one '' discussed above
, we will assume that @xmath254 , @xmath255 . by construction
, we have the inequalities @xmath256 .
we claim the following inequalities also hold , @xmath257 consider the first of these inequalities .
the quantity @xmath29 is half the total number of @xmath134 s and @xmath246 s in the operator .
the quantity @xmath258 is the number of @xmath229 and @xmath231 fields that we can not pair with with @xmath230 and @xmath232 fields respectively .
the @xmath217 indices of these excess @xmath259 fields must be contracted in a gauge invariant way . to minimize @xmath228 we can try to contract as many of these excess fields as possible with the @xmath233 , @xmath234 , @xmath235 , and @xmath236 fields . however ,
if the excess number of @xmath229 and @xmath231 fields is larger than @xmath29 , then we will still have at least @xmath260 @xmath229 and @xmath231 fields left over which we will be forced to contract together , yielding the lower bound on @xmath228 .
now we impose the three - term relations that can be derived from and .
these relations break the symmetries corresponding to @xmath261 ; only the sum @xmath262 is preserved .
we will let @xmath263 denote the heaviside step function , and @xmath264 .
any monomial that has @xmath244 greater than the minimum @xmath265 can be reduced to a sum of monomials with @xmath266 . therefore , the density of operators with a given @xmath267 is @xmath268 the factor of @xmath269 comes from the fact that @xmath270 is always even .
if @xmath271 , then in piecewise form we have @xmath272 when we replace @xmath273 with @xmath274 and multiply by @xmath275 , we get agreement with the densities computed by the matrix model
. we may also compare the @xmath276 values with numbers of operators in the chiral ring . a conjecture in @xcite states that for a bifundamental field @xmath277 that transforms in the anti fundamental of the @xmath50th @xmath1 gauge group and the fundamental of the @xmath51th group , _
r / ( x_ab)^(1,1 ) ( r , rx/ )
= ( x ) [ y_b(x ) - y_a(x ) + r(x_ab ) ] where @xmath278 is the r - charge of @xmath277 . in this case , @xmath279 because of the @xmath0 susy .
first , we count the operators that are not divisible by @xmath280 . dividing by @xmath280 decreases @xmath228 by one .
since all operators have @xmath281 , any operator with @xmath238 is not divisible by @xmath280 .
we can check that all other operators are divisible by @xmath280 . since @xmath282
, operators with @xmath238 exist only if @xmath283 .
therefore , @xmath284 the factor of @xmath269 comes from the fact that @xmath285 must be even .
therefore , operator counting predicts that @xmath286 we can check that this agrees with our matrix model calculations .
similarly , @xmath287 finally , we count the operators that are not divisible by @xmath288 . dividing by @xmath288 decreases @xmath289 , @xmath228 and @xmath29 by one and increases @xmath290 by one .
since @xmath228 and @xmath291 must both be nonnegative , operators with @xmath238 or @xmath292 are not divisible by @xmath288 .
there are two cases to consider .
if @xmath238 , then for @xmath293 , the value of @xmath294 must be 1/2 as we had before . the second case to consider is @xmath295 .
this constraint implies that @xmath296 is even and we do not need an additional factor of 1/2 .
all operators must satisfy the constraint @xmath297 .
thus in the range @xmath298 , the value of @xmath294 must be one . in summary
, we find @xmath299 operator counting predicts that @xmath300 and this agrees with our matrix model results .
in this paper , we saw that there is an ade classification of @xmath0 susy cs theories in 2 + 1 dimensions with bifundamental matter fields whose partition function @xmath7 has a nice large @xmath3 limit . by nice limit , we mean that when we use the matrix model of @xcite to calculate @xmath7 , there is cancellation in the long range forces between the eigenvalues .
the classification raises a number of questions which we were not able to answer in this paper .
leading among them is the significance of the @xmath100 condition ( [ adecond ] ) . for a 3 + 1 dimensional gauge theory ,
this condition implies the vanishing of the nsvz beta function and the existence of a conformal fixed point , but in the present context the meaning is obscure .
it is tempting to speculate that @xmath301 is necessary in 2 + 1 dimensions for the existence of the fixed point .
however , it could be that the condition is just a technical limitation of the method introduced in @xcite for calculating the large @xmath3 limit of the matrix model .
another question has to do with the restriction to @xmath1 gauge groups
. it would be interesting to study quivers of @xmath302 and @xmath303 groups in the large @xmath3 limit .
we conjecture that `` good '' theories can all be described by extended dynkin diagrams .
we believe that the `` good '' quivers containing just @xmath302 and @xmath303 gauge groups are simply laced extended dynkin diagrams , while the `` good '' quivers that also contain @xmath1 are the ones shown in figure [ fig : sosp ]
. it would be interesting to study this correspondence further .
we note that an @xmath304-@xmath305 version of the simply laced @xmath229 quiver was studied in @xcite .
we would like to end this paper by describing some progress we made in writing @xmath12 in a way that is manifestly invariant under seiberg duality . in the introduction , we reviewed the fact that @xmath306 where @xmath17 is the tri - sasaki einstein space in the dual eleven dimensional supergravity description of the cs theory . in the @xmath4 and @xmath5 cases
, we have found that @xmath307 can be written as a rational function of the cs levels @xmath46 .
the numerator of this function takes the form _ _ 1 , , _
n r ( _ 1 , ,
_ n)^2 _ a=1^n |_a p| where @xmath207 is the set of roots of the @xmath4 , @xmath42 , or @xmath43 lie algebra .
recall that the cs levels can be reconstructed from the simple roots via @xmath116 .
we would like to thank d. berenstein , i. klebanov , j. mcgreevy , t. nishioka , s. pufu , m. roek , and e. silverstein for discussion .
this work was supported in part by the us nsf under grants no .
phy-0844827 , phy-0756966 , and phy-0551164 .
ch thanks the kitp for hospitality and the sloan foundation for partial support .
in this section , we find a simple class of solutions for any ade quiver .
let us start by looking at the @xmath308 quiver . at the risk of unacceptably proliferating the number of indices , we label the cs levels of the external nodes @xmath309 , of the internal nodes @xmath310 , and of the central node @xmath311 with the constraint @xmath312 .
the imaginary parts of the eigenvalues are correspondingly labeled @xmath313 , @xmath314 and @xmath315 . for small enough @xmath135 , the equations of motion yield ( x )
= ; y_1a - y_2a , i = ; y_2a , i - y_3,j = . by small enough ,
we mean that no difference between the @xmath128 s has saturated to @xmath93 .
more precisely , we require @xmath316 and @xmath317 for all @xmath50 . as in the @xmath5 case , for a certain class of solutions the eigenvalue distribution ends inside this central region .
let us try to extremize the free energy as a function of the endpoint @xmath318 .
the on shell free energy as a function of @xmath318 is f(x _ * ) = n^3/2 . which leads to the condition that
x _ * = ^-1/4 .
in order for @xmath318 to lie in the central region , we must take @xmath319 or @xmath320 , leading to the final result f = n^3/2 4 , and the result for the volume ( y ) = . the tri - sasaki einstein manifold in question is a level surface of the hyperkhler quotient @xmath321 .
the solution above has some characteristic features that we can try to generalize for arbitrary ade quivers .
the @xmath322 and @xmath323 all saturate to @xmath93 at @xmath324 . also , the absolute values of the @xmath325 are equal , @xmath326 . for the arbitrary ade case ,
let the cs level of the @xmath327 gauge group be @xmath328 , where we choose @xmath329 so that neighboring gauge groups have opposite sign .
this choice of cs levels is consistent with the condition @xmath85 .
the equations of motion following from ( [ finalf ] ) are satisfied if we set all the @xmath330 equal to each other and the eigenvalue density to a constant : y_ab , ij = ( a ) , = . with a little more work
, we can find the chemical potential and the volume of the associated tri - sasaki einstein manifold ( y ) = = ( ) ^2 . | we consider @xmath0 supersymmetric chern - simons ( cs ) theories that contain product @xmath1 gauge groups and bifundamental matter fields .
using the matrix model of kapustin , willett and yaakov , we examine the euclidean partition function of these theories on an @xmath2 in the large @xmath3 limit .
we show that the only such cs theories for which the long range forces between the eigenvalues cancel have quivers which are in one - to - one correspondence with the simply laced affine dynkin diagrams .
as the @xmath4 series was studied in detail before , in this paper we compute the partition function for the @xmath5 quiver .
the @xmath5 example gives further evidence for a conjecture that the saddle point eigenvalue distribution is determined by the distribution of gauge invariant chiral operators .
we also see that the partition function is invariant under a generalized seiberg duality for cs theories .
pupt-2396 + nsf - kitp-11 - 220 daniel r. gulotta , j. p. ang , and christopher p. herzog department of physics , princeton university + princeton , nj 08544 , usa dgulotta , jang , [email protected] + |
refsdal ( 1964a 1964b 1966 ) outlined a method for determining a global value for the hubble constant by measuring the light travel delay between different images of a gravitationally lensed system . together with a model that describes the gravitational potential of the lens
, the time delay defines the geometrical dimensions of the lens system and therefore may lead to the determination of the hubble constant .
the criteria that make a gravitational lens system ( gls ) suitable for measuring the hubble constant can be summarized as follows : ( a ) the sky positions and redshifts of the lens and of the images of the lensed object should be measurable accurately ; ( b ) the mass distribution of the lens should be subject to accurate and reliable estimation . when the lens has multiple components , as for q0957 + 561 , this estimation becomes especially difficult and benefits from many different types of observations and also from many and extended images , such as rings .
for example , if the types of observations include radio and optical emission lines , which are less sensitive to microlensing , flux ratio measurements are more useful in constraining the lens model ; and ( c ) the differences in propagation time from the lensed object to us through ( at least two ) different images ( `` time delay '' ) should be measurable with high accuracy .
the source , i.e. , the lensed object , must therefore be variable on time scales far shorter than the time delays , to allow accurate estimates to be made of the latter .
several potential candidates for determining @xmath7 are the quasar q0957 + 561 with two images and a measured time delay of 1.1 years ( pelt et al .
1996 ; haarsma et al . 1997 ; kundi et al .
1997 ) , the quadruple image system pg 1115 + 080 with a measured time delay between components b and c of about 25 days ( schechter et al . 1997 ; bar - kana 1997 ) , and b0218 + 357 with a time delay of 12 @xmath8 3 days ( geiger & schneider 1996 ; corbett et al .
1996 ) . combining measured time delays with a detailed lensing model , refsdal
s suggestion has been applied to q0957 + 561 ( falco , gorenstein , & shapiro 1991 ; grogin & narayan 1996 , hearafter gn ) and pg1115 + 080 ( keeton & kochanek 1997 ) . in this paper we focus on a study of the properties of q0957 + 561 ( walsh , carswell & weyman 1979 ) , one of the most extensively studied lensing systems which has been monitored since its discovery in both the optical and radio ( see , for example , schild & thomson 1995 ; vanderriest et al .
1989 ; haarsma et al .
the primary lensing components are an elliptical cd galaxy at a redshift of 0.355 , usually referred to as the g1 galaxy , a cluster of galaxies containing the g1 galaxy , and a group of galaxies at a redshift of 0.5 .
the most thorough optical spectroscopic and photometric study to date of the field around q0957 + 561 has been presented by angonin - willaime , soucail , & vanderriest ( 1994 ) .
one of the major concerns in modeling q0957 + 561 is a mass sheet degeneracy which arises from the presence of the cluster of galaxies at z=0.36 . in particular
if one were to modify the assumed radial mass density profile @xmath9 of the lens to @xmath10 + ( 1 - @xmath11 ) , where 1 - @xmath11 represents a constant surface mass density sheet term , then the new mass distribution would also satisfy the observational constraints .
however the resulting value for @xmath7 would be scaled by the constant 1 - @xmath11 .
since @xmath11 must always be positive one may always determine an upper limit on the hubble constant for @xmath11 = 0 ( falco et al .
1991 ) . as also noted by falco et al .
( 1991 ) and later gn , this degeneracy may be broken if one were to measure the velocity dispersion or total mass distribution of either the cluster of galaxies or the principal lensing galaxy g1 .
recently falco et al . (
1997 ) have obtained a velocity dispersion for the central lens galaxy g1 of @xmath12 km s@xmath6 or @xmath13 km s@xmath6 depending on the interpretation of the spectroscopic slit data .
+ in this paper we focus on improving the cluster model and we present an estimate of the total mass of this lensing cluster of galaxies through x - ray measurements of thermal emission from the intracluster gas . in section 2
we describe the spatial analysis of the x - ray data of q0957 + 561 obtained from several rosat hri observations and we present an estimate for the convergence parameter @xmath9 of the cluster .
section 3 describes the spectroscopic analysis of rosat position sensitive proportional counter ( pspc ) and asca gas imaging spectrometer ( gis ) x - ray observations at different epochs in order to investigate the variability of different spectral components , and presents various scenarios to explain the observed variability in the x - ray flux .
finally section 4 presents our conclusions from the analysis of the x - ray observations of the q0957 + 561 gl system .
in particular , we discuss how our results allow bounds to be placed on the hubble constant , albeit with these bounds still subject to systematic errors that we can not yet evaluate accurately .
the limits on the mass and luminosity of the cluster of galaxies that contains the primary lensing cd elliptical galaxy g1 presented in chartas et al .
( 1995 ) were based on the rosat hri observation of q0957 + 561 from 1992 may 1 .
we now estimate the mass and luminosity of this cluster utilizing additional data obtained from the three subsequent rosat hri observations in 1992 october 20 , 1992 october 22 , and 1994 april 11 ( see table 1 ) .
the october 20 rosat hri observation is centered on the galaxy ngc 3079 with the quasar q0957 + 561 located 12 arcmin off - axis . because of the dependence of the properties of the point spread function ( psf ) with off - axis angle , we performed the analysis for this off - axis observation separately and combined the results with those obtained from the on - axis hri observations of q0957 + 561 .
to look for extended cluster emission centered on the galaxy g1 , we first merged the on - axis observations of 1994 april , 1992 october , and 1991 may .
the image files containing the data for these observations were shifted appropriately such that the center of q0957 + 561 image b from each observation was co - aligned .
the shifts were relatively small , of the order of a few arcsecs .
figure 1 shows the resulting merged image of the three hri observations .
x - ray emission surrounding the lensed images can be seen . to determine the significance of this extended emission and whether it originates from hot cluster gas emission or from the psf wings of the lensed images , we searched for a statistically significant excess in detected counts originating from cluster emission around the g1 galaxy .
in particular we determined the number of counts within annuli of inner radii ranging from 15@xmath14 to 100@xmath14 in steps of 5@xmath14 and outer radius of 120@xmath14 centered on q0957 + 561 for the merged rosat hri observations and separately for the off - axis observation .
the background was taken from an annulus of inner and outer radii of 340@xmath14 and 460@xmath14 .
for inner radii less than about 25@xmath14 the contribution from the psf s of images a and b of q0957 + 561 is significant whereas for inner radii greater than about 60@xmath14 the signal to noise drops considerably leading to a large uncertainty in the determined source counts , as seen in figure 2 .
we modeled two point sources at quasar image locations a and b. each modeled source psf was normalized such that the number of counts within a radius of 3@xmath14 from the center of the modeled source was equal to the number of counts within a radius of 3@xmath14 of the observed source .
the net counts @xmath15 detected by the rosat hri originating from cluster emission within an annulus of inner radius @xmath16 and outer radius @xmath17 were determined by subtracting background and point source contributions and is given by the expression , @xmath18 where @xmath19 are the background counts in the annular region @xmath20 and @xmath21 , @xmath22 are the counts within the region @xmath20 originating from the a and b quasar images of q0957 + 561 , respectively . to compute the total cluster counts , we estimated a correction term for the cluster counts beyond the range of the inner and outer radii of the annulus ( as described in chartas et al .
the correction term depends on the adopted value for the cluster core radius and cluster radius . for our calculations we consider core radii ranging between @xmath23 and @xmath24 and cluster radii ranging between 120@xmath14 and 280@xmath14 . for a typical cluster core radius of @xmath25 ( 0.067 mpc , assuming @xmath26 km s@xmath6 mpc@xmath6 and @xmath27 ) , we find that the number of counts within a 20@xmath14 circle centered on q0957 + 561 is about 60@xmath28 of the counts found in the annulus with inner radius of 20@xmath14 and outer radius of 120@xmath14 , whereas for @xmath29 we find that about 18 @xmath28 of the x - ray counts originate within 20@xmath14 .
no x - ray sources detected within the hri field of view lie within the annuli used as source and background extraction regions .
for the four hri observations of q0957 + 561 with a total exposure time of 85154sec , we estimate that a total of about 300 @xmath8 100 counts ( with the quoted error being at the 1@xmath0 level ) originated from cluster emission and approximately 2900 counts from the quasar images . to convert the measured hri counts into luminosities we have assumed a raymond - smith spectrum characterized by a temperature @xmath30 , a galactic column density @xmath31 = @xmath32 ( stark et al . 1992 ) , 30@xmath28 cosmic abundances ( see , e.g. , henriksen 1985 ; hughes et al . 1988 ; and arnaud et al . 1987 ) and the @xmath33 - @xmath30 correlation determined by david et al .
the relation between cluster luminosity and temperature for a value of the hubble constant of @xmath7 = 100h km s@xmath6 mpc@xmath6 is , @xmath34 with @xmath35 = 0.29 .
there is considerable scatter in the observational data used by david et al .
( 1993 ) to produce the @xmath33 vs. @xmath30 correlation .
a recent study however by markevitch ( 1998 ) has shown that the scatter in the @xmath33 - @xmath30 relation is greatly reduced when the emission from cooling flow regions is excluded in the analysis . to derive the cluster mass , we assume a typical density profile @xmath36 of the hot gas ( e.g. , jones & forman 1996 ) given by : @xmath37 and
apply the equation of hydrostatic equilibrium to the intracluster hot gas , which we assume to be isothermal .
the total mass of the lens within a radius r is given by @xmath38 } } } { \ , } \ ] ] where @xmath39 is the core radius of the cluster , @xmath40@xmath41 is the mean molecular weight of the gas , and @xmath42 is the ratio of kinetic energy per unit mass in galaxies to kinetic energy per unit mass in the gas .
the derived cluster temperature depends on the assumed value of @xmath7 and the assumed core radius of the cluster .
therefore , we determined the cluster mass within a radius of 1mpc for a range of cluster core radii and for @xmath7 values of 50 and 75 km s@xmath6 mpc@xmath6 .
the results of this analysis are shown in figure 3 .
we use the estimate of the cluster mass to evaluate the convergence parameter @xmath1 , the ratio of the local surface mass density of the cluster to the critical surface mass density for lensing .
the convergence parameter together with the time delay @xmath43 of the gls are two remaining principal parameters with the largest uncertainties that enter in the gravitational models recently developed by gn used to determine the hubble constant .
gn obtain , @xmath44 recent deep hst observations of q0957 + 561 by bernstein et al .
( 1997 ) have detected several new features and provide improved positions for the lens components .
in particular the position of the center of g1 is found to lie at the position previously determined by vlbi ( see , e.g. , falco et al .
1991 ) and significantly distant from the position used in the gn derivation of equation 4 .
the previously published estimates for @xmath1 are very uncertain because they depend ( strongly ) on the assumption that the cluster potential may be represented as a softened isothermal sphere .
the velocity dispersion and its relation to the mass of the cluster , the cluster core radius , the cluster shape and the position of the cluster center with respect to the lensed object also need to be known .
the present hri observations provide a partly independent , but still model - dependent , estimate of the total mass of the lens and hence of the convergence parameter @xmath1 through the expression @xmath45 where @xmath46 is the surface mass density of the cluster lens and @xmath47 is the critical surface mass density ( see , e.g. , blandford @xmath48 narayan 1992 ) , the density the cluster must have to produce multiple images if it were the only lensing agent , and is given by : @xmath49 where d@xmath50 , d@xmath51 , and d@xmath52 are the angular - diameter distances between source and observer , deflector and source , and deflector and observer , respectively .
we evaluated the critical surface mass density for the gravitationally lensed system q0957 + 561 to be @xmath47 = 0.9452(0.9224)@xmath53 gr @xmath54 , for @xmath55 = 0(0.5 ) . in estimating @xmath7 from the gls , we need to know the cluster contribution to the lensing of the quasar . in this treatment , we linearly superpose the effects of g1 and the cluster .
we first view g1 s contribution to the lensing as producing a pair of images , a@xmath56 and b@xmath56 , as shown in figures 4 and 5 .
the contribution of the central galaxy g1 has been represented by spls and fgs models ( gn , falco et al .
1991 ) . for the cluster ,
we assume a spherically symmetric potential which produces an outward radial offset of the lensed components with respect to the cluster center ( see figures 4 and 5 ) . to simplify the calculations we incorporate the thin lens approximation in which the cluster is projected onto a plane perpendicular to the line of sight , on the plane of the sky .
to calculate the angle , @xmath57 , by which a ray is deflected by the cluster lens , we integrated the deflection over the entire cluster .
the scaled deflection angle ( schneider , ehlers , & falco 1992 ) is given by @xmath58 where @xmath59 represents the scaled location of a projected mass element in the lens plane with respect to the cluster center and @xmath60 represents the scaled location of the impact ray which is deflected by @xmath61 .
the scaled deflection angle @xmath57 is related to the true deflection angle @xmath62 by @xmath63 , where @xmath64 is an arbitrary scale factor .
if the lens is symmetric around the line of sight , equation 7 reduces to a function of only the radial component @xmath65 , and @xmath57 lies along the radial direction @xmath65 .
@xmath66 thus , to determine the lensing contribution of the cluster , we must first model @xmath67 . from equation 3 , we find that the three dimensional mass density at a radius r can be expressed as , @xmath68 this quantity can be projected on the plane of the sky ( normal to the line of sight ) , resulting in the following expression for the surface mass density , @xmath69 , as a function of the cylindrical radius @xmath70 : @xmath71 where,@xmath72 and r is the radial extent of the cluster .
notice that the surface mass density of the cluster and the critical mass density both depend linearly upon @xmath73 , thus implying that @xmath1 does not depend on @xmath73 .
however , in our derivation of the convergence we infer the temperature of the cluster through an empirical luminosity - temperature relation and therefore our results do depend on the assumed value for @xmath7 .
in determining @xmath67 we have examined a variety of models in which we used a combination of different cluster core parameters .
we chose cluster limits ranging between 120@xmath14 and 280@xmath14 .
we also varied the core radius between the values of @xmath74 and @xmath75 and use @xmath7 values of 50 and 75 km s@xmath6 mpc@xmath6 .
note that at the smaller radii , the model approaches the limit of a singular isothermal sphere .
we find that the average value of @xmath1 , at the location of the images , varies in the range of 0.07 to 0.21 as shown in figure 6 .
the upper and lower bounds on @xmath1 place respective upper and lower bounds on @xmath73 in accordance with the models of falco et al .
( 1991 ) and gn . for the estimation of @xmath7 we have taken a self consistent iterative approach .
we begin with an initial trial value for @xmath7 and derive the total number of counts and cluster luminosity for an assumed core radius and cluster extent . using the l@xmath76 - t relation ,
we derive the cluster temperature and from equations 3 and 5 we obtain the total mass and convergence of the cluster , respectively .
the gn result ( equation 4 ) provides an estimate for h@xmath3 .
we reuse this value for h@xmath3 to initialize the iterative procedure and repeat it until the value for h@xmath3 converges ( @xmath77h@xmath3 @xmath78 1 ) .
the result is very insensitive to the initial guess for h@xmath3 .
the error for h@xmath3 is calculated by applying the self consistent iterative approach for cluster core radii ranging between 5@xmath14 and 45@xmath14 , cluster radii ranging between 120@xmath14 and 280@xmath14 and changing the mean hri count rate by @xmath8 2@xmath0 .
assuming @xmath79 , we obtain @xmath2 h@xmath3 @xmath4 82 @xmath5 km @xmath5 s@xmath6 @xmath5 mpc@xmath6 for a time delay of 1.1 years . the effect of the cluster involves a rotation and an elongation of the separation vector connecting the two images .
we may express the deflection angles for images a and b as @xmath80 and @xmath81 , respectively , where the subscript on @xmath1 indicates the point of evaluation and where @xmath82 and @xmath83 are the angular distances from the cluster center to the image locations of a and b , respectively .
the calculated values for the deflections @xmath84 and @xmath85 are shown in figure 7 .
the fractional elongation of the angular separation of images a and b produced by the cluster potential alone is @xmath86 and the calculated values for the percent elongation are presented in figure 8 ; here @xmath87 and @xmath88 denote , respectively , the vector separations of the images with and without the cluster contribution . to calculate the convergences , @xmath89 and @xmath90 , we use equation 8 with angular distances between images a and b and the cluster center given by @xmath91 = 26.57@xmath14 and @xmath92 = 21.04@xmath14 , respectively , based on the published position ( -13.7@xmath14 , -19.6@xmath14 ) ( angonin - willaime et al .
1994 ) for the estimated cluster center of mass with respect to image b and the measured position of image a with respect to image b of ( -1.25252@xmath14 , 6.04662@xmath14 ) ( gorenstein et al .
again , we use a grid of models in which we vary the cluster core radius from 5@xmath14 to 45@xmath14 and the cluster extent from 120@xmath14 to 280@xmath14 .
the quasar q0957 + 561 also was observed with the _ rosat _ pspc on 1991 november 14 for 19006 sec , on 1993 november 8 for 4972 sec , and with the asca gis on 1993 may 9 for 29000 sec .
a summary of information related to each of these observations is presented in table 1 . the spectral analysis for the 1991 november rosat observation
was presented in chartas et al .
the 1993 november pspc observation was centered on q0954 + 556 leaving the quasar images located at an off - axis angle of about 47 arcmin . for the analysis we extracted the source counts within a circular region
centered on q0957 + 561 with a radius of 6 arcmin .
background counts were extracted from several other circles of 6 arcmin radius with centers located at an off - axis angle of 47 arcmin , equal to that of q0957 + 561 . to model the spectral response of the pspc
, we used the september 1994 pspc detector response matrix pspcb_gain2_256.rmf provided by the rosat guest observer facility ( gof ) .
the spectrum was grouped to obtain a minimum of 20 counts in each energy bin , allowing use of @xmath93 statistics .
a simple redshifted power law plus galactic absorption due to cold material with solar abundances , provides an acceptable @xmath94 per degree of freedom ( see table 2 ) . however , the resulting power law is slightly steeper ( @xmath95 ) than expected for a radio loud quasar at a redshift of 1.4 ( wilkes & elvis 1987 ; shastri et al .
1993 , laor et al .
1997 , schartel et al .
1996 ) . in general the results obtained from the spectral fits to the 1993 november pspc data are consistent with the results from the 1991 november pspc observation of q0956 + 561 . in particular we find in both cases that an absorbed redshifted power law of photon index @xmath96 plus a thermal raymond - smith model provides an acceptable @xmath94 fit to the data and yields values for the fitted parameters @xmath96 , @xmath30 and @xmath97 that are close to what is expected for a radio loud quasar at z=1.4 .
the x - ray flux of the thermal component derived from the fit to the pspc spectrum of q0957 + 561 is about 5 @xmath98 10@xmath99 ergs s@xmath6 @xmath54 , at least a factor of five greater than the x - ray flux of the cluster as derived from the rosat hri observations ( see table 3 ) .
this result implies that the observed rosat pspc thermal component most likely originates from a source other than the cluster ( see , e.g. , chartas et al .
1995 ) . to constrain the model parameters even further
we performed simultaneous fits to the 1991 november and 1993 november pspc data .
we allowed the normalization of the power law and thermal component of each observation to vary independently .
a comparison of the 0.2 - 2kev x - ray fluxes between the 1991 and 1993 rosat pspc observations of q0957 + 561 implies a decrease in flux over this interval by @xmath100 with the quoted error being at the 90@xmath28 confidence level .
q0957 + 561 was also observed with the asca satellite during a pointed observation of ngc 3079 on 1993 may 9 for 29072 sec . in this observation q0957
+ 561 is located 14.1@xmath56 off axis in the gis2 image and 9.28@xmath56 off axis in the gis3 image just within the field of view of the gis2 and gis3 , but unfortunately outside the solid - state imaging spectrometer s ( sis s ) field of view .
the source counts were extracted from circles of radius 3.5 arcmin centered on q0957 + 561 .
the background was extracted from several other circles of the same radius centered at an off - axis angle equal to that of q0957 + 561 , but far away in azimuth . for the data analysis of the gis2 and gis3 spectra we used the spectral response functions gis2v4_0.rmf and gis3v4_0.rmf , respectively , provided by the asca gof .
the spectral data were grouped to obtain a minimum of 20 counts per bin .
the observed rosat pspc and asca gis spectra are shown in figure 9 .
the few source counts obtained in the asca observation prevent us from fitting any complex models ; however , a comparison of the spectral slope of a simple power law model fitted to the 0.5 - 6 kev gis data to that of the slope obtained from fits to the 0.2 - 2 kev rosat data was possible . in particular a simultaneous fit of a simple absorbed power law model between 0.5 and 5kev to the gis2 and gis3 spectra of q0957 + 561 yields a photon index @xmath101 , quite typical for radio loud quasars at z=1.4 ; a simultaneous fit between 0.2 and 2kev of a similar model to the november 1991 and november 1993 rosat pspc spectra gave a photon index of @xmath102 .
this difference may indicate a break in the spectrum at some energy above 2 kev .
alternatively , including a thermal component ( fits 13 and 14 ) results in values of the photon index of the power law component consistent with that for radio loud quasars .
the need for a detailed measurement of the mass distribution of the cluster lens has been emphasized by falco et al.(1991 ) and by gn ( 1996 ) in analyses of lens models for q0957 + 561 that incorporate the cluster contribution through a convergence parameter @xmath103 .
the most important result of the present analysis of the rosat hri data of q0957 + 561 , that we hope will eventually lead to an accurate determination of the mass distribution of the cluster lens , is the 3@xmath0 detection of x - rays from the cluster of galaxies that contains the principal lens galaxy g1 . + as indicated in figure 3 , the uncertainty in our estimated values for the mass distribution of the cluster spans a large range as do the consequent estimates of the convergence parameter of the lens ( figure 6 ) .
the main contributors to this uncertainty are the lack of knowledge of the shape of the cluster , the core radius of the cluster , the temperature profile of the hot gas , the cluster extent , and the location of the cluster center as well as the sensitivity to the value of @xmath104 used for deriving the cluster hot gas temperature from the x - ray luminosity .
the present analysis yields a total cluster mass in the range 1.5 - 3.2 @xmath98 10@xmath105m@xmath106 , and does not include allowances for errors in several of the above characteristics , but does consider the effects of cluster core radii between 5@xmath14 and 45@xmath14 , and cluster extents between 120@xmath14 and 280@xmath14 .
the calculations of the x - ray luminosity were performed for hubble constants of 50 and 75 km s@xmath6 mpc@xmath6 .
the average convergence parameter @xmath107 of the cluster at the image locations a and b was found to lie between 0.07 and 0.21 .
+ one of the other assumptions in the present calculation is that the temperature of the cluster follows the luminosity temperature relation of david et al .
( 1993 ) for a large sample of clusters of galaxies .
the derived cluster temperature depends weakly on the cluster x - ray luminosity .
an error of a factor of 2 in the x - ray luminosity derived from the hri data will propagate as an error of about 30@xmath28 in the cluster temperature and about 7@xmath28 ( for @xmath1 = 0.2 ) in the derived value of the hubble constant .
the residual rms scatter in the @xmath33 - t relation along the temperature axis is @xmath108 = 0.104/2.1 according to markevitch ( 1998 ) and propagates as an error of about 3@xmath28 in the derived value of the hubble constant .
our calculations for the deflection angles produced from considering the influence of the cluster alone ( not including g1 ) found the deflection for images a and b to lie in the range of 2@xmath14 to 16@xmath14 as shown in figure 6 ; the differences in these deflections for images a and b are much smaller .
the resulting percent contribution , @xmath109 , to the observed 6.@xmath141 separation of images a and b due to the cluster alone , lies in the range of 4% to 22% .
this range includes only the effects of varying the cluster core radius between 5@xmath14 and 45@xmath14 and the cluster limit between 120@xmath14 and 280@xmath14 .
this result is in contrast with the commonly accepted notion that the large separation of the two images of q0957 + 561 is an indicator that the lens contains a cluster of galaxies . to test the sensitivity of this result to the adopted cluster center position , we computed the cluster contribution to the image separation for cluster center locations within a distance of 5@xmath14 of the location derived from the optical observations by angonin - willaime et al .
( 1994 ) . as expected , when the cluster center is moved closer to the locations of images a and b
, the derived cluster contribution to the image separation becomes more significant reaching a value as high as @xmath109 = 31% for a cluster center at a distance of about 16@xmath14 from image b. + recently the mass distribution of the lens of q0957 + 561 was determined by fischer et al .
( 1997 ) based on the distortion of background galaxies produced by the lens .
their estimate of 3.7 @xmath8 1.2 @xmath98 10@xmath105m@xmath106 within a radius of 1mpc is consistent with the x - ray estimates .
as pointed out by schneider & seitz ( 1995 ) however , the kaiser & squires ( 1993 ) mass reconstruction technique also is insensitive to constant density sheets of matter . an interesting conclusion from the mass reconstruction analysis by fischer et al .
( 1997 ) is the relatively low value of 5@xmath14 derived for the cluster core radius . the cluster core radius , r@xmath110 , as defined in fischer et al .
( 1997 ) , however , differs from the cluster core radius , r@xmath111 , used in this paper and defined in equation 2 .
the relation between r@xmath111 and r@xmath110 is given approximately by r@xmath110 = -0.987 + 0.685r@xmath111 for 5@xmath14 @xmath4 r@xmath111 @xmath4 50@xmath14 .
we will have to await future x - ray observations of q0957 + 561 to compare more accurately x - ray and weak - lens results for the cluster core radius .
+ the present rosat detection of cluster emission suggests that future x - ray observations will greatly add to our understanding of q0957 + 561 .
axaf observations should provide direct determination of the temperature profile , the core radius and the shape of the cluster lens , reveal possible clumpiness in the total mass profile , and ultimately provide a tighter constraint on the hubble constant .
+ we are grateful to emilio falco for providing many useful comments and to donald schneider for making many useful suggestions on several early drafts of this paper .
this research was supported by nasa through grant nas 8 - 38252 .
llllll + + & & & & & + & & & & & + date & & file name i d & & angle & + & & & @xmath112 & @xmath113 & + 1991 may 1 & _ rosat _ hri & rh700100 & 20847 & 45 & 473 @xmath8 30 + 1992 october 20 & _ rosat _ hri & wh600411n00 & 20676 & 740 & 812 @xmath8 34 + 1992 october 22 & _ rosat _ hri & rh700889n00 & 18548 & 50 & 800 @xmath8 33 + 1994 april 11 & _ rosat _ hri & rh701294n00 & 25083 & 50 & 1077 @xmath8 39 + 1991 november 15 & _ rosat _ pspc & rp300319 & 19006 & 993 & 3677 @xmath8 70 + 1993
november 8 & _ rosat _ pspc & rp701401n00 & 4472 & 2821 & 449 @xmath8 27 + 1993 may 9 & _ asca _ gis2 & ad60000000g200470 & 29072 & 846 & 140 @xmath8 20 + 1993 may 9 & _ asca _ gis3 & ad60000000g300470 & 29072 & 557 & 121 @xmath8 18 + note- a description of the format for the rosat high resolution instrument ( hri ) , rosat position sensitive proportional counter ( pspc ) and asca gas imaging spectrometer ( gis ) file names is available at the rosat and asca guest observer facility web pages .
lllllll + + + & & & & & & + & & & & & & + & & & & @xmath114@xmath115 & & + 1 & 1 & @xmath116 & @xmath117 & 0.87 & & 0.95(13 ) + 2 & 2 & @xmath116 & @xmath118 & @xmath119 & & 0.80(12 ) + 3 & 3 & @xmath116 & @xmath120 & 0.87 & @xmath121 & 0.73(11 ) + 4 & 1 & @xmath122 & @xmath123 & 0.87 & & 0.88(63 ) + 5 & 2 & @xmath122 & @xmath124 & @xmath125 & & 0.69(62 ) + 6 & 3 & @xmath122 & @xmath126 & 0.87 & @xmath127 & 0.62(60 ) + 7 & 4 & @xmath122 & @xmath128 & @xmath129 & @xmath130 & 0.63(59 ) + 8 & 1 & gis2 & @xmath131 & 0.87 & & 1.85(12 ) + 9 & 1 & gis3 & @xmath132 & 0.87 & & 1.34(7 ) + 10&1 & gis2 + gis3 & @xmath133 & 0.87 & & 1.99(20 ) + 11&1 & @xmath122 + & & & & + & & gis2 + gis3 & @xmath134 & 0.87 & & 1.42(84 ) + 12&2 & @xmath122 + & & & & + & & gis2 + gis3 & @xmath135 & @xmath136 & & 1.42(83 ) + 13 & 3 & @xmath122 + & & & & + & & gis2 + gis3 & @xmath137 & 0.87 & @xmath138 & 1.09(79 ) + 14 & 4 & @xmath122 + & & & & + & & gis2 + gis3 & @xmath139 & @xmath140 & @xmath141 & 1.01(78 ) + notes- + @xmath142 model 1 incorporates a redshifted power law plus absorption due to cold material at solar abundances fixed to the galactic value .
model 2 incorporates a redshifted power law plus absorption due to cold material at solar abundances set as a free parameter in the fit .
model 3 incorporates a redshifted power law plus thermal raymond - smith model plus galactic absorption .
model 4 incorporates a redshifted power law plus thermal raymond - smith model plus absorption due to cold material at solar abundances set as a free parameter in the fit .
+ @xmath143 @xmath144 and @xmath116 correspond to the rosat pspc observations made on 1991
november 15 and on 1993 november 8 , respectively .
+ llllll + + & & & & & + & @xmath145 & & & & + & & @xmath146 erg @xmath54 s@xmath6&@xmath146 erg @xmath54 s@xmath6&@xmath147 erg s@xmath6&@xmath147 erg s@xmath6 + 6 & @xmath144 & @xmath148 & @xmath149 & 1.29 & 1.68 + 6 & @xmath116 & @xmath150 & @xmath151 & 1.42 & 1.11 + | analysis of several recent rosat hri observations of the gravitationally lensed system q0957 + 561 has led to the detection at the 3@xmath0 level of the cluster lens containing the primary galaxy g1 .
the total mass was estimated by applying the equation of hydrostatic equilibrium to the detected hot intracluster gas for a range of cluster core radii , cluster sizes and for different values of the hubble constant .
x - ray estimates of the lensing cluster mass provide a means to determine the cluster contribution to the deflection of rays originating from the quasar q0957 + 561 .
the present mass estimates were used to evaluate the convergence parameter @xmath1 , the ratio of the local surface mass density of the cluster to the critical surface mass density for lensing .
the convergence parameter , @xmath1 , calculated in the vicinity of the lensed images , was found to range between 0.07 and 0.21 , depending on the assumed cluster core radius and cluster extent .
this range of uncertainty in @xmath1 does not include possible systematic errors arising from the estimation of the cluster temperature through the use of the cluster luminosity - temperature relation and the assumption of spherical symmetry of the cluster gas . applying this range of values of @xmath1 to the lensing model of grogin & narayan ( 1996 ) for q0957 + 561 but not accounting for uncertainties in that model yields a range of values for the hubble constant : @xmath2 h@xmath3 @xmath4 82 @xmath5 km @xmath5 s@xmath6 @xmath5 mpc@xmath6 , for a time delay of 1.1 years .
# 1 # 1 |
recall that for each coxeter group @xmath11 one has a group extension @xmath12 usually called _ artin group _ of type @xmath11 ( see section [ sec : prelim ] ) . in this paper
we give a detailed calculation of the cohomology of some artin groups with non - trivial local coefficients .
let @xmath13 $ ] be the ring of two - parameters laurent polynomials .
the main result ( theorem [ t : cohomqt ] ) is the computation of the cohomology of the artin group @xmath14 ( of type @xmath2 ) with coefficients in the module @xmath15 the latter is the ring @xmath16 with the module structure defined as follows : the generators associated to the first @xmath17 nodes of the dynkin diagram of @xmath2 act by @xmath5multiplication ; the one associated to the last node acts by @xmath6multiplication .
let @xmath18 be the @xmath19-th cyclotomic polynomial in the variable @xmath20 .
define the @xmath16-modules ( @xmath21 ) @xmath22 and for @xmath23 set : @xmath24 notice that the modules @xmath25 are all non isomorphic as @xmath16-modules . @xmath26 and @xmath27 are isomorphic as @xmath28$]-modules if and only if @xmath29 and are isomorphic as @xmath30$]-modules if and only if @xmath31 ( @xmath32 is the euler function ) and @xmath33 . our main result is the following [ t : cohomqt ] @xmath34 the proof uses the spectral sequence associated to a natural filtration of the algebraic complex exhibited in @xcite , plus some technical results from @xcite .
we apply shapiro s lemma to a well known inclusion of @xmath35 into @xmath36 to derive the cohomology of @xmath35 over the module @xmath37,$ ] the action of each standard generator being @xmath5multiplication . by considering another natural inclusion of @xmath38 into the classical braid group @xmath39 we also use shapiro s lemma in order to identify the cohomology of @xmath14 with coefficients in @xmath40 with that of @xmath41 with coefficients in the irreducible @xmath42dimensional representation of @xmath41 found in @xcite , twisted by an abelian representation .
we derive the trivial @xmath43cohomology of @xmath35 as well as the cohomology of the braid group over the irreducible representation in @xcite .
computation of the cohomology of artin groups was done by several people : for classical _ braid groups _ and trivial coefficients it was first given by f. cohen @xcite , and independently by a. vanten @xcite ( see also @xcite ) . for artin groups of type @xmath44 @xmath45 see @xcite , while for the exceptional cases see @xcite , where the @xmath46-module structure was given , while the ring structure was computed in @xcite .
the case of non - trivial coefficients over the module of laurent polynomials @xmath47 $ ] is interesting because of its relation with the trivial @xmath48-cohomology of the milnor fibre of the naturally associated bundle . for the classical braid groups
see @xcite , while for cases @xmath49 see @xcite . for computations over the integral laurent polynomials @xmath50,$ ]
see @xcite for the exceptional cases and recently @xcite for the case of braid groups , and @xcite for the top cohomologies in all cases . in the case of artin groups of non - finite type , some computations were given in @xcite and @xcite .
the computations of theorem [ t : cohomqt ] could be partially extended to integral coefficients ; however , major complications occur because the laurent polynomial ring @xmath50 $ ] is not a p.i.d .. in the last part we also indicate ( see @xcite ) an explicit construction of finite cw - complexes which are retracts of _ orbit spaces _ associated to artin groups .
the construction works as in @xcite , with few variations necessary for infinite type artin groups ( see also @xcite for a different construction ) .
the artin group identifies with the fundamental group of the orbit space , and the standard presentation follows easily ( see @xcite ) .
the euler characteristic of the orbit space reduces to that of a simplicial complex and in some cases one has a particularly simple formula .
it is conjectured that such orbit spaces are always @xmath10 spaces ; for the affine groups , this is known in case @xmath51 ( see @xcite ) and recently also for @xmath52 ( @xcite ) ( see also @xcite for a different class of artin groups of infinite type ) .
notice also the geometrical meaning of the two - parameters cohomology of @xmath53 similar to the one - parameter case , it is equivalent to the trivial cohomology of the `` homotopy - milnor fibre '' associated to the natural map of the orbit space onto a two - dimensional torus .
the main results of this paper were announced ( without proof ) in @xcite .
in this section we briefly fix the notation and recall some preliminary results . a _ coxeter graph _
is a finite undirected graph , whose edges are labelled with integers @xmath54 or with the symbol @xmath55 .
let @xmath56 be the vertex set of a coxeter graph .
for every pair of vertices @xmath57 ( @xmath58 ) joined by an edge , define @xmath59 to be the label of the edge joining them . if @xmath60 are not joined by an edge , set by convention @xmath61 .
let also @xmath62 ( see @xcite ) .
two groups are associated to a coxeter graph : the _ coxeter group _
@xmath11 defined by @xmath63 and the _ artin group _
@xmath64 defined by ( see @xcite ) : @xmath65 loosely speaking , @xmath64 is the group obtained by dropping the relations @xmath66 ( @xmath67 ) in the presentation for @xmath11 . in this paper , we are primarily interested in artin groups associated to coxeter graph of type @xmath68 , @xmath2 and @xmath69 ( see figure [ fig : dinkyn ] ) .
( 0,0 ) , @xmath2 ( @xmath70 ) and @xmath69 ( @xmath71 ) .
labels equal to @xmath72 , as usual , are not shown .
moreover , to fix notation , every vertex is labelled with the corresponding generator in the artin group.,title="fig : " ] # 1#2#3#4#5 @font ( 5720,1937)(-4070,-4328 ) ( -1076,-3076)(0,0)[lb ] ( 1510,-3891)(0,0)[lb ] ( 1650,-3228)(0,0)[lb ] ( 994,-4289)(0,0)[lb ] ( 534,-3376)(0,0)[lb ] ( -4070,-4036)(0,0)[lb ] ( -3630,-4043)(0,0)[lb ] ( -1313,-3834)(0,0)[lb ] ( -2598,-3729)(0,0)[lb ] ( -1552,-4015)(0,0)[lb ] ( -4047,-3090)(0,0)[lb ] ( -3607,-3097)(0,0)[lb ] ( -1529,-3069)(0,0)[lb ] ( -2575,-2783)(0,0)[lb ] ( -278,-3870)(0,0)[lb ] ( 317,-4297)(0,0)[lb ] ( -1092,-4022)(0,0)[lb ] let @xmath73 be the braid group on @xmath74 strands and @xmath75 be the subgroup of braids fixing the @xmath76-st strand .
the group @xmath77 is called the annular braid group .
let @xmath78 be a set of @xmath74 distinct points in @xmath79 .
the classical braid group @xmath80 can be realized as the fundamental group of the space of unordered configurations of @xmath74 points in @xmath79 with basepoint @xmath81 ( see the left part of figure [ fig : cylinder ] ) , with @xmath82 ) .
we can now think to the subgroup @xmath75 as the fundamental group of the space of unordered configurations of @xmath8 points in @xmath83 : in fact if we take @xmath84 and @xmath85 for @xmath86 , since for a braid @xmath87 the orbit of the @xmath76-st point can be thought constant , up to homotopy , we can think to @xmath88 as a braid with @xmath8 strands in the anulus ( see the right part of figure [ fig : cylinder ] ) .
( 0,0 ) represented as an annular braid on @xmath89 strands.,title="fig : " ] # 1#2#3#4#5 @font ( 5501,2851)(-2605,-4336 ) ( 2588,-1561)(0,0)[lb ] ( 1708,-1561)(0,0)[lb ] ( 2598,-3602)(0,0)[lb ] ( 1728,-3591)(0,0)[lb ] ( -1566,-1635)(0,0)[lb ] ( -1041,-1636)(0,0)[lb ] ( 1640,-2238)(0,0)[lb ] ( 2655,-2238)(0,0)[lb ] ( 2200,-2428)(0,0)[lb ] ( 1634,-4113)(0,0)[lb ] ( 2200,-4284)(0,0)[lb ] ( 2639,-4116)(0,0)[lb ] ( -2583,-1637)(0,0)[lb ] ( -2077,-1640)(0,0)[lb ] ( -530,-1633)(0,0)[lb ] ( -10,-1628)(0,0)[lb ] it is well known that the annular braid group is isomorphic to the artin group @xmath14 of type @xmath2 . for
a proof of the following theorem see @xcite or @xcite .
let @xmath90 and @xmath91 be respectively the standard generators for @xmath92 and @xmath14 .
then , the map @xmath93 is an isomorphism . using the suggestion given by the identification with the annular braid group , a new interesting presentation for @xmath14
can be worked out .
let @xmath94 .
it is easy to verify that : @xmath95 i.e. conjugation by @xmath96 shifts forward the first @xmath97 standard generators . by analogy ,
let @xmath98 .
we have the following [ teo : peifer ] the group @xmath14 has presentation @xmath99 where @xmath100 where are
all indexes should be considered modulo @xmath8 .
letting @xmath101 be the standard generator of the artin group of type @xmath69 , we have the following straightforward corollary : [ cor : peifer ] the map @xmath102 gives an isomorphism between the group @xmath35 and the subgroup of @xmath14 generated by @xmath103 .
moreover , we have a semidirect product decomposition @xmath104 .
we have thus a `` curious '' inclusion of the artin group of infinite type @xmath69 into the artin group of finite type @xmath105 .
the proof of theorem [ teo : peifer ] presented in the original paper is algebraic and based on tietze moves ; a somewhat more coincise proof can however be obtained by standard topological constructions . indeed , one can exhibit an explicit infinite cyclic covering @xmath106 ( see @xcite ) .
for future use in cohomology computations , we are interested in a @xmath109-analog of the usual poincar series for @xmath2 , that is an analog of the poincar series with coefficients in the ring @xmath110 $ ] of laurent polynomials . this result and similar ones
are studied in @xcite , to which we refer for details .
we also use classical results from @xcite without further reference . consider the coxeter group @xmath11 of type @xmath2 with its standard generating reflections @xmath111 . for @xmath112 ,
let @xmath113 be the number of times @xmath114 appears in a reduced expression for @xmath115 . by standard facts ,
@xmath113 is well - defined .
+ we define the @xmath109-weighted poincar series for the coxeter group of type @xmath2 as the sum @xmath116 where @xmath117 is the length function .
we recall some notation .
we define the @xmath20-analog of the number @xmath19 by the polynomial @xmath118_q : = 1 + q + \cdots q^{m-1 } = \frac{q^m -1}{q-1}.\ ] ] notice that @xmath119 = \prod_{i \mid m , \ , i\neq 1 } { \varphi}_m(q)$ ] , where we denote with @xmath18 the @xmath19-th cyclotomic polynomial in the variable @xmath20 .
moreover we define the @xmath20-factorial analog @xmath120_q!$ ] as the product @xmath121_q\ ] ] and the @xmath20-analog of the binomial @xmath122 as the polynomial @xmath123 } _ { q } : = \frac{[m]_q!}{[i]_q ! [ m - i]_q!}.\ ] ] we can also define the @xmath109-analog of an even number @xmath124_{q , t } : = [ m]_q ( 1+tq^{m-1})\ ] ] and of the double factorial @xmath124_{q , t } ! ! :
= \prod_{i=1}^m [ 2i]_{q , t}\ = \ [ m ] _ q
! \prod_{i=0}^{m-1 }
( 1+tq^i).\ ] ] finally , we define the polynomial @xmath123 } _ { q , t } ' : = \frac{[2m]_{q , t}!!}{[2i]_{q , t } ! !
[ m - i]_q ! } \
= { \left [ \begin{array}{c } m \\
i \end{array } \right ] } _ { q}\prod_{j = i}^{m-1}(1+tq^j).\ ] ] [ p : qtpoincare ] @xmath125_{q , t}!!.\ ] ] consider the parabolic subgroup @xmath126 associated to the subset of reflections @xmath127 .
notice that @xmath126 is isomorphic to the symmetric group on @xmath8 letters @xmath128 and that it has index @xmath129 in @xmath2 .
let @xmath130 be the set of minimal coset representatives for @xmath131 .
then , by multiplicative properties on reduced expressions : @xmath132 clearly , for elements @xmath133 , we have @xmath134 ; so the second factor in ( [ formula:1 ] ) reduces to the well - known poincar series for @xmath128 : @xmath135_q!.\ ] ] to deal with the first factor , instead , we explicitly enumerate the elements of @xmath130 .
let @xmath136 for @xmath137 .
then , it can be easily verified that @xmath138 .
notice that @xmath139 and @xmath140 .
thus , @xmath141 finally , @xmath142_q!=[2n]_{q , t}!!.\ ] ]
in this section we prove theorem [ t : cohomqt ] enunciated in the introduction .
we use the notations given in the introduction . to perform our computation
we will use the complex discovered in @xcite , @xcite ( notice : an equivalent complex was discovered by different methods in @xcite ) , and the spectral sequence induced by a natural filtration .
the complex that computes the cohomology of @xmath14 over @xmath40 is given as follows ( see @xcite ) : @xmath143 where @xmath144 denote the set @xmath145 and the graduation is given by @xmath146 .
the set @xmath144 corresponds to the set of nodes of the dynkin diagram of @xmath2 and in particular the last element , @xmath8 , corresponds to the last node .
it is useful to consider also the complex @xmath147 for the cohomology of @xmath92 on the local system @xmath40 . in this case
the action associated to a standard generator is always the @xmath148-multiplication and so the complex @xmath147 and its cohomology are free as @xmath149$]-modules .
the complex @xmath147 is isomorphic to @xmath150 as a @xmath16-module . in both complexes
the coboundary map is @xmath151 where @xmath152 is the number of elements of @xmath153 that are less than @xmath154 . in the case
@xmath68 @xmath155 is the poincar polynomial of the parabolic subgroup @xmath156 generated by the elements in the set @xmath153 , with weight @xmath157 for each standard generator , while in the case @xmath2 @xmath158 is the poincar polynomial of the parabolic subgroup @xmath159 generated by the elements in the set @xmath153 , with weight @xmath157 for the first @xmath17 generators and @xmath160 for the last generator .
using proposition [ p : qtpoincare ] we can give an explicit computation of the coefficients @xmath161 .
for any @xmath162 , let @xmath163 be the subgraph of the dynkin diagram @xmath2 which is spanned by @xmath153 . recall that if @xmath163 is a connected component of the dynkin diagram of @xmath2 without the last element , then @xmath164_q!,\ ] ] where @xmath165 . if @xmath163 is connected and contains the last element of @xmath2 , then @xmath166_{q , t}!!,\ ] ] where @xmath167 .
if @xmath163 is the union of several connected components of the dynkin diagram , @xmath168 , then @xmath169 is the product @xmath170 of the factors corresponding to the different components . if @xmath171 we can write @xmath172 for the connected component of @xmath173 containing @xmath154
suppose that @xmath174 and @xmath175 is the number of elements in @xmath176 greater than @xmath154 .
then , if @xmath177 we have @xmath178 } _ { q , t}'\ ] ] and @xmath179 } _ { q}\ ]
] otherwise .
it is convenient to represent generators @xmath180 by their characteristic functions @xmath181 so , simply by strings of @xmath182s and @xmath183s of length @xmath8 .
we define a decreasing filtration @xmath184 on the complex @xmath185 : @xmath186 is the subcomplex generated by the strings of type @xmath187 ( ending with a string of @xmath188 @xmath183s ) and we have the inclusions @xmath189 we have the following isomorphism of complexes : @xmath190\ ] ] where @xmath191 is the complex for @xmath192 and the notation @xmath193 $ ] means that the degree is shifted by @xmath188 .
let @xmath194 be the spectral sequence associated to the filtration @xmath184 .
the equality [ e : shift ] tells us how the @xmath195 term of the spectral sequence looks like .
in fact for @xmath196 we have @xmath197_q)[t^{{\pm 1}}]\ ] ] since the @xmath198-action is trivial . for @xmath199 and @xmath200 the only non trivial elements in the spectral sequence are @xmath201 in order to prove theorem [ t : cohomqt ] we need to state the following lemmas .
[ l : ideali2 ] let @xmath202 be the ideal generated by the polynomials @xmath203 } ' _ { q , t } \qquad\textrm { for $ d \mid n$ and $ d \leq k$}\ ] ] if @xmath204 the map @xmath205 induced by the multiplication by @xmath206 } ' _ { q , t}$ ] is well defined and is injective .
* remark . * the fact that this map is well defined will follow automatically from the general theory of spectral sequences , as it is clear from the proof of theorem [ t : cohomqt ] .
however , below we prove it by other means .
let @xmath207 be positive integers such that @xmath208 and @xmath209 .
we can observe that @xmath210 } _ q = { \left [ \begin{array}{c } n \\ n - k \end{array } \right ] } _ q $ ] if and only if @xmath211 .
moreover each factor @xmath212 appears in @xmath213 } _ q$ ] at most with exponent @xmath183 .
let @xmath214 be the ideal generated by the polynomials @xmath215 } _
q$ ] for @xmath208 and @xmath216 .
it is easy to see that we have the following inclusion : @xmath217 moreover @xmath214 is a principal ideal and is generated by the product @xmath218 it follows that @xmath206 } _ q { \varphi}_k(q ) \in j(n , k-1)$ ] and so @xmath206 } ' _ { q , t } { \varphi}_k(q ) \in i(n , k-1)$ ] .
this proves that the map @xmath219 is well defined .
now we notice that the factor @xmath220 divides each generator of @xmath221 , but does not divide @xmath206 } ' _ { q , t}$ ] .
this imply that @xmath219 is not the zero map and that every polynomial in the kernel of @xmath219 must be a multiple of @xmath220 , hence the map must be injective .
[ l : ideali1 ] let @xmath222 be the ideal generated by the polynomials @xmath203 } ' _ { q , t } \qquad \textrm { for $ d \mid n$}\ ] ] then @xmath222 is the direct product of the ideals @xmath223 for @xmath224 and @xmath225 and of the ideal @xmath226 .
moreover the ideals @xmath227 and @xmath228 are pairwise coprime .
notice that the polynomial @xmath229 divides each generator of the ideal @xmath222 , so we can write @xmath230 where @xmath231 is the ideal generated by the polynomials @xmath232 } } ' _ { q , t } : = { \left [ \begin{array}{c } n \\ n - d \end{array } \right ] } ' _ { q , t}/(1+tq^{n-1})\ ] ] let @xmath233 be the list of all the divisors of @xmath8 in decreasing order .
if we set @xmath234 we can rewrite our ideal as @xmath235 } , \ , { \left [ \begin{array}{c } n \\
n - d_{h-1 } \end{array } \right ] } q_{h-1},\ , { \left [ \begin{array}{c } n \\ n - d_{h-2 } \end{array } \right ] } q_{h-2}q_{h-1},\ , \ldots \right . \\ { } & \left .
\ldots , { \left [ \begin{array}{c } n \\ n - d_2 \end{array } \right ] } q_2\cdots q_{h-1 } , \ ,
q_1 \cdots q_{h-1 } \right ) \end{split}\ ] ] we claim that we can reduce to the following set of generators : @xmath236 the first generator is the same in both equations and the @xmath154-th generator in equation [ e : ideali2 ] divides the corresponding generator in equation [ e : ideali1 ] .
now suppose that a factor @xmath18 divides @xmath237 } $ ] but does not divide @xmath238 .
we may distinguish two cases : * suppose that @xmath239 . then we can get rid of the factor @xmath18 in @xmath237 } $ ] with an opportune combination with the polynomial @xmath240 * suppose @xmath241 .
then @xmath242 for some @xmath243 and we can get rid of @xmath18 using a suitable combination with the polynomial @xmath244 we may now proceed inductively . supposing we have already reduced the first @xmath245 terms , we can reduce the @xmath154-th term of the ideal in equation [ e : ideali1 ] to the corresponding term in equation [ e : ideali2 ] .
now we observe that if @xmath246 are ideals and @xmath247 , then @xmath248 .
since the polynomials @xmath249 are all coprime , we can apply this fact to the ideal @xmath231 @xmath250 times . at the @xmath175-th step we set @xmath251 @xmath252 @xmath253
so we can factor @xmath231 as @xmath254 @xmath255 finally we can split @xmath256 as the product @xmath257 so we have reduced the ideal @xmath222 in the product stated in the lemma and it is easy to check that all the ideals of the splitting are coprime .
[ proof of theorem [ t : cohomqt ] ] we can now prove our theorem using the spectral sequence described in the equations [ e : sseq1 ] and [ e : sseq2 ] .
we introduce , as in @xcite , the following notation for the generators of the spectral sequence : @xmath258 we write @xmath259 $ ] for the module @xmath260 .
the @xmath195-term of the spectral sequence has a module @xmath259 $ ] in position @xmath261 if and only if one of the following condition is satisfied : a)@xmath262 and @xmath263 ; b)@xmath264 and @xmath265 .
moreover we have modules @xmath16 in position @xmath266 and @xmath267 .
we now look at the @xmath268 map between these two modules .
notice that @xmath269 is generated by the string @xmath270 and @xmath271 is generated by the string @xmath272 .
furthermore the map @xmath273 is given by the multiplication by @xmath274 } _ { q , t } ' = [ n]_q(1+tq^{n-1 } ) $ ] and is injective .
it turns out that @xmath275 and @xmath276_q(1+tq^{n-1}))$ ] .
moreover all the following terms @xmath277 are quotient of @xmath278 .
notice that every map between modules of kind @xmath259 $ ] and @xmath279 $ ] must be zero if @xmath280 .
so we can study our spectral sequence considering only maps between the same kind of modules .
first let us consider an integer @xmath19 that does nt divide @xmath8 .
say that @xmath281 with @xmath282 and set @xmath283 .
the modules of type @xmath284 $ ] are : @xmath285 for @xmath286 . here is a diagram for this case ( we use the notation @xmath287 for @xmath259 $ ] ) : [ cols="<",options="header " , ] the map @xmath288 is given by the multiplication by @xmath289 } ' _ { q , t } = [ \lambda m]_q(1+tq^{\lambda m-1})$ ] , but in this case the coefficient is zero in the module @xmath290 $ ] because @xmath291_q$ ] and so we have that @xmath292 .
so we have to consider the map @xmath293 for @xmath294 .
this map corresponds to the multiplication by @xmath295 } ' _ { q , t } = { \left [ \begin{array}{c } ( \lambda + 1)m-1 \\ \lambda m \end{array } \right ] } _ q \prod_{j = \lambda m + 1}^{(\lambda + 1)m-1}(1+tq^{j-1}).\ ] ] it is easy to see that the polynomial @xmath296 }
_ q$ ] is prime with the torsion @xmath18 and so the map @xmath297 is injective and the cokernel is isomorphic to @xmath298 as a consequence we have that @xmath299 and all these modules collapse to @xmath300 .
this means that we can find @xmath18-torsion only in @xmath301 and for @xmath302 the summand is given by @xmath303 for @xmath304 .
we still have to consider all the terms @xmath305 $ ] for @xmath241 . here
the maps we have to look at are the following : @xmath306 these maps correspond to multiplication by the polynomials @xmath307 } ' _ { q , t } $ ] . moreover recall that @xmath308 } ' _ { q , t } \right ) \right . .\ ] ] we can now use lemma [ l : ideali2 ] to say that all the maps @xmath309 are injective and lemma [ l : ideali1 ] to say that @xmath310 since @xmath311 , this complete the proof of the theorem .
we may consider the cohomology of @xmath14 over the module @xmath312,$ ] where the action is trivial for the generators @xmath313 and @xmath314-multiplication for the last generator @xmath315 .
this cohomology is computed by the complex @xmath150 of section 3 where we specialize @xmath20 to @xmath316 so we may use similar filtration and associated spectral sequence .
we used this argument in @xcite . here
we briefly indicate a different and more concise method , using the results of theorem [ t : cohomqt ] .
we have : [ teo : ratio ] @xmath317 ) & = & { \mathbb{q}}[t^{{{\pm 1}}}]/(1+t ) & \quad \
1\leq k\leq n-1 \\ & & & \\ h^n(g_{b_n},{\mathbb{q}}[t^{{{\pm 1 } } } ] ) & = & { \mathbb{q}}[t^{{{\pm 1}}}]/(1+t ) & \quad \text{for odd $ n$}\\ & & & \\
h^n(g_{b_n},{\mathbb{q}}[t^{{{\pm 1 } } } ] ) & = & { \mathbb{q}}[t^{{{\pm 1}}}]/(1-t^2 ) & \quad \text{for even $ n$. } \end{array}\ ] ] consider the short exact sequence : @xmath318 \stackrel{1+q}{{\longrightarrow } } { \mathbb{q}}[q^{{{\pm 1}}},t^{{{\pm 1 } } } ] \to { \mathbb{q}}[t^{{{\pm 1 } } } ] \to 0\ ] ] and the induced long exact sequence for cohomology @xmath319)\ !
\stackrel{1+q}{{\longrightarrow } } \ !
h^i(g_{b_n},\ ! { \mathbb{q}}[q^{{{\pm 1 } } } \ ! , \
t^{{{\pm 1}}}])\ ! \to \ !
h^i(g_{b_n } , \ ! { \mathbb{q}}[t^{{{\pm 1}}}])\ ! \to \ ! \cdots .\ ] ] the result is now a straightforward consequence of theorem [ t : cohomqt ] .
by means of shapiro s lemma ( see for instance @xcite ) , the inclusions introduced in section [ sec : inclusions ] can be exploited to link the cohomology of the artin group of type @xmath69 , @xmath68 to the cohomology of @xmath320 .
let @xmath322 be any domain and let @xmath20 be a unit of @xmath323 we indicate by @xmath324 the ring @xmath322 with the @xmath35-module structure where the action of the standard generators is given by @xmath148-multiplication .
[ prop : sha1 ] we have @xmath325_{q , t } ) \\
h^*(g_{\tilde{a}_{n-1 } } , m_q ) \cong & h^*(g_{b_{n } } , m[[t^{\pm 1}]]_{q , t})\end{aligned}\ ] ] where the action of @xmath14 on @xmath326_{q , t}$ ] ( and on @xmath327_{q , t}$ ] ) is given by @xmath148-multiplication for the generators @xmath313 and @xmath314-multiplication for the last generator @xmath328 .
applying shapiro s lemma to the inclusion @xmath329 , one obtains : @xmath330 by corollary [ cor : peifer ] , any element of @xmath331 \otimes_{g_{\tilde{a}_{n-1 } } } m_q$ ] can be represented as a sum of elements of the form @xmath332 .
now , we have an isomorphism of @xmath333$]-modules @xmath334 \otimes_{g_{\tilde{a}_{n-1 } } } m_q & \to m[t^{\pm 1}]_{q , t}\end{aligned}\ ] ] defined by sending @xmath335 and the result follows .
in cohomology we have similarly : @xmath336,m_q ) \cong m[[t^{\pm 1}]]_{q , t}.\end{aligned}\ ] ] by propositions [ prop : sha1 ] , in order to determine the cohomology @xmath337 it is necessary to know the cohomology of @xmath320 with values in the module @xmath338 $ ] of laurent series in the variable @xmath198 .
the latter is linked to the cohomology with values in the module of laurent polynomials by : [ prop : shift ] @xmath339_{q , t})\cong h^{*+1}(g_{b_n } , m[t^{\pm 1 } ] _ { q , t}).\ ] ] this result was obtained in @xcite in a slightly weaker form , but it is possible to extend it to our case with little effort .
let from now on @xmath340.$ ] in this case we have @xmath341_{q , t}= r_{q , t},$ ] so we obtain the cohomology of the artin group of affine type @xmath69 with @xmath342coefficients by means of theorem [ t : cohomqt ] . in a similar way we get the rational cohomology of @xmath343 [ prop : shaq ] we have @xmath344 ) \\
h^*(g_{\tilde{a}_{n-1 } } , { \mathbb{q}})\cong & h^*(g_{b_{n } } , { \mathbb{q}}[[t^{\pm 1}]])\end{aligned}\ ] ] where the action of @xmath14 on @xmath312 $ ] ( and on @xmath345 $ ] ) is trivial for the generators @xmath346 and @xmath314-multiplication for the last generator @xmath315 . to obtain the rational cohomology of @xmath35
we may apply proposition [ prop : shift ] together with theorem [ teo : ratio ] .
the tong - yang - ma representation is an @xmath76-dimensional representation of the classical braid group @xmath347 discovered in @xcite .
below we just recall it , referring to @xcite for a discussion of its relevance in braid group representation theory . [
df : tong ] let @xmath348 be the free @xmath349$]-module of rank @xmath74 . the tong - yang - ma representation is the representation @xmath350}(v)\ ] ] defined w.r.t .
the basis @xmath351 of @xmath348 by : @xmath352 where @xmath353 denote the @xmath154-dimensional identity matrix and all other entries are zero .
notice that the image of the pure braid group under the tong - yang - ma representation is abelian ; hence this representation factors through the _ extended coxeter group _ presented in @xcite .
[ prop : sha2 ] we have @xmath354_{q , t})\cong h_*(g_{a_n } , m_q \otimes v)\ ] ] @xmath355_{q , t})\cong h^*(g_{a_n } , m_q \otimes v)\ ] ] where each generator of @xmath92 acts on @xmath356 by @xmath148-multiplication on the first factor and by the tong - yang - ma representation the second factor . for the statement in homology , by shapiro s lemma ,
it is enough to show that @xmath357_{q , t } \cong m_q \otimes v$ ] .
+ notice that @xmath358=n+1 $ ] and let choose as coset representatives for @xmath359 the elements @xmath360 for @xmath361 , @xmath362 , @xmath363 . + then by definition of induced representation , there is an isomorphism of left @xmath92-modules , @xmath364_{q , t}=\bigoplus_{i=1}^{n+1 } m[t^{\pm 1}]e_i\ ] ] where the action is on the r.h.s .
is as follows . for an element @xmath365 ,
write @xmath366 with @xmath367 . then
@xmath368 acts on an element @xmath369e_i$ ] as @xmath370 .
+ computing explicitly this action for the standard generators of @xmath92 , we can write the representation in the following matrix form : @xmath371 for @xmath372 , whereas @xmath373 conjugating by @xmath374 and setting @xmath375 , one obtains the desired result .
+ finally , since @xmath358=n+1<\infty$ ] , the induced and coinduced representation are isomorphic ; so the analogous statement in cohomology follows .
in particular the cohomology of @xmath320 determined in theorem [ t : cohomqt ] is isomorphic to the cohomology of @xmath92 with coefficient in the tong - yang - ma representation twisted by an abelian representation . by means of shapiro s lemma
, we may as well determine the cohomology of @xmath347 with coefficient in the tong - yang - ma representation .
indeed : [ prop : shaq2 ] we have @xmath376)&\cong h_*(g_{a_n } , v)\\ h^*(g_{b_n } , { \mathbb{q}}[t^{\pm 1}])&\cong h^*(g_{a_n } , v)\end{aligned}\ ] ] where @xmath348 is the representation of @xmath92 defined in [ df : tong ] . as a consequence
we have let @xmath348 be the @xmath76-dimensional representation of the braid group @xmath41 defined in [ df : tong ] .
then the cohomology @xmath377 is given as in theorem [ teo : ratio ] .
in particular the homology of @xmath35 with trivial coefficients is isomorphic to the homology of @xmath92 with coefficients in the tong - yang - ma representation .
we refer to @xcite for the few changes which have to be done to the construction given in @xcite ( see also @xcite ) for non - finite type artin groups ( but still finitely generated ) .
we obtain a _
finite _ cw - complex @xmath378 , explicitly described , which is a deformation retract of the _ orbit space _ of the artin group .
the latter is defined as the quotient space @xmath380\ \setminus\ \bigcup_{h\in { \mathcal a}}\ h_{{\mathbb{c}}}\ ] ] @xmath381 being the interior part of the _ tits cone _ of @xmath382 while @xmath383 is the hyperplane arrangement of @xmath384 the associated artin group @xmath385 is the fundamental group of the orbit space ( see @xcite ) .
the simplest way to realize @xmath378 is by taking one point @xmath386 inside a _ chamber _ @xmath387 and , for any maximal subset @xmath388 such that the parabolic subgroup @xmath389 is finite , construct a @xmath390-cell ( a polyhedron ) in @xmath391 as the `` convex hull '' of the @xmath389-orbit of @xmath386 in @xmath392 so , we obtain a finite cell complex which is the union of ( in general , different dimensional ) polyhedra .
next , there are identifications on the faces of these polyhedra , which are the same as described in @xcite for the finite case .
the resulting quotient space is a cw - complex @xmath378 which has a @xmath390-cell for each @xmath393 such that @xmath389 is finite .
we show an example in the case @xmath394 ( fig .
[ fig : affine ] ) .
let @xmath395 with the natural structure of simplicial complex .
then the euler characteristic of the orbit space ( so , of the group @xmath385 when such space is of type @xmath10 ) ) equals @xmath396 if @xmath11 is affine of rank @xmath74 we have @xmath397 if @xmath11 is @xmath398 ( so , all 3-subsets of @xmath56 generate an infinite group ) of rank @xmath8 then @xmath399 where @xmath400 is the number of pairs in @xmath56 having finite weight ( @xmath401 if there are no @xmath55-edges in the coxeter graph ) .
the cohomology of the orbit space in case @xmath402 with trivial coefficients is deduced from corollary [ prop : shaq ] and from theorem [ teo : ratio ] ; that with local coefficients in the @xmath403-module @xmath28 $ ] is deduced from theorem [ t : cohomqt ] .
n. bourbaki , _ lments de mathmatique .
groupes et algbres de lie .
chapitre iv : groupes de coxeter et systmes de tits .
chapitre v : groupes engendrs par des rflexions .
chapitre vi : systmes de racines _ , actualits scientifiques et industrielles , no . 1337 , hermann , paris , 1968 .
c. de concini , m. salvetti , and f. stumbo , _ the top - cohomology of artin groups with coefficients in rank-@xmath183 local systems over @xmath408 _ , special issue on braid groups and related topics ( jerusalem , 1995 ) , vol .
78 , topology appl .
, no . 1 - 2 , 1997
, pp . 520 .
r. p. kent , iv and d. peifer , _ a geometric and algebraic description of annular braid groups _ , internat
. j. algebra comput .
* 12 * ( 2002 ) , no . 1 - 2 , 8597 , international conference on geometric and combinatorial methods in group theory and semigroup theory ( lincoln , ne , 2000 )
. s. s. f. lambropoulou , _ solid torus links and hecke algebras of @xmath412-type _ , proceedings of the conference on quantum topology ( manhattan , ks , 1993 ) , world sci .
publ . , river edge , nj , 1994 , pp . | the result of this paper is the determination of the cohomology of artin groups of type @xmath0 and @xmath1 with non - trivial local coefficients .
the main result ( theorem [ t : cohomqt ] ) is an explicit computation of the cohomology of the artin group of type @xmath2 with coefficients over the module @xmath3.$ ] here the first @xmath4 standard generators of the group act by @xmath5multiplication , while the last one acts by @xmath6multiplication .
the proof uses some technical results from previous papers plus computations over a suitable spectral sequence .
the remaining cases follow from an application of shapiro s lemma , by considering some well - known inclusions : we obtain the rational cohomology of the artin group of affine type @xmath1 as well as the cohomology of the classical braid group @xmath7 with coefficients in the @xmath8-dimensional representation presented in @xcite .
the topological counterpart is the explicit construction of finite @xmath9complexes endowed with a free action of the artin groups , which are known to be @xmath10 spaces in some cases ( including finite type groups ) .
particularly simple formulas for the euler - characteristic of these orbit spaces are derived . |
millisecond pulsars have been the subject of intense discovery over the past few years .
thanks to the current generation of large - scale pulsar surveys , we find ourselves in an era where we have large samples of both millisecond and normal pulsars and , in particular for millisecond pulsars , there are many opportunities to learn about the population of objects as a whole based upon the ones we see . the contributions by keith , ng , lazarus and lynch elsewhere in these proceedings provide the latest status of these searches , and further details can be found in lorimer ( 2011 ) .
these discoveries are due to the modern surveys having low - noise receiver systems with a large fractional bandwidth and employing state - of - the - art digital data acquisition systems which are now close to optimal ( e.g. duplain et al .
2008 ) , as well as substantial computing resources using sophisticated search processing algorithms which can largely remove radio - frequency interference and combat the effects of binary motion during the observation .
in addition , multiple analyses of the data often result in additional discoveries ( e.g. keith et al .
2009 ; eatough et al .
2010 ; mickaliger et al . 2012 ) .
along with the current generation of ongoing pulsar surveys at green bank , parkes , arecibo and effelsberg , a key set of surveys that provide the backbone for much of the current population analyses are the parkes multibeam surveys .
these groundbreaking experiments were carried out using analog filterbanks and had substantially better sensitivity to millisecond pulsars compared to previous efforts .
the main surveys of interest here are the galactic plane survey ( manchester et al .
2001 ) , the swinburne intermediate ( edwards et al . 2001 ) and high - latitude ( jacoby et al . 2009 ) pulsar surveys , the high latitude survey ( burgay et al . 2006 ) and perseus arm survey ( burgay et al .
2012 ) as well as a deep multibeam survey of the northern galactic plane ( lorimer , camilo & mclaughlin 2013 ) . in spite of their good sensitivity ,
these surveys were ultimately limited by a number of selection effects which bias their pulsar samples and need to be accounted for by population analyses .
these selection effects include the inverse - square law , pulse dispersion and scattering , pulsar intermittency , interstellar scintillation and also binary motion . for further details of these effects ,
the interested reader is referred to earlier reviews on this subject ( lorimer 2009 , 2011 ) .
although early efforts to correct for observational selection were done via analytical treatments ( e.g. gunn & ostriker 1970 ) , nowadays this is best done in a monte carlo fashion to create realizations of the true of the underlying pulsar population which are then searched using models of survey detection thresholds which account for propagation in the interstellar medium .
the contribution by cordes in this volume describes current work to update the so - called `` ne2001 '' electron density model ( cordes & lazio 2002 ) which is currently used to model the propagation effects in these population syntheses .
the reader is also referred to recent work on this subject by schnitzeler ( 2012 ) . from statistical analyses of samples of artificial pulsars that satisfy the criteria for detection ,
it is possible to generate and optimize models for the pulsar population which inform us about the underlying distribution functions and make predictions for future survey yields .
although a variety of different approaches have been employed , the monte carlo simulations follow two basic strategies . in the `` snapshot '' approach
, no assumptions are made concerning the prior evolution of pulsars .
instead , the populations are simply generated according to various distribution functions ( typically in galactocentric radius , @xmath1 , height with respect to the plane , @xmath2 , spin period , @xmath3 and luminosity , @xmath4 ) which are optimized in order to find the best match to the sample .
alternatively , one may carry out `` evolution '' approaches where the model pulsars are evolved forward in time from a set of initial distributions .
software to carry out both approaches has been developed by a number of groups and some of this is freely available .
the snapshot approach was applied to the normal pulsar population by lorimer et al .
( 2006 ) who were able to derive best - fitting probability density functions in @xmath1 , @xmath4 , @xmath2 and @xmath3 for the present - day population of objects .
one result of this work was that the radial distribution of pulsars could not be decoupled from the radial distribution of free electrons in the pulsar distribution .
for the evolution approach on the normal population , the current state - of - the - art is the work of faucher - gigure & kaspi ( 2006 ) who generated excellent fits to the pulsar @xmath3-@xmath5 diagram using a model in which the luminosity has a power - law dependence on @xmath3 and @xmath5 . in their optimal model ,
@xmath4 scales as @xmath6 , i.e. the square root of the spin - down luminosity .
one interesting result from this paper is that the luminosity function of the present - day pulsar population appears to be log - normal in form .
the smooth tails of this distribution ( which are integrable over all luminosities to give a finite result ) offer a distinct advantage over previous studies which parameterized the luminosity as a power law which is divergent and requires a somewhat unphysical minimum luminosity .
similar results were found by ridley & lorimer ( 2010 ) .
the log - normal form of the luminosity distribution has subsequently been adopted as a starting point by a number of other studies ( e.g. , boyles et al .
2011 ; bagchi et al . 2011 and chennamangalam et al . 2012 ; see also these proceedings ) .
one of the first efforts to quantify the millisecond pulsar population was the work of kulkarni & narayan ( 1988 ) who used a @xmath7 approach to estimate the number of similar objects to those observed by surveys at that time . with a sample of only three millisecond pulsars ,
their study was subject to large uncertainties , but it began a significant discussion on the so - called `` birthrate problem '' for millisecond pulsars . based on their results
kulkarni & narayan ( 1989 ) claimed that the birthrate of millisecond pulsars was substantially greater than that of their proposed progenitors , the low - mass x - ray binaries .
this problem has largely disappeared as better constraints have become available from larger samples ( lorimer 2009 ) .
rathnasree ( 1993 ) attempted to synthesize the population of millisecond pulsars from low - mass x - ray binaries by carrying out monte carlo simulations to model their evolution since birth .
the current state of the art of this approach is discussed in the contribution by tauris in this proceedings .
a prescient paper by johnston & bailes ( 1991 ) demonstrated that the local population of millisecond pulsars revealed by all - sky surveys at @xmath8 ghz should be largely isotropic .
this work , and early discoveries of two recycled radio pulsars at high galactic latitudes ( wolszczsan 1990 ) inspired a number of 400 mhz pulsar surveys during the 1990s which led to a sample of about 30 objects by the end of the decade . during that time , studies of the scale height , velocity distribution and luminosity function were performed ( lorimer 1995 ; cordes & chernoff 1997 ; lyne et al .
1998 ) and it was found that the local ( within a few kpc ) millisecond pulsar population potentially observable was comparable in size to the equivalent population of normal pulsars .
one conclusion from these studies is that the populations of millisecond and normal pulsars are consistent with a single velocity distribution applied to all neutron stars at birth ( tauris & bailes 1996 ) .
we are now in an era where significant further understanding of the millisecond pulsar population should be possible in the coming years . as a starting point ,
i present here a snapshot analysis of the sample of millisecond pulsars detectable by the parkes multibeam surveys prior to the current high time - resolution universe surveys ( see keith s contribution in this proceedings , and keith et al .
the total number of millisecond pulsars from these surveys now numbers 58 .
this number turns out to have more - or - less asymptoted , but may still increase further thanks to a number of new discoveries by reanalyses of the parkes multibeam survey of the galactic plane . using the snapshot approach ,
i have developed a model ( hereafter referred to as model a ) which has the following parameters : ( i ) a log - normal luminosity function with an identical mean and standard deviation ( i.e. 1.1 and 0.9 ) to that found by faucher - gigure & kaspi ( 2006 ) .
this function was found to be consistent with recycled pulsars in globular clusters recently by bagchi et al .
( 2012 ) ; ( ii ) a manually - tweaked period distribution with a peak at 3 ms ; ( iii ) an exponential scale height with a mean of 500 pc ; ( iv ) a gaussian radial distribution with a standard deviation of 7.5 kpc .
the period distribution was arrived at by initially choosing periods from a distribution which is uniform in @xmath9 between 1 and 30 ms .
i adjusted the relative weighting of the bins to arrive at a distribution which most closely matches the observed sample .
the @xmath2 distribution was motivated by my earlier results ( lorimer 1995 ) based on the low - frequency surveys . as can be seen in fig . 1
, this model provides a reasonable match to the observed sample .
i define reasonable in this context in terms of a comparison of the observed and predicted survey yields and also by looking at the observed distributions of spin period , @xmath3 , dispersion measure , dm , galactic longitude , @xmath10 , and galactic latitude , @xmath11 . in table 1 , along with the total number of potentially observable i tabulate two figures of merit for this model : @xmath12 and @xmath13 .
the former is the base 10 logarithm of the product of the four individual kolmogorov - smirnoff ( ks ) tests between @xmath3 , dm , @xmath10 and @xmath11 .
the latter is the reduced @xmath14 computed from the observed and predicted numbers of pulsars .
.summary of the simulation results obtained from our snapshot modeling of the millisecond pulsar population . from left
to right we list the model , base-10 logarithm of the combined kolmogorov smirnoff probability ( @xmath12 ) , the reduced chi - squared value ( @xmath13 ) and the number of potentially observable millisecond pulsars in the galaxy ( @xmath15 ) . [
cols=">,>,>,>,>",options="header " , ] also listed in the table are the equivalent numbers for a number of other models b i .
the approach i took here was to investigate the impact of each assumption made in model a by changing it , but keeping all other assumptions constant . in model b , to show the impact on the choice of luminosity function on the results , i made only a small change to mean of the log - normal function , i.e. 1.1 to 1.2 .
the figures of merit are comparable to model a , but the number of pulsars increases by about 14% . for model
b , to show that the sample does require a radial dependence on number density , i generated pulsars assuming constant number density on the plane throughout the model galaxy . here , both figures of merit are substantially poorer due to the gain in number of detections at higher galactic longitudes .
the number of pulsars required in the model is substantially reduced , since it is now much easier to detect the population at larger @xmath1 .
models g and h , where the scale length is varied by @xmath16 kpc from the value in model a show that our ability to constrain the scale length is currently not very good . in model d ,
i adopt the cordes & chernoff ( 1997 ) spin period distribution .
although the relative survey yields are satisfactory , this distribution predicts a substantial fraction of millisecond pulsars with @xmath17 ms which is much higher than that observed .
a similar result is found using a gaussian period distribution ( model i ) with a mean of zero and a standard deviation of 10 ms .
reducing the scale height of the population in model e to 100 pc substantially worsens the agreement with the observed data , while increasing the scale height to 1 kpc ( model f ) has less of an effect .
the analysis presented here will be described further in a forthcoming parkes multibeam survey paper .
it represents a first step towards a more detailed understanding of the millisecond pulsar population in the galaxy .
further work is encouraged to account for the following subtleties not included here . of particular interest
are studies of the motion of millisecond pulsars in the @xmath18 diagram and the relationship to the low - mass x - ray binary population .
the work of kiziltan & thorsett ( 2010 ) and tauris et al .
( 2012 ) relates directly to the first issue .
further work in this area , along the lines of the population syntheses carried out by story et al .
( 2007 ) seem to be the next logical step .
significant progress is now being made in modeling the binary evolutionary steps and predicting distributions for orbital parameters for the binary population ( see , for example , belszcynski et al .
2008 ) . combining all these elements into an all - encompassing synthesis of the millisecond pulsar population which accounts ( as far as possible ) for the observational selection effects
is now a major goal of future studies . on the road to such a lofty goal
, past experience with the normal pulsar population ( see , for example , lorimer 2009 ) suggests that it will be extremely profitable to break up the steps into a number of smaller problems .
one such example is the radio - selected sample of millisecond pulsars revealed by _
( ray et al .
2012 ) . a careful study of the selection effects impacting this sample should now be undertaken in order to fully understand the impact of these discoveries on our knowledge of the millisecond pulsar population | among the current sample of over 2000 radio pulsars known primarily in the disk of our galaxy , millisecond pulsars now number almost 200 . due to the phenomenal success of blind surveys of the galactic field , and
targeted searches of _ fermi _ gamma - ray sources , for the first time in over a decade , galactic millisecond pulsars now outnumber their counterparts in globular clusters ! in this paper , i briefly review earlier results from studies of the galactic millisecond pulsar population and present new constraints based on a sample of 60 millisecond pulsars discovered by 20 cm parkes multibeam surveys .
i present a simple model of the population containing @xmath0 potentially observable millisecond pulsars with a luminosity function , radial distribution and scale height that matches the observed sample of objects .
this study represents only a first step towards a more complete understanding of the parent population of millisecond pulsars in the galaxy and i conclude with some suggestions for further study in this area . |
given an undirected graph @xmath0 with vertex set @xmath1 and edge set @xmath2 , a k - coloring of @xmath3 is a function @xmath4 that assigns to each vertex @xmath5 a color @xmath6 , where @xmath7 . a k - coloring is considered legal if each pair of vertices @xmath8 connected by an edge @xmath9 receive different colors @xmath10 . the minimum sum coloring problem ( mscp ) is to find a legal k - coloring @xmath11 such that the total sum of colors over all the vertices @xmath12 is minimized .
the minimum value of this sum is called the chromatic sum of @xmath3 and denoted by @xmath13 .
the number of colors related to the chromatic sum is called the strength of the graph and denoted by @xmath14 .
the mscp is np - hard for general graphs @xcite and provides applications mainly including vlsi design , scheduling and resource allocation @xcite .
given the theoretical and practical significance of the mscp , effective approximation algorithms and polynomial algorithms have been presented for some special cases of graphs , such as trees , interval graphs and bipartite graphs @xcite . for the purpose of practical solving of the general mscp ,
a variety of heuristics have been proposed in recent years , comprising a parallel genetic algorithm @xcite , a greedy algorithm @xcite , a tabu search algorithm @xcite , a hybrid local search algorithm @xcite , an independent set extraction based algorithm @xcite and a local search algorithm @xcite . on the other hand ,
binary quadratic programming ( bqp ) has emerged during the past decade as a unified model for a wide range of combinatorial optimization problems , such as set packing @xcite , set partitioning @xcite , generalized independent set @xcite , maximum edge weight clique @xcite and maxcut @xcite . a review concerning the additional applications and the reformulation procedures can be found in @xcite .
this bqp approach has the advantage of directly applying an algorithm designed for bqp to solve other classes of problems rather than resorting to a specialized solution method .
moreover , this approach proves to be competitive or even better than the special algorithms proposed for several problems . in this paper , we investigate for the first time the application of this bqp approach to solve the mscp problem .
we propose a binary quadratic formulation for the mscp which is solved by our path relinking algorithm previously designed for the general bqp . to assess the performance of the proposed approach , we present computational results on a set of 23 benchmark instances from the literature and contrast these results with those of several reference algorithms specifically dedicated to the mscp .
the rest of this paper is organized as follows .
section [ sec_trans ] illustrates how to transform the mscp into the bqp formulation .
section [ sec_pra ] presents an overview of our path relinking algorithm for the general bqp .
section [ sec_results ] is dedicated to computational results and comparisons with other reference algorithms in the literature .
the paper is concluded in section [ sec_conclusion ] .
given an undirected graph @xmath0 with vertex set @xmath1 ( @xmath15 ) and edge set @xmath2 .
let @xmath16 be 1 if vertex @xmath17 is assigned color @xmath18 , and 0 otherwise .
the linear programming model for the mscp can be formulated as follows : @xmath19 the linear model of the mscp can be recast into the form of the bqp according to the following steps : for the constraints @xmath20 , we represent these linear equations by a matrix @xmath21 and incorporate the following penalty transformation @xcite : @xmath22 \ \ \ \ \ \ \ \ \ \ \ \ \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ = p[x^{t}(a^{t}a)x - x^{t}(a^{t}b)-(b^{t}a)x]+pb^{t}b \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ = xd_{1}x+c \ \ \ \ \ \ \ \end{split}\ ] ] for the constraints @xmath23 , we utilize the quadratic penalty function @xmath24 to replace each inequality @xmath25 in @xmath23 and add them up as follows @xcite : @xmath26 where @xmath27 if @xmath28 and 0 otherwise . to construct the nonlinear bqp formulation @xmath29 , we first inverse the minimum objective of the mscp to be @xmath30 in accordance with the general bqp model under a maximum objective , which becomes the first component of @xmath29
. then we add the penalty function @xmath31 into @xmath29 such that @xmath32 if all the linear equations in @xmath20 are satisfied and otherwise @xmath31 is a penalty term with large negative values .
in the same way , we add the penalty function @xmath33 into @xmath29 .
hence , the resulting bqp formulation for the mscp can be expressed as follows : @xmath34 once the optimal objective value for this bqp formulation is obtained , the minimum sum coloring value can be readily obtained by taking its inverse value .
further , a penalty scalar @xmath35 is considered to be suitable as long as its absolute value @xmath36 is larger than half of the maximum color ( @xmath37 ) . consider that penalty functions should be negative under the case of a maximal objective , we select @xmath38 for the benchmark instances experimented in this paper . the optimized solution @xmath39 obtained by solving the nonlinear bqp formulation indicates that such selection ensures both @xmath31 and @xmath33 equal to 0 .
in other words , each variable @xmath16 with the assignment of 1 in the optimized solution @xmath39 forms a feasible k - coloring in which vertex @xmath17 gets the color @xmath18 . to illustrate the transformation from the mscp to the bqp formulation , we consider the following graph with @xmath40 and expect to find a legal @xmath41-coloring with @xmath42 .
+ its linear formulation according to equation ( [ lp ] ) is : @xmath43 choosing the scalar penalty @xmath44 , we obtain the following bqp model : @xmath45 where @xmath46 is written as : @xmath47 solving this bqp model yields @xmath48 ( all other variables equal zero ) and the optimal objective function value @xmath49 .
reversing this objective function value leads to the optimum ( the minimum sum coloring ) of 6 for this graph .
path relinking @xcite is a general search strategy closely associated with tabu search and its underlying ideas share a significant intersection with the tabu search perspective @xcite , with applications in a variety of contexts where it has proved to be very effective in solving difficult problems .
our previous path relinking algorithm ( pr ) is directly utilized to solve the mscp in its nonlinear bqp form as expressed in eq .
the proposed path relinking algorithm is mainly composed of the following components : _ refset initialization _ , _ solution improvement _ , _ relinking _ , _ solution selection _ , _ refset updating _ and _ refset rebuilding_. below we overview the general scheme of our path relinking algorithm .
more details can be found in @xcite .
firstly , the _ refset initialization
_ method is used to create an initial set of elite solutions _ refset _ where each solution is obtained by applying a tabu search procedure to a randomly generated solution .
secondly , for each pair of solutions ( @xmath50 , @xmath51 ) in _ refset _ we undertake the following operations : ( 1 ) apply a _ relinking _ method to generate two paths ( from @xmath50 to @xmath51 and from @xmath51 to @xmath50 ) by exploring trajectories ( strictly confined to the neighborhood space ) that connect high - quality solutions .
the first and last solutions of the path are respectively called the initiating and guiding solutions . at each step of building the path from the initiating solution to the guiding solution , we randomly select a variable from the set of variables for which @xmath50 and @xmath51 ) have different values ; ( 2 ) apply a _ solution selection _ method to select one solution from those generated on the path by reference both to its quality and to the hamming distance of this solution to the initiating solution and the guiding solution . this selected solution is then submitted to the tabu search based _ improvement method _ ; ( 3 ) apply a _ refset updating _ method to decide if the newly improved solution is inserted in _
refset _ to replace the worst solution .
the above procedure is called a round of our path relinking procedure . finally , once a round of path relinking procedure is completed , a _ refset rebuilding
_ method is launched to rebuild _ refset _ that is used by the next round of the path relinking procedure .
the path relinking algorithm terminates as soon as its stop condition ( e.g. a fixed computing time ) is satisfied .
to assess this bqp approach for the mscp , we carry out experiments on a set of 23 graphs , which are the most used benchmark instances in the literature .
our experiments are conducted on a pc with pentium 2.83ghz cpu and 8 gb ram . the time limit for a single run of our path relinking algorithm
is set as follows : 1 hour for the first 16 instances in table [ table_results ] ; 10 hours for dsjc125.1 , dsjc125.5 , dsjc125.9 , dsjc250.1 and dsjc250.5 ; 20 hours for dsjc250.9 and dsjc500.1 .
given the stochastic nature of our pr algorithm , each problem instance is independently solved 20 times .
the tabu tenure @xmath52 and the improvement cutoff @xmath53 @xcite are two parameters in the tabu search based improvement method a key component of the pr algorithm . according to preliminary experiments , we set @xmath54 ( where @xmath55 denotes the number of variables in the resulting bqp model and rand(50 ) receives a random integer ranging from 1 to 50 ) .
in addition , we set @xmath56 for the improvement of the initial solutions in @xmath57 and @xmath58 for the improvement of the solutions on the path , respectively .
table [ table_results ] presents the computational statistics of the bqp model for the mscp .
columns 1 to 3 give the instance names _ instances _ along with the vertex number @xmath1 and edge number @xmath2 of the graphs . columns 4 and 5 show the number of colors @xmath41 to be used and the number of variables @xmath55 in the bqp formulation .
column 6 summarizes the best known results _
bkr _ from the previous literature @xcite .
the columns under the heading of bqp - pr report our results of the bqp model solved by the pr algorithm : the best objective values @xmath59 , the average objective values @xmath60 , the standard deviation @xmath61 , the average time @xmath62 ( in seconds ) to reach the best objective value @xmath59 over 20 runs , and the average time ( in seconds ) @xmath63 consumed to reach the best objective value obtained in each run .
the last row shows the average performance over a total of 23 tested instances in terms of the deviations of the best and average solution values from the _
bkr_. notice that the results marked in bold in the @xmath59 column indicate that bqp - pr reaches the _ bkr _ on these instances .
we also applied cplex v12.2 to the linear model ( [ lp ] ) ( section [ linear model for the mscp ] ) to solve these mscp instances .
cplex was successful in finding an optimal solution for 15 instances ( marked with an asterisk ) , but terminated abnormally for the remaining 8 instances due to excess requirements of memory .
.computational statistics of the bqp - pr approach for the mscp [ cols= " < , < , < , < , < , < , < , < , < , < , < , < , < " , ] as we can observe in table [ table_cmp ] , the proposed bqp - pr approach outperforms hls , mrlf , pga and ts in terms of the best solution values .
specifically , bqp - pr finds better solutions than hls , mrlf , pga and ts for 3 , 14 , 8 and 7 instances , respectively .
finally , bqp - pr performs less well compared with the most effective mscp heuristics exscol and mds(5 ) . from this experiment
, we conclude that our bqp approach combined with the pr algorithm for tacking the mscp constitutes an interesting alternative to specific algorithms tailored to this problem . in this section ,
we discuss some limitations of the proposed bqp - pr approach for solving the mscp .
first , the proposed method may require considerable computing time to reach its best solutions for large graph instances ( see column @xmath63 in table [ table_results ] ) .
this can be partially explained by the fact that the number of the bqp variables ( equaling @xmath64 where @xmath1 is the number of vertices and @xmath41 is the number of colors ) sharply increases with the growth of @xmath1 and @xmath41 .
additionally , at present our approach is not able to solve graph instances with bqp variables surpassing the threshold value of 20,000 because of the memory limitation .
these obstacles could be overcome by designing more effective data structures used by the bqp algorithms .
we have investigated the possibility of solving the np - hard minimum sum coloring problem ( mscp ) via binary quadratic programming ( bqp ) .
we have shown how the mscp can be recast into the bqp model and explained the key ideas of the path - relinking algorithm designed for the general bqp .
experiments on a set of benchmark instances demonstrate that this general bqp approach is able to reach competitive solutions compared with several special purpose mscp algorithms even though considerable computing time may be required .
the work is partially supported by the pays de la loire " region ( france ) within the radapop ( 2009 - 2013 ) and ligero ( 2010 - 2013 ) projects . | in recent years , binary quadratic programming ( bqp ) has been successively applied to solve several combinatorial optimization problems .
we consider in this paper a study of using the bqp model to solve the minimum sum coloring problem ( mscp ) .
for this purpose , we recast the mscp with a quadratic model which is then solved via a recently proposed path relinking ( pr ) algorithm designed for the general bqp . based on a set of mscp benchmark instances , we investigate the performance of this solution approach compared with existing methods .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
for a given contact 3-manifold , let us consider the 5-dimensional open book whose page is a stein filling of the contact 3-manifold and whose monodromy is the identity map .
this open book supports a contact structure on the resulting closed 5-manifold , such contact structures were extensively studied , for example in @xcite ding - geiges - van koert classified a certain class of contact 5-manifolds . from the view point of 4-dimensional symplectic topology ,
it is natural to ask how the choices of stein fillings of a fixed contact 3-manifold affect the resulting contactomorphism types of @xmath3-manifolds .
however , not much is known to the best of the authors knowledge . here
we use the following terminology .
we say that a family of 5-dimensional open books has pairwise exotic stein pages , if pages of its members are pairwise exotic ( i.e. homeomorphic but non - diffeomorphic as smooth 4-manifolds ) , and are stein fillings of the same contact 3-manifold .
recently , ozbagci - van koert @xcite showed that infinitely many open books with exotic stein pages and the identity monodromy can support pairwise distinct contact structures on the same smooth 5-manifold .
they used the exotic stein manifolds of @xcite as pages , and used @xcite to distinguish their contact structures . in this paper , we show that distinct open books with exotic stein pages can support the same contact 5-manifold , contrary to the expectation from the result of ozbagci - van koert .
[ intro : thm : main ] there exists a contact 5-manifold supported by infinitely many distinct open books with pairwise exotic stein pages and the identity monodromy .
in fact , any closed contact 5-manifold with @xmath4 , that is induced from the boundary of a subcritical stein manifold without 1-handles , is such an example .
note that by @xcite any 6-dimensional subcritical stein manifold is the product of @xmath5 and a 4-dimensional stein manifold .
therefore , contact 5-manifolds in this theorem are the contact boundary of the product of @xmath5 and stein manifolds without 1-handles .
such smooth 5-manifolds are always diffeomorphic to either @xmath6 or @xmath7 with @xmath8 , and each of them admits infinitely many contact structures ( cf .
@xcite , see also remark [ rem : subcritical ] ) .
in contrast to the above theorem , we also extend the aforementioned result of ozbagci - van koert .
let @xmath9 denote the total space of the non - trivial @xmath10-bundle over @xmath11 , then the following holds ( they proved the @xmath12 case ) . [ intro : thm : extension of ov ] for a closed 5-manifold diffeomorphic to either @xmath6 or @xmath7 with @xmath8 , there exist infinitely many open books with pairwise exotic stein pages and the identity monodromy which support pairwise distinct contact structures on the 5-manifold .
here we outline our proofs of these results .
these exotic stein pages are diffeomorphic to exotic stein fillings obtained in @xcite and @xcite . to obtain the results ,
we construct new stein structures ( and hence boundary contact structures ) on these 4-manifolds .
we distinguish 5-dimensional contact structures using results of @xcite .
regarding theorem [ intro : thm : extension of ov ] , we use the argument similar to @xcite .
the new ingredient is 4-dimensional exotic stein fillings obtained in @xcite .
also , we use yet another new stein structures on these exotic 4-manifolds . + * acknowledgements*. this work was done during the second author s stay at michigan state university .
he would like to thank them for their hospitality .
in this section , we construct new stein structures on certain exotic smooth 4-manifolds . for basics of handlebody , stein structures and contact structures , the readers can consult @xcite , @xcite and @xcite .
we first recall the infinitely many pairwise exotic 4-manifolds which are stein fillings of the same contact 3-manifolds . for integers
@xmath13 satisfying @xmath14 , @xmath15 , @xmath16 , and @xmath17 , let @xmath18 and @xmath19 be the simply connected compact smooth 4-manifolds with @xmath20 given by the handlebody diagrams in figure [ fig : smooth_nuclei ] , where we put @xmath21 .
the @xmath22 case of these manifolds was studied in @xcite using stein handlebody , and the general case was subsequently discussed in @xcite using lefschetz fibration .
we remark that each @xmath19 is obtained from @xmath18 by a luttinger surgery ( i.e. a logarithmic transformation with the multiplicity one ) , as shown in @xcite and @xcite . and @xmath23 ( @xmath14 , @xmath24 , @xmath16 , @xmath17).,width=384 ] these 4-manifolds have the following properties , where we fix a given @xmath25-tuple @xmath26 of positive integers satisfying @xmath14 and @xmath24 .
[ thm : recall : nuclei ] the infinite family @xmath27 contains infinitely many pairwise homeomorphic but non - diffeomorphic smooth 4-manifolds which are stein fillings of the same contact 3-manifold .
[ thm : recall : boundary_sum ] for a given stein filling @xmath28 of a contact 3-manifold , the infinite family @xmath29 contains infinitely many pairwise homeomorphic but non - diffeomorphic smooth 4-manifolds which are stein fillings of the same contact 3-manifold .
[ remark : exotic_stein ] these theorems hold for any choices of stein structures on @xmath30 s , since every closed 3-manifold admits at most finitely many stein fillable contact structures up to contactomorphism ( see lemma 3.4 in @xcite ) .
note that the boundary contact structures in these theorems depend on the choices of stein structures on these 4-manifolds .
next we construct a new stein structure on @xmath30 to prove theorem [ intro : thm : main ] .
( for theorem [ intro : thm : extension of ov ] , we use yet another stein structure . ) by canceling 1-handles of @xmath30 , we obtain figure [ fig:2-handle_nuclei ] . from this picture
, we obtain the stein handlebody diagram of @xmath30 in figure [ fig : stein_nuclei_odd ] , in the case where @xmath31 is odd . here the boxes in the stein picture means the legendrian version of left handed full twists shown in figure [ fig : legendrian_left_twists ] .
note that each framing in the diagram is one less than the thurston - bennequin number .
we equip @xmath23 with the stein structure given by this picture .
one can check the lemma below .
[ lem : picture : rotation ] the rotation numbers of the attaching circles of the stein handlebody @xmath23 in figure [ fig : stein_nuclei_odd ] are @xmath32 and @xmath33 . , width=384 ] with @xmath34,width=364 ]
in this section we prove theorem [ intro : thm : main ] applying stein fillings in the previous section . throughout this section , we assume @xmath21 satisfies @xmath34 .
let @xmath35 be the contact 5-manifold supported by the open book @xmath36 .
for a positive integer @xmath37 , let @xmath38 be a 4-dimensional stein handlebody with @xmath39 and without 1-handles such that the rotation number of the attaching circle of each 2-handle is @xmath32 .
one can easily find such a @xmath38 for each @xmath37 .
we denote by @xmath40 the boundary connected sum @xmath41 , and equip it with the stein structure induced from its natural stein handlebody .
let @xmath42 be the contact 5-manifold supported by the open book @xmath43 .
first we recall the following useful results ( @xmath9 denotes the total space of the non - trivial @xmath10-bundle over @xmath11 ) .
[ thm : dgv : knot ] let @xmath44 @xmath45 be a legendrian knot in the standard contact @xmath10 , and let @xmath46 be the contact 5-manifold supported by the open book with the identity monodromy such that the page is the stein filling obtained from @xmath47 by attaching a 2-handle along @xmath44 with contact @xmath48 framing .
then @xmath49 and @xmath50 are contactomorphic to each other if and only if the absolute values of the rotation numbers of @xmath51 ad @xmath52 are eaqual to each other .
furthermore , @xmath53 is diffeomorphic to @xmath54 @xmath55resp .
@xmath9@xmath56 , if the rotation number of @xmath51 is even @xmath55resp .
odd@xmath56 .
[ thm : dgv]let @xmath49 and @xmath57 be two simply connected closed contact 5-manifolds admitting subcritical stein fillings without 1-handles .
if there exists an isomorphism @xmath58 that sends @xmath59 to @xmath60 , then @xmath49 and @xmath57 are contactomorphic to each other . for simplicity of the notation , we use the following terminology . for a 4-dimensional stein handlebody @xmath61 with @xmath62 and without 1-handles ,
we call the maximal common divisor of the rotation numbers of the attaching circles of the 2-handles of @xmath61 as the rotation divisor of @xmath61 , and denote it by @xmath63 . in the case where all the rotation numbers of the attaching circles are @xmath32 , @xmath63 is defined to be @xmath32 . note that each attaching circle is a legendrian knot in the standard contact @xmath10 .
we show the following proposition using the above theorems .
[ prop : dgv : rotation ] let @xmath64 be 4-dimensional stein handlebodies with @xmath39 and without 1-handles . then two contact 5-manifolds @xmath65 and @xmath50 supported by the open books @xmath66 and @xmath67 are contactomorphic to each other , if and only if @xmath68 .
furthermore , @xmath53 is diffeomorphic to @xmath69 @xmath55resp .
@xmath7@xmath56 , if @xmath70 is even @xmath55resp .
odd@xmath56 . due to the proof of proposition 4.5 in @xcite
, we see that there exists an isomorphism @xmath71 which sends @xmath59 to @xmath72 .
therefore , according to theorem [ thm : dgv ] , there exists an isomorphism @xmath73 which maps @xmath72 to @xmath74 , if and only if @xmath65 and @xmath50 are contactomorphic to each other .
a result of gompf @xcite tells that @xmath72 is represented by a cocycle whose value on the 2-chain corresponding to each 2-handle of @xmath75 is the rotation number of the attaching circle of the 2-handle . since @xmath75 is simply connected , we see that @xmath76 is isomorphic to @xmath77 .
the maximal divisor of @xmath72 is thus equal to @xmath70 .
therefore , there exists an isomorphism @xmath73 which maps @xmath72 to @xmath74 if and only if @xmath68 .
note that , for any two primitive elements of @xmath78 , there exists an automorphism of @xmath78 which maps one to the other .
the former claim of the proposition is now straightforward .
we check the latter claim of the proposition . due to the former claim
, we may assume that @xmath75 is the boundary sum of stein handlebodies each of which is obtained from @xmath47 by attaching a 2-handle along a legendrian unknot with the contact @xmath48 framing , and that the parity of the rotation number of each legendrian unknot coincides with the one of @xmath70 .
theorem [ thm : dgv : knot ] thus implies the latter claim of the proposition . as a consequence of this proposition
, we see that rotation divisors determine supporting contact structures .
we denote the resulting contact structure as follows .
[ def : classification]for a 4-dimensional stein handlebody with @xmath39 and without 1-handles , its rotation divisor @xmath79 uniquely determines a contact 5-manifold supported by the open book whose page is the given stein handlebody and whose monodromy is the identity .
we denote the resulting contact structure by @xmath80 .
note that the resulting 5-manifold is diffeomorphic to @xmath81 ( resp .
@xmath82 ) , if @xmath79 is even ( resp .
[ rem : subcritical ] @xmath83 due to the above proposition , @xmath80 is contactomorphic to @xmath84 if and only if @xmath85 and @xmath86 .
+ @xmath87 any closed contact 5-manifold admitting a subcritical stein filling without 1-handles is contactomorphic to some @xmath80 .
this is because a contact 5-manifold admits a subcritical stein filling without 1-handles if and only if the contact structure admits an open book with the identity monodromy whose page is a stein handlebody without 1-handle .
in fact , such a 6-dimensional subcritical stein filling is the product of @xmath5 and a 4-dimensional stein filling without 1-handles . for these facts ,
see for example @xcite .
note that the boundary 5-manifold is simply connected , since it is a double of a 5-dimensional handlebody consisting of only 0- and 2-handles .
now we can immediately determine the contactomorphism types of our contact 5-manifolds from lemma [ lem : picture : rotation ] and definition [ def : classification ] .
[ prop : contact_type_5-dim ] @xmath83 @xmath88 is contactomorphic to @xmath89 if @xmath90 , and to @xmath91 if @xmath92 .
+ @xmath87 @xmath93 is contactomorphic to @xmath94 if @xmath90 , and to @xmath95 if @xmath92 .
we are ready to prove theorem [ intro : thm : main ] . due to remark [ rem : subcritical ]
, it suffices to prove that , for each @xmath96 and each @xmath97 , the contact structure @xmath98 is supported by infinitely many distinct open books with pairwise exotic stein pages and the identity monodromy .
we assume @xmath21 satisfies @xmath14 , @xmath99 , @xmath100 and @xmath16 . for @xmath101 ,
proposition [ prop : contact_type_5-dim ] tells that , each @xmath93 is contactomorphic to @xmath102 if @xmath103 is even , and to @xmath95 if @xmath103 is odd .
therefore , varying @xmath104 and applying theorem [ thm : recall : boundary_sum ] , we obtain the desired claim in the @xmath101 case . applying the same argument to @xmath88
, we obtain the @xmath105 case .
in this section we prove theorem [ intro : thm : extension of ov ] by constructing a new stein structure on @xmath30 . in the rest of this section
, we use the same symbol as those in the previous sections .
however , we give different stein or contact structures for the smooth manifolds @xmath106 , @xmath107 , @xmath108 and @xmath109 . for a positive integer @xmath104 and a 3-tuple @xmath21 of positive integers with @xmath14 , we obtain the stein handlebody diagram of @xmath30 in figure [ fig : many_contact ] using figure [ fig:2-handle_nuclei ] .
we equip @xmath30 with the stein structure given by this stein handlebody diagram .
note that we do not need the extra condition @xmath34 unlike the previous section .
one can check the rotation numbers .
we define contact 5-manifolds @xmath35 and @xmath42 in the same way as section [ sec : main ] , with respect to the above stein structure on @xmath23 . similarly to the proof of proposition [ prop : contact_type_5-dim ]
, we can determine these contact structures .
[ prop : contact_type_many ] @xmath83 @xmath88 is contactomorphic to @xmath111 if @xmath112 , and to @xmath113 if @xmath114 .
+ @xmath87 @xmath93 is contactomorphic to @xmath115 if @xmath112 , and to @xmath116 if @xmath114 .
we fix the parameter @xmath21 with @xmath14 and @xmath24 .
proposition [ prop : contact_type_many ] tells that each @xmath93 is contactomorphic to @xmath117 if @xmath118 is even , and to @xmath116 if @xmath118 is odd .
therefore , varying the parameter @xmath104 and using theorem [ thm : recall : boundary_sum ] and remark [ rem : subcritical ] , we see that both @xmath119 and @xmath120 satisfies the desired claim in the case @xmath101 .
applying the same argument to @xmath88 , we obtain the @xmath105 case
. therefore the theorem follows .
the proof of theorem 4.8 in the paper @xcite of ding - geiges - van koert together with the proof of our theorem [ intro : thm : main ] tells that , for any fixed @xmath37 and @xmath121 , our stein handlebodies @xmath107 in section [ sec : main ] are related to each other by certain moves of 2-handles ( the same holds for @xmath106 s ) .
according to the recent paper @xcite of ozbagci - van koert , for 4-dimensional stein handlebodies @xmath75 and @xmath122 , the 6-dimensional subcritical stein filling @xmath123 is symplectically deformation equivalent to @xmath124 with the contactomorphic boundary , if @xmath75 is related to @xmath122 by the above mentioned moves of handles .
hence , assuming their result , we see that infinitely many exotic 4-dimensional stein fillings of a fixed contact 3-manifold can become the same 6-dimensional subcritical stein filling ( up to symplectic deformation ) after taking the product with @xmath5 .
this seems to dash the hopes that smooth exotic structures of stein @xmath125-manifolds might be detected by the complex structures of the @xmath126-manifolds obtained by the map @xmath127 .
k. cieliebak , _ subcritical stein manifolds are split _ , arxiv : math/0204351v1 f. ding , h. geiges and o. van koert , _
diagrams for contact 5-manifolds _ , j. lond . math .
* 86 * ( 2012 ) , no . 3 , 657682 . y. eliashberg
. _ topological characterization of stein manifolds of dimension @xmath128 _ , international j. of math .
vol . 1 ( 1990 ) ,
29 - 46 .
gompf , _ handlebody construction of stein surfaces _ , ann . of math .
( 2 ) * 148 * ( 1998 ) , no . 2 , 619693 .
gompf and a.i .
stipsicz , _ @xmath125-manifolds and kirby calculus _ , graduate studies in mathematics , * 20*. american mathematical society , 1999 .
b. ozbagci and a i. stipsicz , _ surgery on contact 3-manifolds and stein surfaces _ ,
bolyai society mathematical studies , 13 .
springer - verlag , berlin ; janos bolyai mathematical society , budapest , 2004
. b. ozbagci and o. van koert , _ contact open books with exotic pages _ , arxiv:1502.03233 . | we construct a contact 5-manifold supported by infinitely many distinct open books with the identity monodromy and pairwise exotic stein pages ( i.e. pages are pairwise homeomorphic but non - diffeomorphic stein fillings of a fixed contact 3-manifold ) , moreover we describe a process of generating infinitely many such examples .
in contrast to this result , on each of @xmath0 and @xmath1 @xmath2 we construct infinitely many open books with pairwise exotic stein pages ( and identity monodromy ) supporting mutually distinct contact structures . |
hydroxyl ( oh ) masers are found in high - mass star - forming regions .
they trace the local velocity and magnetic field and therefore provide clues to understanding the physical conditions of material surrounding newly - formed high - mass stars . but an unanswered question is whether the physical conditions in masing regions are representative of the surrounding material . in order for oh masing to occur , a large column density of
oh must have velocity coherence such that the velocity gradient along the amplification path of the masing clump does not exceed the masing linewidth . in principle
, sufficient column density can be achieved in two ways .
if a masing clump is much denser than the ambient gas , the total oh column density can be large even if the physical extent of the clump is not .
alternatively , masing may occur along favored paths of velocity coherence even in a medium of homogeneous density . while it is uncertain which of these two scenarios is prevalent , two pieces of evidence suggest that density enhancements may not be necessary for oh masing to occur .
first , 13434 mhz oh _ absorption _ in w3(oh ) shows zeeman splitting indicating a line - of - sight magnetic field strength of 3.1 mg @xcite .
vlbi measurements of 1665 and 1667 mhz oh _ masers _ in w3(oh ) imply similar magnetic field strengths ( e.g. , * ? ? ?
the magnetic field strength of collapsing ( or collapsed ) material scales as the density @xmath10 , where @xmath11 , a result supported both by theoretical modelling @xcite and observations of molecular clouds in various stages of collapse ( e.g. , * ? ? ?
since the oh absorption and maser magnetic fields are comparable , this suggests that the density at masing sites is similar to that of the ambient cloud of oh . second , ammonia ( nh@xmath12 ) observations of w3(oh ) indicate that the density of material in the clumps of maser emission is roughly the same ( to within a factor of two ) as the density of the interclump material @xcite . since the velocity and extent of nh@xmath12 absorption is similar to that of the oh emission in w3(oh ) , it is a reasonable assumption that nh@xmath12 and oh exist in the same cloud of material .
finally , @xcite , when modeling multi - transition oh observations of w3(oh ) , find that maser emission in certain lines and absorption in others can be explained for the same range of densities , between @xmath13 and a few times @xmath14 @xmath15 ( for a temperature of 150 k ) . is w3(oh ) a special case , or is it representative of all interstellar oh maser sources ?
in particular , is any density enhancement required for the onset of oh masing ?
these are questions that motivated us to observe a wider range of sources than @xcite with the high sensitivity that the gbt can afford .
the observations were performed on 2004 april 11 and 12 using the national radio astronomy observatory s robert c. byrd green bank telescope ( gbt ) in green bank , wv .
the gbt has an effective diameter of 100 m. observations were taken in both circular polarizations with the gregorian focus ku - band receiver .
the gbt spectrometer was configured in 9-level mode to provide 8192 uniform - weighted spectral channels in each of two ifs covering a bandwidth of 12.5 mhz each , centered on 13441.4173 and 13434.6374 mhz , doppler - shifted to the lsr velocity of each source .
these correspond to the frequencies of the @xmath5 and @xmath1 main - line @xmath0 transitions , respectively @xcite .
each spectral channel had a resolution of 1.5 khz , corresponding to a resolution of 0.034 km s@xmath9 in velocity space .
thus , the 13433.982 mhz ( @xmath16 ) and 13442.072 mhz ( @xmath17 ) satellite lines were also in the observed frequency range . at these frequencies ,
the fwhm beamwidth of the gbt is 55 and the gain is 1.7 k / jy .
double beam switching was employed such that the source appeared in each of the dual beams alternately for a period of two minutes .
the beams were separated by 330 in azimuth .
dynamic pointing and focusing were used , and the pointing and focus were checked hourly as well as each time the telescope was pointed at a source .
variable rain , heavy at times , fell throughout the data collection period . during the brief interludes without precipitation ,
the ( non - zenith ) system temperature approached being receiver - limited at about 30 k. at times , the system temperature was over 100 k , due to heavy precipitation and low elevation angles .
our sources were chosen according to two criteria .
first , they must contain a strong background region so that absorption might be observed .
second , since it was our intent to compare the magnetic field strength in the oh gas seen in absorption with that seen in maser emission , we sought sources in which the masers indicated a uniform field distribution and eliminated any with a reversal of the line - of - sight direction of the magnetic field as determined from ground - state oh maser zeeman splitting @xcite .
only about 10 sources observable at the latitude of the gbt meet these criteria .
we observed the six sources listed in table [ source - table ] .
lllll w3(oh ) & 02 27 03.70 & @xmath1861 52 25.4 & 276 & 0.006 + [email protected] & 18 10 28.61 & @xmath419 55 49.7 & 270 & 0.008 + [email protected] & 18 42 58.04 &
@xmath404 13 58.0&10 & 0.026 + w49 & 19 10 11.04 & @xmath1809 05 20.2&58 & 0.028 + k3@xmath450 & 20 01 45.73 & @xmath1833 32 45.3 & 178 & 0.010 + on 1 & 20 10 09.05 & @xmath1831 31 35.2&20 & 0.015 + asingle - channel rms noise in stokes i and v.
the masers we detected are shown in figures [ w3ohlr1 ] to [ on1lr1a ] and discussed in [ masers ] .
the absorption features we detected are shown in figures [ w3ohi1 ] to [ on1i2 ] and discussed in [ absorption ] .
the absorption spectra have been hanning weighted for clarity , although the analyses are based on the uniform - weighted data .
the antenna temperatures are related to the flux density by @xmath19 , where @xmath20 is the effective collecting area of the telescope .
we have used the convention that stokes @xmath21 and stokes @xmath22 rcp to be consistent with @xcite . ]
thus , @xmath23 k in both rcp and lcp would result in @xmath24 jy in stokes i. zeeman measurements in the oh absorption are discussed in [ zeeman ] .
remarks about absorption line ratios are presented in [ line - ratios ] .
masers were found in two sources : w3(oh ) and on 1 .
parameters of the maser lines are listed in table [ maser - table ] .
magnetic field estimates assume a zeeman splitting coefficient of 0.178 km s@xmath9 mg@xmath9 for the @xmath25transition and 0.230 km s@xmath9 mg@xmath9 for the @xmath26transition @xcite .
zeeman splitting of oh masers is sensitive to the strength of the _ full , three - dimensional _ magnetic field , independent of its inclination to the line of sight , when the zeeman splitting exceeds the linewidth . in w3(oh ) , we find maser lines in both the @xmath0 , @xmath5 and @xmath6 transitions .
we detect seven or eight maser line components in each circular polarization in the @xmath5 transition , compared to the three previously detected with other single dish antennas @xcite .
nearly all of these lines can be grouped into zeeman pairs with implied full magnetic field strength ranging from 6.9 to 11.3 mg .
this range agrees with previous observations by @xcite with the vlba , whose angular resolution is sufficient to unambiguously pair most zeeman components .
the strongest masers we detect have counterparts in table 1 of baudry & diamond .
it is not possible to identify unambiguous counterparts to our weaker masers , which may be blends of features resolved at the submilliarcsecond resolution of the vlba .
the detection of the lcp and rcp masers centered at @xmath8 km s@xmath9 is new .
there is some ambiguity as to whether the lcp maser at @xmath27 km s@xmath9 should be paired with the rcp maser at @xmath28 or @xmath29 km s@xmath9 , but the former is more likely because the linewidth of the latter is much greater than for the lcp maser . in the @xmath1 transition , we find one zeeman pair implying a line - of - sight magnetic field component of 10.3 mg , consistent with the detection by @xcite .
on 1 is the only other source in which we find @xmath0 masers .
the @xmath5 maser at 14 km s@xmath9 was also detected by @xcite , although their velocity resolution was insufficient to find measurable zeeman splitting .
we do not see the strong ( 0.5 jy ) maser at @xmath40.13 km s@xmath9 that they do ; however , we find evidence of a weak maser at @xmath30 km s@xmath9 ( lcp ) and @xmath31 km s@xmath9 ( rcp ) . while these lines are weak , we believe they are real .
the peaks of the lcp and rcp lines are only @xmath32 and @xmath33 times the single - channel rms noise of 0.02 k , respectively , but the line widths are 8 times a single channel width . these masers can be interpreted either as @xmath34 lines at 0.3 km s@xmath9 or as @xmath17 lines at 14.9 km s@xmath9 .
the 1665 mhz ( @xmath35 ) masers in on 1 are grouped in two velocity ranges : 1 to 4 km s@xmath9 and 10 to 16 km s@xmath9 @xcite ; likewise , the 6035 mhz ( @xmath36 ) masers are seen from @xmath37 to 2 km s@xmath9 and 13 to 16 km s@xmath9 @xcite .
thus , either the @xmath38 or @xmath17 interpretation of these maser lines is consistent with the ground - state oh maser velocities .
nevertheless , two arguments suggest that the weak maser feature in on 1 is a maser in the @xmath5 transition .
first , satellite - line emission in interstellar oh masers is generally weak compared to main - line masers .
this is true in the @xmath35 lines , where the 1665 and 1667 mhz masers are typically much stronger than 1612 and 1720 mhz masers . in the @xmath36 lines ,
strong masers are found in the 6035 and 6030 mhz main lines , but 6049 mhz satellite - line emission is weak and rare @xcite , while the 6016 mhz satellite line has not been seen in emission @xcite . indeed ,
if photon trapping is important , inversion in the 6016 mhz satellite line may be impossible @xcite .
as for the @xmath0 lines , while about a dozen masers have been found in the 13441 mhz main line transition , the 13434 mhz maser in w3(oh ) remains the single known maser in the other main line transition , despite two searches encompassing 77 distinct interstellar masing sources @xcite .
empirically , this suggests that satellite - line @xmath0 masers should be extremely rare if they exist at all .
second , due to a smaller zeeman coefficient , the magnetic field strength implied by a constant - velocity zeeman splitting is higher for satellite lines than for main lines . if the weak maser in on 1 is actually in the @xmath39 transition , the implied magnetic field is @xmath40 mg .
this field strength would be much larger than that derived from any other zeeman pair in on 1 in the @xmath41 ladder ( see also * ? ? ?
* ; * ? ? ?
lllllllll w3(oh ) & @xmath26 & @xmath442.53 & 0.15 & 0.32 & @xmath442.29 & 0.15 & 0.29 & 10.3 + & @xmath25 & @xmath444.94 & 1.93 & 0.18 & @xmath444.75 & 1.82 & 0.20 & 10.5 + & & @xmath443.97 & 4.01 & 0.38 & @xmath443.84 & 0.56 & 0.22 & 7.0 + & & @xmath443.66 & 9.77 & 0.24 & @xmath443.54 & 7.02 & 0.19 & 6.9 + & & & & & @xmath443.51 & 6.22 & 0.70 & a + & & @xmath443.33 & 12.31 & 0.15 & @xmath443.19 & 12.36 & 0.17 & 8.2 + & & @xmath443.08 & 62.65 & 0.30 & @xmath442.93 & 63.27 & 0.31 & 8.5 + & & @xmath442.46 & 64.50 & 0.27 & @xmath442.26 & 63.35 & 0.28 & 11.3 + on 1 & @xmath25&14.09&1.11 & 0.20&14.02&1.14 & 0.21 & @xmath43.8 + & & 0.38b&0.07 & 0.26&0.24b & 0.06 & 0.30 & @xmath48.3b + amagnetic field strength is 8.5 mg if paired with lcp maser at @xmath27 km s@xmath9 .
see [ masers ] for details .
btable velocities and magnetic field assume that these are @xmath5 masers .
see [ masers ] for details .
unlike maser emission , absorption is clearly seen in all sources in our sample in both the @xmath1 and @xmath5 transitions . in some sources
, absorption is seen in the @xmath17 and/or @xmath16 transitions as well .
we believe that this is the first clear detection of absorption in the @xmath0 satellite lines .
parameters of the absorption lines are given in table [ absorption - table ] .
we summarize results for individual sources below .
w3(oh ) : absorption is seen in all four hyperfine transitions in w3(oh ) .
the presence of multiple strong maser lines ( see figures [ w3ohlr1 ] and [ w3ohi1 ] ) in the @xmath5 transition is problematic for fitting gaussian components to the absorption .
the data in the velocity range from @xmath42 to @xmath43 km s@xmath9 have been excluded from the fitting of the two gaussian components listed in table [ absorption - table ] .
[email protected] : absorption is seen in the @xmath17 transition as well as the main lines .
[email protected] : absorption is seen in the main lines only .
the noise is high due to the short time spent on source .
w49 : very broad absorption is seen in the main lines .
this absorption is broader than the frequency difference between adjacent main- and satellite - line transitions .
it is possible that the absorption marked at @xmath44 km s@xmath9 in the @xmath17 transition is actually at @xmath45 km s@xmath9 in the @xmath5 transition .
however , the velocity assuming that this component is due to the @xmath17 transition is more consistent both with the other two transitions and with the h92@xmath46/he92@xmath46 velocity of @xmath47 km s@xmath9 for the associated region @xcite .
k3@xmath450 : absorption is seen in all four hyperfine transitions . on 1
: absorption is seen in the main lines only . while the observations occurred during a period devoid of precipitation ,
the noise is still high due to the short time spent on source .
the magnetic field can be measured by zeeman splitting in absorption , as well as in maser emission .
while the full three - dimensional magnetic field strength is obtained from zeeman splitting of velocity - separated maser components , absorption lines are much broader than the velocity separation between lcp and rcp components , so only the line - of - sight component of the magnetic field can be measured ( e.g. , * ? ? ?
the stokes v curve is related to the derivative of the stokes i curve by the following equation : @xmath48 where @xmath49 and @xmath50 are the brightness temperatures of the stokes v and i spectra , and @xmath51 hz g@xmath9 for the @xmath1 transition and @xmath52 hz g@xmath9 for the @xmath7 transition @xcite . hence , if the stokes v spectrum of an absorption component is nonzero , the line - of - sight magnetic field can be measured .
w3(oh ) is the only source in which the stokes v absorption spectrum shows clear evidence of zeeman splitting .
figure [ w3ohv2 ] shows the stokes v spectrum of the @xmath1 transition of w3(oh ) .
the feature near @xmath53 km s@xmath9 is due to the maser . in the direction of lower velocity from this feature
is an `` s - curve . ''
superposed atop the data are two curves corresponding to the derivative of the gaussian fits to the stokes i curve .
the curve labelled `` main '' indicates the scaled derivative of the gaussian centered at @xmath54 km s@xmath9 ( listed in table [ absorption - table ] ) , while the curve labelled `` both '' indicates the scaled derivative of both this gaussian and the one centered at @xmath55 km s@xmath9 .
a positive magnetic field shifts rcp to higher velocity than lcp ; in _ absorption _ , this corresponds to the positive bump of the stokes v curve being at higher velocity than the negative bump .
note that in _ emission _
, a positive magnetic field results in the positive bump of the stokes v curve being at _ lower velocity _ than the negative bump , as is seen for the 42.4 km s@xmath9 maser zeeman pair in figure [ w3ohv2 ] . thus the absorption and emission magnetic fields in w3(oh )
are consistent in sign even though the s - curves appear inverted with respect to each other .
the magnetic field value that provides the best fit to the data is @xmath56 mg for the main feature and @xmath57 mg for both features combined .
this is consistent with the values of @xmath58 mg and @xmath59 mg obtained by @xcite . for the other sources ,
no s - curve is observed in the stokes v spectra , but we can place upper limits on the possible magnetic field strength . from equation [ deriv - equation ] , the measurement error of the line - of - sight magnetic field component in a single channel is @xmath60 in principle , each spectral channel constitutes an independent measurement of @xmath61 .
the overall error of the parallel magnetic field strength can be estimated as @xmath62 where the summation is taken over all spectral channels .
table [ zeeman - table ] gives the @xmath63 upper limits on magnetic field strengths from absorption measurements .
for comparison , the average three - dimensional magnetic field strengths obtained from oh maser zeeman splitting are also provided .
masers from each of the @xmath64 , @xmath65 , and @xmath66 sets of transitions are considered separately . when zeeman splitting is detected in more than one set of transitions ,
the minimum and maximum average magnetic strengths obtained from oh maser zeeman splitting in sets of transitions corresponding to different values of @xmath67 are quoted .
the actual range of magnetic field strengths seen in maser zeeman splitting may be greater that that listed in table [ zeeman - table ] , but averages seem appropriate for comparison with our absorption data , since the gbt beamwidth is much greater than the angular extent of any of our sources . with the exception of the zeeman pair centered at @xmath8 km s@xmath9 , zeeman pairs in the @xmath5 transition detected in this work
are excluded from the calculation of the average magnetic field strength in w3(oh ) obtained from @xmath68 masers , in deference to the much higher angular resolution obtained by @xcite .
note that the magnetic fields derived from the zeeman pairs we detect agree with those obtained by baudry & diamond in sign and magnitude to better than 1 mg .
in addition to obtaining a positive field measurement for w3(oh ) , we are able to place upper limits of several milligauss on @xmath3 for two other sources : k3@xmath450 and [email protected] . in k3@xmath450
the magntitude of the line - of - sight component deduced from the zeeman splitting of the main absorption feature is less than 2.9 mg . measurements of oh masers give three - dimensional magnetic field strengths of @xmath69 to @xmath70 mg in the @xmath35 lines @xcite and @xmath71 to @xmath72 mg in the 6035 mhz @xmath73 line @xcite .
our @xmath63 limit of 1.2 mg for [email protected] improves by a factor of 2 on previous observations in the oh @xmath0 @xmath1 line and is comparable to results obtained from observations in so absorption @xcite . in [email protected] ,
vlba observations of the 1667 mhz @xmath74 line show one zeeman pair with a magnetic field of @xmath75 mg @xcite .
four possibilities , alone or in combination , can explain the non - detection of a zeeman pattern in the absorption in k3@xmath450 and [email protected] .
( 1 ) a reversal of the line - of - sight direction of the magnetic field exists across each source .
( 2 ) the magnetic field in k3@xmath450 and [email protected] is inclined at a large angle relative to the line of sight .
( 3 ) the average magnetic field strength of the material sampled by @xmath0 oh is smaller than that sampled by @xmath73 or @xmath74 oh .
( 4 ) masers are on average denser than the surrounding oh material , and thus the magnetic field strength is greater at masing sites than for the oh as a whole .
the first explanation is possible for both these sources .
while five @xmath35 maser zeeman pairs were found in vlba imaging of k3@xmath450 , they are all located on the periphery of the region to the north and east .
it is possible that the line - of - sight direction of the magnetic field reverses on the south or west side of the source . even if there is no reversal in k3@xmath450 , four of the five zeeman pairs found in the vlba images of the 1665 and 1667 mhz maser emission imply full ( unprojected ) magnetic field strengths less than the detectability limits in table [ zeeman - table ] @xcite .
only one ground - state oh maser has been found in vlba imaging of [email protected] .
it would be a @xmath76 and @xmath77 detection for the @xmath1 and @xmath7 transitions , respectively , if the average line - of - sight magnetic field component in the region of the @xmath78 absorption were the same as the full three - dimensional magnetic field strength implied by zeeman splitting of the 1667 mhz maser .
but with only one zeeman pair detected through synthesis imaging , it is impossible to conclude whether or not there exists a line - of - sight field reversal across the source .
the second explanation simply states that if the magnetic field is inclined at a large angle to the line of sight , its projection along the line of sight may be insufficient to produce a detectable zeeman splitting in stokes v , for the case where the splitting is less than the linewidth . assuming that the magnetic field strength of 6.0 mg for the single zeeman pair detected with vlbi resolution in [email protected] is a typical value for the source as a whole , the required inclination angle of the magnetic field to produce a line - of - sight component less than 1.2 mg is @xmath79 .
if the magnetic field actually is inclined @xmath80 to the line of sight , the expected linear polarization fraction of the two @xmath81-components of the zeeman pair is @xmath82 @xcite . but
no linear polarization is detected in the maser @xcite , ruling out a high inclination at the maser site , on the east side of the region .
linear polarization fractions consistent with inclination angles as large as @xmath83 are seen on the west side of the region , although most masers imply inclination angles much smaller than this value . incomplete spatial coverage of oh masers across the source render it difficult to estimate with certainty the inclination angle averaged across the source , but it is almost certainly less than @xmath80 . thus , while @xmath84 mg in [email protected] , the inclination angle is likely not large enough such that @xmath85 mg . still , an inclination angle smaller than @xmath80 in combination with another factor ( such as a field reversal across the source ) may suffice to reduce the average line - of - sight magnetic field strength below our detection threshold .
the third explanation is unlikely because a higher temperature is required to populate the @xmath0 states than for either the @xmath86 or @xmath73 states .
these higher temperatures likely require that the distribution of oh in the @xmath68 states be peaked closer to the central heat source than for the @xmath74 and @xmath73 states , as is noted for masers in w3(oh ) @xcite .
since the density probably increases with decreasing radius , the density ( and therefore average field strength ) of the material sampled by oh in the @xmath68 states should be higher than that of the lower states .
there is no evidence that the magnetic field strengths derived from zeeman splitting in @xmath68 masers @xcite are weaker than those derived from masers in the @xmath73 @xcite or @xmath74 transitions ( e.g. , * ? ? ? * ) . indeed ,
if w3(oh ) is not atypical , there is reason to believe that field strengths deduced from oh zeeman splitting actually increase as measured by higher excitation states in the @xmath41 ladder @xcite .
the fourth explanation differs from the third in that it suggests that masers may inherently occur at density enhancements in the surrounding medium .
this is plausible given the conditions of formation of interstellar oh masers .
observed maser strengths require some combination of oh enrichment and density enhancement @xcite .
oh masers are believed to form in the zone between the ionization and shock fronts @xcite , where instabilities lead to inhomogeneous density enhancement ( e.g. , simulations of * ? ? ?
the density conditions under which oh masers form have implications which affect the physical interpretation of their phenomenology .
if masers are formed preferentially at density enhancements , magnetic field measurements obtained at oh maser sites should be higher than the average magnetic field strength in the surrounding region .
it would also be strong evidence that observed proper motions of oh masers ( as in * ? ? ?
* ) are due to discrete material motions , not shifting coherence paths that could be unrepresentative of the motion of the material .
it is difficult to tell conclusively whether or not the material sampled by @xmath68 absorption is at a lower density than the @xmath41 masers in most of our sources . in k3@xmath450 ,
the upper limit on the magnetic field strength in the absorption falls within the range obtained from masers . in [email protected] ,
the inclination of the magnetic field to the line of sight alone is unlikely to explain the discrepancy between our upper limit and maser field strength values .
but it is quite possible that a field direction reversal occurs across the source , which would reduce the effective average line - of - sight strength detectable through absorption .
it is also not possible to rule out that there exists a field direction reversal _ along the line of sight_. however , magnetic field directions deduced from oh maser zeeman splitting in massive star - forming regions show an overwhelming tendency to fall into one of two categories : ( 1 ) a constant line - of - sight direction throughout the source , and ( 2 ) a single reversal across the source with the property that a line can be drawn dividing the source into a region of positive magnetic field and a region of negative magnetic field @xcite . a reversal of the field direction _ along the line of sight _ would produce projected regions of mixed magnetic field direction , suggesting that they are not common in massive star - forming regions . as for w3(oh ) , the magnetic field measurement obtained from absorption @xmath87 mg is less than those obtained from masers projected atop the region in the @xmath88 transitions ( average @xmath89 standard error of the mean : @xmath90 mg , from * ? ? ?
* ) , @xmath73 transitions ( @xmath91 mg , from * ? ? ?
* ) , as well as the @xmath68 transitions ( @xmath92 mg , from this work and * ? ? ?
the average magnetic field measurements from the masers projected atop the region exceed the absorption magnetic field measurement by @xmath93 , @xmath94 , and @xmath95 for the @xmath74 , @xmath65 , and @xmath66 transitions , respectively .
most masers in w3(oh ) have no detectable linear polarization @xcite , suggesting that the magnetic field in w3(oh ) is oriented close to the line of sight ( i.e. , @xmath96 ) and therefore that the average magnetic field sampled by @xmath68 absorption is less than that sampled by maser emission .
the average magnetic field strength obtained from the @xmath68 masers is greater than those obtained from @xmath74 and @xmath65 masers .
as previously noted , the @xmath68 masers in w3(oh ) are much more tightly distributed near the center of the source than in either of the other masing @xmath41 transitions . yet
the magnetic field strength measured from @xmath68 absorption is significantly less than the value obtained from @xmath68 masers . taken together , these results suggest that @xmath68 masers occur only in the highest - density portions of the oh cloud , even though oh likely exists in the @xmath68 excited state throughout the source
. two possible scenarios could explain the prevalence of high - density material necessary for excited - state oh maser activity
. there may be a large region of higher density coincident with the distribution of @xmath68 and methanol masers @xcite but with little small - scale clumping .
alternatively , small - scale ( @xmath97 cm ) clumping may occur throughout w3(oh ) but with increased prevalence in the region were @xmath68
oh masers are observed .
our data do not directly distinguish between these two scenarios .
our observations of zeeman splitting in excited - state oh absorption allow us to measure the magnetic field ( and therefore density ) averaged over the entire source , but not the length scale on which density fluctuations occur .
nevertheless , other evidence suggests that small - scale density variations are responsible for excited - state oh maser activity .
the distribution of ground - state oh masers in w3(oh ) @xcite and other massive star - forming regions @xcite show clustering on a scale of @xmath98 cm .
this is unlikely to be caused by kolmogorov turbulence , which is a scale - free process .
also , ammonia observations by @xcite show that while the optical depth of nh@xmath12 is fairly constant across w3(oh ) , the beam - filling factor decreases from west to east .
this suggests that the number of clumps is decreasing , not the density of any one clump .
lllll w3(oh ) & @xmath26 & @xmath447.22 & @xmath40.177 & 3.75 + & & @xmath445.03 & @xmath40.406 & 2.40 + & @xmath99 & @xmath446.03 & @xmath40.039 & 2.84 + & @xmath100 & @xmath444.71 & @xmath40.013 & 3.40 + & @xmath25 & @xmath448.16 & @xmath40.156 & 2.70 + & & @xmath444.99 & @xmath40.620 & 3.25 + [email protected] & @xmath26 & @xmath42.54 & @xmath40.330 & 5.57 + & & 1.02 & @xmath40.114 & 3.80 + & @xmath99 & @xmath42.36 & @xmath40.030 & 6.64 + & @xmath25 & @xmath41.86 & @xmath40.536 & 6.13 + & & 0.99 & @xmath40.047 & 1.89 + [email protected] & @xmath26 & 92.45 & @xmath40.027 & 6.12 + & @xmath25 & 93.82 & @xmath40.038 & 8.64 + w49 & @xmath26&4.34 & @xmath40.152 & 4.11 + & & 6.22 & @xmath40.219 & 21.69 + & @xmath99&5.59 & @xmath40.107 & 23.07 + & @xmath25&4.27 & @xmath40.232 & 5.07 + & & 8.23 & @xmath40.347 & 19.87 + k3@xmath450 & @xmath26 & @xmath425.07 & @xmath40.216 & 5.98 + & & @xmath420.10 & @xmath40.040 & 2.90 + & @xmath99 & @xmath425.50 & @xmath40.018 & 7.11 + & @xmath100 & @xmath425.08 & @xmath40.010 & 3.19 + & @xmath25 & @xmath425.02 & @xmath40.276 & 6.34 + & & @xmath420.24 & @xmath40.039 & 2.14 + on 1 & @xmath26 & 10.88 & @xmath40.014 & 7.13 + & @xmath25 & 10.74 & @xmath40.016 & 7.34 lllll w3(oh ) & @xmath57 & @xmath101b & @xmath102 & 1,2,3,4 + [email protected] & @xmath103 & @xmath104 & @xmath75 & 5 + [email protected] & @xmath105 & @xmath106 & @xmath107 & 6 + w49 & @xmath108 & @xmath109 & @xmath110 & 7 + k3@xmath450 & @xmath111 & @xmath112 & @xmath113 & 5,8 + on 1 & @xmath114 & @xmath115 & @xmath116 & 1,5,8 aaverage magnetic field strengths deduced from zeeman splitting in one or more @xmath41 oh transitions are presented . note that magnetic field strengths determined from zeeman splitting of oh masers are _ full _ three - dimensional values , while those obtained from absorption are sensitive to only the _ line - of - sight projection _ of the full magnetic field . see
[ zeeman ] for more details .
bcontamination from strong masers and uncertainty in the strength and location of the absorption features precludes obtaining a lower estimate of the zeeman splitting . in local thermodynamic equilibrium ,
the relative strengths of the @xmath117 and @xmath118 absorption lines are @xmath119 .
the ratio of the @xmath5 to @xmath6 main lines ranges from 0.41 ( [email protected] , 1 km s@xmath9 ) or 0.97 to 1.62 in our sources , with 1.30 being the lte value for optically thin lines .
the satellite lines , when detected in absorption , are always enhanced relative to the expected lte value for the @xmath34 line and almost always enhanced relative to the @xmath1 transition , the single exception being the @xmath16 line in w3(oh ) .
@xcite surmise that the excitation temperatures in the hyperfine lines are not equal due to line overlaps in the 84 @xmath120 m lines connecting the @xmath36 and @xmath0 states of oh .
@xcite note that collisions will preferentially excite the @xmath0 negative - parity state , but that the 84.42 @xmath120 m line that excites the positive - parity state may be in resonance with co emission at 84.41 @xmath120 m .
matthews et al .
point out that far infrared line overlaps will equalize populations between the @xmath121 and @xmath122 states , but the @xmath123 state will depopulate relative to the @xmath124 state .
hence , the @xmath5 and @xmath118 absorption lines will have a lower excitation temperature than the @xmath1 and @xmath125 lines .
for the two cases in which both satellite lines are detected , the @xmath39 line is stronger than the @xmath16 line .
additionally , the @xmath118 line is detected in two sources in which the @xmath125 line is not detected .
this is consistent with a lower excitation temperature for the @xmath118 state than for the @xmath125 state .
however , we do not find evidence that the @xmath5 lines are enhanced relative to the @xmath6 lines , as predicted by the matthews et al . model .
we have detected @xmath0 oh absorption toward six massive star - forming regions .
main - line absorption was detected in both the @xmath1 and @xmath7 lines toward all sources .
additionally , we detected at least one satellite line in absorption toward four of the six sources .
we believe that this is the first detection of the @xmath0 satellite lines in interstellar sources .
stokes v spectra of the main lines were produced for these six sources . in the case of w3(oh ) , zeeman splitting of the @xmath1 absorption results in a magnetic field measurement of @xmath126 mg , consistent in magnitude and sign with the @xmath127 mg obtained by @xcite .
w3(oh ) is the only source in which a positive detection of zeeman splitting in the @xmath68 absorption has been obtained .
the component of the magnetic field along the line of sight is comparable to the full magnetic field strength measured in @xmath74 and @xmath73 masers , suggesting that these masers do not preferentially form in high density , low filling factor regions where the density significantly exceeds that of the surrounding , non - masing oh . however , the line - of - sight component of the magnetic field determined from @xmath68 absorption is much smaller than that determined from @xmath68 masers .
the grand question is whether the small line - of - sight magnetic field strength measured in @xmath0 absorption necessarily implies density enhancement at maser sites . if @xmath68 absorption occurs only where the @xmath68 masers occur , then the full magnetic field strength ( as deduced from the @xmath68 masers ) in this region ranges from 5.6 to 11.3 mg @xcite , consistent with the range of field strengths determined from @xmath74 masers at this site at the northern limb of the region @xcite .
restricting consideration to the @xmath1 transition , the transition in which a positive result for zeeman splitting is obtained in absorption , favors the upper end of this range of field strengths , since the only detected maser zeeman pair implies a full magnetic field strength of 10.3 mg ( table [ maser - table ] ) .
the magnetic field in this region does not appear to be significantly inclined to the line of sight @xcite , suggesting that the line - of - sight magnetic field strength is comparable to the full magnetic field strength .
since the average magnetic field strength derived from masers in this region is 9.1 mg , three times the field strength measured in absorption , it might be concluded that masers are denser than the surrounding , non - masing region by a factor of 9 , under the reasonable assumption that @xmath128 .
however , it is likely that @xmath68 absorption comes from other areas in front of the region as well .
@xcite identify three other regions of @xmath35 masers projected atop the region in w3(oh ) : a central clump with magnetic fields of 6.2 and 7.1 mg , a southern clump with fields of 2.3 to 6.0 mg , and a western clump with a magnetic field of 1.8 mg . observations of the @xmath129 transitions , with an excitation temperature of 511 k above ground , show strong absorption at the northern clump as well as weaker absorption at the central and southern clumps @xcite . in the @xmath130 ( 270 k above ground ) transitions ,
significant absorption is seen at and between all four @xmath35maser clumps , while in the @xmath131 ( 415 k above ground ) transitions , absorption is seen mainly along a line running through the northern , central , and southern clumps , with weaker absorption from the western clump @xcite .
there is no published map of @xmath0 ( 290 k above ground ) absorption in w3(oh ) , but the distribution of oh in other comparable excited states suggests that strong @xmath0 absorption would be seen over the majority of the western half of the region , including the sites of all four @xmath35 maser clumps .
this would imply that the density at @xmath35 maser sites is about 2 to 4 times that of the non - masing oh . on the other hand , 7820 mhz ( @xmath132 ) and 8190 mhz ( @xmath133 ) _ emission _ is seen exclusively near the northern maser clump @xcite , and 6031 mhz ( @xmath36 , @xmath134 ) and 6035 mhz ( @xmath135 ) _ maser emission _ is strongest near the northern maser clump @xcite . in total , this suggests that while @xmath0 absorption occurs over most of the western half of w3(oh ) , the strongest contribution to the absorption most likely comes from the northern clump , which has the highest average magnetic field strength .
the unweighted average magnetic field determined from oh masers in all transitions atop the region is 6.9 mg .
due to the distribution of maser spots , this effectively gives highest weight to the northern clump and lowest weight to the eastern clump , consistent with our arguments based on the distribution of excited - state @xmath136 and @xmath41 oh absorption and emission .
thus , the density at maser sites is likely several ( @xmath137 ) times that of the non - masing regions in w3(oh ) .
more precise quantitative results would require an interferometric map of @xmath0 absorption .
we are able to place upper limits of several milligauss on the line - of - sight component of the magnetic field in two other sources .
the @xmath138 mg limit on k3@xmath450 is about 50% of the full three - dimensional magnetic field strengths obtained from zeeman splitting of the @xmath86 @xcite and @xmath73 masers @xcite .
the @xmath139 mg limit on [email protected] is a factor of 5 smaller than the @xmath140 mg obtained from a @xmath74 maser zeeman pair @xcite .
it is not clear if this occurs because the masers are in higher density ( and hence higher magnetic field ) clumps or if other effects , such as field reversals or large angles of the magnetic field to the line of sight , are present in portions of the source .
we have detected @xmath17 and/or @xmath125 satellite line absorption in four sources .
of these two , absorption in the @xmath118 line is always stronger , as predicted by @xcite .
satellite line absorption appears enhanced relative to the main lines , from that which would be expected in local thermodynamic equilibrium .
maser emission was also observed toward two sources . in w3(oh ) we find seven pairs of lcp and rcp masers implying magnetic field strengths from @xmath141 to @xmath142 mg .
we detect line components at all velocities where strong maser features were previously observed , although we do not have the spatial resolution to separate the multiple lines detected by @xcite using the vlba .
in addition , we find a pair of previously undetected masers centered at @xmath8 km s@xmath9 . in on 1
, we find the maser at @xmath143 km s@xmath9previously detected by @xcite .
we do not see the strong maser at @xmath144 km s@xmath9 that they did , but we find a new weak maser centered at @xmath145 km s@xmath9 .
maser strength variability , previously noted in the @xmath0 masers of w3(oh ) ( e.g. , * ? ? ?
* ) , appears to be operating in on 1 as well .
support for this work was provided by the nsf through award gssp04 - 0001 from the nrao .
we thank t. minter for help in setting up the observations , j. braatz and g. langston for assistance in data reduction , and m. modjaz for helpful comments in the interpretation of the data .
we also thank an anonymous referee for helpful comments .
km s@xmath9 in the @xmath5 transition of on 1 .
the lcp and rcp data are plotted as histograms , and the best gaussian fits are plotted as curves .
the rcp data have been shifted by 0.2 k for clarity.[on1lr1a],width=384 ] and @xmath118 transtitions of w3(oh ) .
data points are hanning weighted and plotted as a histrogram .
the gaussian fit parameters listed in table [ absorption - table ] are plotted as a curve .
the masers are plotted in figure [ w3ohlr1 ] with a much larger scale to show the strong emission .
see [ absorption ] for details regarding the fits to the main - line absorption features .
the @xmath17 line is shown in greater detail in figure [ w3ohi1a ] .
[ w3ohi1],width=384 ] and @xmath125 transitions of w3(oh ) .
see figure [ w3ohi1 ] caption for more details .
the maser at @xmath146 km s@xmath9 is plotted in figure [ w3ohlr2 ] .
the @xmath16 line is shown in greater detail in figure [ w3ohi2a].[w3ohi2],width=384 ] transition of [email protected] .
see figure [ w3ohi1 ] caption for more details .
the arrow indicates where @xmath16 absorption would appear if at the velocity of the main absorption component.[g10i2],width=384 ] transition of w3(oh ) .
the solid line shows the best fit for the derivative of stokes i for the main absorption feature at @xmath148 km s@xmath9 , and the dashed line shows the best fit for both @xmath1 absorption features .
the fit lines correspond to a magnetic field of @xmath183.0 mg , as explained in
[ zeeman ] .
the large value of stokes v near @xmath53 km s@xmath9 is due to a pair of maser features , which have been excluded from the fit.[w3ohv2],width=384 ] | we have observed six high - mass star - forming regions in the @xmath0 lines of oh using the gbt in order to investigate whether the magnetic field , and hence the density , measured in absorption differs from that implied by maser zeeman splitting .
we detect absorption in both the 13441 and 13434 mhz main lines in all six sources .
zeeman splitting in the @xmath1 absorption line in w3(oh ) implies a line - of - sight magnetic field strength of @xmath2 mg .
this is significantly less than full magnetic field strengths detected from oh maser zeeman splitting , suggesting that oh maser regions may be denser than the non - masing oh material by a factor of several .
zeeman splitting is not detected in other sources , but we are able to place upper limits on @xmath3 of 1.2 mg in [email protected] and 2.9 mg in k3@xmath450 .
these results are consistent with a density enhancement of the masers , but other explanations for the lower magnetic field in absorption compared to maser emission are possible for these two sources .
absorption in one or both of the 13442 and 13433 mhz satellite lines is also seen in four sources .
this is the very first detection of the @xmath0 satellite lines
. ratios of satellite - line to main - line absorption suggest enhancement of the satellite lines from local thermodynamic equilibrium values .
masers are seen in the @xmath5 and @xmath6 transitions of w3(oh ) and the @xmath7 transition of on 1 .
a previously undetected @xmath7 maser is seen near @xmath8 km s@xmath9 in w3(oh ) . |
in the last years , magnetized quark matter has attracted the attention of the physics community due to its relevance for heavy ion collisions at very high energies @xcite , to the understanding of the first phases of the universe @xcite and for studies involving compact objects like magnetars @xcite . in the presence of an external magnetic field ,
the behavior of quark matter is determined by the competition between two different mechanisms : the enhancement of the quark condensate because of the opening of the gap between the landau states leading to the increase of low - energy contributions to the formation of the chiral condensate ; and the suppression of the quark condensate due to the partial restoration of chiral symmetry .
it was shown that in the region of low momenta relevant for chiral symmetry breaking there is a strong screening effect of the gluon interactions , which suppresses the condensate @xcite . in this region ,
the gluons acquire a mass @xmath0 of the order of @xmath1 , due to the coupling of the gluon field to a quark - antiquark interacting state . in the presence of a strong enough magnetic field ,
this mass @xmath0 for gluons becomes larger .
this , along with the property that the strong coupling @xmath2 decreases with increasing @xmath3 , @xmath4 with @xmath5 , @xcite leads to an effective weakening of the interaction between the quarks in the presence of an external magnetic field , and damps the chiral condensate .
the suppression of the quark condensate , also known as inverse magnetic catalysis ( imc ) , manifests itself on the decrease of the pseudocritical chiral transition temperature obtained in lattice qcd ( lqcd ) calculations with physical quark masses and an increase of the polyakov field @xcite .
recent results in two - flavor lqcd with dynamical overlap fermions in an external magnetic field also support the imc scenario @xcite .
in particular , in @xcite it is argued that the imc may be a consequence of how the gluonic sector reacts to the presence of a magnetic field , and , it is shown that the magnetic field drives up the expectation value of the polyakov field .
the distribution of gluon fields changes as a consequence of the distortion of the quark loops in the magnetic field background .
therefore , the backreaction of the quarks on the gauge fields should be incorporated in effective models in order to describe the imc .
almost all low - energy effective models , at zero chemical potential , including the nambu jona - lasinio ( njl)-type models , find an enhancement of the condensate due to the magnetic field , and no reduction of the pseudocritical chiral transition temperature with the magnetic field @xcite .
however , a recent study using the polyakov nambu jona - lasinio ( pnjl ) @xcite has shown that if the lqcd data @xcite is fitted by making the pure - gauge critical temperature @xmath6 , a parameter of the pnjl model , @xmath3 dependent , the model is able to describe an imc . several recent studies @xcite discuss the origin of the imc phenomenon . in particular , the magnetic inhibition can be a possible explanation for the decreasing behavior of the chiral restoration temperature with increasing @xmath3 @xcite ; another mechanism to explain the imc around the critical temperature as induced by sphalerons was proposed in @xcite .
the discussion above points out that the gluon distribution reacts to the magnetic background and suggests that the effective interaction between the quarks should get this dependence . with this motivation , we adopt a running coupling constant of the chiral invariant quartic quark interaction in njl and pnjl models with the magnetic field .
the damping of the strength of the effective quartic interaction is built phenomenologically , keeping su(3 ) flavor symmetry , under different assumptions inspired by lattice results for the quark condensate at finite temperature and magnetic field .
this paper is organized as follows . in sec .
ii , we briefly present the pnjl model used in this work , the polyakov loop potential , and the parametrizations chosen . in sec .
iii , the importance of the running coupling in the ( p)njl models for magnetized quark matter is discussed .
also , the behavior of the condensates with temperature and the magnetic field intensity is compared with the lqcd results . finally , in sec .
iv , the main conclusions are drawn .
the pnjl lagrangian with explicit chiral symmetry breaking , where the quarks couple to a ( spatially constant ) temporal background gauge field , represented in terms of the polyakov loop , and in the presence of an external magnetic field is given by @xcite @xmath7 q ~+~ { \cal l}_{sym}~+~{\cal l}_{det } \nonumber\\ & + & \mathcal{u}\left(\phi,\bar\phi;t\right ) - \frac{1}{4}f_{\mu \nu}f^{\mu \nu } , \label{pnjl}\end{aligned}\ ] ] where the quark sector is described by the su(3 ) version of the njl model which includes scalar - pseudoscalar and the t hooft six fermion interactions that models the axial @xmath8 symmetry breaking @xcite , with @xmath9 and @xmath10 given by @xcite , @xmath11 , \end{aligned}\ ] ] @xmath12 + { \rm det}\left [ { \bar q}(1-\gamma_5)q\right ] \right \}\end{aligned}\ ] ] where @xmath13 represents a quark field with three flavors , @xmath14 is the corresponding ( current ) mass matrix , @xmath15 where @xmath16 is the unit matrix in the three - flavor space , and @xmath17 denote the gell - mann matrices .
the coupling between the ( electro)magnetic field @xmath18 and quarks , and between the effective gluon field and quarks is implemented _ via _ the covariant derivative @xmath19 where @xmath20 represents the quark electric charge ( @xmath21 ) , @xmath22 and @xmath23 are used to account for the external magnetic field and @xmath24 where @xmath25 is the su@xmath26 gauge field .
we consider a static and constant magnetic field in the @xmath27 direction , @xmath28 . in the polyakov gauge and at finite temperature the spatial components of the gluon field are neglected : @xmath29 .
the trace of the polyakov line defined by @xmath30 is the polyakov loop which is the order parameter of the @xmath31 symmetric / broken phase transition in pure gauge . to describe the pure - gauge sector
an effective potential @xmath32 is chosen in order to reproduce the results obtained in lattice calculations @xcite , @xmath33 , \label{ueff}\end{aligned}\ ] ] where @xmath34 , @xmath35 .
the standard choice of the parameters for the effective potential @xmath36 is @xmath37 , @xmath38 , @xmath39 , and @xmath40 .
the value of @xmath41 is fixed in order to reproduce lqcd results ( @xmath42 170 mev @xcite ) , we use as a regularization scheme , a sharp cutoff , @xmath43 , in three - momentum space , only for the divergent ultraviolet sea quark integrals .
the parameters of the model , @xmath43 , the coupling constants @xmath44 and @xmath45 , and the current quark masses @xmath46 and @xmath47 are determined by fitting @xmath48 , @xmath49 , @xmath50 and @xmath51 to their empirical values .
we consider @xmath52 , mev , @xmath53 , mev , @xmath54 mev , @xmath55 and @xmath56 as in @xcite .
the thermodynamical potential for the three - flavor quark sector @xmath57 is written as @xmath58 where the flavor contributions from vacuum @xmath59 , medium @xmath60 , and magnetic field @xmath61 @xcite are given by @xmath62 } } \end{aligned}\ ] ] where @xmath63 , @xmath64 and @xmath65 , @xmath66 , and @xmath67 , where @xmath68 is the riemann - hurwitz zeta function . at zero chemical potential the quark distribution functions @xmath69 and @xmath70 read @xmath71 once @xmath72 .
as already referred , the presence of an external magnetic field has two competing mechanisms : on one hand it enhances the chiral condensate due to the increase of low - energy contributions ; on the other hand there is a suppression of the condensate because in the region of the low momenta relevant for the chiral symmetry breaking mechanism there is a strong screening effect of the gluon interactions @xcite .
this suppression of the condensate , also known as imc , manifests itself as the decrease of the pseudocritical chiral transition temperature obtained in lqcd calculations with physical quark masses @xcite and in the increasing of the polyakov loop @xcite . within the njl and pnjl models
the inclusion of the magnetic field in the lagrangian density allows us to describe the magnetic catalysis effect , but fails to account for the imc . in the njl model
the quarks interact through local current - current couplings , assuming that the gluonic degrees of freedom can be frozen into pointlike effective interactions between quarks .
therefore , we may expect that the screening of the gluon interaction discussed above weakens the interaction and which is translated into a decrease of the scalar coupling with the intensity of the magnetic field .
mev ) as a function of @xmath3 in the njl model with a magnetic field dependent coupling @xmath73 ( blue dashed ) and a constant coupling @xmath74 ( black line ) , and the lattice results ( red dots ) @xcite . ] in @xcite , it was shown that the running coupling decreases with the magnetic field strength , @xmath75 with @xmath5 .
+ consequently , in the njl model the coupling @xmath44 , which can be seen as @xmath76 , must decrease with an increasing magnetic field strength .
a first attempt to include the impact of the running coupling in the njl model can be done by introducing the simple ansatz , to the lqcd results for the chiral transition pseudocritical temperature . ]
@xmath77 in the limit case @xmath78 , we obtain @xmath79 , and for @xmath80 , we get @xmath81 . the pseudocritical temperatures for the chiral transitions @xmath82 ( being @xmath83 and @xmath84 the transition temperatures for @xmath85 and @xmath86 quarks , respectively ) , are calculated using the location of the susceptibility peaks , @xmath87 , with @xmath88 .
the multiplication by @xmath49 is only to ensure that the susceptibilities are dimensionless .
other methods to define the temperature transitions , such as those from the magnitude of the order parameters , are equally useful , see , for instance , ref .
the calculated chiral pseudocritical temperatures are shown in fig .
[ fig : temp_crit ] : when @xmath89 the model always shows a magnetic catalyzes , with increasing @xmath90 for all range of magnetic fields ; when @xmath91 , defined by eq .
( [ geb1 ] ) , an imc is seen until @xmath92 gev@xmath93 , with the decrease of the pseudocritical temperature for low magnetic fields , but with the increase of @xmath90 for high @xmath3 values .
thus , with this simple ansatz , the model predicts an imc at low @xmath3 and magnetic catalysis at high @xmath3 .
this is in agreement with lattice results at high @xmath3 @xcite .
it is worth noting that the logarithm behavior of the running coupling of qcd @xmath94 , occurs for high momentum transfers @xmath95 1 gev . in this way
, the @xmath96 behavior , may not be suitable for the low magnetic field range , @xmath97 gev@xmath93 .
+ dependence calculated in the njl model in order to reproduce the lqcd renormalized chiral transition temperature @xcite .
] since there is no lqcd data for @xmath98 available , we will use another approach , in particular we will fit @xmath73 in order to reproduce @xmath99 obtained in lqcd calculations @xcite .
the resulting fit function of @xmath73 , that reproduces the @xmath99 ( see fig . [
fig : t_c ] ) , is plotted in fig .
[ fig : fit ] and is written as @xmath100 with @xmath101 , @xmath102 , @xmath103 , and @xmath104 and where @xmath105 .
we have used @xmath106 mev . as a function of @xmath3 , for several values of temperature in mev , in the njl model , with a magnetic field dependent coupling @xmath73 from eq .
( [ eq : fit ] ) ( top ) and a constant coupling @xmath74 ( bottom ) . ] in @xcite , the authors have shown that in the presence of an external magnetic field and in the intermediate regime , corresponding to an energy scale below the magnetic field scale but larger than the dynamical quark mass , the gluon acquires a mass @xmath107 .
thus in this limit of interest precisely at the chiral symmetry restoration transition , we have @xmath108 .
the above polynomial form insures that for @xmath78 , @xmath44 goes as @xmath109 . as a function of temperature , for several values of @xmath3 in gev@xmath93 , in the njl model , with a magnetic field dependent coupling @xmath73 from eq .
( [ eq : fit ] ) compared with lqcd results @xcite ( top ) , and a constant coupling @xmath74 ( bottom ) .
the lqcd data was renormalized by @xmath110 mev @xcite and the njl model results by @xmath111 mev . ] in the following , we focus on the order parameter for the chiral transition , and , according to @xcite , we define the change of the light quark condensate due to the magnetic field as @xmath112 with @xmath113 + 1\ ] ] where the factor @xmath114 in the denominator contains the pion mass in the vacuum ( @xmath115 mev ) and ( the chiral limit of the ) pion decay constant ( @xmath116 mev ) in njl model .
the behavior of the quark condensates with the magnetic field is shown in figs .
[ fig : chiral_cond_soma]-[fig : chiral_cond_dif ] . in fig .
[ fig : chiral_cond_soma ] , the change of the renormalized condensate @xmath117 as a function of the magnetic field intensity at several temperatures is shown for @xmath73 defined in eq .
( [ eq : fit ] ) ( top panel ) and @xmath74 ( bottom panel ) .
the average of the light condensates calculated with @xmath73 shows the same behavior as lqcd calculations : at low and high temperatures the magnetic field enhances the condensates but at temperatures near the pseudocritical chiral transition temperature , the magnetic field suppresses the condensates .
using @xmath89 , an enhancement is predicted at any temperature @xcite : the magnetic catalysis is the result of an enhancement of the spectral density at low energies which increases the number of participants in the chiral condensate . as a function of temperature , for several values of @xmath3 in gev@xmath93 , in the njl model , calculated with a magnetic field dependent coupling @xmath73 [ eq .
( [ eq : fit ] ) ] compared with lqcd results @xcite ( top ) , and a constant coupling @xmath74 ( bottom ) .
the lqcd data was renormalized by @xmath110 mev @xcite and the njl model results by @xmath111 mev . ]
if @xmath89 , and for @xmath118 , @xmath119 increases with @xmath3 due to the magnetic catalysis effect , being its value even larger as the temperature is higher @xcite . on the other hand , when @xmath120 we are in the region where the partial restoration of chiral symmetry already took place . in this region
there are two competitive effects : the partial restoration of chiral symmetry , that prevails at lower values of @xmath3 , making the condensate average approximately zero , and the magnetic catalysis , that becomes dominant as the magnetic field increases . when the strength of the interaction decreases as @xmath3 increases , the coupling of a quark - antiquark pair interaction is weakened leading to the occurrence of an earlier partial restoration of chiral symmetry , so this effect is dominant preventing the magnetic catalysis to occur .
these same conclusions are obtained from fig .
[ fig : cond_soma_t ] where the average light quark condensate is plotted as functions of @xmath121 for several values of @xmath3 .
the lattice results extracted from @xcite have also been included in the top panel together with the results obtained with @xmath91 from eq .
( [ eq : fit ] ) .
the qualitative agreement between both calculations is quite good , with the main features being reproduced by the njl model .
a very different behavior is obtained with a constant coupling @xmath74 , see fig .
[ fig : cond_soma_t ] bottom , where the transition for larger values of @xmath3 occurs for larger temperatures . as a function of @xmath3 , for several values of temperature in mev , in the njl model , with a magnetic field dependent coupling @xmath73 ,
( [ eq : fit ] ) . ] in figs .
[ fig : cond_soma_dif_t ] and [ fig : chiral_cond_dif ] the difference between light quark condensate are plotted , respectively , as a function of @xmath121 for several values of @xmath3 , and as a function of @xmath3 for several temperatures .
the lattice results from @xcite have been included in fig .
[ fig : cond_soma_dif_t ] ( top ) together with the results for @xmath91 . for comparison
we also show the results for @xmath89 ( bottom ) . in the njl model , with the magnetic field
dependent coupling @xmath73 [ eq .
( [ eq : fit ] ) ] ( blue line ) and lqcd results ( red dots ) @xcite . ]
the bumps present in curves obtained with @xmath89 around the transition temperatures , a characteristic of the njl with constant coupling , do not appear when @xmath73 is used , and a reasonable agreement with the lqcd results is achieved . as pointed out in @xcite
, these bumps are the result of a stronger magnetic catalysis effect for the @xmath85 quark , due to its larger electric charge , ( the larger the magnetic field the larger the difference between @xmath85 and @xmath86 condensates , and the respective chiral transition temperatures ) , being this feature particularly strong close to the transition temperature , where the curves for stronger fields have a larger bump .
when @xmath91 , the effect due to the partial restoration of chiral symmetry will prevail over the magnetic catalysis due to a weaker interaction and , as already pointed out , the bumps will disappear in accordance with lattice results . in fig .
[ fig : chiral_cond_dif ] the condensate difference @xmath122 is plotted as a function of @xmath3 for several temperatures and it is clearly seen that it always decreases with the temperature .
we next analyze the @xmath123 phase diagram obtained within the njl with a coupling dependent on the magnetic field .
the parametrization @xmath73 was obtained using the available lattice results for the pseudocritical temperatures in the range @xmath124 gev@xmath93 .
the calculated pseudocritical temperatures are shown in fig .
[ fig : t_c ] for a range of magnetic field intensities larger than the one used in the fit . for @xmath125 gev@xmath93 ,
the pseudocritical temperature starts to increase with @xmath3 .
this behavior was also obtained in lattice calculations @xcite , which also predict magnetic catalysis at high values of @xmath3 . for @xmath126 gev@xmath93 a chiral
first order phase transition appears .
the lqcd as well as the njl results from fig .
[ fig : cond_soma_t ] show that the average chiral condensate slope increases with increasing magnetic field .
thus , if this behavior persists for high magnetic fields , it is expected also from the lqcd results , that at some critical @xmath3 , the transition turns into a first order . in the pnjl , using the magnetic field dependent coupling @xmath73 [ eq .
( [ eq : fit ] ) ] . ] in the present section , we consider the pnjl model . in this model
the the quark degrees of freedom are coupled to a polyakov loop field which allows us to simulate a confinement - deconfinement phase transition at finite temperature .
several studies about the deconfinement and chiral symmetry restoration of hot qcd matter in an external magnetic field have recently been made @xcite .
now , we will take for the scalar coupling parameter the same magnetic field dependent parametrization obtained in the previous section [ eq .
( [ eq : fit ] ) ] .
next we discuss the effect of the magnetic field on the polyakov loop and on the quark condensates within this model . for several values of @xmath121 ( mev ) ( top ) and versus @xmath121 for several values of @xmath3 in gev@xmath93 ( bottom ) .
, title="fig : " ] + for several values of @xmath121 ( mev ) ( top ) and versus @xmath121 for several values of @xmath3 in gev@xmath93 ( bottom ) .
, title="fig : " ] some remarks are in order concerning the applicability of the pnjl model .
it should be noticed that in this model , besides the chiral point - like coupling between quarks , the gluon dynamics is reduced to a simple static background field representing the polyakov loop . as referred in sec . [
sec : model ] , we consider the parameter @xmath6 in the polyakov loop @xmath127 mev , which takes into account the quark backreaction and reproduces the lattice deconfinement pseudocritical temperature @xmath128 mev . as
shown in the following , we obtain within the pnjl model several features discussed in the previous section , e. g. the deconfinement transition and the chiral transition pseudocritical temperatures are both decreasing functions with @xmath3 until a limiting magnetic field of @xmath129 gev@xmath93 , as in lqcd , see fig . [ fig : pnjl_tc ] . due to the existing coupling between the polyakov loop field and quarks within pnjl
, the @xmath73 does not only affect the chiral transition but also the deconfinement transition .
as a function of @xmath3 , for several values of @xmath121 in mev , in the pnjl model . ]
the effect of the magnetic field on the polyakov loop is more clearly seen in fig .
[ fig : polyakov ] where @xmath130 is plotted as a function of the magnetic field intensity for different values of the temperature ( top ) , and as a function of temperature , for several magnetic field strengths ( bottom ) .
the suppression of the condensates achieved by the magnetic field dependence of the coupling parameter translates into an increase of the polyakov loop .
the effect of the magnetic field on @xmath130 is stronger precisely for the temperatures close to the transition temperature , see fig .
[ fig : polyakov ] ( top ) , in close agreement with the lqcd results @xcite .
( top ) and the difference @xmath131 ( bottom ) of the light chiral condensates as a function of temperature , for several values of @xmath3 in gev@xmath93 , and the lqcd results @xcite .
the lqcd data was renormalized by @xmath110 mev @xcite and the pnjl model results by @xmath132 mev . ] in fig . [
fig : pnjl_sum ] we plot the average chiral condensate @xmath133 as a function of @xmath3 , for several temperatures . as in the lqcd @xcite , for temperatures smaller and higher than the transition temperatures ,
the model predicts a monotonously increase with @xmath3 , and for temperatures near the transition temperature , a nonmonotonic behavior is obtained .
thus , for @xmath134 , at magnetic field intensities higher than some pseudocritical value , the condensates are suppressed by the presence of the magnetic field . in fig .
[ fig : pnjl_cond2 ] the chiral condensate sum @xmath135 and the chiral condensate difference @xmath122 are plotted as a function of the temperature , renormalized by the pseudocritical temperature at zero magnetic field , for several magnetic field strengths and compared with the lqcd results from @xcite . just as already obtained for njl , general features of the lqcd results
are reproduced .
we observe that su(3 ) symmetry of the pointlike effective interactions between quarks is assumed in the magnetic background , however the comparison with the lqcd results for the difference in the quark condensates in fig .
[ fig : pnjl_cond2 ] bottom , suggests that the up quark interaction is depleted with respect to the down quark one . that , seems reasonable
as the effect of the magnetic field on the up quark is larger than in the down quark , and therefore the interaction between the up quarks should decrease with respect to the down quarks as the magnetic field increases .
a more detailed calculation should account for magnetic su(3 ) flavor breaking and it is postponed for a future work . in ref .
@xcite the possibility of describing the imc effect within the models pnjl and entangled pnjl ( epnjl ) by including a magnetic field dependent parameter @xmath136 in the parametrization of the polyakov potential was studied . the main argument in favor was that backreaction effects on the polyakov loop due to the presence of a strong magnetic field should introduce screening effects leading to a reduction of the pseudocritical transition temperatures .
a magnetic field dependent effective polyakov potential could indeed describe the imc effect but only within epnjl .
neither the pnjl model @xcite nor the two - flavor thermal quark - meson model @xcite were able to obtain the imc effect with a @xmath6 parameter dependent on the magnetic field .
these results are in accordance with the results of the present work : in the epnjl the coupling @xmath44 depends on the polyakov loop , and , therefore , at the crossover when the polyakov loop increases the coupling @xmath44 becomes weaker .
this is shown in fig .
[ epnjl ] top panel , where the coupling @xmath137 $ ] of ref .
@xcite is plotted for several temperatures ( dashed curves ) and , for comparison , the parametrization @xmath73 given in eq .
( [ eq : fit ] ) and fig .
[ fig : fit ] is also included ( full black line ) .
it is interesting to realize that in the range @xmath138 gev@xmath93 , the curve obtained for @xmath139 mev , which is close to the deconfinement pseudocritical temperature ( @xmath140 mev ) , has a behavior in accordance with the results of the present work . within the pnjl
no imc effect was obtained because the parameter @xmath136 does not affect the coupling @xmath44 .
top : the scalar coupling @xmath44 versus the magnetic field , the black full line is the parametrization defined in ( [ eq : fit ] ) and plotted in fig .
[ fig : fit ] ; middle : the chiral and deconfinement pseudocritical temperatures versus the magnetic field ; bottom : the polyakov loop versus the temperature renormalized by the deconfinement pseudocritical temperature @xmath141 for @xmath142 , respectively , 171 mev [ pnjl with @xmath73 ] and 214 mev [ epnjl with @xmath136 ] .
, title="fig : " ] + @xcite .
top : the scalar coupling @xmath44 versus the magnetic field , the black full line is the parametrization defined in ( [ eq : fit ] ) and plotted in fig . [
fig : fit ] ; middle : the chiral and deconfinement pseudocritical temperatures versus the magnetic field ; bottom : the polyakov loop versus the temperature renormalized by the deconfinement pseudocritical temperature @xmath141 for @xmath142 , respectively , 171 mev [ pnjl with @xmath73 ] and 214 mev [ epnjl with @xmath136 ] .
, title="fig : " ] in fig .
[ epnjl ] middle and bottom panels , we compare the results obtained in the present work with the ones of @xcite for the pseudocritical temperatures and the polyakov loop . in @xcite , the pseudocritical temperatures have a much flatter behavior at small values of the magnetic field reflecting the softer decrease of the coupling @xmath44 for small values of the magnetic field shown in fig .
[ epnjl ] top panel . also , within the epnjl with @xmath136 the difference between the pseudocritical temperatures @xmath145 and @xmath141 is much smaller because the polyakov loop and the quark condensates are strongly coupled . for @xmath3=0 these temperatures are almost coincident , but a finite strong magnetic field destroys this coincidence .
pnjl does not have this feature and a @xmath73 coupling is not changing its normal behavior predicting different temperatures for @xmath145 and @xmath141 . on the other hand ,
the effect of the parametrization @xmath136 on the polyakov loop within epnjl is much stronger than the one obtained in the present work , which is an indirect effect occurring due to the dependence of the quark distributions , eq .
( [ eq : z ] ) , on the polyakov loop .
in the present work we study quark condensates and the qcd phase diagram at zero chemical potential and finite temperature subject to an external magnetic field within the njl and pnjl models .
we have shown that recent results from lqcd , for quark matter in the presence of an external magnetic field , can be reproduced within njl / pnjl models , if a magnetic field dependent coupling constant is used . a decreasing magnetic field dependent four quark coupling is essential , within effective quark models , to mimic the expected running of the coupling constant with the magnetic field strength and to incorporate the backreaction of the sea quarks in order to explain the imc .
we have calculated the @xmath73 coupling constant in the njl model , that reproduces the qualitative behavior of chiral pseudocritical temperature given by lqcd .
all the qualitative results predicted by lqcd , can be reproduced using the calculated @xmath73 coupling : ( a ) the nonmonotonic behavior of the average condensate @xmath133 as a function of @xmath3 is obtained in the temperature region of the chiral transition ; ( b ) the polyakov loop increases with @xmath3 , and this increase is stronger for temperatures in the temperature region of the chiral transition ; ( c ) the difference between the @xmath85 and @xmath86 quarks decreases monotonically with temperature , contrary to the prediction of the njl and pnjl with constant couplings that predict an increase of this difference until the transition temperature .
furthermore , lqcd results suggest that the su(3 ) symmetry of the pointlike effective interaction between quarks should be broken in the magnetic environment . :
this work was partially supported by project no .
ptdc / fis/ 113292/2009 developed under the initiative qren financed by the ue / feder through the program compete : `` programa operacional factores de competitividade '' , and by grant no .
sfrh / bd/51717/2011 .
we are also grateful to j. moreira for useful discussions .
thanks the conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) and t.f . and
thank the fundao de amparo a pesquisa do estado de so paulo ( fapesp ) . v. skokov , a. y. illarionov , and v. toneev , int .
a * 24 * , 5925 ( 2009 ) ; v. voronyuk , v. toneev , w. cassing , e. bratkovskaya , v. konchakovski , and s. voloshin , phys .
c * 83 * , 054911 ( 2011 ) . v. a. miransky and i. a. shovkovy , phys
. rev .
d * 66 * , 045006 ( 2002 ) .
g. s. bali , f. bruckmann , g. endrdi , z. fodor , s. d. katz , s. krieg , a. schfer , and k. k. szab , j. high energy phys . * 02 * ( 2012 ) 044 .
m. ferreira , p. costa , d.p .
menezes , c. providncia and n.n .
scoccola , phys .
d * 89 * , 016002 ( 2014 ) .
r. l. s. farias , k. p. gomes , g. i. krein and m. b. pinto , arxiv:1404.3931 [ hep - ph ] . k. fukushima and y. hidaka , phys .
* 110 * ( 2013 ) 031601 .
k. kamikado and t. kanazawa , j. high energy phys .
* 03 * ( 2014 ) 009
. j. o. andersen , w. r. naylor and a. tranberg , j. high energy phys . * 04 * ( 2014 ) 187 .
e. s. fraga , b. w. mintz and j. schaffner - bielich , phys .
b * 731 * , 154 ( 2014 ) .
k. fukushima , phys .
b * 591 * , 277 ( 2004 ) ; c. ratti , m. a. thaler , and w. weise , phys .
d * 73 * , 014019 ( 2006 ) .
d. p. menezes , m. b. pinto , s. s. avancini , a. perez martinez and c. providncia , phys . rev .
c * 79 * , 035807 ( 2009 ) ; d. p. menezes , m. b. pinto , s. s. avancini and c. providncia , phys .
c * 80 * , 065805 ( 2009 ) .
m. dutra , o. loureno , a. delfino , t. frederico , and m. malheiro , phys . rev .
d 88 , 114013 ( 2013 ) .
e. -m .
ilgenfritz , m. muller - preussker , b. petersson and a. schreiber , phys .
d * 89 * , 054512 ( 2014 ) .
m. delia , s. mukherjee and f. sanfilippo , phys .
d * 82 * , 051501 ( 2010 ) .
p. costa , m. ferreira , h. hansen , d. p. menezes and c. providncia , phys .
d * 89 * , 056013 ( 2014 ) . | the qcd phase diagram at zero chemical potential and finite temperature subject to an external magnetic field is studied within the three - flavor nambu jona - lasinio ( njl ) model and the njl model with the polyakov loop .
a scalar coupling parameter dependent on the magnetic field intensity is considered .
the scalar coupling has been fitted so that the lattice qcd pseudocritical chiral transition temperatures are reproduced and in the limit of large magnetic field decreases with the inverse of the magnetic field intensity .
this dependence of the coupling allows us to reproduce the lattice qcd results with respect to the quark condensates and polyakov loop : due to the magnetic field the quark condensates are enhanced at low and high temperatures and suppressed for temperatures close to the transition temperatures and the polyakov loop increases with the magnetic field . |
the term _ statistical properties _ of a dynamical system @xmath0 , where @xmath1 is a measurable space and @xmath2 a measurable map , refers to the behavior of large sets of trajectories of the system .
it is well known that this relates to the properties of the evolution of measures by the dynamics .
statistical properties are often a better object to be studied than pointwise behavior . in fact
, the future behavior of initial data can be unpredictable , but statistical properties are often regular and their description simpler .
suitable results can be established in many cases , to relate the evolution of measures with that of large sets of points ( ergodic theorems , large deviations , central limit , logarithm law , etc ... ) . in this paper
we take the point of view of studying the evolution of measures and its speed of convergence to equilibrium to understand the statistical properties of a class of dynamical systems .
we consider fiber - contracting maps which are a skew - product whose base transformation has exponential convergence to equilibrium ( a measure of how fast iterates of the lesbegue measure converge to the physical measure ) and deduce that the whole system has exponential decay of correlations . by this
it is possible to obtain other consequences regarding the statistical behavior of the dynamics , as the logarithm law , which will be introduced below .
the regularity required by the techiniques we use to prove decay of correlations is quite low , so we can apply it to a class of systems including poincar return maps of lorenz like systems .
we apply indeed these results to suitable poincar maps of an open dense subset of a large class of attractors of three - dimensional partially hyperbolic flows , including the _ singular - hyperbolic ( or lorenz - like ) attractors _ , introduced in @xcite .
these last kind of attractors present equilibria accumulated by regular orbits , and the lorenz attractor as well the geometric lorenz attractor are the most studied examples of this class of attractors @xcite .
the main feature of these systems we exploit to estimate the decay of correlations is the existence of a quite regular stable foliation .
we can disintegrate a measure along the foliation , and estimate the convergence of its iterates separately , along the stable and along the unstable direction , also by the use of suitable anysotropic norms .
let us mention that it was proved in @xcite that a generic @xmath3 vector field on a closed @xmath4-manifold either has infinitely many sinks or sources , or else its non - wandering set admits a finite decomposition into compact invariant sets , each one being either a uniformly hyperbolic set or a singular - hyperbolic attractor or a singular - hyperbolic repeller ( which is a singular - hyperbolic attractor for the time reversed flow ) .
this shows that the class of singular - hyperbolic attractors is a good representative of the limit dynamics of many flows on three - dimensional manifolds .
more important , _ the class of singular - hyperbolic attractors contains every @xmath3 robustly transitive isolated set _ for flows on compact three - manifolds ; see @xcite .
we obtain exponential decay of correlations for the poincar return maps of a well - chosen finite family of cross - sections for these flows . in @xcite , exponential decay of correlations and
limit laws for the two - dimensional poincar first return map of a class of geometrical lorenz attractors were proved .
we now provide similar results in a much more general context .
we remark that our results on decay of correlations are finer that the one of bunimovich @xcite and afraimovich , chernov , sataev @xcite , which can be seen as developments on the work of pesin @xcite . from the decay of correlations
we then deduce a logarithm law for the hitting times to small targets in these cross - sections .
then , exploiting the fact that _ the flow _ on these attractors can be seen as a suspension flow over poincar return maps , we also obtain a logarithm law for the hitting time for the singular hyperbolic flow itself .
we observe that quantitative dynamical properties of this class of systems are still quite unknown .
a logarithm law is a statement on the time a typical orbit hits a sequence of decreasing targets which is strictly related to the so called dynamical borel cantelli lemma and to diophantine approximation ( see @xcite and @xcite for relations with the borel cantelli lemma and other approaches ) . roughly speaking
, a system has a logarithm law for a decreasing sequence of targets @xmath5 if the time which is needed for a typical orbit to hit the @xmath6-target is in some sense inversely proportional to the measure of the target .
hence it is a quantitative statement indicating how fast a typical orbit `` fills '' the space .
hitting time results of this kind have been proved in many continuous time dynamical systems of geometrical interest : geodesic flows , unipotent flows , homogeneous spaces etc . ; see e.g. @xcite .
other examples of connections with diophantine approximation can be found in @xcite and @xcite .
we also remark that , in the symbolic setting , similar results about the hitting time are also used in information theory ( see e.g. @xcite ) .
the logarithm law automatically holds for fastly mixing systems ( see e.g , theorem [ maine ] ) , but there are mixing systems for which the law does not hold ( @xcite ) . as it will be mentioned in the next section ,
some subclass of singular hyperbolic flows has been proved to be fastly mixing but in general the speed of mixing ( and several statistical properties ) of such flows are still unknown .
the referee of the current paper has pointed out to us that , in the case of poincar maps for geometric lorenz attractors , as considered in @xcite , exponential decay of correlations can be also obtained as an application of the landmark work of young @xcite .
indeed , results on the statistical behavior of more or less general classes of lorenz like flows have also been obtained by markov tower like constructions ; see e.g. @xcite .
the approach we present here to estimate decay of correlations decouples the statistical properties of the base map of a skew - product with contracting fibers from the rest of the estimation about the speed of mixing . in the application to singular hyperbolic flows , we use the functional analytic approach to estimate the convergence to equilibrium for the base map ( a piecewise expanding map ) , but in other cases
also tower like constructions could be used in this step .
we also remark that we need very weak regularity assumptions on the base map apart from the speed of convergence to equilibrium ; see section [ sec : primeiroretorno ] and remark [ rmk : slowconverg ]
. this will be essential when studying higher - dimensional partially hyperbolic systems , like the sectional - hyperbolic attractors introduced by bonatti , pumarino , viana in @xcite and metzger - morales in @xcite , where the quotient map over the contracting fibers in general fails even to have a derivative ; it should be not more than a locally holder homeomorphism . in section [ 2p1 ]
we establish results on the decay of correlations and convergence to equilibrium for fiber contracting maps with a base transformation which rapidly converges to equilibrium .
those results imply the following statement ( which in turn will be applied to singular hyperbolic attractors ) .
we denote by @xmath7 the unit square , where @xmath8 $ ] . for a function @xmath9
we denote by @xmath10 the best lipschitz constant of @xmath11 , that is , @xmath12 where @xmath13 is the euclidean distance .
we define the lipschitz norm by setting @xmath14 where , as usual , @xmath15 and set @xmath16 .
[ mthm : expdecay ] let us consider a map @xmath17 from the unit square into itself such that : 1 .
@xmath2 has the form @xmath18 ( is a skew - product and preserves the natural vertical foliation of the square ) ; 2 .
@xmath19 is @xmath20-lipschitz with @xmath21 ( hence is uniformly contracting ) on each leaf @xmath22 of the vertical foliation of the square ; 3 .
@xmath23 ( a kind of variation in one direction ; see beginning of section [ sec : p - bounded - variation ] for the definition ) , 4 .
@xmath24 is piecewise monotonic , with @xmath25 , @xmath26 increasing branches on the intervals @xmath27, ... ,@xmath28 , ... , @xmath29 and @xmath30 .
@xmath31 has finite universal @xmath32bounded variation ( a generalization of the notion of bounded variation ; see again section [ sec : p - bounded - variation ] for the definition ) ; 6 .
@xmath33 has only one absolutely continuous ( w.r.t .
lebesgue on @xmath34 ) invariant probability measure ( a.c.i.m . ) for which it is weakly mixing . then _
the unique physical measure _ @xmath35 of @xmath2 has exponential decay of correlation with respect to lipschitz observables , that is , there are @xmath36 , such that @xmath37 the notion of physical measure is central in smooth ergodic theory .
we say that a @xmath2-invariant probability measure @xmath38 is _ physical _ if the _ ergodic basin _ of @xmath38 @xmath39 has positive area in @xmath40 ( two - dimensional lebesgue measure ) .
we remark that items ( 4 ) to ( 6 ) of the assumptions on the above theorem can be replaced by ( much more general ) exponential convergence to equilibrium on the base map under suitable observables : see theorem [ resuno ] and theorem [ thm : summary - exp - decay ] in section [ 2p1 ] .
we also remark that lipschitz norms can be replaced by suitable weaker anisotropic norms .
next , we establish a logarithm law for the dynamics of the flow .
this is a relation between hitting time to small targets and the local dimension of the invariant measure we consider .
let us consider a family of target sets @xmath41 , where @xmath42 is a given point , indexed by a real parameter @xmath43 , and let us denote the time needed for the orbit of a point @xmath44 to enter in @xmath41 by @xmath45 a logarithm law is the statement that as @xmath46 the hitting time scales like @xmath47 . to express this ,
let us consider the local dimensions of a measure @xmath48 @xmath49 representing the scaling rate of the measure of small balls as the radius goes to @xmath50 .
when the above limits coincide for @xmath51-almost every point , we say that @xmath51 is _ exact dimensional _ and set @xmath52 . from the main results of @xcite it follows that in the kind of systems mentioned above ( under mild extra conditions on the second coordinate @xmath53 of @xmath2 ) @xmath38 is exact dimensional , and the logarithm law holds .
[ cor : hitting - law ] if @xmath18 is as in the setting of theorem [ mthm : expdecay ] and * @xmath2 is injective ; * @xmath53 is @xmath3 on @xmath54 , where @xmath55 is any monotonicity interval of @xmath33 ; * @xmath56 whenever defined , @xmath57 and @xmath58 ; then the physical measure @xmath38 is exact dimensional and for @xmath38-almost every @xmath44 it holds @xmath59 we now apply these results to three - dimensional flows having a singular - hyperbolic attractor .
let @xmath60 be the family of all @xmath61 vector fields @xmath1 on a compact three - manifold @xmath62 having an open trapping region @xmath63 , i.e. , @xmath64 for all @xmath65 , such that its maximal invariant subset @xmath66 is a compact transitive singular - hyperbolic set .
this means that , for each @xmath67 , there exists a continuous invariant and dominated decomposition @xmath68 of the tangent space at each @xmath69 , where @xmath70 is a one - dimensional sub - bundle uniformly contracted by the derivative @xmath71 , and @xmath72 is a two - dimensional sub - bundle , containing the flow direction , and whose area is uniformly expanded by the action of @xmath71 , that is , there are constants @xmath73 such that @xmath74 .
in addition , there are finitely many hyperbolic singularities @xmath75 in @xmath76 , which are _ lorenz - like _ , i.e. , the eigenvalues @xmath77 , @xmath78 , of @xmath79 are real and satisfy @xmath80 ( see section [ sec : lorenzmodel ] for a more precise description ) .
we consider the @xmath61 topology of vector fields in @xmath60 in what follows .
given a flow having a singular hyperbolic attractor , there exists a finite family @xmath81 of well - adapted cross - sections of the flow where we can define a poincar return map @xmath2 which satisfies the properties in the statement of theorem [ mthm : expdecay ] after a suitable choice of coordinates . to take advantage of the above cited result from steinberger @xcite on exact dimensionality and
obtain proposition [ cor : hitting - law ] , we need that the poincar return map @xmath2 be injective , which is not evident in the construction and choice of these cross - sections done at @xcite . because of this , the construction presented here is slightly different from the ones presented elsewhere .
once constructed a suitable section , taking some nonresonance conditions on the eigenvalues of the equilibria of @xmath1 inside @xmath76 we can reduce the dynamics on @xmath76 to a map @xmath2 as in theorem [ mthm : expdecay ] , and obtain [ cor : decaycorr - sing - hyp ] there exists an open dense set @xmath82 of vector fields ( satisfying a nonresonance condition ) in @xmath60 such that , for each @xmath83 , we can find a finite family @xmath81 of cross - sections to the flow @xmath84 of @xmath1 such that an iterate of the poincar first return map @xmath85 has a finite set of srb measures @xmath86 , each of them has exponential decay of correlations with respect to lipschitz observables : there are @xmath87 satisfying for every pair @xmath88 of lipschitz functions @xmath89 we remark that , since the work of ruelle @xcite , it is well - known that exponentially mixing for the base transformation of a suspension flow does not imply fast mixing for the suspension flow .
in fact , the suspension flow might be mixing but without exponential decay of correlations as in @xcite ; or it may have arbitrarily slow decay of correlations , as shown by pollicott in @xcite . hence we can not deduce in general any kind of fast mixing results for the flow on a singular - hyperbolic attractor from corollary [ cor : decaycorr - sing - hyp ] .
however , in a recent work of one of the authors with varandas @xcite , it has been proved the existence of a @xmath61 open subset of vector fields having a geometrical lorenz attractor with exponential decay of correlations for the flow on @xmath3 observables , from which follows the exponential decay of correlations for the corresponding poincar map . but
this @xmath61 open subset was obtained under very strong conditions which can not hold in such generality as in corollary [ cor : decaycorr - sing - hyp ] . under mild ( open and dense ) conditions on the eigenvalues of the hyperbolic singularities of @xmath76
, we can show that the second coordinate @xmath53 of @xmath2 is in the setting of proposition [ cor : hitting - law ] .
hence we can also deduce the logarithm law for the poincar return map ; see section [ sec : logarithm - law - syngul ] .
we can then use this to obtain a log - law for the hitting times for the singular - hyperbolic flow , in a way similar to what was done in @xcite .
let @xmath90 and @xmath91 be the time needed for the @xmath1-orbit of a point @xmath44 to enter for the _ first time _ in a ball @xmath92 .
the number @xmath93 is the _ hitting time _ associated to the flow @xmath84 and target @xmath94 .
let us consider @xmath95 , the local dimension at @xmath42 of the unique physical measure @xmath96 supported on the singular - hyperbolic attractor ( which was constructed in @xcite ) .
[ cor : hitting - flow ] if @xmath84 is a flow over a singular - hyperbolic attractor in the setting of corollary [ cor : decaycorr - sing - hyp ] and if , in addition , the eigenvalues of every equilibrium point @xmath75 of @xmath1 in @xmath76 satisfy @xmath97 ( which includes the classical lorenz system of odes ) , then for each regular point @xmath98 such that @xmath99 exists , we have @xmath100 for @xmath96-almost each @xmath101 we begin by proving exponential decay of correlations for fiber - contracting maps in section [ 2p1 ] .
we use the exponential convergence to equilibrium of the base transformation to prove exponential convergence to equilibrium of the fibered contracting map . for this
we compare the disintegrations of the push - forward measures along the fibers using the wasserstein - kantorovich distance between measures .
we choose appropriate norms to be able to compare the size of the correlation functions and deduce the exponential decay of correlations from the exponential convergence to equilibrium . to help the reader follow the diverse norms used throughout we have compiled a list of norms in the last section [ sec : notati - norms - used ] . in section [ sec : p - bounded - variation ]
, we explain the meaning of generalized @xmath102-bounded variation and the existence of absolutely continuous invariant probability measures for @xmath103 piecewise expanding maps ( not necessarily markov maps ) whose densities are of generalized @xmath102-bounded variation .
this is based on the work of keller in @xcite .
we present singular - hyperbolic attractors and the related properties which we use in the remaining part of the paper in section [ sec : lorenzmodel ] , mainly following the works @xcite and @xcite . in section [ sec : decay - correl - poincar ]
, we analyze the poincar return map associated to a singular - hyperbolic attractor and show that this map satisfies the conditions needed to deduce exponential decay of correlations for lipschitz observables . here
we take advantage of the constructions presented in the previous section . in section [ sec : muexata ]
we also show that the physical measure of the flow is exact dimensional .
this is done using the results of @xcite .
following @xcite , in section [ sec : logarithm - law - syngul ] we use the results proved in the previous sections to establish the logarithm law for the hitting times for the poincar map and flow of a singular - hyperbolic attractor with mild conditions on the eigenvalues of its hyperbolic singularities .
* acknowledgments : * s. g. wishes to thank impa , puc , and ufrj ( rio de janeiro ) , where a part of this work has been done , for their warm hospitality .
all authors whish to thank carlangelo liverani for illuminating discussions about convergence to equilibrium and decay of correlations .
in this section we will prove that a certain class of maps , with contracting fibers and low regularity have exponential decay of correlation .
this class contains suitable poincar maps of singular - hyperbolic attractors as we will se in what follows .
the methods we use are related both to coupling and to spectral techniques , and will be implemented by adequate anisotropic norms .
we will quantify the speed of convergence of the iterates of the lesbegue measure to the physical invariant measure through the help of a certain distance on the space of probability measures . if @xmath104 is a bounded metric space , we denote by @xmath105 the set of borel probability measures on @xmath104 .
we denote by @xmath10 the best lipschitz constant of @xmath106 , that is , @xmath107 and set @xmath108 let us consider the following notion of distance between measures : given two probability measures @xmath109 and @xmath110 on @xmath104 @xmath111 we remark that adding a constant to the test function @xmath11 does not change the above difference of integrals .
we also recall that if @xmath112 is a map , then it induces an associated map @xmath113 defined as @xmath114 .
the above defined distance has the following basic properties ; see e.g ( * ? ? ?
* prop 7.1.5 ) and @xcite .
[ p2 ] [ pr : propert - wkdist][remark ] @xmath115 is a distance and if @xmath104 is separable and complete , then @xmath116 with this distance becomes a separable and complete metric space .
a sequence is convergent for the @xmath117 metrics if and only if it is convergent for the weak - star topology .
moreover , let @xmath118 be a @xmath20-contracting map , let us consider two probability measures @xmath51 , @xmath119 on @xmath120 .
then @xmath121 [ r1 ] [ conv](distance and convex combinations ) if @xmath122 , @xmath123 for @xmath124 , then @xmath125 [ lipw]if @xmath11 is @xmath126-lipschitz and @xmath127 are probability measures then @xmath128 we will see that the decay of correlation of the system is related to the speed of convergence of iterates of a starting measure to the invariant one . in @xcite the rate of convergence with respect to the @xmath115 distance was considered to obtain exponential decay of correlation for the poincar map of a geometric lorenz flow . here ,
being in a more general case , having less regularity , we can not perform the same construction .
but this idea will be part of the construction we are going to implement .
we will also use the following variation distance on the space of probability measures @xmath129 we remark that when the two measures have a density , this is the @xmath130 distance between the densities .
we denote by @xmath8 $ ] and @xmath7 .
we consider the @xmath131 distance on @xmath40 so that the diameter is one : @xmath132 .
this choice is not essential , but will avoid the presence of many multiplicative constants in the following , making notations cleaner
. the square @xmath40 will be foliated by stable , vertical leaves .
we will denote the leaf with @xmath44 coordinate by @xmath133 or , with a small abuse of notation , when no confusion is possible , we will denote both the leaf and its coordinate by @xmath134 .
let @xmath135 be the measure @xmath109 such that @xmath136 .
let @xmath51 be a probability measure on @xmath40 .
such measures on @xmath40 will be often disintegrated in the following way : for each borel set @xmath137 @xmath138 with @xmath139 being probability measures on the leaves @xmath120 and @xmath140 the marginal on the @xmath44 axis , which will be absolutely continuous with respect to the lebesgue length measure .
we denote by @xmath141 the density of @xmath142 .
now we consider the integral of a suitable observable @xmath9 .
let us consider the following anisotropic norm , considering lipschitz regularity only on the vertical direction .
let @xmath143 be defined by @xmath144 where for simplicity we consider @xmath145}{\sup } |g(x , y)| \quad\text{and}\quad { \operatorname{lip}}_y(g):=\sup_{\substack { x , y_{1},y_{2}\in \lbrack 0,1 ] \\
y_1\neq y_2 } } \frac{|g(x , y_{2})-g(x , y_{1})|}{|y_{2}-y_{1}|}.\ ] ] if @xmath146 and @xmath147 are two disintegrated measures as above , the integral of an observable @xmath11 can be estimated as function of the above norm , some distance between their respective marginals on the @xmath44 axis and measures on the leaves .
[ prod]let @xmath146 , @xmath147 be measures on @xmath40 as above , such that for each borel set @xmath137 @xmath148where @xmath149 is absolutely continuous with respect to the lebesgue measure .
in addition , let us assume that 1 .
@xmath150 2 .
@xmath151 .
then ) that @xmath152 is defined in some way , for example @xmath153 ( the one dimensional lebesgue measure on the leaf ) for each leaf where the density of @xmath154 is null .
] @xmath155 disintegrating @xmath146 and @xmath147 we get @xmath156adding and subtracting @xmath157 the last expression is equivalent to @xmath158 this is bounded by ( see remark [ lipw ] ) @xmath159for almost each @xmath120 it holds @xmath160 . thus , by assumption ( 2 ) in the statement , the proposition is proved .
let us consider a manifold @xmath161 ( possibly with boundary ) and the dynamics on @xmath161 generated by the iteration of a function @xmath162 . we will consider a notion of speed of approach of an absolutely continuous initial measure , with density @xmath163 , to an invariant measure @xmath51 . of course
, many generalizations are possible , but we will consider here only absolutely continuous measures as starting measures .
[ def : exp - conv - equil ] we say that @xmath33 has exponential convergence to equilibrium with respect to norms @xmath164 and @xmath165 , if there are @xmath166 such that for @xmath167 @xmath168 where @xmath169 is the lebesgue measure on @xmath161 . we remark that in several situations , exponential convergence to equilibrium can be obtained as consequence of spectral gap of the transfer operator restricted to suitable funcion spaces , see lemma [ eigen2 ] .
now we consider maps preserving a regular foliation , which contracts the leaves and whose quotient map ( the induced map on the space of leaves ) has exponential convergence to equilibrium .
we will give an estimation of the speed of convergence to equilibrium for this kind of maps , establishing that it is also exponential .
[ def : pi - f ] if @xmath170 is integrable , we denote by @xmath171 the function @xmath172 [ resuno ]
let @xmath17 be a borel function such that @xmath18 .
let @xmath51 be an @xmath2-invariant measure with marginal @xmath140 on the @xmath44-axis which , moreover , is @xmath33-invariant .
let us suppose that 1 .
@xmath173 has exponential convergence to equilibrium with respect to the norm @xmath174 ( the @xmath175 norm ) and to a norm which we denote by @xmath176 .
@xmath33 is nonsingular with respect to the lesbegue measure , piecewise continuous and monotonic : there is a collection of intervals @xmath177 , @xmath178 such that on each @xmath179 , @xmath33 is an homeomorphism onto its image .
3 . @xmath2 is a contraction on each vertical leaf : @xmath53 is @xmath20 -lipschitz in @xmath180 with @xmath181 .
then @xmath182 has exponential convergence to equilibrium in the following sense .
there are @xmath166 such that @xmath183 for each @xmath184 .
let us take an integrable non - negative @xmath185 and an essentially bounded @xmath186 .
we can divide and multiply by @xmath187 obtaining @xmath188we denote @xmath189 and disintegrate the measures .
let us suppose having iterated the system @xmath190 times , and then continue to iterate again @xmath191 more times . by item ( 1 ) after the first @xmath190 iterations , the marginals become exponentially close:@xmath192 now iterating @xmath191 more times we estimate the distance on the leaves , and then the speed of convergence by theorem [ prod ] .
let us consider @xmath193 , the intervals where the continuous branches of @xmath194 are defined .
let us consider @xmath195 and @xmath196 , @xmath197 so that @xmath198 , @xmath199 .
remark that by item ( 1 ) after the first @xmath190 iterations , also the marginals of @xmath200 and @xmath201 are exponentially close , and since @xmath202 have disjoint support:@xmath203 by triangle inequality@xmath204 let us denote @xmath205 , remark that this is injective . then by proposition [ prod]@xmath206 and by uniform contraction along the leaves we can bound @xmath207 summarizing , we obtain @xmath208 since @xmath209 is a @xmath210 contraction , from this is less or equal than @xmath211 finally we can deduce @xmath212\end{aligned}\ ] ] which implies the statement and concludes the proof of theorem [ resuno ] .
[ rem1 ] since @xmath213 we can also state : under the assumptions of theorem [ resuno ] , if @xmath184 @xmath214 [ rmk : slowconverg ] we remark that if , instead of exponential convergence to equilibrium for the base map ( @xmath215 $ ] ) , we had a slower rate of convergence ( @xmath216 with @xmath217 slower than exponential ) , then because @xmath218 will converge to @xmath50 much faster than @xmath219 ( see the last equation in the above proof ) , the same arguments lead to an estimation for the convergence to equilibrium for the skew product with the same kind of asymptotical behavior as @xmath219 .
now let us relate convergence to equilibrium to decay of correlations .
we consider the general case of a measure preserving map .
[ lemmone ] let us consider a measurable map @xmath220 , two probability measures @xmath221 on @xmath222 , such that @xmath51 is invariant .
if we have a convergence to equilibrium with speed @xmath223 and norms @xmath224 and @xmath225 , that is @xmath226then , for each @xmath227 , @xmath228 is bounded by @xmath229 where @xmath230 is the constant function with value @xmath231 adding and subtracting we rewrite @xmath232 as @xmath233 applying the assumption to each line we obtain the three summands in the statement .
now we use the above results to deduce exponential decay of correlations from exponential speed of convergence to equilibrium .
[ resdue ] let @xmath2 be a map satisfying the assumptions of theorem resuno and let us suppose that there are @xmath234 and a seminorm @xmath235 such that for each @xmath236 @xmath237then @xmath2 has exponential decay of correlations : there are @xmath238 such that for @xmath236 @xmath239 let us consider bounded observables @xmath240 .
since adding a constant to @xmath163 the correlation integral does not change , we can suppose @xmath184 and we can apply theorem [ resuno ] . by theorem [ resuno ] , remark [ rem1 ] and lemma [ lemmone ] , there are @xmath241 s.t .
for each @xmath227@xmath242 by assumption ( [ square ] ) , we take @xmath243 ( the integer part ) with @xmath244 so small that @xmath245 goes to zero exponentially ; that is , there are @xmath246 such that @xmath247 let us evaluate each term of the above sum .
we recall the definition of @xmath248 from ( [ eq : lip_y ] ) .
we observe that we can bound the first summand in the above inequality in the following way @xmath249since @xmath2 is contracting on the vertical direction , when @xmath191 grows we get latexmath:[\ ] ] and @xmath316 + 1}{\operatorname{osc}}(h,\frac{a}{2}+ia,\frac{a}{2})\geq { \operatorname{osc}}(f,\frac{1}{2},\frac{1}{2})$ ] by the triangle inequality .
thus @xmath317 finally @xmath318 and then @xmath319 let us consider the iteration of piecewise expanding maps whose inverse of the derivative has @xmath32bounded variation .
let @xmath320 a finite interval partition of @xmath34 and @xmath321 be a transformation which is monotone and continuous on each interval @xmath179 .
we assume that 1 .
@xmath33 is nonsingular with respect to the measure @xmath169 ; 2 .
@xmath322 exists inside the intervals and @xmath31 is bounded almost everywhere ; under these assumptions let us consider a measurable and bounded function @xmath163 and the perron - frobenius operator related to @xmath33 @xmath323 this operator represent the action of the transfer operator on densities of absolutely continuous measures , and extends to a linear contraction on @xmath324 . under these assumptions , if the map @xmath33 is piecewise expanding , and @xmath325 has universal @xmath32bounded variation , a kind of lasota yorke inequality can be proved ( see ( * ? ? ? * thm 3.2 and 3.5 ) ) .
[ th - keller ] let @xmath326 be as described above . 1 .
if @xmath327 ( where @xmath297 ) and 2 . there is an @xmath328 with @xmath329 , then there are @xmath330 and @xmath331 such that for each @xmath332 @xmath333 since @xmath334 is a @xmath324-contraction , the theorem of ionescu - tulcea - marinescu @xcite provides a description of the spectral properties of @xmath334 ; see ( * ? ? ?
3.3 ) . 1 .
@xmath335 has a finite number of eingenvalues @xmath336 of modulus @xmath230 ; 2 .
let @xmath337 , then @xmath338 and @xmath339 ( @xmath340 ) ; 3 .
@xmath341 , where @xmath342 are projections onto the eigenspaces @xmath343 with @xmath344 and @xmath345 is a linear operator on @xmath324 with @xmath346 , @xmath347 and @xmath348 for some @xmath349 .
furthermore @xmath350 ( @xmath351 ) and @xmath352 ( for all @xmath6 ) ; 4 .
@xmath230 is an eigenvalue of @xmath334 , and assuming @xmath353 and @xmath354 , @xmath355 is the greatest @xmath33-invariant probability on @xmath1 absolutely continuous with respect to @xmath169 , i.e. , if @xmath356 is @xmath33-invariant and @xmath357 , then @xmath358 ; 5 .
there is a finite partition @xmath359 of @xmath360 $ ] such that @xmath361 for @xmath362 and @xmath363 ; moreover , @xmath364 is weakly mixing for each @xmath365 .
[ eigen1 ] the above result ( item ( 5 ) ) tells that some iterate @xmath366 of the transfer operator has a finite set of a.c.i.m . with no eigenvalues other than @xmath230 on the unit circle .
an iterated of the system can be hence decomposed into a finite union of invariant sets and each invarant subsystem has a unique a.c.i.m and no other eigenvalues on the unit circle .
this implies exponential _ convergence to equilibrium _ for some iterate of the map , as needed in theorem [ resdue ] .
[ eigen2 ] under the same assumptions as above , if @xmath367 , @xmath48 is an a.c.i.m .
, the associated transfer operator has @xmath230 as a simple eigenvalue and there are no more eigenvalues on the unit circle , then there is @xmath368 s.t .
@xmath369 let us set @xmath370 .
since @xmath230 is a single eigenvalue and there are no more eigenvalues on the unit circle , then @xmath371 ( this decomposition is invariant by @xmath334 ) , and @xmath372 is contained in a disc with radius @xmath373 . without loss of generality
we can suppose that @xmath374 .
by item ( 4 ) above , @xmath375 is the density of the invariant measure @xmath51 , therefore @xmath376 where @xmath377 is the spectral projection onto @xmath378 then , by lemma [ lemaa ] and the spectral radius theorem , we get the needed estimation @xmath379
in this section we will introduce what nowadays is called a singular - hyperbolic attractor for a @xmath4-dimensional vector field @xmath1 ( or flow @xmath84 ) .
the singular - hyperbolic class of attractors we consider contains the partially hyperbolic attractors @xmath76 of a @xmath61 vector field @xmath1 , with finitely many equilibria accumulated by regular orbits of @xmath1 in @xmath76 , and admitting a continuous and @xmath71-invariant splitting of the tangent bundle over @xmath76 into a pair @xmath380 of vector sub - bundles . here
@xmath70 has one - dimensional fibers and is uniformly contracted by @xmath71 , and @xmath381 has two - dimensional fibers whose area is uniformly expanded by @xmath71 ; see below for precise definitions .
as shown in @xcite , this class is an extension of the class of uniformly hyperbolic attractors , since every singular - hyperbolic attractor containing no equilibria of @xmath1 is uniformly hyperbolic , that is , the sub - bundle @xmath381 admits a further continuous and @xmath71-invariant splitting @xmath382 into the line bundle generated by the flow direction over @xmath76 , and the bundle @xmath383 with one - dimensional fibers uniformly expanded by @xmath71 . in particular , anosov ( or globally hyperbolic ) flows on three - dimensional manifolds
belong to this class .
moreover , and more important , _ the class of singular - hyperbolic attractors contains every @xmath3 robustly transitive isolated set _ for flows on compact three - manifolds @xmath161 .
that is , as proved in @xcite , if for a given open subset @xmath63 of @xmath161 there exists a @xmath3 open subset @xmath384 of @xmath385 ( the family of @xmath3 vector fields of @xmath161 ) such that the maximal invariant subset @xmath386 of @xmath63 contains a non - trivial transitive orbit of @xmath104 for every @xmath387 , then @xmath386 is a singular - hyperbolic attractor and contains at most finitely many hyperbolic saddle - type equilibria , either for @xmath104 of for @xmath388 , for each @xmath387 .
next we describe precisely what we mean by a singular - hyperbolic attractor .
we start by recalling some definitions and notations and then we shall list some facts about this kind of attractors proved elsewhere but that will be useful for us .
we also adapt some known results to our case and add some new ones , to meet the requirements needed to prove the decay of correlations and the logarithm law ; the main results proved are listed in theorem [ thm : propert - singhyp - attractor ] below .
let @xmath161 be a @xmath4-dimensional compact riemanian manifold and @xmath389 be set of @xmath390 vector fields ( or flows ) defined on @xmath161 .
given a compact invariant set @xmath76 of @xmath391 , we say that @xmath76 is _ isolated _ if there exists an open set @xmath392 such that @xmath393 if @xmath63 above can be chosen such that @xmath394 for @xmath65 , we say that @xmath76 is an _ attracting set_. given @xmath395 , a point @xmath396 is _ regular _ if @xmath397 .
in this case we refer to its orbit as a _
regular @xmath1-orbit or regular orbit for short_. given @xmath398 , we define @xmath399 as the set of accumulation points of the positive orbit @xmath400 of @xmath102 .
we also define @xmath401 , where @xmath402 is the time reversed vector field .
a subset @xmath403 is _ transitive _ if it has a full dense orbit , that is , there is @xmath404 such that @xmath405 .
[ def : attractor ] an _ attractor _ is a transitive attracting set , and a _ repeller _ is an attractor for the reversed vector field @xmath406 .
an attractor , or repeller , is _ proper _ if it is not the whole manifold .
an invariant set of @xmath1 is _ non - trivial _ if it is neither a periodic orbit nor an equilibrium of @xmath1 .
recall that a point @xmath407 is an equilibrium of @xmath1 if @xmath408 .
[ d.dominado ] let @xmath76 be a compact invariant set of @xmath395 , @xmath409 , and @xmath410 .
we say that @xmath76 has a @xmath411-dominated splitting if the bundle over @xmath76 can be written as a continuous @xmath71-invariant sum of sub - bundles @xmath412 such that for every @xmath413 and every @xmath414 , we have @xmath415 we say that @xmath76 is _ partially hyperbolic _ if it has a @xmath411-dominated splitting such that the sub - bundle @xmath416 is uniformly contracting , that is , for some @xmath409 and every @xmath413 and each @xmath414 we have @xmath417 in this case we denote the one - dimensional bundle @xmath416 by @xmath418 and the two - dimensional bundle @xmath419 by @xmath420 .
we refer to @xmath418 as the contracting direction and to @xmath420 as the central or center - unstable direction of the splitting .
the next proposition is an immediate consequence of ( * ? ? ? * theorem 1 ) and will simplify many of the arguments .
[ nikolaz ] there exists an adapted riemannian metric @xmath421 , equivalent to the original one , and @xmath422 such that @xmath423 for all @xmath424 , and all @xmath69 . throughout the remaining of the paper
we assume that the riemannian metric is an adapted metric and the first sub - bundle @xmath416 is one - dimensional . for simplicity
we denote the adapted metric by @xmath425 . for @xmath426 and @xmath427
we let @xmath428 be the absolute value of the determinant of the linear map @xmath429 we say that the sub - bundle @xmath381 of the partially hyperbolic invariant set @xmath76 is _ volume expanding _
if @xmath430 in this case we say that @xmath381 is _
@xmath431-volume expanding _ to indicate the dependence on @xmath432 . this condition is weaker than the uniform exponential expansion along the central direction . the geometric lorenz attractor has a volume expanding central direction that is not uniformly expanding @xcite . the domination condition , together with the volume expanding condition ( [ volume ] ) along the central direction ,
imply that the direction of the flow is contained in the central bundle @xmath420 ( * ? ? ?
* lemma 6.1 , pg 163 ) . recall that an equilibrium @xmath75 of @xmath1 is hyperbolic only if the real part of every eigenvalue of @xmath79 is distinct of zero , see ( * ? ? ?
* section 2.1,pp 6 ) .
recall also the definition of a special type of equilibrium of a vector field @xmath1 in a @xmath4-manifold .
[ lorenz - like ] we say that an equilibrium @xmath75 of a @xmath4-vector field @xmath1 is lorenz - like if the eigenvalues @xmath433 , @xmath434 , of @xmath79 are real and satisfy @xmath435 if @xmath75 is a lorenz - like equilibrium , letting @xmath436 and @xmath437 we obtain @xmath438 [ d.singularset ] let @xmath76 be a compact invariant set of @xmath395 .
we say that @xmath76 is a _ singular - hyperbolic set _ for @xmath1 if all the equilibria of @xmath76 are hyperbolic , and @xmath76 is partially hyperbolic with volume expanding central direction .
a singular - hyperbolic set which is also an attractor will be called a _ singular - hyperbolic attractor_. the next result is the content of ( * ? ? ?
* lemma 5.27 ) : [ singularidadelorenz - like ] let @xmath76 be a singular - hyperbolic attractor of a @xmath4-dimensional vector field @xmath1 .
then , every equilibrium @xmath75 properly accumulated by regular orbits within @xmath76 is lorenz - like either for @xmath1 or for @xmath406 .
an isolated set @xmath76 of a @xmath3 vector field @xmath1 is robustly transitive if it has an open neighborhood @xmath63 such that @xmath439 is both transitive and non - trivial ( i.e. , _ it is neither an equilibrium point nor a periodic orbit _ ) for any vector field @xmath104 @xmath3-close to @xmath1 . roughly speaking , @xmath76 is robustly transitive if it can not be destroyed by small @xmath3 perturbations .
morales , pacifico , and pujals proved in @xcite that any _ transitive robust invariant set _ of a @xmath4-dimensional flow containing some equilibrium is a singular - hyperbolic _ attractor or repeller_. in the absence of equilibria , robustness implies uniform hyperbolicity .
the most meaningful examples of singular - hyperbolic attractors are the lorenz attractor @xcite and the so called geometric lorenz attractor @xcite .
the main results we explain and derive in this section that will be used in the rest of the paper are stated in the following .
[ thm : propert - singhyp - attractor ] for an open and dense subset of @xmath61 vector fields @xmath1 having a singular hyperbolic attractor @xmath76 on a @xmath4-manifold , there exists a finite family @xmath81 of cross - sections and a global ( @xmath191-th return ) poincar map @xmath440 , @xmath441 such that 1 .
the domain @xmath442 is the entire cross - sections with a family @xmath443 of finitely many smooth arcs removed and @xmath444 is a smooth function bounded away from zero by some uniform constant @xmath445 .
2 . we can choose coordinates on @xmath81 so that the map @xmath446 can be written as @xmath447 , @xmath18 , where @xmath7 , @xmath8 $ ] and @xmath448 with @xmath449 and @xmath450 a finite set of points .
the map @xmath451 is @xmath103 piecewise monotonic with @xmath25 branches defined on the connected components of @xmath452 and has a finite set of a.c.i.m .
, @xmath453 .
also @xmath454 where it is defined , @xmath325 has universal bounded @xmath102-variation and then @xmath455 has bounded @xmath102-variation .
the map @xmath456 preserves and uniformly contracts the vertical foliation @xmath457 of @xmath40 : there exists @xmath458 such that @xmath459 for each @xmath460 .
in addition , the map @xmath53 satisfies @xmath461 .
the map @xmath2 admits a finite family of physical probability measures @xmath462 which are induced by @xmath453 in a standard way .
the poincar time @xmath463 is integrable both with respect to each @xmath462 and with respect to the two - dimensional lebesgue area measure of @xmath40 .
6 . moreover if , for all singularities @xmath464 , we have the eigenvalue relation @xmath465 , then the second coordinate map @xmath53 of @xmath2 has a bounded partial derivative with respect to the first coordinate , i.e. , there exists @xmath331 such that @xmath466 for all @xmath467 .
next we give the notions and establish notations needed to describe the properties of singular - hyperbolic attractors stated in theorem [ thm : propert - singhyp - attractor ] .
hereafter , @xmath76 is a singular - hyperbolic attractor of @xmath468 with invariant splitting @xmath469 with @xmath470 .
let @xmath471 be a continuous extension of this splitting to a small neighborhood @xmath472 of @xmath76 . for convenience , we take @xmath472 to be forward invariant and such that @xmath473 .
we will see in subsection [ sec : extens - laminat - wss ] , that @xmath474 may be chosen invariant under the derivative .
in general , the extension @xmath475 of the center - unstable direction can not be assumed to be invariant ; see @xcite and also ( * ? ? ? * appendix b ) .
however , we can always consider a cone field around it on @xmath472 @xmath476 which is forward invariant for some @xmath477 , that is , there is a large @xmath478 depending on @xmath479 , but not depending on @xmath480 , such that @xmath481 moreover , we may take @xmath477 arbitrarily small , reducing @xmath472 if necessary . for notational simplicity , we write @xmath418 and @xmath420 for @xmath482 and @xmath483 in all that follows . next let us recall a few classical facts about partially hyperbolic systems , especially existence of stable and strong - stable foliations , and center - unstable foliations .
the standard reference are ( * ? ? ?
* theorems 4.1 , 5.1 and 5.5 ) and ( * ? ? ?
* theorem iv.1 ) . before stating the appropriate result we need to review some terms in @xcite ;
namely , the notion of immediate relative @xmath484 pseudo - hyperbolic splitting . to define this ,
recall that if @xmath485 is an injective linear map and @xmath486 is a subspace , then the _ conorm of l restricted to g _ is @xmath487 the conorm gives the minimal expansion of @xmath488 on @xmath53 .
let @xmath489 be a flow and @xmath222 be an @xmath1-invariant compact set .
a splitting @xmath490 is _ immediate relatively @xmath484 pseudo - hyperbolic _ relative to @xmath1 if there exists a continuous function @xmath491 such that @xmath492 using proposition [ nikolaz ] we have the following [ l.new ] let @xmath1 be a @xmath61 flow and @xmath76 be a singular - hyperbolic attractor with splitting @xmath493 .
then @xmath494 is an immediate relatively @xmath484 pseudo - hyperbolic splitting over @xmath76 relative to @xmath1 for some continuous function @xmath495 .
the domination assumption ensures that @xmath496 for all @xmath69 , so @xmath497 and the above function @xmath498 is continuous since a dominated splitting is continuous .
we then have the following restatement of theorems ( * ? ? ?
* theorem iv.1 ) and ( * ? ? ?
* theorem 5.5 ) : [ contributions ] let @xmath1 be a @xmath61 flow and @xmath76 be a compact @xmath1-invariant singular - hyperbolic attractor @xmath76 having a dominated splitting @xmath499 .
there are @xmath500 and @xmath501 such that for every point @xmath502 there are two embedded discs @xmath503 and @xmath504 tangent at @xmath44 to @xmath505 and @xmath506 respectively , and satisfying , for all @xmath65 1 .
@xmath503 is a @xmath61 embedded disc , 2 .
@xmath507 3 .
@xmath508 , 4 .
the embedding @xmath503 depends @xmath61 smoothly on @xmath44 in the following sense : there is a neighborhood @xmath509 of @xmath44 in @xmath76 and a continuous map @xmath510 such that @xmath511 and @xmath512 where @xmath513 is the collection of all embeddings @xmath514 endowed with the @xmath61 distance ; 5 .
@xmath515 , where @xmath516 6 .
the @xmath504 depend @xmath61 smoothly on @xmath44 as in .
that is , there exists a continuous map @xmath517 such that @xmath511 and @xmath518 , where @xmath519 is the collection of all embeddings @xmath520 from the unit @xmath521-disk @xmath522 into @xmath161 endowed with the @xmath61 distance .
note that , for @xmath69 , @xmath503 and @xmath504 are embedded discs , and so , sub - manifolds of @xmath161 .
we refer to @xmath503 as local _ strong - stable _ manifold and to @xmath504 as local _ center - unstable _ manifold . since @xmath418 is uniformly contracting we have that @xmath503 is uniquely defined .
but we stress that the center - unstable @xmath504 manifold _ is not _
unique without further assumptions ; see @xcite .
the set @xmath523 is called the _ stable manifold _ at @xmath69 .
the proof that @xmath524 is a manifold is contained in ( * ? ? ?
* theorem 5.5 ) .
denoting @xmath525 , where @xmath526 is the direction of the flow at @xmath44 , it follows that @xmath527 here we show that the collection of @xmath61 strong - stable leaves through the points of @xmath76 , depending continuously on the base point , which is known as a lamination , can be extended to an invariant foliation ( in the usual sense from differential topology ) of a open neighborhood of @xmath76 , whose leaves are @xmath61 submanifolds and whose foliated charts are of class @xmath3 . in addition , these leaves are uniformly contracted by the action of the flow .
the argument we present below follows ( * ? ? ?
* appendix 1 ) closely .
let us fix a neighborhood @xmath472 of the singular - hyperbolic attractor @xmath76 on the manifold @xmath161 such that the closure of @xmath529 is contained in @xmath472 for all @xmath65 .
we consider the following family of directions through the points of @xmath472 @xmath530 this set is clearly a smooth manifold of dimension @xmath531 .
the time - one map of the flow @xmath532 induces a map @xmath533 given by @xmath534 naturally , this map @xmath535 is of class @xmath3 if @xmath163 is of class @xmath61 .
we note that the subset @xmath536 is compact and fixed by @xmath535 .
we claim that it is also a partially hyperbolic subset for @xmath535 .
to prove this , we consider the inverse map @xmath537 . indeed , for every @xmath538 we can calculate the derivative @xmath539 of @xmath540 at @xmath538 using coordinates provided by the splitting of @xmath541 , as follows .
directions on @xmath542 near the stable direction can be seen as graphs of a linear map @xmath543 which in fact is given by the vector @xmath544 , since @xmath545 is one - dimensional .
the action of @xmath546 on the directions of @xmath542 can be represented by the graph transform @xmath547 .
so in this coordinates we have @xmath548 .
then we have , since a tangent direction at @xmath549 is another linear map @xmath550 given by the vector @xmath551 @xmath552 which we may represent by the following isomorphism of @xmath553 @xmath554 we have now using the domination assumption from @xmath555 and by condition we get @xmath556 since the assumption ensures that @xmath557 which is the same as @xmath558 this shows the partial hyperbolicity of @xmath222 .
hence the map @xmath539 is _ immediate relative @xmath484-pseudo - hyperbolic _ for some function @xmath484 strictly smaller than @xmath230 over @xmath222
. we may now use ( * ? ? ?
* theorem iv.1 ) and ( * ? ? ?
* theorem 5.5 ) to obtain a @xmath3 center - unstable manifold @xmath559 for @xmath540 through each point @xmath560 , which is a center - stable manifold for @xmath535 , with dimension @xmath4 and tangent to @xmath561 at @xmath562 .
this manifold projects to a neighborhood of @xmath69 on @xmath161 through the canonical projection of @xmath563 on the first coordinate .
this means that on a neighborhood @xmath564 of @xmath76 we can define a field of directions @xmath565 such that @xmath566 for some @xmath69 . integrating these direction we obtain @xmath61 one - dimensional submanifolds @xmath567 passing through @xmath568 (
note that the field of direction is @xmath3 smooth because @xmath569 was @xmath3 ) . since the lamination @xmath569 is @xmath535-invariant , we can deduce that @xmath570 for all @xmath571 .
_ we have obtained a one - dimensional foliation of a neighborhood @xmath63 of @xmath76 which extends the strong - stable lamination provided by theorem [ contributions ] . for small enough @xmath63 the leaves of this foliation
are uniformly contracted by a rate close to @xmath572 under the action of @xmath163 .
_ this results extend to the action of @xmath84 in a standard way : we have the same conclusions for the diffeomorphism @xmath84 for @xmath65 in the place of @xmath532 .
[ cor : extension - stable - foliation ] for any compact invariant subset @xmath76 of a @xmath61 flow @xmath84 which is partially hyperbolic , that is , @xmath76 satisfies conditions and for a continuous @xmath71-invariant splitting @xmath573 with one - dimensional stable direction , there exists a neighborhood @xmath63 of @xmath76 in the ambient manifold @xmath161 where a extension @xmath574 of the local stable lamination @xmath575 is defined , is a locally @xmath84-invariant foliation and its leaves are uniformly contracted by @xmath84 , for all @xmath65 .
each leaf is a @xmath61 one - dimensional sub - manifold of @xmath161 .
the dependence of @xmath503 on @xmath44 is similar to item ( 4 ) of theorem [ contributions ] but in the @xmath3 topology .
we stress that the smoothness of @xmath576 is in general _ not _ related to the differentiability of the foliation @xmath577 of the neighborhood @xmath63 of @xmath76 ; see e.g. @xcite .
now let @xmath578 be a _
cross - section _ to the flow , that is , a @xmath61 embedded compact disk transverse to @xmath1 at every point @xmath579 .
we assume from now on that cross - sections are contained in the open neighborhood of @xmath76 where a contracting foliation @xmath580 which extends the strong - stable lamination through the points of @xmath76 is defined . for @xmath581
we define @xmath582 to be the connected component of @xmath583 that contains @xmath44 , where @xmath584 is the central - stable leaf obtained from the strong - stable leaf @xmath574 in a manner similar to ( [ eq : center - stable - manifold ] ) .
since the flow @xmath585 is @xmath61 , @xmath582 is a @xmath61 co - dimension one embedded curve for every @xmath586 .
these leaves form a foliation @xmath587 of @xmath578 . from the celebrated work of anosov @xcite and more recent developments in the partially hyperbolic setting by pugh - shub @xcite and brin - pesin ( * ? ? ?
* theorem 3.1 ) , it is known that the holonomies ( projection along leaves ) between pairs of transverse surfaces to @xmath580 admit a hlder jacobian with respect to lebesgue induced measure .
this naturally implies a similar statement for holonomies transverse to @xmath588 . in this setting the holonomy ( projection ) between pairs of transverse curves to @xmath589 along the lines of @xmath589 can be seen as maps between intervals of the real line having a hlder jacobian with respect to lebesgue measure .
hence these holonomies are @xmath103 maps , for some @xmath590 which depends on @xmath1 only ; see e.g. ( * ? ? ?
* section 2.7.2 ) . in this case
the leaves @xmath582 , for @xmath591 , define a foliation @xmath589 of @xmath578 whose transversal smoothness is hlder-@xmath3 .
[ r.foliated ] given a cross - section @xmath578 there is no loss of generality in assuming that it is the image of the square @xmath592 by a @xmath103 diffeomorphism @xmath593 , for some @xmath590 , which sends vertical lines inside leaves of @xmath594 .
we denote by @xmath595 the image of @xmath596 under the above - mentioned diffeomorphism , which we call the _ interior _ of @xmath578 .
we also assume that each cross - section @xmath578 is contained in @xmath472 , so that every @xmath581 is such that @xmath597 . from now on
we always assume that cross - sections are of this kind .
given any two cross - sections @xmath578 and @xmath598 to the flow , a _
poincar map _ is a map defined by @xmath599 ( for a suitable hitting time @xmath600 which will be precised later ) .
we note that , in general , @xmath446 needs not correspond to the first time the orbits of @xmath578 encounter @xmath598 , nor it is defined everywhere in @xmath578 . if @xmath446 is defined at @xmath591 , a time @xmath601 so that @xmath602 is called a _
poincar time _ of @xmath44 .
the continuity of the flow implies that , if @xmath446 is defined at @xmath591 , then it is defined in an open neighborhood @xmath603 of @xmath44 in @xmath578 and it is a @xmath3 local diffeomorphism , see ( * ? ? ?
* proposition 1.2 , pp 94 ) .
let @xmath578 be a cross - section to @xmath1 and @xmath604 be a poincar map @xmath605 to another cross - section @xmath598 ( possibly @xmath606 ) , defined as above .
the splitting @xmath499 over @xmath472 induces a continuous splitting @xmath607 of the tangent bundle @xmath608 to @xmath578 ( and analogously for @xmath598 ) , defined by ( recall for the use of @xmath609 ) @xmath610 the next result establishes that if the poincar time @xmath611 is sufficiently large then ( [ eq.splitting ] ) defines a hyperbolic splitting for the transformation @xmath446 on the cross - sections . given a pair @xmath612 of cross - sections in @xmath81 , we write @xmath613 for the subset of points of @xmath578 whose poincar map is defined and hits @xmath598 .
[ p.secaohiperbolica ] ( * ? ? ?
* proposition 6.15 , pp 172 ) let @xmath614 be a poincar map as before with poincar time @xmath615
. then @xmath616 at every @xmath617 and @xmath618 at every @xmath619 .
moreover , for every given @xmath458 there exists @xmath620 such that if @xmath621 at every point , then @xmath622 given a cross - section @xmath578 , a positive number @xmath484 , and a point @xmath591 , we define the unstable cone of width @xmath484 at @xmath44 by @xmath623 ( we omit the dependence on the cross - section in our notations ) .
we note that @xmath624 .
let @xmath625 be a small constant .
in the following consequence of proposition [ p.secaohiperbolica ] ( which is itself a consequence of partial hyperbolicity for the splitting @xmath499 ) we assume that the neighborhood @xmath472 has been chosen sufficiently small , depending on @xmath484 and on a bound on the angles between the flow and the cross - sections .
* corollary 6.17 , pp 173 ) [ ccone ] for @xmath604 as in proposition [ p.secaohiperbolica ] , with @xmath621 , and any @xmath626 , we have @xmath627 the proof of this corollary is based on the observation that , for small @xmath628 , the vectors in @xmath629 can be written as the direct sum of a vector in @xmath630 , which is expanded at a rate @xmath631 , with a vector in @xmath632 , which is contracted at a rate @xmath572 .
hence , for small @xmath484 , the center - unstable component dominates the stable component and the length of the vector is increased at a rate close to @xmath631 . in this way
we can always achieve an arbitrarily large expansion rate along the directions of the unstable cone as long as we take @xmath572 sufficiently close to zero and , consequently , a big enough threshold time @xmath633 .
let us introduce the following notion : a _ cu - curve _ in @xmath578 is a _
curve contained in a cross - section @xmath634 whose tangent direction @xmath635 is contained in a center - unstable cone _
@xmath636 for all @xmath637 ; see below .
[ rmk : boundary - cu - curve ] the cone @xmath638 is defined at every @xmath581 and we can choose @xmath578 so that the diffeomorphism @xmath593 sends horizontal lines into @xmath639-curves , i.e. , curves whose tangent directions are contained in the @xmath639-cone at every point . the next step is to exhibit stable manifolds for poincar transformations @xmath604 .
the natural candidates are the intersections @xmath640 we introduced previously . by construction , these leaves are contracted by the action of the flow and so they are contracted by the transformation @xmath446 . moreover , as already commented before , these intersections define a @xmath103 stable foliation @xmath589 of @xmath578 with a hlder-@xmath3 holonomy . for our purposes
it is also important that this foliation be invariant : @xmath641 in order to have this we restrict our class of cross - sections so that the center - unstable boundary is disjoint from @xmath76 .
we recall ( see remark [ r.foliated ] ) that we are considering cross - sections @xmath578 that are diffeomorphic to the square @xmath592 , with the vertical lines @xmath642 being mapped to stable sets @xmath643 .
the _ stable boundary _
@xmath644 is the image of @xmath645 .
the _ center - unstable ( or @xmath639-)boundary _ @xmath646 is the image of @xmath647 .
the cross - section is _
@xmath648-adapted _ if @xmath649 where @xmath650 is the intrinsic distance in @xmath578 .
we also recall that , from remark [ rmk : boundary - cu - curve ] , we choose the cross - sections so that the @xmath639-boundary is in fact formed by @xmath639-curves .
we call _ vertical strip _ of @xmath578 the image @xmath651 for any compact subinterval @xmath55 , where @xmath652 is the coordinate system on @xmath578 as in remark [ r.foliated ] .
notice that every vertical strip is an @xmath648-adapted cross - section .
.,height=151 ] ( * ? ? ?
* lemma 6.22 , pp 177)[l.existeadaptada ] let @xmath414 be a regular point , that is , such that @xmath653 .
then there exists @xmath654 for which we can find a @xmath648-adapted cross - section @xmath578 at @xmath44 .
given cross - sections @xmath578 and @xmath598 we set @xmath655 the domain of the return map from @xmath578 to @xmath598 .
the next lemma establishes that if the cross - sections are adapted , then we have the invariance property .
* lemma 6.23 , pp 178 ) [ stablereturnmap ] given @xmath654 and @xmath648-adapted cross - sections @xmath578 and @xmath598 , there exists @xmath656 , where @xmath633 is as in proposition [ p.secaohiperbolica ] , such that if @xmath614 defined by @xmath657 is a poincar map with time @xmath658 , then 1 .
@xmath659 for every @xmath617 , and also 2 .
@xmath660 for every @xmath180 , @xmath661 and @xmath617 .
this lemma provides a sufficient condition for having partial hyperbolicity for the poincar return map . indeed ,
if @xmath662 , then the stable leaves are sent strictly inside stable leaves and uniformly contracted by the ratio @xmath663 ; and the unstable cones on cross - sections are preserved .
here we consider the poincar maps of the flow near the singularities .
we recall that , since the equilibria @xmath664 in our setting are all lorenz - like , the unstable manifold @xmath665 is one - dimensional , and there is a one - dimensional strong - stable manifold @xmath666 contained in the two - dimensional stable manifold @xmath667 . by the smooth linearization results provided by hartman @xcite in the absence of resonances , orbits of the flow in a small neighborhood @xmath668 of the given equilibrium @xmath75 are solutions of the following linear system , modulo a smooth change of coordinates : @xmath669 with @xmath670 .
more precisely , let us consider and use the following smooth linearization result .
[ thm : smooth - linear ] let @xmath671 be given .
then there exists an integer @xmath672 such that : if @xmath443 is a real non - singular @xmath673 matrix with eigenvalues @xmath674 satisfying @xmath675 and if @xmath676 and @xmath677 , where @xmath678 and @xmath81 is of class @xmath679 for small @xmath680 with @xmath681 ; then there exists a @xmath682 diffeomorphism @xmath446 from a neighborhood of @xmath683 to a neighborhood of @xmath684 such that @xmath685 for all @xmath427 and initial conditions for which the flows @xmath686 and @xmath687 are defined in the corresponding neighborhood of the origin
. see ( * ? ? ?
* theorem 12.1 , p. 257 ) .
we recall that , in general , hyperbolic singularities are only linearizable by an at most hlder homeomorphism according to the standard hartman - grobman theorem @xcite . by theorem [ thm : smooth - linear ] , hence it is enough for us to choose the eigenvalues @xmath688 of @xmath75 satisfying a _ finite set of non - resonance relations _ for a certain @xmath689 and for each singularity @xmath690 in @xmath76 . for this condition
defines an open and dense set in @xmath691 and so all small @xmath3 perturbations @xmath104 of the vector field @xmath1 will have a singularity whose eigenvalues @xmath692 are still in the @xmath61 linearizing region .
we note that in ( [ eq : linearlorenz ] ) @xmath693 corresponds to the strong - stable direction at @xmath75 , @xmath694 to the expanding direction and @xmath695 to the weak - stable direction .
then for some @xmath654 we may choose cross - sections contained in @xmath668 * @xmath696 at points @xmath697 in different components of @xmath698 * @xmath699 at points @xmath700 in different components of @xmath701 and poincar first hitting time maps @xmath702 , where @xmath703 satisfying ( see figure [ fig : singularbox0 ] ) 1 .
every orbit in the attractor passing through a small neighborhood of the equilibrium @xmath75 intersects some of the incoming cross - sections @xmath699 ; 2 .
@xmath704 maps each connected component of @xmath705 diffeomorphically inside a different outgoing cross - section @xmath696 , preserving the corresponding stable foliations .
here we write @xmath706 for the local invariant stable , strong - stable and unstable manifolds of the hyperbolic saddle - type singularity @xmath75 ( see e.g. @xcite ) , so that these invariant manifold extend up to the cross - sections @xmath707 and @xmath708 .
we note that at each flow - box near a singularity there are four cross - sections : two `` ingoing '' @xmath709 and two `` outgoing '' @xmath710 .
cross - sections near a lorenz - like equilibrium.,height=188 ] using @xmath61 linearizing coordinates in a flow - box near a singularity , with the appropriate rescaling , we can assume without loss of generality that , for a small @xmath654 , see figure [ fig : singularbox0 ] @xmath711 then from we can determine the expression of the poincar maps between ingoing and outgoing cross - sections easily @xmath712 the cases corresponding to the other ingoing / outgoing pairs and signs of @xmath713 are similar .
this shows that the map obtained by identifying points with the same @xmath694 coordinate , i.e. , points in the same stable leaf , is simply @xmath714 where @xmath715 .
analogously , the coordinate transverse to the stable leaves transforms according to the map @xmath716 where @xmath717 .
[ rmk : g - bdd - derivative ] here @xmath718 is bounded if , and only if , @xmath719 or , equivalently @xmath720 . in these coordinates
it is easy to see that for points @xmath721 the time @xmath722 taken by the flow starting at @xmath723 to reach one of @xmath708 depends on @xmath693 only and is given by @xmath724 this in particular shows that the return time on a ingoing cross - section near a singularity is constant on stable leaves .
in this section we exibit a global poincar map for the flow near the singular - hyperbolic attractor @xmath76 .
the construction we perform here is slightly different from the one presented at @xcite : we need injetiveness of the poincar return map to prove exact dimensionality of the physical measure . for this
we need to cover the attractor by flow boxes through pairwise disjoint cross - sections and then consider a fixed iterate of the poincar first return map between these cross - sections .
this is the main difference with respect to the usual construction presented elsewhere .
we observe first that by lemma [ l.existeadaptada ] we can take a @xmath725-adapted cross - section at each non - singular point @xmath69 .
we know also that near each singularity @xmath690 of @xmath76 there is a flow - box @xmath726 containing @xmath690 in its interior .
let @xmath727 denote the finite set of equilibria contained in @xmath76 , all of which are lorenz - like .
step 1 : : choose a flow - box @xmath668 near each singularity @xmath728 as explained in section [ sec : poincare - maps - near ] with the extra conditions + 1 . for any pair of distinct @xmath729 the flow - boxes @xmath730 are pairwise disjoint , and 2 .
the smallest time needed for the positive orbit of a point in @xmath707 to reach @xmath708 is bigger than @xmath633 .
+ we note that since we may take @xmath731 arbitrarily close to @xmath75 , these conditions can always be achieved .
we denote by @xmath732 the family of all such cross - sections near the singularities of @xmath76 .
step 2 : : consider the open set @xmath733 and the compact subset @xmath734 of @xmath76 . for any @xmath735
we know that @xmath44 is a regular point .
hence we have a @xmath648-adapted cross - section @xmath736 through @xmath44 .
we consider the @xmath737-flow - box @xmath738 for a given fixed @xmath739 small and @xmath740 .
we note @xmath741 ensures that @xmath742 does not contain any singularity and , in fact , does not intersect any of the cross - sections fixed at step 1 .
+ the collection @xmath743 is an open cover of the compact set @xmath744 .
we fix a finite subcover @xmath745 in what follows and also consider the corresponding finite family of cross - sections @xmath746 .
step 3 : : now we adjust the construction so that the poincar first return time between elements of @xmath747 is bigger than some uniform positive constant . + for any given pair @xmath748 , if we have @xmath749 , then we may assume without loss of generality that the intersection is transversal .
for otherwise , if @xmath750 is the coordinate system of @xmath578 given according to remark [ r.foliated ] , we may find a @xmath103 embedding @xmath751 close enough to @xmath593 so that @xmath752 is a @xmath648-adapted cross - section and there exists @xmath753 such that @xmath754 and both pairs @xmath755 and @xmath756 are transversal . in particular , @xmath757 and @xmath758 must be disjoint because the ambient space is three - dimensional .
+ hence we may assume that the ( transversal ) intersection @xmath759 is formed by finitely many smooth closed curves .
we consider the sub - strip of @xmath578 given by @xmath760 and also the sub - strip of @xmath761 given by @xmath762 see figure [ fig:2cross ] .
according to the definition of @xmath763 there are @xmath764 and @xmath765 such that for each @xmath766 @xmath767 and @xmath768 . since , by theorem [ contributions ] , the stable manifolds in the neighborhood @xmath472 of @xmath76 depend @xmath769-smoothly on the base point @xmath770 is a @xmath648-adapted cross - section whose flow - box with time @xmath771 covers the @xmath737-flow - box of @xmath772 and @xmath773 . + we replace @xmath578 and @xmath774 in @xmath747 by the following strips : the closure of the connected components of @xmath775 ; together with the closure of the connected components of @xmath776 ; and the closure of @xmath777 ; see figure [ fig:2cross ] .
the number of such components is finite and , moreover , their flow boxes with time @xmath771 cover at least the same portion of @xmath76 as the flow - boxes of @xmath578 and @xmath761 .
this procedure ensures that , given any pair @xmath778 in @xmath747 , their interiors do not intersect , and the minimum poincar first return time between these sections is strictly positive . at this point
we redefine @xmath779 .
cross - sections which intersect and their adaptation.,height=188 ] we define @xmath780 the family of all cross - sections chosen in the above steps .
[ rmk : xi - family ] after this construction we note that 1 .
we can ensure that each of the open sets @xmath781 and @xmath782 is contained in the trapping region @xmath472 , which we also assume is a neighborhood of @xmath76 where the extension of the strong - stable foliation is defined ; 2 . given @xmath783 and @xmath784 ,
the poincar first return time for the positive orbit of @xmath44 to reach some cross - section in @xmath81 is strictly positive ; since the number of cross - sections is finite and each cross - section is compact , there exists @xmath785 such that @xmath786 3 . since @xmath76 is an attractor , the omega - limit set @xmath787 of any @xmath788
is contained in @xmath76 .
let us assume that @xmath723 is a regular point .
thus , with the exception of the local stable manifolds of @xmath75 in @xmath789 , a point @xmath790 has regular orbit under the flow which cross some cross - section in @xmath81 in some future time .
therefore , the orbit of @xmath723 , which accumulates in @xmath787 , must cross some cross - section of @xmath81 . [
def : global - poincare1-map ] [ global poincar first return map ] for any point @xmath723 in the interior of the cross - sections in @xmath81 , we consider the first hit at a cross - section from @xmath81 .
this gives the _ global poincar first return map _ @xmath791 and we say that @xmath792 is the _ poincar time _ of @xmath723 .
if the point @xmath723 never returns to one of the cross - sections , then the map @xmath793 is not defined at @xmath723 .
[ rmk : injective0 ] this construction ensures that @xmath793 is injective , since it is a first return map between cross - sections of a flow . in this way
we can not yet ensure that the poincar time is big enough to guarantee hyperbolicity of the return map . to obtain such big enough poincar time we consider an iterate of @xmath793 , as follows . using proposition [ p.secaohiperbolica ] and lemma [ stablereturnmap ] , for the collection @xmath81 of @xmath648-adapted cross - sections , we consider the following threshold time @xmath794 hence , if we choose a big enough iterate @xmath795 of the poincar first return map so that the return time for @xmath796 is bigger than @xmath33 , then the tangent map to @xmath797 is hyperbolic between any pair of cross - sections . [
def : global - poincare - map ] [ global poincar map ] we choose @xmath798 such that @xmath799 and set @xmath800 .
we note that @xmath446 in the definition above is guaranteed to have hyperbolic derivative at every point where it is defined , since we can write @xmath801 if @xmath802 is defined , @xmath803 .
the function @xmath463 is the _ global poincar time _ and ( [ eq : globalptime ] ) shows that both proposition [ p.secaohiperbolica ] and lemma [ stablereturnmap ] simultaneously hold .
in addition , by lemma [ stablereturnmap ] , if @xmath446 is defined for @xmath581 on some @xmath783 , then @xmath446 is defined for every point in @xmath582 . hence _
the domain of @xmath804 consists of strips of @xmath578_. the smoothness of @xmath805 ensures that the strips @xmath806 have non - empty interior in @xmath578 for every @xmath807 .
[ rmk : r - injective ] since @xmath446 is a fixed iterate of the injective map @xmath793 , we see that @xmath446 is also injective .
moreover , by item ( 3 ) of remark [ rmk : xi - family ] , the family of all @xmath808 for @xmath809 covers @xmath578 except the points where @xmath446 is not defined .
the next result shows that , fixing a cross - section @xmath783 , the points where @xmath446 is not defined are contained in finitely many stable leaves .
thus , after remark [ rmk : r - injective ] , the family of all possible strips , defined as in by the set of points @xmath808 which move from @xmath578 to some strip @xmath809 , covers @xmath578 except for finitely many stable leaves @xmath810 .
thus the number of strips in each cross - section is finite .
we note that @xmath446 is locally smooth for all points @xmath784 such that @xmath811 by the tubular flow theorem , ( * ? ? ?
* theorem 1.1 , pp 40 ) , and the smoothness of the flow , where @xmath812 is the union of the interiors of each cross - section of @xmath81 .
let @xmath813 denote the union of all the leaves forming the stable boundary of every cross - section in @xmath81 .
* lemma 6.29 , pp 182 ) [ le : descont ] the subset @xmath814 of points for which @xmath446 is not defined in @xmath815 is contained in the set of points @xmath816 such that : 1 . either @xmath817 is defined and belongs to @xmath813 ; 2 . or there is some time @xmath818 , @xmath819 given at lemma [ stablereturnmap ] , such that @xmath820 for some singularity @xmath75 of @xmath76 and @xmath821 .
moreover this set is contained in a finite number of stable leaves of the cross - sections @xmath783 .
the proof of this lemma depends on the fact that the poincar time is finite for points where @xmath817 is defined , so the other points must be inside the attractor but never cross other cross - sections of @xmath81 , so they must be converging to an equilibrium point of @xmath1 in @xmath76 ; or they will hit the stable boundary of some cross - section of @xmath81 .
let @xmath443 be the finite set of stable leaves of @xmath81 provided by lemma [ le : descont ] together with @xmath813 .
then the complement @xmath822 of this set is formed by finitely many open strips @xmath823 , where @xmath446 _ is smooth , i.e. , of class @xmath61_. each of these strips is then a connected component of the sets @xmath613 for @xmath807 , where @xmath446 is a @xmath61 diffeomorphism .
we may now obtain a crucial property for the construction of the physical measure for singular hyperbolic attractors and to study its properties in what follows .
[ le : global_poincare_time_int ] the global poincar time @xmath463 is integrable with respect to the lebesgue area measure in @xmath81 , induced by the riemannian lebesgue volume form on the manifold . indeed , given @xmath803 , the point @xmath824
is also given by @xmath825 where @xmath826 .
hence @xmath463 is bounded by the sum of at most @xmath827 exit - time functions of flow - boxes of @xmath727 ( all of them integrable with respect to lebesgue measure ) plus the sum of at most @xmath827 bounded poincar first return time functions between cross - sections in @xmath828 , away from singularities .
thus the poincar time of @xmath446 on @xmath81 is lebesgue integrable .
> from now on , @xmath81 is the collection of all strips @xmath808 where the poincar return map is smooth .
we still denote the strips by the letter @xmath578 in what follows .
we choose a @xmath61 @xmath639-curve @xmath829 transverse to @xmath589 in each @xmath783 .
then the projection @xmath830 along leaves of @xmath589 onto @xmath829 is a @xmath103 map , since the stable leaves @xmath582 are defined through every point of @xmath783 and holonomies depend @xmath103 smoothly on the base point .
given a set @xmath137 , @xmath831 means the closure of @xmath137 .
we define @xmath832 as the number of strips is finite , by the properties of @xmath833 obtained earlier , the set @xmath834 is @xmath61-diffeomorphic to a closed interval @xmath8 $ ] with finitely many points @xmath835 removed , and @xmath836 becomes a @xmath3 submersion .
the set @xmath837 is @xmath103-diffeomorphic to a _ non - degenerate closed rectangle _
@xmath838\times [ 0,1]$ ] with finitely many vertical lines @xmath839 removed .
we denote by @xmath840 the @xmath103-diffeomorphism @xmath841 which sends stable leaves to vertical lines and consider the composition map @xmath842 according to lemma [ stablereturnmap ] , proposition [ p.secaohiperbolica ] and corollary [ ccone ] , the poincar map @xmath843 takes stable leaves of @xmath589 inside stable leaves of the same foliation and is @xmath3 piecewise hyperbolic .
in addition , by corollary [ ccone ] , a @xmath639-curve @xmath844 is taken by @xmath446 into a @xmath639-curve @xmath845 in the image cross - section .
we can define unstable cones on @xmath40 using the smoothness of @xmath840 as @xmath846 .
this ensures , in particular , that @xmath639-curves are taken by @xmath847 into @xmath639-curves .
hence , the map @xmath848 can be written as @xmath849 where @xmath850 moreover , by construction , we have that the following hold : 1 . @xmath851 is not defined at a finite number of points @xmath852 , and it is @xmath3 at @xmath853 .
the points @xmath852 correspond either to the projection of a line @xmath854 of points which fall in the stable manifold of an equilibrium @xmath75 , or to the projection of the boundary of a strip @xmath783 .
@xmath855 is not defined at a finite number of vertical lines @xmath856 in @xmath40 , corresponding to @xmath857 , where @xmath858 are as in ( a ) and @xmath53 is @xmath3 restricted to @xmath859 .
the choice of @xmath446 as an iterate of the first return map @xmath793 between the finite family @xmath81 of cross - sections ensures that @xmath2 is injective in the following sense : if @xmath351 then @xmath860)\cap f((c_j , c_{j+1})\times [ 0,1])=\emptyset$ ] .
finally , with the convention @xmath861 and @xmath862 , the restriction of @xmath2 to each strip @xmath863 \subset q$ ] , @xmath864 , is given by @xmath865 we note that lemma [ stablereturnmap](b ) together with the fact that @xmath840 is a @xmath103-diffeomorphism imply [ glipschitz ] the map @xmath866 preserves the vertical foliation @xmath867 of @xmath40 , and @xmath868 is @xmath572-lipschitz with @xmath869 on each leaf @xmath870 .
[ rmk : horiz - cu - curve ] by taking the cross - sections @xmath578 small enough we can ensure that the unstable cone has very small variation along the strip . we can then ensure that all curves which are at a constant distance from @xmath871 along @xmath872 are also @xmath639-curves . since @xmath872 is sent to vertical lines in @xmath40 through @xmath840 , this in turn ensures that all horizontal lines in @xmath40 which do not intersect the vertical lines @xmath873 are @xmath639-curves , because these horizontal lines correspond through @xmath840 to curves which are essentially at a constant distance from @xmath871 along the stable leaves .
as already mentioned , since the flow @xmath585 is @xmath61 , the leaves @xmath582 , @xmath591 , define a @xmath26 foliation @xmath589 of each @xmath783 with a hlder-@xmath3 holonomy ( since the leaves are one - dimensional ) .
these properties taken together with the expansion provided by corollary [ ccone ] imply ( see the proof in ( * ? ? ?
7.3.2 , pp 222 ) ) .
[ fholder ] the one - dimensional map @xmath33 obtained above is in fact a @xmath103 piecewise expanding map such that @xmath874 is @xmath875-hlder , for some @xmath876 , restricted to each @xmath877 .
the uniform expansion is a consequence of the existence of a uniform bound for the angles between @xmath872 and the curves @xmath871 , once we have fixed the set @xmath81 of cross - sections , and our ability to obtain an arbitrarily large expansion rate along the unstable cones if we choose the threshold @xmath878 large enough . as seen in section [ sec : p - bounded - variation ] ( see theorem [ th - keller ] and consequences ) , if @xmath33 is piecewise expanding and @xmath879 has finite universal @xmath102-bounded variation then there is an absolutely continuous invariant measure with @xmath102-bounded variation density .
it is easy to see that if @xmath874 is piecewise @xmath875-hlder for some @xmath880 , then it is of universal @xmath102-bounded variation .
moreover , by taking an iterate @xmath881 of @xmath33 if necessary , we can assume that each ergodic absolutely continuous invariant probability measure for @xmath33 is decomposed into a finite family of of probability measures invariant for @xmath881 and having exponential speed of convergence to equilibrium .
thus , lemma [ fholder ] together with the results from section [ sec : p - bounded - variation ] imply the following result .
[ ftemacim ] the one - dimensional map @xmath33 obtained above has finitely many ergodic physical measures @xmath882 , whose density is a function of @xmath102-bounded variation , and whose ergodic basins cover lebesgue almost all points of @xmath34 .
we recall that , given a @xmath540-invariant borel probability measure @xmath48 with respect to a map @xmath883 on a metric space @xmath1 , we denote by @xmath884 the _ ergodic basin _ of @xmath48 and say that @xmath48 is _ physical _ if the volume of @xmath885 ( or some other natural measure ) is positive . according to standard constructions described in @xcite and (
* section 7.3 , pp .
225 - 235 ) , each absolutely continuous ergodic probability measure @xmath886 for @xmath33 can be lifted to a unique physical ergodic probability measure @xmath887 for the map @xmath2 , in such a way that @xmath888 , where @xmath889 is the projection on the first coordinate and we have @xmath890 .
hence the ergodic basin @xmath891 of @xmath887 is given by @xmath892 and is a finite collection of strips with non - empty interior and the interior of the supports of distinct @xmath893 and @xmath894 are disjoint .
moreover , each probability measure @xmath887 can be lifted to a physical ergodic probability measure @xmath895 for the flow of @xmath1 supported in the attractor @xmath76 ; more on this in subsection [ sec : integr - global - poinca ] . since
a singular - hyperbolic attractor is transitive , that is , it has a dense orbit , it follows that there can be only one such physical measure for the flow in the basin of attraction of @xmath76 ; see ( * ? ? ?
* section 7.3.8 , pp .
234 - 235 ) .
[ funicamedida ] for each absolutely continuous ergodic probability measure @xmath886 for @xmath33 there exists a unique physical ergodic probability measure @xmath887 for the map @xmath2 such that @xmath888 . for each physical measure @xmath894 for @xmath2
there exists a unique ergodic physical measure @xmath896 for @xmath1 .
we note that since @xmath2 is given by a power of the first return map @xmath793 , the uniqueness statement above does not imply uniqueness of the absolutely continuous invariant probability measure @xmath897 for @xmath33 .
we will also need the fact that the entropy @xmath898 of the map @xmath2 with respect to each physical measure @xmath894 is positive .
since we know that @xmath890 and @xmath899 , where @xmath453 is one ergodic absolutely continuous @xmath33-invariant probability measure , we see that @xmath900 , for each @xmath2-invariant ergodic and physical probability measure @xmath894 .
since @xmath901 lemma [ glipschitz ] means that there is @xmath902 such that for all fixed @xmath44 , @xmath903 the form of the singularities ensures the following result .
[ le : domination - derivative ] if for all singularities @xmath75 we have that the eigenvalues @xmath904 satisfy the non - resonance conditions expressed in theorem [ thm : smooth - linear ] , then the map @xmath905 satisfies @xmath461 . from the choices made in section [ sec
: poincare - maps - near ] we see that the expression of the poincar maps between ingoing and outgoing cross - sections implies the following . for a maximal interval @xmath906 , where the first coordinate map @xmath33 is monotonous , corresponding to an ingoing cross - section near a equilibrium point of the flow @xmath907 with the eigenvalue ratios @xmath908 and @xmath909 ,
we have that @xmath910 is given by @xmath911 for the remaining cases , @xmath912 is just a poincar map between tubular neighborhoods of regular points for the flow , whose derivatives are bounded : @xmath913 , @xmath914 and @xmath915 are bounded functions . since there are finitely many such tubular neighborhoods , we let @xmath916 be an upper bound for these derivatives .
let us now consider the estimation of the total variation .
let @xmath917 be a partition of @xmath34 and @xmath918 arbitrary points in @xmath34 .
for a sequence @xmath919 in @xmath906 we consider two cases .
case a : : @xmath920 is the domain corresponding to an ingoing cross - section near an equilibrium point of the flow .
we can write , since @xmath921 @xmath922 where we have assumed that @xmath923 .
the other case , with @xmath924 in the place of @xmath858 , is similar .
case b : : @xmath920 is the domain corresponding to a tubular neighborhood away from a singular flow - box
. we can now write by the mean value theorem @xmath925 for some @xmath926 .
finally , the case @xmath927 can also be bounded @xmath928 we note that the bounds we have obtained do not depend on the choice of the @xmath929 .
since there are finitely many such tubular neighborhoods and flow - boxes to consider , we see that @xmath930 is bounded above by a constant times the number of smooth branches of the first coordinate function @xmath163 .
now we state a straightforward consequence of remark [ rmk : g - bdd - derivative ] . [
le : f - g - bdd - derivative ] if for all singularities @xmath464 we have the eigenvalue relation @xmath465 , then the second coordinate map @xmath53 of @xmath2 has a bounded partial derivative with respect to the first coordinate , i.e. , there exists @xmath331 such that @xmath466 for all @xmath467 .
we just have to observe that in the domains corresponding to ingoing cross - sections near singularities , the expression of the map @xmath11 is given by ( [ eq : expr - g - near - sing ] ) .
thus we can apply remark [ rmk : g - bdd - derivative ] to conclude that @xmath931 is bounded if , and only if , the stated eigenvalue relation holds for the eigenvalues of @xmath932 at the equilibrium point . for the other domains , since @xmath2 is just a poincar map between tubular neighborhoods of regular points for the flow , its partial derivatives are bounded .
now we show that the global poincar time @xmath463 is integrable with respect to the @xmath2-invariant physical measures @xmath894 on @xmath40 , which lifts to the physical measure @xmath935 for the flow on the singular - hyperbolic attractor and itself is a lift of the @xmath33-invariant absolutely continuous probability measure @xmath897 on @xmath34 .
[ pr : tau_muf - int ] the global poincar time @xmath463 is integrable with respect to each @xmath2-invariant physical probability measure @xmath894 .
we recall the main steps of the construction of @xmath38 from @xmath897 .
we denote by @xmath457 the vertical foliation on the square @xmath40 and by @xmath936 the leaves corresponding to discontinuities of the map @xmath33 .
we have already shown that @xmath937 is * _ invariant : _ the image of any @xmath938 distinct from @xmath443 is contained in some element @xmath939 of @xmath937 ; * _ contracting : _ the diameter of @xmath940 goes to zero when @xmath941 , uniformly over all the @xmath938 for which @xmath940 is defined .
let @xmath453 be an absolutely continuous probability measure on @xmath942 invariant under the transformation @xmath33 obtained in subsection [ sec : f ] .
for each bounded function @xmath943 , let @xmath944 and @xmath945 be defined by @xmath946 [ l.fconvergence ] ( * ? ? ? * lemma 7.21 ) given any continuous function @xmath943 , both limits @xmath947 exist , and they coincide .
> from this it is straighforward to prove the following .
* corollaries 7.22 , 7.25 & subsection 7.3.5 ) [ c.poincareinvariant ] there exists a unique probability measure @xmath894 on @xmath40 such that @xmath948 for every continuous function @xmath949 . besides , @xmath894 is invariant under @xmath2 and is @xmath2-ergodic .
in addition , the basin @xmath950 of @xmath38 equals @xmath40 except for a zero lebesgue ( area ) measure subset .
now , for the integrability of @xmath463 with respect to @xmath894 , we use the construction in lemma [ l.fconvergence ] .
we observe first that for each @xmath951 we have a constant @xmath409 such that @xmath952 since , as explained in subsection [ sec : global - poincare - retu ] , the poincar time on points of a ingoing cross - section near a equilibrium point is constant on the stable leaves ; and on other cross - sections is just a uniformly bounded smooth function of @xmath180 .
moreover , because the density @xmath455 is bounded from above ( since it is a function of generalized bounded variation on the interval ; see section [ sec : p - bounded - variation ] ) , by a constant @xmath953 say , we have @xmath954 since , as explained in , the integral with respect to lebesgue ( length ) measure over an ingoing cross - section is finite , and the return time function is globally bounded otherwise .
hence we may define @xmath955 and write @xmath956 for all @xmath957 , @xmath236 and @xmath897-almost every @xmath958 .
thus we obtain @xmath959 and from lemma [ l.fconvergence ] we see that @xmath960 .
this completes the proof of the proposition .
some other integrability or regularity properties will be needed in the sequel .
we obtain them here .
[ pr : logtprimedyg - int ] we have the following properties : 1 .
@xmath961 ; 2 .
@xmath962 ; 3 .
the maps @xmath963 are uniformly equicontinuous for @xmath964 , i.e. , outside the singularities of the map @xmath33 . for the integrability , we repeat the arguments in the proof of proposition [ pr : tau_muf - int ] . indeed , at a maximal interval @xmath906 , where the first coordinate map @xmath33 is monotonous
, corresponding to an ingoing cross - section near a equilibrium point @xmath907 with the eigenvalue ratios @xmath908 and @xmath909 , we have the expression , thus we obtain @xmath965 and @xmath966 with @xmath967 or @xmath968 . for the remaining cases , @xmath912 is just a poincar map between tubular neighborhoods of regular points for the flow bounded away from equilibria , whose derivatives are bounded : @xmath913 and @xmath915 are bounded functions .
in addition , because @xmath969 we have that @xmath970 is also bounded away from zero . since there are finitely many such tubular neighborhoods , we let @xmath916 be an upper bound for these derivatives
. we can now argue exactly as in the proof of proposition [ pr : tau_muf - int ] to conclude that the integrals in items 1 and 2 of the statement are finite , since both @xmath971 and @xmath972 are comparable to the logarithm of the distance to the singular set @xmath973 , and this function in integrable , as explained in .
moreover , since @xmath974 in @xmath975 we also obtain that the integral in item 1 is positive .
finally , on the one hand , we have seen that @xmath976 does not depend on @xmath180 at monotonicity intervals associated to ingoing cross - sections near singularities . on the other hand , on other monotonicity intervals ,
the function @xmath970 is bounded above and also away from zero , and it is moreover a @xmath3 function with bounded derivatives .
this is enough to conclude item 3 of the statement .
the proof is complete .
now we are ready to consider the poincar maps of singular hyperbolic attractors , and use the results of the previous section to deduce exponential decay of correlations for these maps for each one of the physical measures . to apply theorem [ resdue ] we have to consider a suitable seminorm .
we will use @xmath977 as defined at beginning of section [ sec : p - bounded - variation ] ( recall also theorem [ thm : propert - singhyp - attractor ] , item 4 ) .
the following lemma will establish an estimation ( see equation [ square ] ) which will allow to apply theorem [ resdue ] .
[ so_hard]if @xmath2 has the following properties * @xmath978 is of the form @xmath18 and * the map uniformly contracts the vertical leaves * @xmath23 * @xmath979 has @xmath980 , increasing branches on the intervals @xmath981, ...
,@xmath982 , ... , @xmath983 $ ] , if moreover @xmath984 , then there are @xmath257 such that @xmath985 for each @xmath986 . before proving the above lemma
, we need the following [ 4 ] @xmath987 let us fix @xmath180 and set @xmath988 , then @xmath989 and @xmath990 since @xmath991 then @xmath992 for any @xmath993 . by proposition [ cmp ] equation [ 4.6 ]
we then have the required result .
we first remark that @xmath994so it is sufficient to prove that @xmath995 let us fix @xmath996,@xmath997 suppose that there are no @xmath998 between @xmath273 and @xmath999 . since @xmath163 is lipchitz along the @xmath180 direction , by the third item in the assumptions @xmath1000 now suppose that @xmath1001 .
since @xmath1002 is increasing , then @xmath1003 and @xmath1004 hence , putting all togheter one obtains @xmath1005this gives @xmath1006 and the statement is proved . by theorem [ thm : summary - exp - decay ] and equation ,
we obtain exponential decay of correlations with suitable norms for a class of maps containing the ones found as poincar sections of singular hyperbolic flows .
[ restre]let @xmath978 a borel function such that @xmath1007 .
let @xmath48 be an invariant measure for @xmath2 with marginal @xmath1008 on the @xmath44-axis ( which is absolutely continuous and invariant for @xmath24 ) .
let us suppose that * @xmath2 is a contraction on each vertical leaf : @xmath53 is @xmath20-lipschitz in @xmath180 with @xmath181 for each @xmath44 ; * @xmath1009 ; * @xmath979 is piecewise monotonic , with @xmath25 @xmath26 , increasing branches on the intervals @xmath1010, ... ,@xmath1011 , ... , @xmath1012 and @xmath1013 ;
* @xmath31 has finite universal @xmath32bounded variation ; * @xmath1008 is weakly mixing .
then , @xmath2 has exponential decay of correlations with respect to suitable norms : there are @xmath1014 @xmath1015 since in next section we need to use decay of correlation with respect to lipshitz observables , we show the following estimation .
let @xmath269 , then @xmath1016 let @xmath1017 be lipschitz .
by lemma [ 4 ] we have only to prove @xmath1018 , but @xmath1019 altogether we obtain exponential decay of correlations with respect to lipschitz observables , as stated . [ dec - lip ] if @xmath2 satisfies the assumptions of theorem [ restre ] then it has exponential decay of correlations with respect to lipshitz observables : there are @xmath166 such that @xmath1020 [ dec - sec ] we have seen that if we consider a suitable iterate of the first return map on a section of a flow having a singular hyperbolic attractor , then by theorem [ thm : propert - singhyp - attractor ] and remark [ eigen1 ] all the needed assumptions are satisfied .
hence we conclude that this iterate has exponential decay of correlations over lipschitz observables for each of its physical invariant measures .
to establish the logarithm law for a singular hyperbolic system , we need to establish that the local dimension exists at a section of the system ( see proposition [ maine ] ) .
we recall and use a result of steinberger @xcite and prove that for a singular hyperbolic system , under certain general assumptions the local dimension is defined at almost every point .
let us consider a map @xmath1021 , @xmath18 where 1 .
@xmath1022\to [ 0,1]$ ] is piecewise monotonic : there are @xmath1023 $ ] for @xmath1024 with @xmath1025 such that @xmath1026 is continuous and monotone for @xmath1027 .
furthermore , for @xmath1028 , @xmath1029 is @xmath3 and that @xmath1030 holds where @xmath1031\setminus \cup_{0\leq i < n}c_i$ ] .
2 . @xmath1032 is @xmath3 on @xmath1033 $ ] .
furthermore , @xmath1034 , @xmath1035 and @xmath1036 for @xmath1037 $ ] .
@xmath1038)\cap f((c_j , c_{j+1})\times [ 0,1])=\emptyset$ ] for distinct @xmath1039 with @xmath1040 .
now consider the projection @xmath1041 , set @xmath1042 and @xmath1043 .
for @xmath1044 let @xmath1045 be the unique element of @xmath1046 which contains @xmath44 .
we say that @xmath509 _ is a generator _ if the length of the intervals @xmath1045 tends to zero for @xmath1047 for any given @xmath44 . in piecewise expanding maps it is easy to see that @xmath1048 is a generator .
set @xmath1049 the result of steinberger that we shall use is the following ( * ? ? ? * theorem 1 ) [ th : steinberg ] let @xmath2 be a two - dimensional map as above and @xmath38 an ergodic , @xmath2-invariant probability measure on @xmath40 with the entropy @xmath1050 .
suppose @xmath509 is a generator , @xmath1051 and @xmath1052 .
if the maps @xmath1053 are uniformly equicontinuous for @xmath1054 and @xmath325 has finite universal @xmath102- bounded variation , then @xmath1055 for @xmath48-almost all @xmath1056 .
now , in the systems we consider , item ( 3 ) above is satisfied because the map is induced by a first return poincar map induced by a flow ; see remark [ rmk : injective0 ] and definition [ def : global - poincare - map ] in section [ sec : cross - sections - poinc ] .
moreover @xmath1034 in item ( 2 ) above is established at item ( 6 ) of theorem [ thm : propert - singhyp - attractor ] , provided that for all equilibria @xmath464 we have the eigenvalue relation @xmath465 .
let us also observe that , for the first return map @xmath1057 , associated to the singular - hyperbolic flow , the entropy is positive @xmath1058 ; see section [ sec : positive - entropy - two ] .
so , all we need to prove that @xmath1059 is exact dimensional is to verify that @xmath1060 satisfies the hypothesis of theorem [ th : steinberg ] .
however , proposition [ pr : logtprimedyg - int ] provides precisely that for the functions @xmath1061 defined above in ( [ eq : log - psi - vfi ] ) : we have 1 . @xmath1062 ; 2 . @xmath1063 ; and 3 . the maps @xmath1064 are uniformly equicontinuous for @xmath964 ; where @xmath894 is each one of the invariant ergodic srb measure described in section [ sec : lorenzmodel ] .
this all together completes what is necessary to use theorem [ th : steinberg ] , establishing that each @xmath1065 is exact dimensional .
the exact dimensionality of the measure on the section implies the exact dimensionality of the measure @xmath48 on the flow at almost each point , and the dimension satisfies @xmath1066 at almost every point @xmath44 .
in this section we prove the logarithm law for singular hyperbolic flows .
more precisely we prove [ thm : exactdim - loglaw ] let @xmath1067 be a @xmath61 flow having a singular hyperbolic attractor @xmath76 satisfying the nonresonance condition in theorem [ thm : smooth - linear ] and @xmath465 at each fixed point @xmath75 of the flow in @xmath76 .
let us consider its physical invariant measure @xmath48 .
let us consider @xmath42 and the local dimension at @xmath42 @xmath1068 ( which was above proved to exist almost everywhere ) , then for @xmath48 almost every @xmath44 @xmath1069 where @xmath1070 is the time needed for the orbit of @xmath44 to hit the ball @xmath41 as defined in ( [ circlen ] ) .
the remaining part of the section is devoted to the proof of the theorem [ thm : exactdim - loglaw ] .
the strategy is to first establish the law for the poincar map associated to the flow , and then extend it to the flow itself .
the first step is based on a result about discrete time systems which we recall : let @xmath1071 be an ergodic , measure preserving transformation on a metric space . in this setting the following is proved in @xcite ( see also @xcite for a generalization to targets different than balls ) .
[ maine]for each @xmath42 @xmath1072 hold for @xmath1073-almost every @xmath44 . moreover , if the system has super - polynomial decay of correlations under lipschitz observables and @xmath1074 exists , then for @xmath48-almost every @xmath44 it holds @xmath1075 [ log - sec ] we show how this result can be applied to deduce a logarithm law for the poincar map of singular hyperbolic systems as described in theorem [ thm : propert - singhyp - attractor ] , which is a suitable iteration of the first return map and then to the first return map . by proposition
[ dec - lip ] and remark [ dec - sec ] , we know that the poincar map we consider has super - polynomial decay of correlations for lipshitz observables with respect to each physical invariant measure @xmath1076 .
since in the previous section we proved exact dimensionality of those measures , we can apply proposition [ maine ] to our poincar map .
more precisely , to apply this result to the poincar map @xmath2 and establish the logarithm law , suppose @xmath42 is in the basin of @xmath1076 and consider the system @xmath1077 .
now note that since we are dealing with a ratio of logarithms , and always hold , if we establish the logarithm law for some iterate @xmath1078 , then it will hold also for @xmath1079 ; indeed , applying proposition [ maine ] , we know that there is a set @xmath137 with @xmath1080 such that @xmath1081 then we also have @xmath1082 for each @xmath1083 . but this is also true for each @xmath1084 for each @xmath6 , which by ergodicity of the first return map @xmath1079 covers a full measure set for the invariant measure @xmath1085 of the first return map @xmath1079 .
finally we remark that , since each @xmath1065 is exact dimensional , and each @xmath1065 give rise to the ergodic physical measure @xmath48 of the flow by suspension , then the a.e
. local dimension of each @xmath1065 will be the same for each @xmath6 .
in remark [ log - sec ] we showed that , on the section , the logarithm law holds for the first return map .
let us extend this to the flow .
we consider a general measure preserving flow and note that , just like in discrete time systems , one inequality between hitting time behavior and dimension of the measure is valid in general ( see ( * ? ? ?
* remark 2.4 ) ) .
[ easyineqflow ] let us consider a @xmath26 flow preserving the measure @xmath1086 , then for each @xmath1087 where @xmath1088 is defined @xmath1089 hold for amost each @xmath44 .
the other inequality can be established by the behavior of the system on a section , as considered before .
let @xmath578 be a section of a measure preserving flow @xmath1090 .
we will show that , if the flow is ergodic and the return time is integrable , then the hitting time scaling behavior of the flow can be estimated by the one of the system induced on the section . given any @xmath1091 ,
let us denote by @xmath611 the smallest strictly positive time such that @xmath1092 .
we also consider @xmath1093 , the smallest non negative time such that @xmath1094 .
we remark that these two times differ on the section @xmath578 , where @xmath1095 , while @xmath600 is the return time to the section .
we define @xmath1096 as @xmath1097 , the projection on @xmath1098 . as before denote by @xmath1099 the invariant measure for the first return poincar map @xmath1079 which is induced by the invariant measure @xmath48 of the flow .
[ 4log ] let us suppose that the flow @xmath1100 is ergodic and has a transverse section @xmath1098 with an induced first return map @xmath1079 and an ergodic invariant measure @xmath1101 such that @xmath1102 let @xmath1103 and @xmath1104 be balls centered in @xmath42 , with @xmath1105 .
then , there is a full measure set @xmath1106 not depending on @xmath42 , such that if @xmath1107@xmath1108 ( we recall that @xmath1109 is the time needed for the flow to take @xmath44 to @xmath41 and @xmath1110 is the time needed for the induced map @xmath1079 to take @xmath42 to the same set . ) once we have the logarithm law for the first return map , by this proposition we can extend it to the flow .
indeed , on the section there will be a set @xmath1111 of points where the logarithm law is satistied for each @xmath1107 .
let us consider @xmath1112 .
then @xmath1113 ( recall that @xmath51 is the physical measure of the flow ) because the flow is ergodic and , by definition of @xmath137 , for each @xmath1114 there is @xmath1115 such that @xmath1116 , where the constant represent the time which is needed for @xmath44 to arrive in @xmath1117 by the flow , and does not depend on @xmath43 . hence the same logarithm law which is satisfied on @xmath1111 is satisfied on @xmath137 , showing that , for @xmath44 in the full measure subset of @xmath137 @xmath1118 a similar construction can be done if the target point @xmath42 is not on the section , and extend the result to @xmath1119 .
the other inequality is provided by proposition [ easyineqflow ] .
all together this proves the logarithm law for the flow , once it was proved on the section .
this was done in remark [ log - sec ] .
hence we have established the logarithm law for the flow stated in theorem [ thm : exactdim - loglaw ] .
to help the reader , in this section we give a list of notations related to the various norms which are used in the paper and some explanations to explain their role in the construction .
@xmath263 : : the `` regularity '' norm on the base ( in a skew product the norm for which convergence to equilibrium is established for the base transformation ) .
@xmath143 : : norm evaluating the lipschitz constant in the vertical direction .
@xmath235 : : a seminorm on the square having some suitable properties with respect to the previous two above norms ( see theorem [ resdue ] ) .
@xmath1120 : : a sort of bounded variation seminorm on the square ( but looking only on increments on the horizontal direction ) which is an example of @xmath235 ( for the definition , see beginning of section [ sec : p - bounded - variation ] ) .
@xmath1121 : : the sup norm on all points ( not neglecting zero lebesgue measure sets , this norm will be used where we have to consider both lebesgue and the invariant measure , which are singular with respect to each other ) .
@xmath1122 : : the usual @xmath175 norm with respect to lebesgue measure .
@xmath1123 : : the usual @xmath1124 norm with respect to lebesgue measure .
@xmath224 : : the usual @xmath210 norm with respect to lebesgue measure .
@xmath299 : : the generalized bounded variation norm with respect to lebesgue measure ( see section [ sec : p - bounded - variation ] ) .
@xmath1125 : : the usual lipschitz norm .
v. arajo and m. j. pacifico .
, volume 53 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 , heidelberg , 2010 . with a foreword by marcelo viana .
p. hartman .
, volume 38 of _ classics in applied mathematics_. society for industrial and applied mathematics ( siam ) , philadelphia , pa , 2002 . corrected reprint of the second ( 1982 ) edition [ birkhuser , boston , ma ; mr0658490 ( 83e:34002 ) ] , with a foreword by peter bates . | we consider maps preserving a foliation which is uniformly contracting and a one - dimensional associated quotient map having exponential convergence to equilibrium ( iterates of lebesgue measure converge exponentially fast to physical measure ) .
we prove that these maps have exponential decay of correlations over a large class of observables .
we use this result to deduce exponential decay of correlations for suitable poincar maps of a large class of singular hyperbolic flows . from this
we deduce a logarithm law for these flows . |
this section presents the details of the derivation of the effective model , equation ( 3 ) of the main text , from the fundamental model , equation ( 1 ) of the main text .
the exact hamiltonian may be written ( here we neglect the bare electron - electron repulsion for simplicity and denote the original boson operators by @xmath149 ) : @xmath150 with on - site electron density operator @xmath7 .
we treat the quadratic electron - phonon coupling ( third term of equation ) via a squeezing transformation @xmath151 with @xmath152 where @xmath153 are the phonon displacement and momentum operators and @xmath154 . \label{zetadef}\ ] ] denoting the operator creating a squeezed phonon state on site @xmath6 as @xmath155 we obtain @xmath156\left(\beta_{i}^{\dagger}\beta_{i}+\frac{1}{2}\right ) , \label{eq : squeezed}\ ] ] with @xmath157 and @xmath158=\omega_0\sqrt{1 + 2g\hat{n}}. \label{omegaofn}\ ] ] we begin an analysis of equation with the second term , which may be expanded in powers of @xmath4 which is small in the situations of interest here .
expanding equation to @xmath159 , using @xmath160 , noting that @xmath161 and inserting the result into equation we obtain three terms : @xmath162 the first term is the bare phonon hamiltonian , expressed in terms of the squeezed states , the second term is a phonon state - dependent potential felt by the electrons , and the third term is the phonon - induced electron - electron interaction , which as noted is attractive .
we replace the phonon operators in the last term by their expectation value and discuss the middle term below .
we now consider the transformation of the hopping ( @xmath163 ) term . noting that operators on different sites commute and writing the hopping term in the electron occupation number basis we obtain ( here the @xmath164 refer to the occupation numbers in the initial states before hopping and @xmath165 ) @xmath166=-\sum_{ij\sigma}j_{ij } c^\dagger_{i\sigma}c_{j\sigma}e^{i\left(\zeta[n_i+1]-\zeta[n_i]\right ) \mathcal{o}_i}e^{i\left(\zeta[n_j-1]-\zeta[n_j]\right)\mathcal{o}_j}. \label{htilde1}\ ] ] the phonon creation / annihilation operator content of @xmath24 means that hopping processes may involve the creation or annihilation of pairs of phonons .
as is commonly done in the linearly coupled polaron problem we approximate the hopping by its expectation value ( the adequacy of this approximation is addressed in the next section ) .
computing the expectation value exactly is awkward because the commutator @xmath167 $ ] does not commute with @xmath168 or @xmath169 so the exponentials can not easily be disentangled into the product of an exponential of creation operators times an exponential of annihilation operators times a c - number .
we now expand the exponentials , noting that @xmath170-\zeta[n_i]= \pm \frac{g}{2}+\mathcal{o}(g^2\hat{n})$ ] and retaining only terms that give a non - vanishing expectation value in the decohered ( phonon density matrix diagonal in occupation number basis ) state .
the leading term with a non - zero expectation value is then @xmath171 . evaluating the expectation value of this term in the decohered state and assuming all sites are the same on average and re - exponentiating we obtain @xmath172 where in the second equality we assumed that the occupation number eigenvalues are poisson - distributed with mean occupancy @xmath173 .
we see that the expansion parameter is @xmath174 times a favorable numerical factor .
. in order to gain insight into the physics of the model and the limits of validity of the approximations leading to the effective model , in this section we compare the results of numerically exact solutions of the two site version of the fundamental ( equation ( 1 ) of the main text ) and approximate ( equation ( 3 ) of main text ) hamiltonians .
the two - site version of the exact hamiltonian is @xmath175 while the two - site version of the approximate hamiltonian is @xmath176 with @xmath177 given by equation and @xmath178 . in the absence of electron - phonon coupling
the pump pulse used in the experiment of mitrano _ et al .
_ @xcite would produce an equally phased oscillator coherent state on each site .
we therefore integrate equations and forward in time from an initial state taken to be the direct product of the noninteracting ( @xmath179 ) two electron ground state and a phonon coherent state of mean occupancy @xmath180 and identical phase on each site and compute both time - dependent and time - averaged electron and phonon properties .
( as long as @xmath181 and @xmath182 the comparison is insensitive to the precise values of @xmath4 and @xmath15 ) . in all of the plots presented in this section
the unit of time is @xmath183 . using the estimate @xmath33 means that @xmath184 corresponds to @xmath185 . , starting from coherent states with boson expectation value @xmath186 and non - interacting electron ground state for electron - boson coupling @xmath187 ( upper panels ) , @xmath49 ( middle panels ) and @xmath188 ( lower panels ) .
left panels : time dependence of expectation value of position coordinate of the oscillator on site 1 .
middle panels , time dependence of expectation value of square of position coordinate of the oscillator on site 1 .
right panels , left - right correlation function @xmath189 . ]
we begin by using the exact two - site model to study the evolution of the oscillator states from the initial condition of equally phased coherence states .
the left panel of fig .
[ fig : coherence ] shows the time dependence of the expectation value of the position coordinate of one of the two oscillators ( the other is identical ) .
the curves display the expectec free - boson coherent state oscillations @xmath190 t)$ ] but the oscillations decay within about 15 cycles , showing that coherence in the initial state is rapidly lost .
the central panel shows @xmath191 ; we see that the oscillator state converges to one with small fluctuations about a reasonably well defined mean oscillation level , indicating that the oscillators remain in a relatively highly excited state described to good approximation by a time independent distribution with negligible phase coherence between different number states .
finally the right panels show that the intersite correlation , which is perfect in the initial state , decays to a small ( albeit non - zero ) value , about @xmath192 of the mean value of @xmath193 .
the electronic timescales relevant to superconductivity are of the order of hundreds to thousands of @xmath194 , so these plots justify the neglect of phonon coherence ( either between sites or between different occupation numbers on site ) in our approximate model .
the results of the previous subsection establish that the initial phonon coherence decays rapidly , so that in the phonon occupation number basis the density matrix describing the oscillators on one site is approximately diagonal . in this subsection
we examine the time dependence of the diagonal elements of the density matrix in more detail .
the key issue is that the approximate model conserves phonon number on each site , so in the approximate model an initially diagonal ( in the occupation number basis ) density matrix does not change with time .
show the time dependence of the probability @xmath195 of measuring the number eigenstate @xmath196 computed using the exact two - site polaronic model equation with initial state a coherent state with @xmath180 on each site and @xmath197 .
panels @xmath198 show the time average of @xmath199 ( dots ) compared with the poisson distribution ( line ) characterizing the initial state . ] in the initial coherent state the occupation probabilities ( diagonal density matrix elements ) are poisson - distributed ; if the mean boson occupancy is @xmath15 then the occupation probabilities @xmath200 of the @xmath201-phonon occupation number eigenstates are @xmath202
. the left hand panels @xmath203 of fig . [ two_site_comp_phonons ] present the time dependence of @xmath200 obtained by solving the exact model for different values of the dimensionless electron - phonon coupling @xmath4 for @xmath204 .
we see that the diagonal matrix elements of the initial density matrix are to a good approximation preserved over times long compared to the basic electronic hopping scale .
a different perspective on these results is shown in panels @xmath205 , which compare the time averaged @xmath206 to the poisson distribution corresponding to the initial state ( which is preserved under the dynamics defined by the approximate model ) .
these results show that for the values of dimensionless coupling @xmath4 and mean boson occupancy @xmath173 relevant here , the changes of the diagonal density matrix elements from the initial poisson distribution values are not important : approximation of the phonon distribution as an incoherent average over poisson - distributed occupation - number eigenstates is reasonable . in this subsection
we investigate the electronic state .
we begin with the double occupancy operator @xmath207 fig .
[ two_site_comp_electrons ] shows the expectation value of the double occupancy averaged over the time interval from 0 to @xmath208 .
we compare the expectation values obtained from the exact ( black line ) and the approximate ( yellow line ) hamiltonians as a function of phonon frequency at different values of coupling constant @xmath4 .
the exact and approximate theories both produce an enhancement of the double occupancy and agree well even at @xmath188 .
this enhancement of the double occupancy stems from ( a ) the induced disorder ( second term of equation ) and ( b ) the negative interaction induced by the phonons ( third term of equation ) .
the two contributions are comparable in magnitude and we thus conclude that the quadratic electron - phonon coupling indeed produces an attractive interaction as indicated by the effective model .
[ two_site_comp_electrons ] reveals a non - monotonic dependence of the double occupancy on the interaction strength @xmath4 .
our calculation is in essence an interaction quench and the non - monotonic dependence is a known consequence of quench dynamics . in particular , at very large values of the effective interaction it is easy to see that projecting the initial non - interacting state onto the large @xmath209 eigenstates leads to a @xmath210 , while for small enough effective interactions ( @xmath211 , the regime of interest in this paper ) the double occupancy found in the quenched problem agrees well with the one found in the corresponding ground state .
finally , fig .
[ double_occ_wide ] presents a comparison of the double occupancy computed in the exact and effective models , over a wide range of parameters .
bosons on each site using @xmath197 and values of @xmath4 shown . ]
bosons on each site as a function of boson frequency and coupling constant . ] in this subsection we further investigate the adequacy of our approximation for the electronic dynamics by considering the electron spectral function . the computation of response functions of strongly nonequilibrium systems presents deep conceptual issues which are beyond the scope of this paper .
they depend in general on two time arguments : the center of time @xmath212 and relative time @xmath213 . to investigate the approximations used here we consider the proxy spectral function defined by fourier transforming with respect to @xmath213 and simply averaging over @xmath212 .
for the electron addition part of the spectral function we have @xmath214 here @xmath215 labels the exact eigenstates of the two - site hamiltonian and we have decomposed the initial state @xmath216 , which was chosen as above .
we also introduce the energy @xmath217 . in the approximate model ,
we additionally must keep track of the operator transformations , so we have @xmath218 in evaluating equation we expanded @xmath219 to @xmath220 and define @xmath221 and the eigenstates @xmath215 with respect to the approximate hamiltonian . ( a ) and @xmath222 ( b ) and @xmath180 .
we introduce a gaussian broadening of the delta functions @xmath223 with @xmath224 .
@xmath4 is chosen to keep the induced effective interaction @xmath225 constant .
note that properly transforming the electron creation operator @xmath226 to order @xmath4 is crucial here . ]
[ spectral ] shows a comparison of the exact and approximate spectral functions computed for two phonon frequencies @xmath227 ( a ) and @xmath222 ( b ) .
@xmath4 is chosen to keep the induced effective interaction @xmath225 constant in both panels and we used and initial state corresponding to @xmath180 in the coherent state ( exact model ) or decohered state ( approximate ) .
we see that evolving the electron state for fixed phonon distribution , neglecting the inelastic terms in the hopping , produces a very good approximation to the exact spectral function even for relatively low phonon frequency @xmath228 . is a measure of the localization of the corresponding single particle eigenstate with @xmath229 being extended ( localized ) .
we use a cubic lattice of @xmath230 lattice sites and a nearest neighbor hopping of amplitude @xmath177 given by equation as well as a disorder given by the middle term in equation .
we choose @xmath49 as in the main text and vary @xmath15 .
up to @xmath231 disorder has a minor effect on the localization of the states and we therefore disregard its influence in the superconductivity study of the main text ( where @xmath232 per site ) . ]
in this section we analyse the middle term of equation .
the analysis of the time dependence of the oscillator coordinates and electron spectral function shows that the initial coherent state evolves rapidly to a state characterized by a time - independent , incoherent , approximately poissonian distribution of occupation numbers on each site . in this limit
the middle term in equation corresponds to an on - site random potential , and the effect of this randomness on the properties of interest must be considered . in this paper
we focus on s - wave superconductivity , which is not sensitive to moderate disorder , but is suppressed if the disorder becomes strong enough to localize the electrons .
localization can not be assessed in the two - site model considered in previous sections . to investigate localization effects we consider non - interacting electrons on a three dimensional cubic lattice with nearest neighbor hopping @xmath177 with an on - site disorder modeled by the middle term of equation with phonon occupancies chosen from the poisson distribution with mean occupancy @xmath15 ( the non - vanishing intersite correlations shown in fig .
[ fig : coherence ] means that this is a modest overestimate of the true disorder effects ) and coupling strength determined by @xmath4 . to assess localization physics we compute the participation ratio @xcite @xmath233 for each single - particle eigenstate @xmath234 .
extended states are characterized by a partipation ratio close to zero ; localized states by a particpation ratio close to @xmath235 .
[ fig : disorder_panels ] shows the participation ratio computed for @xmath49 ( in accordance with the main text ) and varying @xmath15 .
we find that up to @xmath231 disorder has a minor effect on the localization of the states and we therefore disregard its influence in the superconductivity study of the main text ( where @xmath232 per site ) .
we note , however , that these values of @xmath15 are not far away from the borderline where disorder should become relevant such that it might be possible to study the disorder dominated phase experimentally by increasing the fluence .
figure [ fig : max_gap ] shows the maximum value of the time dependent nonequilibrium gap function , for different values of @xmath4 and mean boson excitation level @xmath15 roughly consistent with the experiment of mitrano _ et al .
_ we see that values of @xmath4 as small as @xmath236 and reasonable levels of pumping can produce significant enhancement of superconductivity . of time - dependent nonequilibrium gap function @xmath237
* calculated as described in the main text .
different colors indicate different temperatures and circles ( squares ) are @xmath49 ( @xmath187 ) . for both values of @xmath4
we show two values of @xmath15 as indicated in the plot .
the other parameters are chosen as in fig . 2 of the main text .
] the timescales of experimental relevance are long compared to the timescales associated with the inverse of the maximum gap or of the variation of the gap , justifying an adiabatic approximation .
while the adiabatic approximation was not used to obtain the results presented for the time dependence of the gap function , it is used in the discussion of the conductivity and therefore we present here an adiabatic analysis . to formulate the adiabatic analysis
we first introduce the time - dependent rotation @xmath238 that diagonalizes @xmath239 at a given instant of time @xmath240 ( we suppress the momentum index for simplicity ) : @xmath241 with @xmath242 in the above , @xmath243 are pauli matrices , @xmath244 is the unit vector in the superconducting ( @xmath245 ) subspace that is parallel to @xmath246\tau_{1}+{\rm im}[\delta(t)]\tau_{2}$ ] and @xmath247 is the rotation angle in nambu space which rotates the instantaneous time - dependent hamiltonian to a diagonal form with eigenvalues @xmath248 writing @xmath249 and inserting the result into the equations of motion ( see methods section of main text ) we obtain @xmath250 in the adiabtic limit we expect @xmath251 as are all time derivatives of @xmath252 one can then write @xmath253 which gives @xmath254 and @xmath255 to this order we have while @xmath258 lies in the @xmath245 plane and rotates as @xmath259 @xmath260 once we obtain @xmath261 we re - exponentiate to preserve the unitary properties of @xmath262 in fig .
[ fig : adiabtic ] we show the comparison between the exact time - dependent gap and the one computed via the adiabtic approximation .
these results indicate that the system is indeed in the adiabtic limit for parameters relevant to the experiment . | * advances in light sources and time resolved spectroscopy have made it possible to excite specific atomic vibrations in solids and to observe the resulting changes in electronic properties but the mechanism by which phonon excitation causes qualitative changes in electronic properties has remained unclear . here
we show that the dominant symmetry - allowed coupling between electron density and dipole active modes implies an electron density - dependent squeezing of the phonon state which provides an attractive contribution to the electron - electron interaction , independent of the sign of the bare electron - phonon coupling and with a magnitude proportional to the degree of laser - induced phonon excitation .
reasonable excitation amplitudes lead to non - negligible attractive interactions that may cause significant transient changes in electronic properties including superconductivity .
the mechanism is generically applicable to a wide range of systems , offering a promising route to manipulating and controlling electronic phase behavior in novel materials . * strong mode - specific excitation of specific atomic vibrations ( phonon modes ) in solids
@xcite has been shown @xcite to drive drastic changes in collective electronic properties .
of the many effects observed , perhaps the most dramatic is the superconducting - like behavior observed at temperatures far above the equilibrium transition temperature in strongly irradiated @xmath0 @xcite and @xmath1 @xcite . because optically addressable phonon modes are dipole active ( odd parity ) zone center vibrations which typically do not couple linearly to local electronic quantities such as the density or orbital occupancy , the mechanism by which phonon excitation changes electronic properties has not been clear , although many interesting proposals have been made @xcite . here
we present a new and general mechanism that naturally explains how optical phonon excitation can lead to changes in electronic properties .
the key idea is that the dominant symmetry - allowed coupling between electron density and dipole active modes is quadratic , implying an electron density - dependent squeezing of the phonon state .
we show that this electron density dependence of the phonon squeezing provides a sizable attractive contribution to the electron - electron interaction .
the attractive contribution is independent of the sign of the bare electron - phonon coupling , has a magnitude proportional to the degree of laser - induced phonon excitation , and is large enough that reasonable values of the excitation amplitude lead to non - negligible attractive interactions .
the mechanism is generically applicable to a wide range of systems , offering a promising route to manipulating and controlling electronic phase behavior . to demonstrate the mechanism we consider a minimal tight - binding model of electrons that can hop between sites of a lattice , are subject to electronic interactions , and
are coupled quadratically to an optical phonon that is excited by an external field .
the dimensionality of the lattice and the precise nature of the bare electronic interactions will not be qualitatively important in what follows , but we will invoke realistic and explicit coupling parameters and dimensionality when making connection to recent experiments .
the optical mode we study can be considered as a proxy for the @xmath2 phonons which are the dominant optically excited modes in the @xmath1 experiments of mitrano _ et al .
_ @xcite and are known to couple quadratically to electrons .
the hamiltonian may be written as ( circumflexes denote quantum operators in cases where the symbol could be confused with its classical meaning ) @xmath3 here @xmath4 is a dimensionless measure of the quadratic electron - phonon coupling , @xmath5 denotes the electron density operator on site @xmath6 , namely @xmath7 , and @xmath8 is the phonon displacement operator on site @xmath6 .
the usual linear electron - phonon coupling term can be trivially incorporated but is suppressed here for clarity : we assume that it acts primarily to modify the strength of the hopping and electron - electron interaction terms in equilibrium , which may then be taken as effective parameters .
the quadratic electron - phonon coupling implies a density - dependent change of the oscillator stiffness @xmath9 ( @xmath10 is required by stability ) implying that the phonon frequency @xmath11 becomes @xmath12 . to see that the electron density dependence of the phonon frequency induces an attractive interaction we note that the effective electron - electron interaction @xmath13 is the energy of a site with two electrons plus the energy of an empty site , minus twice the energy of the singly occupied site : @xmath14 .
thus the interaction energy associated to a site with @xmath15 excited phonons is @xmath16 the downward concavity of the square root means that @xmath17 is positive _ for either sign _ of @xmath4 .
the magnitude of the induced interaction is proportional to the phonon frequency , demonstrating the essentially quantum aspect of the effect and also to the number of excited phonons .
the magnitude of the induced interaction is also proportional to the number of excited phonon quanta , which can be controlled by the pump fluence .
thus , this simple mechanism offers a direct , robust and generic means to controllably change the magnitude and even the sign of the electron - electron interaction by phonon excitation .
to analyse this effect mathematically we note that the quadratic electron - phonon coupling of equation gives rise to an electron density - dependent change of oscillator stiffness without a corresponding change of mass , in other words to a density - dependent _ squeezing _ @xcite of the oscillator states , in contrast to the usual linear ( holstein or frhlich ) electron - phonon coupling which leads to an electron density - dependent _ shift _ of the oscillator equilibrium position . just as the linear coupling
can be treated with a lang - firsov canonical transformation @xcite that shifts the phonon coordinate , the quadratic coupling can be treated with a squeezing transformation @xmath18 that rescales oscillator position @xmath19 and momentum @xmath20 and transforms the hamiltonian as @xmath21 . the squeezing parameter is @xmath22 $ ] .
the details of the transformation are given in the supplementary information .
the transformation of the phonon and electron - electron interaction terms is straightforward . as in the lang - firsov case
the transformation of the hopping @xmath23 generates expressions involving inelastic ( phonon pair creation / annihilation ) processes .
the nonlinear dependence of @xmath24 on the phonon operators means that the standard disentangling @xcite formulas that simplify the lang - firsov case do not apply , while the nonlinear dependence of @xmath24 on electron density means that the transformed hopping depends on the occupancies of the states between which the electron hops .
however the physically relevant situation involves small values of the dimensionless coupling @xmath4 , with interesting behavior occurring when the product of @xmath4 and the number of excited phonons @xmath15 is large enough . in this situation
we find by comparison to the exactly solvable two - site version of the model ( see supplementary information ) that inelastic effects lead primarily to a rapid decoherence of the initially prepared phonon state , so that on experimentally relevant timescales one may consider the phonons to be characterized by a density matrix which is diagonal in the site occupation number with poisson - distributed eigenvalues initially determined by the pump fluence and decaying slowly back to the equilibrium while the electronic physics is described by a renormalized hopping @xmath25 so the effective hamiltonian becomes ( @xmath26 creates an eigenstate of the squeezed phonon hamiltonian ) @xmath27 with @xmath13 given by equation . and square of pump fluence @xmath28 ( equivalently , mean boson occupancy @xmath15 ) , assuming a half filled band .
expected phases include a mott insulator ( mi ) phase occurring at large @xmath29 and small fluence , metallic ( m ) and superconducting ( sc ) phases occurring at small to intermediate @xmath29 and small to intermediate fluence , and anderson / charge density wave insulator phases ( ai / cdw ) occurring at larger fluence . ]
some aspects of the remarkably rich physics of equation are summarized in the phase diagram shown in fig .
[ phase_diagram ] .
@xmath30 is of course only an approximate description of a nonequilibrium situation and the phases shown indicate the qualitative characteristics expected of the intermediate time behavior as seen explicitly in our discussion of superconductivity below .
the on - site phonon number operator @xmath31 is a constant of the motion , so that in the absence of inelastic effects and phonon - phonon coupling ( not included in equation [ happrox ] ) the phonon distribution is fixed by the pump field .
for all but the last term in @xmath30 we may replace the phonon number operator by its average over the phonon distribution function .
the last term expresses the physics that different phonon occupancies on different sites give rise to fluctuations of root mean square magnitude @xmath32 in the on - site potential energy of the electrons , i.e. to an effective disorder .
for simplicity we sketch the phase diagram for a half - filled non - nested band .
the pump pulse determines the initial phonon density matrix and thus the mean value of the boson occupancy .
if the pump fluence is zero ( only thermally excited phonons , occupation number @xmath15 negligibly small at temperatures of order room temperature or less ) , then at large @xmath29 a mott insulating phase occurs ; as @xmath29 decreases the mott insulator gives way to a metal and then at negative bare @xmath29 to a superconductor .
as the fluence increases the effective @xmath29 ( which is @xmath13 ) decreases and the effective disorder also increases . the positive @xmath29 mott insulating phase will cross over to an anderson ( disorder - dominated ) insulator or a metal ( depending on the bare value of @xmath29 ) and , when the @xmath13 becomes negative enough , the material becomes a charge density wave ( interaction dominated ) insulator .
of course it is important to emphasize again that the effective model is an approximate representation of a dynamical physical situation : the phase represented here by the anderson / cdw insulator is actually a slowly fluctuating polaronically trapped state .
similarly if the initial interaction is not too negative ( equilibrium metal ) , phonon excitation produces a superconductor , while if the initial interaction is attractive ( equilibrium superconductor ) , then the strength of the superconductivity is increased . here
it is important to note that s - wave superconductivity is robust to moderate disorder ( anderson theorem ) ; only when the disorder becomes strong enough to localize the electrons will the superconducting state give way to an anderson / cdw phase .
while induced superconductivity is a focus of current experimental attention , the evolution of an initially mott insulating regime will be of interest in the context of optically pumped mott insulators .
we now present physical consequences , focusing on the superconducting regime potentially relevant to recent reports of transient superconductivity in phonon - pumped @xmath1 .
we note that while the simplified model introduced here omits many of the specifics of @xmath1 , including the three - fold orbital degeneracy of the electronic states and the multiplicity of on - ball phonons , the basic physics of electrons subject to a phonon - mediated attractive interaction and a repulsion of electronic origin ( non - negligible but in the end weaker than the phonon attraction ) is agreed to describe the fullerides @xcite . in mapping this material onto our simplified model
we use a one band featureless semielliptical density of states ( reasonably representative of typical densities of states of three dimensional materials ) with a bare hopping parameter @xmath33 corresponding to the @xmath1 bandwidth of approximately @xmath34 reported in @xcite .
we take the phonon frequency to be @xmath35 , the frequency of the @xmath36 mode excited in the experiment and assume that the combination of an initial laser pulse and inelastic scattering places the phonons into a state described by a density matrix which is diagonal in the site and occupation number basis .
the density matrix is characterized by a mean boson number @xmath37 with rise time and maximal value set by the experimental protocol of mitrano _ et al . _ and a phenomenological exponential decay representing the physically present phonon decay processes not included in equation . for simplicity
we treat the electronic physics within a bcs approximation using the time dependent interaction determined by the phonon occupation via equation and assume that the combination of bare electron - electron and linear electron - phonon coupling leads to a net attractive interaction producing superconductivity with an equilibrium transition temperature @xmath38
. we also incorporated a weak phenomenological electronic damping to represent inelastic processes not well captured by the time - dependent bcs approximation . if this damping is not included the model maps onto an integrable model @xcite whose behavior , while of great theoretical interest , is likely wiped out by damping and dephasing in real materials .
estimates for the strength of the quadratic electron - phonon coupling @xmath4 are difficult to make . via the correlation between mode frequencies and doping
, we use equation along with experimental data to estimate @xmath39 to fall in the range @xmath40 ( see methods section ) . as we shall see
, producing effects at temperatures as high as those reported in the experiment requires a @xmath41 within our model , several times larger than that estimated from the phonon frequency shift , but even significantly smaller @xmath4 values would produce a sizable effect with an induced @xmath42 in the range of @xmath43 ( see supplementary information ) .
the crudeness of the estimation itself as well as the fact that our model is a simplified one - band , one mode model , suggests that it is reasonable to allow for a fair degree of latitude in choosing this value .
future work will be devoted to the study of a more realistic multi - band model to assess if a closer connection with the experimental setting can be obtained . a rough estimate of the maximum boson number @xmath15 corresponding to the experimental situation can be made by comparing the harmonic oscillator approximation for the relevant mode ( here the @xmath44 mode ) to the reported mode amplitude of @xmath45 @xmath46 of the equilibrium bond length , yielding @xmath47 per lattice site .
it should be noted that for a realistic anharmonic mode this estimate may be conservative because the level spacing will decrease for larger @xmath15 values .
as shown in the supplementary information , for these values of @xmath15 and the @xmath4 values we use the system is outside of , but close to , the disorder dominated region of the phase diagram . in the following calculations we use a values @xmath48 , @xmath49 and neglect the disorder .
the time dependence of the assumed pump field , interaction and effective hopping are shown in panels ( a ) and ( b ) of fig .
[ fig : gapoft ] .
note that the induced interaction and hopping renormalization are proportional to the time integral of the fluence so that the maximum in the interaction amplitude lags the maximum in the pump field by a time delay related to the pump profile width . as described in the methods section with @xmath50 ( corresponding to an equilibrium superconducting critical temperature @xmath51 ) , @xmath52 , @xmath35 , @xmath49 , peak boson occupancy @xmath53 and temperatures [ @xmath54 @xmath55 , @xmath56 @xmath57 , @xmath58 @xmath59 and @xmath60 @xmath61 .
for panels [ @xmath62 , the blue shaded area indicates the time window over which the induced interaction is strong enough to stabilize superconductivity at this temperature in equilibrium .
the dashed horizontal lines indicate times at which the probe field is applied in the conductivity calculation ( see fig .
[ fig : cond ] ) ] the remaining four panels of fig .
[ fig : gapoft ] show the time dependence of the superconducting gap amplitude , computed from the time - dependent bcs approximation using standard methods ( see supplementary information ) .
we present results for four representative temperatures .
the lowest temperature ( @xmath63 ) illustrates that if the system is initially in the superconducting state , the gap amplitude is dramatically increased by the change in interaction , and follows the time dependence of the interaction closely .
for the three cases in which the equilibrium material is in the normal state we see that the gap is non - zero only over part of the pulse duration , with the time window where the gap is non - zero is decreasing as the temperature increases .
the shaded regions show the time regime in which the effective interaction is strong enough to sustain equilibrium superconductivity at the given temperature .
we see that gap appears significantly after the dynamical interaction becomes strong enough to sustain equilibrium superconductivity .
the delay occurs because the non - superconducting state is a marginally unstable fixed point of the bcs approximation to the hamiltonian , so the initial time dependence of the gap is controlled by an exponential amplification of small fluctuations .
( left column ) and @xmath64 ( right colum ) using the gap profiles @xmath65 of fig .
[ fig : gapoft ] .
the x - axis is scaled with respect to the maximum value of the corresponding gap profiles shown in fig .
[ fig : gapoft ] .
the insets show @xmath66 $ ] for a larger y - scale and a zoom of the low @xmath67 scale .
, title="fig : " ] the optical response is an important probe of transient superconductivity .
a complete discussion presents subtle experimental and theoretical issues which are beyond the scope of this paper . here
we _ define _ the nonequilibrium conductivity @xmath68 via the relation @xmath69 connecting the time dependent current @xmath70 to the time dependent amplitude of a probe electric field @xmath71 .
this definition reduces to the familiar result at equilibrium and corresponds approximately to the quantity presented in the experimental literature .
we assume for simplity that the probe pulse has negligible width , an integrated amplitude @xmath72 and is applied after a delay @xmath73 defined relative to the maximum in the induced interaction ( here set to be @xmath74 ; note that this occurs somewhat after the maximum in the pump field ) .
we fourier transform @xmath70 and define a frequency - dependent nonequilibrium conductivity @xmath75 as the ratio of @xmath76 to @xmath72 .
we compute the conductivity using an adiabatic approximation and the dirty limit ( mattis - bardeen @xcite ) formula ( details are given in the methods section . ) .
representative results are shown in fig .
[ fig : cond ] .
the real part of the conductivity displays a gap - like structure that depends on temperature and on the delay @xmath73 between the pump that establishes the transient superconducting state and the probe electric field .
the imaginary part correspondingly displays an intermediate frequency enhancement .
the subgap conductivity displays oscillations arising physically from the transient nature of the superconducting state ; the real part may even become slightly negative . for frequencies below a scale @xmath77 that depends on the lifetime of the superconducting state the system behaves like a normal conductor , with the imaginary part of the conductivity vanishing linearly and the real part saturating .
the results , including the subgap oscillations and variation with delay between the pump and probe pulses , bear a striking similarity to the measurements reported in @xcite ( the @xmath78 regime has not yet been experimentally resolved ) . in summary ,
we have introduced a robust and general mechanism by which a large amplitude coherent phonon excitation can dramatically change the electronic behavior of a system .
the effect is generically symmetry - allowed and is the dominant coupling mechanism in centro - symmetric structures .
the main requirements are an optically active phonon of a reasonably high frequency with a frequency that depends modestly on the electron concentration ( @xmath79 over the range @xmath80 ) . under these circumstances
a high degree of phonon excitation produces a component to the electron - electron interaction which is always attractive , is proportional to the number of excited phonons and to the bare phonon frequency and for reasonable couplings may be large enough to appreciably affect electronic properties .
we have shown here that for reasonable parameters the mechanism can produce transient high temperature superconductivity , but the applications are expected to extend far beyond this case .
for example , in transition metal oxides or dicalcogenides , the dependence of breathing - mode and jahn - teller phonons on d - shell occupancy implies an excitation - induced negative-@xmath29 that has the potential to tune systems across the mott metal - insulator or charge density wave transition .
the results presented here raise many questions now under active investigation , including the effects of phonon bandwidth and anharmonicity on the evolution of phonon coherence during and after the initial phonon pulse , the consequences of electronic relaxation and the dynamical aspects of the electron - phonon coupling , the implications of electronic orbital degeneracy and multiple phonon modes , and the physics of the quasilocalization induced at larger fluence . realistic _
ab - initio _ calculations of the crucial coupling constant @xmath4 in different materials are also of interest .
_ acknowledgements _ ajm and eyw were supported by the basic energy sciences program of the us department of energy under grant sc-0012592 .
dmk was supported by dfg ke 2115/1 - 1 .
drr was supported by nsf che-1464802 + _ author contributions _ all authors contributed to planning the research , developing the methods interpreting the results and writing the paper .
eyw and dmk performed the numerical calculations .
+ _ competing financial interests : _ the authors declare no competing financial interests .
10 url # 1`#1 ` urlprefix [ 2]#2 [ 2][]#2 , , and .
_ _ * * , ( ) . , , , , and .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * ( ) .
_ et al . _ .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , , and . _ _ ( ) .
_ et al . _ .
_ _ ( ) . and .
_ _ ( ) . and .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) . and
. _ _ * * , ( ) .
_ _ ( , ) . , and .
_ _ * * , ( ) . , and .
_ _ * * , ( ) . and .
_ _ * * , ( ) . and .
_ _ * * , ( ) .
* methods * _ estimation of dimensionless coupling : _ the parameter @xmath4 is estimated from the density dependence of the phonon frequency via @xmath81 with the bare phonon frequency @xmath82 .
it should be noted that this is a crude estimate since experimental doping also induces structural changes , and our model is highly simplified .
measured vibrational spectra show a frequency shift of @xmath83 for the non - symmetric @xmath84 respectively for @xmath85 with @xmath86 @xcite . using our prediction for the frequency shift in equation
we find @xmath87 for non - zero shift values when we associate @xmath86 ( filled three bands ) to @xmath88 ( filled single band ) .
we have also examined the case @xmath89 with similar results .
this estimated value of @xmath39 is @xmath90 times smaller then the value used in the main text @xmath49 , for which our theory stabilizes superconductivity on comparable scales as found in the experiment .
even lower @xmath4 values produce sizable effects .
_ time dependence of boson occupancy _ in the experiment the phonons are excited via a light pulse with the approximate temporal profile @xmath91 with @xmath92 determined from the fluence of @xmath93 given by mitrano et al @xcite , mean frequency @xmath67 tuned to the c@xmath94 @xmath2 optical phonon frequency ( @xmath95 in the notation of the present paper ) and pulse width @xmath96 such that the envelope corresponds to about 50 cycles of the phonon . in our calculations
we used @xmath97 .
the light field is , to good approximation laterally uniform across the sample and couples linearly to the mode amplitude ( the decay across the sample due to the finite optical penetration depth is included in the estimate of the fluence given by mitrano et al ) . in the absence of electron - phonon coupling textbook quantum mechanics
shows that the resulting phonon state would be a coherent state with maximum oscillator amplitude @xmath98 determined by : @xmath99 here we have included a phenomenological damping @xmath100 to account for the leakage of energy out of the phonon modes . using the textbook relation between oscillator amplitude and coherent state and setting @xmath101 equal to the mass of a @xmath102 atom we estimate a maximum of @xmath103 in the experiment of ref .
@xcite per lattice site from the reported mode amplitude of @xmath45 @xmath46 of the equilibrium bond length . because the electron - phonon coupling destroys the coherence between different occupation number states on a time scale comparable to the pump pulse width we approximate the phonon state at all times in terms of a density matrix which is diagonal in the phonon occupation number basis with occupation probabilities corresponding to the projection of the coherent state with oscillation amplitude @xmath104 onto an occupation number diagonal density matrix .
_ transient superconductivity : _
we treat equation in a time - dependent bcs approximation and neglect disorder effects , so the retarded ( = 0 for @xmath105 ) and advanced ( = 0 for @xmath106 ) electrons green s functions are obtained from ( we do not denote the spatial indices explicitly ) @xmath107 where @xmath108 and @xmath109 are matrices in nambu space . within the time dependence bcs approximation the electron distribution does not relax and the keldysh function is @xmath110 with initial distribution parametrized by @xmath111 , which we typically take to be the fermi - dirac distribution ( non = superconducting initial state ) . for a superconducting initial state
we choose @xmath112 , where @xmath113 are the rotations that diagonalize the initial hamiltonian in nambu - space and in @xmath111 we use the superconducting energies .
the time - dependent hamiltonian is @xmath114 where the time dependence of the gap function @xmath65 is computed from @xmath115,\ ] ] with the time dependence of @xmath116 and @xmath29 given by the time - dependent phonon distribution .
we write the time - dependent gap equation as @xmath117 where @xmath118\hat{e}_{1}-{\rm im}[\delta\left(t\right)]\hat{e}_{2}. \label{beffdef}\ ] ] we solve equation by propagating forward in finite time steps @xmath119 .
we additionally introduce a phenomenological decay to represent electronic dephasing and energy loss processes by taking at each time step the weighted average of 99% of the equal time keldysh green s function obtained by equation and 1% of the instantaneous equilibrium keldysh green s function .
this weighting produces a decay towards equilibrium with relaxation rate @xmath120 .
essentially , equation was written down and solved by barankov _
et al . _
@xcite for an undamped integrable system and a piecewise constant @xmath121 .
our solution is slightly different because we employ the @xmath121 implied by the experimental pump pulse and the phenomenological decay and include electronic damping as described above . _
optical conductivity : _ it is beyond the scope of this paper to model the details of pump - probe nonequilibrium conductivity measurements . for the purposes of this paper we assume that a weak applied ( ` probe ' ) electric field @xmath122 produces an electric current @xmath70 which is non - vanishing only at @xmath123 ( recall that the drive pulse excites a zone - center optical phonon which does not directly produce an electronic current ) .
the relationship @xmath124 between the current and the strength @xmath92 of the probe field then _ defines _ the conductivity .
we introduce the center time @xmath125 and relative time @xmath126 and fourier transform the defining relationship to obtain @xmath127 we further suppose that the pump pulse is short on time - scales of interest , and is peaked at a delay time @xmath128 relative to the time at which the interaction @xmath121 is maximal @xmath129 so that @xmath130 and @xmath131 _ defining _ @xmath132 as the right hand side of equation .
we approximate the conductivity @xmath133 as the equilibrium conductivity appropriate to the gap at the average time , i.e. @xmath134 so that equation becomes @xmath135 we evaluate @xmath136 using the fourier transform of the dirty - limit ( mattis - bardeen ) formula ( equations ( 3.9 ) and ( 3.10 ) of @xcite ) .
the computation of the fourier transform of the dirty - limit conductivity is simplified by the observation that the superconducting contribution is a universal function of @xmath137 and @xmath138 as well as by adding and subtracting the normal state conductivity .
* electronic squeezing of pumped phonons : negative @xmath29 and transient superconductivity - supplementary information * dante m. kennes,@xmath139 eli y. wilner,@xmath140 david r. reichman,@xmath141 and andrew j. millis@xmath142 @xmath143 _ department of physics , columbia university , new york , ny , 10027 , usa _ @xmath144 _ department of chemistry , columbia university , new york , ny , 10027 , usa _
+ @xmath145 [email protected] + @xmath146 [email protected] + @xmath147 [email protected] + @xmath148 [email protected] |
one of the main goal of modern cosmology is the determination of the hubble constant to an accuracy of 1 - 2% ( * ? ? ? * and references therein ) .
the distance to the lmc is a critical step in the problem of determining the scale of the universe .
it is , in fact , considered as a benchmark in the determination of distances to other galaxies , being a useful place to compare and test different distance estimators .
therefore , any error in its distance contributes a substantial fraction of the uncertainty in the hubble constant @xcite . in the last two decades
there have been improvements in the determination of the lmc distance both theoretically ( e.g. red clump , tip of red giant branch , model fitting of variable stars light curves ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and observationally ( e.g. parallaxes for galactic cepheids and rr lyrae , eclipsing binaries ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the literature contains a huge number of estimated distance values ranging from 18.10 mag @xcite to 18.80 mag @xcite , a very unsatisfactory situation that indicates the presence of significant systematic errors in most of the methods .
a summary of the results of different methods with the corresponding references can be found in ( * ? ? ? * tab.10 ) . since the discovery by @xcite , the most used absolute calibration of the extragalactic distance scale is based on the period
luminosity relation of lmc cepheids .
unfortunately , this procedure typically gives an indirect measure of the distance , because it calibrates the period
luminosity relation for lmc on the basis of the same relation obtained for galactic cepheids ( see e.g. * ? ? ?
a strong drawback in this procedure is the role played by metallicity in the period luminosity relation . from both the observational and theoretical perspective , some authors argue that this dependence is present ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , even if the size and the sign of the effect is still disputed , while others suggest that the effect is small and perhaps ill defined ( e.g. * ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the hubble space telescope h@xmath1 key project by @xcite
a lmc distance modulus value of [email protected] mag has been adopted to calibrate several secondary distance indicators using the cepheid variables .
the adopted value of the lmc distance by freedman and collaborators can be considered as a converging value for the lmc distance and as the watershed for the dichotomy between the so called `` short distance scale ( values lower than 18.50 mag ) and the ' ' long distance scale
( values larger than 18.50 mag ) . a direct measurement of the distance to the lmc can be obtained by using a version of the baade
wesselink technique called the surface brightness method ( * ? ? ?
* and references therein ) .
this method utilizes radial changes of the stellar surface brightness and the pulsational velocity for the determination of the linear radius and angular diameter of pulsating stars .
@xcite applied the near infrared surface brightness technique to a sample of 13 cepheids in the lmc and obtained a distance modulus of [email protected] mag .
the same method has been applied recently by @xcite who derived the distances of 36 cepheids in the lmc and obtained a distance modulus of [email protected] mag . in the present work we are going to face the problem of the lmc distance by means of the cors version of the baade - wesselink surface brightness technique @xcite , using recent photometric and spectroscopic data of a sample of cepheids observed in the populous cluster ngc 1866 .
this object is one of the few young ( @xmath2 yr * ? ? ?
* ) lmc clusters that are close enough to allow the detailed observation of individual stars ( see e.g. * ? ?
* ; * ? ? ?
* and references therein ) .
moreover , @xcite found that the ngc 1866 cepheids are close to the average lmc cepheid distance , so we will assume the distance to this cluster to be the distance of the lmc itself .
to date it is known that ngc 1866 harbors at least 23 cepheids @xcite , the largest number among all the lmc clusters , and this makes it an excellent system for distance determination .
indeed it provides a sizeable sample of variable stars contained in a limited volume and then all at the same distance .
furthermore , we expect they are characterized by the same chemical composition , age and reddening , so that we can derive their distance with no influence of differences in the above quantities among stars
. the first baade
wesselink distance to ngc 1866 was derived by @xcite , obtaining the radii of seven variables in the field of ngc 1866 and using the b - v color to calibrate the surface brightness .
the final result , [email protected] mag , is affected by a large error , probably due to the used color index , which is not a good surface brightness calibrator @xcite .
@xcite , using the surface brightness modification of the classical baade wesselink method , obtained the distances of four cepheids in the field of ngc 1866 .
they used the v - r color index and the radial velocities obtained from spectroscopic data taken at the las campanas 2.5 m du pont reflector @xcite .
the obtained distance modulus to ngc 1866 was [email protected] mag , considering only three cepheids ( hv 12198 , hv 12199 and hv 12203 ) .
the fourth ,
hv 12204 , was excluded because the authors suspected it was not a member of the cluster . finally , using optical and near infrared data , @xcite derived the distance to ngc 1866 through the surface brightness method applied on five cepheids and found [email protected] mag .
we aim at improving their result by increasing the analyzed sample of cepheids , thanks to new photometric and spectroscopic data , and by relying on an accurate calibration of the surface brightness obtained from grids of theoretical atmospheres .
[ sec - data ] contains a description of the photometric and spectroscopic data used in this work .
the procedures followed to phase the light curve the radial velocity curve and to correct for reddening are described in sec .
[ sec - analysis ] .
the cors baade wesselink method is introduced in sec .
[ sec - theory ] together with the procedure adopted to calibrate the surface brightness function using atmosphere models . sec .
[ sec - radii ] contains the derived cepheid angular diameter and the linear radii .
the distance to ngc 1866 and the comparison with other results from the literature are discussed in sec .
[ sec - dist1866 ] while conclusions are contained in sec .
[ sec - discus ] .
ngc 1866 harbors at least 23 cepheids @xcite . among them
we have selected those stars such that both near infrared and radial velocity data were available .
the final sample consists of 11 cepheids whose data are presented in the next two sections and are resumed in tab .
[ tab - stars ] .
we used photometric data in the ultraviolet u band , optical b , v , i bands and near infrared k band .
the u band data were obtained at cerro tololo inter american observatory ( ctio ) with 0.9 m , 1.5 m and 4.0 m telescopes .
the optical data were taken at vlt with fors1 imager and consist in 69 images in b , 90 images in v and 62 images in i. both ultraviolet and optical data were calibrated in the johnson
cousins photometric system .
these vlt data have already been published in @xcite .
the complete ubvri dataset used in this work , will be published in a forthcoming paper ( musella et al . in preparation ) , including a detailed discussion of the adopted procedure to reduce the data and to face the crowding problem . as for the near infrared k band
, we have used the dataset described in @xcite and @xcite .
furthermore , we used unpublished data collected by j. storm for the cepheids v4 , v7 and v8 , which were calibrated using the data by @xcite as reference .
the data from @xcite were calibrated in the lco photometric system using the relations from @xcite .
k band data from @xcite have been calibrated in the cit photometric system . to compare the photometric data with theoretical models ( see sec .
[ sec - sb ] ) we have transformed the infrared data from cit and lco to saao photometric system through the simple transformation from @xcite : @xmath3 where we have used also the relation @xmath4 from @xcite .
we used the radial velocity data from @xcite , @xcite and @xcite , obtained through the classical cross correlation method . for three cepheids , namely hv 12197 , hv 12199 and we2 ,
we used new data coming from the flames vlt dataset by ( * ? ? ?
* appendix [ app - spec ] ) .
we note that our radial velocities are the first ones ever published for we2 .
the radial velocities for the quoted three cepheids are listed in tab .
[ tab - new - vr ] . [ cols="^,^,^,^",options="header " , ] [ tab - coeff ]
to investigate how the errors in the various parameters , involved in the cors method , influence the linear radius estimation , we have performed monte carlo simulations .
in particular we have considered the uncertainties on the colors ( v - k ) and ( u - b ) , on the reddening e(b - v ) and the error in the phase matching between the photometric and radial velocity curves .
we have run 1000 simulations for each of the cited parameters .
they consist in varying each analyzed parameter using random numbers extracted from a gaussian distribution with an rms equal to the error on the parameter itself .
for all the extracted displacements the radius is recalculated and the rms of the obtained values is assumed as the uncertainty in the radius due to the error on the analyzed parameter . as for the colors ( v - k ) and ( u - b )
we have derived their errors from the scatter around the interpolated light curves .
here we do not give all the derived errors in the colors , but remind only that typical values are @xmath50.03 mag for both the ( v - k ) and ( u - b ) , except for few cases with errors of @xmath50.06 mag because of less sampled light curve in the k and/or u bands . for the three cepheids hv12197 , hv12203 , and we2 ,
an additional systematic error has been considered for the ( u - b ) color , leading to a total uncertainty of the order of 0.085 - 0.10 mag .
the effect of the reddening on the radius estimation has been analyzed by considering the error of 0.02 mag obtained as described in sec .
[ sec - analysis ] , while for the phase matching we have chosen a typical error of 0.01 in phase .
this value has been chosen to be conservative , because the error on the phase matching , due to the uncertainty on the period propagated between the epoch of photometry and that of spectroscopy , is typically smaller .
furthermore , we have taken into account the fact that the star hv 12202 is a spectroscopic binary @xcite and consequently its photometry and radial velocity can be influenced by the flux and the motion of the companion respectively . to be conservative
we have doubled the errors on the photometry and the radial velocity data .
the total effect of the various errors on the radius is always less than 5@xmath6 , with exception of v7 and v8 for which it is @xmath57@xmath6 ( tab .
[ tab - radii ] ) .
monte carlo simulations have been used also to estimate the errors on the surface brightness .
as it depends on the two colors ( v - k ) and ( u - b ) through the equations given in appendix [ app - fit ] , we have run 1000 simulations for each color following the same procedure described before .
the derived error on the surface brightness has been used to calculate the uncertainty on the angular diameter , obtained from the eq.([eq - sv ] ) and reported in tab .
[ tab - radii ] .
alibert , y. , baraffe , i. , hauschildt , p. & allard , f. , 1999 , , 344 , 551 barnes t. g. , iii , 2009 , in guzik j. a. , bradley p. a. , eds , aip conf .
proc . vol . 1170 ,
stellar pulsation : challenges for theory and observation .
phys . , new york , p. 3
benedict , f.g .
, et al . , 2002 , , 123 , 473 benedict , f.g . , et al . , 2007 , , 133 , 1810 benedict , g.f . , mcarthur , b.e . , feast , m.w . , et al . , 2011 , arxiv:1109.5631 bessell , m.s . & brett , j.m . , 1988 , , 100 , 1134 bessell , m.s . , castelli f. & plez , b. , 1998 , , 333 , 231 bonanos , a.z . , castro , n. , macri , l.m .
& kudritzki , r.p . , 2011 , , 729 , 9 bono , g. , castellani , v. & marconi , m. , 2002 , , 565 , l83 bono , g. , caputo , f. , fiorentino , g. , marconi , m. & musella , i. , 2008 , , 684 , 102 bono , g. , caputo , f. , marconi , m. & musella , i. , 2010 , , 715 , 277 brocato , e. , buonanno , r. , castellani , v. & walker , a.r . , 1989 , , 71 , 25 brocato , e. , castellani , v. , di carlo , e. , raimondo , g. & walker , a.r . , 2003 , , 125 , 3111 caccin , r. , onnembo , a. , russo , g. , sollazzo , c. , 1981 , , 97 , 104 cardelli , j.a . , clayton , g.c .
& mathis , j.s . , 1989 , , 345 , 245 carpenter , j.m . , 2001 , , 121 , 2851 ct , p. , welch , d.l . , mateo , m. , fischer , p. & madore , b. , 1991 , , 101 , 5 coulson , i.m . , caldwell , j.a.r .
& gieren , w.p . , 1986 , , 303 , 273 dallora , m. , et al . , 2004 , , 610 , 269 dean , j.f . ,
warren , p.r . & cousins , a.w.j .
, 1978 , , 183 , 569 feast , m.w . &
walker , a.r .
, 1987 , , 25 , 345 feast , m.w .
& , catchpole , r.m .
, 1997 , , 286 , l1 fitzpatrick , e.l . ,
ribas , i. , guinan , e.f . ,
dewarf , l.e . ,
maloney , f.p . &
massa , d. , 2002 , , 564 , 260 fitzpatrick , e.l . ,
ribas , i. , guinan , e.f . ,
maloney , f.p .
& claret , a. , 2003 , , 587 , 685 fouqu , p. , & gieren , w.p . , 1997 , , 320 , 799 freedman , w.l .
1988 , , 326 , 691 freedman , w.l .
, et al . , 2001 , , 553 , 47 freedman , w.l .
& madore , b.f . , 2011 , , 734 , 46 freedman , w.l . ,
madore , b.f . ,
scowcroft , v. , et al .
, 2011 , arxiv:1109.3802 gould , a. & popowski , p. , 1998
, , 433 , 73 gieren , w.p . ,
fouqu , p. , & gomez , m.i . , 1997 , , 488 , 74 gieren , w.p . , storm , j. , barnes iii , t.g . , fouqu , p. , pietrzyski , g. & kienzle , f. , 2005 , , 627 , 224 grocholski , a.j . ,
sarajedini , a. , olsen , k.a.g . ,
tiede , g.p . & mancone , g.l . , 2007 , , 134 , 680 groenewegen , m.a.t .
& salaris , m. , 2003 , , 410 , 887 groenewegen , m.a.t . , & oudmaijer , r.d . ,
2000 , , 356 , 849 groenewegen , m.a.t .
, 2007 , , 474 , 975 keller , s.c . & wood , p.r . , 2002 , , 578 , 144 keller , s.c .
& wood , p.r .
, 2006 , , 642 , 834 kervella , p. , et al . , 2004 , , 428 , 587 koerwer , j.f . , 2009 , , 138 , 1 kurucz , r. l. , 1993 , synthe spectral synthesis programs and line data .
kurucz cd - rom no .
19 , cambridge , mass . : smithsonian astrophysical observatory , 1993 , 13 labhardt , l. , sandage , a. , tammann , g.a . , 1997 , , 322 , 751 leavitt , h. , 1908 , annals of harvard college observatory , vol .
lx , iv macri , l.m . ,
stanek , k.z .
, bersier , d. , greenhill , l. & reid , m. , 2006 , , 652 , 1133 marconi , m. , clementini , g. , 2005 , , 129 , 2257 marconi , m. , musella , i. & fiorentino , g. , 2005 , , 632 , 590 molinaro , r. et al .
, 2011 , , 413 , 942 mould , j.r . , 2000 , , 529 , 786 mucciarelli , a. , et al . , 2011 , , 413 , 837 musella , i. , et al . , 2006 , , 77 , 291 nardetto , n. , et al . , 2009 , , 502 , 951 onnembo , a. , buonaura , b. , caccin , b. , russo , g. & sollazzo , g. , 1985 , , 152 , 349 osterbrock , d. e. , fulbright , j. p. , martel , a. r. , keane , m. j. , trager , s. c. & basri , g. , 1996 , , 108 , 277 persson , s.e . ,
madore , b.f . ,
krzemiski , w. , et al . , 2004 , , 128 , 2239 popowski , p. & gould , a. , 1998 , , 506 , 259 ripepi , v. , barone , f. , milano , f. & russo , g. , 1997 , , 318 , 797 ripepi , v. , russo , g. , bono , g. , & marconi , m. , 2000 , , 354 , 77 romaniello , m. , salaris , m. , cassisi , s. & panagia , n. , 2000 , , 530 , 738 romaniello , m. , primas , f. , mottini , m. , groenewegen , m. , bono , g. & franois , p. , 2005
, , 429 , l37 ruoppo , a. , ripepi , v. , marconi , m. , & russo , g. , 2004 , , 422 , 253 sakai , s. , ferrarese , l. , kennicutt , jr.,r.c .
& saha , a. , 2004 , , 608 , 42 salaris , m. , percival , s. , brocato , e. , raimondo , g. & walker , a.r . , 2003 , , 588 , 801 storm , j. , carney , b.w . ,
gieren , w.p . ,
fouqu , p. , latham , d.w . & fry , a.m. , 2003 , , 415 , 531 storm , j. , et al , 2004 , , 415 , 521 storm , j. , gieren , w.p . ,
fouqu , p. , barnes iii , t.g .
& gmez , m. , 2005 , , 440 , 487 storm , j. , gieren , w. , fouque , p. , et al .
, 2011 , arxiv:1109.2017 storm , j. , gieren , w. , fouque , p. , et al .
, 2011 , arxiv:1109.2016 szabados , l. , 2003 , information bulletin on variable stars , 5394 , 1 tammann , g.a . , sandage , a. & reindl , b. , 2003 , , 404 , 423 testa , v. , et al . , 2007 , , 462 , 599 tonry , j. & davis , m. , 1979 , , 84 , 1511 udalski , a. , 1998 , , 48 , 113 walker , a.r . , raimondo , g. , di carlo , e. , brocato , e. , castellani , v. & hill , v. , 2001 , , 560 , 139 welch , d.l . ,
mateo , m. , ct , p. , fischer , p. & madore , b.f . , 1991 , , 101 , 490 welch , d.l .
& stetson , p.b . , 1993 , , 105 , 1813 wesselink , a.j . , 1946 , , 368 , 91 whitelock , p. & feast , m. , 2000 , , 319 , 759 | we used optical , near infrared photometry and radial velocity data for a sample of 11 cepheids belonging to the young lmc blue populous cluster ngc 1866 to estimate their radii and distances on the basis of the cors baade - wesselink method .
this technique , based on an accurate calibration of the surface brightness as a function of ( u - b ) , ( v - k ) colors , allows us to estimate , simultaneously , the linear radius and the angular diameter of cepheid variables , and consequently to derive their distance . a rigorous error estimate on radius and distances was derived by using monte carlo simulations .
our analysis gives a distance modulus for ngc 1866 of [email protected] mag , which is in agreement with several independent results . |
let @xmath0 be a finitely generated group and let @xmath1 be a subgroup of @xmath0 .
we say that @xmath1 is a co - dimension 1 subgroup if @xmath2 has more than 1 end , where @xmath3 is the cayley graph of @xmath0 .
if @xmath0 splits over @xmath1 then one easily sees that @xmath1 is co - dimension 1 .
the opposite is not true , for example any closed geodesic on a surface group gives a cyclic codimension 1 subgroup of the fundamental group of the surface . on the other hand
only simple closed geodesics correspond to splittings .
the surface example can be generalized to @xmath4 complexes to produce examples of codimension 1 subgroups : if @xmath5 is a finite @xmath4 complex of ( say ) dimension 2 and if @xmath6 is a locally geodesic track on @xmath5 then the subgroup of @xmath7 corresponding to @xmath6 is a codimension 1 free subgroup of @xmath0 .
wise ( @xcite ) has exploited this idea producing codimension 1 subgroups for small cancellation groups . in the setting of small cancellation groups
of course one needs some combinatorial analog for the convexity property of geodesics ( or tracks ) and wise develops such a notion .
pride ( @xcite ) has shown that there are small cancellation groups that have property fa , so such groups have codimension 1 subgroups but do not split .
stallings showed that if a compact set separates the cayley graph of a finitely generated group @xmath0 , into at least two unbounded components , then @xmath0 splits over a finite group .
bowditch ( @xcite ) showed something similar for hyperbolic groups : if the boundary @xmath8 of a 1-ended hyperbolic group @xmath0 has a local cut point , then the group splits over a 2-ended group , unless it is a triangle group .
there have been other generalizations of stallings theorem similar in spirit .
the general idea is that if a ` small ' set ( coarsely ) separates the cayley graph of a group then the group splits over a subgroup quasi - isometric to the ` small set ' . for a precise conjecture
see @xcite .
the main purpose of this paper is to show the limitations of this ` philosophy ' . given any @xmath9 , we produce an example of a hyperbolic group @xmath0 , such that @xmath10 , @xmath8 is separated by a set of dimension 0 ( a cantor set ) and @xmath0 has property @xmath11 ( so it does not split over any subgroup ) .
our example is based on wise s construction which we generalize to the setting of small cancellation theory over free products .
a diagram is a finite connected planar graph .
the faces of a diagram @xmath12 are the closures of the bounded components of @xmath13 . in what follows
we assume always that each interior vertex ( i.e. not on @xmath14 ) of a diagram has degree at least 3 .
we can always achieve this by erasing all interior vertices of degree 2 .
we will need some small cancellation results about diagrams shown by mccammond and wise in @xcite .
for the reader s convenience and also because our setting is slightly different we include these results here .
these results strengthen classical small cancellation results ( see e.g. @xcite ) .
we need some notation : if @xmath12 is a diagram we denote by @xmath14 the boundary of the unbounded component of @xmath13 ( so if @xmath15 is the unbounded component of @xmath13 , @xmath16 ) .
we say that the diagram is _ non singular _ if @xmath14 is homeomorphic to @xmath17 .
we say that an edge of @xmath12 is interior edge if it does not lie on @xmath14 . if @xmath12 is a diagram we denote by @xmath18 respectively the total number of edges , faces and vertices of the diagram .
we denote by @xmath19 respectively the number of edges of the diagram that lie ( do not lie ) on @xmath14 .
we denote by @xmath20 the number of vertices on @xmath14 that lie in exactly one face and by @xmath21 the number of vertices on @xmath14 that lie on more than one face .
we denote by @xmath22 the number of vertices of @xmath12 that do not lie on @xmath14 .
we say that a diagram verifies the @xmath23 condition if every face of the diagram has at least 6 sides .
we have the following version of greedlinger s lemma ( see @xcite ) : [ greed ] let @xmath12 be a non singular diagram which verifies the condition @xmath23
. then @xmath24 .
we have the following inequalities : @xmath25 this is because each face has at least 6 edges and each interior edge lies in at most 2 faces while boundary edges lie in one face .
@xmath26 this is because each edge has at most 2 vertices and each interior edge has degree at least 3 . using euler s formula and the first inequality we obtain : @xmath27 we remark now that @xmath28 substituting @xmath29 above and using the second inequality for @xmath30 we obtain @xmath31 @xmath32 we recall some definitions from @xcite : let @xmath12 be a non singular diagram . a face @xmath33 of the diagram
is called an @xmath34-spur if the intersection of @xmath33 with the boundary of @xmath12 is connected and exactly @xmath34-edges of @xmath33 are interior edges of @xmath12 .
we say that a diagram @xmath12 is a ladder if there are at most two faces @xmath35 of @xmath12 such that @xmath36 are connected while for every other face @xmath33 of @xmath12 , @xmath37 has exactly 2 components .
we have the following corollary from lemma [ greed ] : [ ladder ] let @xmath12 be a non singular disc diagram which is @xmath23 and contains no @xmath38-spurs and at most two @xmath34-spurs for @xmath39 . then either @xmath12 has a single region or it contains exactly two @xmath34-spurs with @xmath39 and it is a ladder .
we modify @xmath12 as follows : if a face @xmath33 of @xmath12 has more than 6 edges and it intersects the boundary we erase successively vertices of @xmath33 that do not lie on any other face till @xmath33 has 6 edges ( or there are no more vertices to erase ) .
let s call @xmath40 the new diagram .
@xmath40 is still a @xmath23 diagram .
@xmath40 contains also no @xmath38-spurs and at most two @xmath34-spurs for @xmath39 .
we consider now a face @xmath33 of @xmath40 that intersects the boundary and we see how it contributes to @xmath41 . if @xmath33 is not an @xmath34-spur then 2 vertices of @xmath33 contribute to @xmath20 while at least 4 vertices of @xmath33 contribute to @xmath21 .
so the total contribution of all such faces to the difference @xmath42 is at most 0 ( note that the contribution is not necessarily negative as we count twice the @xmath21 vertices as they lie in at least 2 faces ) .
the contribution of an @xmath34-spur to the difference @xmath42 is @xmath43 .
since @xmath12 contains no 3-spurs and at most 2 @xmath34-spurs for @xmath39 , the inequality @xmath44 implies that if @xmath40 has more than one face then @xmath40 has exactly two 1-spurs , say @xmath35 .
if we erase @xmath45 we obtain a diagram @xmath46 which is again @xmath23 .
we note that @xmath45 intersects exactly one face of @xmath40 so after erasing it the diagram @xmath46 has still the other 1-spur of @xmath40 and at most one new @xmath34-spur for some @xmath47 . by the inequality @xmath24
again we conclude as before that either @xmath46 has only one face or it has exactly two 1-spurs @xmath48 .
inductively we see that @xmath40 is a ladder hence @xmath12 is also a ladder . we will need a more technical result . if @xmath49 is a vertex in a diagram we denote by @xmath50 the valency of @xmath49 .
the result below will be used to show that small cancellation products of word hyperbolic groups are word hyperbolic . [ isop ] let @xmath12 be a non singular diagram that verifies the condition @xmath51 .
then @xmath52 in particular @xmath53 i.e. @xmath12 satisfies a linear isoperimetric inequality .
we denote by @xmath54 the set of vertices of @xmath12 ( the 0-skeleton ) .
clearly @xmath55 we have also the following inequality : @xmath56 this is because each face has at least 7 edges and each interior edge lies in at most 2 faces while boundary edges lie in one face . using euler s formula and the inequality above
we obtain : @xmath57 since @xmath58 for every @xmath49 in the interior of @xmath12 : @xmath59 by ( 1 ) and ( 2 ) we have @xmath60 since @xmath61 we have @xmath62 and using ( 3 ) @xmath63
small cancellation theory can be developed over free products ( see @xcite ) .
we show in this section that small cancellation products have codimension 1 subgroups .
this generalizes a result of wise ( @xcite ) .
we recall that the free product factors embed in small cancellation products ( @xcite cor .
9.4 , p.278 ) .
osin ( @xcite , lemma 4.4 ) showed that free product factors embed quasi - isometrically in small cancellation products ( this also follows from @xcite ) . for the reader
s convenience we include a proof of this below .
let @xmath64 be a presentation of a group @xmath0 .
we say that @xmath64 is symmetrized if for any @xmath65 all @xmath66 cyclic permutations of @xmath67 are also in @xmath6 and @xmath68 is in @xmath6 too .
we assume that all elements of @xmath6 are reduced words .
if @xmath69 and the words @xmath70 are reduced we call @xmath71 a _ piece _ of the presentation
. let now @xmath64 be a symmetrized presentation .
we have then the following small cancellation conditions : * _ condition _ @xmath72 : if @xmath73 and @xmath74 with @xmath75 reduced word and @xmath71 a piece then @xmath76 . *
_ condition _ @xmath77 : no element of @xmath6 is a product of fewer than @xmath78 pieces . * _ condition _ @xmath79 : if @xmath80 and @xmath81 is a product of @xmath78 pieces then @xmath82 .
wise showed in @xcite that groups that admit a presentation in which all relators have even length and condition @xmath83 is satisfied , have codimension 1 subgroups .
clearly condition @xmath84 is stronger than condition @xmath83 so wise s result holds for these groups too .
we recall now how that the small cancellation conditions can be given for free products too ( @xcite ch v , sec .
let @xmath0 be the free product of the groups @xmath85 .
we say that a word @xmath86 is reduced if each @xmath87 represents an element of one of the @xmath85 and @xmath88 belong to different factors for any @xmath89 .
any element @xmath90 can be represented in a unique way as a reduced word ( normal form of @xmath91 ) . if @xmath92 is the normal form of @xmath91 we define @xmath93 .
if @xmath94 are reduced words we say that the word @xmath95 is _ semi - reduced _ if @xmath96 .
note however that @xmath97 might lie in the same factor .
we say that a word @xmath98 is _ weakly cyclically reduced _ if it is reduced and @xmath99 .
we say that a sequence of words @xmath6 is symmetrized if whenever @xmath73 all weakly cyclically reduced conjugates of @xmath67 and @xmath68 are in @xmath6 .
we say that @xmath71 is a _ piece _ if there are distinct @xmath100 such that @xmath101 and the words @xmath102 are semi - reduced . as before we have the condition @xmath72 : _ condition _ @xmath72 : if @xmath73 and @xmath74 with @xmath75 semi - reduced word and @xmath71 a piece then @xmath103 .
let now @xmath33 be a free product @xmath104 and let @xmath6 be a symmetrized subset of @xmath33 .
the group @xmath0 defined by the free product presentation @xmath105 is the quotient @xmath106 where @xmath107 is the normal closure of @xmath6 in @xmath33 .
we show now that if a group @xmath0 has a free product presentation @xmath105 that satisfies the @xmath108 condition then @xmath0 has a codimension 1 subgroup .
we start first by considering van - kampen diagrams over @xmath0 .
we consider a usual presentation of @xmath0 with a set of generators @xmath109 given by the generators of @xmath110 and a set of relators consisting of relators of the @xmath111s together with a set @xmath112 such that @xmath6 is obtained by taking all weak cyclic conjugates of elements in @xmath112 and their inverses .
if @xmath112 is finite we say that @xmath0 has a _ finite free product presentation .
_ let now @xmath113 be a word in @xmath109 representing the identity in @xmath0 .
let @xmath12 be a reduced van - kampen diagram for @xmath113 for the presentation given above .
we remark that if @xmath78 is a simple closed path in the 1-skeleton of @xmath12 such that all edges of @xmath78 lie in a single factor @xmath85 then the word corresponding to @xmath78 represents the identity in @xmath85 ( see @xcite , cor .
call such a simple closed path maximal if there is no other such simple closed path @xmath114 in the interior of @xmath78 .
we modify now the diagram @xmath12 as follows : for each maximal simple closed path @xmath78 we erase all edges of @xmath78 and all edges of @xmath12 inside @xmath78 and we introduce a new vertex @xmath115 which we join with all vertices of @xmath78 .
now each edge @xmath116 of @xmath78 has been replaced by two edges @xmath117 .
we label @xmath117 by elements of @xmath85 so that the product of their labels is equal to the label of @xmath116 .
this is clearly possible since @xmath78 corresponds to the trivial element in @xmath85 .
we are allowed here to label an edge by the identity .
after this operation some of the edges of @xmath12 are subdivided. we subdivide the rest of the edges of @xmath12 so that the labels of the new edges lie in the same factor as the old ones and the product of their labels is equal to the label of the old edge .
we call this diagram van - kampen diagram over the free product .
we remark now that the @xmath72 condition holds for this new diagram , i.e. if @xmath118 are adjacent regions of the diagram then @xmath119 [ codim ] let @xmath0 be a finitely generated group with a finite free product @xmath108 presentation @xmath120 . assume further that all @xmath121 are cyclically reduced words and @xmath122 has even length
. then @xmath0 has a codimension 1 subgroup .
we construct a complex for @xmath0 as usual . if @xmath104 and @xmath123 are complexes with a single vertex @xmath124 such that @xmath125 we take the wedge product of the @xmath126s identifying all @xmath124 s . for each @xmath127
we glue a 2-cell to @xmath128 in the obvious way to obtain a complex @xmath129 such that @xmath130 .
we argue now in a way similar to wise ( @xcite ) .
we slightly change approach and we consider bouquets of circles that go through @xmath131 rather than tracks .
we explain now how we construct a bouquet of circles @xmath132 which will give the codimension 1 subgroup .
let @xmath133 be the normal form in @xmath33 of @xmath127 .
we represent the 2-cell @xmath134 corresponding to @xmath67 as a polygon where the @xmath135 are the labels of the sides of this polygon .
the bouquet of circles has a single vertex @xmath131 and a set of edges corresponding to ` diagonals ' of these polygons .
we fix @xmath127 as above we pick the diagonal joining the beginning of the @xmath136-edge to the vertex opposite to it , i.e. the end of the @xmath137 edge .
we remark now that since @xmath134 has an even number of sides each vertex has a vertex opposite to it , so we associate to this vertex the diagonal joining it with the opposite vertex .
we remark that any vertex is determined by the edges adjacent to it .
for example the beginning of the @xmath136 edge is the vertex corresponding to the consecutive edges @xmath138 .
we consider now the equivalence relation on vertices of the @xmath139 generated by the following relation : the vertex @xmath140 of @xmath141 is equivalent to the vertex @xmath142 of @xmath143 if @xmath140 lie in the same free factors as @xmath142 .
we remark that @xmath144 might be equal to @xmath145 in this definition . now for each @xmath67 we consider all vertices equivalent to the vertices of the chosen diagonal .
we add to the bouquet of circles all diagonals corresponding to these vertices and we call the graph obtained in this way @xmath132 .
we remark that @xmath132 is a bouquet of circles if we see it as an abstract graph but if we see it as immersed in @xmath129 its edges are likely to intersect each other in the middle point of the polygons .
@xmath132 corresponds to a subgroup of @xmath0 .
indeed each diagonal gives a generator , for example the diagonal joining @xmath146 gives the generator @xmath147 .
let s call this subgroup @xmath1 .
we will show that @xmath1 is a codimension 1 subgroup of @xmath0 .
there is a tree @xmath148 which has as edges diagonals of 2-cells which is invariant under @xmath1 .
@xmath149 separates @xmath150 in at least 2 components .
let @xmath151 be a vertex .
we define now a connected graph in @xmath150 as follows : we say that two vertices are related if they are opposite .
we take the equivalence relation generated by this relation and we consider the equivalence class of @xmath49 .
let @xmath149 be the graph obtained by joining opposite vertices in this equivalence class by diagonals .
we claim that @xmath149 is a tree .
if it is not a tree then there is a path @xmath78 in @xmath149 such that both endpoints of @xmath78 lie on the same 2-cell of @xmath112 and @xmath78 is not a single edge ( a diagonal ) .
let s say @xmath152 are the endpoints of @xmath78 and they lie on a 2-cell @xmath153 .
let @xmath114 be a path on @xmath154 joining @xmath152 .
we may assume @xmath114 to have minimal normal form length in the free product among the 2 possible paths .
now @xmath155 is a closed loop .
we change now @xmath78 by replacing each diagonal by the corresponding path on the boundary on which the diagonal lie .
we note that we have two choices and we replace the diagonals so that the path we obtain by replacing all of them corresponds to a reduced word of @xmath33 . let @xmath156 be the path we obtain in this way .
we may arrange also that @xmath157 is reduced at the vertex @xmath81 ( unless @xmath158 ) .
we consider now the van - kampen diagram over the free product for @xmath157 and we remark that if it has an @xmath34-spur for @xmath39 then the boundary of this @xmath34-spur contains a neighborhood of the vertex @xmath159 . but
this contradicts corollary [ ladder ] .
it follows that @xmath149 is a tree . by construction @xmath149
is invariant under @xmath1 and separates locally ( hence also globally ) @xmath150 .
the first part of the next lemma follows also from work of osin ( @xcite , see also @xcite ) .
we include a proof here for the sake of completeness .
the vertex groups @xmath85 and @xmath1 embed quasi - isometrically in @xmath0 .
@xmath1 is a codimension 1 subgroup of @xmath0 .
let @xmath81 be a geodesic word in the cayley graph of @xmath85 .
we will show that @xmath81 is a quasi - geodesic in @xmath150 .
let @xmath109 be the generating set of @xmath0 and let @xmath160 be the length of a word in @xmath109 .
let @xmath161 we define a new length function @xmath162 for words of @xmath109 : @xmath163 it is clear that an @xmath162-geodesic is a quasi - geodesic .
indeed let @xmath78 be a geodesic in the 1-skeleton of @xmath150 with respect to the length function @xmath162 with the same endpoints as @xmath81 .
we consider the van - kampen diagram over the free product for @xmath164 .
we may assume that @xmath165 is equal to the endpoints of @xmath166 since along the intersection of @xmath166 , @xmath81 is quasi - geodesic .
we remark now that this diagram has at most 2 @xmath34-spurs ( for @xmath39 ) , corresponding to the endpoints of @xmath78 so by corollary [ ladder ] this diagram is a ladder . by considering now the usual van - kampen diagram for @xmath164 we have that @xmath167
so @xmath81 is quasi - geodesic .
we prove now that @xmath1 is quasi - isometrically embedded .
since @xmath1 acts freely on @xmath149 it is enough to show that @xmath168 is quasi - isometrically embedded .
let @xmath78 be a geodesic path in @xmath149 joining two vertices @xmath169 of @xmath168 .
we change @xmath78 by replacing each diagonal by the corresponding path on the boundary of the cell on which lies the diagonal .
we pick this path so that the word of @xmath33 corresponding to the path is reduced .
let @xmath156 be the path we obtain in this way .
let @xmath114 be an @xmath162-geodesic path joining @xmath169 .
as before we may assume that @xmath170 intersect only at their endpoints .
again by the definition of @xmath170 if we consider the van - kampen diagram over the free product for @xmath157 we remark that it has at most 2 @xmath34-spurs for @xmath39 .
hence this diagram is a ladder . by considering the usual van - kampen diagram
we have that @xmath171 . since @xmath114 is a quasi - geodesic we have that @xmath156 is a quasi - geodesic ,
so @xmath1 is quasi - isometrically embedded .
finally we show that @xmath1 is a codimension 1 subgroup .
it suffices to show that @xmath149 separates @xmath150 in at least 2 components which are not contained in a finite neighborhood of @xmath149 . by
its definition @xmath149 separates locally @xmath150 .
since @xmath150 is simply connected , @xmath168 separates @xmath150 .
we introduce now some useful terminology .
let @xmath127 and let @xmath172 be the normal form of @xmath67 in @xmath33 . recall that @xmath67 is cyclically reduced .
let @xmath173 be smallest such that @xmath174 we say then that @xmath175 is a _ piece _ of @xmath67 .
similarly we define pieces of all cyclic permutations of @xmath172 .
let @xmath176 be a 2-cell in @xmath150 which intersects @xmath168 on an edge @xmath116 .
let @xmath169 be the vertices of @xmath116 .
let @xmath172 be the label of @xmath6 starting from @xmath49 and written in free product normal form .
let @xmath177 be the vertex corresponding to the endpoint of a piece @xmath178 of @xmath6 starting at @xmath49 .
we construct a path starting from @xmath49 and lying in the same component of @xmath179 as @xmath177 .
the path starts by @xmath180 . at @xmath177
we continue @xmath180 by a piece @xmath181 of another 2-cell @xmath182 corresponding to @xmath183 .
we pick @xmath184 and so that @xmath185 is reduced in @xmath33 .
we continue inductively in the same way picking each time a new 2-cell and a piece so that the word we obtain is reduced in @xmath33 .
let @xmath186 be the path we obtain after @xmath66 steps . if @xmath187 is the endpoint of @xmath188 we claim that @xmath189 as @xmath190 .
indeed let @xmath114 be a geodesic joining @xmath187 to a closest vertex @xmath191 .
we consider a geodesic @xmath192 in @xmath149 joining @xmath49 to @xmath193 .
we distinguish now two cases .
assume first that @xmath194 does not lie on @xmath195 .
we change @xmath195 by replacing each diagonal by the corresponding path on the boundary on which the diagonal lie to obtain a path @xmath196 .
we make these replacements so that the word of @xmath33 corresponding to the new path is reduced and @xmath197 corresponds also to a reduced word in @xmath33 .
clearly this is possible since we have two choices for replacing each diagonal and the normal form of each starts from a different free factor .
we consider now the loop @xmath198 since @xmath114 is geodesic the van - kampen diagram for free products for this loop has at most 2 @xmath34-spurs with @xmath39 which appear around the endpoints of @xmath114 .
it follows that this diagram is a ladder ( see corollary [ ladder ] ) hence the lengths of @xmath114 and @xmath199 are comparable so the length of @xmath114 goes to infinity as @xmath200 .
we deal now with the second case , i.e. we assume that @xmath194 lies on @xmath195 .
we modify then @xmath188 as follows .
we replace @xmath180 by the path @xmath201 on the boundary of @xmath6 joining @xmath177 to @xmath194 .
we note that the new path @xmath202 might not be reduced at the endpoint of @xmath201 .
we replace @xmath195 by a path @xmath203 in the 1-skeleton of @xmath150 as before so that @xmath204 is reduced in the free product @xmath33 .
we remark that the van - kampen diagram over free products for the loop @xmath205 is a ladder in this case too hence the length of @xmath114 goes to infinity as @xmath200 .
similarly we construct we see that the component of @xmath206 that does not contain @xmath49 is not contained in a finite neighborhood of @xmath149 .
it follows that @xmath1 is co - dimension 1 .
given any @xmath9 there is a one - ended hyperbolic group @xmath0 such that let @xmath208 be a torsion free 1-ended hyperbolic group with property @xmath209 and such that @xmath210 ( eg a lattice in @xmath211 ) .
let s say @xmath212 .
we may assume that @xmath213 for any @xmath214 and @xmath215 .
we take now another copy of @xmath208 . for notational convenience
we denote the second copy by @xmath216 and its generators by @xmath217 .
we consider now the free product @xmath218 and we define @xmath0 to be the small cancellation quotient of @xmath218 given by the relations : @xmath219 by theorem [ codim ] of the previous section @xmath0 has a free codimension 1 subgroup @xmath1 . as we showed in the proof of the theorem @xmath1
is quasi - isometrically embedded so a cantor set separates @xmath8 .
we show now that @xmath0 has property @xmath11 ( i.e. it does not split ) .
clearly @xmath0 is not an @xmath220 extension since the abelianization of @xmath208 is trivial , so the abelianization of @xmath0 is trivial .
we show now that @xmath0 does not split as an amalgamated product .
let s say @xmath221 . without loss of generality we may assume that @xmath222 and @xmath223 or @xmath224 .
let @xmath225 be the normal form of @xmath0 in the free product decomposition . by replacing @xmath226 by conjugates
we may assume that either @xmath227 and @xmath228 or @xmath229 .
however we see then that the word @xmath230 is reduced in @xmath231 unless @xmath232 or @xmath233 is in @xmath234 .
as all @xmath235 are equal to the identity this implies that @xmath236 and @xmath216 contained in @xmath237 or @xmath238 and @xmath208 contained in @xmath5 but in both cases , the splitting would be trivial .
we claim finally that @xmath0 is hyperbolic .
indeed this follows by lemma 4.4 of @xcite , and @xcite . for the reader
s convenience we sketch a proof here using lemma [ isop ] .
it is enough to show that @xmath0 satisfies a linear isoperimetric inequality .
let @xmath113 be a word on the generators of @xmath0 and let @xmath12 be a reduced van - kampen diagram for @xmath0 . as we describe in section 3
one obtains from @xmath12 a new diagram , let s say @xmath40 , which is called the diagram for @xmath113 over the free product .
since @xmath226 are hyperbolic they satisfy some isoperimetric inequality of the form @xmath239 for any simple closed path @xmath78 in the cayley graph of @xmath208 or of @xmath216 .
it follows that if @xmath78 is a simple closed path of @xmath12 such that all edges of @xmath78 lie in @xmath208 ( or in @xmath216 ) and if @xmath49 is the vertex of @xmath40 that we obtain by collapsing @xmath78 to a point then @xmath240 where @xmath50 is the degree of @xmath49 .
it follows that @xmath241 the above example shows also that for any @xmath66 there is a finitely presented group @xmath0 with @xmath250 which is separated coarsely by a uniformly embedded set @xmath1 of @xmath251 and which does not split .
this answers a question in @xcite . | we construct hyperbolic groups with the following properties : the boundary of the group has big dimension , it is separated by a cantor set and the group does not split .
this shows that bowditch s theorem that characterizes splittings of hyperbolic groups over 2-ended groups in terms of the boundary can not be extended to splittings over more complicated subgroups . |
nuclear matter calculations in the brueckner approximation @xcite do not consider a possible pairing in the ground state of nuclear matter @xcite , with the exception of refs .
this approach is justified by the fact that the superfluid energy gap is expected to be small in normal density nuclear matter @xcite . when addressing the spectral properties of nucleons in medium , calculations using t - matrix approximation where performed @xcite .
the t - matrix approximation for the two - particle green s functions leads to a much stronger pairing than the brueckner approximation .
the t - matrix calculation and the bcs theory give the same prediction for the critical temperature @xcite .
brueckner type calculations of the effective in - medium nuclear interactions , permit to estimate approximate nuclear matter ground state energy , in - medium nucleon masses , and effective cross sections from the free nucleon - nucleon interaction .
although , discrepancies between different calculations and different parameterization of the nuclear interaction remain , a tremendous improvement over a simple hartree - fock approach is achieved .
it is an interesting question , how to include the pairing force into this picture .
the generalization of the t - matrix nuclear matter calculations below the superfluid transition temperature ( @xmath0 ) , requires by itself a procedure combining the ladder summation of the self - energy diagrams and a possible fermion pairing . on the other hand
, the t - matrix scheme seems to be a natural starting point for the inclusion of superfluidity in nuclear matter calculations @xcite . in this work
we show that such a procedure is indeed possible , merging a t - matrix calculation of the normal fermion self - energy with a mean - field , frequency independent , superfluid energy gap .
the interaction potential , which enters in the superfluid gap equation is the free nucleon - nucleon interaction .
however , the scattering of nucleons and the pseudogap formation above the critical temperature reduces the value of the energy gap and of the critical temperature , as compared to the bcs result .
this can be achieved only using self - consistent off - shell propagators in the t - matrix diagrams @xcite .
in such a way the formation of a pseudogap influences strongly the result for the t - matrix in attractive channels , shifting the appearance of a singularity in the t - matrix at the fermi energy to much lower temperatures .
a consistent treatment of the superfluid transition requires then both the use of the off - shell nucleon propagators ( below and above @xmath0 ) and of the anomalous self - energy ( below @xmath0 ) . to illustrate the formation of the pseudogap above @xmath0
, we present first the results of two calculations above @xmath0 ( sec .
[ pseudosec ] ) , a self - consistent t - matrix resummation , using off - shell propagators and a t - matrix calculation using on - shell nucleon propagators with hartree - fock single - particle energies .
the results of that section indicate the necessity of using off - shell propagators in the t - matrix equation around @xmath0 , rather than a quasi - particle approximation . in section [ pairing ]
we derive the equation for the t - matrix self - energy and the gap equations , in the superfluid phase .
a numerical example of a self - consistent calculation of the superfluid energy gap and of the t - matrix self - energy is presented is section [ super ] .
the nuclear matter can be described as an infinite system of interacting fermions .
we consider a system of neutrons and protons interacting via two - body forces @xmath2 note that we define the energies with respect to the fermi energy ( chemical potential ) @xmath3 ; also for real - time calculations .
this will be helpful when dealing with the anomalous propagator in the superfluid phase . in the following
we shall use equivalently the term zero energy and the fermi energy .
nuclear forces can decomposed in ( jst ) ( total angular - momentum , spin , isospin ) channels @xcite @xmath4 the sum over the partial waves @xmath5 is restricted to @xmath6 for uncoupled states and to @xmath7 for the scattering of triplet states .
one body observables of a systems of fermions at equilibrium can be described by the spectral function @xmath8 where @xmath9 denote the retarded(advanced ) green s function .
we restrict the discussion to symmetric nuclear matter so that the green s function is diagonal in spin - isospin indices .
the t - matrix approximation consist of summing particle - particle and hole - hole ladder diagrams for the self - energy @xcite .
the self - energy for in - medium propagators describes the dressing of the quasi - particles due to scatterings with other particles . in general
one can not approximate the spectral function by a @xmath10-function ( quasi - particle peak ) at some quasi - particle energy @xmath11 .
the particles acquire a width , which enters self - consistently in the equation for the t - matrix .
moreover , more complicated structures in the spectral function seem to appear when the temperature is in a range of few mev above the superfluid transition @xcite . using a quasiparticle ansatz for the spectral - function the authors of ref . @xcite
have shown that the spectral function has two ( or sometimes three ) peaks .
this effect by itself invalidates the quasi - particle approximation for the t - matrix . on the other hand , it is possible to calculate the t - matrix and the corresponding spectral - function self - consistently using off - shell propagators in the ladder diagrams for the self - energy @xcite .
the self - consistent calculation was performed in a simple model with separable interactions in the @xmath12-wave . in the present work we use this simple interaction for numerical examples .
the self - consistent solution encompasses in a consistent way the non - fermi liquid behavior of the spectral functions , and , thus , gives quantitatively different results for the critical temperature and for the temperature where the pseudogap appears , than the quasi - particle approximation .
off - shell propagators were used previously in a nuclear matter calculation by jong and lenske @xcite . the spectral function in that work was approximated as a sum of a quasiparticle peak and a continuum part on the other side of the fermi energy , the contribution to the two particle propagator with two continuum spectral functions was neglected to simplify the numerics .
this neglected part could be calculated using the methods developed in the present work . on the other hand a mixed ansatz using a quasi - particle and a continuum part for the spectral function ,
like the one used in @xcite or similar , is unavoidable at small temperatures .
the authors of ref @xcite use a self - consistent scheme a zero temperature , but do not consider pairing .
however , even if the expected pairing is small it is more natural to use a scheme allowing for the creation of pair condensate , since the normal fermi liquid is unstable .
the t - matrix for a system with a two - body interaction @xmath13 is defined as @xcite : @xmath14 with @xmath15 and @xmath16 is the fermi distribution .
the in - medium t - matrix equation can be decomposed in partial waves if an angle averaged two - particle propagator @xmath17 is used : @xmath18 where we average over the angle between @xmath19 and @xmath20 .
@xmath21 ) .
the imaginary part of the self - energy is defined by the t - matrix in the following way @xmath22 where @xmath23 is the bose distribution .
the real part of the self - energy is the sum @xmath24 of the hartree - fock term @xmath25 and the dispersive contribution to real part of the self - energy @xmath26 the solution of eqs .
( [ teqp ] ) , ( [ imsc ] ) , ( [ hfsc ] ) , ( [ disc ] ) , and ( [ spectral ] ) is obtained by iteration with the constraint @xmath27 where @xmath28 is the assumed density of the nuclear matter .
numerical methods used in the solution of the self - consistent t - matrix equations are presented in an appendix . * * quasi - particle approximation * + the quasi - particle solution is obtained by using mean - field propagators in the t - matrix equation . @xmath29 where @xmath30 is the mean - field energy measured with respect to the fermi energy , and the mean - field quasi - particle energy is of course @xmath31 .
more generally , the self - consistent real part of the self - energy ( [ realpart ] ) can also be used to define the quasi - particle energy .
the density is given by @xmath32 analogously the expressions for the self - energies in the quasi - particle approximation take the form @xmath33 @xmath34 the dispersive contribution to the real part of the self - energy can be obtained from the dispersion relation ( [ disc ] ) .
note that there is no need for an iteration procedure in the solution of the t - matrix equation in the quasi - particle approximation , when using hartree - fock single particle energies , as in this work .
once the mean - field self - energy is calculated , the t - matrix can be obtained as @xmath35 for numerical convenience , we calculate the angle average separately in the numerator and in the denominator of the two - particle propagator in the above equation . the spectral function
is then calculated using eq .
( [ imqp ] ) , the dispersion relation ( [ disc ] ) , and the definition ( [ spectral ] ) .
if the spectral function presents a quasi - particle peak around the energy @xmath36 , the quasi - particle approximation can be trusted .
+ of course an iterative improvement of the single particle energies using the real part of the self - energy obtained in the t - matrix or g - matrix approximation is possible @xcite .
such an iterative scheme requires in principle the introduction of bcs - like single particle energies in the presence of pairing .
this can be calculated in a quasi - particle version of the formalism used in sect .
[ pairing ] .
however the actual implementation of this approach is quite tedious and we restrict ourselves to the hartree - fock single particle energies and to the normal fermi - liquid phase for the the quasi - particle approximation . a common property of all the t - matrix approaches to the nuclear matter is the appearance of strong pairing in channels with attractive interactions . as the temperature is lowered a singularity builds up in the t - matrix at zero total momentum and energy . for the quasi - particle scheme it leads to the thouless criterion for @xmath0 , which is equivalent to weak coupling bcs @xcite .
the inclusion of self - energy corrections in the normal fermion propagators generalizes this criterion , giving more general equations for the critical temperature . in section [ pairing ]
we shall derive equations valid also below @xmath0 .
numerical results presented in this , and subsequent sections were obtained using a yamaguchi separable potential of rank one @xcite .
the momenta of all the nucleons are restricted to @xmath37 mev .
the coupling strengths of the original yamaguchi parameterization are slightly adjusted to reproduce the scattering length in vacuum as obtain from the yamaguchi potential without cutoff in momenta of nucleons @xmath38 with @xmath39 , @xmath40 , and @xmath41 . for large momenta
the effect of the cutoff is significant .
nevertheless , the following calculation is a helpful illustration , presenting all the basic phenomena related to the superfluidity in nuclear matter , and describing qualitatively the behavior around the fermi energy .
the binding energy of the deuteron and the critical temperature ( in the quasi - particle approximation ) are changed within few percent by the introduction of the cutoff .
all the results are calculated for nuclear density @xmath42 . at higher densities
, a resonance appears in the two - particle green s function below the fermi energy .
this happens for temperatures above @xmath0 , according to the thouless criterion .
this statement is strongly dependent on the assumed interaction and the resulting deuteron wave function .
this phenomenon is not related to the main subject of this work .
we postpone the discussion of this topic , using more realistic parameterizations of the momentum distribution of the deuteron , to another work and restrict ourselves to low density nuclear matter in the following .
the yamaguchi potential has also the practical advantage , that the critical temperature and the energy gap are relatively ( somewhat unrealistically ) high @xcite .
the strength of the pairing simplifies the numerics , and is helpful in illustrating qualitatively the physical effects . in ref .
@xcite we have studied the self - consistent t - matrix approximation at normal nuclear density and high temperature .
however we were not able to perform similar calculations at small temperatures ( around and below the critical temperature ) with the limited resolution in energy , because the single - particle width becomes very small at the fermi energy ( see appendix ) .
the spectral function in the quasi - particle approximation has been calculated in several works @xcite , in the t - matrix or brueckner schemes .
calculations using the t - matrix approximation for the fermion self - energy show very strong pairing in channels with attractive interactions . at a critical temperature
a pole appears in the t - matrix . for the interaction here chosen ,
it appears first in the in the @xmath43 channel .
it signals the formation of neutron - proton bound states ( with zero binding energy , since the pole in the t - matrix appears at zero total momentum and at the fermi energy ) @xcite , for even lower temperature it is preferable for the system to condense into nucleon - nucleon pairs instead of remaining in the fermi - liquid phase . at temperatures slightly above @xmath0 ,
the t - matrix is strongly peaked for small values of total momentum and small energy .
it is a precursor of the singularity in the t - matrix at the critical temperature .
one of the consequences of this singularity is a very unusual form of the imaginary part of the self - energy ( fig .
[ gamqp ] ) .
the imaginary part of the self - energy @xmath44 has a peak for @xmath45 .
not surprisingly , the spectral function exhibits also a nontrivial structure , with two or more peaks ( fig .
[ spec51qp ] ) .
the spectral function is not of a fermi - liquid type .
the appearance of two peaks in the spectral function above @xmath0 was broadly discussed in high @xmath0 superconductivity under the name of pseudogap formation @xcite .
the quasi - particle approximation represents only the first iteration in the self - consistent calculation of the nucleon spectral function .
thus , it fails to describe the feedback of the pseudogap on the value of the t - matrix in the vicinity of the fermi energy .
this leads to quantitative difference between the self - consistent and the quasi - particle calculations .
the critical temperature is @xmath46 mev for the self - consistent calculation and @xmath47 mev in the quasi - particle approximation .
the form of the spectral function is different in the two approximations @xcite .
the strong differences between the quasi - particle approximation and the self - consistent solution indicate that for strongly interacting systems ( having bound or almost bound states in vacuum ) the usual approximation of the spectral function by a quasi - particle peak fails already above @xmath0 .
another difference already mentioned in @xcite is that the imaginary part of the self - energy has a singularity in the quasi - particle approximation ; in the self - consistent calculation this singularity is smeared out . in fig .
[ gap51qp ] we plot the position of the maxima of the peaks in the spectral function as function of the position of the quasi - particle pole @xmath48 ; @xmath49 clearly two peaks are visible in the spectral function on two sides of the fermi energy .
the dominant peak follows approximately the position of the quasi - particle pole .
however , a subdominant peak is found on the other side of the fermi energy , recalling the effect of an energy gap in the superfluid phase .
the effect was named a pseudogap , since the energy gap in the spectral function occurs without pairing condensate and the superfluid density is zero .
close to the fermi energy the subdominant peak disappears .
the position of the quasi - particle peak ( solution of eq .
[ sieq ] ) is relatively smooth as a function of momentum , in both the quasi - particle approximation and in the self - consistent solution .
there is no sharp variation of the effective mass near the fermi energy .
the self - consistent solution has a similar behavior around the critical temperature as the quasi - particle approximation : at the critical temperature the t - matrix has a singularity at zero total momentum and energy , and for temperatures above @xmath0 , the t - matrix is strongly peaked for small values of energy and total momentum .
this leads to a pseudogap structure in the spectral function ( fig .
[ spec17 ] ) .
this pseudogap structure is again similar to the superfluid gap .
two peaks of the spectral function are located on both sides of the fermi energy .
the dominant peak ( with the larger weight ) is the one whose energy is the closest to the energy of the quasi - particle pole .
please note that for momenta close to the fermi momentum there is only one peak in the spectral function ( the curve corresponding to @xmath50 mev in fig .
[ spec17 ] ) . in fig .
[ gap17 ] the dispersion of the two peaks of the spectral function is presented .
the dispersion of the maximum of the two peaks is very reminiscent of the superfluid dispersion relations @xmath51 , where @xmath52 is the pseudogap . at the fermi energy
the separation between the two peaks should be equal to twice the value of the pseudogap .
however , close to the fermi energy only one , dominant peak of the spectral function survives .
its width is small close to the fermi energy , and the position of its maximum is given by the energy of the quasi - particle pole @xmath48 .
it is difficult to devise a self - consistent approximate scheme , using a two - pole ansatz for the spectral function .
usually such a procedure would require , when iterating eq .
( [ imsc ] ) , the introduction of additional poles in the spectral function . a self - consistent approximate solution in the pseudogap region
can be obtained only in the pairing approximation @xcite , i.e. using a mean - field propagator on the right - hand side of eq .
( [ imsc ] ) . if additional assumptions are made for the t - matrix in the pseudogap region , a self - consistent approximation scheme with superfluid - like ansatz for the spectral function can be obtained @xcite .
however , the self - consistent solution indicates ( fig .
[ gap17 ] ) , that the pseudogap picture @xcite of single - particle excitations above @xmath0 is not entirely correct .
moreover , the energy of the quasi - particle pole is different from the hartree - fock energy .
the imaginary part of the self - energy and its energy dependence are usually non - negligible . in fig .
[ wag17 ] the integrated weights of the two peaks in the spectral function are presented .
the weight of the peak is obtained by integrating the spectral function on the same side of the fermi energy .
although qualitatively the weight of the two parts of the spectral function follows the superfluid result @xmath53 , we could not fit the weight @xmath54 using this law .
this is not surprising since the two peaks are very broad , and , moreover close to the fermi energy where the effect of the pseudogap should be the clearest , the two peaks merge into one .
the sum of the weights of the two peaks gives of course @xmath55 . the nucleon momentum distribution is presented in fig .
[ fer17 ] .
the solid line was obtained using the self - consistently calculated spectral function @xmath56 the dotted line denotes the result of the quasi - particle approximation for the spectral function @xmath57 where @xmath48 denotes the position of the quasi - particle pole , i.e the solution of eq .
( [ sieq ] ) .
nucleon scattering ( nonzero imaginary part of the self - energy in @xmath58 ) leads to a considerable broadening of the momentum distribution .
this is another indication that the quasi - particle approximation is not reliable for nuclear matter around the critical temperature .
in the superfluid phase the nuclear matter develops a superfluid energy gap and non - zero anomalous fermion propagators . the usual bcs theory is limited to mean - field approximations to the self - energy and to the superfluid gap .
formally t - matrix equations can be written for the normal propagators and anomalous propagators @xcite .
as before , t - matrix enters in the calculation of the self - energy in place of the interaction potential .
but also a generalized t - matrix , between the anomalous propagators enters in the calculation of the anomalous self - energy .
the resulting equations are , however , very difficult and no solution in this scheme has been presented . in the following
we will present the simplest generalization of the t - matrix and bcs schemes , allowing for the calculation of the spectral function in the superfluid phase .
it is known that there is a link between the t - matrix and the bcs theory at the critical temperature .
the critical temperature in the bcs theory is related to the appearance of a singularity in the t - matrix at zero energy @xcite .
it is a manifestation of more general relation between the t - matrix fermion scattering and pairing , and the bcs gap equation .
this relation was observed long time ago by kadanoff and martin @xcite .
they proposed to use one mean - field fermion propagator and one bcs two - pole fermion propagator in the ladder of the t - matrix diagram : @xmath59 where @xmath58 is the spectral function without anomalous self - energy , which we will call in the following normal fermion spectral function , and which takes the form @xmath60 in the quasi - particle approximation , and @xmath61 is the spectral function with anomalous self - energy , which we call full fermion spectral function in the rest of the paper ( fig [ tsfig ] ) . in the mean - field approximation it takes the usual bcs form @xmath62 where @xmath63 using ( [ teqkm ] ) the gap equation can be written as @xmath64 the relation between @xmath65 and @xmath66 is given later .
the limit of the above equation when @xmath67 in the t - matrix equation , gives the thouless criterion for @xmath0 .
the meaning of eq .
( [ intgap ] ) is , however , more general .
it states that the gap equation is equivalent to requiring that the singularity of the t - matrix remains at zero energy and momentum for all temperatures below @xmath0 .
the principle of conserving approximation , in the sense of kadanoff and baym @xcite , requires the use of the normal fermion mean field propagator ( without anomalous self - energy ) on the right hand side of eqs .
( [ imsc ] ) and ( [ hfsc ] ) . in the above equations we have assumed that the ground state is time - reversal invariant .
thus , the full fermion propagator @xmath68 is still diagonal in spin - isospin indices @xcite .
however , the full spectral function and the full propagator depend in general on the direction of the momentum . also the t - matrix would then depend on the direction of the total momentum of the pair . in principle
the pairing approximation of kadanoff and martin can be used to calculate the spectral properties of the nuclear matter in the superfluid phase using the quasi - particle approximation . however , as we have seen , the quasiparticle approximation becomes unreliable in the case of strong attractive interaction between fermions , with bound or almost bound states in vacuum . as discussed in models of high @xmath0 superconductivity , the crossover region between the bcs pairing and the boson condensation ,
can not be described using mean - field fermion propagators for the calculation of the t - matrix .
the same is true in nuclear matter for with the yamaguchi interaction in the t - matrix approximation , because of the pseudogap formation in the spectral function ( sec .
[ pseudosec ] ) . at higher densities , in particular at the saturation density
, the pairing gap is expected to be smaller and the bcs theory is correct .
certainly calculations using more realistic interactions for several densities , including the modification of single - particle energies and finite quasi - particle width , would clarify at which densities , if any , the crossover - like behavior between the bcs pairing and the boson condensation occurs .
one way to generalize the kadanoff - martin approach was proposed by the chicago group @xcite . from the observation that a pseudogap in the spectral function forms above @xmath0 , the authors of refs .
@xcite proposed to use one mean - field and one full propagator in the t - matrix equation .
the full propagator would include self - consistently the normal self - energy in the t - matrix approximation ( defined by ( [ simfqp ] ) and ( [ imqp ] ) ) , and , below @xmath0 , also the anomalous self - energy ( defined by the superfluid gap ( [ intgap ] ) ) . in actual calculations ,
the authors of @xcite used an approximate bcs - like form also the normal self - energy in the vicinity and below @xmath0 .
the spectral function is then given in an analytical form by the superfluid gap and by a pseudogap , defined by the singularity of the t - matrix for small energy and small total momentum .
this approach represents an improvement over the quasi - particle approximation , because it takes into account the pseudogap formation in the calculation of the t - matrix .
also its a very nice illustration of the mechanism of the pseudogap formation and of the two - fermion excitations close to the fermi energy @xcite .
the pseudogap , as well as the superfluid gap , causes the formation of an energy gap in the two - fermion excitations around the fermi energy , very similar to the one postulated in @xcite .
however , the self - consistent spectral function calculated in sec .
[ pseudosecsc ] shows only partly a pseudogap .
the position of the subdominant peak in the spectral function does not follow the relation expected from a pseudogap energy dispersion relation . also close to the fermi energy
only one peak in the spectral function is present .
one could use the full spectral function instead of the pseudogap approximation for one of the propagators in the t - matrix equation .
thus , the correct dispersion of the peaks in the spectral function would be taken into account self - consistently .
however , it turns out that the numerical cost of a calculation in the pairing approximation with one off - shell propagator and of a full self - consistent calculation is similar .
also the use of the mean - field propagator in the calculation of the self - energy ( [ imqp ] ) , as required in the pairing approximation , leads to a singularity in the self - energy at @xmath0 .
no such effect is present in the self - energy from the self - consistent calculation @xcite .
therefore , we shall seek a generalization of the self - consistent t - matrix equations in the superfluid phase , including the off - shell fermion propagators in the t - matrix equation and in the self - energy calculation . since the t - matrix equation in the pairing approximation is related to weak - coupling bcs equation @xcite
, we will use a mean - field approximation for the anomalous self - energy . indeed the superfluid gap equation ( fig [ dgapfig ] ) @xmath69 can be related to the t - matrix equation , using strong coupling bcs equations , with frequency independent gap parameter @xcite ( fig [ dbcsfig ] ) @xmath70 where @xmath68 is the full fermion green s function , including the anomalous self - energy ( @xmath71 is reserved , from now on , to the normal green s function including only the normal self - energy ) .
the anomalous propagator spectral function is @xmath72 using ( [ strongbcs2 ] ) in the gap equation ( [ tengap ] ) , and using the spectral representation of the propagators @xmath71 and @xmath68 one obtains @xmath73 thus , the gap equation for the superfluid gap is equivalent to the condition that @xmath65 is the zero eigenvalue vector of the real part of the inverse t - matrix ( eq . [ intgap ] ) , i.e eqs .
( [ intgap ] ) , ( [ gapzwyk ] ) and ( [ tengap ] ) are equivalent .
however , here the t - matrix is given by the expression ( [ teqkm ] ) , with the spectral functions @xmath58 and @xmath61 being obtained from the solution of ( [ strongbcs ] ) ( @xmath74 , @xmath75 ) . the set of equations ( [ teqkm ] ) , ( [ strongbcs ] ) , and ( [ tengap ] ) , defines a consistent approximation scheme for the nucleon self - energy below @xmath0 . in this approach
the singularity of the t - matrix is always at zero energy and zero total momentum , the same is true in the bcs approach @xcite .
however here we include the contribution of the scattering and resonant pairs to the self - energy , through eq .
( [ imsc ] ) , as well as of the condensed pairs , through the anomalous self - energy . in the region of intermediate coupling strength , between the weak bcs and the bose condensation ,
the contribution of the t - matrix to the normal self - energy is important @xcite , and leads to a pseudogap formation .
there is one more technical obstacle before using this scheme in actual calculations . in the superfluid phase the full propagator
@xmath76 depends on the direction of the momentum @xmath20 , and the t - matrix depends on the direction of the total momentum .
this dependence introduces tremendous difficulties in the numerical solution of the t - matrix ( [ teqkm ] ) and the bcs dyson equation ( [ strongbcs ] ) .
also in the t - matrix calculation we have used partial - wave expansion , enabled by the angle averaging of the two - fermion propagator in the ladder diagrams . on the other hand , the solution of the superfluid gap equation , which is equivalent to the calculation t - matrix equation at zero total momentum , requires no angular averaging of the two propagators in ( [ strongbcs2 ] ) . in the bcs theory an approximate decoupling of different angular momentum gaps
is obtained @xcite . in that case the gap equation in each partial wave approximately decouple , i.e. we can use a different propagator @xmath68 in each of the partial wave gap equations . in nuclear physics
the situation is more complicated due to the coupled partial waves in the t - matrix . if we include the coupled partial waves , we can not use the partial wave decoupling of the gap equations ( e.g. @xmath77 ) . also in the nuclear matter we can in general have two superfluid gaps for the same partial wave but with different total angular momentum and isospin ( e.g. @xmath78 and @xmath79 ) .
moreover , we want to use the angular averaged two - nucleon propagator in the t - matrix equation .
thus , the corresponding gap equations in different channels will be coupled through the dependence of the full fermion propagator @xmath68 on the anomalous self - energy . in practice
only the gap corresponding to one channel would be non - zero , the one with the strongest pairing at a given temperature and density . the t - matrix in other channels
would have the singularity pushed away from the fermi energy due to the energy gap originating form the paired channel .
of course this does not exclude phase transitions between pairing in different channels , when changing the temperature or density . in order to simplify the description of the full fermion propagator
we shall use an angle averaged anomalous self - energy in ( [ strongbcs ] ) .
this approximation simplifies also the gap equation @xcite , in particular the superfluid gap for non - zero angular momentum can be taken as real and independent of the projection of the angular momentum @xcite .
of course such an assumption for the superfluid gap is only an approximation , and in some cases the true ground state of the nuclear matter may be described by more complicated angle dependent anomalous propagators .
the example of the liquid helium teaches us that the structure of the anisotropic superfluid order parameter can be very rich @xcite .
in particular the real order parameter would not lead to the state of lowest energy .
however , it is sufficient for consistent and stable t - matrix calculation of the nuclear matter
. it would be extremely difficult to implement the most general angle dependent superfluid gap and full fermion propagator in the t - matrix calculation . as a result
the self - consistent t - matrix can be decomposed in partial waves @xmath80 the full fermion green s function can be written as @xmath81 where the normal fermion propagator is given by the standard expression and @xmath82 is the angle averaged total energy gap .
the self - energy @xmath83 is given by eqs .
( [ hfsc ] ) and ( [ imsc ] ) , with the normal spectral function @xmath58 on the right hand side .
the set of gap equations in ( jst ) channels takes the form @xmath84 the existence of a non - zero solution of the above equation in a particular channel is equivalent to the presence of a singularity in the t - matrix in that channel at zero total momentum and zero energy . the total
, angle averaged energy gap is given by an incoherent sum of contributions in different channels @xmath85 where in the last equality we have used the fact that only one channel @xmath86 gives nonzero gap . it is the channel which gives the larger value of @xmath82 .
one can notice that using mean - field spectral function in ( [ gapsc ] ) , one recovers the bcs gap equations in ( jst ) channels from @xcite .
the final equation of the calculation scheme is the expression for the nuclear density @xmath87 it is instructive to write explicitly the full fermion spectral function in terms of the energy gap and the self - energy @xmath88 it is clear that in order to find the spectral function for energy @xmath89 , one has to know the self energy for energy @xmath89 and @xmath90 .
therefore , the numerical solution must be performed in a symmetrical interval around the fermi energy ( @xmath91 ) .
we solve numerically the set of equation ( [ tsf ] ) , ( [ imsc ] ) , ( [ as ] ) , ( [ gapsc ] ) , with the constraint ( [ rhosf ] ) for the same interaction as used in sec . [ pseudosec ] .
( [ rhosf ] ) and ( [ gapsc ] ) are solved simultaneously for the superfluid gap @xmath52 and for the chemical potential @xmath3 .
below @xmath0 the gap equation has a nontrivial solution , corresponding to the superfluid phase . for the case of the yamaguchi interaction ,
the pairing occurs always in the @xmath79 channel .
the t - matrix has a singularity at zero momentum and zero energy in this channel , for all temperatures below @xmath0 .
the t - matrix in the @xmath78 channel has no singularity but is strongly peaked near the fermi energy . for more realistic interaction
we expect also pairing in the @xmath92 channel and pairing with higher angular momentum at larger densities .
the calculation gives two spectral functions , the normal spectral function @xmath93 , without the anomalous self - energy , and the full spectral function @xmath94 , obtained from the bcs dyson equation ( [ strongbcs ] ) .
we have observed that due to strong attractive interaction , and due to the appearance of the singularity in the t - matrix , a bcs - like two peak structure appears in the spectral function @xmath58 in the pseudogap region .
the same is true below @xmath0 ( figs .
[ spec14a ] and [ spec14c ] ) .
the spectral function without anomalous self - energy is very similar as above @xmath0 ( fig .
[ spec17 ] ) . indeed the imaginary part of the self - energy ( fig .
[ gam17 ] ) is almost indistinguishable for the two temperatures , except near the fermi energy . in the full spectral function
another mechanism is also responsible for the double peak structure . for all momenta the full spectral function
@xmath61 has two peaks situated on both sides of the fermi energy .
close to the fermi energy , the superfluid gap in the spectral function is clearly visible ( fig [ spec14b ] ) .
far from the fermi energy the modifications of the dominant peak are negligible .
the subdominant peak of the spectral function is only slightly modified by the anomalous self - energy .
this is what is expected form the usual bcs theory , since the weight of the subdominant superfluid peak in the spectral function @xmath95 is becoming small far from the fermi energy .
the dispersion of the maximum of the peaks of the spectral functions @xmath58 and @xmath61 illustrates the formation of the superfluid gap ( fig . [ gap14 ] ) . on the abscissa we plot the energy of the quasi - particle pole in the normal fermion green s function @xmath71 .
this energy coincides approximately with the maximum of the dominant peak in the spectral function @xmath58 .
this is especially well fulfilled in the vicinity of the fermi energy .
the subdominant peak in the spectral function @xmath58 is visible for energies further away from the fermi energy , also its energy does not follow the branch @xmath96 .
like for the case above @xmath0 , the pseudogap in the spectral function is only qualitatively similar to a superfluid gap . close to the fermi energy ,
the spectral function @xmath58 is similar as in the normal fermi liquid . on the other hand , the positions of the peaks of the full spectral function
@xmath61 follow the dispersion relation expected for a superfluid matter , especially close to the fermi energy .
an energy gap is clearly visible for @xmath97 .
away from the fermi energy , the positions of the two peaks approach @xmath98 for the dominant peak and @xmath96 for the subdominant peak in the full fermion spectral function . for @xmath99 mev , the positions of the peaks do not follow the bcs energy branches any more .
however , one can observe that the dominant peaks in the normal fermion spectral function and in the full fermion spectral function are very close away from the fermi energy , as expected . in fig .
[ wag14 ] are presented the weights of the peaks in the spectral functions @xmath58 and @xmath61 as function of the energy of the quasi - particle pole of the spectral function @xmath58 .
the peaks in the two spectral functions have qualitatively similar behavior . however , the weight of the dominant ( subdominant ) peak is smaller ( larger ) in the full spectral function , than in the normal spectral function .
it means that part of the weight of the spectral function is shifted from the dominant to the subdominant peak , as expected from the presence of an additional mechanism of peak doubling in the superfluid phase . the fermion momentum distribution is plotted in fig .
[ fer14 ] . the finite imaginary part of the spectral function modifies strongly the momentum distribution .
the momentum distributions obtained from the spectral functions @xmath58 and @xmath61 are much broader then the one obtained taking the quasi - particle pole approximation for the spectral function . as expected , the nucleon scattering produces a high momentum tail in the distributions .
the effect of the superfluid gap can be seen in the difference between the solid and dashed lines in fig .
[ fer14 ] .
as expected @xcite the full spectral function @xmath61 gives broader momentum distributions .
the bcs broadening with small energy gap results in the smearing of the momentum distribution in an interval @xmath100 around the fermi energy .
the density given by the quasi - particle approximation to the spectral function @xmath58 is @xmath101 times the actual fermion density given by ( [ rhosf ] ) .
the densities calculated from the spectral function @xmath58 and @xmath61 are almost the same . in fig .
[ szcz ] we plot the temperature dependence of the superfluid gap @xmath102 as function of the temperature .
below @xmath103 mev the superfluid gap rapidly sets in .
we could not perform calculations below @xmath104 mev , due to limited numerical resolution in energy .
however , a saturation in the dependence of the superfluid gap can already be observed at the lowest temperatures studied .
the dashed line denotes the result of the bcs theory .
the critical temperature for the bcs theory is the same as in the quasi - particle approximation , @xmath105mev .
the scattering of nucleons and the pseudogap formation above @xmath0 make the pairing more difficult . in the pairing approximation
@xcite part of the actual energy gap is due to the pseudogap and part to the superfluid gap .
thus the superfluid transition occurs only when the pseudogap alone is not enough to exclude the pairing .
it occurs at a much lower temperature .
another effect may be related to quantitative differences between the quasi - particle and the self - consistent calculations .
the real part of the self - energy and the t - matrices are different in the two cases .
below @xmath0 the self - consistent energy gap is always significantly smaller than the bcs result .
this is different from the pseudogap pairing approximation @xcite , where at zero temperature the pseudogap goes to zero and the usual bcs result is recovered .
we could not perform calculations at low temperatures , due to limitations in computational resources . however , already in fig .
[ szcz ] , a saturation of the energy gap in the self - consistent calculation is visible , around @xmath106 mev . for comparison we plot also the dispersion relation for the peaks in the bcs spectral function at @xmath107 mev : @xmath108 .
since the energy gap is larger than in the self - consistent calculation the splitting of the two energies in the spectral function is much more pronounced .
it should be noted that the condition of weak bcs , @xmath109 , is not really fulfilled in the above case .
it is another indication that weak coupling bcs theory with free nucleon - nucleon interaction should not be used in the nuclear matter .
the strong modification of the spectral function by the anomalous self - energy in the bcs case is visible also in the fermion momentum distribution ( fig .
[ ferbcs ] ) .
the bcs fermion distribution is much broader than the hartree - fock distribution .
also the density , which is taken to be @xmath110 for the bcs fermion distribution is @xmath111 for the hartree - fock distribution .
we have presented a new way of performing nuclear matter calculations in the presence of nuclear superfluidity .
the starting point is the t - matrix approximation for the nucleon self - energy .
it allows to calculate the spectral properties of nucleons in an interacting system , treating on equal footing particle - particle and hole - hole diagrams .
it leads also to the formation of cooper pairs below some critical temperature .
strong scattering in the vicinity of the critical point , leads to a breakdown of the quasi - particle picture , confirming the observation of @xcite .
this requires the use of self - consistent , and not quasi - particle , fermion green s functions for the calculation of the self - energy . a more important generalization of the formalism is required to address the superfluid transition . in section [ pairing ] such a scheme is presented at least for the case , where the t - matrix approach indicates a second order transition .
it consists in the use of off - shell propagators in the t - matrix ladder , one of them including also the anomalous self - energy .
in such a way the superfluid gap equation is equivalent to the condition that the t - matrix has a singularity at fermi energy and zero total momentum for all temperatures below @xmath0 .
the full nucleon spectral function includes the anomalous self - energy from the condensed pairs , and the normal self - energy due to nucleon - nucleon scattering .
both contributions are important for strongly attractive potentials in nuclear matter leading to an interplay of the pseudogap and superfluid gap around @xmath0 .
the method presented in this work has several important properties * above @xmath0 , it leads to a self - consistent t - matrix resummation , which allows for the description of the feedback of the pseudogap formation and of the scattering on the t - matrix . *
it provides a link between the gap equation and the t - matrix equation , at the critical temperature and below . *
it gives the bcs theory in the limit of small scattering rates , unlike the approach @xcite .
* the value of the energy gap and the critical temperature are strongly reduced , in comparison to the bcs ( quasi - particle ) results . when used with realistic nuclear forces
, this approach presents a procedure for calculating the nuclear matter properties , taking into account self - consistently single - particle spectral properties and fermion pairing .
both these important questions were not studied up to now in actual nuclear matter calculations .
the first of these problems was discussed in a number of works related to high @xmath0 superconductivity @xcite ; the first such calculation being presented by hausmann @xcite .
the second question was addressed , to the author s knowledge , only by two groups in condensed matter physics @xcite . the work @xcite uses a t - matrix resummation with both fermion propagators in the ladder including the anomalous self - energy . however , due to the truncation of the resummation in the anomalous sector , the usual bcs gap equation is used . in that way ,
the gap equation is not consistent with the t - matrix resummation of the normal self - energy .
the approach @xcite is similar , i.e. it uses two full propagators in the ladder diagrams .
thus , the gap equation is recovered only for small gap , or is used in its linear form .
the work @xcite uses the pairing approximation , with one mean - field propagator and one with normal and anomalous self - energies . both self - energies have a bcs like form .
our approach is a generalization of this procedure , using two off - shell propagators in the t - matrix it allows us to use the off - shell propagator for the calculation of the hartree - fock and scattering self - energies ( [ hfsc ] ) and ( [ imsc ] ) , it is not restricted to a bcs like form of the normal self - energy , and it gives the most general t - matrix self - energy scheme above @xmath0 .
it should be stressed that there is no reason to neglect off - shellness in one of the propagators in the t - matrix ladder .
the consistency with the gap equation requires only to neglect the anomalous self - energy in one of these propagators .
the value of the pseudogap goes to zero at low temperature in the pairing approximation @xcite . on the other ,
hand the energy gap in the self - consistent calculation is much smaller than the bcs gap also at low temperatures ( fig .
[ szcz ] ) .
the numerical results are given for a very simple interaction .
we plan to study the superfluid nuclear matter using more realistic separable interactions and also at zero temperature .
this would allow to study questions like the effect of pairing interaction on the ground state energy , the interplay between pairing in isospin @xmath112 and @xmath55 channels , and the pairing in neutron matter .
some more fundamental questions remain also open .
we use for the energy gap a mean - field approximation .
moreover , the interaction potential in the gap equation is the free one .
obviously modification of the pairing potential in medium , due to screening , are expected . for consistency
the same screened potential should be used in the t - matrix equation for normal nucleon self - energy , meaning significant complications @xcite .
another , question is related to the cause of the reduction of @xmath0 . in our calculation both the pseudogap and the nucleon scattering effects are included . to disentangle them the self - consistent calculation
should be compared to results of the quasi - particle approximation and of the pairing approximation below @xmath0 .
this work was partly supported by the national science foundation under grant phy-9605207 .
in this appendix we sketch the numerical method used in the iterative solution of the coupled equations for the t - matrix , the the self - energy and the spectral function .
special methods must be used for an efficient estimation of multidimensional integrals , over energies and momenta , involved in the calculation of the t - matrix and the self - energy when using off - shell propagators .
let us enumerate below the steps of the iterative procedure , explaining in details the calculation of the energy integrals .
we present the case of a separable potential of rank 1 , as used in the actual calculations in this work . in that case the t - matrix in a given channel @xmath113 is given by @xmath114 where the function @xmath115 depending only on the total momentum and energy is given by @xmath116 for rank one separable interaction , a nonzero solution of the gap equation ( [ intgap ] ) is equivalent to the condition @xmath117 and the momentum dependence of the pairing gap is @xmath118 .
let us present in the following , step by step , one iteration starting from the calculation of the spectral function : 1 .
let us assume that the imaginary part of the self - energy is known .
obtained earlier for a similar temperature and density is used . ] the dispersive contribution to the real part of the self - energy can be obtained from ( [ disc ] ) .
the calculation of the spectral functions @xmath58 and @xmath61 requires also the knowledge of the chemical potential @xmath3 and the pairing gap @xmath65 . the condition for superfluidity is checked and if necessary the gap equation ( [ intgap ] ) is solved , using a numerical procedure for the solution of a nonlinear equation . for each trial value of @xmath65 the chemical potential and the corresponding hartree - fock energy are obtained with the constraint ( [ rhosf ] ) and using eqs .
( [ as ] ) and ( [ hfsc ] ) . for each trial value of @xmath3 the coupled equations ( [ hfsc ] ) , ( [ as ] ) , ( [ spectral ] ) , ( [ rhosf ] ) are solved by iteration . as a result we obtain the spectral functions @xmath58 and @xmath61 .
2 . in this step
we calculate the t - matrix @xmath119 for a range of values of total momentum and energy .
let us first calculate the imaginary part of the function @xmath120 @xmath121 the dots represent a similar term with factors @xmath122 instead of @xmath123 .
the above @xmath124 integral is a convolution integral and can be performed using fast fourier transform algorithms for numerical convolutions .
first the two factors in the energy integral are fourier transformed @xmath125 \ .\ ] ] then the function @xmath126 is calculated using standard integration procedures for the two - dimensional momentum integration . the first term in eq . ( [ nuj ] )
is obtained by inverse fourier transform of @xmath127 .
analogously the second term in ( [ nuj ] ) can be calculated .
the real part of @xmath128 is calculated using a dispersion relation @xmath129 3 . the imaginary part of the self - energy ( [ imsc ] )
can also be written as a sum of two convolution integrals in energy and can be calculated in a similar way as @xmath130 in the previous point .
this gives a next iteration of the single - particle width and the iteration returns to the first point .
a generalization to a separable potential of higher rank is obvious . since the fourier transforms are performed before the loop over the total momentum , the most expensive numerical cost comes from the two dimensional integrals , when calculating @xmath131 for a range of values of @xmath132 and @xmath133 .
very similar integrals must be performed when calculating the t - matrix and the self - energy in the quasi - particle approximation , except that here we have several such integrals . in practice , when starting the iteration from a self - energy obtained in a previous calculation at similar temperature and density , around 10 iteration are sufficient to converge with a relative deviation @xmath134 .
the numerical procedure in its present form introduces a cutoff in momenta of nucleons , to limit the energy range .
thus , we can not treat more realistic potentials , whose cutoff functions @xmath135 involve large momenta . on the other hand , with decreasing temperature ,
the spectral function becomes narrow near the fermi energy , and can not be discretized on a finite grid of energy .
the extension of the present work to low energies and to large nucleon momenta requires a combined use of quasi - particle approximation near the fermi energy and for large momenta and a continuum spectral function elsewhere , similar to the ansatz used in ref .
@xcite .
sprung , in _ advances in nuclear physics _ , edited by m. baranger and e. vogt , ( plenum press , new york , 1972 ) ; c. mahaux and r. sartor , in _ advances in nuclear physics _ , edited by j.w .
negele and e. vogt , ( plenum press , new york , 1991 ) a.l .
fetter and j.d .
walecka , _ quantum theory of many - particle systems _ ( mcgraw - hill , new york , 1971 ) .
cooper , r.l . mills and a.m. sessler ,
* 114 * ( 1959 ) 1377 .
dickhoff , phys . lett .
* b210 * ( 1988 ) 15 ; b.e .
vonderfecht , c.c .
gearhart , w.h .
dickhoff , a. polls and a. ramos , phys . lett .
* b253 * ( 1991 ) 1 .
vonderfecht , w.h .
dickhoff , a. polls and a. ramos , nucl . phys .
* a555 * ( 1993 ) 1 .
vonderfecht , w.h .
dickhoff , a. polls and a. ramos , phys . rev .
* c44 * ( 1991 ) r1265 .
khler , phys . rev .
* c46 * 1687 ( 1992 ) 1687 . t. alm , g. rpke , a. schnell , n.h .
kwong and s. khler , phys . rev .
* c53 * ( 1996 ) 2181 ; a. schnell , t. alm and g. rpke , phys . lett .
* b387 * ( 1996 ) 443 .
thouless , ann .
( n.y . ) * 10 * ( 1960 ) 553 .
kadanoff and p.c .
martin , phys . rev . * 124 * ( 1961 ) 670 .
p. bozek , nucl - th/9811073 .
f. de jong and h. lenske , phys . rev .
* c56 * ( 1997 ) 154 .
haftel and f. tabakin , nucl . phys . *
a158 * ( 1970 ) 1 ; m. golberger and k. watson , _ collision theory _
( john wiley and sons , new york , 1964 ) .
kadanoff and g. baym , _ quantum statistical mechanics _ ( benjamin , new york , 1962 ) .
p. danielewicz , ann . phys . * 152 * ( 1984 ) 239 .
w. botermans and r. malfliet , phys .
* 198 * ( 1990 ) 115 .
m. schmidt , g. rpke and h. schulz , ann .
* 202 * ( 1990 ) 57 .
y. yamaguchi , phys .
* 95 * ( 1954 ) 1628 .
m. baldo , i. bombaci , g. giansiracusa , u. lombardo , c. mahaux and r. sator , nucl . phys .
* a545 * ( 1992 ) 741 .
o. benhar , a. fabrocini and s. fantoni , nucl . phys .
* a550 * ( 1992 ) 201 .
m. randeria , n. trivedi , a. moreo and t. scalettar , phys . rev
* 69 * ( 1992 ) 2001 ; p.g .
mcqueen , d.w .
hess and j.w serene , phys . rev . *
b 50 * ( 1994 ) 7304 ; n. trivedi and m. randeira , phys .
* 75 * ( 1995 ) 312 ; r. micnas , m.h .
pedersen , s. schafroth , t. schneider , j.j .
rodrguez - nez and h. beck , phys . rev .
* b52 * ( 1995 ) 16223 ; m.y .
kagan , r. frsard , m. capezzali and h. beck , phys .
* b57 * ( 1998 ) 5995 ; m. randeira , cond - mat/9710223 m.h .
pedersen , j.j .
rodrguez - nez , h. beck , t. schneider and s. schafroth , zeit .
* b103 * ( 1997 ) 21 .
j. maly , b. jank and k. levin , cond - mat/9805018 ; j. maly , b. jank and k. levin phys . rev .
* b56 * ( 1997 ) r11407 ; i. kosztin , q. chen , b. jank and k. levin , phys . rev .
* b58 * ( 1998 ) r5936 . q. chen , i. kosztin , b. jank and k. levin , phys . rev .
* 81 * ( 1998 ) 4708
. b. kyung , e. g. klepfish and p.e .
kornilovitch , phys .
* 80 * ( 1998 ) 3109 .
r. hausmann , phys .
* b49 * ( 1994 ) 12975 .
r. hausmann , zeit .
* b91 * ( 1993 ) 291 .
a.m. sessler , in _ liquid helium _ , proceedings of the international school of physics enrico fermi , edited by g. careri , ( academic press , new york , 1963 ) ; v.g .
valeev , g.f .
zharkov and yu.a .
kukharenko , in _ nonequilibrium superconductivity _ , proceedings of the lebedev physics institute vol .
174 , edited by v.l .
ginzburg , ( nova science publishers , new york , 1988 ) ; d. rainer and j.a .
sauls , in _ superconductivity _ , lecture notes of the ictp spring college in condensed matter on `` superconductivity '' , edited by p.n . butcher and yu lu , ( world scientific , singapore , 1995 ) .
goodman , nucl .
a186 * ( 1972 ) 475 .
m. baldo , u. lombardo and p. schuck , phys .
* c52 * ( 1995 ) 975 .
migdal , _ theory of finite fermi systems _ ( john wiley and sons , new york , 1967 ) . j.r .
schrieffer , _ theory of superconductivity _
( w.a . benjamin , inc . , massachusetts , 1964 ) ; g.d .
mahan , _ many - particle physics _
( plenum press , new york , 1981 ) .
anderson and p. morel , phys . rev .
* 123 * ( 1960 ) 1911 .
r. balian and n.r .
werthamer , phys . rev .
* 131 * ( 1963 ) 1553 .
anderson and w.f .
brinkman , in _ the physics of liquid and solid helium , part ii _ , edited by k.h .
benneman and j.b .
ketterson , ( john wiley and sons , new york , 1978 ) .
jackson , a. lande and r.a .
smith , phys .
* 86 * ( 1982 ) 55 . | we present a method to calculate nuclear matter properties in the superfluid phase .
the method is based on the use of self - consistent off - shell nucleon propagators in the t - matrix equation .
such a complete treatment of the spectral function , is required below and around @xmath0 due to a pseudogap formation in the spectral function . in the superfluid phase
we introduce the anomalous self - energy in the fermion propagators and in the t - matrix equation , consistently with the strong coupling bcs equations .
the equations for the nucleon spectral function include both a contribution of condensed and scattering pairs .
the method is illustrated by numerical calculations . above @xmath0 pseudogap formation
is visible in the spectral function and below @xmath0 a superfluid gap also appears .
16.3 cm 24.2 cm -2.5 cm 2.0 cm 2.0 cm -2.cm -1.5 cm = 1 . *
pacs : * 24.10cn , 21.65+f * keywords : * nuclear matter ,
superfluidity , spectral function @xmath1electronic address : [email protected] |
time - resolved pump - probe optical spectroscopy represents a new powerful tool to study materials with strongly correlated electrons and offers a direct approach to far - from - equilibrium phenomena , probing in particular the relaxation and thermalization processes @xcite .
it appears that strongly correlated systems are in general characterized very fast relaxation processes , emerging from the inherent strong interactions .
theoretical studies of nonequilibrium dynamics and transient phenomena in correlated models confirm this , even in particular states as the mott - hubbard insulator @xcite . in theory , probing
the transient nonequilibrium state with a weak electromagnetic pulse naturally leads to the linear - response approach .
the optical conductivity @xmath0 is the time - dependent dynamical quantity directly relevant to pump - probe optical spectroscopy and represents the response to the probing electric field in a nonequilibrium situation .
some care is needed to define and properly extract the sum rules and a possible dissipationless component drude weight ( charge stiffness ) .
recently , such a formalism has been proposed for continuous correlated systems @xcite .
adapted for the application of the dynamical mean - field theory ( dmft ) , it has been used for the analysis of the hubbard model @xcite or with the emphasis on the description of time - dependent photoemission spectroscopy @xcite .
another definition has been used to examine the dynamics of the hubbard - holstein model @xcite . a stationary response , as
a characteristic of a nonequilibrium quenched state , has also been considered recently and studied explicitly for a hard - core boson model @xcite and in connection with the fluctuation - dissipation relation @xcite .
there are also studies in which the effect of the probe field after the intensive pulse excitation is directly followed by introduction of the ( classical ) driving time - dependent electric field @xcite .
the aim of this paper is to introduce the differential optical conductivity @xmath0 representing the causal linear response of the lattice current to an arbitrary electric - field pulse @xmath3 , acting on a general nonequilibrium many - body wave function @xmath4 within a tight - binding model of correlated electrons .
such a formulation allows the definition and the consideration of the optical sum rule at any time , as well as the possible existence of the drude weight @xmath5 as the dissipationless response . as a nontrivial example we test the formalism by numerical investigation of a single highly excited charged particle ( hole ) within mott - hubbard insulator , as represented by the two - dimensional ( 2d ) @xmath1-@xmath2 model .
it seems plausible that in the long - time and thermodynamic limit , anomalous @xmath0 should approach the ground state ( g.s . )
response @xmath6 , since the particle is expected to relax to the g.s . by emitting the extra energy via relevant bosonic excitations ,
i.e. , via magnons . on the other hand , in a closed finite system one would expect the response to approach a thermal equilibrium response @xmath7 , characterized by @xmath8 .
our numerical solutions indicate that these presumptions are only partly realized for concrete examples and that for the long - time response also the initial state plays a role , clearly visible at least within the limitation of our finite systems and finite evolution times .
the paper is organized as follows . in sec .
ii we present the linear response formalism for the optical conductivity @xmath0 within the tight - binding model for the general nonequilibrium many - body wave function and related density matrix .
since the calculation of the time - dependent @xmath0 in principle involves a two - time evolution , implementation with a single - time evolution is developed in order to reduce the numerical complexity as described in sec .
section iv is devoted to the numerical study of a nontrivial test case , representing the optical response of the excited particle in the strongly correlated background , as given by the @xmath1-@xmath2 model with a single hole .
conclusions and open questions are discussed in sec .
in a general single - band tight - binding model of correlated electrons ( with charge @xmath9 , assuming the system with periodic boundary conditions ( pbcs ) , the action of the electromagnetic field can be introduced via the vector potential @xmath10 through the usual gauge ( peierls ) construction .
the latter neglects the inter - band tunneling in multi - orbital models , and the field - induced distortions of the orbitals , but remains appropriate for the single - orbital case and weak fields .
we consider the tight - binding hamiltonian ( using @xmath11 ) up to @xmath12 , @xmath13 written with the particle current @xmath14 and the stress tensor @xmath15 operators , @xmath16 where @xmath17 . the electrical current @xmath18 is a sum of the particle current and the diamagnetic contribution .
we treat the case in which the unperturbed system is described by a pure ( nonequilibrium ) many - body wave function ( wf . ) @xmath19 , with the time evolution operator @xmath20 $ ] corresponding to the time - independent @xmath21 .
we use the standard formalism @xcite to evaluate the linear response of @xmath22 to a general @xmath23 applied at @xmath1 , whereas the diamagnetic part is already linear in @xmath24 . introducing the notation for expectation values @xmath25 and the interaction representation @xmath26 , @xmath27 \rangle_t . \ ] ] differential conductivity @xmath28 is defined through the response to electric field @xmath29 , @xmath30 @xmath31 being the volume of the system . taking into account @xmath32 and eqs .
( [ jet]),([djet ] ) we get @xmath33 .\ ] ] for a nonstationary state there is no unique definition of the frequency - dependent @xmath34 @xcite .
we choose plausible relation reflecting the causality and switching - on of the field at time @xmath1 , i.e. , @xmath35 , @xmath36 where @xmath37 is the width of window in which we do the fourier transformation , so that formally @xmath38 but is in practice the maximum time of probe duration .
with such a definition , eq .
( [ sigot ] ) , we avoid the ambiguity of including times prior to the pump pulse .
the sum rule for so defined @xmath39 then follows directly from eq .
( [ sigot ] ) , @xmath40 it is evident that the sum rule , eq .
( [ sumr ] ) , is a time - dependent quantity , i.e. , @xmath41 evaluated at the time @xmath1 when the probe field is applied .
moreover , independent of the precise form of the fourier transform it remains proportional to the @xmath41 at the time held fixed in the transformation .
one can define also the drude weight @xmath5 as the dissipationless component , @xmath42 , \label{dt}\end{aligned}\ ] ] again for @xmath43 .
equation ( [ dt ] ) is a generalization of the equilibrium expression , @xmath44 @xcite .
in contrast to the sum rule , @xmath5 following from eq .
( [ dt ] ) is expected to be dominated by @xmath45 in eq .
( [ sigtt ] ) .
its dependence on @xmath1 is revealed if written in the basis of eigenstates @xmath46 of @xmath21 . in the standard notation for matrix elements @xmath47 and amplitudes @xmath48 ( @xmath49 chosen arbitrarily , e.g. , when the nonequilibrium state is prepared ) , we can express @xmath5 in the eigenbasis , assuming that there are no degeneracies , @xmath50 + \notag \\
+ \frac{1}{2v}\sum_{m ,
n\neq m } a_m^ * a_n \frac{j_{mn}(j_{mm}-j_{nn})}{(\epsilon_m-\epsilon_n ) } \ e^{i(\epsilon_m-\epsilon_n)t}.\end{aligned}\ ] ] as relevant for the isotropic cases @xmath51 is the diagonal term , with @xmath52 . obviously , the last term in eq .
( [ drudebasis ] ) provides dependence on @xmath1 if nonzero .
however , it is expected to vanish if averaged over @xmath1 @xcite , yielding stationary @xmath53 which is dependent only on the ( initial ) nonequilibrium state @xmath54 through @xmath55 . moreover , the first two terms in eq .
( [ drudebasis ] ) resemble the equilibrium expression @xcite , with thermal weights substituted by projection weights @xmath56 . however , the latter derivation is feasible only for the time - independent @xmath21 .
one can express limiting @xmath57 also for the case with degeneracies , where it is of more general form , @xmath58,\end{aligned}\ ] ] still being independent of the choice of the basis within the degenerate sector .
we have so far considered the response of a pure state @xmath19 and a @xmath1-independent unperturbed @xmath21 .
the formalism can be extended also to the more general density matrix as well as the time dependent @xmath59 , e.g. , representing the presence of the pump . in this case , the response to the perturbation @xmath60 for an ensemble of pure states or time - dependent @xmath59 is derived by expanding the density matrix up to the first order in @xmath61 , @xmath62 , and using the von neumann equation @xmath63 $ ] .
the linear response of general operator @xmath64 at time @xmath65 to perturbation applied at @xmath1 is then @xmath66 \\ & = -i \int_{t}^{t ' } dt '' f(t'')tr[\rho_0(t ) [ b^{i}(t'),v^{i}(t '' ) ] ] , \\
\rho_0(t)&=u(t,0)\rho_0(0)u^\dagger(t,0 ) , \notag \\
\rho_1(t')&=i \int_t^{t ' } dt '' f(t'')u(t',t '' ) [ \rho_0(t''),v ] u^\dagger(t',t '' ) , \notag\end{aligned}\ ] ] where the time - evolution operator is @xmath67 $ ] .
in such a formulatiom the linear response of the particle current to the field applied at @xmath1 , written with density matrix for a single pure state @xmath68 is @xmath69.\end{aligned}\ ] ] optical conductivity , possibly generalized also to an ensemble of pure states , is then @xmath70\right),\ ] ] where now @xmath71 $ ] .
let us discuss here the numerical implementation for @xmath0 only for the isotropic system in the case of a single w.f . @xmath19 and time - independent @xmath21 .
the apparent disadvantage of the definition with eq .
( [ sigtt ] ) is that for a fixed @xmath1 the evaluation of @xmath72 via eq .
( [ jet ] ) requires the propagation @xmath73 for each @xmath74 , and finally for the numerical calculation at chosen @xmath37 , @xmath75 operations are needed with @xmath76 being the integration time step .
instead it is more efficient to calculate the integral inside the matrix element , eq.([jet ] ) , performing discrete steps , @xmath77 where @xmath78 and @xmath79 propagation for @xmath76 . the sum eq .
( [ intsum ] ) can be then evaluated recursively , @xmath80 such a procedure reduces the number of operations to @xmath81 .
we note that quite an analogous procedure can be applied for other transient correlation functions or for the time - dependent @xmath59 .
in order to test the feasibility of the above formalism and contribute to the discussion of transient optical response of nonequilibrium strongly correlated systems , we investigate in the following the case of a single excited charge carrier ( hole ) @xmath82 , doped in an antiferromagnetic ( afm ) mott - hubbard insulator .
we consider the standard single - band @xmath1-@xmath2 model , @xmath83 where @xmath84 are fermion operators , projected onto the space with no double occupancy , describing hopping between the nearest neighbor sites only .
we consider in the concrete example the 2d square lattice , relevant for cuprates , with @xmath85 .
further on we use @xmath86 as the unit of energy , as well of time @xmath87 .
lattice spacing is set at @xmath88 so that @xmath89 , as well as @xmath90 .
the intention is to consider the situation relevant for the pump - probe experiments on cuprates @xcite .
one can imagine two different situations : \a ) first is the photodoping of the mott - hubbard insulator , where a low concentration of highly excited charge carriers ( holons and doublons ) is created within an otherwise insulating afm system .
both types of carriers would exhibit , at least in the transient stage , an independent response to the probe pulse .
to simulate this situation we consider as the initial w.f . a single hole localized on one site ,
@xmath91 being the eigenstate of the model , eq .
( [ tj ] ) , where effective @xmath92 is put to zero .
\b ) another setting is a weakly doped afm insulator , represented by the g.s . of a single hole within the @xmath1-@xmath2 model .
the effect of a strong pump pulse applied to it can be simulated by introduction of a phase shift in the hopping term of eq .
( [ tj ] ) , changing @xmath93 , where we perform the maximum shift in the chosen direction @xmath94 , i.e. @xmath95 . both scenarios correspond to the same change in the total kinetic energy @xmath96 , calculated from the expectation value of the @xmath92 term in eq .
( [ tj ] ) .
as obvious for the initially localized hole with @xmath97 , the excited state has @xmath98 . for
the choice @xmath99 the effect of the particular phase shift is to change the sign of hopping in the @xmath94 direction . due to rotational invariance
this again yields zero total kinetic energy .
results for a single hole within the @xmath1-@xmath2 model are obtained via two numerical methods .
one is the exact diagonalization ( ed ) of small systems employing lanczos method , where we study the 2d square lattices with @xmath100 sites and pbc .
first we find the g.s .
@xmath54 . by solving time - dependent schrdinger equation , the time evolution of @xmath101 and evaluation of recursive relations , eq .
( [ iter ] ) , are obtained by employing the lanczos basis @xcite .
since the available square lattices ( @xmath102 ) are in general not rotationally invariant we perform the averaging of @xmath103 for case b ) together with corresponding pulses @xmath104 .
another method to evaluate @xmath0 is the diagonalization within the limited functional space ( edlfs ) @xcite the advantage of the edlfs method in the equilibrium regime follows from a systematical construction of states with distinct configurations of local spin excitations in the proximity of the hole . in this way ( in contrast to the ed on small system ) the method in principle deals with an infinite system . in practice
the effective size of the system is larger than in ed , but still limited by the number of basis states taken into account .
the edlfs remains efficient even when applied to nonequilibrium systems , as long as the spin disturbance caused by the local quench remains within the confines of generated spin excitations @xcite . in the considered case
spin excitations extend up to @xmath105 lattice sites away from the hole .
first we analyze the time variation of @xmath106 , representing the sum rule eq .
( [ sumr ] ) , which is for the tight - binding model ( [ tj ] ) related to the kinetic energy , @xmath107 , where only hopping in the @xmath108 direction is taken into account . in fig . 1 we present ed and edlfs results for @xmath41 calculated directly from kinetic energy for @xmath54 corresponding to localised and @xmath109-pulse excitations , respectively . averaging @xmath110 is employed in ed calculations , whereas for edlfs @xmath111 . for comparison
we show also the g.s .
@xmath112 . from fig . 1
it follows that for both types of initial excited states the decay of @xmath41 is very fast .
the corresponding short time @xmath113 can be related to the formation of the spin polaron and can be explained with the generation of string states @xcite .
it is expected to scale as @xmath114 .
we confirm in fig
. 1 that both methods , the ed and the edlfs , give quite consistent result for @xmath115 for short times @xmath116 ( for chosen @xmath117 ) . for intermediate times
@xmath118 the decay of @xmath119 evaluated with ed is somewhat slower , which could indicate the influence of finite - size and pbc effects .
namely , in small systems considered with the ed spin excitations populate the lattice and consequently influence further hole relaxation .
the spread of spin excitation is expected to saturate in @xmath120 , yielding an approximately stationary response afterwards . within the edlfs with in principle an infinite lattice ,
such effects are not present or appear only at later times due to the restricted basis .
more than an artifact , ed results on small lattices should be relevant for the optical sum rules of systems with finite density of carriers , while those from edlfs corresponds to systems with vanishing density .
although for both types of excitations @xmath121 , the distinction in sum rules is apparent for times shorter than @xmath122 , featuring the fact that the state of the initially localized hole has the rotational symmetry , whereas with the @xmath109pulse this symmetry is broken , with @xmath123 .
vs time @xmath1 for @xmath124 as obtained using the ed on @xmath125 sites as well as the edlfs .
results are presented for initial states of : a ) the localized hole and b ) the @xmath109-pulsed excited hole . the g.s .
@xmath126 value is also displayed.,title="fig:",scaledwidth=30.0% ] + vs time @xmath1 for @xmath124 as obtained using the ed on @xmath125 sites as well as the edlfs .
results are presented for initial states of : a ) the localized hole and b ) the @xmath109-pulsed excited hole . the g.s .
@xmath126 value is also displayed.,title="fig:",scaledwidth=30.0% ] it is evident that for long times @xmath127 kinetic energy approaches or oscillates around a quasi - stationary value , denoted by @xmath128 . within the ed the latter still somewhat depends on the system size @xmath129 .
the larger the system under consideration , the closer this value is to the g.s .
@xmath130 , as marked in fig .
[ fig1a ] .
since the ed simulates a fixed - size system , and the excited state is quite far from the g.s .
, a plausible interpretation could be investigated within the concept of thermalization , i.e. the approach to the equilibrium state with a finite effective temperature @xmath131 .
one could argue that different @xmath128 originate in different @xmath132 , depending on the type of quench and size of the system . in a finite system @xmath132
is set by the excitation energy so that the canonical expectation value of the energy equals the total initial energy @xmath133 . at @xmath125
the effective temperature is approximately @xmath134 for the initially localized hole and @xmath109-pulsed hole , respectively , and increases as @xmath129 decreases .
our observation is that @xmath135 is still lower than @xmath128 for all finite systems considered , as seen in fig .
[ fig1a ] this suggests that the system can not completely thermalize , possibly due to the discreteness of spin excitations in finite systems . in this connection
we notice that the @xmath8 calculation of the same model , eq .
( [ tj ] ) , with a single hole using the finite temperature lanczos method @xcite surprisingly reveals that @xmath136 does not essentially differ from g.s .
@xmath130 , fig .
[ fig1a ] . vs. temperature @xmath137 for single hole on a system with @xmath125 sites .
dots mark @xmath138 for both types of excitations in @xmath139 systems and g.s .. ,scaledwidth=30.0% ] in fig .
[ fig2 ] we finally present results for the time - dependent optical spectra per hole @xmath140 as obtained by ed and edlfs , respectively , following the described procedure .
the cases of initially localized and @xmath109-pulse excited holes are compared .
with respect to fig . 1 chosen times represent different evolution stages : a ) @xmath141 response of the initial excited state @xmath54 , b ) response at approximately characteristic decay time @xmath142 , and c ) @xmath143 already relaxed but not yet fully stationary response . to mimic the stationary response for finite systems @xmath144 , obtained by average over responses in interval @xmath145 \gg t_d$ ] ,
is presented in fig .
[ fig3 ] and compared with the g.s . @xmath146 and with the thermal - equilibrium result @xmath147 at effective @xmath148 , all obtained within the ed . in the initial stage
the response is very incoherent and for the @xmath149pulse case even predominantly negative @xmath150 , which is compatible with the sum rule @xmath151 in fig . 1 .
the latter indicates on a highly nonequilibrium state @xmath54 corresponding to an inverse particle population @xcite .
( per hole ) vs. @xmath152 as calculated at different times @xmath153 for : a ) the initially localized hole with the ed on @xmath125 sites , b ) localized hole with with the edlfs , c ) the @xmath109-pulse excited hole with the ed , and d ) the @xmath109-pulse excited hole within edlfs . for comparison the g.s .
@xmath146 is shown .
broadening of spectra @xmath154 is used .
, title="fig:",scaledwidth=30.0% ] + ( per hole ) vs. @xmath152 as calculated at different times @xmath153 for : a ) the initially localized hole with the ed on @xmath125 sites , b ) localized hole with with the edlfs , c ) the @xmath109-pulse excited hole with the ed , and d ) the @xmath109-pulse excited hole within edlfs . for comparison the g.s .
@xmath146 is shown .
broadening of spectra @xmath154 is used .
, title="fig:",scaledwidth=30.0% ] + ( per hole ) vs. @xmath152 as calculated at different times @xmath153 for : a ) the initially localized hole with the ed on @xmath125 sites , b ) localized hole with with the edlfs , c ) the @xmath109-pulse excited hole with the ed , and d ) the @xmath109-pulse excited hole within edlfs . for comparison the g.s .
@xmath146 is shown .
broadening of spectra @xmath154 is used .
, title="fig:",scaledwidth=30.0% ] + ( per hole ) vs. @xmath152 as calculated at different times @xmath153 for : a ) the initially localized hole with the ed on @xmath125 sites , b ) localized hole with with the edlfs , c ) the @xmath109-pulse excited hole with the ed , and d ) the @xmath109-pulse excited hole within edlfs . for comparison the g.s .
@xmath146 is shown .
broadening of spectra @xmath154 is used .
, title="fig:",scaledwidth=30.0% ] with some quantitative difference between both methods , within the g.s .
@xmath146 two features are well visible and of particular interest .
one is the mid - infrared ( mir ) peak at @xmath155 , which is the signature of the string - like excited particle states within the 2d afm background .
we see in figs .
[ fig2]a , b that the mir - like peak appears also in @xmath156 and stabilizes very fast at @xmath157 for the localized hole , independent of the numerical method employed . on the other hand ,
after the @xmath109-pulse ( fig . [ fig2]c , d ) the mir peak is less pronounced . especially within the ed , @xmath158 is closer to the thermal @xmath147 at appropriate @xmath159 . in the equilibrium at such high @xmath160 the mir peak ( figs .
[ fig2]c , [ fig3]b ) is already smeared out due to thermally disordered afm spin background . within the edlfs results ( fig .
[ fig2]d ) the mir peak recovers better , which could be attributed to a lower density of spin excitations in effectively bigger systems .
the second feature , pronounced within the g.s . is the drude weight @xmath161 , i.e. peak at @xmath162 which accounts for @xmath163 of the weight in the sum rule at @xmath164 .
it should acquire a finite width in the equilibrium at @xmath8 due to scattering processes ( i.e. in the strict sense @xmath165 is expected for a nonintegrable system ) .
although within both methods used ( eq .
[ drudebasis ] would require the full ed method ) we can not strictly establish and determine the value of @xmath5 as defined in eqs .
( [ dt],[drudebasis ] ) , it is evident that with respect to the low-@xmath152 response studied , excited particles reveal quite different behavior . within the ed the @xmath109-pulse excited hole shows essentially no drude peak , i.e. , no corresponding low-@xmath152 remainder at @xmath166 .
on the other hand , the initially localized hole displays a substantial low-@xmath152 peak and presumably @xmath167 ( note that we use broadening @xmath168 ) at all @xmath166 , although the weight is smaller than in @xmath146 . using the edlfs we notice a weak remainder of the low-@xmath152 contribution even for the @xmath109 pulse , though much smaller than for the localized hole .
these findings indirectly support eq .
( [ drudebasis ] ) , that also the initial wf .
( and not just its total energy ) determines the limiting drude weight value @xmath57 .
( color online ) long - time response @xmath169 vs. @xmath152 as calculated with ed on @xmath125 , obtained by average over responses in interval @xmath145 \gg t_d$ ] for a ) initially localized hole , and b ) the @xmath109-pulse excited hole .
results are compared to the g.s . @xmath146 and the thermal @xmath147 at corresponding effective temperatures @xmath170 .
spectra are broadened with @xmath154 .
, title="fig:",scaledwidth=30.0% ] ( color online ) long - time response @xmath169 vs. @xmath152 as calculated with ed on @xmath125 , obtained by average over responses in interval @xmath145 \gg t_d$ ] for a ) initially localized hole , and b ) the @xmath109-pulse excited hole .
results are compared to the g.s . @xmath146 and the thermal @xmath147 at corresponding effective temperatures @xmath170 .
spectra are broadened with @xmath154 .
, title="fig:",scaledwidth=30.0% ]
we have presented a formalism for the linear optical conductivity response @xmath0 of a nonequilibrium ( excited ) state of a strongly correlated system , where the probe pulse is taken as a perturbation in the linear order . in the absence of an unique approach we have chosen the definition reflecting the onset of the probe electric field pulse at time @xmath1 , which is in contrast to some other studies .
such an approach allows the discussion of the optical sum rule as well as the dissipationless drude weight at any time @xmath166 . on the other hand ,
the definition introduces a complication due to an additional time integration which we circumvent by a particular numerical implementation .
the presented test case of a single highly excited charge carrier within the @xmath1-@xmath2 model already shows several features and opens questions relevant for the theoretical analysis of the pump - probe spectroscopy results .
independently of the initial state we observe a fast relaxation of several observables , e.g. the kinetic energy and the optical sum rule , towards the respective g.s .
value . still , more specific features of the transient and long - time optical response , such as the mir peak and the drude component @xmath5 , appear non - universal .
our results reveal that they do not depend merely on the excitation energy , as e.g. expected from the canonical thermalization , but as well on the character and w.f . of the initial excited state .
the persistance of this feature in the thermodynamic limit remains an open question .
on one hand , it appears plausible that in an infinite system the local state of the quasiparticle will correspond to the ground state .
nevertheless , this statement is , e.g. , not evident for a quantum system with gapped bosonic excitations . to this
end our findings show a lack of canonical thermalization , observed before in theoretical @xcite as well as experimental studies @xcite . since we address excited systems , the concepts of thermalization and relaxation to the g.s .
response are only partly applicable , especially for dynamical quantities , and remain the challenge also for further studies .
we should note as well that in the application of our formalism and results to the pump - probe experiments some care is needed when energy absorption of particular probe pulses is measures , which can not be directly compared to time - dependent @xmath171 calculated in the present study .
z. l. and d. g. acknowledge discussions with m. eckstein and a. silva .
j. b. acknowledges stimulating discussion with s. a. trugman and the financial support by the center for integrated nanotechnologies , an office of science user facility operated for the u.s .
this work has been supported by the program p1 - 0044 and the project j1 - 4244 of the slovenian research agency ( arrs ) .
99 k. matsuda , i. hirabayashi , k. kawamoto , t. nabatame , t. tokizaki and a. nakamura , phys .
b * 50 * , 4097 ( 1994 ) .
h. okamoto , t. miyagoe , k. kobayashi , h. uemura , h. nishioka , h. matsuzaki , a. sawa and y. tokura , phys .
b * 82 * , 060513 ( 2010 ) ; phys . rev .
b * 83 * , 125102 ( 2011 ) .
s. wall et al , nature phys .
* 7 * , 114 ( 2011 ) .
f. novelli , d. fausti , j. reul , f. cilento , p. h. m. van loosdrecht , a. a. nugroho , t. t. m. palstra , m. grninger and f. parmigiani , phys .
b * 86 * , 165135 ( 2012 )
. s. dal conte _
et al . _ , science * 335 * , 1600 ( 2012 ) .
c. giannetti _ et al .
_ , nature communications * 2 * , 353 ( 2011 ) . m. eckstein and p. werner , phys . rev .
b * 84 * , 035122 ( 2011 ) .
m. eckstein and p. werner , phys .
lett . * 110 * , 126401 ( 2013 ) .
z. lenari and p. prelovek , phys .
lett . * 111 * , 016401 ( 2013 ) .
a. shimizu and t. yuge , j. phys .
soc . jpn . * 80 * , 093706 ( 2011 ) .
m. eckstein and m. kollar , phys .
b * 78 * , 205119 ( 2008 ) .
m. eckstein , m. kollar , p. werner , phys .
b * 81 * , 115131 ( 2010 ) .
j. k. freericks , h. r. krishnamurthy , t. pruschke , phys .
102 * , 136401 ( 2009 ) .
g. de filippis , v. cataudella , e. a. nowadnick , t. p. devereaux , a. s. mishchenko and n. nagaosa , phys .
* 109 * , 176402 ( 2012 ) .
d. rossini , r. fazio , v. giovannetti , and a. silva , arxiv 1310.4757 .
l. foini , l. f. cugliandolo and a. gambassi , phys .
b * 84 * , 212404 ( 2011 ) . l. foini , l. f. cugliandolo and a. gambassi , j. stat . mech . * 2012 * , p09011 ( 2012 ) . b. moritz , a. f. kemper , m. sentef , t. p. devereaux , and j. k. freericks , phys . rev . lett . * 111 * , 077401 ( 2013 ) .
p. nozires , _ theory of interacting fermi systems _
( addison - wesley , reading , 1964 ) j. jakli and p. prelovek , adv .
* 49 * , 1 - 92 ( 2000 ) .
h. castella , x. zotos and p. prelovek , phys .
lett . * 74 * , 972 ( 1995 ) .
t. j. park , j. c. light , j. chem .
85 * , 5870 ( 1986 ) .
j. bona , s. maekawa , and t. tohyama , phys .
b * 76 * , 035121 ( 2007 ) . for a review
see , p. prelovek and j. bona , in _ strongly correlated systems - numerical methods _ , edited by a. avella and f. mancini , springer series in solid - state sciences 176 , ( springer , berlin , 2013 ) , pp . 1 - 29
. m. mierzejewski , l. vidmar , j. bona , and p. prelovek , phys .
lett . * 106 * , 196401 ( 2011 ) .
l. vidmar , j. bona , t. tohyama and s. maekawa , phys .
* 107 * , 246404 ( 2011 ) .
d. gole , j. bona , m. mierzejewski , and l. vidmar , arxiv 1311.5574 ( 2013 ) . m. m. zemlji and p. prelovek , phys .
b * 72 * , 075108 ( 2005 ) . n. tsuji , t. oka , h. aoki , phys . rev
. lett . * 103 * , 047403 ( 2009 ) . m. rigol , v. dunjko and maxim olshanii , nature * 452 * , 854 ( 2008 ) . s. r. manmana , s. wessel , r. m. noack and a. muramatsu , phys .
lett . * 98 * , 210405 ( 2007 ) . c. giannetti
b * 80 * , 235129 ( 2009 ) . | we present a linear - response formalism for a system of correlated electrons out of equilibrium , as relevant for the probe optical absorption in pump - probe experiments .
we consider the time dependent optical conductivity @xmath0 and its nonequilibrium properties . as an application we numerically study a single
highly excited charged particle in the spin background , as described within the two - dimensional @xmath1-@xmath2 model .
our results show that the optical sum rule approaches the equilibrium - like one very fast , however , the time evolution and the final asymptotic behavior of the absorption spectra in the finite systems considered still reveal dependence on the type of initial pump perturbation .
this is observed in the evolution of its main features : the mid - infrared peak and the drude weight . |
the @xmath0-continued fractions introduced by nakada @xcite give a one - dimensional family of interval maps , @xmath2 with @xmath3 $ ] .
this family has been the focus of a recent spate of publications , focusing especially on the entropy of the maps .
here we answer a natural question arising in this context : for @xmath1 $ ] there is a cross section of the geodesic flow on the modular surface for which @xmath2 is a factor .
kraaikamp , schmidt and steiner @xcite determined an explicit planar model @xmath4 of the natural extension of the @xmath0-continued fraction map for positive @xmath0 .
denote by @xmath5 the measure on @xmath6 given by @xmath7 .
they also showed that @xmath5 is a @xmath8-invariant measure on @xmath9 ; let @xmath10 denote the _ entropy _ of the interval map with respect to @xmath11 , the invariant probability measure on @xmath9 induced by @xmath5 .
the following is theorem 2 in @xcite : for any @xmath1 $ ] , @xmath12 with @xmath13 denoting the unit tangent bundle of the modular surface ( see below for definitions ) , we show the following . [ t : geoflow ] for any @xmath1 $ ] , equation holds if and only if there exists a cross section @xmath14 to the geodesic flow with the following two properties : 1 .
there is a two - to - one ( a.e . )
map @xmath15 that projects the normalized transverse invariant measure on @xmath16 induced by liouville measure to @xmath17 ; 2 .
denote by @xmath18 the first return map to @xmath19 of the geodesic flow , the following diagram commutes : + @xmath20 since a system is by definition a factor of its natural extension , with equation we have the following .
[ t : factor ] for @xmath1 $ ] , the system defined by @xmath2 is a factor of a cross section of the geodesic flow on the modular surface .
that whenever @xmath2 is the factor of a cross section for the geodesic flow there is an equation of the form of equation is well - known to be a consequence of a formula of abramov , see section [ ss : firstreturn ] for related ideas .
the main contribution of this paper is thus the construction of the appropriate measurable cross sections .
even this is simply an adaption of the construction given in @xcite ; however , in that setting , the fact that the map on the cross section was a _ first _ return was evident . that this is true in our more general setting requires equation ; and ,
we must admit , careful consideration of the normalizations involved in calculating the entropy of the return map to a cross section .
+ we show that for each @xmath21 , the natural extension of @xmath2 is _
double covered _ by a corresponding cross section of the geodesic flow on the modular surface it is this cardinality two that accounts for the difference in the constant in equation from the volume of the unit tangent bundle of the modular surface . on the fiber over an @xmath22 with @xmath23 given by a matrix of determinant
@xmath24 acting on @xmath22 in the standard linear fractional manner , the first return of the flow switches sheets , over @xmath22 where the map is given by a matrix of determinant @xmath25 the flow returns to the same sheet .
the case of @xmath26 is that of the regular continued fractions , and is thus the case that is treated in @xcite ; there , above any @xmath22 the flow changes sheet .
for all other values of @xmath0 , there are @xmath22 above which one returns to the sheet of departure .
our techniques are easily adapted to the special case of @xmath27 .
the interval map @xmath28 is known as the by - excess continued fraction map , defined on the interval @xmath29 $ ] it has an infinite invariant measure given by @xmath30 .
although one thus can not define the entropy of @xmath28 , luzzi and marmi @xcite show that its entropy in what they call krengel s sense is zero . for all @xmath22 , @xmath31
is given by some matrix of positive determinant and indeed , one can extend our discussion here to show that the natural extension of the system of @xmath28 can be expressed as a cross section of flow on the modular surface , with the cross section of infinite measure .
+ the study of the intertwined nature of continued fractions and the geodesic flow on the modular surface has a rich history .
one of the first significant steps was e. artin s @xcite coding of geodesics in terms of the regular continued fraction expansions of the real endpoints of their lifts .
hedlund @xcite then used this to show the ergodicity of the geodesic flow on the modular surface .
a few years later , e. hopf proved ergodicity for the geodesic flow on any hyperbolic surface of finite volume , see his reprisal @xcite . taking up on an approach revived by moeckel @xcite , series @xcite gave an explicit cross section for the geodesic flow on the modular surface such that the regular continued fraction map is given as a factor .
adler and flatto gave a second approach , see especially their @xcite .
quite recently , d. mayer and co - authors @xcite , @xcite have given explicit cross sections for the geodesic flow on the surfaces uniformized by the hecke triangle groups such that a variant , due to nakada @xcite , of the rosen continued fractions @xcite is given as a factor .
( the second - named author first learned of such possibilities from a. haas , see the related treatment of another variant of the rosen fractions in @xcite . ) for much of this history , further motivation , and also the work of s. katok and co - authors using various means to code geodesics on the modular surface , see @xcite .
as stated above , our work here is directly inspired by another approach to identifying cross sections related to continued fractions , that of arnoux @xcite . that work can be viewed as an elaboration of veech s @xcite notion of `` zippered rectangles '' in the special case of surfaces of genus one .
here we proceed by , in a sense , extracting the algebraic expression for the cross section determined in @xcite , and adjusting this appropriately so as to pass from the planar models of the natural extensions given by @xcite to our cross sections .
thus , whereas the various cross sections mentioned in the previous paragraph were each constructed in what can be called a geometric fashion , we use a more algebraic approach .
+ there has been a great deal of recent interest in the @xmath0-continued fractions maps , and especially in the function associating to @xmath0 the entropy of the corresponding map .
nakada @xcite computed the entropy of his maps for @xmath32 .
kraaikamp @xcite gave a more direct fashion to compute these .
moussa , cassa and marmi @xcite gave the entropy for the maps with @xmath33 .
luzzi and marmi @xcite presented numeric data showing that the entropy function @xmath34 behaves in a rather complicated fashion as @xmath0 varies .
nakada and natsui @xcite gave explicit intervals on which @xmath34 is respectively constant , increasing , decreasing .
carminati and tiozzo @xcite extended this work by describing the intervals involved .
the same authors have revisited the question of the shape of the entropy function in terms of @xmath0 in @xcite .
luzzi and marmi @xcite also gave strong evidence that the entropy function is continuous .
( they furthermore gave numeric evidence that the function is far from trivial , see the graphs they present in their figures 48 ) .
tiozzo @xcite showed continuity for @xmath0 above a certain positive constant , kraaikamp - schmidt - steiner @xcite show the continuity for positive @xmath0 .
( after that work was completed , tiozzo in an updated version of @xcite showed that hlder continuity holds throughout the full interval . ) the proof of the continuity of @xmath34 given in @xcite proceeds by first showing equation so as then to turn to the key result , the continuity of the function @xmath35 ( that the geometry , and even topology , of the @xmath9 change drastically is evidenced in figure 6 of @xcite . ) the proof of equation consists mainly in showing that for any @xmath36 , the intersection @xmath37 is nontrivial , and the return maps of each of @xmath8 and @xmath38 to this intersection agree ; from this one uses abramov s formula and the fact that the equation certainly holds for @xmath26 .
( the proof of continuity of @xmath35 is much harder , in briefest terms : the shape of @xmath9 is determined mainly by the @xmath2-orbits of @xmath0 and @xmath39 , there are intervals of @xmath0 such that these two orbits meet , and in a common fashion ; continuity is fairly straightforward for each of these intervals ; the remaining set of @xmath0 is such that the key orbits can still be described symbolically , the difference in measure for nearby @xmath40 is bounded above using this description . )
luzzi and marmi @xcite seem to be the first to ask in print if the @xmath0-continued fractions arise as a factor of the geodesic flow on the modular surface .
folklore consensus seems to be that determining whether the @xmath0-continued fractions are or not factors of a cross section for the geodesic flow remained an important unsolved problem .
+ to show the robustness of our basic argument , in section [ s : rosencf ] we sketch a proof that the rosen continued fraction maps and certain of their @xmath0-type variants are also factors of cross sections of geodesic flow over the corresponding hyperbolic surface .
rosen @xcite introduced his analogs of the nearest integer continued fractions to assist in the study of the hecke triangle fuchsian groups , see below for definitions .
there has been increasing interest in the past few decades in these continued fractions , see @xcite for a sketch of the history to the mid-1990s , and the aforementioned @xcite , @xcite for some of the more recent history ; the @xmath0-rosen continued fractions , that we treat as well , were introduced by dajani , kraaikamp and steiner @xcite in 2009 .
+ in light of the extremely complicated geometry and topology of the various cross sections that we determine , it would be interesting to know if the more geometric approaches to the construction of cross sections giving interval maps as factors could be successful in our setting . * thanks * the second - named author takes great pleasure in thanking cor kraaikamp and wolfgang steiner for introducing him to the intricacies of the @xmath0-continued fractions and for comments on this continuation of joint work .
thanks also go to corinna ulcigrai for mentioning the interest in the question of relating cross sections and the @xmath2 , and to an anonymous referee who insisted on the importance of it .
for @xmath3 $ ] , we let @xmath41 $ ] and define the map @xmath42 by @xmath43 for @xmath44 , put @xmath45 with @xmath46 . furthermore ,
for @xmath47 , put @xmath48 this yields the _ @xmath0-continued fraction _ expansion of @xmath49 : @xmath50 where @xmath51 is such that @xmath52 .
( standard convergence arguments justify equality of @xmath22 and its expansion . )
these include the regular continued fractions , given by @xmath26 and the nearest integer continued fractions , given by @xmath53 .
the standard number theoretic planar map associated to continued fractions is defined by @xmath54)\,,\ ] ] and @xmath9 is the closure of the orbits of @xmath55 with @xmath56 .
recall that @xmath5 is given by @xmath7 and that @xmath9 is the planar region determined by @xcite .
let @xmath11 be the probability measure given by normalizing @xmath5 on @xmath57 , and @xmath58 the marginal probability measure obtained by integrating @xmath11 over the fibers @xmath59 , @xmath60 the borel @xmath61-algebra of @xmath62 , and @xmath63 the borel @xmath61-algebra of @xmath57 . that @xmath64 is a natural extension of @xmath65
is shown in @xcite .
much of the following can be found in manning s
chapter @xcite in the text @xcite .
+ using the mbius action of @xmath66 on the poincar upper - half plane @xmath67 , by identifying a matrix with the image of @xmath68 under it , we can identify @xmath69 with @xmath67 .
similarly , @xmath70 can be identified with the _ unit tangent bundle _ of @xmath67 .
we are most interested in the _ modular surface _ , @xmath71 .
the unit tangent bundle of the modular surface , @xmath72 , can be identified as @xmath73 the geodesic flow in our elementary setting is a map on a surface s unit tangent bundle : given a time @xmath74 and a unit tangent vector @xmath75 , since the unit tangent vector uniquely determines a geodesic passing through the vector s base point , we can follow this geodesic for arclength t in the direction of @xmath75 , the unit vector that is tangent to the geodesic at the end point of the geodesic arc is the image , @xmath76 , under the geodesic flow .
the hyperbolic metric on @xmath67 corresponds to an element of arclength satisfying @xmath77 with coordinates @xmath78 . in particular , for @xmath79 , the points @xmath80 and @xmath81 are at distance @xmath74 apart . since @xmath66 acts by isometries on @xmath67 , the geodesic flow on its unit tangent bundle is given by sending @xmath82 to @xmath83 , where @xmath84 .
similarly , on @xmath72 one sends the class represented by @xmath85 to that represented by @xmath83 .
there is a natural measure on the unit tangent bundle @xmath86 : _ liouville measure _ is given as the product of the hyperbolic area measure on @xmath67 with the length measure on the circle of unit vectors at any point .
this measure is @xmath66-invariant , and induces a finite measure on @xmath72 . with the standard choice of this liouville measure ,
the volume of @xmath72 is @xmath87 , see say @xcite .
_ normalized liouville measure _ is the corresponding probability measure on @xmath72 .
gurevich and katok @xcite cite a fairly general result of sullivan @xcite to state that the entropy with respect to _ normalized _ liouville measure of the geodesic flow on the modular surface is one .
+ a measurable _ cross section _ for the geodesic flow is a subset of the unit tangent bundle through which almost every geodesic passes tranversely and infinitely often .
the _ first return map _ to a cross section @xmath88 is the self - map @xmath89 given by @xmath90 with @xmath91 minimal such that this image is indeed in @xmath88 . by definition of @xmath88
, this map is defined almost everywhere on @xmath88 . by a celebrated result of ambrose @xcite , since the geodesic flow leaves normalized liouville measure invariant , there is a measure @xmath92 on @xmath88 invariant under the the first return map such that normalized liouville measure locally factors as the product of @xmath92 with lebesgue measure along geodesics . in subsection
[ ss : mainresults ] , we will refer to this phenomenon as an _
ambrose factorization_. due to a result of abramov @xcite , one defines the entropy of a flow @xmath93 on a probability space as the entropy of the time @xmath94 map , @xmath95 .
a celebrated formula of abramov @xcite then gives that the _ entropy of the first return map _ @xmath96 to a cross section @xmath88 is @xmath97 where @xmath98 is the probability measure on @xmath88 induced by @xmath92 .
the integral of the return times over a cross section gives the volume of the total space . in our setting , this volume is one since we considered normalized liouville measure ; sullivan s result gives that @xmath99 .
thus , we have @xmath100
recall that any locally compact hausdorff topological group has an invariant measure for left multiplication , this _ haar measure _ is unique up to scaling . in this section
, we derive explicit formulas for the haar measure on @xmath66 .
these formulas are presumably well known , and can verified by specializing known expressions for the haar measure on various classes of lie groups . here
we derive them in a straightforward , elementary fashion .
this derivation then allows us to identify a transversal to the geodesic flow see the form of the matrices given in lemma [ l : towardstransv ] that is key to our construction of cross sections
. * notation . * throughout this section , in order to use standard naming conventions of entries of matrices , @xmath0 denotes any real number .
let @xmath101 be the set @xmath102 [ p : haar ] the haar measure @xmath103 on @xmath104 is given , up to a constant , by @xmath105 the natural left action of the group @xmath106 on the set @xmath107 of @xmath108 real matrices preserves the measure induced by lebesgue measure on @xmath109 under the standard identification of @xmath107 with @xmath109
. left multiplication on @xmath107 by a matrix @xmath110 , defines a linear transformation , whose matrix with respect to the canonical basis is @xmath111 the determinant of this new matrix is @xmath112 , and thus equals @xmath113 when @xmath114 .
that is , the left multiplication preserves lebesgue measure .
let @xmath115 be the determinant @xmath116 . on the subset of matrices with @xmath117 ,
the map @xmath118 is injective , and thus @xmath119 gives local coordinates on this subset .
the jacobian of this map equals @xmath120 , and thus lebesgue measure @xmath121 on @xmath122 is given in these coordinates by @xmath123 now , left multiplication by elements of @xmath106 preserves @xmath115 on @xmath107 .
therefore , restricting to the hypersurface where @xmath124 ( that is to @xmath106 ) , we find that the measure @xmath125 is invariant by left multiplication .
this of course implies that the measure is a haar measure .
there are obviously three related expressions : @xmath126 etc . by a jacobian change - of -
coordinates calculation , one easily shows that each of these expressions defines the same measure on the intersection of their respective domains of definition .
we emphasize that the above directly proves the invariance of the measure under left ( and , for that matter , also right ) multiplication .
that is , we have naively solved for an explicit expression of haar measure .
[ l : towardstransv ] let @xmath127 be the connected component of @xmath128 defined by @xmath129 , and let @xmath130 be defined by @xmath131 .
consider local coordinates @xmath132 by letting each @xmath133 be given as @xmath134 then @xmath127 has local coordinates @xmath135 by way of @xmath136 with @xmath133 and @xmath137 as above .
furthermore , @xmath138 gives haar measure restricted to @xmath127 .
suppose that @xmath139 .
then letting @xmath140 gives @xmath141 .
clearly , the set @xmath142 for all @xmath143 , @xmath144 comprises all of @xmath127 . the map @xmath145 has jacobian of absolute value @xmath146 . since haar measure restricted to @xmath127 is ( any nonzero constant times )
@xmath147 , we deduce that @xmath138 does give haar measure here .
( the change in normalization constant by a factor of @xmath148 is innocuous here . ) for an element @xmath149 , let @xmath150 $ ] denote the corresponding element of @xmath151 .
then since haar measure @xmath106 induces liouville measure on @xmath86 , and @xmath127 projects one - to - one to a set of full measure in @xmath151 , we have the following . [
t : liouville ] under the identification of @xmath86 with @xmath152 , liouville measure on the full measure set @xmath153\,|\ , m \in g_{\gamma}\,\}$ ] is proportional to @xmath138 where @xmath154 are as above .
( we identify the above implied constant of proportionality in subsection [ ss : firstreturn ] . ) one can give another parametrization by remarking that a matrix @xmath155 in @xmath156 corresponds to an horizontal tangent vector pointing to the right if and only if @xmath157 , and to the left if and only if @xmath158 . the set of such matrices gives a section , since any geodesic which is not vertical has a highest point , where the tangent vector is horizontal .
one can parametrize such a geodesic by its origin @xmath159 and its extremity @xmath160 .
a simple computation shows that the set of such matrices , in the case @xmath157 , is parametrized by @xmath161 applying the geodesic flow as above gives a parametrization of half of the group of the form @xmath162 a jacobian computation proves that the haar measure , in these coordinates , is proportional to @xmath163 .
note that the measure @xmath164 has a clear geometric signification : the base point of the corresponding unit tangent vector is @xmath165 , so this measure is proportional to the measure defined by the hyperbolic metric on the upper half - plane .
a small modification consists in changing @xmath166 into @xmath167 , giving the following result .
any element of @xmath152 such that @xmath168 and @xmath169 have same sign can be written in a unique way ( up to sign ) : @xmath170 in these coordinates , the haar measure can be written @xmath171 thus , we have sketched a geometric interpretation of the density function for the measure @xmath5 on @xmath6 that is sometimes referred to as the `` standard number theoretic '' measure .
( this latter name comes from the fact that the orbits of points of the form @xmath55 in planar models of natural extensions with @xmath5 an invariant measure reveal diophantine approximation properties of the related interval maps . )
note that a similar derivation of an invariant measure for rosen continued fraction from liouville measure for the geodesic flow has been given by mayer and stroemberg in the section 5 of @xcite .
it is easy to give explicit formulas to change between these systems of coordinates ; we use such a change of coordinates in the next section , the function @xmath172 , to transform the invariant measure to lebesgue measure .
the goal of this section is , for each @xmath1 $ ] , to define a subset @xmath16 of the unit tangent bundle of the modular surface and to give an explicit self - map induced by the geodesic flow .
let the function @xmath173 be defined by @xmath174 and let @xmath175 .
an elementary calculation shows that @xmath172 conjugates @xmath8 to the bijection : @xmath176 a second elementary calculation shows that @xmath5 projects to give lebesgue measure , which is thus invariant for this induced map .
( of course , one can just as easily directly verify this invariance . ) given @xmath177 , let @xmath178 define the following subsets of @xmath156 : @xmath179 \ , \vert \ , ( x , y ) \in \sigma_{\alpha}\}\;\mbox{and } \ ; \mathcal a_{\alpha , + 1 } = \ { \ ; \left [ a_{+1}(x , y)\right ] \ , \vert \ , ( x , y ) \in \sigma_{\alpha}\}\,.\ ] ] in lemma [ l : distinctclasses ] we show that almost all @xmath180 in @xmath181 lie in distinct @xmath182-orbits .
we let @xmath183 denote the set of the classes in @xmath184 represented by the various elements of the @xmath185 and let @xmath186 let @xmath187 on @xmath16 be induced by lebesgue measure on @xmath188 .
the previous section applies to show that the geodesic flow from @xmath16 gives an ambrose factorization of ( some positive constant multiple of ) liouville measure of the form @xmath189 .
denote a point of @xmath16 by @xmath190 in the obvious fashion .
our main result is given by the following two statements ; we prove theorem [ t : first ] in section [ s : firstreturn ] .
[ t : first ] for any @xmath1 $ ] , the first return map of the geodesic flow to @xmath16 is given by @xmath191 by the discussion in subsection [ ss : altnatext ] , the following holds .
[ l : cover ] for @xmath1 $ ] , let @xmath192 then @xmath193 is a @xmath194 surjection , and the following is a commutative diagram .
in this section we give the remaining steps to prove the main result , in particular showing that @xmath18 is a _ first _ return map . to show that this step is in general necessary , in the final subsection we give an example ( which can easily be generalized ) of an interval map that is not given as the factor of first return to a cross section for the geodesic flow .
* notation .
* we use asterisks to denote entries of a determinant one matrix that are not germane to the argument at hand .
we first show that the classes parametrized by the various @xmath196 are almost always distinct .
this implies that @xmath197 [ l : distinctclasses ] let @xmath198 . for almost all distinct @xmath180 in @xmath199
the @xmath200 classes represented by @xmath180 are distinct .
* case 1.*suppose that both @xmath201 . if @xmath202 is such that @xmath203 , then writing @xmath180 in accordance with and multiplying on the right by @xmath204 gives @xmath205 since @xmath206 , we find that @xmath207 .
substitution into the @xmath208-element easily leads to the conclusion that @xmath209 . *
case 2.*if both @xmath210 , then we find @xmath211 and again @xmath212 .
* case 3.*if @xmath213 and @xmath214 , then again @xmath215 , as @xmath216 thus , in all cases , symmetry shows that distinctness holds unless both @xmath217 belong to the countable set @xmath218 .
[ l : flowformula ] for each @xmath219 , the geodesic flow for time @xmath220 sends @xmath85 to @xmath221 .
consider @xmath222 representing a class of @xmath223 .
since @xmath220 gives @xmath224 where @xmath225 , the geodesic flow gives @xmath226 .
when @xmath227 , this is equivalent to @xmath228 where we choose @xmath229 .
when @xmath230 , we choose a different @xmath231-orbit representative of @xmath232 , to wit : @xmath233 similarly , for @xmath234 representing a class in @xmath235 , if @xmath227 , we find @xmath236 if @xmath230 , we have @xmath237 due to the bijection of , and the definition of the @xmath183 , the result holds
. let @xmath238 be the probability measure on @xmath16 induced by lebesgue measure @xmath187 .
the system @xmath239 is a skew product with finite fiber over @xmath240 .
their entropies are thus equal : @xmath241 .
since a system and its natural extension have the same entropy , equation gives @xmath242 . by equation , theorem [ t : liouville ] and equation , the first return map of the geodesic flow to @xmath16 has entropy equal to @xmath243 where @xmath244 is a normalizing constant independent of @xmath0 .
but , @xmath245 _ is _ given by the first return of the geodesic flow ( by , say , @xcite ) , and we thus find that @xmath246 .
therefore , for all positive @xmath0 , we find that the entropy of the first return map to @xmath16 equals @xmath247 .
but , @xmath18 is given ( locally ) by powers of this first return map , and thus the equality of @xmath247 with the entropy of the first return map holds if and only if the two maps are equal a.e .
see 10.6 theorem 2 ( 3 ) of @xcite .
that is , @xmath18 is the _ first _ return map . to reassure the reader that our efforts in the previous subsection are not absurd , we sketch the existence of an interval map that is piecewise fractional linear with integral coefficients with the planar model of its natural extension double covered by a cross section for the geodesic flow on @xmath72 but for which the map is not given by the @xmath248 return map for any single @xmath249 .
we begin with the regular continued fractions . for simplicity ,
let @xmath250 and @xmath251 .
denote cylinder sets in the usual manner : @xmath252 $ ] is the subset of the unit interval of elements whose first @xmath249 partial quotients are the @xmath253 .
in particular , the @xmath254-image of @xmath252 $ ] is @xmath255 $ ] whenever @xmath256 . using the partition of @xmath257 given by @xmath258\ , , \;\mbox{and the various}\;\ ;
a_n = \;\bigcup_{k>1 , l>1}\ ; \delta[k,\underbrace{1 , \dots , 1}_{n-2 \ , \text{times } } , l]\;\ ; \mbox{with}\ ; n\ge 2 \;,\ ] ] let @xmath259 be defined by @xmath260 that is , on each @xmath261 we define @xmath262 to be the @xmath249-fold composition of @xmath254 with itself . again simplifying notation ,
let @xmath263 be the planar natural extension for @xmath264 .
for each @xmath265 , define @xmath266 to be the subset of @xmath267 lying over @xmath268 .
let @xmath269 be defined by letting @xmath270 restricted to @xmath266 be the @xmath249-fold composition of @xmath271 with itself . as usual , for simplicity s sake , we refer to dynamical systems merely by space and function , each time the mentioned have sigma - algebra of borel subsets ; invariant measures in the following are ( the normalization of ) @xmath272 and its marginal measure .
we let @xmath273 as defined in subsection [ ss : mainresults ] .
the map @xmath269 is a natural extension for @xmath262 .
furthermore , @xmath274 is partitioned by sets indexed by @xmath275 such that the map on @xmath276 , sending any element of the @xmath249th partition set to its @xmath249th return under the geodesic flow to @xmath277 , gives a double cover of @xmath270 .
the union of the @xmath266 is clearly all of @xmath278 , up to a set of measure zero .
we claim that also the union of the @xmath279 is @xmath278 , up to measure zero . to see this
, first note that for @xmath256 , @xmath280 = [ 0 , 1/2)\,.\ ] ] now , let @xmath281 ; thus , for @xmath282 with @xmath283 $ ] , @xmath284 a proof by induction shows that for @xmath285 , @xmath286 where @xmath287 is the fibonacci sequence .
since @xmath288 , we have that @xmath289 but , @xmath290 and @xmath291 .
thus , if @xmath292 is even , we have for @xmath292 , @xmath293\;\ ; & \mbox{if}\;n\ ; \mbox{is even};\\ \\ ( f_n / f_{n+1},f_{n-2}/f_{n-1}]\;\ ; & \mbox{otherwise}. \end{cases}\ ] ] for @xmath294 , the sets @xmath295 are disjoint , and their union is @xmath296\setminus \ { \frac{\sqrt{5}-1}{2}\}$ ] ; since @xmath297 \times [ 0,1]$ ] , the union of the @xmath295 for @xmath298 is indeed @xmath267 , up to measure zero .
with this claim , and the fact that @xmath270 is given by applying @xmath271 to @xmath299 , we have that @xmath270 is bijective , up to measure zero .
since @xmath5 is an invariant measure for @xmath271 , it is also an invariant measure for @xmath270 . since @xmath271 on @xmath267 is a natural extension for @xmath254 , the images under the various @xmath300
, @xmath301 of the pull - back of the borel sigma - algebra on @xmath302 gives the borel sigma - algebra on @xmath267 .
recall ( see the proof of theorem 1 of @xcite on p. 2219
there ) that this holds since points can be separated by the integral powers of @xmath271 ; it is easily show that this is also true for @xmath270 , and from this it follows that @xmath269 does give the natural extension of @xmath303 . finally ,
we have that @xmath267 is double covered by @xmath304 in such a way that on fibers above @xmath266 the @xmath248-return by the geodesic flow projects to give @xmath270 .
an analysis of the proofs above ( combined with using the aforementioned result of sullivan @xcite in fuller generality see the theorem on p. 276 there ) , gives the result of this section .
the rosen continued fractions @xcite are defined as follows .
let @xmath305 and @xmath306 . put @xmath307 and define the map @xmath308 by @xmath309 for @xmath310 , put @xmath311 and as usual , @xmath312 .
furthermore , for @xmath47 with @xmath313 put @xmath314 this yields the _ @xmath0-rosen continued fraction _ of @xmath315 : @xmath316 where @xmath317 and @xmath318 .
+ rosen introduced his continued fractions to study the hecke groups .
the hecke ( triangle fuchsian ) group @xmath319 with @xmath320 is the group generated by @xmath321 with @xmath322 as above .
[ t : rosencf ] for each @xmath323 , the rosen fraction map of index @xmath324 is given as the factor of a cross section of the geodesic flow on the unit tangent bundle of @xmath325 .
due to the general nature of sullivan s result , equation holds for the first return map to any cross section @xmath88 for the geodesic flow on unit tangent bundle of @xmath325 for any @xmath324 . for each @xmath324 ,
burton - kraaikamp - schmidt @xcite determined a planar natural extension on a region @xmath326 with the measure @xmath5 as above .
it is easily verified that the analog of equation holds , and we thus define @xmath327 completely analogously to the @xmath16 above .
the analog of lemma [ l : distinctclasses ] goes through , as for any fixed @xmath324 , all elements of @xmath319 have their entries lying in the ( countable ! ) algebraic number field @xmath328 .
the analog of lemma [ l : flowformula ] holds upon replacing each occurrence of @xmath329 there by the appropriate @xmath329 times @xmath322 .
finally , the result holds , since nakada @xcite showed that the entropy of the rosen continued fraction map equals one - half times the quotient of the volume of the unit tangent space of @xmath325 by the @xmath5-area of @xmath326 .
dajani , kraaikamp and steiner @xcite introduced the @xmath0-rosen fractions , a generalization combining the idea of nakada s @xmath0-continued fractions with rosen s continued fractions .
let @xmath305 and @xmath306 . for @xmath330,$ ]
we put @xmath331 and define the map @xmath332 by @xmath333 for @xmath334 , put @xmath335 and @xmath312 . furthermore ,
for @xmath47 with @xmath313 put @xmath314 the _ @xmath0-rosen continued fraction _ of @xmath22 is then defined in what now is the obvious fashion . fixing @xmath336 results in the rosen fractions . on the other hand , fixing @xmath337 and considering general @xmath338
, we have nakada s @xmath0-expansions .
using direct methods , similar to those of @xcite for the classical rosen fractions , planar natural extensions @xmath339 with invariant measure @xmath5 as above for certain of the @xmath0-rosen fractions are given in @xcite . for each index @xmath324 , kraaikamp - schmidt - smeets @xcite determine the value @xmath340 such that @xmath341 $ ] is the maximal interval containing @xmath342 with @xmath343 being connected for each value of @xmath0 in this interval .
the `` quilting '' of @xmath344 to determine these @xmath339 is then fairly straightforward ; @xcite find a subinterval containing @xmath342 on which entropy is constant , since also @xmath5-measure of @xmath339 is constant .
they also give an argument ( see lemma 12 there ) that easily implies the analog of equation for these values of @xmath324 and @xmath0 . combining these results with nakada s entropy calculation for the classical rosen maps
, we have the following . [ t : rosenalpcf ] for each @xmath345 , let @xmath346 be as in @xcite .
then for each @xmath324 and @xmath347 $ ] the @xmath0-rosen fraction map of index @xmath324 is given as the factor of a cross section of the geodesic flow on the unit tangent bundle of @xmath325 .
indeed , when we consider the special case of @xmath348 , the classical gauss map , we see that one can find a cross - section of the geodesic flow which is a double cover of a surface parametrized by @xmath349 .
this cross - section defines a fundamental domain of the action of @xmath350 on @xmath156 .
-fraction with positive determinant , for @xmath353,title="fig:",width=188 ] -fraction with positive determinant , for @xmath353,title="fig:",width=188 ] -fraction with positive determinant , for @xmath353,title="fig:",width=188 ] a technical difficulty in all this work arises from the fact that the linear maps underlying the transformations @xmath2 are in @xmath354 and not necessarily in @xmath355 ; they can have determinant -1 .
thus the necessity of the 2-fold covering of the natural extension since @xmath356 does not act on the hyperbolic plane , we alternate between the two sheets of the covering when the determinant is -1 .
the only case when this problem does not occur is for @xmath28 .
-fraction with negative determinant , for @xmath353,title="fig:",width=188 ] -fraction with negative determinant , for @xmath353,title="fig:",width=188 ] -fraction with negative determinant , for @xmath353,title="fig:",width=188 ] in that case , the fractional linear maps associated with the transformations are always in @xmath355 , so we will not need a double cover of the natural extension .
the limit cases @xmath358 and @xmath359 both have an indifferent fixed point , which is probably the reason why this family has not been very much studied ( although @xmath359 is known as the _ backwards continued fraction _ ) .
experimental studies indicate that the natural extension behaves much more nicely than for the @xmath0-continued fraction ; see fig .
[ fig : dessextp ] , where we have shown the first 200 000 iterates of the point @xmath360 .
in particular , its domains seems to be connected for all @xmath361 , and any vertical line intersects it in an interval , the domain being bounded by two increasing step functions .
hence its study might be easier than that of the @xmath0-continued fraction .
( we note that nakada s @xcite variant of rosen continued fractions , mentioned in our introduction , also has all associated fractional linear maps of determinant one . ) by contrast , the functions @xmath362 , which are perhaps more natural , seem to be more delicate to study in the general case , see fig . [ fig : dessextm ] , where we show the first 200 000 iterates of the same point ; the corresponding fractional linear maps always have determinant -1 , and for small values of @xmath0 , the domain of the natural extension appears to be disconnected , the vertical sections being cantor - like with nonempty interior . c. bonanno ,
c. carminati , s. isola , and g. tiozzo , _ dynamics of continued fractions and kneading sequences of unimodal maps _ , to appear discrete contin .
series a r. burton , c. kraaikamp , and t.a .
schmidt , _ natural extensions for the rosen fractions _ , tams 352 ( 2000 ) , 12771298 . | we adjust arnoux s coding , in terms of regular continued fractions , of the geodesic flow on the modular surface to give a cross section on which the return map is a double cover of the natural extension for the @xmath0-continued fractions , for each @xmath1 $ ] .
the argument is sufficiently robust to apply to the rosen continued fractions and their recently introduced @xmath0-variants . |
the skin effect problem is one of the most important problems in plasma kinetic theory ( see , for example , in refs.@xcite@xcite ) .
the skin effect in plasma is a response of electron gas to external transverse electromagnetic field .
the problem also has great practical importance .
the solution of the skin effect problem with specular reflection boundary conditions is well known @xcite , @xcite .
the analytic solution of the problem with diffuse reflection boundary conditions has been obtained in the middle of the previous century ( see , for example , ref.@xcite ) . the skin effect problem with general specular
diffuse reflection boundary conditions @xcite is not solved till now .
it s well known that specularity coefficient @xmath0 is a very important factor in the kinetic skin effect theory @xcite @xcite .
the limiting cases @xmath1 ( diffuse surface scattering of electrons ) and @xmath2 ( specular surface scattering of electrons ) are only very special cases .
actually the specularity coefficient @xmath0 equals neither to zero , nor unit , and takes some intermediate values on .
so , for example , in the work @xcite it is shown , that the specularity coefficient @xmath0 is equal to @xmath3 in @xmath4 wire . in this connection
the skin effect problem with specular diffuse boundary conditions has exclusively fundamental significance .
its value is great for theories , and for practical applications .
so it s obvious that the solution of the skin effect problem with general specular
diffuse reflection boundary conditions is a very important task .
the method of solution of this problem for degenerate plasma in metal has been developed in ref .
this method is based on the use of von neumann series .
authors in ref .
@xcite have demonstrated high efficiency of the method developed for computation of the skin effect characteristics .
the goal of this work is generalization of the method developed in ref .
@xcite in the case of gaseous plasma . by method of decomposition of the solution by eigenfunctions of the corresponding characteristic equation exact solutions of the skin effect problem in metal for diffuse and specular boundary conditions
are received in refs .
@xcite , @xcite . in the last years interest to skin effect problems continues to grow ( see , for example , refs .
@xcite@xcite ) . in particular , in ref . @xcite in limiting anomalous skin effect conditions , the oblique electromagnetic wave reflection from the sharp plasma boundary in an assumption of mixed ( specular and diffuse ) electron reflection from the boundary is considered .
let s gaseous ( nondegenerate ) plasma occupy a half space @xmath5 .
the distribution function of electrons @xmath6 is normalized by electron numerical density ( concentration of electrons ) : @xmath7 where @xmath8 is the electron momentum , @xmath9 is the electron mass , @xmath10 is the electron charge , @xmath11 .
we consider electromagnetic wave which propagates in direction orthogonal to the plasma surface .
then the external field has only one @xmath12component .
the internal field inside plasma has only @xmath12component @xmath13 too , where @xmath14 is the field frequency . to describe the electron distribution function we will use the vlasov boltzmann kinetic equation . the collision integral will be represented in the form of @xmath15model @xmath16 where @xmath15 is the time between two electrons collisions , @xmath17 , @xmath18 is the effective electron collision frequency , @xmath19 is the equilibrium maxwellian distribution function , @xmath20 for weak fields this equation may be linearized : @xmath21 here @xmath22 where @xmath23 , @xmath24 , @xmath25 is the dimensionless electron velocity , @xmath26 is the boltzmann costant , @xmath27 is the plasma temperature . for function
@xmath28 we have the following kinetic equation : @xmath29 where @xmath30 in the equation ( 1.2 ) @xmath31 is the dimensionless coordinate , @xmath32 @xmath33 is the dimensionless time , @xmath34 is mean electron free path and @xmath35 is the dimensionless electric field : @xmath36 we will neglect displacement current . then the equation for electric field may be written in the form : @xmath37 where @xmath38 is the electric current , @xmath39\,d^3v .
\eqno{(1.4)}\ ] ] let s extend the electric field and the electric distribution function on the `` negative '' half
space @xmath40 in symmetric manner . for the functions
@xmath41 and @xmath42 we will then write : @xmath43 we may rewrite the equation ( 1.4 ) with use of dimensional parameter @xmath44 : @xmath45 here @xmath46 @xmath47 @xmath48 is the electron thermal velocity , @xmath49 is the penetration length of external electric field for normal skin effect , @xmath50 is the plasma conductivity . by extension procedure ( 1.5 ) on the half
space @xmath40 we may include the surface conditions in the equation for skin effect problem .
specular diffuse boundary conditions on the boundaries of positive and negative half
spaces may be written in the form : @xmath51 @xmath52 where @xmath0 is the specularity coefficient , @xmath53 . in accordance with ( 1.5 ) we obtain : @xmath54 @xmath55 the required function @xmath56 and the electric field must decay away from the surface : @xmath57 we assume that the gradient of the electric field is finite and known at the plasma boundary : @xmath58 here , the gradient of the electric field on the plasma boundary @xmath59 is given .
the variable @xmath31 will be denoted again by @xmath60 .
let s include boundary conditions ( 1.7 ) and ( 1.8 ) in the kinetic equation ( 1.2 ) , and boundary condition ( 1.10 ) include in the electric field equation ( 1.6 ) . as a result
we will obtain system of equation for skin effect in half space of the plasma : @xmath61 @xmath62 @xmath63 @xmath64 the impedance is determined by formula @xcite : @xmath65 with the use of dimensionless field @xmath66 this relation may be rewritten in the form : @xmath67 from equation ( 2.1 ) and boundary conditions ( 1.9 ) we obtain the following expression for @xmath68 : @xmath69 in the case @xmath70 we obtain : @xmath71 then we may rewrite the equation ( 2.1 ) in the form : @xmath72 @xmath73 the solution of the equations ( 2.2 ) and ( 2.3 ) we will seek in the form of fourier integrals : @xmath74 @xmath75 @xmath76 then for the function @xmath77 the following expression may be derived : @xmath78 @xmath79 @xmath80 it s
may be proved , that the expression for @xmath81 coincides with the expression for @xmath77 . therefore we have @xmath82 we substitute the expressions ( 2.4 ) , ( 2.5 ) and ( 2.6 ) into the equations ( 2.2 ) and ( 2.3 ) .
this procedure leads to characteristic system of equations : @xmath83 @xmath84 @xmath85 the function @xmath66 is an even function .
then @xmath86 , and equation ( 2.7 ) may be rewritten as @xmath83 @xmath87 let s substitute the expression ( 2.9 ) into the equation ( 2.8 ) .
then we obtain : @xmath88 @xmath89 here @xmath90 is the dispersion function , @xmath91 @xmath92 and @xmath93 is the integral @xmath94 characteristic system consists of two equations ( 2.9 ) and ( 2.10 ) . the integral @xmath95 we will express through the integral exponential function .
we will spread out integrand partial fractions : @xmath96 where @xmath97 thus , we receive , that @xmath98 or @xmath99 where @xmath100 @xmath101
let s expand the solution of equations ( 2.9 ) , ( 2.10 ) by the following series : @xmath102 @xmath103 @xmath104 functions @xmath105 and @xmath106 may be obtained from the characteristic system . for zero approximation
we have : @xmath107 for first approximation we obtain : @xmath108 @xmath109 for @xmath110-approximation the following expression may be derived : @xmath111 @xmath112 @xmath113 @xmath114 we may rewrite the expressions ( 3.1 ) in the form @xmath115 @xmath116 @xmath117 in general case when @xmath118 we have : @xmath119 @xmath120 therefore the series ( 3.1 ) constructed may be expressed in the explicit form : @xmath121@xmath122.\ ] ]
in accordance with ( 2.4 ) and ( 2.5 ) we will construct expressions for electric field and distribution function . using the expressions ( 3.1 ) and ( 3.2 )
we obtain : @xmath123 @xmath124 we rewrite the expression ( 4.1 ) with the use of ( 3.8 ) in the following form : @xmath125 @xmath126.\ ] ] function @xmath42 may be written in the form : @xmath127 @xmath128 \frac{e^{ikx}dk}{z_0+ik\mu } , \eqno{(4.3)}\ ] ] expression ( 4.3 ) may be written also in the next form : @xmath129 @xmath130dk.\ ] ] if we know function @xmath42 we may write down the electron distribution function @xmath131 according to the equation ( 1.1 ) .
let us consider now the calculation of impedance : @xmath132 @xmath133 we decompose @xmath134 in the following series : @xmath135 here @xmath136 now we write down expressions for zero , first and second approximations for impedance ( see ( 4.4 ) and ( 4.5 ) ) @xmath137 @xmath138 @xmath139 @xmath140 expression for general term of series ( 4.4 ) has the form : @xmath141 @xmath142 @xmath143
in the previous sections we have considered the method , leading to exact solution of the skin effect problem with arbitrary specularity coefficient . in the case @xmath2 the method leads to the classical solution ( 4.6 ) of the problem with specular surface conditions ( see , for example @xcite , @xcite , @xcite ) . in @xcite
this classical solution is represented in the form : @xmath144 where @xmath145 the comparison of the expressions ( 5.1 ) and ( 4.6 ) gives @xmath146 . indeed , after the change of variables in the integral we have
: @xmath147 now let us consider second approximation : @xmath148 when @xmath2 this solution is exact .
maximum deviation from exact solution corresponds to the case when @xmath1 .
the exact solution of the problem in the case @xmath1 is also well known @xcite , @xcite : @xmath149\frac{d\tau}{\tau^2}\bigg]^{-1}. \eqno{(5.2)}\ ] ] it s convenient to rewrite the expression ( 5.2 ) with the use of our notations : @xmath150\,dk\bigg]^{-1}.\ ] ] now consider the ratio of real ( and imaginary ) parts of the solutions constructed in zero , first and second approximations to the solution in zero approximation @xmath151 ( @xmath152 ) for the case @xmath1 .
the last solution coincide with the solution of the problem with specular scattering boundary conditions .
we will build two plots ( curves 1 and 2 ) : @xmath153 and @xmath154 or @xmath155 and also analogous plots for the ratios of the imaginary parts . here
@xmath156 is the solution for the case of specular suface conditions .
values @xmath157 and @xmath158 correspond to corrections for the first and the second approximations . the curves 3 on the plots correspond the ratios of impedance for diffuse scattering surface condition to impedance for specular scattering surface condition @xmath159 ( @xmath160 ) .
the derived method has maximum error in the case of extremely anomalous skin effect , when parameter @xmath161 . in this case
ratios defined above are equal to 1.125 .
so it is obvious from the plots , that in zero approximation the method error is equal to @xmath162 .
for the first approximation in this case we have @xmath163 .
so the first approximation error is equal to @xmath164 . for the second approximation in this case
we have @xmath165 .
and the second approximation error is equal to @xmath166 .
the analysis of plots shows , that the considered impedance ratios in first approximation coincide with the exact solution when @xmath167 . for the second approximation the coincidence is observed when @xmath168 .
the effective method of the solution of boundary problems of the kinetic theory is developed .
this method is based on symmetric continuation of the electric field and distribution function of electrons . on parameter @xmath44 for the case @xmath1 .
the curve @xmath170 corresponds to diffuse scattering boundary conditions , the curves @xmath171 correspond to first and second approximations.,width=321,height=170 ] on parameter @xmath44 for the case @xmath1 .
the curve @xmath170 corresponds to diffuse scattering boundary conditions , the curves @xmath171 correspond to first and second approximations.,width=321,height=170 ] | the problem of skin effect with arbitrary specularity in maxwellian plasma with specular diffuse boundary conditions is solved .
new analytical method is developed that makes it possible to obtain a solution up to an arbitrary degree of accuracy .
the method is based on the idea of symmetric continuation of not only the electric field , but also electron distribution function .
the solution is obtained in a form of von neumann series .
* keywords : skin effect , specular diffuse boundary conditions , analytical method , von neumann series . * = -15 mm |
this document outlines some details from the paper full rotational control of levitated silicon nanorods " .
all trapping frequencies of the nanorods can be calculated , based on their geometry as measured through scanning electron microscopy .
the scattered light signal contains information about all degrees of freedom , with a high signal - to - noise ratio ( snr ) . in the manuscript , all data for @xmath68 is extracted from the scattered light detector d4 , and the motion in @xmath22 is extracted from a polarization dependent measurement on the scattered light d5 .
we independently check all motional degrees of freedom to confirm our assignation of frequencies .
the full optical layout is shown in fig .
[ fig : slayout ] .
* motion in @xmath19 and @xmath20 : * we expect the trapping frequencies for @xmath19 and @xmath20 to be @xmath69khz and @xmath70khz respectively . in the data we see a double - peak in the power spectral density ( psd ) of the scattered light signal d4 around these frequencies , as shown in fig .
[ fig : alpha]a ) . to confirm
experimentally which peak is which we implement an additional detection scheme .
the nanorod rotates the polarization of the trapping light , depending on the angle it makes to the polarization axis ( i.e. by an amount proportional to @xmath20 ) . the full optical layout , including mechanisms for independently measuring different trapping frequencies . a small portion of the free - space light is split on a d - shaped mirror , yielding the radial motion @xmath18 on detector d3 .
a portion of the light that is coupled back into the fibers is picked off , and split equally to be sent to two detectors .
d2 monitors the motion in @xmath19 , as described in ref .
@xcite , and d1 the motion in @xmath20 .
we collect the light which the nanorod scatters using a multimode fiber , and monitor its intensity on detector d4 , and use a lens to collect the scattered light to perform a polarization sensitive measurement in the 45@xmath71 basis on detector d5 . ]
we monitor the trapping light that has interacted with the nanorod by collecting some of the light coupled back into the fiber outcouplers , splitting it off with a 99:1 fiber beamsplitter and turning its linear polarization by 45@xmath71 .
this light then goes through a fiber polarizing beamsplitter ( pbs ) , and each arm of the pbs is coupled onto a fast fiber - coupled balanced detector d1 ( thorlabs pdb420c ) . by this method
we measure just the rotation in @xmath20 , as shown in fig .
[ fig : alpha]a ) .
this confirms that the higher frequency peak is due to motion in @xmath20 .
a ) a comparison of the psd of the signal from the scattered light detector d4 ( red ) , which is sensitive to motion in all directions , and the polarization sensitive detector d1 ( blue ) , which is only sensitive to @xmath20 .
this confirms that the higher frequency peak corresponds to the @xmath20 motion , which agrees with our calculations .
b ) a comparison of the psd from the scattered light detectors d4 ( red ) and d5 ( blue , polarization sensitive ) .
d5 is sensitive to the frequency @xmath61 , which is nt visible on detector d4 . ]
we also monitor the intensity fluctuations of this picked - off light on detector d2 , which yields information about the motion in @xmath19 , as described in ref .
the quality of this signal is poor , due to the large beam waist of our trap , and could be improved through a difference measurement .
it reconfirms that the peak at 124khz is due to motion in @xmath19 . *
motion in @xmath62 : * we pick off a small amount of the light that has interacted with the nanorod using a free - space 97:3 beamsplitter .
this light is incident on a d - shaped mirror to cut the beam in half , and the resulting two beams are measured on a differencing photodiode d3 .
this yields the motion in @xmath21 , as described in ref .
@xcite . by performing a half - waveplate operation on the trapping light
we can rotate the nanorod to measure @xmath18 . due to the large beam waist of our trap ,
the snr is poor for this signal .
* motion in @xmath22 : * using a lens we collect the scattered light emitted in the opposite direction to that collected by the multimode fiber .
this light is sent through a polarizing beamsplitter cube , which is rotated by 45@xmath71 with respect to the @xmath18-axis , and monitored with a photodiode d5 .
this yields information about the motion in @xmath22 , as shown in fig .
[ fig : alpha]b ) .
the snr from the scattered light detector d4 is significantly better than that from detectors d1,2,3 . because of this , even though the @xmath19 and @xmath20 peaks overlap , fitting the psd of d4 is the most accurate method for monitoring the dynamics of the nanorod
however , we can always use the other detectors to confirm our findings .
the snr of the scattered light signal is so good because we only collect light that is scattered by the nanorod , with virtually zero background .
we expect the scattered light signal d4 to be only sensitive to the first harmonic of all motions @xmath72 , since it depends on position squared
. however , fig .
[ fig : alpha]a ) shows that we measure the fundamental motions @xmath73 on d4 .
this is due to a slight misalignment between the trapping beams , and because the polarization in each arm of the trap is not perfectly identical .
we confirm this through calculation , which also confirms that d4 is not sensitive to the fundamental frequency @xmath61 .
the polarization ( rather than intensity ) sensitive detectors d1,5 are sensitive to the sign of the motion in @xmath5 respectively , and so the psds in fig .
[ fig : alpha ] show the frequency @xmath53 and not its harmonic .
to extract parameters such as trapping frequencies and damping rates we fit the psd of the scattered light time series with the function @xmath74 where @xmath75 labels the degrees of freedom ( i.e. @xmath76 ) , @xmath77 is the ( angular ) momentum damping rate , @xmath78 is the trapping frequency , @xmath79 is the particle s mass or moment of inertia and @xmath80 is the calibration between our measured signal and absolute motional information .
the derivation of this psd is standard , e.g. @xcite . when recording the time series of the particle s motion , the psd can usually be calibrated to extract @xmath81 and convert from units of v@xmath82hz@xmath83 to m@xmath82hz@xmath83 .
however , our signal contains both translational ( @xmath4 ) and rotational ( @xmath5 ) information , so such a global calibration is nt possible .
[ t ] the psd of the scattered light signal showing all motional degrees of freedom .
solid red lines are fits to the data using eq .
[ eqn : psd ] . ]
each degree of freedom can be fit with this expression .
figure [ fig : fitting ] shows an example of fitting the psd . due to the proximity of the @xmath19 and @xmath20 peaks in the scattered light psd , we fit the data with a sum of two of the functions defined in eq .
[ eqn : psd ] to extract the parameters for both the @xmath19 and @xmath20 motions .
we expect all trapping frequencies to scale with the square root of power , see eq .
( 1 ) of the manuscript .
figure [ fig : powpres ] shows the measured trapping frequencies as a function of power in comparison to the theoretical expectation .
the only unknown experimental parameter is the laser waist @xmath84 , which we extract from the ratio of @xmath54 to @xmath85 .
the ratio of @xmath54 to @xmath86 confirms the length of our nanorod , which agrees with scanning electron microcope images .
we observe excellent agreement between theory and @xmath87 , and also for the axial frequency when the rod is rotating @xmath88 .
the discrepancy ( @xmath89 ) between theory and experiment for @xmath61 is attributed to the fact that the rods have finite diameter @xmath90 and , the generalized rayleigh - gans approximation @xcite is not strictly valid .
[ t ] variation in the trapping frequencies with trap power , markers are data , solid lines are the theoretical predictions .
the frequency @xmath88 is the trap frequency in the @xmath19 direction when the nanorod is rotating in the @xmath20 direction .
experimental uncertainties are smaller than the data markers . ]
due to the rotation of the rod in the plane orthogonal to the trap axis , only the averaged susceptibility @xmath91 enters the trapping potential of the translational motion in @xmath19-direction .
this means that in eq .
( 1 ) of the manuscript , @xmath66 has to be replaced by @xmath92 .
this agrees well with the measured reduction of the trapping frequency , see fig .
[ fig : powpres ] .
when the trapping light is circularly polarized the nanorod rotates in the plane orthogonal to the trap axis ( @xmath20 direction ) with the frequency @xmath56 . as shown in the paper
, @xmath56 has a broad distribution . to analyze this motion
we extract the instantaneous frequency , using time bins 100 times longer than the mean rotational period . in the paper we present the mean value of @xmath56 and display shaded regions around the data points representing the minimum and maximum values of @xmath56 .
the theoretical analysis predicts the maximum rotation frequency @xmath93 , which is determined by the balance between the torque exerted by the light field and the rotational friction due to collisions with gas molecules .
stickler , b. a. , nimmrichter , s. , martinetz , l. , and kuhn , s. , and arndt , m. and hornberger , k. , `` ro - translational cavity cooling of dielectric rods and disks . '' , phys .
a ( to be published ) ( 2016 ) . | we study a nanofabricated silicon rod levitated in an optical trap . by manipulating the polarization of the light we gain full control over the ro - translational dynamics of the rod .
we are able to trap both its centre - of - mass and align it along the linear polarization of the laser field .
the rod can be set into rotation at a tuned frequency by exploiting the radiation pressure exerted by elliptically polarized light .
the rotational motion of the rod dynamically modifies the optical potential , which allows tuning of the rotational frequency over hundreds of kilohertz .
this ability to trap and control the motion and alignment of nanoparticles opens up the field of rotational optomechanics , rotational ground state cooling and the study of rotational thermodynamics in the underdamped regime . _
introduction _ nanofabrication has advanced all areas of science , technology and medicine @xcite , including the field of optomechanics , where the motion of a mechanical oscillator is controlled by light .
the quantum ground state of motion has been reached in optomechanical crystal devices @xcite , and superconducting microwave circuits @xcite .
ground - state cooling enables the coherent transduction of signals @xcite , the production of non - classical states of light and matter @xcite , and the ultra - sensitive detection of motion @xcite and forces @xcite .
coherent optomechanical technology is limited by the coupling between the mechanical device and its environment , which leads to decoherence of quantum states , and by a reduction in mechanical quality factor due to clamping forces on the oscillator .
these limitations can be overcome by optically levitating the mechanical system , such that it oscillates in a harmonic trapping potential .
optical trapping is applicable from atoms in vacuum @xcite , to complex organisms in liquid @xcite . by optically levitating nanoscale objects in vacuum , ultra - high mechanical quality factors ( @xmath0 )
are predicted @xcite , and it may be possible to generate macroscopic quantum superpositions @xcite .
such massive quantum systems could test the limits of quantum physics @xcite , looking for the existence of new mechanisms of wave - function collapse such as spontaneous localisation @xcite or gravitational effects @xcite .
a ) nanofabricated silicon nanorods of length @xmath1(725 @xmath215)nm and diameter @xmath3 ( 130 @xmath213)nm are optically levitated in a standing laser wave at low pressures .
the light they scatter is collected by a multimode optical fiber placed close to the trap waist .
b ) the rods have five degrees of freedom which can be controlled ; three translational ( @xmath4 ) and two rotational ( @xmath5 ) .
c ) by monitoring the scattered light , trapping of all five degrees of freedom can be observed in the power spectral density ( psd ) .
the appearance of the various harmonics can be explained by slight misalignment of the trap as discussed in the supp .
info.,title="fig:",scaledwidth=46.0% ] the field of levitated optomechanics is growing rapidly , with progress including feedback @xcite and cavity cooling @xcite to the milli - kelvin level and below , the sensing of forces on the zepto - netwon scale @xcite , the detection of rotation @xcite and the study of brownian motion @xcite and equilibration in the underdamped regime @xcite .
experiments are often limited by the quality of commercially available nanoparticles .
impurities lead to absorption of the trapping light , causing loss at low pressures , and even graphitization of levitated diamond @xcite . in this work ,
we trap clean , nanofabricated silicon nanorods , and study their center - of - mass and rotational motion .
our particles are of uniform , tailored size and shape , allowing a high degree of repeatability , predictability and control of the dynamics .
we are able to trap the nanorods , control their orientation , and tuneably spin them by using the radiation pressure exerted by the light field . _
experimental setup _ a single silicon nanorod is optically trapped in the focus formed by two counterpropagating laser beams of wavelength @xmath6 nm , see fig .
[ fig : exp]a ) . at this wavelength
, silicon exhibits a high relative permittivity , @xmath7 , and negligible absorption , which is supported by the fact that we see no signature of heating due to light absorption ( following the method in ref .
@xcite ) .
the nanorods are tailored to have a length of @xmath8 nm and a diameter of @xmath9 nm , corresponding to a mass @xmath10amu .
they are fabricated onto a silicon chip following the methods described in ref .
@xcite .
the laser trap is characterized by a beam waist radius @xmath11 m and the total power @xmath12w , making a large volume trap to enhance the capture efficiency .
the nanorods are trapped in a clean n@xmath13 environment at a pressure of @xmath14 mbar , after being launched by laser desorption from a silicon wafer , see ref .
@xcite .
the rods can be stably trapped for weeks at any pressure above 1 mbar , below which they are lost , as observed in experiments with spherical nanoparticles @xcite .
a ) experimental setup .
light at @xmath15 nm is produced by a fiber laser ( keysight 81663a ) , and then goes through an electro - optical in - fiber polarization controller ( eospace ) , allowing us to realise arbitrary waveplate operations .
the light is amplified in a fiber amplifier ( hangzhou huatai optic ha5435b-1 ) and split equally to make the two arms of the trap .
stress induced birefringence in the fibers can be accounted for with polarization controlling paddles ( pcp ) .
the system is completely fiber - based until out - coupled to the aspheric trapping lenses ( f=20 mm ) .
the inset shows an sem micrograph of a rod that was launched and captured on a sample plate .
the scattered light signal reveals the nanorod dynamics in case of b ) co - linear polarization , and c ) the strongly driven rotation of the rod for circularly polarized trapping light .
, title="fig:",scaledwidth=46.0% ] _ trapping of nanorods _ the motional state of the nanorod is described by its center - of - mass position @xmath16 and by its orientation @xmath17 , see fig .
[ fig : exp]b ) , where @xmath18 points counter - parallel to the direction of gravity and @xmath19 along the beam axis .
the orientation of the rod is parametrized by @xmath20 , the angle between the @xmath18-axis and the projection of the rod onto the @xmath18-@xmath21-plane , and @xmath22 the angle between the rod s symmetry axis and the beam propagation axis .
the motion of the nanorod is measured via the light that the rod scatters out of the trap , which is collected with a 1 mm diameter multimode optical fiber as described in ref .
@xcite .
the polarization of the two trapping beams determines the properties of the optical trap . in the case of co - linear polarization
the rod aligns with the field polarization and is thus trapped in all its degrees of freedom .
the resulting trapping frequencies can be measured in the power spectral density ( psd ) of the scattered light signal , as shown in fig .
[ fig : exp]c ) . using a linb - polarization controller , we can perform arbitrary wave - plate operations on the polarization of the trapping light @xcite . the optical setup ( see fig . [
fig : layout]a ) ) is designed such that the rod experiences the same polarization from both arms of the counterpropagating trap . by realising a half - waveplate operation on the linearly polarized trapping beam
we can align the rod along any direction orthogonal to the trap axis .
the trapping frequencies of a harmonically captured rod can be calculated as @xcite @xmath23 where @xmath24 , @xmath25kgm@xmath26 is the density of silicon , @xmath27 is the susceptibility along the rod s symmetry axis and @xmath28 is the susceptibility anisotropy @xcite . at the maximum input power we measure @xmath29khz , @xmath30khz , @xmath31khz and @xmath32khz ,
see fig .
[ fig : exp]c ) . for comparison ,
a silicon nanosphere of the same volume under the same experimental conditions would have @xmath33khz , and a silica sphere would have @xmath34khz , illustrating the great potential for silicon nanorods in cavity cooling experiments @xcite .
we can use the measured frequencies to deduce the trapping waist radius @xmath35 m , which is the only free experimental parameter .
the measured frequencies agree well with the theoretical expectations , as shown in fig .
[ fig : circ]c ) .
the slight ( @xmath36 ) discrepancy between the measured and predicted value of @xmath37 is attributed to the fact that the rods have finite diameter and the generalized rayleigh - gans approximation @xcite is not strictly valid .
_ spinning the nanorods _ when the trapping light is circularly polarized , the trapping potential for the @xmath20 motion vanishes whilst the standing wave structure along @xmath19 is retained .
the radiation pressure of the laser field exerts a constant torque @xmath38 acting on @xmath20 .
adapting the theory presented in ref .
@xcite , the resulting torque is obtained as @xmath39,\end{aligned}\ ] ] where the two functions @xmath40 are given by @xmath41 for short rods , @xmath42 , one has @xmath43 while @xmath44 .
collisions with residual gas molecules lead to damping of the rotational motion .
since the mean free path of the gas molecules exceeds the diameter of the rod ( free molecular regime @xcite ) , the rotational damping rate for diffuse reflection of gas molecules of mass @xmath45 takes the form @xcite @xmath46 where @xmath47 is the gas temperature .
effect of performing a quarter - waveplate ( qwp ) operation on the trapping light , either increasing from from @xmath48 ( crosses ) or decreasing from @xmath49 ( circles ) . at @xmath48 and @xmath49 the trap
is linearly polarized along the @xmath21-axis . at @xmath50
the polarization is circular .
a ) shift of the trapping frequencies for different qwp settings . for small deviations from linear polarization
the trapping frequencies decrease due to a lower trapping potential . at @xmath51 from the starting linear polarization ,
the light is circularly polarized enough to drive @xmath52 , at which point @xmath53 vanish , and @xmath54 drops . at @xmath55 from the starting linear polarization ,
the motion becomes trapped again .
b ) due to this hysteresis the driven rotational frequency @xmath52 can be tuned over several hundred kilohertz via the ellipticity of the trapping field .
the markers indicate the mean value of @xmath56 , and the shaded region represents the range of measured frequencies.,title="fig:",scaledwidth=46.0% ] the maximum steady - state rotation frequency is obtained by balancing the torque eq . with the damping eq . , @xmath57 this expression agrees well with the measured value of the rotation frequency @xmath56 as a function of power and pressure , as shown in fig .
[ fig : circ]d ) and e ) , respectively . a comparison of the psd for the co - linear and the circular polarization traps is shown in fig .
[ fig : circ]c ) .
the peak related to the trapping frequency at @xmath58 vanishes and a pronounced peak at @xmath59 arises .
we are only sensitive to @xmath59 due to the symmetry of the rod .
the rotation of the rod in the circularly polarized field results in a reduced average susceptibility and thus a weaker trapping potential , which shifts the axial trapping frequency to @xmath60khz as discussed in the supp .
info .
the rapid rotation in @xmath20 leads to a stabilization in @xmath22 and hence the complete suppression of the peak at @xmath61 in fig .
[ fig : circ]c ) .
a similar effect has also been observed for spinning microspheres @xcite .
the broad distribution of frequencies about @xmath59 is due to perturbations temporarily decreasing the rotation rate , which then takes time to spin back up to the maximum value .
for example , irregular excursions in the radial @xmath62 directions lead to variations in the instantaneous rotation frequency , as shown in fig .
[ fig : circ]f ) , with the correlation clearly shown in fig .
[ fig : circ]g ) .
_ tuning the rotational frequency _ to study the effect of driven rotation in more detail we use the polarization controller to perform a quarter - waveplate ( qwp ) operation on the trapping light and track the motion of the rod at each setting , see fig .
[ fig : qwp ] . starting from a linear polarization along @xmath18 and increasing its ellipticity at first leads to a shift of all trapping frequencies to lower values due to a reduced trap depth , as shown in fig .
[ fig : qwp]a ) . at a qwp setting of @xmath51 the radiation pressure induced
torque starts driving the rod into rotation over the trapping potential in the direction of @xmath20 , the frequencies @xmath53 vanish , and @xmath54 drops to a steady value of @xmath63khz , as also seen in fig .
[ fig : circ]c ) . rotating beyond @xmath50
, one may expect the nanorod to become trapped again at @xmath64 , however the rod is not trapped until @xmath55 .
when starting at @xmath49 and decreasing the qwp angle , the rod spins at @xmath64 , and is not trapped until @xmath65 , showing a symmetric hysteresis , see fig [ fig : qwp]b ) .
this effect is due to the anisotropy of the susceptibility tensor : a trapped rod experiences the full trap depth related to @xmath66 whereas the trapping potential for a spinning rod is proportional to rotationally averaged susceptibility @xmath67 , which is smaller by a factor of 1.7 .
thus , it requires a greater torque to spin a trapped rod than to maintain the rotation of an already spinning rod .
the value of @xmath52 varies with the ellipticity of the light , as shown in fig .
[ fig : qwp]b ) . by exploiting the dynamical modification of the trap depth
we can extend the range over which the rotation frequency can be tuned to many hundreds of kilohertz . _
conclusions _ in summary , we present a method to capture and levitate nanofabricated silicon nanorods at low pressures , working with telecom wavelengths in a fiber - based setup .
we can precisely control the length and diameter of our nanorods , meaning we can tailor rods to attain particular trapping and rotational frequencies .
we are able to trap all relevant degrees of freedom , and control the orientation of the rods via the polarization of the trapping beam . by using circularly polarized light we can spin the nanorods at more than 1mhz , and tune this frequency over hundreds of kilohertz by introducing ellipticity into the field polarization and through a dynamic modification of the trapping potential .
when the rod is spinning we notice a stabilization of the tilt angle @xmath22 and a coupling to the radial motion @xmath62 .
the system is very well understood as documented in the excellent agreement between experiment and theory .
the high degree of control opens the way to study rotational optomechanics @xcite and orientational decoherence @xcite and rotational thermodynamics .
this is the first use of silicon in an optical trap , and its high susceptibility and low absorption in this frequency band , combined with the shape - enhanced susceptibility of rods , will enable rotational cavity cooling to the quantum level @xcite . _
acknowledgments _ we are grateful for financial support by the austrian science funds ( fwf ) in the project p27297 and dk - coqus ( w1210 - 3 ) .
we acknowledge support by s. puchegger and the faculty center for nanostructure research at the university of vienna in imaging the nanorods .
j.m . acknowledges the marie skodowska curie action ( nano - q ) .
f.p . acknowledges the legacy program ( israel science foundation ) for its support .
44 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , _ _ ( , , ) . , , , , , , , ,
* * , ( ) . , , , , , , , , , , * * , ( ) .
, , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , ,
* * , ( ) . ,
* * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , , , , pp . ( ) , . , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , pp . ( ) , . , , , , * * , ( ) . , * * , ( ) . , , , * * ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , pp . ( ) , . , , , , * * , ( ) . , , , , * * , ( ) .
, , , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , ( ) . ,
_ _ ( , ) . ,
_ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . ,
( ) . , , , ( ) . supplementary information + |
baryons made of two or three heavy quarks offer an interesting alternative to quarkonium for studying the dynamics of non - relativistic systems in qcd and for investigating the transition region from coulombic to confined bound states . the modern approach to quarkonium physics
consists in taking advantage of the hierarchy of non - relativistic energy scales in the system by constructing a suitable hierarchy of effective field theories ( efts ) @xcite .
the energy scales are the heavy - quark mass , @xmath0 , the typical momentum transfer , @xmath1 , whose inverse sets the typical distance , @xmath2 , between the heavy quark and the antiquark , and the typical kinetic energy , @xmath3 , whose inverse sets the typical time scale of the bound state . in the ultimate eft ,
obtained after integrating out gluons of energy and momentum of the order of @xmath0 and @xmath4 , the interaction between heavy quarks is organized as an expansion in powers of @xmath5 and @xmath2 . at zeroth - order in @xmath2 , the interaction
is entirely encoded in the quark - antiquark potential , which , at zeroth - order in @xmath5 , reduces to the static potential .
terms proportional to powers of @xmath5 and @xmath2 can be systematically added .
an analogous hierarchy of energy scales characterizes also baryons that contain at least two heavy quarks .
hence , we may describe these systems by means of efts analogous to the ones suited for heavy quarkonium @xcite .
in particular , the ultimate eft , obtained after integrating out gluons of energy and momentum of the order of the heavy - quark masses and of the typical momentum transfer between heavy quarks , is organized as an expansion in the inverse of the heavy - quark masses and in the distance between the heavy quarks . for equal heavy - quark masses ,
the structure of the lagrangian and its power counting are similar to the quarkonium case ; the lagrangian reads @xmath6 where @xmath7 are terms containing the heavy quark or antiquark fields , which are proportional to @xmath8^j$ ] .
the fields @xmath9 are @xmath10 light - quark fields , assumed to be massless . the heavy quark or antiquark fields in @xmath7
are twice those in the bound state , e.g , in the quarkonium case , @xmath7 contains two quark and two antiquark fields . for different heavy - quark masses ,
more scales are involved ; in the following , we will not consider such cases .
although the efts for quarkonium and baryons made of two or three heavy quarks are similar in structure , they are characterized by different degrees of freedom .
this is best seen when the typical distance between heavy quarks is smaller than the inverse of the hadronic scale @xmath11 , which is the case that we will discuss in this note . at these distances , gluons may resolve coloured degrees of freedom .
these are gluons or light quarks or , in the quarkonium case , @xmath12 states in a colour singlet or in a colour octet configuration . in the case of baryons made of two heavy quarks @xmath13 and a light quark @xmath14 , at distances smaller than @xmath15
, gluons can resolve gluons , light quarks and @xmath16 pairs in a colour antitriplet or in a colour sextet configuration .
it is the binding of the antitriplet with the light quark @xmath14 that forms the @xmath17 baryon .
the system very much resembles a heavy - light meson , with the heavy antiquark replaced by a @xmath16 antitriplet ; this fact may be exploited to deduce some properties of the @xmath17 baryons from the corresponding @xmath18 mesons @xcite . in the case of baryons made of three heavy quarks @xmath13 , at distances smaller than @xmath15
, gluons can resolve gluons , light quarks and @xmath19 states either in a colour singlet or in two different colour octets or in a colour decuplet configuration .
the lattice evaluation of the @xmath19 static potential has a long tradition ( see e.g. @xcite and references therein ) , while the static potential between a @xmath16 pair in the presence of a light quark has been evaluated on the lattice only recently @xcite .
expressions for the @xmath5 and the @xmath20 spin - dependent @xmath19 potentials in terms of wilson loops can be found in @xcite , but have not been calculated on the lattice yet ( while the complete expressions of all the @xmath12 potentials up to order @xmath20 can be found in @xcite and the most recent lattice determinations are in @xcite ) .
perturbative studies of the static potential may help to understand the transition region from the perturbative to the non - perturbative regime . in the case of the @xmath12 potential , this region is characterized by the smooth transition from a coulomb potential to a linear raising one ( for recent studies , see @xcite ) . in the case of the @xmath19 static potential ,
the transition from the perturbative to the non - perturbative regime is accompanied by the emergence of a three - body potential that depends on one length only ( see , for instance , @xcite ) .
this is a rather spectacular phenomenon , which has been investigated only recently from a perturbative perspective @xcite .
there is so far no experimental evidence of @xmath19 baryons , while a few years ago the selex experiment has claimed evidence of possible doubly charmed baryon states @xcite . until today
this evidence has not been confirmed by other experiments , but is mostly behind the revival of interest in this kind of systems at the mid of this decade .
the eft lagrangian that describes @xmath17 baryons below the momentum transfer scale @xmath4 , assumed to be larger than @xmath11 , has the general form of eq . , with @xmath7 made out of four quark fields @xcite .
the term @xmath21 is @xmath22 t + \sigma^\dagger \left [ id_0 - v^{(0)}_\sigma \right ] \sigma,\ ] ] where @xmath23 are the three independent @xmath16 antitriplet fields , @xmath24 @xmath25 are the six independent @xmath16 sextet fields and the gauge fields in the covariant derivatives acting on the antitriplet and sextet fields are understood in the @xmath26 and @xmath27 representations respectively .
the matching coefficients @xmath28 and @xmath29 can be identified with the antitriplet and sextet static potentials respectively .
the @xmath16 antitriplet static potential @xmath28 has been calculated recently up to next - to - next - to - leading order ( nnlo ) @xcite ; it reads @xmath30 \right .
\nonumber\\ & & \hspace{5 mm } \left .
+ \left(\frac{{\alpha_{\rm s}}}{4\pi}\right)^2 \left[66 \zeta(3)+484 \gamma_e^2 + \frac{1976}{3}\gamma_e + \frac{3}{4}\pi^4 + \frac{121}{3}\pi^2 + \frac{4343}{18 } \right.\right . \nonumber\\ & & \hspace{25 mm } -\left(\frac{52}{3 }
\zeta(3 ) + \frac{176}{3}\gamma_e^2 + \frac{916}{9}\gamma_e + \frac{44}{9}\pi^2 + \frac{1229}{27}\right)n_f \nonumber\\ & & \hspace{25 mm } \left.\left .
+ \left(\frac{16}{9}\gamma_e^2 + \frac{80}{27}\gamma_e + \frac{4}{27}\pi^2 + \frac{100}{81}\right)n_f^2 \right ] \right\},\end{aligned}\ ] ] where @xmath31 is the strong - coupling constant in the @xmath32 scheme .
the @xmath16 sextet static potential @xmath29 is repulsive at leading order : @xmath33 .
several terms contribute to @xmath34 .
the one responsible for the hyperfine splitting is @xmath35 where @xmath36 is a matching coefficient of the eft , @xmath37 is the matching coefficient of the chromomagnetic interaction in the heavy quark effective theory , which is known up to three loops @xcite , @xmath38 is a pauli matrix acting on the heavy quark labeled @xmath39 and @xmath40 are the gell - mann matrices in the @xmath26 representation . from eq . , it follows that the hyperfine splitting between the @xmath41-wave ground state of a doubly heavy baryon of spin @xmath42 ( @xmath43 ) and the corresponding state of spin @xmath44 ( @xmath45 ) may be related to the hyperfine splitting between the @xmath41-wave ground state of a heavy - light meson of spin 0 ( @xmath46 ) and the corresponding state of spin 1 ( @xmath47 ) : @xmath48,\end{aligned}\ ] ] where we have kept different the mass , @xmath49 , of the heavy quarks in the baryons from the one , @xmath50 , in the mesons .
the obtained figures compare well with existing lattice determinations ( see discussion and references in @xcite , a more recent unquenched lattice determination of @xmath51 , which is consistent with previous quenched determinations , may be found in @xcite ) .
other terms in the effective lagrangian have been derived in @xcite .
the eft lagrangian that describes @xmath19 baryons below the momentum transfer scale @xmath4 , assumed to be larger than @xmath11 , has the general form of eq . , with @xmath7 made out of six quark fields @xcite .
the term @xmath21 is @xmath52 s + o^{\dagger } \left [ id_0 - v^{(0)}_{o } \right ] o + \delta^\dagger \left [ id_0 - v^{(0)}_{\delta } \right ] \delta , \ ] ] where @xmath41 is the singlet field , @xmath53 , with @xmath54 and @xmath55 , are the fields that parameterize the two possible octet configurations of three quarks , @xmath56 @xmath57 are the ten independent @xmath19 decuplet fields and the gauge fields in the covariant derivatives acting on the octets and decuplet fields are understood in the @xmath58 and @xmath59 representations respectively .
the matching coefficients @xmath60 , @xmath61 and @xmath62 can be identified with the singlet , octet and decuplet static potentials respectively ; note that the octet potential @xmath61 is a @xmath63 matrix . the @xmath19
singlet static potential @xmath60 has been calculated up to nnlo in @xcite ; it reads @xmath64 \right .
\nonumber\\ & & \hspace{5 mm } \left .
+ \left(\frac{{\alpha_{\rm s}}}{4\pi}\right)^2 \left[66 \zeta(3)+484 \gamma_e^2 + \frac{1976}{3}\gamma_e + \frac{3}{4}\pi^4 + \frac{121}{3}\pi^2 + \frac{4343}{18 } \right.\right .
\nonumber\\ & & \hspace{25 mm } -\left(\frac{52}{3 } \zeta(3 ) + \frac{176}{3}\gamma_e^2 + \frac{916}{9}\gamma_e + \frac{44}{9}\pi^2 + \frac{1229}{27}\right)n_f \nonumber\\ & & \hspace{25 mm } \left.\left .
+ \left(\frac{16}{9}\gamma_e^2 + \frac{80}{27}\gamma_e + \frac{4}{27}\pi^2 + \frac{100}{81}\right)n_f^2 \right ] \right\ } \nonumber\\ & & \!\
! + v^\mathrm{3body}_s({\boldsymbol{r}}_1,{\boldsymbol{r}}_2,{\boldsymbol{r}}_3),\end{aligned}\ ] ] where @xmath65 , @xmath66 and @xmath67 are the distances between the heavy quarks ; only two of them are independent : if we call @xmath68 , @xmath69 and @xmath70 the coordinates of the three quarks , our choice is @xmath71 , @xmath72 and @xmath73 , which implies that @xmath74 .
@xmath75 is the three - body part of the perturbative potential , defined as the part of the potential that vanishes when putting one of the quarks at infinite distance from the other two .
the three - body part is a specific feature of the two - loop potential ; it reads @xmath76,\ ] ] where @xmath77 \nonumber\\ & & + 16\pi \hat{{\mbox{\boldmath $ \rho$}}}^i\hat{{\mbox{\boldmath $ \lambda$}}}^j \int_0 ^ 1dx\int_0 ^ 1dy \ , \frac{\hat{{\boldsymbol{r}}}^i\hat{{\boldsymbol{r}}}^j}{r }
\left[\left(1 + 3\frac{m^2}{r^2}\right)\arctan\frac{r}{m}-3\frac{m}{r}\right],\end{aligned}\ ] ] with @xmath78 , @xmath79 and @xmath80 .
this term has a different dependence on the positions of the three quarks with respect to the coulomb potential .
it is finite for all configurations : it vanishes when one of the quarks is pulled at infinite distance ( note that in this case @xmath60 becomes @xmath28 ) and still remains finite when two quarks are put in the same position ( note that in this case @xmath60 becomes the @xmath12 singlet static potential ) .
hence , its dependence on the geometry is much smoother than the dependence on the geometry of the coulomb potential , although it still clearly depends on it .
these features can be seen by plotting @xmath75 as a function of @xmath81 for two different set of configurations ; @xmath81 is the sum of the distances of the three quarks from the so - called fermat ( or torricelli ) point , which has minimum distance from the quarks .
figure [ fig : threebody](a ) shows the three - body potential when we place the three quarks in a plane @xmath82 , fix the position of the first quark in @xmath83 , the position of the second one in @xmath84 and move the third one in @xmath85 for @xmath86 and @xmath87 .
the plot shows a clear dependence on the geometry at fixed @xmath81 ( i.e. different configurations with same @xmath81 give different potentials ) , although weaker than in the coulombic case .
figure [ fig : threebody](b ) shows the three - body potential in a geometry used in @xcite , which consists in placing the three quarks along the axes of a three - dimensional lattice : @xmath88 , @xmath89 and @xmath90 , for @xmath91 and @xmath92 .
the plot shows a weaker dependence on the geometry than in the previous case . as the comparison with fig .
[ fig : threebody](a ) indicates , the weaker dependence is an artifact of the chosen configurations rather than a physical effect . clearly , the precise identification of the transition region from a two - body dominated potential to a three - body dominated one would require to perform lattice calculations in geometries different from the one used in fig .
[ fig : threebody](b ) .
the use of different geometries could also be important to assess the nature of the long - range three - body potential .
( 8,0)=5.5truecm ( 188,0)=5.5truecm ( 0,90)*(a ) * ( 180,90)*(b ) * the one - gluon exchange mixes the octet fields , so that the octet potential @xmath61 is a non - diagonal @xmath63 matrix already at leading order .
choosing @xmath93 and @xmath94 to be respectively symmetric and antisymmetric for exchanges of the quarks located in @xmath68 and @xmath69 ( a different choice of the octet fields would correspond to a field redefinition leading to a different octet potential ) , we obtain @xmath95,\ ] ] while at leading order the decuplet potential is @xmath96 we mention that there exist lattice data that show very clearly the singlet , octet and decuplet @xmath19 static potentials , although in an equilateral geometry , where the two octets are degenerate @xcite .
other terms in the effective lagrangian have been derived in @xcite .
systems made of two or three heavy quarks or antiquarks develop similar hierarchies of energy scales and may be treated in similar eft frameworks . in the case of @xmath12 mesons ,
the static potential has been determined up to next - to - next - to - next - to - leading order in perturbation theory and to a high accuracy on the lattice .
the coulomb behaviour starts getting substantial modifications at distances around 0.2 fm turning over a linearly raising potential at larger distances .
therefore the transition region can be studied to a large extent with perturbative methods .
terms proportional to powers of @xmath5 and @xmath2 have been also calculated and included systematically in physical observables . in the case of @xmath17 baryons ,
the static potential has been determined up to nnlo in perturbation theory and recently also on the lattice .
terms proportional to powers of @xmath5 and @xmath2 in the lagrangian have been matched ( mostly ) at leading order and used to determine , for instance , the expected hyperfine splitting of the ground state of these systems .
if early experimental evidences will get confirmation in future experimental facilities , this will mark the beginning of a future new spectroscopy .
finally , in the case of @xmath19 baryons , the static potential has been determined up to nnlo in perturbation theory and also on the lattice .
the transition region from a coulomb to a linearly raising potential is characterized in this case also by the emergence of a three - body potential apparently parameterized by only one length .
while we have argued that more lattice studies employing different geometries would be necessary to precisely identify the transition region , we have also shown that in perturbation theory a smooth genuine three - body potential shows up at two loops .
n. brambilla , a. pineda , j. soto and a. vairo , `` effective field theories for heavy quarkonium , '' rev .
phys . * 77 * , 1423 ( 2005 ) [ arxiv : hep - ph/0410047 ] .
n. brambilla , t. rsch and a. vairo , `` effective field theory lagrangians for baryons with two and three heavy quarks '' , phys .
d * 72 * , 034021 ( 2005 ) [ arxiv : hep - ph/0506065 ] .
s. fleming and t. mehen , `` doubly heavy baryons , heavy quark - diquark symmetry and nrqcd , '' phys .
d * 73 * , 034502 ( 2006 ) [ arxiv : hep - ph/0509313 ]
. m. j. savage and m. b. wise , `` spectrum of baryons with two heavy quarks , '' phys .
b * 248 * , 177 ( 1990 ) .
t. t. takahashi , h. matsufuru , y. nemoto and h. suganuma , `` the three - quark potential in the su(3 ) lattice qcd , '' phys .
lett . * 86 * , 18 ( 2001 ) [ hep - lat/0006005 ] ; `` detailed analysis of the three quark potential in su(3 ) , '' phys .
d * 65 * , 114509 ( 2002 ) [ hep - lat/0204011 ] ; a. yamamoto , h. suganuma and h. iida , `` heavy - heavy - light quark potential in su(3 ) lattice qcd , '' phys .
b * 664 * , 129 ( 2008 ) [ arxiv:0708.3610 [ hep - lat ] ] ; `` lattice qcd study of the heavy - heavy - light quark potential , '' phys . rev .
d * 78 * , 014513 ( 2008 ) [ arxiv:0806.3554 [ hep - lat ] ] ; j. najjar and g. bali , `` static - static - light baryonic potentials , '' arxiv:0910.2824 [ hep - lat ] .
n. brambilla , a. pineda , j. soto and a. vairo , `` the qcd potential at @xmath97 , '' phys .
d * 63 * , 014023 ( 2001 ) [ arxiv : hep - ph/0002250 ] ; a. pineda and a. vairo , `` the qcd potential at @xmath98 : complete spin dependent and spin independent result , '' phys . rev .
d * 63 * , 054007 ( 2001 ) [ erratum - ibid .
d * 64 * , 039902 ( 2001 ) ] [ arxiv : hep - ph/0009145 ] ; n. brambilla , a. pineda , j. soto and a. vairo , `` the @xmath99 scale in heavy quarkonium , '' phys .
b * 580 * , 60 ( 2004 ) [ arxiv : hep - ph/0307159 ] .
y. koma , m. koma and h. wittig , `` nonperturbative determination of the qcd potential at @xmath97 , '' phys .
lett . * 97 * , 122003 ( 2006 ) [ arxiv : hep - lat/0607009 ] ; y. koma and m. koma , `` spin - dependent potentials from lattice qcd , '' nucl .
b * 769 * , 79 ( 2007 ) [ arxiv : hep - lat/0609078 ] ; y. koma , m. koma and h. wittig , `` relativistic corrections to the static potential at @xmath97 and @xmath100 , '' pos * lat2007 * , 111 ( 2007 ) [ arxiv:0711.2322 [ hep - lat ] ] ; y. koma and m. koma , `` scaling study of the relativistic corrections to the static potential , '' arxiv:0911.3204 [ hep - lat ] . n. brambilla , x. garcia i tormo , j. soto and a. vairo , `` precision determination of @xmath101 from the qcd static energy , '' arxiv:1006.2066 [ hep - ph ] ; `` the qcd static energy at nnnll , '' phys .
d * 80 * , 034016 ( 2009 ) [ arxiv:0906.1390 [ hep - ph ] ] .
n. brambilla , j. ghiglieri and a. vairo , `` the three - quark static potential in perturbation theory , '' phys .
d * 81 * , 054031 ( 2010 ) [ arxiv:0911.3541 [ hep - ph ] ] .
m. mattson [ selex collaboration ] , `` first observation of the doubly charmed baryon @xmath102 , '' phys .
* 89 * , 112001 ( 2002 ) [ arxiv : hep - ex/0208014 ] ; a. ocherashvili _ et al . _
[ selex collaboration ] , `` confirmation of the double charm baryon @xmath103 via its decay to @xmath104 , '' phys .
b * 628 * , 18 ( 2005 ) [ arxiv : hep - ex/0406033 ] .
a. g. grozin , p. marquard , j. h. piclum and m. steinhauser , `` three - loop chromomagnetic interaction in hqet , '' nucl . phys .
b * 789 * , 277 ( 2008 ) [ arxiv:0707.1388 [ hep - ph ] ] .
r. lewis and r. m. woloshyn , `` bottom baryons from a dynamical lattice qcd simulation , '' phys .
d * 79 * , 014502 ( 2009 ) [ arxiv:0806.4783 [ hep - lat ] ] .
k. hbner , f. karsch , o. kaczmarek and o. vogt , `` heavy quark free energies for three quark systems at finite temperature , '' phys .
d * 77 * , 074504 ( 2008 ) [ arxiv:0710.5147 [ hep - lat ] ] . | baryons made of two or three heavy quarks can be described in the modern language of non - relativistic effective field theories .
these , besides allowing a rigorous treatment of the systems , provide new insight in the nature of the three - body interaction in qcd . |
the study of the cosmic microwave background ( cmb ) anisotropies is providing strong constraints on theories of structure formation .
these theories are statistical in essence , so the extraction of the information must be done in a statistical way . in particular ,
the standard method for analysing a cmb experiment is the maximum likelihood estimator ( ml ) .
the procedure is straightforward : maximise the probability of the parameters of the model given the data , p(parameters@xmath2data ) , over the allowed parameter space .
usually , we take the prior probability for the parameters to be constant , so this is equivalent to maximising the likelihood , p(data@xmath2parameters ) , via the bayes theorem .
the ml method has been widely applied in cmb analyses , for power spectrum or parameters estimation , @xcite . when computing the likelihood in these problems , we have to deal with the inversion of the covariance matrix of the data , which usually involves @xmath3 operations , being @xmath4 the number of pixels of the map .
the increasing size of the datasets makes this method computationally costfull for new experiments , so other methods have been investigated in the last few years to confront the problem
. there have been several proposals on this matter .
pioneering work on the problem of power spectrum estimation @xcite , based on an evaluation of the @xmath5 s coefficients of the multipole expansion of the observed map in the spherical harmonics basis , have been applied to cobe data @xcite .
quadratic estimators have been proposed by several authors @xcite as statistics that give the same parameters that maximise the likelihood , but requiring less computational work .
nevertheless , alternative statistical methods are required in the field to extract the cosmological information from future cmb experiments ( as planck ) where the number of data points will be very large ( see , e.g. @xcite for an estimation of the scaling of the computing time with the dataset size ) . here , we propose a new statistical method to analyse a cmb map . in order to illustrate it
, we will use the two - point correlation function ( cf ) .
we first replace the likelihood of the full map by the likelihood of the fluctuations of an estimator of the cf .
then , we derive the cosmological parameters from it in an efficient manner .
if we assume gaussianity for the primordial cmb fluctuations , the cf completely characterises the statistical properties of the field . in this line , it has been suggested @xcite that it can be used to obtain the power spectrum or for parameter estimation , because it encodes all the relevant information for that purpose .
this approach of considering the cf in cmb analyses has been recently used by other authors @xcite to estimate the power spectrum .
they obtain the cf using different estimators , and integrate it , projecting over the legendre polynomials , to obtain the @xmath6 s .
the advantage of this estimator is that it only needs at the most @xmath7 operations to be computed , and not @xmath3 , as is required for the likelihood .
we construct a modified version of the standard @xmath0 test , using the cf evaluated at a certain set of points .
the estimate of the parameters of the model is given by the minimum of this statistic , as in the standard analysis .
we give a very good approximation to the distribution function of this modified @xmath0 , so the confidence limits can be obtained without using simulations , by integration below that curve , as for the ml .
we show that , in several problems , choosing a large enough set of points to evaluate the cf , our method has the same power as the maximum likelihood , while being two different methods .
in this section we will introduce the test , using for this purpose the two - point cf .
nevertheless , all the procedure described below can be applied to any other estimator . for a certain map of the cmb anisotropies , @xmath8 with @xmath4 pixels , and errors
@xmath9 , we can estimate the cf , @xmath10 , in a set of @xmath11 angular distances , @xmath12 . in this work
we have used the following estimator , @xmath13 = \frac { \sum _ { i , j \in \ { k \ } } \ ; x_i x_j } { \sum _ { i , j \in \ { k \ } } 1 } \label{est_cf}\ ] ] where @xmath14 stands for the set of all pixel pair @xmath15 such that their angular distance is @xmath16 , but our proposal and techniques can be applied to other estimators for the cf ( see , for example , @xcite , or @xcite ) .
hereafter , we will write the estimate of a certain parameter as @xmath17 $ ] . if we have a map with zero mean and no noise , eq .
( [ est_cf ] ) is a unbiased estimator of the theoretical two - point cf , which for a experiment with a symmetric beam is given by @xmath18 where @xmath19 stands for the window function of the experiment .
we have explicitly removed the dipole contribution in the previous equation , because these coefficients are dominated by the kinematic dipole in a real map .
if the cmb signal is gaussian , the power spectrum ( or its fourier transform , the cf ) , encodes all the information about the model .
so , under this assumption , we can parameterise a model through the @xmath6 s themselves , or through the cosmological parameters , by writing @xmath20 . in general , we will write @xmath21 , being @xmath22 the parameters of the model . as an example of the general procedure that we propose in this paper , we consider in detail the estimator derived from the following statistic : @xmath23 - c(\theta_k|\hat{m } ) - c_n(\theta_k ) ) ^2 } { \sigma^2 ( c(\theta_k ) ) } \label{chi2}\ ] ] where @xmath24 s are certain weights to be defined below , and @xmath25 stands for the variance of the estimator @xmath26 $ ] .
@xmath27 represents the discrete cf of the noise .
if we have an experiment with uncorrelated noise , this function takes the form @xmath28 this method is a modification of the standard form of a @xmath0-test , for the case when the error of each of the estimates entering ( [ chi2 ] ) are independent and gaussianly distributed ( hereafter , we mean by standard @xmath0-test the case when @xmath29 , @xmath30 , and all the terms of the sum in ( [ chi2 ] ) are independent ) . in the present case , the @xmath31 $ ] quantities follow very closely a gaussian , but are correlated .
for this reason , we have introduced some weights ( @xmath24 ) , that will be determined by minimising the dispersion of the estimator derived from equation ( [ chi2 ] ) , as we will see in the next section .
those quantities will account for the different degree of correlation between terms , and in a general problem , they will be a function @xmath32 , where @xmath33 stands for the correlation matrix between the errors of the estimates of @xmath34 and @xmath35 , i.e. @xmath36 - c(\theta_i|\hat{m } ) - c_n(\theta_i ) \bigg ) \times \\ \times \bigg ( e [ c(\theta_j ) ] - c(\theta_j|\hat{m } ) - c_n(\theta_j ) \bigg ) > \label{cprima}\end{aligned}\ ] ] so the variance @xmath25 is related to the @xmath37-matrix by @xmath38 the brackets @xmath39 represent an average over an ensemble of universes , i.e. , an average over realizations for one fixed cmb model . in principle
, our construction seems to miss information about the correlations when compared to the usual @xmath0 procedure to analyse correlated datasets , because in equation ( [ chi2 ] ) only the diagonal terms of the covariance matrix ( @xmath37 ) are explicitly shown . nevertheless , as we will see in the next section , the @xmath24 weights depend on the full covariance matrix and not on its inverse ,
so all the correlations implicitly enter in that expression .
we will not make a detailed comparison between the usual @xmath0 method with uses the full covariance matrix c ( we will refer to this method as the `` usual @xmath0 '' ) and our @xmath1 .
however , we will illustrate this with an example ( see section 6.1 ) .
in addition , in appendix a we present a brief comparison of some characteristics of both methods ( @xmath1 and the usual @xmath0 ) for the case of linear problems .
it is an interesting result that , for gaussian linear problems , if we are estimating only one parameter , the estimates from both methods are exactly equal , while being different statistics ( i.e. they will give different probability contours ) .
hereafter , we will concentrate in our method , and its application to cmb problems .
it is worth to notice that our @xmath1 statistic is a different approach to the ml , in the sense that it provides different estimates and probability contours .
however , in the case of cmb analyses , we will see below that it has a similar power to the ml , but avoiding the problem of the inversion of the covariance matrix .
there is however a minor sense in which our test may formally be considered as an approximation to the ml .
it is well - known that the ml is an asymptotically efficient estimator for our problem .
thus , in the limit of infinite size of the sample ( or for those problems where an efficient estimator exists , as in linear gaussian problems ) , then the ml is the only one statistic which renders the minimum variance , and thus any other estimator may be regarded as approximate .
once we have constructed the @xmath1 test , the estimate for this method is given by the set of values for the parameters that minimise eq .
( [ chi2 ] ) , i.e. the solution to the set of equations @xmath40 . for a given problem ,
we proceed as follows .
if we want to estimate a set of @xmath41 parameters , @xmath42 , we first compute the @xmath37 matrix by assuming an initial value for those parameters , @xmath43 . using this matrix , we obtain our estimate by solving the following system of equations , @xmath44 when computing these derivatives with respect to @xmath22 , we neglect the dependence of @xmath37 on the parameters , which is equivalent to assuming that @xmath25 and @xmath24 are constants in the derivation .
this process is iterated until convergence .
the reason to keep the @xmath37 matrix fixed in the derivation is that we want to have an unbiased estimator of the parameters .
the evaluation of the @xmath37 matrix can be done by monte carlo simulations , but we also propose an analytical approach .
it is possible to evaluate equation ( [ cprima ] ) using the quantities @xmath45 , for a multivariate - gaussian field , as is the case for the cmb ( see appendix b ) . in the particular case of power spectrum estimation , the set of parameters we have to determine are the @xmath6 s themselves , or the band powers in a certain number of multipole bands centred at multipoles @xmath46 ( i.e. , @xmath47 ) .
as the theoretical cf ( [ cf_th ] ) is linear in the @xmath6 s , we have that eq . ( [ ps_est ] ) is a linear system of @xmath41 equations , and the solution can easily be found . as an example , we present here the equations for the determination of the total power measured by a certain experiment . in this case
, we only have one parameter , @xmath48 , defined as @xmath49 which is essentially a normalisation of the spectrum . from here
, we define the function @xmath50 , which is independent of @xmath51 .
we can now obtain the analytic expression for the estimate of @xmath51 by minimising eq .
( [ chi2 ] ) with respect to @xmath51 , which in this case takes the form @xmath52 - \sigma_{sky}^2 f(\theta_k ) ) ^2 } { \sigma^2 ( c(\theta_k ) ) } \ ] ] for simplicity , we will not write the term of the noise cf , but it can be easily included inside the true cf . inserting the previous expression in equation ( [ ps_est ] ) ,
we obtain , for fixed @xmath24 , an equation for @xmath51 .
it must be noted that due to the dependence of @xmath25 on @xmath51 , this equation is not exactly linear .
we could solve it iteratively , starting with certain fixed value for @xmath25 .
however , for all the values of these quantities within the current limits , a first iteration is enough , as we will see , so that the equation determining @xmath51 is effectively linear , and therefore its solution is given by @xmath53 = \bigg ( \sum_i \frac { p_i f_i^2 } { \sigma^2 ( c(\theta_i ) ) } \bigg ) ^{-1 } \sum_i p_i f_i \frac { e [ c(\theta_i ) ] } { \sigma^2 ( c(\theta_i ) ) } \label{e_ssky}\ ] ] the @xmath54 of this estimator is given by @xmath55 ) = \bigg ( \sum_i \frac { p_i f_i^2 } { \sigma^2 ( c(\theta_i ) ) } \bigg ) ^{-1 } \times \\ \qquad \times \bigg [ \sum_i \frac{p_i^2 f_i^2}{\sigma^2 ( c(\theta_i ) ) } + \sum_{i \neq j } \frac { f_i f_j p_i p_j c'_{ij } } { \sigma^2 ( c(\theta_i ) ) \sigma^2 ( c(\theta_j ) ) } \bigg]^{1/2 } \label{rms_ssky}\end{aligned}\ ] ] where we have defined @xmath56 .
the expression within large parentheses in this equation is the @xmath54 of the second sum in equation ( [ e_ssky ] ) .
the first sum within the parentheses correspond to the quadratic addition of the contributions of each term in ( [ e_ssky ] ) , which is present even when the random variables @xmath31 $ ] are independently distributed .
the second sum is due to the correlations between any pair of these variables .
it should be noted that equation ( [ rms_ssky ] ) has been obtained assuming that the quantities @xmath31 $ ] follow a multivariate gaussian distribution .
this is a good approximation if there are enough pixel pairs entering in the sum in ( [ est_cf ] ) ( see , for example , @xcite , for the cf of the cobe data ) .
similar calculations for the standard @xmath0 and the likelihood function can be found in @xcite ( hereafter , b93 ) .
the matricial expression of the estimate and the rms for a general linear problem are shown in appendix a.
the @xmath24 s weights in equation ( [ chi2 ] ) are introduced in order to take into account the different degree of correlation of the terms of the sum .
their expression can be obtained once we define exactly what we are interested in .
for example , one common criteria for one parameter estimation is to use the estimator which has the minimum @xmath54 .
we will consider this criteria here .
for the problem of one parameter estimation described in the previous section , once we have the analytic expression for the @xmath54 , and an initial guess for the @xmath37 matrix , we can obtain the optimum set of @xmath24 s using the `` minimum @xmath54 criteria '' .
we minimise eq .
( [ rms_ssky ] ) with respect to the @xmath24 s quantities .
we obtain that the @xmath24 s quantities are given by the solution to the implicit set of equations @xmath57 @xmath58 which can be solved numerically , using a newton - raphson scheme for nonlinear systems of equations .
it should be noted that in the case when @xmath59 , @xmath60 , equation ( [ p_k_ssky ] ) has the trivial solution @xmath29 , as we expected for the standard case without correlations .
the estimates obtained with these @xmath24 give us a better guess for @xmath37 , that could be used in equation ( [ p_k_ssky ] ) to obtain more appropriate values of the @xmath24 .
however , in practice , we have checked that for all the cases that we consider in this paper , this iteration is not necessary , since over the a priori uncertainty region of the parameter , the variation of the @xmath24 is negligible . the previous expression , derived for the problem of total power estimation , can also be applied to any problem of one parameter estimation , as follows .
let @xmath61 be the parameter we are interested in .
if we expand the cf in a taylor series around an initial guess , @xmath62 , we obtain , up to first order , @xmath63 @xmath64 with @xmath65 , so we can use equation ( [ p_k_ssky ] ) , with @xmath66 if we use an initial guess close to the real value , this linear approximation will give good results .
the @xmath67 s can be obtained numerically for each problem .
when we deal with a problem of several parameters estimation , it is not well defined what has to be minimised . a reasonable criteria for these problems ,
if we want to estimate the set @xmath68 , is to minimise @xmath69 ) $ ] , where we define @xmath70)$ ] as the @xmath54 for each individual parameter .
linearising around our initial guess , @xmath43 , we can derive a simple expression for the @xmath54 of each parameter . in this case , we have that @xmath71 where we define @xmath72 the estimate of the parameters is given by the solution to the linear system of equations @xmath73 , where we have again neglected the dependence of @xmath74 on the parameters .
the general expression of the covariance matrix of the parameters is shown , for a general linear problem , in appendix a. if we expand those matrices , we obtain that the general form of this estimate , for our problem , is given by @xmath75 = \frac{1}{r } \sum_{k } j_{k , i } p_k \frac { e [ c(\theta_k ) ] - c(\theta_k,\hat{m}_0 ) } { \sigma^2(c(\theta_k ) ) } , \qquad i=1, ... ,p \label{est_m_k}\ ] ] where @xmath76 = e[m_i ] - ( m_0)_i$ ] , and @xmath77 and @xmath78 are numbers obtained from the @xmath79 s . from ( [ est_m_k ] ) we can infer the general expression for the @xmath54 in the case of several parameters , obtaining @xmath80 ) = \frac{1}{r } \bigg [ \sum_{k , j } j_{k , i } j_{j , i } p_k p_j \frac { c'_{kj } } { \sigma^2(c(\theta_k ) ) \sigma^2(c(\theta_j ) ) } \bigg]^{1/2 } \label{rms_several}\ ] ] where @xmath81 . using the previous equation , we can obtain the @xmath24 quantities for any problem , just minimising the product @xmath82 ) $ ] numerically .
summarising , we will have a different expression for the @xmath24 s for each particular problem .
an application of these equations for the problem of two parameters estimation can be found in section 6.1 .
the @xmath1 proposed in eq .
( [ chi2 ] ) corresponds to a sum of quantities which are not independent .
if we had set @xmath83 , we would have the standard @xmath0 statistic , but still with correlations among the terms .
therefore , its distribution function will not be the standard one .
the formal expression for the distribution of a @xmath0 constructed from variables which are distributed following a multivariate gaussian distribution is given in appendix c. this distribution , when applied to cmb analyses , was studied in b93 .
there , they proposed that the distribution function for the statistic ( [ chi2 ] ) is given by an standard ( rescaled ) @xmath0 function , but with an effective number of degrees of freedom .
this proposal is not exact ( see also appendix c ) , but it turns out to be a very good approximation for the true distribution function , as we shall see in the following section . in a general case , the error in the distribution function using our approximation will be a few percent .
the basis of the approximation is to assume that correlations only reduce the degrees of freedom , but do not change the shape of the distribution . quantifying this argument
, there exist a certain constant , @xmath84 , which makes the statistic @xmath85 to be distributed as an ordinary @xmath0 , with an effective number of degrees of freedom , @xmath86 .
this number , and the constant @xmath84 , are obtained just by imposing the new distribution to have the mean and the variance of a standard @xmath0 , i.e. , @xmath87 @xmath88 where @xmath89 means mean square .
summarising , our proposal is that once the first and second moments are fixed , the whole distribution will follow a @xmath0 very closely . for our problem ,
we obtain @xmath90 @xmath91 in this calculation , we needed to compute the @xmath89 of eq .
( [ chi2 ] ) .
this result is obtained using the fact that the data points follow a multivariate gaussian distribution , and it can be found in b93 ( see also appendix b for similar calculations ) . in general , @xmath86 is a real number , so we have to consider the analytic extension of a standard @xmath0 distribution ( which is known as the gamma distribution function , see for example @xcite ) , @xmath92 @xmath93 just by replacing the factorial with the gamma function ( @xmath94 ) . in ( [ dfchi2 ] ) , @xmath95 is the probability of finding a value for @xmath96 between @xmath97 and @xmath98 , and @xmath99 stands for the probability density function .
once we know the distribution function , the confidence limits are given by integration bellow this curve .
we assign a weight to each hypothesis as in a standard @xmath0 analysis , integrating the distribution from the obtained value up to infinite , @xmath100 @xmath101 i.e. , the probability of finding a value of @xmath97 bigger than or equal to @xmath102 . here
, we explicitly write where the dependence in the data ( @xmath103 ) and in the parameters ( @xmath22 ) is .
in this section , we will test the whole method in the problem of one parameter estimation , but first , we will study the quality of our approximation to the distribution function of a @xmath0 with correlations .
these two points will be done by means of monte carlo simulations . in order to do that
, we have chosen the jb - iac 33 ghz interferometer @xcite as the reference experiment .
this experiment is a two element interferometer , which operates at @xmath104 ghz , at the teide observatory .
it has two configurations , with angular resolutions @xmath105 ( @xmath106 ) , and @xmath107 ( @xmath108 ) , respectively .
the window function in both configurations is very narrow , so the results are quoted in terms of total power inside the band ( band power ) .
the experiment has given measurements on the power spectrum on both scales @xcite , which are consistent with the boomerang data @xcite .
we have the likelihood analysis implemented for this experiment , so the comparison with the new method will be straightforward . in our analyses ,
we have used the compact configuration , and only one of the two channels ( i.e. , the real part of the complex visibility ) .
the cmb realizations have been done assuming the following values for the cosmological parameters : @xmath109 , @xmath110 , @xmath111 , @xmath112 and @xmath113 . for this model ,
the total power inside the window function ( or band power ) is @xmath114 for the short configuration ( @xmath115 ) .
this number is related with @xmath116 by a conversion factor , which is obtained using the flat band power approximation ( i.e. @xmath117 constant inside the window function ) in equation ( [ def_ssky ] ) . for our instrument , this conversion factor is @xmath118 , which gives @xmath119 for the previous model . for this experiment ,
the sensitivity in a @xmath120s integration is given by @xmath121 , where @xmath122 is the number of observing days .
therefore , the signal - to - noise ratio is given by @xmath123 . with correlations with @xmath124 terms .
we show the histogram with the frequencies for the rescaled @xmath0 , obtained from @xmath125 realizations , for four different cases , varying @xmath126 ( signal - to - noise ratio ) .
we use @xmath127 bins of equal size to sample the distribution function . in all the figures , the dots represent the numbers coming from the realizations , and the error bars show their sampling error .
the solid line is our approximation to the distribution function , using the value of @xmath86 from the formula ( shown within parentheses in the figure ) .
it is also shown , using dot - dashed lines , the distribution function of a ( rescaled ) @xmath0 with @xmath124 .
we can see the effect of the correlations as @xmath126 increases . ]
the first point is to check the validity of our approximation to the distribution function for a @xmath0 with correlations .
we will study the case when the @xmath128 quantities entering in the @xmath0 follow a multivariate gaussian distribution .
this is the case for an ( ideal ) cmb map , where the temperature at each pixel has two contributions , one coming from gaussian noise , and another one from the cosmological fluctuation field , which is supposed to be also gaussian . using the cmb terminology from section 2
, we will study the distribution of the statistic @xmath129 and we will compare it with the proposed approximation .
this is a particular case of ( [ chi2 ] ) , when @xmath83 .
we will study in detail this case here , but our results are completely general .
it should be noted that our proposal is exact , by definition , in the two limit cases of no correlations at all , and totally correlated points .
the first one correspond to the definition of the @xmath0 distribution function , and the second one is the case of a @xmath0 with @xmath130 . in appendix c
we present the formal aspect of the distribution function for ( [ chi3 ] ) , and we study in detail , analytically , the case @xmath131 .
the cases with low @xmath4 turn out to be the critical ones , because the shape of the distribution function differs strongly from a gaussian . in the limit of high @xmath4 , both our approximation and the real distribution function tend to a gaussian distribution ( the same one , by definition ) , as a consequence of the central limit theorem .
therefore , it is interesting to test our proposal for a intermediate range of values of @xmath4 .
we have done so , and we will present here , as an example , the case for @xmath124 .
we generate cmb realizations with noise , for a ten pixels map . from each simulation , we compute ( [ chi3 ] ) , and from the whole set of values obtained , we study the histogram with the frequencies , and we compare it with the proposed one , for several values of the signal - to - noise ratio .
we show these results in figure [ fig_histog ] . in general ,
the asymptotic shape of the distribution is very well reproduced , and the largest differences always occur for low values of @xmath0 .
this is precisely the kind of approximation we need , because in a statistical analysis we usually are interested in the tail of the distributions .
we can see that the distribution of the @xmath0 with correlations among terms is compatible , inside the numerical precision , with an standard ( rescaled ) @xmath0 , with an effective number of degrees of freedom , smaller than @xmath4 .
we have also checked that the numerical values for @xmath86 and @xmath84 are correctly given by equations ( [ neff ] ) and ( [ a ] ) . in table
[ tabla2 ] we compare the values obtained from the simulations with the predicted ones given by the theoretical formulae .
their difference is in all cases smaller than the sampling errors , so we conclude that our expressions give the correct values for these parameters .
.values of @xmath86 and @xmath84 , for @xmath124 .
[ cols="^,^,^,^,^,^ " , ] @xmath132 adopted values for the @xmath24 s . the last row is the optimum set of @xmath24 values ( see figure [ figura2 ] ) .
+ @xmath133 results from 100 monte carlo simulations of cobe data ( see details in the text ) .
the first number is the average value for the parameter , and the second one is the @xmath54 from the simulations .
+ @xmath134 @xmath54 values obtained analytically , using the linear approximation to the cf ( see section 3 ) .
we will use the @xmath24 quantities given by the minimum of the function @xmath135 ) \times rms(e[q_{rms - ps}])$ ] , as we have discussed in section 3 .
those @xmath54 s can be derived , using the linear approximation to the cf , from equation ( [ rms_several ] ) . in this problem ,
the @xmath77 and @xmath78 quantities are given by @xmath136 @xmath137 where @xmath138 stands for @xmath11 , and @xmath139 for @xmath140 .
the equation for @xmath141 can be obtained from @xmath142 , just interchanging @xmath143 . in order to check the previous expressions for the @xmath54
, we use 100 of the above mentioned realizations of cobe like maps ( @xmath144 ; @xmath145 ) , and we analyse them using several sets of @xmath24 s . in this way , we can obtain the real value of the @xmath54 , and compare it with the number coming from the formula .
the results are summarised in table [ tabla3 ] . in all cases ,
the theoretical numbers obtained from equation ( [ rms_several ] ) are in agreement with the numerical results , so we conclude that the linear approximation to the true cf works well in computing the @xmath54 .
the largest differences occur when we obtain a large @xmath54 , due to the fact that , in that case , fails the linear approximation to the cf . the average values recovered for @xmath11 and @xmath146 from monte - carlo simulations show that the estimator is unbiased , as we would expect .
it should be noticed that , in table [ tabla3 ] , the effective number of degrees of freedom is quite small . in all cases ,
we obtain @xmath147 , but we are using @xmath148 .
the reason is that the cf contains long - range terms , coming from low multipoles ( @xmath149 ) .
this fact reduces the degrees of freedom drastically , so the choice of the @xmath24 will be critical in this problem .
the numbers obtained when we consider the whole cf and @xmath29 are compatible with those in bhs94 , but slightly better because we consider the noise of the 4-year cobe map . in that paper
, they obtained , using the same galactic cut ( @xmath150 ) , and the noise from the 2-year map , the values @xmath151 ) = 0.96 $ ] and @xmath152 ) = 253\mu k^2 $ ] ( in our units ) .
nevertheless , we see that considering only the first points , and setting to zero the others , strongly reduces the @xmath54 of the estimate , even below the values obtained when they do not consider noise and incomplete sky coverage ( they have @xmath151 ) = 0.36 $ ] and @xmath152 ) = 175 \mu k^2 $ ] ) .
s for the @xmath153 4-year _ cobe _ map ( see details in the text ) , using the galactic cut @xmath150 .
these values were obtained by numerical minimisation of the product @xmath151 ) \times rms ( e[q_{rms - ps } ] ) $ ] , using the linear approximation to the cf .
each value for @xmath154 ( @xmath155 ) corresponds to an angular distance , given by @xmath156 .
an explanation of this peculiar structure can be found in the text ( section 6.1 ) . ] finally , we can obtain the optimum set of @xmath24 s by numerical minimisation of the product of @xmath54 s .
we have used the fortran program @xmath157 @xcite .
the obtained value for these quantities is shown in figure [ figura2 ] .
these numbers do not depend on the initial guess for @xmath11 and @xmath140 within the a priori region of uncertainty .
the values obtained have a peculiar form , but it can be understood as follows .
we see that the terms which contribute to the optimum estimator ( minimum @xmath54 ) correspond to the first @xmath158 degrees , i.e. the first part of the cf , as we would expect .
but there are also two regions , one at @xmath159 , and another at @xmath160 , which contribute to the estimator .
these peaks are located just in the zeros of the quadrupole ( @xmath161 ) which is the multipole with the largest cosmic variance . by using those points ,
we add some information to the estimator ( coming from higher @xmath162 values ) , but we do not increase its variance . in any case , to consider or not those points do not affect too much to the power of the method ( in table [ tabla3 ] , the values obtained when using the first 10 points from the cf are close to those obtained with the optimal set ) . following this interpretation , one could think about other combinations of the @xmath24 parameters using others points ( for example , taking two points at @xmath163 and @xmath164 .
we have explored this possibility as an illustration , and we find the values @xmath165 and @xmath166 , which are similar but slightly higher than those obtained for our optimal @xmath24 .
thus , we can conclude that for this problem , we can find a set of estimators ( one for each one of these sets of @xmath24 values ) which give similar rms values when estimating @xmath11 and @xmath146 , and as we show below , these values are comparable to those given by the likelihood method .
finally , we have applied our @xmath1-test for the cf to the actual 4-year cobe data .
our estimates , from the analysis of the @xmath153 map , using the galactic cut @xmath150 , a step of @xmath167 to sample the cf , and the optimum set of @xmath24 s , are @xmath168 = 1.08 $ ] , and @xmath169 = 15.2 \mu k$ ] , so our estimate of the parameters , using the true @xmath54 from @xmath170 realizations , will be @xmath171 , and @xmath172 .
this result has to be compared with the likelihood analysis using these data ( see @xcite , table 1 ) .
they obtain for this map @xmath173 , @xmath174 .
the estimates from both methods are nearly the same , and now the error bars coming from the cf are compatible in size with those coming from the maximum likelihood method . for comparison , using no weights ( @xmath29 ) , we would obtain @xmath168 = 1.10 \pm 0.54 $ ] , and @xmath169 = ( 16.0 \pm 4.8 ) \mu k$ ] , so we can see that using the @xmath24 s for this problem is essential . finally , we will apply our method to obtain band power spectra for the cobe data .
we have used the realizations from the previous subsection ( @xmath144 ; @xmath145 ) , and the same cobe map .
we will compare our results with those from @xcite ( see table 2 in that paper ) .
we have used exactly the same @xmath162 range : four multipole bands between @xmath161 and @xmath175 .
those bands are : @xmath176 ; @xmath177 ; @xmath178 and @xmath179 . using those realizations , we have checked that the method is unbiased when applied to power spectrum estimation . when applied to the @xmath153 4-year cobe map
, we obtain the results that are shown in table [ tabla4 ] .
we quote the band power values in terms of the quadrupole normalisation expected for a scale - invariant power - law spectrum within the specified range of @xmath162 .
the quoted values for our method have been obtained from the optimal set of @xmath24 for this problem . when compared with the ml data
, we can see that the error bars are of the same order in both cases , as in the previous subsection .
so we again obtain a method of a similar power to the likelihood on the pixel map .
@ccccc@ & + + method & @xmath176 & @xmath177 & @xmath178 & @xmath179 + @xmath1 , optimal & @xmath180 & @xmath181 & @xmath182 & @xmath183 + ml @xmath133 & @xmath184 & @xmath185 & @xmath186 & @xmath187 + @xmath132 these band power amplitudes are expressed in terms of @xmath188 , i.e. the quadrupole normalisation expected for a scale - invariant power - law spectrum within the specified range of @xmath162 .
the units are @xmath189 .
+ @xmath133 ml values from @xcite for the same map .
the estimates from both methods are consistent in all bins , except the apparent inconsistency at the second one , where the two estimates differ in more than 3 times the size of one error bar .
nevertheless , the statistical significance of that deviation has to be computed as follows : given that we do not know the true value for the band power , when comparing two results we have to consider the difference of both estimates , and so we have to compute the variance of the difference .
in this case , the difference is @xmath190 , and the @xmath54 of the difference of those two estimates is @xmath191 .
here we are assuming the fact that both methods are almost independent , following the results for the case of one parameter estimation , with signal - to - noise ratios of the order of @xmath138 .
therefore , we find a deviation at the @xmath192-sigma level , which corresponds to a fluctuation of 1 in 47 ( for a normal distribution ) .
this can be understood given that the ml and the @xmath1 are two different methods : the first is based on the full map , and the second on the correlation function .
therefore , the estimates will be different in general , although as we can see , both methods have similar power , so there is no reason to consider any one of them as `` the estimate '' .
summarising , we have seen that it is possible with our method to perform an analysis with a similar power to the ml , even for the case of several parameters .
in this paper we have presented an statistical method to analyse cmb maps .
it consists in a variation of an standard @xmath0-test for the case when we have correlated points . here
, our test has been constructed from the two - point correlation function , following the proposal from other authors @xcite that the cf contains all the relevant information concerning the cosmological parameter estimation . in this line
, we propose a @xmath1 based on the cf , which is a different approach from the `` usual @xmath0 '' method ( which uses the full covariance matrix @xmath37 ) .
our proposal explicitly uses only the diagonal terms of the covariance matrix of the data , but we introduce certain weights , which now implicitly contain all the correlations .
this approach has two important computational advantages compared with the `` usual @xmath0 '' , or with a likelihood based on the cf : * we do not need to invert the covariance matrix , @xmath37 ; all the quantities ( the @xmath1 , @xmath86 , @xmath84 and the expression for the @xmath54 ) depend on @xmath37 directly . *
even more , if we have a problem with a low value for @xmath86 , the effective number of @xmath24 to use ( i.e. the number of @xmath24 s which are significantly different from zero ) will be also small , typically , of order @xmath193 .
so it is not even necessary to use all the terms of the diagonal .
these advantages are more important when comparing our method with the standard ml analysis on the full map .
we do not need to invert the covariance matrix of the map ( @xmath194 ) , and we only need to concentrate on a few numbers , so the problem is computationally accessible if we have to deal with large datasets .
it it important to stress here that the @xmath1 is not an approximation to the ml on the full map , but a different approach ( i.e. both methods will give different estimates and probability contours for a given problem ) .
so the @xmath1 can be applied to any problem , but it has to be checked , by means of monte carlo simulations , that the method has a similar power to the maximum likelihood . as we have seen , this is the case for several cmb common problems ( power spectrum estimation , and cosmological parameter estimation ) .
the largest computational effort in our method has to be done estimating the cf , which is a @xmath195 operation .
nevertheless , there are estimators for the cf more efficient @xcite , so our procedure could be applied to current _ wmap _ data , and _ planck _
simulated data , but this will be treated in detail in future works .
our method can be extended for general noise covariance matrices .
we only need to compute @xmath196 in the same way as the cf , and introduce it in equation ( [ chi2 ] ) .
if we have the noise matrix , it is straightforward to obtain @xmath196 .
but , if we do not have the noise matrix , we can obtain an estimate of the correlation function of the noise using mc simulations .
this idea has been used recently by other authors @xcite .
it must be noted that if the noise changes substantially from pixel to pixel , then we would have to use weights in equation ( [ est_cf ] ) to compute the cf in a more efficient way .
we have also presented an approximation which gives very accurately the distribution function for a @xmath0 constructed from a set of multivariate gaussian variables .
this proposal can be extended to approximate the distribution function of any quantity made of a sum of squares , each of them distributed ( exactly or approximately ) following a gaussian distribution . to conclude , we propose that , if we are interested in obtaining a certain set of parameters , @xmath22 , we can use an unbiased estimator of certain quantities depending on these parameters , provided that they contain all the relevant information to these parameters ( in our case , we have used the cf at certain angles ) . from it , we may obtain , by varying the @xmath24 s , the best estimator of those parameters . in this paper , we have tested this proposal , using the two - point cf as the reference estimator , in cmb problems . for the case of one parameter estimation ( the normalisation of the spectrum ) ,
our method , with the cf , turns out to be as powerful as the ml .
when applied to cobe data , we have shown the importance of choosing the right set of @xmath24 s . in the optimum case
, we obtain a value for the @xmath54 two or three times smaller than the one obtained without weighting at all . in this problem ,
the @xmath24 s are critical because the effective number the degrees of freedom is very small .
the reason is that when we have strong correlations ( @xmath86 small compared with @xmath11 ) , the structure of these correlations , which is encoded in the @xmath24 s , will be relevant , so there will be a considerable difference between the optimal weights and @xmath197 . when analysing cmb data which contain large scales ( low multipoles ) , we are considering correlations over long distances .
all the points are correlated with the others due to these low multipoles ( quadrupole , octupole , ... ) . in the case of boomerang data @xcite
, the effective number of degrees of freedom will be larger ( if we throw away the large scales ) , so the @xmath24 s will be closer to 1 , and a standard @xmath0 test based on the cf should produce good results .
we would like to thank m.p.hobson , r.a.watson , r.rebolo , c.gutirrez and r.kneissl for their useful comments , and b.barreiro for her cobe - like maps simulation program .
in this appendix we will study the relationship between the usual @xmath0 analysis ( the standard approach for the case of correlated data points , which uses the full covariance matrix ) , and our approach , for gaussian problems in which the model depends linearly on the parameters , and with a covariance matrix independent of the parameters .
this case is not exactly equal to the usual one in cmb analyses , but it is very close and is particularly suitable for illustration .
we will use here the following notation .
let @xmath198 = @xmath199 be a @xmath200 matrix containing the @xmath11 data points , which , by hypothesis , are distributed following a multivariate gaussian distribution .
let @xmath201 = @xmath202 be a @xmath203 matrix whose elements are the @xmath154 parameters of the model .
let @xmath204 be a @xmath205 matrix , also given by the model .
their elements are defined so that the mean value of @xmath198 , @xmath206 , is given by the matrix multiplication @xmath207 .
finally , let @xmath208 be the covariance matrix of the problem , defined as @xmath209 where @xmath210 stands for the transpose .
using the previous definitions , the usual @xmath0 and our @xmath1 are given , respectively , by @xmath211 @xmath212 where we have defined the matrix @xmath213 using the diagonal of the covariance matrix , and our weights ( @xmath214 ) , in the following way : @xmath213@xmath215 , for @xmath216 , and @xmath213@xmath217 for @xmath60 .
it should be noted that the @xmath213 matrix depends implicitly on the weights ( @xmath214 ) . for this family of models under consideration ,
the optimum estimator is the maximum likelihood , which is given by @xmath218 until this point , we have not made use of the fact that the covariance matrix is independent of the parameters .
if we use it now , from the last equation we have that the maximum likelihood reduces to the usual @xmath0 for this problem . the estimate for both methods is obtained by minimising the previous expressions with respect to the parameters , so we have @xmath219 = ( { \bf { \hat{y } } } { \bf { \hat{m}}}^{-1 } { \bf { \hat{x}}}^t ) ( { \bf { \hat{x}}}{\bf { \hat{m}}}^{-1 } { \bf { \hat{x}}}^t ) ^{-1 } \label{estimate1}\ ] ] @xmath220 = ( { \bf { \hat{y } } } { \bf { \hat{v } } \it{}}^{-1 } { \bf { \hat{x}}}^t ) ( { \bf { \hat{x}}}{\bf { \hat{v } } \it{}}^{-1 } { \bf { \hat{x}}}^t ) ^{-1 } \label{estimate2}\ ] ] hereafter in this section , we will use subscript @xmath138 for the standard ( usual @xmath0 ) method , and @xmath139 for the @xmath1 .
we compute now the covariance matrices of the parameters , @xmath221 , which are given by @xmath222 - < e_1 [ { \bf { \hat{\alpha } } } ] > ) = ( { \bf { \hat{x}}}{\bf { \hat{m}}}^{-1 } { \bf { \hat{x}}}^t ) ^{-1}\ ] ] @xmath223 - < e_2 [ { \bf { \hat{\alpha } } } ] > ) = \ ] ] @xmath224 using this notation , the @xmath89 for each parameter is given by the corresponding element in the diagonal of @xmath221 . the following point is to compare the estimates of both methods . for the case of one parameter estimation , it can be argued that both method give the same estimate , and therefore have the same @xmath54 .
the argument is as follows : in this case , the estimate for both methods is a linear combination of the @xmath11 quantities @xmath198 .
so it could be possible , in principle , to fix the @xmath11 quantities ( @xmath214 ) to equalise the @xmath11 coefficients in expressions ( [ estimate1 ] ) and ( [ estimate2 ] ) . given that for this particular problem the usual @xmath0 is the optimal method ( i.e. the one with the minimum variance ) , and that the corresponding estimator is the only linear one for which this variance is obtained , those @xmath214 quantities which set equal the coefficients in ( [ estimate1 ] ) and ( [ estimate2 ] ) , are exactly the same that would be obtained by minimisation of the @xmath54 of that parameter .
we have checked this statement for the critical case where we have a @xmath0 with only @xmath225 terms . in figure [ comparachi2s ]
we present several particular examples , showing that for the set of @xmath214 quantities that minimise the @xmath54 for the @xmath1 , we always obtain the same @xmath54 as in the case of the usual @xmath0 .
we have also checked that , for those values of the @xmath214 , the estimates from both methods are exactly the same .
method with correlations , and our @xmath1 method , for several linear gaussian models with only one parameter , @xmath226 , and @xmath225 variables , named ( @xmath227,@xmath228 ) , normally distributed with means @xmath229 and @xmath230 .
we also assume a covariance matrix independent on the parameter , and we parameterise it as @xmath231 @xmath232 , @xmath233 @xmath234 , and @xmath235 @xmath236 @xmath237 , but these results are completely general .
we plot here four typical cases for this problem .
the dot - dashed line in all four panels corresponds to the usual @xmath0 value for the rms in the estimate of @xmath226 .
the solid line is the rms obtained for the @xmath1 , using the shown value of @xmath238 .
the @xmath239 is obtained from the normalisation equation @xmath240 .
we see that it is always possible to find a certain value of @xmath238 , for which we have exactly the same rms for both methods . for that @xmath238 value ,
the estimates are also obtained to be equal ( see text for details ) . ]
let s consider now the case of several ( @xmath241 ) parameters .
for this problem , it is not possible in general to obtain a set of @xmath214 quantities which render the @xmath54 of the usual @xmath0 . in any case
, we have checked that it is always possible to choose those quantities to make the estimate for just one of the parameters exactly equal , and so its @xmath54 .
we have studied this problem in detail for the case of @xmath242 terms in the @xmath0 , and @xmath243 parameters .
the results are the following : * if we choose to minimise the @xmath54 of only one parameter to find the optimal @xmath214 quantities , we can always make the estimate for that parameter exactly equal to the usual @xmath0 . for those @xmath214 values , the estimator for the other parameter is very close to the optimum , in the sense that the @xmath54 for the other parameter at the most @xmath244 bigger that the optimal one . *
if we minimise the product of the two @xmath54 s , we find in all cases that both estimates are close to the optimal ones , and the largest relative differences between @xmath245 and @xmath246 are smaller than @xmath247 .
therefore , we can conclude that the criteria to obtain the @xmath214 quantities has an small ambiguity , in the sense that if we are interested in one parameter in particular , we should minimise the @xmath54 for that parameter only . in practice , this ambiguity is not relevant because , for this problem , the differences between the estimates and the obtained @xmath54 values are negligible .
therefore , we will maintain the original proposal of minimising the product of @xmath54 s , which in some sense is equivalent to minimise the average size of the confidence region .
all these arguments can be applied to cases with @xmath248 parameters .
the probability density function for a multivariate gaussian distribution of n variables ( @xmath249mgd ) , @xmath250 , is given by @xmath251 where @xmath252 is called the _ covariance matrix _
, @xmath253 stands for its determinant , and @xmath254 is the mean of @xmath103 .
the elements of @xmath252 are given by @xmath255 we define @xmath256 .
when computing the mean square of equation ( [ chi2 ] ) , or the @xmath37 matrix given in ( [ est_cf ] ) , we need to know the following quantities : @xmath257 , @xmath258 and @xmath259 .
we will obtain them here . for a @xmath260mgd
, we can obtain the quantity @xmath261 @xmath262 } dx_1 dx_2 = \ ] ] @xmath263 for a @xmath264mgd , we obtain @xmath265 @xmath266 where @xmath267 stands for the elements of the inverse of the c matrix . using the same notation , for a @xmath268mgd , we obtain @xmath269 @xmath270 @xmath271 @xmath272
the moment generating function of the quadratic form @xmath273 , where the @xmath250 variables follow a @xmath249mgd with zero mean ( eq .
[ mgd ] ) , is given by ( see @xcite ) in this equation , @xmath275 are the eigenvalues of the matrix @xmath276 , where @xmath252 is the covariance matrix ( [ cov_matrix ] ) , and @xmath277 is defined as @xmath278 , being @xmath279 the kronecker - delta . from this expression ,
the distribution function can be obtained by the laplace inverse transform .
formally , we have where @xmath281 stands for the inverse laplace transform @xcite , and we use the notation @xmath282 for the exact distribution function of the @xmath0 . as an example , we will study in detail the case @xmath225 , comparing the exact distribution function with our approximation .
we compare both the distribution function and the @xmath154-order moments up to @xmath289 with the values obtained using our approximation .
the results quoted here correspond to the case @xmath290 , and @xmath291 , so @xmath292 varies in the range @xmath293 $ ] .
nevertheless , the results are completely general . for our problem
, we can write @xmath294 , @xmath295 , and @xmath296 . in figure [ app1 ] we present the @xmath297 function for the value of @xmath292 which gives us the maximum percentage difference between the exact and the approximated functions .
this value corresponds to @xmath298 .
the largest percentage difference in the distribution function for this case is reached at @xmath299 , and has a value of @xmath300 . in terms of the weights ,
we obtain for this point a difference of a @xmath301 ( @xmath302 , and @xmath303 ) .
nevertheless , the power of our approximation is that the largest differences always occur at low values of @xmath0 . the asymptotic shape of the exact distribution function is well reproduced , as we need for a @xmath0 analysis .
to conclude , we show in figure [ app2 ] the third and fourth order moments , both for the real distribution and the approximation , in the whole range of values for @xmath292 . by definition ,
the first and second moments are equal for the true and the approximate distribution .
we see again that the approximation follows quite closely the true function , as we have found from the simulations in section 5.1 . with correlations , with @xmath304 terms .
we have used @xmath290 , and @xmath305 .
we show the case @xmath298 , because for that value we have the largest percentage difference between the true function and our approximation , which is given by @xmath306 and @xmath307 .
we can see that the largest differences occur at low values of @xmath0 .
the asymptotic values are well reproduced . ] | we present a new general procedure for determining a given set of quantities . to this end , we define certain statistic , that we call modified @xmath0 ( @xmath1 ) , because of its similarity with the standard @xmath0 .
the terms of this @xmath1 are made up of the fluctuations of an unbiased estimator of some statistical quantities , and certain weights .
only the diagonal terms of the covariance matrix explicitly appear in our statistic , while the full covariance matrix ( and not its inverse ) is implicitly included in the calculation of the weights . choosing these weights we may obtain , through minimising the @xmath1 , the estimator that provides the minimum rms , either for those quantities or for the parameters on which these quantities depend . in this paper , we describe our method in the context of cosmic microwave background experiments , in order to obtain either the statistical properties of the maps , or the cosmological parameters .
the test here is constructed out of some estimator of the two - point correlation function at different angles . for the problem of one parameter estimation ,
we show that our method has the same power as the maximum likelihood method .
we have also applied this method to monte carlo simulations of the cobe - dmr data , as well as to the actual 4-year data , obtaining consistent results with previous analyses .
we also provide a very good analytical approximation to the distribution function of our statistic , which could also be useful in other contexts .
[ firstpage ] methods : statistical cosmology : cosmic microwave background cosmology : cosmological parameters |
the electronic potentials and dipole moments are calculated with the configuration - interaction ( ci ) valence bond method .
both nonrelativistic and relativistic implementations of the method are used . the basic idea behind the valence - bond method
is that the electronic molecular wavefunction @xmath11 is constructed from wavefunctions , which are products of wavefunctions that describe the constituent atoms @xmath12 and @xmath13 .
in essence , the molecular wavefunction is given by @xmath14 where each @xmath15 is an antisymmetrized ( @xmath16 ) product of two atomic slater determinants @xmath17 the atomic slater determinants @xmath18 and @xmath19 are constructed from one - electron functions centered on the nucleus of atom @xmath12 and @xmath13 , respectively .
the variational ci coefficients @xmath20 in eq .
( [ psi ] ) are obtained by solving a generalized eigenvalue matrix problem , since in our approach the @xmath21 are not orthogonal .
the permanent molecular dipole moment of the molecular wavefunction @xmath11 is calculated from @xmath22 where @xmath23 is the total number of electrons in the molecule and @xmath24 is the electron charge .
the first term of eq .
[ dipole ] depends on the position @xmath25 and charge @xmath26 of nuclei @xmath27 and @xmath13 and is zero by choosing an appropriate coordinate origin . the second term of eq .
[ dipole ] depends on the electronic molecular wavefunction and the electron positions @xmath28 . in the valence - bond method @xmath29 where @xmath30 denotes both the single - electron coordinate @xmath31 and the variable @xmath32 . for a nonrelativistic or relativistic calculation
@xmath32 has two or four values , respectively .
the single - electron density matrix is @xmath33 where the transition density matrix between determinants @xmath34 and @xmath35 is defined by @xmath36 @xmath37 is the determinant of the @xmath38 overlap matrix @xmath39 = @xmath40 , and the @xmath41 and @xmath42 denote the single - electron functions used in constructing the basis functions @xmath34 and @xmath35 , respectively .
moreover , @xmath43 and @xmath44 are the determinants of the matrices @xmath45 and @xmath46 .
the one - electron wavefunctions @xmath47 and @xmath48 are obtained by self - consistently solving hartree - fock and sturmian equations for a nonrelativistic calculation and dirac - fock and sturmian equations for a relativistic calculation @xcite .
transition dipole moments can be derived in a similar fashion by assuming different initial and final states in eq .
[ dipole ] .
the atomic determinants for the krb dimer are constructed from single - electron hartree - fock or dirac - fock functions / orbitals for electrons in closed and valence shells .
sturmian wavefunctions complement the basis functions and are used to describe virtual orbitals .
one - electron functions are characterized by a main quantum number @xmath49 and orbital angular momentum @xmath50 for both nonrelativistic and relativistic calculations .
in addition for relativistic calculations the one - electron orbital is labeled by the total electron spin @xmath51 .
the closed shells @xmath52 of k and @xmath53 of rb form the core of the heteronuclear molecule .
the superscript denotes the number of electrons in shell @xmath54 . for relativistic calculations
@xmath55 is short for @xmath56 etcetera . in our calculations
excitations from these closed shells to valence and virtual orbitals are not included , i.e. all atomic determinants in the molecular basis contain the same number of electrons in these orbitals .
single electron excitations from the closed @xmath57 shell of k and @xmath58 shell of rb are allowed and introduce core - valence correlations in the ci .
the @xmath59 , @xmath60 , and @xmath61 valence orbitals of k and @xmath62 , @xmath63 , and @xmath64 orbitals of rb are allowed to contain at most two electrons .
in addition , for the nonrelativistic calculation we allow at most two electrons in the @xmath64 , @xmath63 , and @xmath65 virtual orbitals of potassium and the @xmath66 , @xmath65 , and @xmath67 virtual orbitals of rubidium .
for the relativistic calculation computational limitations restricted us to the @xmath64 and @xmath63 virtual orbitals of k and @xmath66 and @xmath65 virtual orbitals of rb . both covalent and ionic configurations
are constructed .
the ci expansions with these basis sets have 493 nonrelativistic and 1459 relativistic configurations .
the ground state potentials have equilibrium distances @xmath68 and dissociation energies @xmath69 that agree with experimental rkr values @xcite and other theoretical potentials .
figure [ potentials ] shows the ground state @xmath6 and @xmath2 potentials of the krb dimer as a function of the internuclear separation @xmath5 .
the solid curves of fig . [ potentials ] describe results of our relativistic ( @xmath10 = 0 ) _
ab initio _ calculation whereas the dashed curves are the theoretical data of ref .
the dotted line of fig . [ potentials ] shows the singlet rkr potential of ref .
@xcite . for the @xmath6 state the difference between the two theoretical calculations shown in fig .
[ potentials ] is about 3% at @xmath68 .
a 20% difference exists at @xmath68 of the @xmath2 potential .
our estimate of the @xmath70 van - der - waals coefficient for the @xmath6 potential , obtained by fitting to the dispersion potential @xmath71 , agrees to 2 - 3% with the value of derevianko et .
the experimentally determined dissociation energy for the @xmath6 potential @xcite is 26 @xmath72 smaller than ours .
we performed calculations of the @xmath6 and @xmath2 state potentials and their permanent dipole moments in both nonrelativistic and relativistic approximation .
for the same set of one - electron orbitals the influence of relativistic effects on the potential energy is very small whereas its influence on the dipole moments is large . at @xmath68
a relativistic calculation leads to a 30% increase in the absolute value of the dipole moment for the @xmath0 state and a 50% decrease for the @xmath2 state .
moreover , we find large correlation effects in the electric dipole moment in both nonrelativistic and relativistic calculations . the implementation of core polarization , for instance , leads a 50% of reduction in the absolute value of the dipole moment at the bottom of the singlet potential .
figures [ 1s_dipole ] and [ 3s_dipole ] show the electric dipole moments of the @xmath0 and @xmath2 potentials of krb as a function of internuclear separation , respectively .
a negative dipole moment implies an excess electron charge on the potassium atom .
the presented dipole moments are calculated with the relativistic and nonrelativistic hamiltonian .
curve 1 in figs .
[ 1s_dipole ] and [ 3s_dipole ] is determined from the most accurate nonrelativistic calculation , which includes the two - electron occupation of all orbitals upto @xmath65 for the k atom and @xmath67 or the rb atom .
curve 2 in figs .
[ 1s_dipole ] and [ 3s_dipole ] shows the @xmath73 relativistic calculation . the number of excited orbitals is smaller than in the nonrelativistic basis set and limited to excitations upto @xmath64 and @xmath63 for k atom and @xmath66 and @xmath65 for rb atom .
the difference between the dipole moments of the @xmath73 and 1 component of the @xmath2 potential is much smaller than the uncertainties of our calculations .
tables [ tab1 ] and [ tab2 ] tabulate the permanent dipole moment of the @xmath0 and @xmath2 potentials shown in the figures .
our calculation shows that the distribution of the charge density between k and rb is very diffuse .
it means that the dipole moment depends not only on the electron charge transfer from rb to k atom but also on the induced polarization from the charge transfer , i.e. the electrons occupy excited @xmath74 and @xmath75 orbitals .
charge transfer and the induced polarization are of opposite sign .
this is especially true for the @xmath76 state .
we find that the excess charge is almost equal for the singlet and triplet states , but the latter has a smaller dipole moment .
the permanent dipole moments of the @xmath0 and @xmath2 state multiplied by @xmath77 , based on a the relativistic calculation are shown in fig .
it shows that the long - range behavior of the dipole moment of the two states are equal and proportional to @xmath78 .
this long - range behavior is due to the modification of the molecular wave function by the dipole - dipole and dipole - quadrupole multipole interactions @xcite .
we believe that our convergence with respect to correlation or the number of basis function is of the same order as the difference between nonrelativistic and relativistic calculation .
consequently , we feel that the most accurate dipole moment is obtained from averaging the nonrelativistic and relativistic calculation and assuming an one - standard deviation uncertainty equal to half the difference between the two calculations . for the @xmath0 state
the dipole moment is @xmath790.30(2 ) @xmath1 at @xmath80 @xmath81 , while for the @xmath2 state the dipole moment is @xmath790.02(1 ) @xmath1 at @xmath68=11.2 @xmath81 .
in this section we analyze the possibility of the creation of dipolar krb molecules in an optical lattice .
we assume that colliding atoms are initially in the doubly spin - polarized state , which only allow them to come together on the state @xmath7 potential .
thus photoassociation via an excited molecular @xmath82 or @xmath83 state produces @xmath7 molecular levels . in principle
this makes the production of molecules in the @xmath6 state problematical , since one has to get from the triplet to singlet spin manifold .
there is , however , a viable route from the doubly spin - polarized colliding ground state atoms to ground @xmath84 levels via the excited state . when the detuning of the photoassociation laser from atomic resonance is small compared to the excited state spin - orbit splitting , as will be the case for most practical photoassociation schemes , then an excited @xmath10 = 0 or 1 hund s case ( c ) state can have both singlet and triplet character .
thus , an excited molecular state can be formed from excitation from a @xmath7 state that can re - emit light in a transition to a @xmath6 state .
this process is absent in homonuclear dimers , since the additional gerade- ungerade selection rule for electronic transitions prevents it .
figure [ excpot ] shows the @xmath85 and @xmath86 relativistic excited state potentials as a function of internuclear separation .
the excited state potentials are obtained with the basis described in sec .
[ gs ] . at short internuclear separation
the potentials are described by the @xmath87 hund s case ( a ) labeling following ref .
@xcite where @xmath88 is a projection of the electronic orbital angular momentum on the molecular axis and @xmath89 is the total electron spin . at longer @xmath5 the curves
are better described by @xmath90 hund s case ( c ) labeling and relativistic effects are important .
the potentials dissociate to the excited k(4s)+rb(5p @xmath91 ) and k(4p @xmath91 ) + rb(5s ) fine structure limits .
the two lowest dissociation limits correspond to the two fine structure states of the excited rb atom plus a ground state k atom .
the long - range behavior of the relativistic potentials dissociating to these two limits is attractive .
there are seven attractive potentials with @xmath73 and @xmath86 .
our potentials agree with the calculations of refs .
@xcite .
the attractive excited - state potentials are the most likely candidates for use in two - color raman photoassociation experiments .
we calculate transition dipole moments relevant for transitions from the @xmath92 state , through the attractive @xmath85 and 1 potentials , to the @xmath93 state .
figures [ 3s_trans ] and [ 1s_trans ] show the nonzero transition dipole moments from the @xmath2 and @xmath0 states , respectively .
curves with the same style in the two figures indicate the same intermediate excited state .
photon selection rules ensure that @xmath94 @xmath95 @xmath96 transition are not allowed . at long - range
the transition dipole moments become independent of @xmath5 and their absolute values approach the rb 5s @xmath95 5p(@xmath97 ) transition dipole moment when the corresponding excited potential dissociates to the k(4s)+rb(5p @xmath91 ) limit . at short - range internuclear separation
the dipole moments strongly depend on @xmath5 .
this behavior reflects the change from a hund s case ( a ) to a relativistic coupling scheme between 20 @xmath81 and 30 @xmath81 . in fig .
[ 3s_trans ] the @xmath98 to @xmath99 and in fig .
[ 1s_trans ] the @xmath93 to @xmath100 transition dipole moments approach zero at short @xmath5 because singlet to triplet transitions are not allowed . in fig .
[ 3s_trans ] both @xmath101 to @xmath100 and @xmath102 to @xmath100 have a small dipole moment at small internuclear separation .
this can be understood by noting that the structure of the ground and excited state potentials shown in figs .
[ potentials ] and [ excpot ] is similar to that of homonuclear alkali - metal dimers , where gerade / ungerade symmetry is valid . in homonuclear dimers electric dipole transitions between gerade and ungerade states
are forbidden . the sudden change in dipole moment in fig .
[ 1s_trans ] near 13 @xmath81 is related to the avoided crossing indicated by the circle in fig .
[ excpot ] between the @xmath103 components of 1@xmath104 and 2@xmath105 potentials .
we estimate that the uncertainty of the transition dipole moments is 0.1 @xmath1 based on a comparison of the calculated dipole moments at @xmath106 with the known 5s to 5p dipole moments of rb@xcite .
we determined the permanent dipole moments of the @xmath4 states of the ground configuration of the krb heteronuclear molecule using a nonrelativistic and relativistic ci valence bond method .
the krb permanent dipole moments are small compared to `` truly '' polar molecules , such as nacl .
it might , however , be large enough for experiments that aim to confine krb in optical traps .
in addition we calculated the potential energy curves and transition dipole moments to excited states correlating to k(4s)+rb(5p ) atomic limits .
we have shown that there exist allowed transitions starting from colliding doubly polarized k and rb atoms via an exited state to the singlet x@xmath107 state .
the calculation of the electric dipole moments is a first step towards obtaining quantitative estimates of photoabsorption and molecular production rates in a gas of k and rb atoms . in the future we plan to evaluate frank - condon factors between vibrational levels of ground and excited state potentials .
in addition we need to investigate the effect of black - body radiation on the dipolar molecule by evaluating frank - condon factors between vibrational levels of the ground state .
we wish to acknowledge helpful discussions with ilia tupitsyn and andrea simoni . .
the electric dipole moment of the @xmath0 state . both nonrelativistic ( @xmath108 ) and relativistic ( @xmath109 )
results are presented .
the internuclear separation @xmath5 is in units of @xmath81 and the dipole moment is in units of @xmath1 . [
cols="<,^,^",options="header " , ] | the relativistic configuration interaction valence bond method has been used to calculate permanent and transition electric dipole moments of the krb heteronuclear molecule as a function of internuclear separation .
the permanent dipole moment of the ground state @xmath0 potential is found to be 0.30(2 ) @xmath1 at the equilibrium internuclear separation with excess negative charge on the potassium atom . for the @xmath2 potential
the dipole moment is an order of magnitude smaller ( 1 @xmath3 cm ) in addition , we calculate transition dipole moments between the two ground - state and excited - state potentials that dissociate to the k(4s)+rb(5p ) limits . using this data we propose a way to produce singlet @xmath0 krb molecules by a two - photon raman process starting from an ultracold mixture of doubly spin - polarized ground state k and rb atoms .
this raman process is only allowed due to relativistic spin - orbit couplings and the absence of gerade / ungerade selection rules in heteronuclear dimers . in heteronuclear diatomic molecules
the electrons can be distributed between the two nuclei unequally . as a result , there can be a small excess of negative charge near the more electronegative atom and thus heteronuclear dimers can have a permanent dipole moment . in this study
we calculate the permanent dipole moments of the @xmath4 electronic states of the krb ground configuration , which dissociate to the k(4s)+rb(5s ) limit , and transition dipole moments between these states and excited states , which dissociate to k(4s)+rb(5p ) limits .
simple analysis of the ionization energies of k and rb atoms , 4.339 ev and 4.176 ev , respectively , shows that the k atom is most likely to have an excess of a negative charge when the two atoms interact .
there are several applications of dipolar molecules in the physics of ultra - cold molecular gases .
the long - range interaction between two dipolar molecules is governed by a electric dipole - dipole interaction and can , for example , significantly modify the many - body dynamics of trapped ultra - cold molecular bose einstein condensates ( bec s ) @xcite .
the atom - atom interactions in current atomic bec s are spherically symmetric in origin . following the proposal for homonuclear molecules of ref .
@xcite , dipolar molecules might also be formed in an optical lattice@xcite an ultra - cold atom of each of the atomic species is held in an individual lattice site . a molecule in the electronic ground state
can then be formed under the influence of laser light by inducing a two - photon transition . for any of the proposed applications ,
knowledge of the permanent or transition dipole moments is essential .
the dipole moments have never been determined for krb states of the ground configuration either theoretically or experimentally .
we calculate the dipole moment of the singlet @xmath0 and triplet @xmath2 states of krb as a function of internuclear separation @xmath5 using an _ ab initio _ configuration interaction valence bond ( vb ) method @xcite .
both relativistic and nonrelativistic calculations are presented , which allows us to determine the influence of relativistic effects on the dipole moment .
the use of configuration interaction ( ci ) in the vb method is essential due to the role of correlation in the formation of the molecular bond and nonzero dipole moment .
the main contribution to the dipole moment of krb is not due to the ground state atomic configuration , even though it predominantly determines the energy , but is due to the population of excited valence and virtual orbitals .
the accuracy of the calculation of the @xmath6 wave function is tested by comparing our _ ab initio _ potential with existing experimental @xcite and theoretical @xcite singlet potentials .
we also present the @xmath7 potential , compare it with the calculations of refs .
@xcite , and indicate the sensitivity of this shallow potential to the computational method . furthermore , we analyze the possibility of the creation of dipolar krb molecules in an optical lattice @xcite . in this paper
we calculate the transition electric dipole moments between the @xmath4 states of the ground configuration and relativistic @xmath8 components of excited @xmath4 and @xmath9 states dissociating to the k(4s)+rb(5p ) limits , where @xmath10 is the projection of the total electronic angular momentum of the two atoms on the internuclear axis .
this data provides us information about the most efficient scheme of forming krb molecules in the singlet @xmath6 ground state from a k(4s)+rb(5s ) collision on the triplet @xmath2 potential .
moreover , we present our _ ab initio _ potentials for the excited states which can be used to identify the complex behavior of the transition dipole moment as function of @xmath5 . large - scale theoretical studies of the excited potentials of krb were previously performed in refs .
@xcite .
it was shown in refs .
@xcite that the dispersion interaction of excited krb states is considerably stronger than in any other heteronuclear alkali dimer and that the krb system is a good candidate for photoassociation experiments .
transition dipole moments to more - highly excited krb states have been calculated in ref .
@xcite . |
consider a population of sequences having a common time ( or location ) index
. signals , when they occur , are present in a small fraction of the sequences and aligned in time . in the detection of copy number variants ( cnv ) in multiple dna sequences , efron and zhang @xcite used local f.d.r . , zhang et al .
@xcite and siegmund , yakir and zhang @xcite applied scans of weighted @xmath0-statistics , jeng , cai and li @xcite applied higher - criticism test statistics .
tartakovsky and veeravalli @xcite , mei @xcite and xie and siegmund @xcite considered the analogous sequential detection of sparse aligned changes of distribution in parallel streams of data , with applications in communications , disease surveillance , engineering and hospital management .
these advances have brought in an added multi - sample dimension to traditional scan statistics works ( see , e.g. , the papers in @xcite ) that consider a single stream of data . in this paper , we tackle the problem of detectability of aligned sparse signals , extending sparse mixture detection ( cf .
@xcite ) to aligned signals , and extending multiscale detection ( cf .
@xcite ) to multiple sequences . hence not surprisingly , we incorporate ideas developed by the sparse mixture and multiscale detection communities to find the critical boundary separating detectable from nondetectable hypotheses . in arias - castro ,
donoho and huo @xcite , there are also links between sparse mixtures and multiscale detection methods in the detection of a sparse component on an unknown low - dimensional curve within a higher - dimensional space .
our work here is less geometrical in nature as the aligned - signal assumption allows us to reduce the problem to one dimension by summarizing across sample first .
we supply optimal adaptive max - type tests : penalized scans of the higher criticism and berk jones test statistics .
we also supply an optimal bayesian test : an average likelihood ratio ( alr ) that tests against alternatives lying on the critical boundary .
the rationale behind the alr is to focus testing at the most sensitive parameter values , where small perturbations can result in sharp differences of detection powers .
we state the main results in section [ sec2 ] .
we describe the detectable region of aligned sparse signals in the multi - sample setting , and show that the penalized scans achieve asymptotic detection power 1 there .
we learn from the detection boundary the surprising result that the requirement to locate the signal in the time domain does not affect the overall difficulty of the detection problem , unless the sequence length to signal length ratio grows exponentially with the number of sequences . in section [ sec3 ]
, we show the optimality of the alr and consider special cases of our model that have been well studied in the literature using max - type tests : the detection of a signal with unknown location and scale in a single sequence , and the detection of a sparse mixture in many sequences of length 1 .
we show that the general form of our alr provides optimal detection in these important special cases .
we also illustrate the detectability and detection of multi - sample signals on a cnv dataset . in section [ extensions ] , the detection problem is extended to heteroscedastic signals .
the extension illustrates the adaptivity of the penalized scans .
even though the detection boundary has to be extended to take into account the heteroscedasticity , the penalized scans as described in section [ sec2 ] are still optimal .
on the other hand , the alr tests have to be re - designed to ensure optimality under heteroscedasticity .
the model set - up here is similar to that in jeng , cai and li @xcite .
there optimality is possible without imposing penalties on the scan of the higher - criticism test because the signal length was assumed to be very short .
let @xmath1 be a population of sequences .
we consider the prototypical set - up @xmath2 under the null hypothesis @xmath3 of no signals , @xmath4 for all @xmath5 and @xmath6 . under the alternative hypothesis @xmath7 of aligned signals
, there exists an unknown @xmath8 of disjoint intervals @xmath9 $ ] such that for the @xmath10th interval , @xmath11 , there is a probability @xmath12 that this interval has an elevated mean @xmath13 \\[-8pt ] \nonumber i_n^{(k ) } & \sim & \operatorname { bernoulli } \bigl ( \pi_n^{(k ) } \bigr),\end{aligned}\ ] ] with @xmath14 and the @xmath15 s and @xmath16 s jointly independent .
let @xmath17 , @xmath18 and so forth .
model ( [ munt ] ) extends sparse mixture detection by adding a time - dimension , and there is a similar extension in potential applications .
for example , in the detection of bioweapons use , as introduced in @xcite , we can assume that there are @xmath19 observational units in a geographical region , each accumulating information over time on bioweapons usage .
the bioweapons are in use over a specific but unknown time period , and only a small fraction of the units are affected . alternatively in covert communications detection , only a small fraction of @xmath19 detectors , each tuned to a distinct signal spectrum , observes unusual activities during the period in which communications are taking place . in the detection of genes that are linked to cancer , readings of dna copy numbers are taken from the chromosomes of @xmath19 cancer patients , and only a small fraction of the patients exhibits copy number changes at the gene locations . in section [ extensions ] , we shall consider an extension of ( [ proto ] ) and ( [ munt ] ) to signals carrying a noise component . in the detection of copy number changes , the common practice was to process samples one at a time ;
see lai et al .
in contrast , efron and zhang @xcite , zhang et al .
@xcite and jeng et al .
@xcite proposed procedures that pool across samples first .
our analysis here shows that the alignment information is important , and we should indeed pool across samples first . in appendix
[ appc ] , we provide a comparison between pooling information across sample versus pooling information within sample first .
consider @xmath20 for some @xmath21 .
ingster @xcite and donoho and jin @xcite showed that in the special case @xmath22 ( hence @xmath23 , @xmath24 , @xmath25 ) , as @xmath26 , the critical detectable value of @xmath27 is @xmath28 , where @xmath29 that is , if @xmath30 with @xmath31 , then no test can detect that @xmath32 in the sense that the sum of type i and type ii error probabilities tends to 1 for any test .
donoho and jin @xcite further showed that tukey s higher criticism as well as the berk
jones statistic achieve the detection boundary @xmath33 ; that is , if @xmath34 , then the sum of types i and ii error probabilities tends to 0 . jager and wellner @xcite showed that tukey s test is a member of a family of goodness - of - fit @xmath35-divergence tests that can each achieve the detection boundary .
when @xmath36 , we need to deal with the complication of multiple comparisons over @xmath37 and @xmath38 , and the question arises of how much harder the detection problem becomes .
the number of disjoint intervals in @xmath39 $ ] with common length @xmath38 is approximately @xmath40 .
this ratio has to be factored into the computation of the detection boundary .
the main message of section [ sec2.1 ] is that the difficulty of the detection problem is not noticeably affected unless this ratio of sequence length to signal length grows exponentially with the number @xmath19 of sequences .
sections [ sec2.2 ] and [ sec2.3 ] provide optimal max - type tests that attain the detection boundary .
let @xmath41 if @xmath42 and @xmath43 if @xmath44 for some constant @xmath45 .
let @xmath46 be the greatest integer function and @xmath47 the number of elements in a set @xmath48 .
let @xmath49 denote expectation under @xmath50 .
we are interested here in the signal length @xmath51 in ( [ munt ] ) satisfying @xmath52 the case of @xmath53 varying sub - exponentially with @xmath19 will be considered in section [ extensions ] . we shall show that under ( [ c1 ] ) with @xmath26 , the asymptotic threshold detectable value of @xmath27 when @xmath20 and @xmath54 is @xmath55 \\[-8pt ] \nonumber & & \qquad = \cases { \sqrt{\log \bigl(1+n^{2 \beta-1+\zeta } \bigr ) } , & \quad $ \mbox{if } 0 \leq\zeta\leq1 - 4 \beta/3,$ \vspace*{2pt}\cr ( \sqrt{1-\zeta}-\sqrt{1- \zeta-\beta } ) \sqrt{2 \log n } , & \quad$\mbox{if } 1 - 4 \beta/3 < \zeta\leq1-\beta,$ \vspace*{2pt}\cr
\sqrt{n^{\beta+\zeta-1 } } , & \quad $ \mbox{if } \zeta > 1-\beta.$}\end{aligned}\ ] ] the first case @xmath56 can be further sub - divided into : ( a ) @xmath57 , under which @xmath58 and ( b ) @xmath59 , under which @xmath60 formula ( [ bnbz ] ) specifies the functional form of @xmath61 as a function of @xmath62 . since @xmath62 appears in the exponent in ( [ poly ] ) and in the third case of ( [ bnbz ] ) , @xmath61 is specified only up to multiplicative constants in these cases .
the boundary @xmath61 is an extension of the donoho
ingster jin boundary @xmath33 . in the case of a sparse mixture , @xmath63 , and ( [ c1 ] ) is satisfied with @xmath64 . by the second case in ( [ bnbz ] ) and by ( [ grow ] ) , @xmath65 when
furthermore , @xmath67 in ( [ poly ] ) recovers the detection boundary in the dense case @xmath68 established by cai , jeng and jin @xcite .
formula ( [ bnbz ] ) likewise recovers the detection boundary for the special case of only one sequence . for the scaled mean @xmath27 in ( [ munt ] ) , this boundary is known to be @xmath69 and is attained by the penalized scan ; see , for example , @xcite . to see how this special case is subsumed in the general setting above , set @xmath70 so that it suffices to consider @xmath71 in ( [ c1 ] ) to parametrize the scale of the signal @xmath72 . then set @xmath73 so that the signal is present in each of the @xmath19 sequences .
since the signals are aligned and have the same means , by sufficiency one can equivalently consider the one sequence @xmath74 of length @xmath53 obtained by summing the @xmath75 over @xmath5 .
dividing by @xmath76 to restore unit variance and formally plugging @xmath73 into ( [ poly ] ) gives a detection threshold for @xmath77 of @xmath78 .
this yields the above detection threshold for the one sequence problem apart from the multiplicative constant @xmath79 , which can be recovered with a more refined analysis in ( [ bnbz ] ) .
the general formula ( [ bnbz ] ) shows how the growth coefficient and the phase transitions of the @xmath80 growth are altered by the effect of multiple comparisons in the location of signals .
the formula also shows that in the case @xmath81 , the signal detection thresholds can grow polynomially with @xmath19 .
[ thmm1 ] assume that ( [ munt ] ) and ( [ c1 ] ) hold for @xmath82 , with @xmath83 and @xmath84 for some @xmath85 and @xmath86 .
under these conditions , there is no test that can achieve , at all @xmath87 , @xmath11 , @xmath88 the simple likelihood ratio of @xmath89 , for @xmath3 against ( [ munt ] ) , is @xmath90 , where @xmath91 with @xmath92 .
the key to proving theorem [ thmm1 ] ( details in section [ sec5 ] ) is to show that under the conditions of theorem [ thmm1 ] , @xmath93 that is , the likelihood ratio of the signal does not grow fast enough to overcome the noise due to the @xmath94 independent comparisons of length @xmath38 .
theorem [ thmm1 ] follows because the likelihood ratio test is the most powerful test . as an illustration ,
first consider sparse mixture detection .
that is , let @xmath22 and test @xmath95 for @xmath3 : @xmath96 against @xmath7 : @xmath97 and @xmath98 .
let @xmath99 be the ordered @xmath100-values of the @xmath101 s .
donoho and jin @xcite proposed to separate @xmath3 from @xmath7 by applying tukey s higher - criticism test statistic @xmath102 they showed that the higher - criticism test is optimal for sparse mixture detection . under @xmath3 ,
@xmath103 ; see @xcite , theorem 1 . under @xmath7 , the argument of @xmath104 at some @xmath105 is asymptotically larger than @xmath106 , when @xmath20 for some @xmath21 , and @xmath107 lies above the detection boundary @xmath108 . for @xmath27 lying below the detection boundary ,
it is not possible to separate @xmath3 from @xmath7 .
cai et al .
@xcite showed that optimality extends to @xmath109 .
we motivate the extension of the higher - criticism test to @xmath36 by first considering a fixed , known signal on the interval @xmath110 $ ] . by sufficiency ,
testing for an aligned signal there is the same as testing @xmath3 against @xmath7 for the sample @xmath111 .
let @xmath112 be the ordered @xmath100-values of the sample , and let @xmath113 .
we define the higher - criticism test statistic on this interval to be @xmath114 for @xmath115 , the constraint in ( [ hcnlj ] ) becomes @xmath116 , which agrees with the constraint in ( [ hcn ] ) .
as explained in @xcite , section 3 , the standardization of @xmath105 given in ( [ hcnlj ] ) has increasingly heavy tails as @xmath5 becomes smaller , so if hc@xmath117 is defined without constraints on @xmath105 , then it has large values frequently due to the smallest @xmath105 . for @xmath118 , the multiple comparisons when maximizing hc@xmath119 over @xmath120
necessitates a more restrictive constraint of @xmath121 .
the term @xmath122 appears also in the scan of the higher - criticism test statistic @xmath123 as a penalty that increases with @xmath124 to counter - balance the generally higher scores under @xmath3 for larger @xmath124 when maximizing hc@xmath125 over @xmath120 .
we will now specify the scanning set @xmath126 in ( [ phc ] ) . in applications
@xmath53 is often large , so maximizing hc@xmath117 over all @xmath120 and @xmath127 is computationally expensive ; the cost is @xmath128 .
we construct below an approximating set @xmath126 , similar to that in walther @xcite and rivera and walther @xcite , which has a computation cost of @xmath129 .
construction of @xmath126 : let @xmath130 , and let @xmath131 we define @xmath132 , where @xmath133 .
the specification of @xmath134 is so that for any @xmath135 , we can find @xmath136 for some @xmath137 such that @xmath138 and @xmath139 this property plays a part in ensuring that the loss of information due to restriction to @xmath126 is negligible .
[ thmm2a ] assume ( [ munt ] ) and that for some @xmath11 , ( [ c1 ] ) holds and @xmath140 , @xmath141 for some @xmath142 and @xmath143 .
under these conditions , @xmath144 can be achieved by testing with @xmath145 . for signal identification ,
when applying the penalized higher - criticism test statistics at a threshold @xmath146 : rank the pairs @xmath147 in order of descending values of hc@xmath148 , and remove those pairs with values less than @xmath146 .
starting with the highest - ranked pair and moving downward , remove a pair from the list if its interval overlaps with that of a higher - ranked pair still on the list by more than a fraction @xmath149 of its length .
jeng et al .
@xcite focused on the detection of signal segments that are well separated .
hence their signal identification procedure is restricted to @xmath150 .
zhang et al .
@xcite focused on both the detection of signal segments that are well separated , as well as the detection of overlapping or nested signal segments .
hence their procedure allows for @xmath151 .
if there are a finite number of well - separated signal segments , then intuitively , all the segments are identified with probability converging to 1 , in the sense that a segment with local maximum score , in a suitably defined neighborhood of each signal segment , is identified .
let @xmath152 if @xmath153 and @xmath154 otherwise .
this is the berk
jones @xcite test statistic that was first proposed as a more powerful alternative to the kolmogorov
smirnov test statistic for testing a distribution function ; see also owen @xcite .
jager and wellner @xcite showed that there is a class of test statistics that includes the berk jones and higher - criticism test statistics as special cases that can be used to detect sparse mixtures ( [ sparse ] ) optimally .
specifically for @xmath22 , the testing of @xmath97 and @xmath98 in ( [ sparse ] ) can be detected optimally by @xmath155 therefore , analogously to ( [ hcnlj ] ) , @xmath156 can optimally detect aligned signals on the interval @xmath110 $ ] . in theorem [ thmm2b ] below , we shall show that analogously to ( [ phc ] ) , the penalized berk
jones test statistic @xmath157 is optimal for aligned signals detection when the signal locations are unknown . [ thmm2b ] assume ( [ munt ] ) and that for some @xmath11 , ( [ c1 ] ) holds and @xmath140 , @xmath141 for some @xmath85 and @xmath158
. under these conditions , @xmath159 can be achieved by testing with @xmath160 . as in section [ sec2.2 ] ,
a sequential approach can be used to identify signals when the penalized berk
jones exceeds a specified threshold .
we shall introduce in section [ sec3.1 ] an alr that is optimal for detecting multi - sample aligned signals .
we then consider the special cases of detecting a sparse mixture ( @xmath22 with @xmath26 ) in section [ sec3.2 ] and multiscale detection in a single sequence ( @xmath161 with @xmath162 ) in section [ sec3.3 ] .
the alr builds upon the likelihood ratios @xmath163 as defined in ( [ lnlj ] ) , first by substituting @xmath27 by its asymptotic threshold detectable value , followed by integrating @xmath20 over @xmath62 and finally by summing over an approximating set for @xmath38 and @xmath37 . in view of ( [ c1 ] ) , let @xmath164 $ ] , and let @xmath165 see ( [ bnbz ] ) for the definition of @xmath61 . in the case of the alr
, we consider @xmath166 \,d \beta,\ ] ] where @xmath167 and @xmath168 are given in section [ sec2.2 ] . by ( [ brt ] ) , @xmath169 . the weights in ( [ ant ] )
are chosen for the following reason : since @xmath170 is a likelihood ratio for @xmath3 , it has expectation 1 under @xmath3 .
hence it follows from ( [ ant ] ) that @xmath171 from ( [ 6pi ] ) , it follows that under @xmath3 , @xmath172 uniformly over @xmath19 and @xmath53 . if the aligned signals under @xmath7 are such that @xmath173 as @xmath26 , then @xmath174 is achieved by simply selecting rejection thresholds going to infinity slowly enough . the alr ( [ ant ] ) is , by its construction , optimal when @xmath175 .
it is not designed to be optimal at other @xmath27 .
however , there is really no point in being optimal at smaller @xmath27 , where the maximum power that can be attained is small . at larger @xmath27 , the alr test has power close to 1 , so there is not much more to be gained in being optimal there . by focusing only on the boundary detectable values ,
we remove the noise due to the consideration of unproductive likelihood ratios associated with too large and too small @xmath27 .
[ thmm2 ] assume ( [ munt ] ) and that for some @xmath11 ( [ c1 ] ) holds and @xmath140 , @xmath141 for some @xmath85 and @xmath176 .
under these conditions , @xmath173 . hence @xmath177 can be achieved by testing with @xmath178 . among the three optimal tests that we propose here ,
the alr is the most intuitive , and the proof of its optimality is also the most straightforward .
however , its computation involves the evaluation of a nonstandard integral , and its form is closely linked to normal errors . on the other hand , the penalized scans involve no integrations in their computations , and the @xmath100-values in their expressions are not tied to normal errors .
this setting has been studied in @xcite and discussed briefly in section [ sec2 ] .
it corresponds to the special case @xmath22 in the above theory , and our test statistic @xmath178 simplifies as follows : @xmath22 implies @xmath179 , and @xmath126 contains only @xmath180 .
hence @xmath181 where @xmath182 is essentially the cai
jeng jin detection boundary @xmath183 for @xmath184 , and the donoho
ingster jin detection boundary @xmath185 for @xmath186 .
[ thmm4 ] assume ( [ sparse ] ) with @xmath187 and @xmath188 for some @xmath189 and @xmath190 .
under these conditions , @xmath191 and ( [ typei ] ) can be achieved by testing with @xmath192 .
let @xmath161 and @xmath193 .
the resulting @xmath73 is not covered by our general theory , but it is a boundary case , and therefore it is of interest to see whether our general statistic @xmath178 still allows optimal detection in this important special case . for this testing problem in a single sequence with @xmath162 , it is known that the critical detectable value of @xmath194 is @xmath195 , where @xmath196 , and that the popular scan statistic is suboptimal except for signals on the smallest scales ; see chan and walther @xcite .
it is also shown there that optimal detection can be achieved by modifying the scan with the penalty method introduced by dmbgen and spokoiny @xcite , or by employing the condensed average likelihood ratio .
note that when analyzing a single sequence we know a priori that @xmath73 , and therefore it makes sense to set @xmath62 to 0 in the definition of @xmath178 rather than integrating @xmath62 over @xmath197 .
the resulting statistic is @xmath198,\ ] ] where @xmath199 .
the test statistic @xmath200 is able to achieve the detection boundary @xmath201 simply because it optimizes detection power there : [ thmm3 ] if there exist @xmath38 and @xmath37 such that @xmath202 , with @xmath203 as @xmath162 , then @xmath204 and ( [ typei ] ) can be achieved by testing with @xmath200 .
efron and zhang @xcite applied local f.d.r . to detect cnv in multi - sample dna sequences .
measurements from @xmath205 probes were taken on each chromosome 1 of @xmath206 glioblastoma subjects from the cancer genome atlas project @xcite . at each probe on each sequence ,
the moving averages of the readings over windows of length @xmath207 were normalized .
these normalized averages correspond to the @xmath208 scores defined just before ( [ lnlj ] ) . the computed local f.d.r .
of the scores at each @xmath120 determined the conclusion of an aligned signal there .
the scientific purpose is to detect rare inherited cnv that may occur in a small fraction , perhaps 5% , of the population .
consider , for example , @xmath209 and @xmath210 .
the solution of @xmath211 [ see ( [ c1 ] ) ] is @xmath212 .
the solution of @xmath213 is @xmath214 .
since @xmath215 , we are under the second case in ( [ bnbz ] ) . based on ( [ bnbz ] ) ,
the signal - to - noise ratio ( for a single observation in a variant segment ) required for successful detection in a mixture with 5% variant is then @xmath216 it is known from earlier studies that the sequence between probes 8800 and 8900 contains two genes that enhance cell death .
copy number losses of these genes promote unregulated cell growth , leading to tumor .
the display in figure [ fig1 ] ( left ) shows that the likelihood at marker @xmath120 ( for a signal on the probe interval @xmath217 ) , @xmath218 \,d \beta,\ ] ] is maximized at @xmath219 .
the display in figure [ fig1 ] ( right ) shows that the likelihood at marker @xmath219 for variant fraction @xmath20 , @xmath220 is maximized at @xmath221 .
this translates to an estimated 4% of the population tested having copy number losses in the probe interval @xmath222 . .
( right ) plot of likelihood against variant fraction @xmath20 at marker @xmath223 . ]
cai , jeng and jin @xcite and cai and wu @xcite showed that the hc test is optimal for heteroscedastic and more general mixtures , respectively .
arias - castro and wang @xcite analyzed the detection capabilities of distribution - free tests for null hypotheses that are not fully specified .
jeng , cai and li @xcite showed that the hc test statistic is optimal for detecting heteroscedastic aligned sparse signals when assuming that the signal length is very small and that @xmath53 does not grow rapidly with @xmath19 .
however , when the aligned signals may range over multiple scales , the penalty terms introduced in section [ sec2.2 ] are critical in ensuring optimality of the hc test .
below , we shall show how the detection boundary of cai et al .
@xcite looks for general @xmath40 asymptotics , and show that the adaptive optimality of the hc and bj tests extends to heteroscedastic signals when their penalties , as given in section [ sec2 ] , are applied .
this brings home the point that the penalties are not tied down to a particular model .
following jeng et al .
@xcite , we assume in place of ( [ proto ] ) that @xmath224 under the null hypothesis @xmath3 of no signals , @xmath225 for all @xmath5 and @xmath6 . under the alternative hypothesis @xmath7 of aligned signals , there exists an unknown @xmath8 of disjoint intervals @xmath9 $ ] , @xmath11 , such that for the @xmath10th interval , @xmath226 , \nonumber \\[-8pt ] \\[-8pt ] \nonumber i_n^{(k ) } & \sim & \operatorname { bernoulli } \bigl ( \pi_n^{(k ) } \bigr),\end{aligned}\ ] ] and @xmath227 otherwise , with @xmath228 , @xmath229 and @xmath230
. we shall denote @xmath231 by @xmath27 , @xmath232 by @xmath38 and so forth .
let @xmath233 be such that @xmath234 , and for @xmath235 and @xmath236 , let @xmath237 [ thmm6 ] assume ( [ x2 ] ) and ( [ unt ] ) .
if for all @xmath11 , ( [ c1 ] ) holds and @xmath238 , @xmath239 for some @xmath142 and @xmath86 , then there is no test that can achieve , at all @xmath87 , @xmath11 , @xmath240 conversely , if for some @xmath11 , ( [ c1 ] ) holds and @xmath241 , @xmath141 for some @xmath142 and @xmath143 , then ( [ pp0 ] ) can be achieved by the penalized hc and bj tests .
it can be checked that setting @xmath242 will recover for us the boundary for aligned signals in jeng et al .
incidentally , they assumed that @xmath243 which effectively brings us to the case @xmath242 .
corollary [ cor2 ] below extends the optimality of the hc test in jeng et al .
@xcite to multiscale signal lengths , by introducing the penalty terms as described in section [ sec2 ] . in place of ( [ c1 ] ) , let @xmath244 .
[ cor2 ] assume ( [ x2 ] ) and ( [ unt ] ) .
theorem [ thmm6 ] holds under ( [ poly2 ] ) with @xmath245 and @xmath246 .
we say that @xmath247 if @xmath248 and @xmath249 , and that @xmath250 if @xmath251 . we start with the proof of theorem [ thmm1 ] in section [ sec5.1 ] , that detection is asymptotically impossible below the detection boundary @xmath61 , followed by the proofs of theorem [ thmm2 ] ( in section [ sec5.2 ] ) and theorem [ thmm3 ] ( in section [ sec5.3 ] ) , that the average likelihood ratio test is optimal .
these proofs are consolidated in this section as they are unified by a likelihood ratio approach .
since the detection problem is easier when @xmath252 compared to @xmath23 , we may assume without loss of generality that @xmath23 under @xmath7 in all the proofs . for @xmath64
the claim of the theorem reduces to theorems proved by ingster @xcite in the sparse case and by cai , jeng and jin @xcite in the dense case .
let @xmath253 , and set @xmath254 , so @xmath255 , set @xmath256 and @xmath257 .
let @xmath258 , and let each @xmath259 , @xmath260 , be of the form @xmath261 , with all @xmath262 $ ] disjoint from each other , and from @xmath263 $ ] .
let @xmath264 , \qquad l_i = \prod_{n=1}^n l_{ni}.\ ] ] since @xmath262 $ ] are disjoint , @xmath265 are independent . we take note that @xmath266 have identical distributions which are unchanged when we switch from @xmath3 to @xmath7 .
in contrast , the distribution of @xmath267 changes when we switch from @xmath3 to @xmath7 .
consider @xmath268 which is the likelihood ratio when @xmath37 is equally likely to take one of @xmath269 possible values spaced at least @xmath38 apart , as explained above .
this we assume without loss of generality . if we are able to find @xmath270 such that both @xmath271 are satisfied , then @xmath272 is unable to achieve ( [ typei ] ) .
if so , then no test is able to achieve ( [ typei ] ) because the likelihood ratio test is the most powerful test
. _ case _ 1 : @xmath273 , @xmath274 , @xmath275 with @xmath276 . under @xmath7 , @xmath277 are i.i.d .
@xmath278n@xmath279 . let @xmath280^n = \exp \bigl\ { \bigl[1+o(1 ) \bigr ] n^{\zeta-2 \varepsilon } \bigr\}.\ ] ] hence ( [ l1 ] ) holds .
let @xmath281 .
since @xmath282 , @xmath283^n = \lambda_n.\ ] ] we check in appendix [ appa ] that lyapunov s condition holds .
hence @xmath284 since @xmath285(n^{\zeta-2 \varepsilon } + n^\zeta)/2 \ } \gg\lambda_n$ ] , ( [ pan ] ) follows from ( [ clt ] ) .
_ case _ 2 : @xmath286 , @xmath287 where @xmath288 , @xmath289 , @xmath290 .
let @xmath291 and define , for @xmath292 , @xmath293.\ ] ] check that @xmath294 \\
& = & 1+\pi_n \bigl[e^{\mu_n^2 } \phi(x \sqrt{2 \log n}-2 \mu_n)-\phi(x \sqrt{2 \log n}-\mu_n ) \bigr].\end{aligned}\ ] ] since @xmath295 and @xmath296 when @xmath297 , it follows that @xmath298^n\nonumber \\ & \leq & \bigl[1+\pi_n^2 e^{\mu_n^2 } \phi(x \sqrt{2 \log n}-2 \mu_n ) \bigr]^n
\nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & \bigl[1+o \bigl(n^{2(y^2-x^2)-2 \varepsilon+2(x - y)^2-(2y - x)^2 } \bigr)/ \sqrt{\log n } \bigr]^n \\ & = & \exp \bigl[o \bigl(n^{1-x^2 - 2 \varepsilon } \bigr)/\sqrt{\log n } \bigr ] = \exp \bigl[o \bigl(n^{\zeta-2 \varepsilon } \bigr)/\sqrt{\log n } \bigr ] .
\nonumber\end{aligned}\ ] ] next we apply @xmath299 to show that @xmath300 \\[-8pt ] \nonumber & = & o_p \bigl(n^{\zeta-\varepsilon } \sqrt{\log n } \bigr ) .
\nonumber\end{aligned}\ ] ] by ( [ 5.6 ] ) , ( [ 5.8 ] ) and @xmath301 , ( [ l1 ] ) holds for @xmath302 .
let @xmath281 .
let @xmath303 { \mathbf i}_{\ { y_{ni } \leq x \sqrt{2 \log n } \}}$ ] and @xmath304 .
we check that @xmath305 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & 1- \bigl[c+o(1 ) \bigr ] n^{-\beta - y^2-\varepsilon}/\sqrt{\log n},\end{aligned}\ ] ] where @xmath306 . from this and @xmath307 , we conclude that @xmath308 n^{\zeta-\varepsilon } /\sqrt { \log n } \bigr\}.\ ] ] since @xmath309 ^ 2 $ ] for @xmath310 , it follows that @xmath311 \\[-8pt ] \nonumber & \geq & \biggl ( \frac{e_0({\widetilde}l_{1i}^2)}{[e_0({\widetilde}l_{1i})]^2}-1 \biggr ) \kappa_n^2 \sim \operatorname{var}_0({\widetilde}l_{1i } ) \kappa_n^2,\end{aligned}\ ] ] and by ( [ e0 t ] ) , @xmath312 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } - \bigl[e_0({\widetilde}l_{1i}-1 ) \bigr]^2 \\ & \sim & c_1 n^{\zeta-1 - 2 \varepsilon}/\sqrt{\log n } , \nonumber\end{aligned}\ ] ] where @xmath313^{-1}$ ] .
we check that lyapunov s condition holds and conclude that @xmath314 by ( [ kappan])([vart ] ) , @xmath315 n^\zeta/2 \ } \gg\lambda_n$ ] , and so ( [ pan ] ) holds , but with @xmath316 replaced by @xmath317 .
the variability of @xmath316 is larger than that of @xmath317 , and hence ( [ pan ] ) for @xmath316 holds as well .
_ case _ 3 : @xmath318 , @xmath319 , @xmath275 with @xmath320 .
let @xmath321 , @xmath292 , and define @xmath322 as in ( [ ln1h ] ) . since @xmath323
, it follows that @xmath324 we next apply the inequality @xmath325 to show that @xmath326 n \pi_n e \bigl(\mu_n y_{11}- \mu_n^2/2|i_1=1 \bigr ) \stackrel{p } { \sim } n^{\zeta-\varepsilon}.\ ] ] it follows from ( [ p1log ] ) , ( [ logln11 ] ) and @xmath301 that ( [ l1 ] ) holds for @xmath327 .
let @xmath281 and @xmath328 for all @xmath329 .
then @xmath330 hence @xmath331^{-1}$ ] and @xmath332
if @xmath333 occurs , then @xmath334 \bigl[-n^{1-\beta-\varepsilon}+ \bigl(n^{1-\beta}/2 \bigr)n^{\zeta -1+\beta}/2 \bigr ] \bigr\ } \\ & = & \exp \bigl\ { \bigl[1+o(1 ) \bigr ] n^\zeta/4 \bigr\ } \gg \lambda_n . \nonumber\end{aligned}\ ] ] since typically @xmath335 , we can conclude ( [ pan ] ) from ( [ hexi ] ) and ( [ liarray ] ) .
let ( [ munt ] ) and ( [ c1 ] ) hold with @xmath256 and @xmath336 .
let @xmath337 satisfy ( [ llr ] ) , and note that by ( [ c1 ] ) and ( [ brt ] ) , @xmath338 .
hence for @xmath253 and @xmath19 large , @xmath339 b_n(\beta,\zeta ) \nonumber \\[-8pt ] \\[-8pt ] \nonumber & \geq & b_n(\eta,\zeta_{\ell_t^*,nt } ) , \qquad \beta_1 \leq\eta \leq \beta_2,\end{aligned}\ ] ] for any @xmath340 lying in the interior of @xmath341 , and this inequality can also be checked for @xmath64 .
let @xmath342 since @xmath343 , in view of ( [ ant ] ) and ( [ e1y ] ) , theorem [ thmm2 ] follows from @xmath344 to show ( [ check ] ) , it suffices to check that @xmath345 when @xmath346 , and @xmath336 with @xmath347 . to see why ( [ check2 ] ) leads to ( [ check ] ) ,
define @xmath348 , let @xmath349 and let @xmath350\dvtx l(\eta ) < m c_{n \zeta } \bigr\}.\ ] ] by ( [ check2 ] ) , @xmath351 and hence @xmath352 .
if @xmath353 , then @xmath354 . since @xmath349 can be chosen arbitrarily large , ( [ check ] ) holds . to cross - reference the results in the proof of theorem [ thmm1 ] more easily , we relabel @xmath355 and @xmath356 in ( [ leta ] ) by @xmath37 and @xmath38 , respectively , and rephrase ( [ check2 ] ) as @xmath357 when @xmath256 and @xmath336 .
_ case _ 1 : @xmath56 , @xmath358 , @xmath336 with @xmath359 . under @xmath7 , @xmath360 , where @xmath361 are
@xmath362 . let @xmath363 $ ] , and @xmath364 } { 1+n^{-\beta } [ \exp(\mu_n z_{n1}-\mu_n^2/2)-1 ] } \biggr).\end{aligned}\ ] ] since for @xmath365 , @xmath366 it follows that @xmath367 \\[-8pt ] \nonumber & \stackrel{p } { \sim } & n \pi_n \phi(-\mu_n ) \times \cases { n^{\beta-1+\zeta } , & \quad $ \mbox{if } 0 \leq\zeta < 1 - 2 \beta,$ \vspace*{2pt}\cr n^{-\beta } \sqrt{2 } , & \quad $ \mbox{if } \zeta = 1 - 2 \beta,$ \vspace*{2pt}\cr n^{-\beta } e^{3 \mu_n^2/2 } , & \quad $ \mbox{if } 1 - 2 \beta < \zeta < 1 - 4 \beta /3,$ \vspace*{2pt}\cr \log2 , & \quad $ \mbox{if } \zeta=1 - 4 \beta/3.$}\end{aligned}\ ] ] in the above
, we apply the relation @xmath368 \sim n^{-\beta } ( e^{\kappa \mu_n^2/2}-1)$ ] for @xmath369 , with the exception @xmath370 \sim\log2\qquad \mbox{when } \zeta = 1 - 4 \beta/3.\ ] ] since @xmath371 when @xmath372 , @xmath373 when @xmath374 and @xmath375 when @xmath376 , it follows from checking each of the cases in ( [ logl11 ] ) that @xmath377 we shall next obtain lower bounds of @xmath378 .
we apply taylor s expansion @xmath379 u^2 $ ] to show that @xmath380 and @xmath381 to show that @xmath382 it follows from ( [ elogln1 ] ) and ( [ elogln12 ] ) that @xmath383 .
since @xmath384 , we can conclude ( [ n3zeta ] ) from ( [ logl11n ] ) .
_ case _ 2 : @xmath286 , @xmath385 where @xmath288 , @xmath386 , @xmath336 .
define @xmath387 { \mathbf i}_{\ { z_{n1 } \leq x \sqrt{2 \log n } \ } } , \\ { \widetilde}l_1 ^ 0 & = & \prod_{n=1}^n { \widetilde}l_{n1 } , \\
{ \widetilde}l_1 ^ 1 & = & \prod_{n : i_n=1 } \biggl ( \frac{1+n^{-\beta } [ \exp ( \mu_n y_{n1}-\mu_n^2/2)-1 ] { \mathbf i}_{\ { z_{n1 } \leq x \sqrt{2 \log n } \ } } } { 1+n^{-\beta } [ \exp(\mu_n z_{n1}-\mu _
n^2/2)-1]{\mathbf i}_{\ { z_{n1 } \leq x \sqrt{2 \log n } \ } } } \biggr).\end{aligned}\ ] ] by ( [ vincrease ] ) , @xmath388}{1+n^{-\beta } [ \exp(\mu_n x \sqrt{2 \log n}-3 \mu _ n^2/2)-1 ] } \biggr ) \\
& \sim & n \pi_n \phi(-y \sqrt{2 \log n } ) \log2 \sim c ( \log2 )
n^{\zeta+\varepsilon}/\sqrt{\log n } , \nonumber\end{aligned}\ ] ] where @xmath306 .
recall that @xmath389^{-1}$ ] .
apply taylor s expansion @xmath379 u^2 $ ] to show that @xmath390 \nonumber \\ & & \qquad\quad { } - \bigl[1/2+o(1 ) \bigr ] n^{-2 \beta } \bigl[e^{\mu_n^2 } \phi \bigl((2y - x ) \sqrt{2 \log n } \bigr ) \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & \hspace*{98pt}\qquad\quad { } -2 \phi(y \sqrt{2 \log n})+\phi(x \sqrt{2 \log n } ) \bigr ] \\ & & \qquad = - \bigl[1+o(1 ) \bigr ] cn^{-\beta - y^2}/\sqrt{\log n}- \bigl[1/2+o(1 ) \bigr ] c_1 n^{-x^2}/\sqrt{\log
n } \nonumber \\ & & \qquad = - \bigl[1+o(1 ) \bigr ] ( c+c_1/2 ) n^{\zeta-1}/\sqrt{\log n } , \nonumber\end{aligned}\ ] ] and @xmath381 to show that @xmath391 it follows from ( [ elogt ] ) and ( [ elogt2 ] ) that @xmath392 if @xmath253 and @xmath393 if @xmath64 . since @xmath394 , we can conclude ( [ n3zeta ] ) from ( [ loghex ] ) . _ case _ 3 : @xmath318 , @xmath395 , @xmath396 .
the inequality @xmath397 \bigr\}\ ] ] leads to @xmath398 n \pi_n \bigl[\mu_n
e_1(y_{n1})- \mu_n^2/2-\beta\log n \bigr ] \\ & = & -2n^{1-\beta}+ \bigl[1+o_p(1 ) \bigr ] n^{\zeta+\varepsilon}/2 \stackrel { p } { \sim } n^{\zeta+\varepsilon}/2,\end{aligned}\ ] ] and from this , we can conclude ( [ n3zeta ] ) .
let @xmath399 and let @xmath400 be such that ( [ llr ] ) holds .
hence @xmath401 \sqrt { \ell_t^*/ \ell_t } = \bigl[1-o \bigl(r^{-1/2 } \bigr ) \bigr ] \bigl[b_t(\ell_t)+c_t \bigr].\ ] ] since @xmath402 and@xmath403 , it follows from ( [ e1ystar ] ) that @xmath404 we check that under ( [ ctprime ] ) , @xmath405}{[\log ( t/\ell_t^*)]^3 ( t/\ell_t^ * ) } \stackrel{p } { \rightarrow } \infty.\ ] ] since @xmath406 ^ 3(t/\ell_t))$ ] , it follows from ( [ at ] ) that @xmath407 .
we shall prove theorem [ thmm2b ] in section [ sec6.1 ] , that the penalized berk jones test is optimal , and theorem [ thmm2a ] in section [ sec6.2 ] , that the penalized higher criticism test is optimal as well . in lemma [ lem1 ]
below , we show that the type i error probability of the penalized berk
jones test statistic goes to zero for the threshold @xmath408 .
we do this in more generality than is required for proving theorem [ thmm2b ] . for each @xmath127 and @xmath120 , we assume only that @xmath409 are i.i.d .
@xmath410 random variables under @xmath3 . hence we allow for @xmath75 to be non - gaussian , and for @xmath411 to be dependent random variables under @xmath3 .
[ lem1 ] assume that for each @xmath127 and @xmath120 , @xmath412 in ( [ bjnlj ] ) and ( [ pbj ] ) are the ordered values of i.i.d .
@xmath410 random variables .
then @xmath413 let @xmath414 .
for each @xmath415 and @xmath416 , let @xmath417 be such that @xmath418 .
let @xmath419= n^{-1}\times\break \sum_{n=1}^n { \mathbf i } _ { \ { y_{n \ell j } \geq z(t ) \}}$ ] , where @xmath420 denotes the upper @xmath6-quantile of the standard normal . by the chernoff hoeffding inequality , @xmath421 and hence by bonferroni s inequality , @xmath422 by ( [ brt ] ) , @xmath423 .
since @xmath424 for @xmath425 , so @xmath426 , and by ( [ pbj ] ) and ( [ bjcs ] ) , @xmath427 and ( [ l1 ] ) holds .
proof of theorem [ thmm2b ] let @xmath428 \in b_{r , t}$ ] be such that @xmath429 and @xmath430 ; see ( [ llr ] ) .
since @xmath431 , in view of lemma [ lem1 ] , it remains for us to show that if @xmath256 and @xmath396 for @xmath432 , then we can find @xmath433 such that in each case below , @xmath434 _ case _
1(a ) : @xmath435 , @xmath436 .
let @xmath437 .
except when @xmath374 , we have @xmath438 and @xmath439 by taylor s expansion , @xmath440 when @xmath441 and @xmath442 .
moreover , the standard error of @xmath443 $ ] is small relative to ( [ e1s ] ) .
hence ( [ pk ] ) holds because @xmath444 when @xmath374 , @xmath445 and @xmath446 this leads to ( [ 1ak ] ) with @xmath447 replaced by @xmath448 , and then to ( [ pk ] ) .
1(b ) : @xmath449 , @xmath450 ( @xmath451 , where @xmath452 ) . let @xmath453 .
then @xmath454 \\[-8pt ] \nonumber & \sim & ( c \sqrt{2 \pi\log n})^{-1 } n^{-2 \beta-(\zeta -1)/2+\varepsilon } , \\
\label{1bv } \operatorname{var}_1 \bar s_n & \sim & n^{-1 } \bigl[t_n+\pi_n \phi \bigl(-b_n ( \beta,\zeta ) \bigr ) \bigr].\end{aligned}\ ] ] we claim that a consequence of ( [ 1btn])([1bv ] ) is that @xmath455 by ( [ 1bdiff2 ] ) , @xmath456 and hence @xmath457 by ( [ 1btn ] ) , the solution in @xmath458 of @xmath459 satisfies @xmath460,\ ] ] where @xmath461 if @xmath462 and @xmath463 if @xmath64 . hence by ( [ 1bgg ] ) and @xmath464 , as @xmath465 and @xmath466 , ( [ pk ] ) holds .
it remains for us to show ( [ 1bdiff2 ] ) . by ( [ 1btn ] ) ,
the exponent of @xmath19 in @xmath467 is @xmath468 , which is smaller than the exponent in @xmath19 of @xmath469 ; see ( [ 1bdiff ] ) .
therefore , @xmath470 the leading exponent of @xmath19 in var@xmath471 is @xmath472,\ ] ] and therefore by ( [ 1bdiff ] ) , var@xmath473 .
this , together with ( [ 6.46 ] ) , implies ( [ 1bdiff2 ] ) .
_ case _ 2 : @xmath474 , @xmath475 , where @xmath288 , @xmath476 .
let @xmath477 $ ] .
then @xmath478 which is large relative to @xmath433 , and @xmath479 \\ & \sim & ( 2y \sqrt{\pi\log n})^{-1 } n^{\zeta-2+\varepsilon } = o \bigl((e_1 \bar s_n)^2 \bigr).\end{aligned}\ ] ] therefore @xmath480 , and by @xmath481 , as @xmath482 and @xmath483 , @xmath484 for some @xmath485 , and ( [ pk ] ) therefore holds .
_ case _ 3 : @xmath318 , @xmath486 .
let @xmath487 .
then @xmath488 therefore @xmath489 and by @xmath490 , as @xmath482 and @xmath483 , @xmath491 and ( [ pk ] ) therefore holds . in lemma [ lem2 ] below
, we show that the type i error probability of the penalized higher criticism test statistic goes to zero for the threshold @xmath492 .
again as in lemma [ lem1 ] , we do this more generally than is required for proving theorem [ thmm2a ] .
[ lem2 ] assume that for each @xmath127 and @xmath120 , @xmath412 in ( [ hcnlj ] ) and ( [ phc ] ) are the ordered values of i.i.d .
@xmath410 random variables .
then @xmath493 we first modify ( [ psn])([6.36a ] ) , in the proof of lemma [ lem1 ] , step - by - step to show that for @xmath494 , @xmath495 we then combine this with ( [ kq2 ] ) in appendix [ appb ] to show that @xmath496 \\[-8pt ] \nonumber & & \qquad \rightarrow0.\end{aligned}\ ] ] therefore it suffices to show that @xmath497 uniformly over @xmath415 .
the left - hand side of ( [ 6.47 ] ) is @xmath498 uniformly over @xmath499 , and @xmath500 uniformly over @xmath501 , so ( [ 6.47 ] ) indeed holds .
proof of theorem [ thmm2a ] we apply the proofs for cases 1 and 2 in theorem [ thmm2b ] to show that there exist @xmath502 and @xmath503 $ ] [ @xmath504 for case 1 and @xmath505 for case 2 ] such that @xmath506 hence by @xmath507 for @xmath153 , @xmath508 since @xmath509 , the type ii error probability indeed goes to zero .
we prove theorem [ thmm6 ] here and in sections [ sec7.1 ] and [ sec7.2 ] . in section [ sec7.3 ] , we prove corollary [ cor2 ] .
let @xmath510 , @xmath511 , and for @xmath512 , let @xmath513 , with all @xmath262 $ ] disjoint from each other , and from @xmath263 $ ] .
let @xmath514 , where @xmath515 - 1 \biggr\}\ ] ] is the likelihood ratio of @xmath516 and @xmath517 .
below , we go over the relevant cases to show that there exists @xmath270 satisfying ( [ l1 ] ) and ( [ pan ] ) when @xmath290 .
this implies that there is no test able to achieve ( [ pp0 ] ) .
we shall only consider @xmath235 as the case @xmath518 has been covered in theorem [ thmm1 ] .
_ case _ 1 : @xmath519 ( @xmath520 ) , @xmath521 .
. by ( [ lni6 ] ) , @xmath523 n^{\zeta -1 - 2 \varepsilon},\ ] ] and therefore ( [ l1 ] ) holds with @xmath524 n^{\zeta-2 \varepsilon } \})$ ] . for @xmath281 , @xmath282 , @xmath525 and @xmath526 .
we check lyapunov s conditions to conclude ( [ clt ] ) and ( [ pan ] ) . _
case _ 2 : @xmath527 , @xmath528 , @xmath529 where @xmath288 and @xmath530 .
let @xmath531 check that @xmath532 and @xmath533 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & 1 + \bigl[c_1+o(1 ) \bigr ] \pi_n n^{\zeta-1 + 2 \beta}/\sqrt{\log n}\end{aligned}\ ] ] for some @xmath534 , hence @xmath535^n\nonumber \\
& = & \bigl\ { 1 + \bigl[c_1+o(1 ) \bigr ] \pi_n^2 n^{\zeta-1 + 2 \beta}/\sqrt{\log n } \bigr\ } ^n \\ \nonumber & = & \exp \bigl\ { \bigl[c_1+o(1 ) \bigr ] n^{\zeta-2 \varepsilon}/ \sqrt{\log n } \bigr\}.\end{aligned}\ ] ] next , we apply @xmath536 to show that @xmath537 \\[-8pt ] \nonumber & = & o_p \bigl ( n \pi_n \phi \bigl(- \sqrt{2(1-\zeta-\beta ) \log n } \bigr ) \sqrt{\log n } \bigr ) = o_p \bigl(n^{\zeta-\varepsilon } \bigr).\end{aligned}\ ] ] by ( [ 62a ] ) , ( [ 62b ] ) and @xmath301 , ( [ l1 ] ) holds for @xmath538 . for @xmath281 , let @xmath539 then by a change - of - measure argument , @xmath540 \\ & = & 1- \bigl[c_2+o(1 ) \bigr ] n^{-\zeta-1-\varepsilon}/\sqrt{\log n},\end{aligned}\ ] ] where @xmath541 , therefore @xmath542 n^{\zeta-\varepsilon } /\sqrt { \log n } \bigr\}.\ ] ] moreover , @xmath309 ^ 2 $ ] for @xmath543 , and therefore , by ( [ vnvar ] ) , @xmath544 \kappa_n^2 \operatorname{var}_0({\widetilde}l_{1i}).\ ] ] by ( [ m01 ] ) and a change - of - measure argument , @xmath545 we check lyapunov s condition to conclude ( [ lya2 ] ) and ( [ pan ] ) . by lemma [ lem1 ] ,
setting @xmath492 leads to @xmath546 . to show @xmath547 , it suffices to find @xmath548 such that @xmath549 , @xmath430 and @xmath550 for some @xmath551 and @xmath552 , where @xmath553 and @xmath336 , @xmath554 .
1(a ) : @xmath435 , @xmath555 . let @xmath556 .
then @xmath557 \sim c_3 n^{-\beta+\varepsilon/2},\ ] ] where @xmath558 , and since @xmath559 therefore @xmath560 .
since @xmath440 when @xmath561 and @xmath562 , ( [ k8 ] ) holds for @xmath563 . _ case _
1(b ) : @xmath564 ( @xmath520 ) , @xmath565 , where @xmath566 . let @xmath567 .
then @xmath568 where @xmath569 , and @xmath570 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & \sim & c_6 n^{-c_4 ^ 2(1+\tau)/[2(1-\tau)]^2-\beta+\varepsilon}/\sqrt { \log n},\end{aligned}\ ] ] where @xmath571 .
we claim that @xmath572 \biggr ) = o \bigl((e_1 \bar s_n)^2 \bigr).\ ] ] by ( [ 8tn])([8claim ] ) and @xmath573 , @xmath574 \\[-8pt ] \nonumber & = & c_7 n^{\zeta-1 + 2 \varepsilon}/\sqrt{\log n}\end{aligned}\ ] ] for some @xmath575 .
check that the inequality @xmath576 reduces to @xmath577 . therefore by ( [ 8tn ] ) , @xmath578 for some @xmath552 , and the root of @xmath579 satisfies @xmath580 . since @xmath581 as @xmath582 and @xmath583 , ( [ 8quad ] ) implies ( [ k8 ] ) .
it remains to show ( [ 8claim ] ) by comparing the leading exponent in @xmath19 of the terms .
that is , it remains to show that @xmath584 summarized as @xmath585 .
the inequality @xmath586 reduces to @xmath587 , which holds trivially , whereas @xmath588 reduces to @xmath589 which holds because @xmath590 when @xmath591 , and it is assumed that @xmath592 . hence ( [ 8claim ] ) holds .
_ case _ 2 : @xmath593 , @xmath528 , @xmath529 where @xmath288 and @xmath594 .
let @xmath595 $ ] .
then @xmath596 where @xmath597 .
moreover , @xmath598 therefore @xmath599 , and since @xmath600 as @xmath482 and @xmath601 , @xmath602 for some @xmath603 , ( [ k8 ] ) holds for @xmath604 . by similar arguments , ( [ k8 ] )
holds for @xmath605 when @xmath606 .
_ case _ 3 : @xmath607 , @xmath608 , @xmath555 .
let @xmath609 $ ] .
then @xmath610 for some @xmath611 .
since @xmath608 , therefore the exponent of @xmath19 in ( [ 3sn ] ) is at least @xmath612 , and so @xmath613 .
we apply the first relation in ( [ c9 ] ) to conclude ( [ k8 ] ) , for both @xmath614 and @xmath615 . by lemma [ lem2 ]
, setting @xmath616 leads to @xmath546 .
let @xmath617 it was shown in ( [ k8 ] ) that in each case above , @xmath618 for some @xmath619 . since @xmath620 for @xmath153 and @xmath621 , @xmath547 .
consider first the penalized hc test . by lemma [ lem2 ] , setting @xmath616 leads to @xmath546 . in the case
@xmath336 , the arguments above and in theorem [ thmm2a ] show that @xmath622 by ( [ poly2 ] ) , @xmath623 , and therefore @xmath547 , and ( [ pp0 ] ) holds . by similar arguments , ( [ pp0 ] )
holds for the penalized bj test .
we check in particular lyapunov s condition to conclude ( [ clt ] ) .
let @xmath619 to be specified .
it follows from taylor s expansion that @xmath624 for some @xmath625 chosen large enough . if @xmath626 , then @xmath627 u^{2+\delta } = ( 3u)^{2+\delta}.\ ] ] by combining ( [ a1 ] ) and ( [ a2 ] ) , we conclude that @xmath628 where @xmath629 .
we apply ( [ a3 ] ) with @xmath630 $ ] on ( [ lni ] ) to show that @xmath631 \\[-8pt ] \nonumber & & \qquad\leq \bigl[1+c^ * p_n^2 \exp
\bigl(\mu_n^2 \bigr)+c_2 \pi _ n^{2+\delta } \exp \bigl ( \mu_n^2 ( 1+\delta ) ( 2+\delta)/2 \bigr ) \bigr]^n.\end{aligned}\ ] ] since @xmath632 and @xmath633 , by ( [ edelta ] ) , @xmath634 \bigl(c^ * n^{\zeta-2 \varepsilon}+c_2 n^{\zeta-2 \varepsilon+\kappa } \bigr ) \bigr\},\ ] ] where @xmath635 .
let @xmath619 be small enough such that @xmath636 .
since @xmath637 and @xmath638 , we get from ( [ a4 ] ) that @xmath639 \bigl(n^\zeta+c_2 n^{\zeta- \varepsilon } \bigr ) \bigr\}.\ ] ] on the other hand , @xmath640 and @xmath641^{1+\delta/2 } = \exp \bigl\ { \bigl[1+\delta/2+o(1 ) \bigr ] \bigl(n^\zeta + n^{\zeta-2 \varepsilon } \bigr ) \bigr\},\ ] ] so lyapunov s condition is satisfied .
for given @xmath127 , @xmath53 , @xmath19 , @xmath643 and @xmath644 ( recall that @xmath645 ) , let @xmath646 be such that @xmath647 = \gamma.\ ] ] we claim that @xmath648 the left inequality of ( [ kq2 ] ) is known . to obtain the right inequality , first note that by ( [ qxt ] ) , @xmath649 since @xmath650 = -1-\log ( \frac{1-x}{1-t } ) \geq-1 $ ] for @xmath153 , therefore @xmath651,\ ] ] where @xmath652 .
check that @xmath653 , and that @xmath654 . for
@xmath655 , apply ( [ qxt ] ) , ( [ c2 ] ) and @xmath656 on @xmath657 to show that @xmath658 the right inequality of ( [ kq2 ] ) holds . for @xmath659 , apply ( [ qxt ] ) , ( [ c2 ] ) and @xmath660 on @xmath661 to show that @xmath662 the right inequality of ( [ kq2 ] ) again holds .
let @xmath663 , where @xmath16 are i.i.d .
@xmath362 . assume that there exists @xmath664 such that for the @xmath5th sequence , @xmath416 , there is an unknown interval @xmath665 $ ] with probability @xmath97 of having an elevated mean @xmath666 \\[-8pt ] \nonumber i_n & \sim & \operatorname{bernoulli}(\pi_n),\end{aligned}\ ] ] with @xmath98 and the @xmath667 s and @xmath16 s jointly independent .
this model is distinct from ( [ munt ] ) in that we do not now assume that the signals are aligned .
we claim that if @xmath275 for some @xmath189 and @xmath432 , @xmath668 and @xmath669 , then there is no test that can achieve at all such @xmath670 , @xmath671 note that though @xmath672 , as in the boundary for cases 1(b ) and 2 in ( [ bnbz ] ) , the growth of @xmath40 that is allowed in ( [ tlt ] ) is considerably smaller than that of ( [ c1 ] ) . as in the proof of theorem [ thmm1 ] , set @xmath673 ,
so that @xmath674 .
let @xmath675 and each @xmath259 , @xmath260 be of the form @xmath676 , with all @xmath262 $ ] disjoint from each other , and from @xmath677 $ ] . assume without loss of generality each @xmath670 is equally likely to take one of the @xmath269 possible values spaced at least @xmath38 apart , as given above .
then when @xmath678 , the detection of ( nonaligned ) signals satisfying ( [ tlt ] ) is at least as difficult as detecting a mixture of @xmath679 normal random variables @xmath680 , with a sparse fraction @xmath681 $ ] of them having mean @xmath27 .
therefore by the results in @xcite , the critical detectable @xmath27 is @xmath682 .
hence when @xmath432 , ( [ d2 ] ) can not be achieved . note that the assumption @xmath683 in ( [ tlt ] ) implies that @xmath684 , so @xmath685 is well - defined .
we would like to thank an associate editor and two referees for their comments that have led to a more realistic model setting and the adaptive optimal test statistics in this paper . | we describe , in the detection of multi - sample aligned sparse signals , the critical boundary separating detectable from nondetectable signals , and construct tests that achieve optimal detectability : penalized versions of the berk jones and the higher - criticism test statistics evaluated over pooled scans , and an average likelihood ratio over the critical boundary .
we show in our results an inter - play between the scale of the sequence length to signal length ratio , and the sparseness of the signals .
in particular the difficulty of the detection problem is not noticeably affected unless this ratio grows exponentially with the number of sequences .
we also recover the multiscale and sparse mixture testing problems as illustrative special cases .
./style / arxiv - general.cfg |
over the last two decades extensive optical / infrared studies of young optically visible pre - main sequence stars have helped to establish a paradigm for the later stages of the star formation process . to date the properties of their precursors
have been inferred largely from photometric measurements .
we have begun a program to examine the spectra of heavily embedded pre - main sequence objects by taking advantage of the fortuitous circumstance that some of these objects illuminate reflection nebulae . in lynds
1641 reflection nebulosity is associated with several of the heavily embedded iras sources that have no other optical manifestation .
the two objects which exhibit the highest surface brightness nebulosity at r are ic 430 ( haro 13a , v883 ori , 05358 - 0704 ) , 19.8 mag/ , and re 50 ( 05380 - 0728 ) , 20.8 mag/. ic 430 appeared in a list of nebulous objects in the orion region contained in the h@xmath0 survey of haro ( 1953 ) . while h@xmath0 emission was not detected from this object , the morphological similarity to hh objects led haro to suspect association with a star formation event . a faint star , below haro s plate limit ,
is seen at the tip of the nebula and has since been designated v883 ori .
this star was interpreted as an embedded middle b main sequence star by allen et al .
( 1975 ) on the basis of an almost featureless image tube spectrogram in the h@xmath0 region , which showed neither h@xmath0 emission nor absorption , and on its absolute k magnitude .
re 50 ( reipurth 1985a ) was first discussed by reipurth and bally ( 1986 ) as the source of a powerful molecular outflow whose associated reflection nebulosity has only recently appeared in the blue lobe of the outflow .
polarization studies ( scarrott & wolstencroft 1988 ) and near infrared imaging ( casali 1991 ) have demonstrated that the nebula is illuminated by the iras source .
on the night of 1993 january 26 ut we obtained spectra of re 50 and ic 430 using the rc spectrograph on the 4 meter telescope .
the slit , of length 5 and width 1.3 , was aligned along the highest surface brightness part of the reflection nebulosity , at a position angle of 90 for ic 430 , 135 for re 50 and 45 for l1551 irs 5 .
the grating had 316 lines / mm and was centered at 7500 ; the detector was a tektronics 2048x2048 ccd .
this combination allowed useful observations to be made from 5500 to 9500 with a spectral resolution of 7 .
three 1800 sec observations were made of re 50 and of l1551 irs5 ; two 1800 sec observations were made of ic 430 .
on the night of 1993 january 27 ut we obtained a spectrum of fu orionis with the same instrumental setup .
the spectra were reduced using the twodspec package within iraf .
the two dimensional spectra in the observational coordinate system were transformed to a two dimensional array in linear wavelength and spatial coordinates by making use of the multiplicity of night sky emission lines crossing the spectra and traces of stellar spectra placed at different locations on the slit .
this procedure allowed improved sky subtraction to be achieved in regions where the nebular surface brightness was low and long slit lengths were coadded .
in figure 1 we show the h@xmath0 profiles for re 50 , ic 430 , l1551 irs5 , and fu ori . re 50 and ic 430 show p cygni h@xmath0 profiles with the blue wings evidencing extreme outflow velocities , extending to 1000 km / s for re 50 .
the profile morphology is very similar to that of l1551 irs 5 which was previously classified as an fu ori object by mundt et al .
( 1985 ) on the basis of the h@xmath0 profile and by carr , harvey and lester ( 1987 ) on the basis of the infrared co bands .
the spectra in both the red region ( 5500 - 7500 ) and the near - ir region ( 7700 - 9200 ) compare well with that of fu ori and a g2 ia supergiant spectrum taken previously with the kpno echelle and smoothed to our present resolution .
the spectra do not compare well with late g main sequence spectra .
this is shown in figure 2 where the normalized spectra in the 5600 - 6800 region are shown .
also shown is the region of the infrared ca triplet , which behaves differently for each object . while the ca lines appear in net absorption in the spectrum of fu ori , welty et al .
( 1992 ) have shown , by subtracting disk photosphere model spectra from the observations of fu ori and v1057 cyg that these stars show residual emission in the ca triplet as well . in z cma , another probable fu ori object ( hartmann et al .
1989 ) , the ca triplet is strongly in emission .
the spectrum of ic 430 shows an emission component to the 8662 triplet line while the 8498 and 8542lines show no evidence for emission .
the spectrum of irs 5 shows a similar phenomenon where the 8662 line is seen in emission and the two shorter wavelength lines are not seen , although the higher noise level from the oh emission in the early evening night sky makes the result of low significance for the 8498 line .
a discussion of the ca infrared triplet anomaly in pre - main sequence stars can be found in hamann & persson ( 1992 ) .
it is difficult to place these objects within that context since the emission components are relatively weak ; models would have to be matched to these spectra and subtracted before a true measure of the emission in each line could be given , something best done with higher resolution spectra .
based primarily on the h@xmath0 profile and apparent low surface gravity and secondarily on the ca infrared triplet , we conclude that re 50 and ic 430 show many of the spectral characteristics similar to those of fu ori objects .
the characteristics that are used to define an fu ori object have evolved with time .
the first identifications were centered on the observation of a sudden large rise in luminosity over a short period of time .
this allowed 3 objects , fu ori , v1057 cyg , and v1735 cyg ( elias 1 - 12 ) to be placed in this group .
v1515 cyg is also included although it has exhibited a slow rise lasting many decades .
it has been recognized however that this phenomenon is characteristic of very early stages in a star s evolution , and is probably associated with multiply occurring sudden mass accretion events ( herbig 1977 ) .
recent theoretical work ( hartmann , kenyon & hartigan 1993 and references therein ) has modeled the fu ori phenomenon as a sudden increase in the mass accretion rate through a circumstellar accretion disk and has made predictions of spectral features , particularly the 2.3@xmath1 m co absorption bands , and line profile doubling to be expected for the disk model . because it is not possible to obtain optical spectra of the resolution necessary to search for line doubling , and we have not yet obtained near infrared spectra , are there other consistency tests that can be applied ?
1)in images of fu ori objects , small arctuate nebulosities are apparent ( goodrich 1987 ) .
the reflection nebulosity for both re 50 and ic 430 is typical of the form found for fu ori objects .
2 ) while the photometric measurements of v883 ori are few ( 2 ) , they are separated by 11 years ( allen et al .
1975 , nakajima et al . 1986 ) , and offer the possibility of detecting a long term decline in the brightness of v883 ori . in this time period , the object declined by 0.58 mag at j , 0.57 mag at h , 0.14 mag at k and 0.07 mag at l. further photometric monitoring will allow us to confirm the decline of this object .
3 ) the bolometric luminosities of the known fu ori objects range from @xmath240 @xmath3(l1551 irs5 ) to @xmath23000 @xmath3 ( z cma ) , ( rodriguez et al . 1990 ; hartmann et al .
1989 ) , although @xmath4 of the luminosity of z cma may be attributable to its infrared companion .
the bolometric luminosity of re 50 is @xmath2 300 @xmath3 while that for ic 430 is @xmath2 400 @xmath3 . while these are the two most luminous iras sources in l1641 , they were separated from the rest of the high luminosity embedded sources by strom et al .
( 1989 ) because they did not possess steep red spectral energy distributions .
instead their spectral energy distributions were relatively flat .
these spectral energy distributions are more characteristic of the fu ori objects .
4 ) the previously classified fu ori objects ( 11 objects , kenyon et al .
1993 ) show optically thin mm continuum emission which measures the mass of cold dust in the immediate environment of the object ( weintraub , sandell , & duncan 1989 , 1991 ; reipurth et al .
1993 ) and predominantly located in a structure with small covering angle , therefore allowing the objects to be seen . both re 50 and ic 430 have been measured at 1300@xmath1 m by reipurth et al .
the deduced masses for the circumstellar gas and dust immediately surrounding these objects are 3.7 @xmath5 for ic 430 and 1.8 @xmath5 for re 50 , well above the masses deduced for most of the optically visible fu ori objects but on the low end of the mass distribution for those objects known to be driving outflows delineated by hh objects .
5 ) several fu ori objects show cm radio emission with a spectral index characteristic of ionized outflows ( rodriguez , hartmann & chavira 1990 , rodriguez & hartmann 1992 ) both ic 430 and re 50 were observed in the vla snapshot survey of morgan , snell and strom ( 1990 ) at 6 cm .
re 50 was clearly detected and the spectral index demonstrated the emission to be thermal .
ic 430 was not detected , but the 3@xmath6 noise level is consistent with the range of flux levels measured for fu ori objects .
6 ) several of these objects are also apparently exciting sources of hh objects ( l1551 irs5 , hh28,29 ( stocke et al . 1988 ) ; v346 nor , hh57 ( reipurth 1985b ) ) , and long narrow jets ( z cma , poetzel et al . 1989 ) indicative of the energtic outflows diagnosed by the h@xmath0 profiles . both of these objects are also associated with hh objects , hh 65 ( reipurth 1989 ) with re 50 and a small emission knot with ic 430 ( strom et al . 1986 ) .
7 ) approximately half of the fu ori objects are known to drive molecular outflows ( see the summary table in hartmann , kenyon and hartigan 1993 ) .
the outflow associated with re 50 has been well documented ( bally & reipurth 1986 , fukui et al .
1986 ) and its interaction with the surrounding cloud material ( casali 1991 ; scarrott & wolstencroft 1988 ) has been described . however , a molecular outflow has not been found to be associated with ic 430 although a search for such evidence has been conducted ( morgan , schloerb , snell & bally 1991 ) . neither is there a cs core associated with this object ( tatematsu et al .
1993 ) although other outflow sources in the cloud are embedded in such cores .
however , there is other evidence that an energetic outflow however variable the rate , may have been present in the past .
the @xmath7co map of the l1641 molecular cloud ( bally et al .
1987 ) shows an elliptical depression in the molecular gas distribution , centered on this object .
the semi - major axis of this cavity is @xmath2 0.9 parsecs .
the presence of this cavity is also clearly indicated in our 100@xmath1 m iras hires image of the region .
the piled up material at the edges of the cavity emits strongly at 100@xmath1 m .
the presence of this object in the index catalog is curious in itself since the surface brightness of the nebulosity is so low .
therefore we traced the origin of its listing to a paper by pickering ( 1890 ) .
this object is described as a nebulous band @xmath8 wide extending @xmath9 north preceeding from dm-7@xmath10 1142 .
it is possible now to see nebulosity of this description on r and i band ccd images , although we know that the bright star is not responsible for its illumination .
however the surface brightness is only @xmath220 mag
/ on our r band ccd image
. the photographic plates of that period were sensitive only in the blue . on our b band ccd image taken 1990 nov 9 ut , it is very difficult to see the nebulosity . the maximum surface brightness near v883 ori is 24 mag / on this image .
it is unlikely that nebulosity of this surface brightness was apparent on the plates taken in 1888 .
therefore it is probable that v883 ori was considerably brighter at that time , and that the epoch of most recent outburst for this object can be placed in or before 1888 , when the two discovery plates were taken .
the total length of this nebulosity , as measured from v883 ori , is @xmath21 pc , in close agreement with the size of the cleared region of the cloud as seen in @xmath7co and 100@xmath1 m images .
spectra of two heavily embedded iras sources in l1641 reveal them to have characteristics of fu ori objects .
the other properties of these objects ( morphology , bolometric luminosities and spectral energy distributions , hh objects , inferred disk masses , thermal radio emission , and molecular outflow ) strengthen the conclusion that they belong to this class . the infrared co bands should be observed to confirm these identifications .
the presence of two fu ori objects within degree of each other in this cloud emphasizes the active star formation occurring in this cloud at the present time .
the authors would like to thank k. michael merrill for obtaining the entry and references in the index catalog for us .
we would also like to thank suzan edwards for discussions and for leading a successful late night search of the new smith college science library for the as yet uncataloged annals of the harvard college observatory .
allen , d. a. , strom , k. m. , grasdalen , g. l. , strom , s. e. , & merrill , k. m. 1975 , , 173 , 47p bally , j. , langer , w. d. , stark , a. a. & wilson , r. w. 1987 , , 312 , l45 carr , j. s. , harvey , p. m. , & lester , d. f. 1987 , , 321 , l71 casali , m. m. 1991 , , 248 , 229 fukui , y. , sugitani , k. , takaba , h. , iwata , t. , mizuno , a. , ogawa , h. , & kawabata , k. 1986 , , 311 , l85 goodrich , r. 1987 , , 99 , 116 hamann , f. & persson , s. e. 1992 , , 82 , 247 haro , g. 1953 , , 117 , 73 hartmann , l. , kenyon , s. j. , & hartigan , p. 1993 , in protostars and planets iii , university of arizona press , tucson hartmann , l. , kenyon , s. j. , hewett , r. , edwards , s. , strom , k. m. , strom , s. e. , & stauffer , j. r. 1989 , , 338 , 1001 herbig , g. h. 1977 , , 217 , 693 kenyon , s. j. , hartmann , l. , gomez , m. , carr , j. s. , & tokunaga , a. 1993 , , ( in press ) morgan , j. a. , snell , r. l. & strom , k. m. 1990 , , 362 , 274 mundt , r. , stocke , j. , strom , s. e. , strom , k. m. and anderson , e. r. 1985 , , 297 , l41 nakajima , t. , nagata , t. , nishida , m. , sato , s. , & kawara , k. 1986 , , 221 , 483 pickering , e. c. 1890 , annals of the harvard college observatory , 18 , 113 poetzel , r. mundt , r. , & ray , t. p. 1989
, , 224 , l13 reipurth , b. 1985a , , 61,319 reipurth , b. 1985b , , 143 , 435 reipurth , b. 1989 , , 220 , 249 reipurth , b. & bally , j. 1986 , nature , 320 , 336 reipurth , b. , chini , r. , krgel , e. , kreysa , e. , & sievers , a. 1993 , , ( in press ) rodriguez , l. f. , hartmann , l. & chavira , e. 1990 , , 102 , 1413 rodriguez , l. f. & hartmann , l. 1992 , rev.mex.astron.astrof . , 24 , 135 scarrott , s. m. & wolstencroft , r. d. 1988 , , 231 , 1019 stocke , j. t. , hartigan , p. m. , strom , s. e. , strom , k. m. , anderson , e. r. , & kenyon , s. j. 1988 , , 68 , 229 strom , k. m. , strom , s. e. , wolff , s. c. , morgan , j. & wenz , m. 1986 , , 62 , 39 strom , k. m. , newton , g. , strom , s. e. , seaman , r. l. , carrasco , l. , cruz - gonzalez , i. , serrano , a. , & grasdalen , g. 1989 , , 71 , 183 tatematsu , k. et al .
1993 , , 404 , 643 weintraub , d. , sandell , g. & duncan , w. d. 1989 , , 340 , l69 weintraub , d. , sandell , g. & duncan , w. d. 1990 , , 382 , 270 welty , a. , strom , s. e. , edwards , s. , kenyon , s. j. , & hartmann , l. w. 1992 , , 397 , 260 | we have obtained spectra of the reflection nebulosity illuminated by two heavily embedded iras sources in the l1641 molecular cloud , re 50 and ic 430 ( v883 ori ) .
examination of these spectra in combination with their other properties indicates that these objects are fu orionis objects .
our spectrum of l1551 irs 5 confirms the fu ori classification made for this star by mundt et al .
( 1985 ) . |
in the recent years , observation of the cosmic microwave background ( cmb ) anisotropy has been greatly improved . after the discovery of the cmb anisotropy by cobe @xcite at the angular scale @xmath1 , there have been many efforts to improve the measurements . in particular , recent balloon and ground - based experiments observed the cmb anisotropy at smaller angular scale of @xmath2 @xcite .
then , the observation of the cmb anisotropy is expected to be greatly improved by the on - going and future satellite experiments , map @xcite and planck @xcite ; after these experiments , the cmb angular power spectrum @xmath3 will be determined at @xmath4 level up to the multipole @xmath5 . with these measurements , our understanding of the evolution of the universe is also being improved .
importantly , since the cmb power spectrum is sensitive to origins and evolutions of the cosmic density perturbations ( as well as to the cosmological parameters ) , we are now able to constrain scenarios of generating the cosmic density perturbations . among various scenarios , inflation
@xcite is probably the most popular and well - motivated one to provide the source of the cosmic density perturbations . in the simplest case ,
all the components in the universe ( other than the cosmological constant ) , like the photon , baryon , cold dark matter ( cdm ) , and so on , originate to decay products of the inflaton field @xmath6 which is the scalar field responsible for the inflation .
the inflaton field fluctuates during the inflation and it becomes the source of the cosmic density perturbations .
one of the most important consequences of such a scenario is that , since all the components in the universe are produced from the inflaton field , there is no entropy perturbation between any of two components and hence the density fluctuations become adiabatic .
importantly , assuming the standard evolution of the universe after the inflation and using a reasonable set of the cosmological parameters , it is now widely believed that the observed cmb power spectrum is consistent with the one predicted from the scale - invariant spectrum which can arise from some class of inflation models @xcite . from the particle - physics point of view
, however , the simplest scenario may not be the case and there exist possible sources of the cosmic density perturbations other than the inflaton field .
in particular , in various scenarios , light scalar fields are introduced which dominate the universe at early epochs .
( we denote such a scalar field as @xmath0 . ) for example , in the scenario of affleck - dine baryogenesis @xcite , condensation of the squark and slepton fields is converted to the baryon - number asymmetry of the universe .
if this is the case , there may exist an epoch when the universe is dominated by the affleck - dine field .
then , at the time of the decay of the affleck - dine field , the late - time entropy production occurs .
other candidate is flat directions in the theory space . in particular , in the superstring theory @xcite , there are various flat directions parameterized by scalar fields called moduli fields .
such moduli fields are expected to acquire masses from the effect of the supersymmetry breaking and hence their masses can be much lighter than the planck scale . if the initial amplitudes of the moduli fields are large , the universe is once dominated by the moduli field and is reheated at the time of the decay of the moduli fields .
( see refs .
@xcite for the scenario with cosmological moduli fields . ) in addition , in refs .
@xcite it is pointed out that the baryon - number asymmetry of the universe can be explained if the universe was once dominated by the condensation of the right - handed scalar neutrino . in this case , the universe also experiences the late - time entropy production .
in addition , axion - like scalar field is proposed as a seed of the cosmic density perturbations @xcite in the pre - big - bang @xcite and the ekpyrotic @xcite scenarios .
such scalar fields may acquire fluctuations of the amplitude in the early universe ( in particular , during the inflation ) .
then , the scalar field eventually decays and reheats the universe .
importantly , in this class of scenario , the cmb radiation we observe today originates to @xmath0 rather than to the inflaton . as a result
, one can expect that the fluctuation of the scalar - field amplitude affects the cosmic density perturbations @xcite .
in particular , in such a scenario , adiabatic and isocurvature perturbations may be generated with cross-correlation.for other mechanisms of generating correlated mixture of the adiabatic and isocurvature perturbations , see @xcite .
indeed , in ref .
@xcite , the cmb angular power spectrum is calculated in such a framework , and it was shown that effects of the correlated isocurvature perturbation may be large enough to be seen in the on - going and future experiments .
in addition , it is possible that the dominant part of the cosmic density perturbations observed today may originate to the primordial perturbation in the amplitude of @xmath0 rather than to the fluctuation of the inflaton amplitude .
( this kind of mechanism is sometimes called `` curvaton mechanism . '' ) in this paper , we consider effects of such scalar - field condensations to the cosmic density perturbations without relying on any particular scenarios . assuming that the universe was once dominated by the scalar - field condensation and that the reheating occurred at a later stage with generating large amount of entropy , we study the evolutions of the cosmic density perturbations and consider the cmb angular power spectrum . as we will see , the cmb power spectrum can be affected in various ways depending on scenarios .
interestingly , if the decay product of the @xmath0 field does not generate the baryon asymmetry or the cdm component , then the adiabatic and isocurvature perturbations can be both generated with cross - correlation . in this case , the cmb power spectrum may significantly change its behavior from the adiabatic result .
in particular , the simple `` curvaton '' scenario , in which the cosmic density perturbations are generated only from the primordial fluctuation of the amplitude of @xmath0 , is severely constrained by the current observations of the cmb angular power spectrum _
if the adiabatic and isocurvature perturbations are generated with cross - correlation_. the organization of the rest of this paper is as follows . in section [ sec : scenario ] , we introduce the scenario we consider and follow the thermal history of the universe .
evolutions of the cosmic density perturbations in such a scenario is discussed in section [ sec : evolution ] . in particular , it is discussed how the density perturbation in the radiation is affected by the primordial fluctuation of the amplitude of @xmath0 . then , in section [ sec : cl ] , we overview the behavior of the cmb anisotropy in our scenario .
detailed discussions on the cmb power spectrum for the cases with and without the isocurvature perturbations are given in sections [ sec : noentropy ] and [ sec : entropy ] , respectively .
section [ sec : conclusion ] is devoted for the conclusion .
let us first introduce the scenario we have in mind . in this paper , we consider the thermal history with late - time entropy production which is due to the decay of the scalar field @xmath0 . for our study , it is important to understand the behavior of the scalar field in the expanding universe .
for this purpose , we start with the lagrangian of the form @xmath7 where @xmath8 is the potential of the scalar field . in our discussion , we adopt the parabolic potential for the scalar field ; @xmath9 from this lagrangian , we can derive the energy density @xmath10 and the pressure @xmath11 of the scalar field condensation ; for the zero - mode , we obtain @xmath12 where the `` dot '' denotes the derivative with respect to time @xmath13 .
in addition , in the expanding universe , equation of motion of the scalar field is given by @xmath14 where @xmath15 is the expansion rate with @xmath16 being the scale factor .
the behavior of @xmath0 depends on the relative size of the expansion rate @xmath17 and the scalar mass @xmath18 . when the expansion rate is much larger than @xmath18 , the second term , which plays the role of the friction term , is significant and @xmath19 ( almost ) vanishes . in this case ,
change of the scalar amplitude can be neglected and @xmath0 takes a constant value .
we call this epoch as `` slow - roll '' epoch since @xmath19 is negligibly small . on the contrary ,
when @xmath20 , the scalar field oscillates around the minimum of the potential .
in particular , when @xmath21 , the oscillation of the scalar field becomes fast enough so that the relation @xmath22 holds , where @xmath23 denotes the average over the time scale much longer than @xmath24 ( but shorter than @xmath25 ) . with this relation , the energy density and the pressure of the scalar field
are given as @xmath26 as we can see , the pressure of the scalar field effectively vanishes at this epoch and the equation - of - state parameter for the scalar field becomes @xmath27 . in this case
, the energy density of the scalar field behaves as that of the non - relativistic component .
consequently , when @xmath28 ( with @xmath29 being the decay rate of @xmath0 ) , @xmath10 decreases as @xmath30 .
now we discuss the thermal history of the universe with the scalar field @xmath0 .
in particular , we consider the case where the initial amplitude of the scalar field @xmath31 is non - vanishing so that @xmath0 may become cosmologically significant .
with such an initial condition , let us follow the evolution of the universe starting with the inflationary period .
as in the conventional cases , we adopt the epoch of ( slow - roll ) inflation in the early stage . as well as solving the horizon and flatness problems , inflation provides the source of the cosmic density perturbations .
we assume that the mass of the scalar field @xmath0 is much smaller than the expansion rate of the universe during the inflation @xmath32 .
if this is the case , the energy density of the scalar field is much smaller than the total energy density of the universe as far as the amplitude of the scalar field is smaller than the planck scale .
then , the potential of the scalar field @xmath0 is effectively flat during inflation and there is no strong reason to have vanishing initial amplitude of @xmath0.if the minimum of the potential of @xmath0 is a symmetry - enhanced point , however , initial value of @xmath0 may vanish @xcite .
in addition , if @xmath33 , @xmath0 is slow - rolling during inflation and it keeps its initial amplitude .
thus , we assume a non - vanishing value of the initial amplitude @xmath31 . in the slow - roll inflation models , the inflaton field @xmath6 oscillates after the inflation and decays .
then , the universe is reheated .
( we call this epoch as the `` first reheating '' since , in the scenarios we consider , the universe is reheated again due to the decay of @xmath0 . )
the energy density of the inflaton field also decreases as @xmath30 in this oscillating period , and hence the expansion rate of the universe is being decreased .
then , at the time when the expansion rate of the universe becomes comparable to the decay rate of the inflaton @xmath34 , the inflaton decays and reheats the universe . equating the expansion rate and the decay rate of the inflaton field ,
the reheating temperature is estimated to be @xmath35 where @xmath36 is the effective number of the massless degrees of freedom , in this section , we neglect the temperature dependence of @xmath37 since it does not significantly change the following discussion .
@xmath38 is the reduced planck scale , and we adopt the approximation of the instantaneous reheating at the time of @xmath39 . in addition , here ( and in the following order - of - estimations in this section ) , we neglect numerical constants of @xmath40 .
evolution of the scalar field @xmath0 depends on the relative size of @xmath18 and @xmath34 .
if the mass @xmath18 is smaller than @xmath34 , @xmath0 starts its oscillation after the decay of the inflaton ; otherwise , @xmath0 starts to move while the inflaton field is still oscillating . in studying the thermal history with the scalar field @xmath0 , it is also important to specify the initial value of the scalar field @xmath31 .
when @xmath0 starts to move , @xmath41 and hence the total energy density of the universe is @xmath42 .
thus , if @xmath31 is smaller than @xmath43 , the scalar field is a sub - dominant component when it starts to move , and hence the evolution of the universe at this epoch is controlled by other components , oscillating inflaton field or radiation . if @xmath44 , on the contrary , energy density of @xmath0 becomes comparable to the total energy density . in this case , the scalar field @xmath0 never becomes minor component until it decays .
let us first consider the case where @xmath45 is high enough so that @xmath0 starts to oscillate after the decay of the inflaton .
this happens when @xmath46 . in this case , the scalar field starts to oscillate at the temperature @xmath47 after this epoch , the energy density of the scalar field @xmath10 decreases as @xmath30 while that of radiation is proportional to @xmath48 , and hence the energy density of the scalar field decreases slower than that of the radiation .
thus , when the amplitude of the scalar field is as large as @xmath49 , the energy density of the scalar field becomes comparable to that of radiation if the scalar field does not decay before this epoch .
expansion rate at this epoch is @xmath50 , and @xmath51 becomes larger than the decay rate of @xmath0 if @xmath52 where @xmath53 is the reheating temperature after the decay of @xmath0 and is estimated as @xmath54 if the condition ( [ phimin ] ) is satisfied , the scalar field dominates the universe before it decays .
if the reheating temperature after the inflation @xmath45 becomes low , the scalar field @xmath0 may start to oscillate before the inflaton field decays ; this is the case if @xmath55 . in this case ,
when the inflaton field decays , the amplitude of the scalar field is @xmath56 .
then , the energy densities of @xmath0 and radiation become comparable when @xmath57 , if @xmath0 does not decay before its amplitude becomes smaller than this value .
then , the condition to have a @xmath0-dominated epoch is @xmath58 if the initial amplitude of @xmath0 becomes larger than @xmath59 , on the other hand , the @xmath0 field slow - rolls even after the expansion rate of the universe becomes comparable to @xmath18 .
then , the universe enters into the de sitter phase again . in this case
, @xmath0 plays the role of the inflaton and the resultant cosmic density perturbation becomes the same as the standard adiabatic result ( if the scale of the secondary inflation is properly chosen).if the @xmath60-folding number due to the secondary inflation is small , however , this is not the case and the cosmic density fluctuation for large scale is generated by the first inflation .
we do not consider such a case in this paper .
thus , we assume that the initial amplitude of @xmath0 is smaller than @xmath43 .
in particular , we assume that @xmath31 is small enough so that the universe has a radiation - dominated epoch before @xmath0 dominates the universe .
( this is the case if @xmath31 is substantially smaller than @xmath43 . ) in fig .
[ fig : phimin ] , we summarize the above results ; on the @xmath53 vs. @xmath18 plane , we plot the contours of the minimum value of @xmath31 with which the universe experiences the @xmath0-dominated epoch . in the following discussion
, we assume that the initial amplitude of the scalar field is large enough so that there exists the @xmath0-dominated epoch . since the universe has to follow the standard big - bang scenario in particular for the big - bang nucleosynthesis , however , the universe has to become radiation - dominated after the @xmath0-dominated epoch .
this is realized by the reheating due to the decay of @xmath0 .
when @xmath61 , the @xmath0 field decays and its energy density is converted to that of radiation . the reheating temperature at the time of the @xmath0-decay depends on the decay rate @xmath29
importantly , the reheating temperature has to be higher than @xmath62 so that the success of the standard big - bang nucleosynthesis is not altered ; we assume that @xmath29 is large enough in order not to affect the standard nucleosynthesis scenario . in summary ,
the thermal history is as follows : * in the early stage , inflation occurs . for the pre - big - bang @xcite and the ekpyrotic @xcite scenarios
, this may not be the case . in these cases , we assume that some other mechanism generates the primordial fluctuation of @xmath0 . during inflation ,
the scalar field @xmath0 has a substantial initial amplitude so that it dominates the universe at the later stage .
* after the inflation , the universe is dominated by the oscillating inflaton field .
( we call this epoch as `` @xmath6d '' epoch . ) at some point , the inflaton decays and reheats the universe .
* when the expansion rate of the universe becomes comparable to @xmath18 , the @xmath0 field starts to oscillate .
this may happen during the @xmath6d or the radiation - dominated epoch .
we assume that the initial amplitude of @xmath0 is smaller than @xmath63 so that there is no inflation induced by the @xmath0 field .
in addition , if @xmath31 is substantially smaller than @xmath43 , the universe experiences a radiation - dominated epoch before @xmath0 dominates the universe .
( we call this epoch as `` rd1 '' epoch . ) * as the universe expands , @xmath0 eventually dominates the universe since @xmath64 while @xmath65 ( where @xmath66 is the energy density of radiation ) .
we call this epoch `` @xmath0d '' epoch . *
when @xmath67 , the @xmath0 field decays and reheats the universe . then the radiation - dominated universe is realized again .
( we call this epoch `` rd2 '' epoch . ) after this epoch , the evolution of the universe is the same as the standard big - bang case as far as the cosmic density perturbations are not concerned .
the scenario is schematically shown in fig .
[ fig : history ] .
( in the following , the subscripts `` @xmath6d , '' `` rd1 , '' `` @xmath0d , '' and `` rd2 '' will be used for variables in each epochs . )
in the previous section , we summarized the scenario and discussed the evolutions of the background variables ( i.e. , the zero - modes ) .
our purpose is , however , to understand the effects of the fluctuation of the amplitude of the scalar field @xmath0 on the cosmic density perturbations .
thus , in this section , we discuss the evolutions of the perturbations in detail .
importantly , properties of the density perturbations are determined at the epoch much earlier than the electron decoupling ( i.e. , @xmath68 , with @xmath69 being the electron mass ) and the neutrino freeze - out .
thus , in this section , we study the behaviors of the perturbations in such an epoch . in this paper
, we give the formalism in the conformal newtonian gauge .
( we follow the notation and convention of @xcite . ) in discussing the evolutions of the perturbations , it is convenient to define the variable @xmath70 as @xmath71 where the subscript @xmath72 denotes the individual components ( like radiation , cdm , baryon , and so on ) and @xmath73 is the density perturbation of the component @xmath72 .
in addition , in the conformal newtonian gauge , the perturbed line element is given by @xmath74 , \label{ds}\end{aligned}\ ] ] where @xmath75 is the conformal time , @xmath76 is the scale factor at the present time , and @xmath77 and @xmath78 are the metric perturbations . substituting eq .
( [ ds ] ) into the einstein equation , we obtain the generalized poisson equation for @xmath78 : @xmath79 , \label{poisson}\end{aligned}\ ] ] and other equation relating @xmath78 , @xmath77 , and the anisotropic stress perturbation of the total matter @xmath80 : @xmath81 here , `` tot '' denotes the total matter and the variable @xmath82 denotes the velocity perturbation of the component @xmath72 .
in addition , @xmath83 is the comoving momentum , @xmath84 the equation - of - state parameter for the total matter , and @xmath85 as indicated in the above equations , evolutions of the metric perturbations are related to those of the density , velocity , and other perturbations of various components . here , we consider the universe with temperature higher than @xmath86 . in such a situation ,
various charged particles become relativistic and are thermally produced .
then , all the relativistic components ( like photon , electron , quarks , neutrinos , and so on ) can be regarded as a single fluid with very short mean free path , much shorter than the horizon scale .
we call such a component as `` radiation '' and use the subscript `` @xmath87 '' for it .
when the mean free path of the radiation is very short , perturbation of the radiation becomes locally isotropic and the anisotropic stress perturbation vanishes : @xmath88 then , the equations for the density and velocity perturbations of the radiation are given by @xmath89 where the `` prime '' denotes the derivative with respect to the conformal time @xmath75 .
in addition , if a very weakly interacting non - relativistic component exists , its perturbations obey the following equations : @xmath90 where the subscript `` @xmath91 '' is for non - relativistic matters .
notice that , when a scalar field is oscillating , the equation - of - state parameter of the scalar condensation vanishes and hence the density and velocity perturbations of the scalar field also obey eqs .
( [ eq : cdm ] ) and ( [ eq : cdm_velocity ] ) . in our scenario
, there are several possible sources of the density perturbations .
the first one is , as in the conventional case , fluctuation of the inflaton amplitude . in our case , however , amplitude of @xmath0 may also fluctuate and such a fluctuation becomes second source of the density fluctuations .
in addition , if there exists independent scalar field ( called @xmath92 ) from which the baryon or the cdm is generated , we also have to take account of its effects .
( in the following , we consider the cases with and without this scalar field @xmath92 . ) in the following , primordial fluctuations of these scalar fields are assumed to be uncorrelated . in the framework of the linear perturbation theory , evolutions of the perturbations from the primordial fluctuations of these scalar fields are independent and hence we can study their evolutions separately .
evolutions of the perturbations from the inflaton fluctuation are intensively studied ; such a fluctuation provides adiabatic density fluctuations .
in addition , density perturbation induced from the fluctuation of @xmath92 is also well known as far as @xmath92 is a minor component in the rd2 epoch ; primordial fluctuation of @xmath92 induces ( purely ) isocurvature fluctuation to the baryonic or the cdm component .
thus , in the following , we study the evolutions of the density perturbations induced by the fluctuation of the amplitude of @xmath0 . for this purpose
, we decompose the scalar field as @xmath93 where @xmath94 denotes the fluctuation of @xmath0 .
( we use the same notation for the total amplitude and for the zero - mode because we believe there is no confusion . )
in particular we study evolutions in very early universe in order to understand how the fluctuations behave when the universe enters into the rd2 epoch . in realistic scenarios ,
the scale we are interested in ( i.e. , the scale corresponding to the multipole @xmath95 ) is far outside of the horizon at this epoch and hence we can expand the solutions to the equations as functions of @xmath96 .
in addition , combining the relation @xmath97 with eq .
( [ psi+phi=0 ] ) , we obtain @xmath98 . in the following , we use this relation to eliminate @xmath78 . in order to discuss the evolutions of the fluctuations ,
it is important to clarify the equation - of - state parameters of the individual components in the universe .
if all the components behave as the relativistic or the non - relativistic matter , evolutions of the perturbations are described by eqs .
( [ dr ] ) @xmath99 ( [ eq : cdm_velocity ] ) . in this case
, it is convenient to distinguish the photon ( or any other components ) from the decay product of @xmath0 from that from the inflaton field , which we call @xmath100 and @xmath101 , respectively.in fact , these photons are mixed each other and they can not be defined separately . in other words , their velocity perturbations should be the same since they form a single fluid .
even so , the following arguments are unchanged as far as we consider the leading terms in the density perturbations since the velocity perturbation is at most @xmath102 . in the following discussion , @xmath100 and @xmath101
should be understood as representatives of the components which are and are not generated from the decay product of @xmath0 , respectively . in order to consider @xmath103 in the rd2 epoch , we can neglect @xmath101 since the cmb radiation at this epoch is dominantly from the @xmath0 field . then , we find that , in the rd2 epoch , @xmath77 and @xmath103 become constant while @xmath104 up to higher order corrections . indeed , combining eq .
( [ poisson ] ) with eqs .
( [ dr ] ) and ( [ vr ] ) , and using @xmath105 and @xmath106 , we obtain @xmath107 and @xmath108 where @xmath109 is the metric perturbation induced by the primordial fluctuation of the amplitude of @xmath0 .
( in the following , the superscript `` @xmath110 '' is for perturbations generated from the primordial fluctuation of @xmath0 . ) as we mentioned , @xmath109 is constant up to a correction of @xmath111 .
behavior of @xmath112 is also easily understood . in discussing the effects of the primordial fluctuation of @xmath0
, we neglect the initial fluctuation of the inflaton field and hence @xmath113 in the deep rd1 epoch .
in addition , from eqs .
( [ dr ] ) and ( [ vr ] ) , @xmath114 becomes higher order in @xmath96 than @xmath112 and @xmath77 .
thus , we obtain @xmath115 the above relation holds in the rd1 , @xmath0d , and rd2 epochs up to corrections of @xmath111 . for the following discussion , it is convenient to define the entropy perturbation between @xmath100 and @xmath101 : @xmath116 for superhorizon modes , this entropy perturbation is a constant of time @xcite .
in addition , for example for the case where the @xmath0 field starts to oscillate during the @xmath6d epoch , @xmath117_{\rm rd2 } = \left [ \delta_{\phi } - \frac{3}{4 } \delta_{\gamma_\chi } \right]_{\rm \phi d , rd1 } = \left [ \delta_{\phi } - \delta_{\chi } \right]_{\rm \chi d},\end{aligned}\ ] ] where the last equality holds for the epoch when @xmath0 and the inflaton field are both oscillating . in the case with the @xmath92 field which is responsible for the baryon or the cdm ,
more careful considerations are needed . adopting the parabolic potential for @xmath92 ,
the fluctuation of the @xmath92 field obeys @xmath123 \delta\psi = -2 m_\psi^2 \psi \psi + \dot{\psi } ( \dot{\psi } - 3 \dot{\phi } ) .
\label{eom_psi}\end{aligned}\ ] ] solving this equation , we see that the fluctuation of @xmath92 at the time of @xmath124 is of the order of @xmath125_{h\sim m_\psi}$ ] .
if the @xmath92 field starts to oscillate in the rd1 epoch , energy density of @xmath92 behaves as that of non - relativistic components even when the metric perturbation @xmath77 is negligibly small . in this case , @xmath125_{h\sim m_\psi}$ ] is small enough to be neglected and the former discussion applies .
if @xmath92 starts to oscillate in the @xmath0d epoch , on the contrary , the situation changes . in this case
, the @xmath92 field is slowly rolling when the @xmath0d epoch starts and @xmath126 at the time of the @xmath0-domination is negligibly small .
then , solving eq .
( [ eom_psi ] ) for the superhorizon mode , we obtain @xmath127 where we used the fact that @xmath128 and @xmath129 are small enough to be neglected in the @xmath0d epoch . with eq .
( [ delpsi ] ) , we can calculate the density fluctuation of @xmath92 as @xmath130_{\rm \phi d}=-3h\dot{\psi}^2\psi^{(\delta\phi)}_{\rm \phi d } t$ ] .
importantly , the relation ( [ delpsi ] ) holds even after the @xmath92 field starts to oscillate , as far as @xmath17 is substantially larger than the decay rate of @xmath92 .
so , we can calculate the entropy between @xmath92 and @xmath0 in the @xmath0d epoch after the @xmath92 field starts to oscillate ; using the relation @xmath131_{\rm\phi d } = -2 \psi^{(\delta\phi)}_{\rm \phi d}$ ] and @xmath132 , we see that the entropy between @xmath92 and @xmath0 vanishes .
thus , if the @xmath92 field is slow - rolling when the @xmath0d epoch starts , density perturbation of the component generated from @xmath92 becomes adiabatic .
if the @xmath92 field starts to oscillate just at the time when the @xmath0d epoch starts , the correlated mixture of the adiabatic and isocurvature perturbations is generated but the ratio of the metric perturbation to the entropy perturbation may deviate from that given in eqs .
( [ sb ] ) @xmath99 ( [ sbc ] ) . before closing this section , let us estimate the size of @xmath118 . as we mentioned before , primordial fluctuation of the @xmath0 field is probably generated during the inflation . in this case , calculating the two - point function of @xmath0 in the de sitter background , we obtain @xmath133 where the fourier amplitude for the case of @xmath33 is given by @xmath134_{k = ah_{\rm inf}}.\end{aligned}\ ] ] ( the subscript `` inf '' means that the correlation function is evaluated during the inflation . )
once the wavelength becomes longer than the horizon scale , the scalar - field fluctuation obeys the following equation : @xmath135 \delta\phi = -2 m_\phi^2 \phi \psi + \dot{\phi } ( \dot{\psi } - 3 \dot{\phi } ) .
\label{dphi_eom}\end{aligned}\ ] ] first , let us consider the case where @xmath136 . in this case ,
when the @xmath0 field starts to move , the energy density of @xmath0 is much smaller than the total energy density of the universe . in studying the density perturbations generated by the primordial fluctuations of @xmath0 , the metric perturbations are negligibly small until the energy density of @xmath0 becomes comparable to the total energy density of the universe . in this case
, we can neglect the terms containing the metric perturbations in eq .
( [ dphi_eom ] ) .
furthermore , the @xmath137-term is also irrelevant for the superhorizon perturbations .
then , we can see that , with the above approximations , @xmath94 obeys the same equation as the zero - mode . neglecting the decaying mode
, we obtain the relation @xmath138 where @xmath139 is the initial value of the fluctuation of @xmath0 , which is @xmath140 .
in addition , @xmath141 ( and hence @xmath142 ) in the early universe , and hence the initial value of @xmath118 is estimated as @xmath143 thus , the entropy @xmath118 increases as the initial amplitude of @xmath0 decreases .
if the initial amplitude of @xmath0 becomes close to @xmath43 , the energy density of @xmath0 becomes comparable to the total energy density of the universe .
then , the metric perturbation @xmath77 becomes comparable to the density perturbation of the @xmath0 field @xmath144 . in this case ,
sizable @xmath77 generates the density perturbations of other components and hence the relation ( [ s_pc ] ) may not hold . even in this case , however , eq .
( [ s_pc ] ) provides a good order - of - estimation of the primordial entropy perturbation between @xmath0 and @xmath6 at the time when the @xmath0 field starts to oscillate ; as far as there is no inflation caused by @xmath0 , @xmath145 when @xmath44 .
in this section , we study the cmb anisotropy generated from fluctuations of various scalar fields .
the cmb anisotropy is characterized by the angular power spectrum @xmath3 which is defined as @xmath146 with @xmath147 being the temperature fluctuation of the cmb radiation pointing to the direction @xmath148 at the position @xmath149 and @xmath150 is the legendre polynomial . the angular power spectrum @xmath3 depends on the properties of the primordial density fluctuations and hence it provides probes for studying how the density perturbations are generated in the early universe .
in the scenario we consider , there are three possible sources of the density perturbations ; the primordial fluctuations of the scalar fields @xmath6 , @xmath0 , and @xmath92 .
since there is no correlation among these fields , the cmb anisotropies from these fluctuations are uncorrelated and the resultant cmb power spectrum is given in the form @xmath151 where @xmath152 , @xmath153 and @xmath154 are contributions from the primordial fluctuations of the inflaton field @xmath6 , the @xmath0 field and the @xmath92 field , respectively.if all the components in the universe are generated from the decay product of @xmath0 , the @xmath92 field is irrelevant and @xmath154 does not exist .
the inflaton contribution @xmath155 is the same as the adiabatic result while @xmath154 becomes the purely isocurvature one .
thus , in the following , we study the properties of @xmath156
. effects of the primordial fluctuation of the @xmath0 field depend on the scenario as we discussed .
if all the components in the universe originate to @xmath0 , then the density perturbations become adiabatic . in this case , it is difficult to experimentally distinguish @xmath156 from the adiabatic cmb power spectrum generated from the inflaton fluctuations . even in this case , however , there are important implications in studying scenarios of generating the cosmic density fluctuations , like inflation , as will be discussed in section [ sec : noentropy ] . on the contrary ,
if the scalar field @xmath92 exists which is responsible for the baryon or the cdm , @xmath156 may become different from the angular power spectrum from the purely adiabatic density perturbation since the adiabatic and isocurvature perturbations can be generated with cross - correlation .
although we have considered the entropy perturbations in the baryonic and the cdm sectors separately in eqs .
( [ sb ] ) @xmath99 ( [ sbc ] ) , we found that the resultant @xmath156 is primarily determined by the entropy between the non - relativistic matter and radiation ; if we parameterize the density perturbation of the non - relativistic component as @xmath157_{\rm rd2 } \equiv \left [ ( \omega_b/\omega_m ) \delta_b^{(\delta\phi ) } + ( \omega_c/\omega_m ) \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } = \frac{3}{4 } \left [ \delta_\gamma^{(\delta\phi ) } \right]_{\rm rd2 } + \kappa_m \psi_{\rm rd2}^{(\delta\phi)},\end{aligned}\ ] ] where @xmath158 , @xmath159 and @xmath160 are the ( present ) density parameters for the baryon , the cdm , and the non - relativistic component ( and hence @xmath161 ) , the parameter @xmath162 determines the shape of @xmath156 . with the relations ( [ sb ] ) , ( [ sc ] ) , and ( [ sbc ] ) , @xmath162 becomes @xmath163 , @xmath164 , and @xmath165 , respectively . on the contrary , if the @xmath92 field is slow - rolling when the @xmath0d epoch starts , @xmath162 vanishes .
in addition , if @xmath92 accidentally starts to oscillate just at the beginning of the @xmath0d epoch , @xmath162 may take any value between 0 and @xmath165 .
this may happen , however , when the expansion rate at the beginning of the @xmath0d epoch accidentally becomes comparable to @xmath166 . in fig .
[ fig : cl s ] , we plot the angular power spectrum with the correlated mixture of the adiabatic and isocurvature perturbations in the baryonic and/or cdm sector , i.e. , the cases with the relations given in eqs .
( [ sb ] ) @xmath99 ( [ sbc ] ) . here , we consider the flat universe with @xmath167 , @xmath168 , and @xmath169 @xcite , where @xmath170 is the hubble constant in units of 100 km / sec / mpc . for comparison
, we also plot the angular power spectrum for the purely adiabatic and isocurvature cases ( i.e. , @xmath155 and @xmath154 ) .
as one can see , the cmb angular power spectrum strongly depends on properties of the primordial density perturbations .
if there exists correlated entropy between the baryon and other components with the relation ( [ sb ] ) , negative interference between the adiabatic and isocurvature ; perturbations suppresses @xmath3 at lower multipole while the effect of the isocurvature perturbation becomes too small to affect the structure at high multipole . as a result
, the angular power spectrum is enhanced at the high multipole rather than at the low multipole .
if the effect of the entropy perturbation becomes more efficient , then @xmath154 at high multipole is suppressed relative to that at low multipole like in the purely isocurvature case .
this happens when the entropy perturbation is in the cdm component with the condition given in eq .
( [ sc ] ) .
in addition , with the relation given in eq .
( [ sbc ] ) , the correlated entropy becomes more effective than the case where only the cdm sector has the correlated entropy .
then , the acoustic peaks are more suppressed relative to the sachs - wolfe ( sw ) tail . as discussed in the previous section , in the general case
, the @xmath162 parameter may vary between 0 and @xmath165 , and hence the cmb anisotropy generated from the primordial fluctuation of @xmath0 changes its behavior as @xmath162 varies .
in particular , if @xmath139 is the only source of the cosmic density perturbations , too large @xmath162 results in angular power spectrum which is inconsistent with observations . in this case , we can derive a constraint on the @xmath162 parameter . in order to derive the constraint
, we calculate the goodness - of - fit parameter @xmath171 , where @xmath172 is the likelihood function , as a function of @xmath162 . in our calculation , the offset log - normal approximation is used @xcite .
we use a data set consisting of 65 data points ; 24 from cobe / dmr @xcite in the range @xmath173 , 19 from boomerang @xcite in the range @xmath174 , 13 from maxima @xcite in the range @xmath175 , and 9 from dasi @xcite in the range @xmath176 . from the figure
, we can read off the constraint on @xmath162 .
requiring @xmath177 , which corresponds to 95 % c.l .
allowed region for the @xmath178 statistics with 64 degrees of freedom , we obtain the constraint @xmath179 for @xmath168 , and @xmath180 for @xmath181 .
interestingly , the @xmath178 variable is relatively suppressed when @xmath182 .
this is because the heights of the acoustic peaks relative to the sw tail once increase then decrease as we increase the @xmath162 parameter . in particular ,
when @xmath182 , the heights of the acoustic peaks relative to the sw tail become similar to those from the adiabatic density perturbations which well agrees with the observations .
of course , if the effects of @xmath155 or @xmath154 become important , the above constraint on @xmath162 does not apply .
if all the component of the universe ( i.e. , the photon , baryon , cdm , neutrino , and so on ) are generated from the decay product of @xmath0 , there is no entropy between any of two components . in this case , the resultant angular power spectrum is given in the form @xmath183 importantly , the two contributions @xmath155 and @xmath156 are both from adiabatic perturbations .
thus , for the case where scale dependence of the sources of @xmath155 and @xmath156 , i.e. , @xmath184 and @xmath118 , are the same , @xmath185 . in this case , it is impossible to test the scenario with @xmath0 using experimental data . in general , however , the scale dependences of @xmath184 and @xmath118 are different . in the slow - roll inflation scenario , @xmath184 is generated from the fluctuation of the inflaton amplitude and is given by @xcite @xmath186_{k = ah_{\rm inf } } , \label{psi(dchi)_rd2}\end{aligned}\ ] ] where @xmath187 with @xmath188 being the inflaton potential .
on the contrary , @xmath118 is related to the fluctuation of the amplitude of @xmath0 , as seen in eq .
( [ s_pc ] ) , and hence @xmath189_{k = ah_{\rm inf}}. \label{s(dphi)}\end{aligned}\ ] ] in many models of slow - roll inflation , the expansion rate @xmath190 is almost constant during the inflation . on the contrary ,
the slope of the inflation potential @xmath191 may significantly vary . as a result ,
@xmath118 becomes ( almost ) scale independent while @xmath184 may have sizable scale dependence .
since the currently measured cmb power spectrum suggests ( almost ) scale invariant primordial density perturbation in the conventional scenario , inflation models are excluded if @xmath184 has too strong scale dependence @xcite .
if the @xmath0 field exists , however , the situation may change . since the variable @xmath118 is expected to be ( almost ) scale invariant
, we can relax the constraint on the inflation models if @xmath156 becomes significantly large , which happens when @xmath192 . as shown in eq .
( [ s_pc ] ) , @xmath118 is inversely proportional to @xmath31 .
thus , if the initial amplitude of @xmath0 is suppressed , this may happen .
in particular , if @xmath193 , the present cosmic density fluctuations are totally from the primordial fluctuation of @xmath0 @xcite .
( such a scenario is sometimes called `` curvaton scenario.''for recent discussions on the curvaton scenario , see @xcite . ) in addition , this mechanism of converting the isocurvature density perturbation in the scalar field to the adiabatic density perturbation may have some significance in the pre - big - bang @xcite and ekpyrotic @xcite scenario . in refs .
@xcite , it was discussed that the present cosmological density perturbations may be generated from the primordial fluctuation of some axion - like fields in the pre - big - bang and ekpyrotic scenarios .
in this section , we discuss the cmb angular power spectrum in cases with entropy perturbations . in particular , we study the effects of the correlated entropy perturbation generated in the cases where the baryonic or the cdm component is generated from @xmath92 , not from @xmath0 .
as we mentioned , if the @xmath92 field starts to oscillate before the @xmath0d epoch starts , the relation ( [ sb ] ) , ( [ sc ] ) or ( [ sbc ] ) holds . on the contrary , if @xmath92 is slow - rolling at the time of the @xmath0-domination , density perturbations become adiabatic .
thus , in this section , we consider the most typical relations ( [ sb ] ) @xmath99 ( [ sbc ] ) when the adiabatic and isocurvature perturbations exist with cross - correlation . in the general case ,
as we mentioned , the relation between the adiabatic and isocurvature perturbations may deviate from eqs .
( [ sb ] ) @xmath99 ( [ sbc ] ) .
this may happen , however , only if the expansion rate at the beginning of the @xmath0d epoch becomes accidentally close to @xmath166 , and we will not consider such a case in this paper .
( see @xcite for such a case . )
as discussed in the previous section , the cmb power spectrum is given in the form @xmath194 here , @xmath155 is from the inflaton fluctuation which is usually parameterized by the metric perturbation generated by the inflaton fluctuation ( in the rd2 epoch ) , @xmath195 . in addition , @xmath156 and @xmath154 are parameterized by the entropy perturbations in the baryon or the cdm . to parameterize these contributions
, we define the ratios @xmath196 and @xmath197 where @xmath198 ( @xmath199 ) is the entropy between the baryon and the photon ( between the cdm and the photon ) generated from the primordial fluctuation of @xmath0 , and @xmath200 and @xmath201 are those from @xmath126 .
( we adopt eqs .
( [ sb ] ) and ( [ sc ] ) , and hence @xmath202 and @xmath203 are equal to @xmath204 if they are non - vanishing . ) the shape of the cmb angular power spectrum depends on the values of these parameters .
the expressions for @xmath205 and @xmath118 are given in eqs .
( [ psi(dchi)_rd2 ] ) and ( [ s(dphi ) ] ) , respectively , and hence , if @xmath92 starts to oscillate before the @xmath0d epoch starts , the @xmath206-parameters are given as @xmath207_{k = ah_{\rm inf}}.\end{aligned}\ ] ] hence @xmath208 is model- and scenario - dependent ; it depends on the scale of inflation , shape of the inflaton potential , and initial amplitude of @xmath0 .
for example in the chaotic inflation model with the parabolic potential @xmath209 , the above expression becomes @xmath210_{\rm chaotic } = \left [ \frac{9m_*^2}{\phi_{\rm init}\chi } \right]_{k = ah_{\rm inf}}.\end{aligned}\ ] ] using the fact that the inflaton amplitude at the time of the horizon crossing of the cobe scale is @xmath211 in the chaotic inflation model , @xmath212_{\rm chaotic}\simeq 0.6m_*/\phi_{\rm init}$ ] .
of course , the values of @xmath213 and @xmath214 depend on the model of inflation , and they vary if we consider different class of inflation models .
the @xmath215-parameters are also estimated in the similar way , and the result is given from the expression for the @xmath206-parameters by replacing @xmath216 .
although our main subject is to investigate effects of the fluctuation of the @xmath0 field , it is instructive to study effects of the uncorrelated isocurvature perturbations .
thus , let us first consider the case with uncorrelated isocurvature perturbations ( i.e. , case with @xmath217 ) . in this case
, the resultant cmb power spectrum is given in the form @xmath218_{\alpha=1},\end{aligned}\ ] ] where @xmath215 here is @xmath219 or @xmath220 depending on the scenario . in fig .
[ fig : cl(iso ) ] , we plot the total angular power spectrum for several values of @xmath219 ( with @xmath221 ) . as can be seen , as the effects of the isocurvature mode become more significant , the power spectrum at the high multipole is more suppressed relative to @xmath3 at small @xmath222 .
in addition , since the effects of @xmath154 at high multipole are very small , the structure of the acoustic peaks are almost unchanged from the adiabatic result unless the @xmath219-parameter becomes extremely large .
since the isocurvature mode changes the shape of the cmb power spectrum , upper bounds on the @xmath215-parameters can be obtained ; if the @xmath215-parameters are too large , the acoustic peaks are so suppressed that the cmb power spectrum becomes inconsistent with the observations.for other discussions on the constraints on the uncorrelated isocurvature perturbations , see @xcite . as in the previous case
, we calculate the @xmath178 variable as a function of @xmath219 , and the results are shown in fig .
[ fig : chi2(iso ) ] . from the figure
, we see that too large @xmath219 is disfavored by the current observations ; taking @xmath168 , which gives the most conservative upper bound on @xmath219 , the constraint @xmath223 at 95 % c.l .
is obtained .
the cmb angular power spectrum with @xmath224 can be also studied .
in fact , the above results with @xmath225 can be translated to the cdm isocurvature cases ; the cmb angular power spectrum for the baryonic and the cdm isocurvature cases are the same if @xmath226 ; using this relation , we can also derive the upper bound on @xmath220 .
thus , in any case , the heights of the acoustic peaks decrease relative to the sw tail as the ( uncorrelated ) isocurvature fluctuation contaminates .
now , we study effects of the correlated entropy fluctuation in the baryonic sector ( i.e. , case with @xmath227 ) . here , we assume other entropy perturbations to vanish ( i.e. , @xmath228 ) .
the general case will be discussed later.notice that the @xmath215-parameters vanish , for example , for the case where the effective mass of the @xmath92 field becomes comparable to ( or larger than ) the expansion rate during the inflation . in this case , the total angular power spectrum is given in the form @xmath229_{r_b=1}.\end{aligned}\ ] ] thus , we expect that , as the @xmath213-parameter increases , the angular power spectrum at the high multipole is more enhanced relative to @xmath3 at low multipole . in fig .
[ fig : cl(phi ) ] , we plot the resultant angular power spectrum with several values of @xmath213 .
as expected , @xmath3 at the high multipole is more enhanced relative to that at low ones as the @xmath213-parameter increases , and hence it can be a signal of the late - decaying scalar condensation .
in fact , if the @xmath213-parameter is too large , the angular power spectrum at high multipole is too enhanced , which becomes inconsistent with the currently available data . for the quantitative discussion
, we calculate the goodness - of - fit parameter as the previous case as a function of @xmath213 .
the results are shown in fig .
[ fig : chi2(rb ) ] for several sets of the cosmological parameters .
the figure shows that too large @xmath213 becomes inconsistent with the observations _ if the total angular power spectrum is given by the sum of @xmath155 and @xmath156_. numerically , even adopting the most conservative constraint , @xmath230 is excluded at 95 % c.l . before closing this subsection
, we would like to comment on the case where the cosmic density perturbations are totally from the primordial fluctuation of @xmath0 , i.e. , the case of the curvaton scenario . in this class of scenario , if the curvaton field @xmath0 does not generate the baryon ( or the cdm ) , the adiabatic and isocurvature perturbations may be generated with cross - correlation with the relation given in eq .
( [ sb ] ) ( or eq .
( [ sc])).in more complicated cases , this may not be the case . if some fraction of the baryon is from the decay product of @xmath0 while the rest is from @xmath92 , for example , the relation given in eq .
( [ sb ] ) does not hold . in this paper , however , we do not consider such a complicated scenario .
the curvaton scenario requires @xmath231 and hence @xmath213 ( or @xmath214 ) becomes much larger than 1 , which is already excluded by the current observations .
( @xmath232 is also excluded as will be shown in the next subsection . )
one caveat to this argument is that , if the @xmath92 field is so light that it starts to oscillate after the curvaton field dominates the universe , @xmath213 ( or @xmath214 ) vanishes and such a scenario is consistent with the current data . in this subsection , we consider the case with the correlated entropy perturbation in the cdm sector .
effects of the correlated entropy perturbation in the cdm sector is different from those in the baryonic sector .
since the density parameter for the cdm @xmath159 is significantly larger than @xmath158 , entropy between the non - relativistic matter and the radiation becomes larger compared to the baryonic case .
then , the effects of the correlation between the adiabatic and isocurvature perturbations become less significant while the effect of the isocurvature perturbation itself becomes more important . as a result , as discussed in the previous section , if there exists the correlated entropy perturbation in the cdm sector with the relation ( [ sc ] ) , the acoustic peaks become lower relative to the sw tail . in fig .
[ fig : cl(rc ) ] , we show the resultant cmb angular power spectrum for several values of @xmath214 ( with @xmath233 ) . as @xmath214 increases ,
the acoustic peaks are more suppressed relative to the @xmath3 at lower multipole . as in the previous cases ,
if the effect of the isocurvature perturbation becomes too effective , the resultant power spectrum becomes inconsistent with the observations .
thus , using the current experimental data , we can put an upper bound on the @xmath214-parameter . in fig .
[ fig : chi2(rc ) ] , we plot the goodness - of - fit parameter @xmath178 as a function of @xmath214 for several sets of the cosmological parameters . as one can see , even adopting the most conservative case , @xmath214 larger than 2.0 is excluded at 95 % c.l .
now , we are ready to study the most general case , the case with @xmath234 and @xmath235 . in this case , the total angular power spectrum is written in the form @xmath236_{r=1 } + \alpha^2 \left[c_l^{(\delta\chi ) } \right]_{\alpha=1}.\end{aligned}\ ] ] effects of the non - vanishing @xmath206- and @xmath215-parameters can be read off from fig .
[ fig : cl s ] ; normalizing the sw tail , non - vanishing @xmath213 increases the heights of the acoustic peaks while non - vanishing @xmath214 , @xmath219 , and @xmath220 suppress the acoustic peaks .
first , let us consider the baryonic isocurvature perturbations . in figs .
[ fig : c1st ] and [ fig : c2nd ] , we show the contours of the constant @xmath237 and @xmath238 on the @xmath213 vs. @xmath219 plane , where @xmath239 and @xmath240 are the heights of the first and second acoustic peaks , respectively .
as one can see , larger value of @xmath213 ( @xmath219 ) corresponds to larger ( smaller ) ratio of @xmath237 .
importantly , the correlated and uncorrelated entropy perturbations affect the heights of the acoustic peaks in the opposite way .
thus , even with extremely large values of the @xmath213- and @xmath219-parameters , the ratio @xmath241 may take a similar value as that from the adiabatic case if @xmath213 and @xmath219 are properly related .
the ratio @xmath242 is less sensitive to the parameters @xmath213 and @xmath219 , and for larger ( smaller ) value of @xmath213 ( @xmath219 ) , the ratio @xmath238 is suppressed .
we also calculated the ratio @xmath243 ( with @xmath244 being the height of the third acoustic peak ) and found that this ratio is almost independent of @xmath213 and @xmath219 .
this is because the structure of the angular power spectrum at high multipole is dominantly determined by the adiabatic part of the perturnations .
we would like to emphasize that the effects of the primordial fluctuations of the @xmath0 and @xmath92 field are quite different ; @xmath156 is enhanced at high multipole while @xmath154 becomes larger at low one .
thus , if the parameters @xmath213 and @xmath219 obey a specific relation , the adiabatic result may be mimiced even when the effects of the entropy perturbation are quite large . to study this issue
, we calculate the goodness - of - fit parameter @xmath178 as a function of @xmath213 and @xmath219 ( with @xmath245 ) . here , the overall normalization of @xmath3 is chosen such that the @xmath178 variable is minimized .
the results are shown in fig .
[ fig : chi2(ra ) ] for several sets of the cosmological parameters . here , all the primordial density perturbations are assumed to be scale - invariant .
for the cases with @xmath181 and 0.3 , it is interesting that large values of @xmath213 and @xmath219 are allowed .
in particular , for the case with @xmath181 , only the case with non - vanishing @xmath213 and @xmath219 is allowed .
importantly , in this parameter region , the cmb anisotropy originates to non - adiabatic density perturbations while the contribution from the inflaton fluctuation @xmath155 becomes negligibly small . in the following , we call this case as `` modified - curvaton '' case .
it should be noticed that , in the modified - curvaton case , the parameters @xmath213 and @xmath219 satisfy the relation @xmath246 for @xmath168 . for @xmath247 , large values of @xmath213 and @xmath219 are not allowed in contrast to other cases .
it is interesting to study whether the angular power spectrum in the modified - curvaton case can be distinguished from the adiabatic one by the on - going and future experiments . in fig .
[ fig : cl(adiiso ) ] , we plot the cmb angular power spectrum in the adiabatic and modified - curvaton cases , @xmath248 , and @xmath249 and @xmath250 ( which minimizes the @xmath178 variable for @xmath249 ) , respectively . in the modified - curvaton case ,
shape of the power spectrum at high multipole is determined by @xmath156 which is similar to that of the adiabatic case .
thus , it is difficult to distinguish two cases just by looking at the power spectrum at high multipole .
however , as one can see , the structure of the angular power spectrum at relatively low multipole is significantly affected .
in particular , the ratio @xmath238 will provide us an interesting probe for studying the modified - curvaton scenario .
since the map is expected to measure the angular power spectrum at @xmath4 level up to the multipole @xmath251 , @xmath3 at the high multipole may be used to calibrate the power spectrum to see the distortion of the cmb angular power spectrum from the adiabatic result .
we can also consider the case where the cdm sector has correlated and uncorrelated entropy perturbations ( i.e. , the case with @xmath252 and @xmath224 ) .
importantly , correlated and uncorrelated entropy perturbations both suppress the angular power spectrum at high multipole relative to the low one .
thus , in this case , too large @xmath214 is always excluded irrespective of the value of @xmath220 if the baryon is from the decay product of @xmath0 .
in this paper , we discussed the effects of the late - time entropy production due to the decay of the scalar - field condensations on the cosmic density perturbations . if the universe is reheated by the decay of the scalar field @xmath0 , many of the components in the present universe are generated from the decay products of the @xmath0 field .
in such a case , cosmic density perturbations are affected by the fluctuation of the amplitude of @xmath0 which may be generated during the inflation .
if all the components in the universe originate to the decay product of @xmath0 , density perturbations generated from the primordial fluctuation of @xmath0 becomes adiabatic . in this case , the cmb angular power spectrum from the fluctuation of @xmath0 becomes the usual adiabatic ones with ( almost ) scale - invariant spectrum .
if this becomes the dominant part of the cosmic density perturbations , then we have seen that the constraints on inflation models from observations of the cmb angular power spectrum are drastically relaxed . if the baryon or the cdm is not generated from @xmath0 but from a new scalar field @xmath92 , on the contrary
, correlated mixture of the adiabatic and isocurvature perturbations may arise .
in particular , if the @xmath92 field starts to oscillate much before the @xmath0 field dominates the universe , the metric and entropy perturbations obey the model - independent relation . in this case
, the cmb angular power spectrum may be significantly affected and the shape of the resultant power spectrum depends on which component has the correlated isocurvature perturbation . if the baryonic component has the correlated isocurvature perturbation , the density perturbations after the decay of @xmath0 become those given in eq .
( [ sb ] ) . in this case , @xmath3 at high multipole is more enhanced relative to that at low multipole . on the contrary , if the cdm is not generated from @xmath0 , correlated isocurvature perturbation can be induced in the cdm component as given in eq .
( [ sc ] ) . in this case ,
heights of the acoustic peaks become lower relative to the sw tail .
if there exists an contamination of these component into the total angular power spectrum , @xmath3 differs from the adiabatic one .
the important point is that , in both cases , too much correlated isocurvature perturbations become inconsistent with the present observation of the cmb power spectrum .
even with the correlated isocurvature perturbation in the baryonic sector , heights of the acoustic peaks can be reduced by introducing uncorrelated isocurvature perturbation .
this may happen , for example , if we take account of the primordial fluctuation of the scalar field @xmath92 .
( for example , this scalar field may be the affleck - dine field . ) in particular , if the sizes of the correlated and uncorrelated entropy perturbations are properly chosen , the resultant cmb power spectrum becomes consistent with the present observations without the effect of the inflaton fluctuation . even in this case , the angular power spectrum is not exactly the same as the conventional adiabatic one and the deviation from the adiabatic result may be observed at on - going and future experiments . in summary , if there exists correlated mixture of the adiabatic and isocurvature perturbations , the total cmb angular power spectrum can be distorted and it can be a signal of the late - time entropy production due to the decay of the scalar field condensation .
the on - going and future experiments may observe such a signal . _
acknowledgment : _ we acknowledge the use of cmbfast @xcite and radpack @xcite packages for our numerical calculations .
this work is supported by the grant - in - aid for scientific research from the ministry of education , science , sports , and culture of japan , no .
12047201 and no .
13740138 .
r. durrer , m. gasperini , m. sakellariadou and g. veneziano , phys .
* d59 * ( 1999 ) 043511 ; r. durrer , m. gasperini , m. sakellariadou and g. veneziano , phys .
* b436 * ( 1998 ) 66 ; a. melchiorri , f. vernizzi , r. durrer and g. veneziano , phys . rev .
* 83 * ( 1999 ) 4464 .
j. khoury , b.a .
ovrut , p.j .
steinhardt and n. turok , phys .
* d64 * ( 2001 ) 123522 ; j. khoury , b.a .
ovrut , n. seiberg , p.j .
steinhardt and n. turok , hep - th/0108187 ; j. khoury , b.a .
ovrut , p.j .
steinhardt and n. turok , hep - th/0109050 .
d. langlois .
* d59 * ( 1999 ) 123512 ; c. gordon , d. wands , b.a .
bassett and r. maartens , phys .
* d63 * ( 2001 ) 023506 ; n. bartolo , s. matarrese and a. riotto , phys .
* d64 * ( 2001 ) 083514 ; phys .
* d64 * ( 2001 ) 123504 ; l. amendola , c. gordon , d. wands and m. sasaki , astro - ph/0107089 .
r. stompor , a.j .
banday and k.m .
gorski , astrophys . j. * 463*(1996 ) 8 ; s.d .
burns , astro - ph/9711303 ; t. kanazawa , m. kawasaki , n. sugiyama and t. yanagida , prog .
* 102 * ( 1999 ) 71 ; k. enqvist , h. kurki - suonio and j. valiviita , phys .
* d62 * ( 2000 ) 03003 ; phys .
* d65 * ( 2002 ) 043002 . | we study the cosmic density perturbations induced from fluctuation of the amplitude of late - decaying scalar condensations ( called @xmath0 ) in the scenario where the scalar field @xmath0 once dominates the universe .
in such a scenario , the cosmic microwave background ( cmb ) radiation originates to decay products of the scalar condensation and hence its anisotropy is affected by the fluctuation of @xmath0 .
it is shown that the present cosmic density perturbations can be dominantly induced from the primordial fluctuation of @xmath0 , not from the fluctuation of the inflaton field .
this scenario may change constraints on the source of the density perturbations , like inflation .
in addition , a correlated mixture of adiabatic and isocurvature perturbations may arise in such a scenario ; possible signals in the cmb power spectrum are discussed .
we also show that the simplest scenario of generating the cosmic density perturbations only from the primordial fluctuation of @xmath0 ( i.e. , so - called `` curvaton '' scenario ) is severely constrained by the current measurements of the cmb angular power spectrum if correlated mixture of the adiabatic and isocurvature perturbations are generated .
tu-654 + hep - ph/0206026 + june , 2002 + .5 in * * cosmic density perturbations + from late - decaying scalar condensations .45 in takeo moroi and tomo takahashi .45 in _ department of physics , tohoku university , sendai 980 - 8578 , japan _ .4 in |
the _ virtual element method _ ( vem ) , introduced in @xcite , is a recent generalization of the finite element method , which is characterized by the capability of dealing with very general polygonal / polyhedral meshes .
the interest in numerical methods that can make use of general polytopal meshes has recently undergone a significant growth in the mathematical and engineering literature ; among the large number of papers on this subject , we cite as a minimal sample @xcite . indeed
, polytopal meshes can be very useful for a wide range of reasons including meshing of the domain , automatic use of hanging nodes , moving meshes and adaptivity .
vem has been applied successfully in a large range of problems ; see for instance @xcite .
the object of this paper is to introduce and analyze an a posteriori error estimator of residual type for the virtual element approximation of the steklov eigenvalue problem .
in fact , due to the large flexibility of the meshes to which the virtual element method is applied , mesh adaptivity becomes an appealing feature as mesh refinement strategies can be implemented very efficiently .
for instance , hanging nodes can be introduced in the mesh to guarantee the mesh conformity without spreading the refined zones .
in fact hanging nodes introduced by the refinement of a neighboring element are simply treated as new nodes since adjacent non matching element interfaces are perfectly acceptable . on the other hand , polygonal cells with very general shapes
are admissible thus allowing us to adopt simple mesh coarsening algorithms .
the approximation of eigenvalue problems has been the object of great interest from both the practical and theoretical points of view , since they appear in many applications .
we refer to @xcite and the references therein for the state of art in this subject area .
in particular , the steklov eigenvalue problem , which involves the laplace operator but is characterized by the presence of the eigenvalue in the boundary condition , appears in many applications ; for example , the study of the vibration modes of a structure in contact with an incompressible fluid ( see @xcite ) and the analysis of the stability of mechanical oscillators immersed in a viscous media ( see @xcite ) .
one of its main applications arises from the dynamics of liquids in moving containers , i.e. , sloshing problems ( see @xcite ) . on the other hand , adaptive mesh refinement strategies based on a posteriori error indicators
play a relevant role in the numerical solution of partial differential equations in a general sense .
for instance , they guarantee achieving errors below a tolerance with a reasonable computer cost in presence of singular solutions .
several approaches have been considered to construct error estimators based on the residual equations ( see @xcite and the references therein ) . in particular , for the steklov eigenvalue problem we mention @xcite . on the other hand , the design and analysis of a posteriori error bounds for the vem is a challenging task .
references @xcite are the only a posteriori error analyses for vem currently available in the literature . in @xcite ,
a posteriori error bounds for the @xmath0-conforming vem for the two - dimensional poisson problem are proposed . in @xcite ,
a posteriori error bounds are introduced for the @xmath1-conforming vem proposed in @xcite for the discretization of second order linear elliptic reaction - convection - diffusion problems with non constant coefficients in two and three dimensions .
we have recently developed in @xcite a virtual element method for the steklov eigenvalue problem . under standard assumptions on the computational domain
, we have established that the resulting scheme provides a correct approximation of the spectrum and proved optimal order error estimates for the eigenfunctions and a double order for the eigenvalues . in order to exploit the capability of vem in the use of general polygonal meshes and its flexibility for the application of mesh adaptive strategies
, we introduce and analyze an a posteriori error estimator for the virtual element approximation introduced in @xcite .
since normal fluxes of the vem solution are not computable , they will be replaced in the estimators by a proper projection . as a consequence of this replacement
, new additional terms appear in the a posteriori error estimator , which represent the virtual inconsistency of vem .
similar terms also appear in the other papers for a posteriori error estimates of vem ( see @xcite ) .
we prove that the error estimator is equivalent to the error and use the corresponding indicator to drive an adaptive scheme .
the outline of this article is as follows : in section [ sec : apriori ] we present the continuous and discrete formulations of the steklov eigenvalue problem together with the spectral characterization .
then , we recall the a priori error estimates for the virtual element approximation analyzed in @xcite . in section [ sec : posteriorierror ] , we define the a posteriori error estimator and proved its reliability and efficiency .
finally , in section [ sec : numerejemplo ] , we report a set of numerical tests that allow us to assess the performance of an adaptive strategy driven by the estimator . we have also made a comparison between the proposed estimator and the standard edge - residual error estimator for a finite element method . throughout the article we will denote by @xmath2 a generic constant independent of the mesh parameter @xmath3 , which may take different values in different occurrences .
let @xmath4 be a bounded domain with polygonal boundary @xmath5 .
let @xmath6 and @xmath7 be disjoint open subsets of @xmath5 such that @xmath8 with @xmath9 .
we denote by @xmath10 the outward unit normal vector to @xmath5 .
we consider the following eigenvalue problem : find @xmath11 , @xmath12 , such that @xmath13 \dfrac{\partial w}{\partial n } = \left\{\begin{array}{ll } \l w & \text{on } \go , \\ 0 & \text{on } \g_1 . \end{array}\right . \end{array}\right.\ ] ] by testing the first equation above with @xmath14 and integrating by parts , we arrive at the following equivalent weak formulation : [ p1 ] find @xmath11 , @xmath12 , such that @xmath15 according to ( * ? ? ? * theorem 2.1 ) , we know that the solutions @xmath16 of the problem above are : * @xmath17 , whose associated eigenspace is the space of constant functions in @xmath18 ; * a sequence of positive finite - multiplicity eigenvalues @xmath19 such that @xmath20 .
the eigenfunctions corresponding to different eigenvalues are orthogonal in @xmath21 .
therefore the eigenfunctions @xmath22 corresponding to @xmath23 satisfy @xmath24 we denote the bounded bilinear symmetric forms appearing in problem [ p1 ] as follows : @xmath25 let @xmath26 be a sequence of decompositions of @xmath18 into polygons @xmath27 .
we assume that for every mesh @xmath28 , @xmath29 and @xmath30 are union of edges of elements @xmath31 .
let @xmath32 denote the diameter of the element @xmath27 and @xmath3 the maximum of the diameters of all the elements of the mesh , i.e. , @xmath33 . for the analysis
, we will make as in @xcite the following assumptions . *
every mesh @xmath28 consists of a finite number of _ simple _ polygons ( i.e. , open simply connected sets with non self intersecting polygonal boundaries ) . * * a2 .
* there exists @xmath34 such that , for all meshes @xmath28 , each polygon @xmath35 is star - shaped with respect to a ball of radius greater than or equal to @xmath36 . *
* there exists @xmath37 such that , for all meshes @xmath28 , for each polygon @xmath35 , the distance between any two of its vertices is greater than or equal to @xmath38 .
we consider now a simple polygon @xmath27 and , for @xmath39 , we define @xmath40 we then consider the finite - dimensional space defined as follows : @xmath41 where , for @xmath42 , we have used the convention that @xmath43 .
we choose in this space the degrees of freedom introduced in ( * ? ? ?
* section 4.1 ) . finally ,
for every decomposition @xmath28 of @xmath18 into simple polygons @xmath27 and for a fixed @xmath39 , we define @xmath44 in what follows , we will use standard sobolev spaces , norms and seminorms and also the broken @xmath45-seminorm @xmath46 which is well defined for every @xmath47 such that @xmath48 for each polygon @xmath35 .
we split the bilinear form @xmath49 as follows : @xmath50 where @xmath51 due to the implicit space definition , we must have into account that we would not know how to compute @xmath52 for @xmath53 .
nevertheless , the final output will be a local matrix on each element @xmath27 whose associated bilinear form can be exactly computed whenever one of the two entries is a polynomial of degree @xmath54 .
this will allow us to retain the optimal approximation properties of the space @xmath55 . with this end , for any @xmath35 and for any sufficiently regular function @xmath56 ,
we define first @xmath57 where @xmath58 , @xmath59 , are the vertices of @xmath27 .
then , we define the projector @xmath60 for each @xmath61 as the solution of @xmath62 on the other hand , let @xmath63 be any symmetric positive definite bilinear form to be chosen as to satisfy @xmath64 for some positive constants @xmath65 and @xmath66 independent of @xmath27 . then , set @xmath67 where @xmath68 is the bilinear form defined on @xmath69 by @xmath70 notice that the bilinear form @xmath63 has to be actually computable for @xmath71 .
the following properties of @xmath68 have been established in ( * ? ? ?
* theorem 4.1 ) . * _ @xmath54-consistency _ : @xmath72 * _ stability _ : there exist two positive constants @xmath73 and @xmath74 , independent of @xmath27 , such that : @xmath75 now , we are in a position to write the virtual element discretization of problem [ p1 ] .
[ p11 ] find @xmath76 , @xmath77 , such that @xmath78 according to ( * ? ? ?
* theorem 3.1 ) we know that the solutions @xmath79 of the problem above are : * @xmath80 , whose associated eigenfunction are the constant functions in @xmath18 .
* @xmath81 , with @xmath82 , which are positive eigenvalues repeated according to their respective multiplicities .
moreover , the eigenfunctions corresponding to different eigenvalues are orthogonal in @xmath21 . therefore the eigenfunctions @xmath83 corresponding to @xmath84 satisfy @xmath85 let @xmath16 be a solution to problem [ p1 ] .
we assume @xmath86 is a simple eigenvalue and we normalize @xmath87 so that @xmath88
. then , for each mesh @xmath89 , there exists a solution @xmath79 of problem [ p11 ] such that @xmath90 , @xmath91 and @xmath92 as @xmath93 .
moreover , according to and , we have that @xmath87 and @xmath94 belong to the space @xmath95 let us remark that the following generalized poincar inequality holds true in this space : there exists @xmath96 such that @xmath97 the following a priori error estimates have been proved in ( * ? ? ?
* theorems 4.24.4 ) : there exists @xmath96 such that for all @xmath98 @xmath99 where the constant @xmath100 is the sobolev exponent for the laplace problem with neumann boundary conditions .
let us remark that @xmath101 , if @xmath18 is convex , and @xmath102 with @xmath103 being the largest re - entrant angle of @xmath18 , otherwise .
the aim of this section is to introduce a suitable residual - based error estimator for the steklov eigenvalue problem which be fully computable , in the sense that it depends only on quantities available from the vem solution .
then , we will show its equivalence with the error .
for this purpose , we introduce the following definitions and notations . for any polygon @xmath104 ,
we denote by @xmath105 the set of edges of @xmath27 and @xmath106 we decompose @xmath107 , where @xmath108 , @xmath109 and @xmath110 . for each inner edge @xmath111 and for any sufficiently smooth function @xmath112
, we define the jump of its normal derivative on @xmath113 by @xmath114\!\!\right]_\ell:=\nabla ( v|_{\e } ) \cdot n_{\e}+\nabla ( v|_{\e ' } ) \cdot n_{\e ' } , \ ] ] where @xmath27 and @xmath115 are the two elements in @xmath89 sharing the edge @xmath113 and @xmath116 and @xmath117 are the respective outer unit normal vectors . as a consequence of the mesh regularity assumptions , we have that each polygon @xmath35 admits a sub - triangulation @xmath118 obtained by joining each vertex of @xmath27 with the midpoint of the ball with respect to which @xmath27 is starred
let @xmath119 .
since we are also assuming * a3 * , @xmath120 is a shape - regular family of triangulations of @xmath18 .
we introduce bubble functions on polygons as follows ( see @xcite ) .
an interior bubble function @xmath121 for a polygon @xmath27 can be constructed piecewise as the sum of the cubic bubble functions for each triangle of the sub - triangulation @xmath118 that attain the value 1 at the barycenter of each triangle . on the other hand ,
an edge bubble function @xmath122 for @xmath123 is a piecewise quadratic function attaining the value 1 at the barycenter of @xmath113 and vanishing on the triangles @xmath124 that do not contain @xmath113 on its boundary .
the following results which establish standard estimates for bubble functions will be useful in what follows ( see @xcite ) .
[ burbujainterior ] for any @xmath104 , let @xmath125 be the corresponding interior bubble function . then , there exists a constant @xmath96 independent of @xmath32 such that @xmath126 [ burbuja ] for any @xmath104 and @xmath127 , let @xmath122 be the corresponding edge bubble function .
then , there exists a constant @xmath96 independent of @xmath32 such that @xmath128 moreover , for all @xmath129 , there exists an extension of @xmath130 ( again denoted by @xmath131 ) such that @xmath132 [ extencion ] a possible way of extending @xmath131 from @xmath127 to @xmath27 so that lemma [ burbuja ] holds is as follows : first we extend @xmath131 to the straight line @xmath133 using the same polynomial function .
then , we extend it to the whole plain through a constant prolongation in the normal direction to @xmath134 . finally , we restrict the latter to @xmath27 .
the following lemma provides an error equation which will be the starting point of our error analysis .
from now on , we will denote by @xmath135 the eigenfunction error and by @xmath136\!\!\right]_{\ell},\quad & \ell\in \ce_{\o } , \\[0.4 cm ] \l_h w_h-\dfrac{\partial ( \pik w_{h})}{\partial n},\quad & \ell\in\ce_{\g_0 } , \\[0.4 cm ] -\dfrac{\partial ( \pik w_{h})}{\partial n},\quad & \ell\in\ce_{\g_1 } , \end{array}\right.\ ] ] the edge residuals . notice that @xmath137 are actually computable since they only involve values of @xmath94 on @xmath138 ( which are computable in terms of the boundary degrees of freedom ) and @xmath139 which is also computable .
[ ext2 ] for any @xmath140 , we have the following identity : @xmath141.\ ] ] using that @xmath16 is a solution of problem [ p1 ] , adding and subtracting @xmath142 and integrating by parts , we obtain @xmath143 \\ & = \l b(w , v)-\sum_{\e\in \ct_h}a^\e(w_h-\pik w_h , v ) -\sum_{\e\in \ct_h}\left[-\int_{\e}\delta(\pik w_h)\,v + \int_{\partial \e}\dfrac{\partial(\pik w_{h } ) } { \partial{n}}\,v\right ] \\ &
= \l b(w , v)-\sum_{\e\in\ct_h}a^\e(w_h-\pik w_h , v ) \\ & \quad + \sum_{\e\in\ct_h}\left[\int_{\e}\delta(\pik w_h)\,v -\!\!\!\!\!\!\sum_{\ell\in\ce_{\e}\cap(\ce_{\go}\cup\ce_{\g_1 } ) } \int_{\ell}\dfrac{\partial(\pik w_{h})}{\partial{n}}\,v + \dfrac{1}{2}\sum_{\ell\in\ce_{\e}\cap\ce_{\o } } \int_{\ell}\left[\!\!\left[\dfrac{\partial(\pik w_{h } ) } { \partial{n}}\right]\!\!\right]_{\ell}v\right].\end{aligned}\ ] ] finally , the proof follows by adding and subtracting the term @xmath144 . for all @xmath31
, we introduce the local terms @xmath145 and @xmath146 and the local error indicator @xmath147 by @xmath148 we also introduce the global error estimator by @xmath149 [ rkk ] the indicators @xmath147 include the terms @xmath145 which do not appear in standard finite element estimators .
this term , which represent the virtual inconsistency of the method , has been introduced in @xcite for a posteriori error estimates of other vem .
let us emphasize that it can be directly computed in terms of the bilinear form @xmath63 .
in fact , @xmath150 first , we provide an upper bound for the error .
[ erroipo ] there exists a constant @xmath96 independent of @xmath3 such that @xmath151 , there exists @xmath152 satisfying ( see ( * ? ? ?
* proposition 4.2 ) ) @xmath153 then , we have that @xmath154}_{t_{2}}\\ & -\underbrace{\sum_{\e\in \ct_h}a^{\e } ( w_h-\pik w_h , e - e_i)}_{t_{3}}+\underbrace{a_h(w_h , e_i)-a ( w_h , e_i)}_{t_{4 } } , \end{split}\ ] ] the last equality thanks to lemma [ ext2 ] .
next , we bound each term @xmath155 separately . for @xmath156
, we use the definition of @xmath157 , the fact that @xmath158 , a trace theorem and to write @xmath159 for @xmath160 , first , we use a local trace inequality ( see ( * ? ? ?
* lemma 14 ) ) and to write @xmath161 hence , using again , we have @xmath162\\\nonumber & \leq c\sum_{\e\in \ct_h}\left[h_\e\|\delta ( \pik w_h)\|_{0,\e}\|e\|_{1,\e}+\sum_{\ell \in \ce_{\e}}h_\e^{1/2}\|j_{\ell}\|_{0,\ell}\|e\|_{1,\e}\right]\\ & \leq c\left\{\sum_{\e\in \ct_h}\left[h_\e^2\|\delta ( \pik w_h)\|_{0,\e}^2+\sum_{\ell \in \ce_{\e}}h_\e\|j_{\ell}\|_{0,\ell}^2\right]\right\}^{1/2}|e|_{1,\o},\end{aligned}\ ] ] where for the last estimate we have used . to bound @xmath163 , we use the _ stability _ property and to write @xmath164 where for the last estimate we have used remark [ rkk ] and again .
finally , to bound @xmath165 , we add and subtract @xmath166 on each @xmath167 and use the _ @xmath54-consistency _ property : @xmath168\\\nonumber & \leq\sum_{\e\in \ct_h}a_h^\e(w_h-\pik w_h , w_h-\pik w_h)^{1/2}a_{h}^\e(e_{i},e_i)^{1/2}\\\nonumber & \quad+\sum_{\e\in \ct_h}a^\e(w_h-\pik w_h , w_h-\pik w_h)^{1/2}a^\e(e_{i},e_i)^{1/2}\\\nonumber & \leq c\sum_{\e\in \ct_h}a_h^\e(w_h-\pik w_h , w_h-\pik w_h)^{1/2}|e_{i}|_{1,\e}\\ & \leq c \left(\sum_{\e\in \ct_h}\theta_{\e}^{2}\right)^{1/2}|e|_{1,\o},\end{aligned}\ ] ] where we have used the _ stability _ property , and for the last two inequalities
. thus , the result follows from .
although the virtual approximate eigenfunction is @xmath94 , this function is not known in practice .
instead of @xmath94 , what can be used as an approximation of the eigenfunction is @xmath169 , where @xmath170 is defined for @xmath171 by @xmath172 notice that @xmath169 is actually computable .
the following result shows that an estimate similar to that of theorem [ erroipo ] holds true for @xmath169 .
[ corolario2 ] there exists a constant @xmath96 independent of @xmath3 such that @xmath173 for each polygon @xmath35 , we have that @xmath174 now , using together with remark [ rkk ] , we have that @xmath175 thus , the result follows from theorem [ erroipo ] . in what follows ,
we prove a convenient upper bound for the eigenvalue approximation .
[ cotalambda ] there exists a constant @xmath96 independent of @xmath3 such that @xmath176 from the symmetry of the bilinear forms together with the facts that @xmath177 for all @xmath140 , @xmath178 for all @xmath179 and @xmath180 , we have @xmath181\\ & \leq c\left[|w - w_h|_{1,\o}^2+|a_h(w_h , w_h)-a(w_h , w_h)|\right],\end{aligned}\ ] ] where we have also used a trace theorem and .
we now bound the last term on the right - hand side above using the definition of @xmath182 and : @xmath183 -\sum_{\e\in\ct_h}a^\e(w_h , w_h)\right| \\ & \,\ , \leq\left|\sum_{\e\in\ct_h } \left[a^\e\big(\pik w_h,\pik w_h\big ) -a^\e(w_h , w_h)\right]\right|+\sum_{\e\in\ct_h } c_1\,a^\e\big(w_h-\pik w_h , w_h-\pik w_h\big ) \\ & \,\ , = \sum_{\e\in\ct_h}\left(1+c_1\right ) a^\e\big(w_h-\pik w_h , w_h-\pik w_h\big)\\ & \leq \left(1+c_1\right)\sum_{\e\in\ct_h}\left(\left|w_h - w\right|_{1,\e}^{2 } + \left|w-\pik w_{h}\right|_{1,\e}^{2}\right).\end{aligned}\ ] ] finally , from the above estimate and we obtain @xmath184 .
according to and , it seems reasonable to expect the term @xmath185 in the estimate of theorem [ erroipo ] to be of higher order than @xmath186 and hence asymptotically negligible .
however this can not be rigorously derived from and , which are only upper error bounds .
in fact , the actual error @xmath186 could be in principle of higher order than the estimate .
our next goal is to prove that the term @xmath185 is actually asymptotically negligible in the estimates of corollaries [ corolario2 ] and [ cotalambda ] . with this aim
, we will modify the estimate and prove that @xmath187 this proof is based on the arguments used in section 4 from @xcite . to avoid repeating them step by step , in what follows we will only report the changes that have to be made in order to prove .
we define in @xmath188 the bilinear form @xmath189 , which is elliptic ( * ? ? ?
* lemma 2.1 ) .
let @xmath190 be the solution of @xmath191 since @xmath177 we have that @xmath192 we also define in @xmath193 the bilinear form @xmath194 , which is elliptic uniformly in @xmath3 ( * ? ? ?
* lemma 3.1 ) .
let @xmath195 be the solution of @xmath196 the arguments in the proof of lemma 4.3 from @xcite can be easily modified to prove that @xmath197 then , using this estimate in the proof of theorem 4.4 from @xcite yields @xmath198 now , since as stated above @xmath199 , we have that @xmath200 to estimate the third term we recall first that @xmath201 then , subtracting this equation divided by @xmath202 from we have that @xmath203 hence , from the uniform ellipticity of @xmath204 in @xmath193 , we obtain @xmath205 therefore @xmath206 the last inequality because of poincar inequality .
then , substituting and into we obtain @xmath207 where we have used for the last inequality . substituting this and estimate into we obtain @xmath208 finally , substituting the above estimate and into , we conclude the proof of the following result .
[ asintotico ] there exists @xmath96 independent of @xmath3 such that @xmath209 using this result , now it easy to prove that the term @xmath185 in corollaries [ corolario2 ] and [ cotalambda ] is asymptotically negligible .
in fact , we have the following result .
there exist positive constants @xmath2 and @xmath210 such that , for all @xmath211 , there holds @xmath212 from lemma [ asintotico ] and corollary [ corolario2 ] we have @xmath213 hence , it is straightforward to check that there exists @xmath214 such that for all @xmath211 holds true . on the other hand , from lemma [ asintotico ]
and we have that for all @xmath211 @xmath215 then , for @xmath3 small enough , follows from corollary [ cotalambda ] and the above estimate .
we will show in this section that the local error indicators @xmath147 are efficient in the sense of pointing out which polygons should be effectively refined .
first , we prove an upper estimate of the volumetric residual term @xmath146 .
[ eficiencia1 ] there exists a constant @xmath96 independent of @xmath32 such that @xmath216 for any @xmath35 , let @xmath125 be the corresponding interior bubble function .
we define @xmath217 .
since @xmath112 vanishes on the boundary of @xmath27 , it may be extended by zero to the whole domain @xmath18 .
this extension , again denoted by @xmath112 , belongs to @xmath188 and from lemma [ ext2 ] we have @xmath218 since @xmath219 , using lemma [ burbujainterior ] and the above equality we obtain @xmath220 where , for the last inequality , we have used again lemma [ burbujainterior ] and together with remark [ rkk ] .
multiplying the above inequality by @xmath221 allows us to conclude the proof .
next goal is to obtain an upper estimate for the local term @xmath222 [ cotar ] there exists @xmath96 independent of @xmath32 such that @xmath223 from the definition of @xmath145 together with remark [ rkk ] and estimate we have @xmath224 the proof is complete .
the following lemma provides an upper estimate for the jump terms of the local error indicator .
there exists a constant @xmath96 independent of @xmath32 such that [ lema4 ] @xmath225 where @xmath226 .
first , for @xmath227 , we extend @xmath228 to the element @xmath27 as in remark [ extencion ] .
let @xmath122 be the corresponding edge bubble function .
we define @xmath229 .
then , @xmath112 may be extended by zero to the whole domain @xmath18 .
this extension , again denoted by @xmath112 , belongs to @xmath230 and from lemma [ ext2 ] we have that @xmath231 for @xmath228 , from lemma [ burbuja ] and the above equality we obtain @xmath232\\ & \leq c\left[\left(\vert e\vert_{1,\e}+\vert w_{h}-\pik w_{h}\vert_{1,\e}\right ) h_\e^{-1/2}\left\|j_{\ell}\right\|_{0,\ell}+h_\e^{-1}\left(\theta_{\e } + \vert e\vert_{1,\e}\right)h_\e^{1/2}\left\|j_{\ell}\right\|_{0,\ell}\right]\\ & \leq ch_\e^{-1/2}\left\|j_{\ell}\right\|_{0,\ell}\big(\vert e\vert_{1,\e}+\theta_{\e}\big),\end{aligned}\ ] ] where we have used again lemma [ burbuja ] together with estimate . multiplying by @xmath233 the above inequality allows us to conclude .
secondly , for @xmath234 , we extend @xmath229 to @xmath188 as in the previous case . taking into account that in this case @xmath235 and @xmath122 is a quadratic bubble function in @xmath27 , from lemma [ ext2 ] we obtain @xmath236 then , repeating the previous arguments we obtain @xmath237.\ ] ] hence , using lemma [ burbuja ] and a local trace inequality we arrive at @xmath238\\ & \leq ch_\e^{-1/2}\|j_{\ell}\|_{0,\ell}\left(\theta_{\e}+|e|_{1,\e}+h_\e^{1/2}\|\l w -\l_h w_h\|_{0,\ell}\right),\end{aligned}\ ] ] where we have used lemma [ burbuja ] again . multiplying by @xmath233 the above inequality
yields @xmath239 .
finally , for @xmath240 , we extend @xmath229 to @xmath188 as above again . taking into account that @xmath228 and @xmath122 is a quadratic bubble function in @xmath27 , from lemma [ ext2 ] we obtain @xmath241 then
, proceeding analogously to the previous case we obtain @xmath242.\ ] ] thus , the proof is complete .
now , we are in a position to prove an upper bound for the local error indicators @xmath243 .
[ eficiencia ] there exists @xmath96 such that @xmath244,\ ] ] @xmath245 .
it follows immediately from lemmas [ eficiencia1][lema4 ] . according to the above theorem
, the error indicators @xmath246 provide lower bounds of the error terms @xmath247 in the neighborhood @xmath248 of @xmath27 .
for those elements @xmath27 with an edge on @xmath138 , the term @xmath249 also appears in the estimate .
let us remark that it is reasonable to expect this terms to be asymptotically negligible .
in fact , this is the case at least for the global estimator @xmath250 as is shown in the following result .
there exists a constant @xmath96 such that @xmath251 from theorem [ eficiencia ] we have that @xmath252 the last term on the right hand side above is bounded as follows : @xmath253 where we have used that @xmath254 .
now , by using a trace inequality and poincar inequality we have @xmath255 on the other hand , using the estimate , we have @xmath256 and we conclude the proof .
in this section , we will investigate the behavior of an adaptive scheme driven by the error indicator in two numerical tests that differ in the shape of the computational domain @xmath18 and , hence , in the regularity of the exact solution . with this aim ,
we have implemented in a matlab code a lowest - order vem ( @xmath42 ) on arbitrary polygonal meshes following the ideas proposed in @xcite . to complete the choice of the vem , we had to choose the bilinear forms @xmath63 satisfying . in this respect , we proceeded as in ( * ? ? ?
* section 4.6 ) : for each polygon @xmath27 with vertices @xmath257 , we used @xmath258 in all our tests we have initiated the adaptive process with a coarse triangular mesh . in order to compare the performance of vem with that of a finite element method ( fem ) , we have used two different algorithms to refine the meshes .
the first one is based on a classical fem strategy for which all the subsequent meshes consist of triangles .
in such a case , for @xmath42 , vem reduces to fem .
the other procedure to refine the meshes is described in @xcite .
it consists of splitting each element into @xmath10 quadrilaterals ( @xmath10 being the number of edges of the polygon ) by connecting the barycenter of the element with the midpoint of each edge as shown in figure [ fig : cero ] ( see @xcite for more details ) .
notice that although this process is initiated with a mesh of triangles , the successively created meshes will contain other kind of convex polygons , as can be seen in figures [ fig : uno ] and [ fig : cinco ] . since we have chosen @xmath42 , according to the definition of the local virtual element space @xmath259 ( cf .
) , the term @xmath260 vanishes .
thus , the error indicators reduce in this case to @xmath261 let us remark that in the case of triangular meshes , the term @xmath262 vanishes too , since @xmath263 and hence @xmath264 is the identity . by the same reason
, the projection @xmath264 also disappears in the definition of @xmath137 .
therefore , for triangular meshes , not only vem reduces to fem , but also the error indicator becomes the classical well - known edge - residual error estimator ( see @xcite ) : @xmath265\!\!\right]_{\ell},\quad & \ell\in \ce_{\o } , \\[0.4 cm ] \l_h w_h-\dfrac{\partial w_{h}}{\partial{n } } , \quad & \ell\in\ce_{\g_0 } , \\[0.4 cm ] -\dfrac{\partial w_{h}}{\partial{n } } , \quad & \ell\in\ce_{\g_1}. \end{array}\right.\ ] ] in what follows , we report the results of a couple of tests . in both cases , we will restrict our attention to the approximation of the eigenvalues .
let us recall that according to corollary [ cotalambda ] , the global error estimator @xmath266 provides an upper bound of the error of the computed eigenvalue .
we have chosen for this test a problem with known analytical solution .
it corresponds to the computation of the sloshing modes of a two - dimensional fluid contained in the domain @xmath267 with a horizontal free surface @xmath6 as shown in figure [ fig : slosh ] .
the solutions of this problem are @xmath268 we have used the two refinement procedures ( vem and fem ) described above .
both schemes are based on the strategy of refining those elements @xmath27 which satisfy @xmath269 figures [ fig : uno ] and [ fig : dos ] show the adaptively refined meshes obtained with vem and fem procedures , respectively .
+ + since the eigenfunctions of this problem are smooth , according to we have that @xmath270 .
therefore , in case of uniformly refined meshes , @xmath271 , where @xmath272 denotes the number of degrees of freedom which is the optimal convergence rate that can be attained .
figure [ fig : tres ] shows the error curves for the computed lowest eigenvalue on uniformly refined meshes and adaptively refined meshes with fem and vem schemes .
the plot also includes a line of slope @xmath273 , which correspond to the optimal convergence rate of the method @xmath274 .
for uniformly refined meshes ( `` uniform fem '' ) , adaptively refined meshes with fem ( `` adaptive fem '' ) and adaptively refined meshes with vem ( `` adaptive vem'').,title="fig:",width=377,height=377 ] [ fig : tres ] it can be seen from figure [ fig : tres ] that the three refinement schemes lead to the correct convergence rate .
moreover , the performance of adaptive vem is slightly better than that of adaptive fem , while this is also better than uniform fem .
we report in table [ tabla : n2 ] , the errors @xmath275 and the estimators @xmath266 at each step of the adaptive vem scheme .
we include in the table the terms @xmath276 which arise from the inconsistency of vem and @xmath277 which arise from the edge residuals .
we also report in the table the effectivity indexes @xmath278 .
components of the error estimator and effectivity indexes on the adaptively refined meshes with vem . [ cols="^,^,^,^,^,^,^",options="header " , ] [ tabla : n3 ]
we have derived an a posteriori error indicator for the vem solution of the steklov eigenvalue problem .
we have proved that it is efficient and reliable . for lowest order elements on triangular meshes , vem coincides with fem and the a posteriori error indicators also coincide with the classical ones
. however vem allows using general polygonal meshes including hanging nodes , which is particularly interesting when designing an adaptive scheme .
we have implemented such a scheme driven by the proposed error indicators .
we have assessed its performance by means of a couple of tests which allow us to confirm that the adaptive scheme yields optimal order of convergence for regular as well as singular solutions .
the authors warmfully thanks loureno beiro da veiga from universit di milano - bicocca , italy , by many helpful discussions on this subject .
the authors also thank paola f. antonietti from politecnico di milano , italy , by allowing us to use her code for vem adaptive mesh refinement .
the first author was partially supported by conicyt ( chile ) through fondecyt project no .
1140791 and by diubb through project 151408 gi / vc , universidad del bo - bo , ( chile ) .
the second author was partially supported by a conicyt ( chile ) fellowship .
the third author was partially supported by basal project , cmm , universidad de chile ( chile ) .
gain , c. talischi and g.h .
paulino , _ on the virtual element method for three - dimensional linear elasticity problems on arbitrary polyhedral meshes _ , comput .
methods appl .
engrg . , 282 , ( 2014 ) , pp . | the paper deals with the a posteriori error analysis of a virtual element method for the steklov eigenvalue problem .
the virtual element method has the advantage of using general polygonal meshes , which allows implementing very efficiently mesh refinement strategies .
we introduce a residual type a posteriori error estimator and prove its reliability and efficiency .
we use the corresponding error estimator to drive an adaptive scheme .
finally , we report the results of a couple of numerical tests , that allow us to assess the performance of this approach . virtual element method , a posteriori error estimates , steklov eigenvalue problem , polygonal meshes 65n25 , 65n30 , 74s99 . |
the hot big - bang model of modern cosmology is based on the assumption that the large scale geometry and dynamics of the universe can be described by an exact flrw model , with more detailed and smaller scale physical phenomena , such as density fluctuations or gravitational waves , being described by perturbed fl models , i.e. by solutions of the efe linearized about an exact fl model .
we begin with the premise that it is important to study cosmological models more general than perturbed fl models .
in particular one is interested in the set @xmath3 , namely , the set of all universes that are compatible with current observations ( ellis 2004 ) .
a member of this set must have an epoch , possibly finite , during which the model is close to fl ; at early or later times it may deviate significantly from fl .
more generally we are interested in the set @xmath4 of _ all _ cosmological models , and in the relationship between @xmath4 , @xmath3 and @xmath5 , the set of all fl models .
for example , we want to study the relation between linearly perturbed fl models and models that are close to fl in some well - defined , gauge - invariant sense , and satisfy the exact efe .
are nonlinear effects perhaps significant in close - to - fl models ?
another question that takes one outside the realm of linearly perturbed fl models is the detailed nature of the generic cosmological singularity . with the preceding discussion as motivation we consider the class @xmath4 of all cosmological models ,
whose state at time @xmath6 can be represented by a vector @xmath7 in a state space @xmath8 .
the evolution of a model universe is then described by a curve @xmath9 in @xmath8 , called an orbit , that will be a solution of a system of first order autonomous evolution equations and constraints of the form @xmath10 where @xmath11 denotes partial differentiation with respect to the spatial coordinates @xmath12 , and @xmath13 denotes possible higher order spatial derivatives .
the goal is to choose state variables that remain bounded during the evolution of the model and a time variable @xmath6 such that @xmath14 at the initial singularity and @xmath15 at late times . within this framework
, classes of models with symmetries will be represented by invariant subsets of @xmath8 , the most important being the subset @xmath16 that describes the fl cosmologies .
one expects that the orbits of close - to - fl models will shadow orbits in @xmath16 . the appropriate mathematical vehicle for implementing the above program is the orthonormal frame formalism , since it expresses the efe directly as first order ( in time ) autonomous evolution equations .
it is also necessary to introduce a process of normalization in order to create bounded variables .
a choice that has proven effective is the so - called _ hubble normalization_. this approach to the study of cosmological dynamics has been discussed in some detail in wainwright & ellis 1997 , and more recently by coley 2003 , with emphasis on spatially homogeneous models . since then the approach has been extended to models without symmetry by uggla 2003 .
the plan of this paper is as follows . in section [ sec_hub ]
we describe the principal features of the hubble - normalized state space , and in section [ sec_c_i ] we discuss close - to - fl models and the related notion of isotropization . in section [ sec_sh ]
we give an overview of the dynamics of a special class of models in the cosmological hierarchy , the nontilted spatially homogeneous cosmologies . in section [ sec_vac_num ]
we describe a candidate for the past attractor in the hubble - normalized state space , and mention some recent numerical simulations which support this proposal . in section [ sec_col ]
we briefly introduce a modification of hubble normalization which provides a description of nontilted spatially homogeneous models that undergo recollapse .
section [ sec_conclusion ] contains our concluding remarks .
in this section we give the motivation for introducing hubble - normalized variables , and describe the principal features of the resulting hubble - normalized state space and evolution equations .
one of the rationals for using dynamical systems methods in cosmology is the hope that one can describe the evolution of cosmological models near the initial singularity by means of a _
past attractor _ of a cosmological dynamical system , and the dynamics at late times in a particular epoch by means of a _
future attractor_. in order to formulate the einstein field equations as a dynamical system it is clear that one has to normalize the variables , since near the initial singularity physical variables typically diverge and at late times typically tend to zero .
physical considerations suggest that in a cosmological setting , normalization with the hubble scalar of the fundamental congruence is an appropriate choice .
firstly , consider the density parameter @xmath17 , which plays a fundamental role in cosmology , in that it measures the influence of the matter on the dynamics . in geometrized units , it is defined to be the matter density @xmath18 divided by @xmath19 , where @xmath20 is the hubble scalar .
secondly , the extent to which the overall expansion of the universe is close to isotropy is measured by the ratio of the shear ( the trace - free part of the expansion tensor ) to the hubble scalar . the importance of hubble normalization appears to have first been mentioned in the literature by kristian & sachs 1966 ( see page 398 ) in connection with their general analysis of the potential constraints that observations of distant galaxies can place on the geometry of spacetime .
the role of hubble - normalized variables emerges again when one analyzes the potential constraints that arise from observations of the cmb ( maartens 1995 , 1996 ) . in both analyses
one finds that bounds are placed on hubble - normalized physical and geometrical quantities .
our principal interest , however , lies in using hubble - normalized variables as dynamical variables , which dates back to collins 1971 , who gave a qualitative analysis of the dynamics of some special classes of cosmological models .
working within the orthonormal frame formalism we introduce hubble - normalized variables and write the efe as first order evolution equations in the hubble - normalized state space . for simple classes of ever - expanding models , for example open fl models , and spatially homogeneous models of bianchi type i , the hubble - normalized state space is bounded .
near the singularity physical variables such as the matter density diverge , but the hubble scalar also diverges and at such a rate that the corresponding hubble - normalized quantity , the density parameter @xmath21 , remains bounded .
in contrast , if @xmath20 is close to zero , then the other physical variables are equally close to zero , so that the hubble - normalized quantities are bounded . for more general models , however , it turns out that this simple picture is not valid _ the hubble - normalized state space is unbounded_. the primary reason is that the hubble - normalized weyl curvature tensor can assume arbitrarily large values .
all available evidence , however , suggests that _ the hubble - normalized variables are bounded into the past _ , i.e. on approach to the initial singularity .
in addition , if there is a positive cosmological constant there is strong evidence that the hubble - normalized variables are also bounded into the future .
thus , even though the hubble - normalized state space is unbounded , one expects that the evolution equations will admit a past attractor and a future attractor .
when one formulates the efe as a dynamical system , one uses a time variable @xmath6 that potentially assumes all real values .
asymptotic regimes _ are then defined by the limits @xmath22 and @xmath23 .
cosmologists model the physical universe as a sequence of epochs in time , an epoch being identified by which source term is dynamically dominant . a typical succession of epochs is i ) inflationary , dominated by a scalar field @xmath24 , \ii ) radiation - dominated , @xmath25 , \iii ) matter - dominated , @xmath26 , \iv ) accelerating , dominated by a cosmological constant @xmath27 . with each source term
is associated a hubble - normalized energy density , @xmath28 and a particular epoch is defined as the time interval during which a particular @xmath29 is dominant . in analytic work
one typically assumes the presence of one or two source terms and one investigates the asymptotic regimes , which are described mathematically by the past attractor and future attractor of the dynamical system .
thus , for example , the past asymptotic regime of the radiation - dominated epoch would physically coincide with the end of the inflationary epoch and the future asymptotic regime would coincide with the beginning of the matter - dominated epoch .
a cosmological model is a spacetime with a preferred timelike vector field @xmath30 whose metric is a solution of the efe with appropriate matter / energy content .
it is assumed that there is an epoch during which the model is expanding , i.e. the preferred vector field satisfies @xmath31 where @xmath20 is referred to as the hubble scalar .
we assume furthermore that @xmath30 is the normal vector field to a family of spacelike hypersurfaces @xmath32 , thereby defining a time variable .
we introduce an orthonormal frame @xmath33 , with @xmath34 , and choose the spatial frame vectors @xmath35 to be fermi - propagated . within this framework ,
the gravitational field variables are the commutation functions @xmath36 associated with the frame , defined by @xmath37 = \gam^c{}_{ab}\ , \mathbf{e}_c.\ ] ] they may conveniently be expressed in terms of geometric quantities \
{ h , _ , _ , a_,n _ } , according to @xmath38 ( see for example we , page 32 ) .
here @xmath39 is the acceleration and @xmath40 is the shear of the congruence @xmath30 , while @xmath41 and @xmath42 determine the curvature of the spacelike hypersurfaces ( we , pages 18 , 19 , 34 ) . as regards matter / energy content we consider a cosmological constant @xmath43 and a perfect fluid with 4-velocity vector field @xmath44 and linear barotropic equation of state @xmath45 , with @xmath46 . in the orthonormal frame formalism the perfect fluid
is described by its energy density @xmath18 relative to @xmath47 and its velocity @xmath48 relative to @xmath47 , i.e. the projection of @xmath44 orthogonal to @xmath47 ( see uggla 2003 , equations ( 2.1)(2.6 ) for details ) .
so the gravitational variables ( [ geo ] ) are augmented by the matter variables \ { , v_,}. the velocity @xmath48 is dimensionless and satisfies @xmath49 we now introduce hubble - normalized variables and differential operators according to @xmath50 it is also necessary to introduce the derivatives of the normalization factor , namely the _ deceleration parameter _ @xmath51 and the _ spatial hubble gradient _ @xmath52 , defined by q+1 : = - 1h _ 0 h , r_:= - 1h _ h. at this stage we introduce the separable volume gauge ( uggla 2003 ) , which is characterized by @xmath53 and the existence of local coordinates @xmath6 , @xmath12 such that @xmath54 with the preferred hypersurfaces being given by @xmath55 . here @xmath56 and @xmath11 denote partial differentiation with respect to @xmath6 and @xmath12 .
the deceleration parameter @xmath51 can then be expressed in terms of the variables in ( [ hubnormvar ] ) and the differential operator @xmath57 , using the raychaudhuri equation . with the above choices of gauge
the hubble - normalized state vector becomes @xmath58 the efe , the jacobi identities and the commutators lead to a system of evolution equations and a system of constraints for the components of @xmath7 . in order to describe the structure of these equations , it is convenient to decompose the state vector @xmath7 as follows : = ( e_^i , r_)^t , where = ( _ , a_,n _ , _ m,_,v_)^t .
the evolution equations have the following form : @xmath59 where the deceleration parameter @xmath51 is given by q = 2 ^2 + 12 _ m -_- 13(e_^i _
i - 2a_)r^. there is also a set of constraints that can be written symbolically as 0 = ( , _ i ) .
one of the constraints , the so - called gauss constraint , is of particular importance , and we thus give it specifically : _
m + _ + _ k + ^2 = 1 , where @xmath60 is the hubble - normalized shear scalar given by ^2 = 16 _ ^ , and @xmath61 is the hubble - normalized spatial curvature is defined by @xmath62 , where @xmath63 is the curvature scalar of the metric induced on the hypersurfaces @xmath55 .
] of the hypersurfaces @xmath55 , given by _
k = -23(_-r _ ) a^ + 16 n _
n^ -112(n_^)^2 + a_a^. the detailed form of equations ( [ bsys])([esys ] ) is given in lim 2004 ( equations ( 2.18)(2.30 ) with @xmath64 ) .
what concerns us here is the overall structure of this system .
we make a number of observations .
* the equations are first order evolution equations in time , but do contain second order spatial derivatives , namely the second derivatives of the spatial hubble gradient @xmath65 .
these derivatives appear in ( [ evor ] ) , on account of ( [ q ] ) . in this respect ,
the equations are reminiscent of a system of quasi - linear diffusion equations .
we shall discuss this matter further in section [ sec_vac_num ] . *
the frame variables @xmath66 enter into the remaining equations only through the spatial differential operator @xmath57 . recalling that @xmath52 is the spatial hubble gradient we thus regard the variables @xmath66 and @xmath52 as controlling the spatial inhomogeneity in the cosmological model , while the variables in @xmath67 directly determine the spacetime geometry ( i.e. the gravitational field ) and the matter content .
we consider three classes of ever - expanding cosmological models , namely , friedmann - lematre ( fl ) models , spatially homogeneous ( sh ) models and general models , i.e. models without symmetry .
the sh models admit a three - parameter local group @xmath68 of isometries acting on spacelike hypersurfaces , while in the fl models the isometry group is a @xmath69 .
for brevity we shall refer to the models without symmetry as _
@xmath70 cosmologies_. so we have the following cosmological hierarchy : @xmath71 we will use @xmath72 to denote the corresponding hubble - normalized state spaces : @xmath73 one can construct a more detailed hierarchy by including models with two or one spacelike killing vectors . classes of models with two killing vectors are variously referred to as gowdy , @xmath74-symmetric or @xmath75 cosmologies
. models with one killing vector are referred to as @xmath76-symmetric or @xmath77 cosmologies .
the sh cosmologies are obtained by requiring that the spatial frame derivatives of the gravitational field and matter variables @xmath67 , and of the normalization factor @xmath20 be zero , i.e. _
= 0,r_=0 . it then follows that all the dimensional commutation functions and matter variables are constant on the hypersurface @xmath55 , which are thus the orbits of a three - parameter group @xmath68 of isometries .
the evolution equations ( [ evor ] ) and ( [ evoy ] ) imply that the sh restrictions ( [ shres ] ) define an invariant set of the full evolution equations , which we shall call the _ sh invariant set_. indeed , equation ( [ evor ] ) is trivially satisfied , and equation ( [ evoy ] ) reduces to a system of ordinary differential equations , namely _
t y_a = f_a ( ) .
the nontrivial constraints become purely algebraic restrictions on @xmath67 , which we write symbolically as ( ) = 0 . an important consequence of this specialization is that the evolution equation ( [ bsys ] ) for @xmath66 decouples from the evolution equation for @xmath67 , which means that _ the dynamics of sh cosmologies can be analyzed using only equations ( [ sh1 ] ) and ( [ sh2])_. in this context , one can think of the variable @xmath67 in equation ( [ y ] ) as defining a _ reduced hubble - normalized state space _ , of finite dimension , for the sh cosmologies .
the bianchi classification of the isometry group then leads to a hierarchy of invariant subsets . in the sh context
the restriction @xmath78 defines an invariant subset , giving the so - called _ nontilted sh cosmologies_. this class of models have been analyzed in detail in the literature .
we give an overview of the known results in section [ sec_sh ] .
we note in passing that tilted sh cosmologies ( i.e. @xmath79 ) were first studied by king & ellis 1973 .
considerable progress has been made recently , using hubble - normalized variables ( see hewitt 2001 , hervik 2004 , coley & hervik 2004 ) .
the fl cosmologies are obtained by imposing the restrictions @xmath80 which is equivalent to requiring that the fluid has zero shear , vorticity and acceleration .
it follows that the density parameters are constant on the hypersurface @xmath55 and the second from the definition of @xmath81 and @xmath52 . ] : @xmath82 in addition the weyl curvature is zero , @xmath83 and the 3-ricci curvature satisfies _ = 0 , _
_ k = 0 .
the dynamics of the fl models is governed by the matter evolution equations which simplify to a system of two odes , @xmath84\om_m \\
\lb{sh_ol } \ptl_t \om_\lambda & = 2(q+1 ) \om_\lambda,\end{aligned}\ ] ] with q = 12(3 - 2)_m -_. the gauss constraint ( [ cgauss ] ) simplifies to _
m + _ k + _ = 1 , where all three quantities are functions of time only . for future reference
we note that the curvature parameter @xmath61 satisfies the evolution equation _
k = 2q _ k , as follows from ( [ sh_om ] ) , ( [ sh_ol ] ) and ( [ sh_cgauss ] ) .
the remaining evolution equations simplify to @xmath85 and hence decouple from ( [ sh_om ] ) and ( [ sh_ol ] ) .
these variables reflect the choice of spatial frame , and do not affect the essential dynamics . since @xmath86 , these equations can be integrated to yield @xmath87 where hatted variables depend only on the spatial coordinates @xmath12 , and @xmath88 .
these quantities are restricted be equation ( [ sab0 ] )
. models with negative , zero and positive spatial curvature are characterized respectively by @xmath89 and in the corresponding canonical spatial frames we have @xmath90 thus , the fl cosmologies with perfect fluid and cosmological constant are described by a reduced hubble - normalized state space , namely the two - dimensional space with the state vector @xmath91 , and evolution equations ( [ sh_om])([sh_q ] ) .
it follows from ( [ sh_cgauss ] ) that for models with negative and zero spatial curvature ( @xmath92 ) , the state space is bounded .
this state space is shown in figure [ fig3.1 ] in section [ sec_c_i ] , using @xmath61 and @xmath81 as variables . in the analysis of the dynamics of sh models using the reduced hubble - normalized state space @xmath93 ,
the equilibrium points ( i.e. fixed points ) of the dynamical system , defined by @xmath94 naturally play a significant role . at this stage
the kasner vacuum solutions come into play , given in hubble - normalized variables by n_=0,a_=0,^2=1 , _
m=0,_=0,q=2 , with @xmath95 these equilibrium points ) , consists only of equilibrium points is a consequence of the fact that we are using a fermi - propagated spatial frame .
see uggla 2003 , page 7 , for a different spatial gauge choice , which leads to a circle of equilibrium points , and frame transition orbits " . ]
form a 4-sphere ( recall the definition ( [ sig_sq ] ) , and that @xmath96 is trace - free ) , which we shall call the _ kasner sphere _ @xmath97 . if @xmath96 is diagonal , then @xmath97 reduces to a circle .
two other important equilibrium points lie in the fl invariant set and satisfy @xmath98 and one of the following sets of conditions : .
see table [ tab1 ] . ]
@xmath99 the local stability of the flat fl and de sitter equilibrium points is of importance in connection with the phenomenon of isotropization , as discussed in section [ sec_iso ] . in the following tables we give the dimensions of the stable and unstable manifolds of these equilibrium points , for both nontilted and tilted sh models .
the sh equilibrium points , with the exception of de sitter , represent self - similar solutions of the efe , admitting a four - parameter group @xmath100 of similarities .
we refer to we , section 9.1 , for a complete list of these self - similar solutions within the class of nontilted sh models . on account of
( [ bsys ] ) and ( [ evor ] ) , the conditions e_^i=0,r_=0 define an invariant set , the so - called _ _ silent boundary _ _ , that forms part of the boundary of the hubble - normalized state space .
solutions of the evolution equations in this invariant set do not correspond to solutions of the efe , however , since the frame variables have to satisfy @xmath101 .
nevertheless , it appears that this invariant set plays a fundamental role in describing the asymptotic behaviour of cosmological models .
indeed current investigations suggest that the past attractor and the future attractor for @xmath70 cosmologies are subsets of the above invariant set ( subject to @xmath27 in the case of the future attractor ) .
we will discuss this matter further in sections [ sec_c_i ] and [ sec_vac_num ] .
the condition @xmath102 in the definition of the silent boundary implies that @xmath103 .
it thus follows that the evolution equations and constraints , when specialized to the silent boundary , are precisely the sh evolution equations and constraints , as given by ( [ sh1 ] ) and ( [ sh2 ] ) .
in addition each self - similar @xmath100 solution determines an equilibrium point in the silent boundary .
it is customary to regard a cosmological model as being close to fl "
if its metric is a linear perturbation of an fl metric ( e.g. bardeen 1980 ) , i.e. the dynamics of such models are governed by the _
linearized _ efe .
an alternative approach is to regard a cosmological model as being
close to fl " if an appropriate set of hubble - normalized anisotropy parameters are small . for simplicity we will restrict our attention to two parameters .
firstly we consider the shear parameter @xmath104 , defined by ( [ sig_sq ] ) , which describes the anisotropy in the rate of expansion of the fundamental congruence .
secondly we consider the weyl parameter @xmath105 defined by @xmath106 where @xmath107 and @xmath108 are the electric and magnetic parts of the weyl curvature tensor relative to the fundamental congruence .
the weyl parameter can be thought of as quantifying the intrinsic anisotropy of the gravitational field .
as described in section [ sec2.3 ] , the fl cosmologies with negative or zero spatial curvature and positive cosmological constant have a two - dimensional hubble - normalized state space . using @xmath61 and @xmath81 as independent variables
, the state space is shown in figure [ fig3.1 ] .
the equilibrium points represent the flat fl model ( f ) , de sitter model ( ds ) , and the milne model ( m ) .
one sees that f is the past attractor and ds is the future attractor .
the milne point m is a saddle , which means that a subset of open fl models with @xmath27 will have a
close - to - milne " epoch of finite duration .
( 120,0)(0,0 ) ( 41,64)(0,0)[b]@xmath81 ( 85,48)(0,0)[b]@xmath109 ( 16,47)(0,0)[b]flat fl , @xmath27 ( 87,4)(0,0)[b]open fl , @xmath110 ( 39,11)(0,0)[b]f ( 30,27)(0,0)[b]@xmath111 ( 38,53)(0,0)[b]ds ( 80,11)(0,0)[b]m ( 92,14)(0,0)[b]@xmath61 when one regards the fl models as a subset of the sh models , a complication arises : a specific fl model is represented by infinitely many orbits in the reduced hubble - normalized sh state space .
this multiple representation occurs because a flat fl model admits a @xmath68 of bianchi type i and one of bianchi type vii@xmath112 , while an open fl model admits a @xmath68 of bianchi type v and one of bianchi type vii@xmath113 , for any @xmath114 ( see ellis & maccallum 1969 ) .
the state space can be described by the variables @xmath115 where @xmath116 here @xmath117 is the group parameter and @xmath118 and @xmath119 are spatial curvature variables defined by @xmath120 the state space , which we will denote by @xmath121 , is shown in figure [ fig3.2 ] .
( 170,0)(0,0 ) ( 42,70)(0,0)[b]@xmath81 ( 115,63)(0,0)[b]@xmath122 , @xmath123 ( 102,72)(0,0)[b]@xmath109 , @xmath124 ( 80,64)(0,0)[b]@xmath109 ( 120,18)(0,0)[b]@xmath109 , @xmath123 ( 79,3)(0,0)[b]@xmath61 ( 39,20)(0,0)[b]f ( 38,58)(0,0)[b]ds ( 113,54)(0,0)[b]@xmath119 ( 82,15)(0,0)[b]m@xmath113 ( 61,6)(0,0)[b]m ( 101,25)(0,0)[b]m@xmath113 ( 120,0)(0,0 ) ( 41,62)(0,0)[b]@xmath81 ( 38,10)(0,0)[b]f ( 79,3)(0,0)[b]flat fl , @xmath110 ( 70,55)(0,0)[b]de sitter ( 33,23)(0,0)[b]@xmath111 ( 17,36)(0,0)[b]flat fl , @xmath27 ( 37,50)(0,0)[b]ds ( 93,11)(0,0)[b]@xmath119 we see that the state space is the union of a one - parameter family of leaves " , and that the milne model is represented by a line of equilibrium points @xmath125 , one in each leaf . the family of orbits in one leaf with @xmath114 are physically equivalent to those in another .
the orbits in the leaf with @xmath126 , and hence @xmath127 , are multiple representations of flat fl models with @xmath123 or @xmath128 . in particular
, the orbit with @xmath123 , @xmath127 is an unbounded orbit that describes the flat fl model , and the orbit with @xmath127 , @xmath109 is an unbounded orbit that describes the de sitter model .
the state space @xmath127 is shown in figure [ fig3.3 ] .
the fl state space @xmath121 is an invariant subset of the sh state space @xmath93 , defined by the requirement that the shear of the congruence normal to the @xmath68 orbits is zero .
it is natural to regard an sh model as being close to fl during some epoch if its orbit lies in a sufficiently small neighbourhood of the fl state space @xmath121 .
the hubble - normalized weyl tensor is unbounded in such a neighbourhood , however , with the result that close - to - fl models can have large hubble - normalized weyl curvature .
in particular there are models with @xmath129 , @xmath130 and @xmath105 large ( close to flat fl ) and models with @xmath131 , @xmath130 and @xmath105 large ( close to de sitter ) , during some epoch .
it is also of interest to consider the connection with perturbed fl models .
are orbits in a neighbourhood of @xmath121 described accurately by solutions of the linearized efe , or do nonlinear effects come into play ?
consideration of sh cosmologies of bianchi type vii@xmath112 shows that the latter occurs .
linearizing the hubble - normalized evolution equations about the fl invariant set @xmath123 and @xmath127 gives @xmath132 while an analysis of the exact equations for large @xmath133 gives @xmath134 ( wainwright 1999 ) .
thus for this class of models , linearization does not give the correct dynamics .
it would be of interest to extend this analysis to @xmath70 cosmologies , i.e. consider @xmath121 as an invariant set of the state space of @xmath70 cosmologies and analyze the dynamics of close - to - fl cosmologies in this context .
it would be necessary to determine all representations of the fl models within the @xmath70 state space , which would give an invariant set larger than @xmath121 . the term isotropization " is used in three different contexts in cosmology : * asymptotic isotropization into the future , * isotropic initial singularity , * intermediate isotropization .
we discuss these dynamic phenomena in turn . a well known result of wald 1983 states that any sh cosmology with zero or negative spatial curvature ( i.e. @xmath92 ) approaches the de sitter solution in the sense that @xmath135 wald proved this result without making specific assumptions about the matter / energy distribution . more detailed analyses of sh cosmologies with a tilted perfect fluid ( goliath & ellis 1999 ) revealed that a bifurcation occurs at @xmath136 : if @xmath137 the peculiar velocity @xmath48 of the fluid does not tend to zero .
specifically , @xmath138 , where @xmath139 if @xmath140 and @xmath141 if @xmath142 .
this behaviour occurs for the following reason . for the de sitter solution the reduced state vector @xmath67 in ( [ y ] )
is given by n_=0,a_=0,_=0 , _
m=0,_=1 , with @xmath143 subject to the evolution equation @xmath144 in the reduced hubble - normalized state space the de sitter solution is thus represented by a set of equilibrium points , given by ( [ ds_full ] ) and with @xmath143 subject to @xmath145 , or @xmath146 ( a point and a sphere ) if @xmath147 , and subject to @xmath148 ( a solid sphere ) if @xmath140 .
the limiting behaviour of @xmath48 determines the future attractor in the reduced hubble - normalized state space , as shown in table [ tab1 ] .
the future attractor for sh cosmologies with @xmath27 forms the basis for the description of @xmath70 cosmologies that are future asymptotic to the de sitter solution , as we now explain .
the frame variables @xmath66 satisfy equation ( [ bsys ] ) , which , when specialized to the de sitter equilibrium solution ( [ ds_full ] ) can be solved to give @xmath149 where the @xmath150 are constants .
it follows that _ t+ e_^i = 0 .
in addition the constraints imply that the spatial hubble gradient is zero ( @xmath151 ) .
thus in the hubble - normalized @xmath70 state space , the de sitter solution is described by orbits that are future asymptotic to a de sitter equilibrium point on the silent boundary .
@xmath70 cosmologies that are future asymptotic to the de sitter solution are described within the hubble - normalized state space as follows . along a fixed timeline ,
the evolution is described by an orbit that is future asymptotic to a de sitter equilibrium point on the silent boundary .
the specific equilibrium point will depend on the equation of state parameter @xmath152 according to table [ tab1 ] .
the idea is that the evolution is described by an orbit that is attracted to the silent boundary , i.e. ( [ ds_silent ] ) holds , and within the silent boundary , the evolution is governed by the sh evolution equations , as described in section [ sec_silent_b ] .
the asymptotic dependence of the hubble - normalized variables for this class of models has been determined by systematically integrating the evolution equations ( lim 2004 ) .
the fact that the orbit is asymptotic to the silent boundary means that the spatial dependence is asymptotically unrestricted , which manifests itself in the appearance of an arbitrary 3-metric in the asymptotic expansion of the spacetime metric ( lim 2004 ) .
the issue of how large is the class of @xmath70 cosmologies that are future asymptotic to de sitter has not been completely resolved . for recent progress
we refer to rendall 2003 .
the second type of asymptotic isotropization , namely isotropization at the initial singularity , arises in connection with the notion of _ quiescent cosmology _ ( cf . ,
e.g. , barrow 1978 ) , which provides an alternative to cosmic inflation .
the idea is that , due to entropy considerations on a cosmological scale , a suitable initial condition for the universe is that the weyl curvature should be zero ( or at least dynamically unimportant ) at the initial singularity ; this is the _ weyl curvature hypothesis _ , proposed by penrose 1979 ( see page 630 ) .
this hypothesis leads to the notion of an _ isotropic initial singularity_. recently , khalatnikov 2003 ( see page 3 ) have suggested that a solution with an isotropic singularity could represent an _ intermediate _
asymptotic regime during expansion of the universe after inflation .
cosmological initial singularities of this type first arose as a special case in the general analysis of initial singularities performed by lifshitz & khalatnikov 1963 ( see page 203 ) .
this type of initial singularity was also encountered in the work of eardley 1972 ( see page 101 ) .
subsequently , goode & wainwright 1985 gave a formal definition of an isotropic initial singularity , using a conformally related metric , and derived various properties .
heuristically , the model approaches the flat fl model near the singularity : @xmath153 for sh models , the occurrence of an isotropic singularity is characterized by the orbit in the reduced hubble - normalized state space being past asymptotic to the flat fl equilibrium point f. in other words , the orbit of a sh model with an isotropic singularity lies in the unstable manifold of the flat fl equilibrium point , which confirms that isotropic singularities occur only for a set of initial conditions of measure zero .
the above behaviour forms the basis for the description of @xmath70 cosmologies that admit an isotropic singularity , as we now explain .
the frame variables @xmath66 satisfy equation ( [ bsys ] ) , which , when specialized to the flat fl equilibrium point can be solved to give @xmath154 where the @xmath150 are constants .
it follows that @xmath155 assuming @xmath156 .
in addition , the constraints imply that the spatial hubble gradient is zero ( @xmath151 ) .
thus in the hubble - normalized @xmath70 state space , the flat fl model is described by an orbit that is past asymptotic to an equilibrium point on the silent boundary .
@xmath70 cosmologies that have an isotropic singularity are described as follows . along a fixed timeline , the evolution
is described by an orbit that is past asymptotic to the flat fl equilibrium point on the silent boundary .
the asymptotic time dependence for this class of models has been determined by systematically integrating the evolution equations ( lim 2004 ) .
the fact that the orbit is past asymptotic to the silent boundary means that the spatial dependence is asymptotically unrestricted , which manifests itself in the appearance of an arbitrary 3-metric in the asymptotic expansion of the spacetime metric ( see lim 2004 ) .
the flat fl equilibrium point f , given by ( [ fl ] ) , is a saddle point in the reduced hubble - normalized state space for sh cosmologies .
thus , for any @xmath157 there is a family of orbits , corresponding to an open set of initial conditions , that pass through an @xmath158-neighbourhood of f but are not asymptotic to f. for the corresponding cosmological models there will be a finite time interval during which the rate of expansion of the model is highly isotropic ( i.e. @xmath159 ) and hence compatible with observational constraints .
this behaviour is called _ intermediate isotropization _ , and occurs in models of all bianchi types except bianchi type i , in which case f is a sink .. ] at this stage , intermediate isotropization for @xmath70 cosmologies has yet to be investigated in detail .
the largest class of cosmologies for which there is a comprehensive and quite detailed knowledge of the dynamics , is the class of nontilted sh cosmologies with @xmath160 and equation of state @xmath161 , with @xmath156 .
the reduced hubble - normalized state space @xmath93 is an unbounded subset of @xmath162 .
this class of cosmologies has two subsets referred to as class a and class b in the classification of ellis & maccallum 1969 .
we focus on the class a models , since they display a wider range of dynamical phenomena , while at the same time permitting a simpler choice of frame . in describing the dynamics of this class of cosmologies ,
it is helpful to make use of the hierarchy of invariant subsets of lower dimension that arise from specializing the matter / energy content and the bianchi type , vii@xmath112 , viii and ix , with the last two being of maximum generality .
we exclude bianchi ix because models of this type can recollapse . ] .
we list these invariant subsets in table [ tab_inv ]
. we will discuss various dynamical phenomena , indicating how likely it is that each will occur .
we say that a dynamical behaviour is * _ generic _ , if it occurs for all initial conditions except for a set of measure zero , * _ typical _ , if it occurs for a set of initial conditions of positive measure , whose complement is also of positive measure .
* _ special _ , it it occurs for a set of initial conditions of measure zero .
one can say that generic behaviour has probability one , typical behavior has probability between zero and one , and special behaviour has probability zero .
we now list the various dynamical phenomena in table [ tab_phe ] and discuss each in turn . within @xmath93
, the mixmaster dynamics is described by a past attractor , which is the union of the kasner circle of equilibrium points and the vacuum bianchi ii orbits , first described by ma & wainwright 1992 .
the attractor contains infinite heteroclinic sequences , i.e. infinite sequences of kasner points joined by vacuum bianchi ii orbits .
a generic orbit that is asymptotic to the attractor shadows one of these heteroclinic sequences .
the evolution of the corresponding cosmological model towards the initial singularity is thus a non - terminating sequence of quasi - equilibrium kasner states , punctuated by increasingly brief curvature transitions .
the early work on the attractor was based solely on heuristic local stability arguments and numerical simulations .
recently , however , ringstrm 2001 proved the existence of the past attractor , using the hubble - normalized evolution equations .
this result may have wider significance in view of the conjectures about the local past attractor for @xmath70 cosmologies that have recently been made ( uggla 2003 and section [ sec_vac_num ] ) . as indicated in table [ tab_phe ] ,
mixmaster dynamics are generic .
examples of special behaviour are provided by models which are past asymptotic to a specific kasner solution and by models with an isotropic singularity . within @xmath93 , inflationary isotropization
is simply described by the fact that the de sitter equilibrium point ( [ ds ] ) , which describes the de sitter solution , is a global sink . ] , and hence forms the future attractor in the state space .
this behaviour is generic , and special behaviour is restricted to the invariant set @xmath123 .
the dimensionless scalars satisfy @xmath163 the asymptotic time dependence can be obtained from lim 2004 and rendall 2003 .
generic orbits in the invariant subset @xmath123 escape to infinity , and satisfy _
this result holds for bianchi viii orbits if the equation of state satisfies @xmath164 ( horwood & wainwright 2004 ) and for bianchi vii@xmath112 orbits of @xmath165 ( wainwright 1999 ) .
in other words the weyl curvature is dynamically dominant at late times . from a mathematical perspective
this result means that the invariant subset @xmath123 does not admit a future attractor . despite this fact
, one can still determine the asymptotic dependence of the state vector as @xmath166 ( horwood & wainwright 2004 ) .
the asymptotic behaviour in the invariant set @xmath123 as @xmath166 , as epitomized by ( [ weyl_infty ] ) , has a significant effect on the dynamics in the full state space @xmath128 , in that _ it approximates the intermediate dynamics _ of a typical class of models with @xmath128 .
one can think of the asymptotic state of models with @xmath123 acting as a _
generalized saddle _ in the full state space , temporarily attracting orbits but eventually repelling them . in summary , for a typical class of models with @xmath128 there will be a finite epoch during which weyl dominance occurs ( i.e. @xmath167 ) .
a cosmological model undergoes intermediate isotropization with @xmath169 if there is a finite epoch during which the model is close to the flat fl model .
this behaviour is created by the local stability properties of the flat fl model .
we have seen that the flat fl model has a multiple representation , as the flat fl equilibrium point f and as an unbounded orbit ( see section [ sec_close_to__fl ] , in particular , figure [ fig3.3 ] ) .
orbits that are future asymptotic to f ( i.e. orbits that form the two dimensional stable manifold of f ) satisfy _
t+=0 , _ t+=0 . on the other hand orbits that are future asymptotic to the unbounded orbit
satisfy invariant set with @xmath123 . ] _
t+= l & + + & we say that models that satisfy ( [ lim_fl ] ) are _ asymptotic to flat fl in the strong sense _ , while those satisfy ( [ lim_fl_infty ] ) are _ asymptotic to flat fl in the weak sense_. both of these asymptotic states can act as ( generalized ) saddles in the full state space , leading to intermediate isotropization with @xmath168 .
thus intermediate isotropization with @xmath168 in the full state space is a result of the occurrence of asymptotic isotropization with @xmath170 in lower dimensional invariant sets .
( 120,0)(0,0 ) ( 27,75)(0,0)[b]@xmath81 ( 23,25)(0,0)[b]f ( 7,1)(0,0)[b]@xmath97 ( 22,64)(0,0)[b]ds ( 117,24)(0,0)[b]@xmath119 intermediate isotropization in the weak sense with @xmath168 is illustrated schematically in figure [ fig_saddle2 ] .
an orbit that is past asymptotic to a kasner equilibrium point @xmath97 spirals around the flat fl orbit , creating an epoch of intermediate isotropization with @xmath168 , before approaching the de sitter equilibrium point . within @xmath93 , a cosmological model with an isotropic singularity
is simply described by an orbit that is past asymptotic to the flat fl equilibrium point ( [ fl ] ) , i.e. an orbit that lies in the unstable manifold of this equilibrium point . recall that this manifold is of dimension 4 ( see table [ tab_dim1 ] ) .
it follows that @xmath171 for these models .
the asymptotic time dependence can be obtained from lim 2004 .
we now describe the attractor in the hubble - normalized @xmath70 state space , following the discussion in uggla 2003 , but using a different choice of gauge .
one of the goals of that paper was to give a precise statement of the so - called bkl conjecture , by describing the past attractor in the hubble - normalized state space .
_ the bkl conjecture .
_ for almost all cosmological solutions of einstein s field equations , a spacelike initial singularity is _ vacuum - dominated , local and oscillatory_. for cosmological models with a perfect fluid matter source , the phrase vacuum - dominated , " or , equivalently ,
matter is not dynamically significant , " is taken to mean that the hubble - normalized matter density @xmath172 tends to zero at the initial singularity .
one might then expect that the past attractor for cosmological models with a perfect fluid matter source would be the same as the past attractor for an idealized vacuum cosmological model .
it turns out ( uggla 2003 ) that this expectation is unwarranted the peculiar velocity @xmath143 plays a role in determining the past attractor .
nevertheless , partly in the interests of simplicity and also because the numerical simulations that have been recently performed ( garfinkle 2003 ) have been for vacuum models , we shall restrict our discussion here to the vacuum case . in terms of hubble - normalized variables and the separable volume gauge , the spacelike initial singularity in a @xmath70 cosmology
is approached as @xmath173 .
we now define a _ silent initial singularity _ to be one which satisfies @xmath174 more precisely , we require that ( [ silent ] ) and ( [ silent2 ] ) are satisfied along typical timelines of @xmath47
. one might initially think that the condition ( [ silent2 ] ) is a consequence of ( [ silent ] ) , since @xmath175 however the analysis of gowdy solutions with so - called spikes shows that the partial derivatives @xmath176 can diverge as @xmath173 .
thus the requirement ( [ silent2 ] ) demands that @xmath66 tend to zero sufficiently fast .
the asymptotic analysis of the gowdy solutions shows that ( [ silent2 ] ) is satisfied along typical timelines even when spikes occur ( see uggla 2003 , section iv.a ) . in more general @xmath75 cosmologies , however , the validity of ( [ silent2 ] ) is still open and further work is needed .
we now refer to uggla 2003 , pages 910 , for further evidence to justify making the following conjecture . * conjecture 1 * : for almost all cosmological solutions of einstein s field equations , a spacelike initial singularity is silent .
proving this conjecture entails establishing the limits ( [ silent])([silent2 ] ) .
we think of the evolution of the hubble - normalized state vector @xmath177 _ for fixed @xmath12 _ , as being described by an orbit in a finite - dimensional hubble - normalized state space . as @xmath173 , this orbit will be asymptotic to a local past attractor @xmath178 , which in accordance with the definition of silent initial singularity , will be contained in the silent boundary , defined by ( [ silent_cond ] ) .
the past attractor is based on the kasner solutions .
we have seen that in the reduced hubble - normalized state space for sh models , the kasner solutions are represented by a sphere of equilibrium points . in the full @xmath70 state space , these solutions
are represented by orbits that are past asymptotic to kasner equilibrium points on the silent boundary .
thus we focus our attention on the _ kasner sphere @xmath97 on the silent boundary _ , defined by @xmath179 in addition to ( [ silent_cond ] ) . each non - exceptional , where @xmath180 denotes tracefree symmetrization .
] equilibrium point on the kasner sphere has one negative eigenvalue and hence has a one - dimensional unstable manifold into the past , which represents a sh vacuum solution of bianchi type ii on the silent boundary .
we refer to these orbits as _ curvature transitions _ , since on each such orbit one degree of freedom of the spatial curvature is activated . that one diagonal component of @xmath181 is unstable into the past . ]
these curvature transitions satisfy @xmath182 corresponding to the fact that two eigenvalues of @xmath181 are zero .
in addition @xmath183 and the gauss constraint simplifies to @xmath184 we now make our second conjecture , motivated by our knowledge of sh dynamics . * conjecture 2 * : the local past attractor @xmath178 for vacuum @xmath70 cosmologies with a silent initial singularity is @xmath185 where @xmath97 is the kasner sphere and @xmath186 is the set of all curvature transitions in the silent boundary . establishing this conjecture entails proving the following limits : @xmath187=0.\end{gathered}\ ] ] the fact that the sequence of transitions is non - terminating implies that @xmath188 do not exist .
however it follows from the gauss constraint ( [ cgauss ] ) and equations ( [ silent ] ) , ( [ silent2 ] ) and ( [ n2 ] ) that @xmath189=1.\ ] ] in addition it follows from one of the other constraints in ( [ esys ] ) , the codacci constraint , and the limits ( [ silent ] ) , ( [ silent2 ] ) and ( [ n2 ] ) that _
t- ^ n _ _ ^=0 .
we thus expect that in a shear eigenframe , @xmath181 will be asymptotically diagonal " as @xmath14 .
the geometrical interpretation of the asymptotic evolution is as follows : the orbit describing the evolution of the state vector @xmath177 for fixed @xmath12 approaches the kasner sphere @xmath97 in the silent boundary and then shadows increasingly closely an infinite sequence of curvature transitions joining kasner equilibrium points . some numerical simulations of the past asymptotic behaviour of vacuum @xmath70 cosmologies using the hubble - normalized evolution equations ( [ bsys])([esys ] ) have been recently performed by garfinkle 2003 .
he found it convenient to use the deceleration parameter as a dynamical variable , with the defining equation ( [ q ] ) forming an additional constraint .
the evolution equation for @xmath51 has the form of a quasilinear diffusion equation , and it is the only evolution equation to contain second order spatial derivatives .
the hubble - normalized evolution equations were solved numerically on a 3-torus , i.e. with periodic boundary conditions , using a crank - nicholson scheme .
the initial conditions were chosen to be of the form @xmath190 where @xmath41 , @xmath191 are arbitrary constants .
the gauss constraint gives a nonlinear elliptic pde for @xmath192 : @xmath193\ , \ ] ] where @xmath194 is the euclidean laplacian in @xmath195 , @xmath196 and @xmath197 .
this pde has to be solved numerically on a 3-torus to obtain explicit initial conditions .
the remaining constraints are satisfied identically .
note that the above choice of @xmath66 corresponds to a conformally flat initial 3-metric .
the numerical simulations show a brief transient epoch after which a sequence of curvature transitions takes place , following the usual bkl transition law , which we now describe .
the kasner exponents @xmath198 are related to the shear variables @xmath96 in a shear eigenframe as follows : @xmath199 the exponents satisfy @xmath200 and thus can be expressed in terms of a single parameter @xmath201 according to @xmath202 the transition law between kasner states can then be written in the form @xmath203 the numerical simulations also suggest that asymptotically @xmath96 and @xmath181 have a common eigenframe in accordance with ( [ eigen ] ) .
the figures show the results of a numerical simulation with @xmath204 initially , and then becoming negative , i.e. evolution towards the initial singularity ( @xmath22 ) .
figures [ fig_u1]a and [ fig_u2]a show the transitions of the kasner parameter @xmath201 along two different timelines and figures [ fig_u1]b and [ fig_u2]b show the diagonal values of @xmath181 in the common asymptotic eigenframe , along the same timelines .
overall , the simulations provide support for the limits ( [ silent ] ) and ( [ silent2 ] ) , and ( [ a])([eigen ] ) along typical timelines .
the numerical simulation is incomplete in two ways .
first , the duration of the kasner epochs in the simulation is far too short , due to the fact that the decay of the eigenvalues of @xmath181 to increasingly small values , which determines when the curvature transitions occur , is not described sufficiently accurately .
the reason for this difficulty is that an arbitrarily chosen fermi - propagated frame is not asymptotic to the eigenframe of @xmath181 as @xmath22 .
second , the simulation does not have high enough numerical resolution to correctly simulate spiky structures .
these structures are known to develop in @xmath75 cosmologies , on approach to the singularity ( berger & moncrief 1993 ) , and we expect that similar structure will develop in @xmath70 cosmologies . from our experience with numerical simulations of @xmath75 cosmologies ,
these spiky structures occur within the particle horizon of timelines where one of the eigenvalues of @xmath181 crosses zero ( lim 2004 ) .
as the particle horizon shrinks into the past , increasingly high numerical resolution is needed to simulate the spiky structures .
it may be prohibitively expensive to perform such simulations in @xmath70 cosmologies .
the principal deficiency of hubble - normalized variables is that they break down in an expanding cosmological model that reaches a maximum state of expansion ( @xmath205 ) , as is possible in an fl model with positive spatial curvature .
this difficulty can be addressed , at least for fl models and sh models of bianchi type ix , by using a modified normalization factor first proposed by uggla ( we section 8.5.2 ) . in the interests of simplicity
we introduce the modified normalization procedure for the class of fl models with positive spatial curvature . for these models
the normalization factor is defined by @xmath206 where @xmath63 is the curvature scalar of the metric induced on the hypersurfaces @xmath55 . since @xmath207 ,
this factor is positive throughout the evolution , including at the time of maximum expansion @xmath205 .
dimensionless variables are defined in the usual way , using @xmath208 instead of @xmath20 : @xmath209 it follows that @xmath210 introducing a time variable @xmath211 according to @xmath212 one obtains the following evolution equations for @xmath213 and @xmath214 : @xmath215 where @xmath216 we note that @xmath217 is related to the usual deceleration parameter @xmath51 according to @xmath218 whenever @xmath219 .
the state space is bounded , being defined by the inequalities @xmath220 and is shown in figure [ fig5.1 ] .
the sign of @xmath213 determines whether the model is expanding ( @xmath221 ) or collapsing ( @xmath222 ) .
the sign of @xmath217 determines whether the model is decelerating ( @xmath223 ) or accelerating .
the fixed points of the evolution equations are as follows : @xmath224 the points @xmath225 and @xmath226 represent expanding models ( @xmath227 ) while @xmath228 and @xmath229 represent the time reversed models ( @xmath230 ) . the orbits @xmath231 represent models that expand from a big - bang singularity and then recollapse to a future singularity .
the orbits @xmath232 represent models that expand indefinitely from a big - bang singularity , enter an accelerating epoch , and approach de sitter at late times .
( 120,0)(0,0 ) ( 95,70)(0,0)[b]@xmath226 ( 17,70)(0,0)[b]@xmath229 ( 35,70)(0,0)[b]contracting ( 75,70)(0,0)[b]expanding ( 53,33)(0,0)[b]e ( 105,28)(0,0)[b]@xmath233 ( 110,45)(0,0)[b]accelerating ( 110,20)(0,0)[b]decelerating ( 55,85)(0,0)[b]@xmath214 ( 55,0)(0,0)[b]@xmath234 ( 95,5)(0,0)[b]@xmath225 ( 95,0)(0,0)[b](flat fl , expanding ) ( 17,5)(0,0)[b]@xmath228 ( 15,0)(0,0)[b](flat fl , contracting ) ( 110,9)(0,0)[b]@xmath213 figure [ fig5.1 ] illustrates the central role played by the einstein static model within the class of fl cosmologies with positive and zero spatial curvature .
the fixed point e is a saddle point whose stable and unstable manifolds are one - dimensional .
the unstable manifold describes a one - parameter family of cosmologies that are past asymptotic to the einstein static model , and hence do not have an initial singularity .
these cosmological models have recently been proposed by ellis & maartens 2004 as a possible description of the early universe , a so - called emergent universe .
we note that this normalization procedure can be generalized to the class of nontilted sh cosmologies of bianchi type ix .
the normalization factor is ( we , equation ( 8.6 ) ) @xmath235 where @xmath236 . the resulting state space , which gives a unified description of the dynamics of nontilted sh cosmologies of bianchi type ix , vii@xmath112 , ii and i , has not been explored in detail . in addition , a variation of this normalization has been given by heinzle 2004 , and has been used to give a detailed analysis of the nontilted sh cosmologies of bianchi type ix that are locally rotationally symmetric .
the dynamical systems approach has two key attributes .
firstly , it provides a geometric framework for analyzing the space of cosmological models .
secondly , it provides a system of first order evolution equations , whose specific form depends on the choice of gauge .
this system of equations is potentially useful for giving a rigorous analysis of the asymptotic dynamics of cosmological models , and also for performing numerical simulations .
the utility of this approach has been amply demonstrated as regards the sh cosmologies , and there are indications that it may be equally useful as regards the @xmath70 cosmologies .
this work has been influenced significantly by our collaboration with claes uggla and henk van elst .
it is a pleasure to acknowledge the many discussions that we have had over the past few years .
we also thank george ellis for his continued interest and encouragement , and david garfinkle for providing figures [ fig_u1 ] and [ fig_u2 ] , and for helpful discussions on the numerical simulations .
j.w . expresses his appreciation to lars andersson and gregory galloway and the department of mathematics at the university of miami for their hospitality during the miami waves conference in january 2004 .
this work was partially supported by a grant to j.w . from the natural sciences and engineering research council of canada .
99 l. andersson , h. van elst & c. uggla 2004 , gowdy phenomenology in scale - invariant variables , cqg * 21 * , s29s57 j.m .
bardeen 1980 , gauge invariant cosmological perturbations , prd * 22 * , 1882905 j.d .
barrow 1978 , quiescent cosmology , _ nature _ * 272 * , 211 b.k .
berger 2004 , hunting local mixmaster dynamics in spatially inhomogneous cosmologies , cqg * 21 * , s81s96 b.k .
berger and v. moncrief 1993 , numerical investigation of cosmological singularities , prd * 48 * , 467687 a.a .
coley 2003 , _ dynamical systems and cosmology _ , kluwer academic publishers a.a .
coley , he & lim 2004 , isotropic singularity in inhomogeneous brane cosmological models , cqg * 21 * , 131142 a.a .
coley & s. hervik 2004 , bianchi cosmologies : a tale of two tilted fluids , cqg * 21 * , 4193208 a.a . coley & r.j .
van den hoogen 2000 , the dynamics of multi - scalar field cosmological models and assisted inflation , prd * 62 * , 023517 c.b .
collins 1971 , more qualitative cosmology , _ commun . math .
* 23 * , 13756 d. eardley , e.p .
liang & r.k .
sachs 1972 , velocity - dominated singularities in irrotational dust cosmologies , _ j. math
_ * 13 * , 99107 g.f.r .
ellis 1967 , dynamics of pressure - free matter in general relativity , _ j. math .
* 8 * , 117194 g.f.r .
ellis 2004 , the space of cosmological space - times , _ preprint _ g.f.r .
ellis & m. bruni 1989 , covariant and gauge - invariant approach to cosmological density fluctuations , prd * 40 * , 180418 g.f.r . ellis , j. hwang & m. bruni 1989 , covariant and gauge - independent perfect - fluid robertson - walker perturbations , prd * 40 * , 181926 g.f.r .
ellis & r. maartens 2004 , the emergent universe : inflationary cosmology with no singularity , cqg * 21 * 22332 g.f.r .
ellis & m.a.h .
maccallum 1969 , a class of homogeneous cosmological models , _ commun .
* 12 * , 10841 h. van elst & c. uggla 1997 , general relativistic 1 + 3 orthonormal frame approach , cqg * 14 * , 2673951 h. van elst , c. uggla & j. wainwright 2002 , dynamical systems approach to @xmath75 cosmology , cqg * 19 * , 5182 d. garfinkle 2003 , numerical simulations of generic singularities , gr - qc/0312117 d. garfinkle & m. weaver 2003 , high velocity spikes in gowdy spacetimes , prd * 67 * 124009 m. goliath & g.f.r .
ellis 1999 , homogeneous cosmologies with a cosmological constant , prd * 60 * , 023502 s.w .
goode & j. wainwright 1985 , isotropic singularities in cosmological models , cqg * 2 * , 99115 j.m .
heinzle , n. rohr & c. uggla 2004 , matter and dynamics in closed cosmologies , gr - qc/0406072 s. hervik 2004 , the asymptotic behaviour of tilted bianchi type vi@xmath112 universes , cqg * 21 * , 230117 c.g .
hewitt , r. bridson & j. wainwright 2001 , the asymptotic regimes of tilted bianchi ii cosmologies , grg * 33 * , 6594 d. hobill , a. burd & a.a .
coley ( ed ) 1994 , _ deterministic chaos in general relativity _ , plenum press j.t .
horwood & j. wainwright 2004 , asymptotic regimes of magnetic bianchi cosmologies , grg * 36 * , 799823 i.m .
khalatnikov , a.yu .
kamenshchik , m. martellini , a.a .
starobinsky 2003 , quasi - isotropic solution of the einstein equations near a cosmological singularity for a two - fluid cosmological model , _ journal of cosmology and astroparticle physics _ * 03 * , 001 a.r . king & g.f.r .
ellis 1973 , tilted homogeneous cosmological models , _ commun . math .
phys . _ * 31 * , 20942 j. kristian & r.k .
sachs 1966 , observations in cosmology , apj * 143 * , 37999 v.g .
leblanc 1998 , bianchi ii magnetic cosmologies , cqg * 15 * , 160726 e.m .
lifshitz & i.m .
khalatnikov 1963 , investigations in relativistic cosmology , _ adv .
* 12 * , 185249 w.c .
lim 2004 , _ the dynamics of inhomogeneous cosmologies _ , phd thesis , university of waterloo w.c .
lim , h. van elst , c. uggla & j. wainwright 2004 , asymptotic isotropization in inhomogeneous cosmology , prd * 69 * , 103507 r. maartens , g.f.r .
ellis & w.r .
stoeger 1995 , limits on anisotropy and inhomogeneity from the cosmic background radiation , prd * 51 * , 152535 r. maartens , g.f.r .
ellis & w.r .
stoeger 1996 , anisotropy and inhomogeneity of the universe from @xmath237 , _ astron .
astrophys . _ * 309 * , l7l10 r. penrose 1979 , singularities and time - asymmetry . in _ general relativity : an einstein centenary survey _
, eds s.w .
hawking & w. israel .
cambridge university press .
rendall 2003 , asymptotics of solutions of the einstein equations with positive cosmological constant , gr - qc/0312020 a.d .
rendall & m. weaver 2001 , manufacture of gowdy spacetimes with spikes , cqg * 18 * , 295976 h. ringstrm 2001 , the bianchi ix attractor , _
annales henri poincar _ * 2 * , 405500 c. uggla , h. van elst , j. wainwright & g.f.r .
ellis 2003 , past attractor in inhomogeneous cosmology , prd * 68 * , 103502 j. wainwright , m.j .
hancock & c. uggla 1999 , asymptotic self - similarity breaking at late times in cosmology , cqg * 16 * , 257798 j. wainwright & g.f.r .
ellis ( ed ) 1997 , _ dynamical systems in cosmology _ , cambridge university press r.m .
wald 1983 , asymptotic behaviour of homogeneous cosmological models in the presence of a positive cosmological constant , prd * 28 * , 211820 | it is useful to study the space of all cosmological models from a dynamical systems perspective , that is , by formulating the einstein field equations as a dynamical system using appropriately normalized variables .
we will discuss various aspects of this work , the choices of normalization factor , multiple representations of models , the past attractor , nonlinear dynamics in close - to - friedmann - lematre models , weyl curvature dominance , and numerical simulations . 1 * cosmological models from a dynamical systems perspective * + j. wainwright@xmath0 and w.c . lim@xmath1 + @xmath0 department of applied mathematics , university of waterloo , + waterloo , ontario , canada n2l 3g1 + @xmath2 department of mathematics and statistics , dalhousie university , + halifax , nova scotia , canada b3h 3j5 + [ 2 mm ] email : [email protected] , [email protected] + |
during the last decade of the past century there has been a vivid discussion about the opacity of spiral galaxies .
this discussion was initiated by disney , davies & phillipps ( 1989 ) and valentijn ( 1990 ) , who countered the conventional view that spiral galaxies are optically thin over their entire optical discs ( holmberg 1958 ; de vaucouleurs , de vaucouleurs & corwin 1976 ; sandage & tammann 1981 ) .
this issue has not yet been settled , although an overall consensus seems to be emerging that spiral galaxies are , in general , at least moderately optically thick within their optical disc , i.e. that they have central face - on optical depth of at least unity .
we will not repeat a detailed overview of the different points of view and the large variety of observational tests , and refer to davies & burstein ( 1995 ) , xilouris et al .
( 1997 ) , kuchinski et al .
( 1998 ) and calzetti ( 2001 ) for excellent reviews .
we just wish to point out an important observational constraint which , in our opinion , is not always given enough attention : the far - infrared ( fir ) emission of spiral galaxies .
this emission originates from the absorption of starlight by dust grains , which is thermally re - emitted at longer wavelengths .
the most recent results show that about 30 per cent of the bolometric luminosity of late - type galaxies is fir emission by interstellar dust ( popescu & tuffs 2002 ) .
notice that this figure refers to normal quiescent spiral galaxies ; special classes of galaxies such as starburst galaxies emit even higher fractions of their energy at fir wavelengths .
these facts demonstrate quite convincingly that interstellar dust forms an important constituent of spiral galaxies .
the physical processes of absorption and scattering by dust grains should hence be taken into account when interpreting and modelling the observed optical properties of galaxies .
we would like to stress that _ all _ optical observables will to some extent be affected by dust , including the observed kinematics .
indeed , when interstellar dust grains absorb or scatter photons , the kinematical information contained within these photons , in the form of the doppler shift , will also be absorbed or scattered .
it is important to realize that most of the kinematical data of galaxies are optical measurements .
firstly , stellar kinematics are always measured from optical absorption lines . in principle , the attenuation effects of interstellar dust could be minimized by using the sensitive @xmath0co absorption feature at 2.29 @xmath1 m ( gaffney , lester & doppmann 1995 ) . measuring stellar kinematics at near - infrared ( nir ) wavelengths is , however , observationally extremely difficult due to the high sky surface brightness .
secondly , the vast majority of spatially resolved rotation curves of spiral galaxies have been measured using optical emission lines ( usually the h@xmath2 line ) .
they can also be measured using the hi or co lines , where dust extinction is not an issue . however , these methods are observationally much more expensive than the more straightforward h@xmath2 observations . currently , the only way to investigate the effects of dust attenuation on the observed kinematics of galaxies is by means of detailed radiative transfer modelling .
many different approaches exist to handle the radiative transfer problem , and various teams have adopted different techniques to study the effect of dust on the photometry and spectral energy distributions ( seds ) of galaxies ( e.g. bruzual , magris & calvet 1988 ; witt , thronson & capuano 1992 ; byun , freeman & kylafis 1994 ; wise & silva 1996 ; bianchi , ferrara & giovanardi 1996 ; corradi , beckman & simonneau 1996 ; baes & dejonghe 2001a ) . the effect of dust attenuation on the observed kinematics , however , is largely unexplored , and limited to the rotation curves of spiral galaxies .
davies ( 1990 ) argued that the effects of dust absorption on the observed optical rotation curve of a disc galaxy can lead to a severe underestimation of the true rotational velocity in the inner regions .
this argument was quantified by bosma et al .
( 1992 ) , who calculated the effect of dust absorption on the observed rotation curve of edge - on galaxies .
the most detailed work on this subject comes from matthews & wood ( 2001 ) , who calculate the effects of dust attenuation on the rotation curve of low surface brightness ( lsb ) galaxies through monte carlo simulations .
we have embarked on a program to systematically investigate the effects of dust on the observed kinematics of galaxies . in previous papers we focused on the effects of dust attenuation on the observed stellar kinematics of elliptical galaxies .
initially , we only considered absorption by dust grains in our models ( baes & dejonghe 2000 ; baes , dejonghe & de rijcke 2000 ) .
the inclusion of absorption in the calculation of the observed kinematics is a fairly simple procedure : it basically adds a weight to the contribution of each individual star along a line of sight .
the radiative transfer problem becomes much more complicated however when scattering is also included . in this case
, photons can leave their original path , such that any star can contribute to the observed kinematics in any line of sight .
moreover , not only the stellar velocities but also the individual dust grain velocities should be taken into account in the calculation of the observed kinematics .
we argue that the monte carlo method is the only radiative transfer modelling method in which kinematical information can be included in an elegant and straightforward way . using a one - dimensional spherical monte carlo code
, we found that the effects of dust scattering are fairly important : dust attenuation can serve as an additional or alternative explanation for the stellar kinematical evidence of a dark matter halo around ellipticals ( baes & dejonghe 2001b , 2002a ) . in this paper
, we tackle the more complicated problem of investigating the effect of dust attenuation on the observed kinematics of disc galaxies .
we have written a new monte carlo code ( skirt , acronym for stellar kinematics including radiative transfer ) that can handle any geometry of stars and dust .
this code is presented in section [ skirt.sec ] , and tested in section [ test.sec ] .
we present our model and the results of our monte carlo simulations in section [ results.sec ] .
these results are discussed in section [ discussion.sec ] , and section [ conclusions.sec ] presents our conclusions .
the basic characteristics of monte carlo radiative transfer have been explained at length by various authors ( e.g. cashwell & everett 1959 ; mattila 1970 ; witt 1977 ; fischer , henning & yorke 1994 ; bianchi et al .
in essence , a monte carlo radiative transfer code follows the life of a very large number of individual photons .
a photon is , at each stage in its existence , characterized by various quantities , such as its position @xmath3 , propagation direction @xmath4 , wavelength @xmath5 and doppler shift @xmath6 as measured in a frame of rest . and
the corresponding line - of - sight velocity @xmath6 , which are related by the expression @xmath7 . ] after being emitted , photons propagate on straight lines through the interstellar medium until they either interact with a dust grain or leave the galaxy .
the various interactions alter the properties of the photon , according to random numbers generated from the appropriate probability functions .
when at last the photon escapes from the galaxy , its final properties are recorded . after recording a large numbers of photons in this way , the global observed properties of the system can be calculated .
we will not repeat an in - depth description of the principles and the numerous equations of monte carlo radiative transfer here , as they are well described in the articles listed above .
instead , we will focus on a number of aspects which make our code different from the existing ones .
these differences are mainly our choice of the dust grid and the possibility to include kinematical information into the radiative transfer calculations .
when no kinematical information is included in the radiative transfer calculations , photons are characterized at each moment by a wavelength @xmath5 , a position @xmath3 and a direction @xmath4 .
initial values for these quantities must be generated randomly from the emitting stellar system , which can be composed of several stellar components .
the initial emission direction @xmath8 can be sampled from the unit sphere , the initial position @xmath9 is sampled from the spatial distribution of the stellar component , and the initial wavelength @xmath10 is sampled from the sed of the stellar component at the position @xmath9 .
the skirt code contains a library with a set of common stellar components ( including spherical jaffe , hernquist , plummer and de vaucouleurs models , and axisymmetric exponential , sech , isothermal disc models ) , from which it is straightforward to sample a random position ( see appendix a ) . another library contains a set of seds , including planck functions and realistic tabulated stellar and galaxy seds ( kinney et al .
1996 ; pickles 1998 ) .
this library also contains so - called monochromatic seds , which are described by a dirac delta function . sampling a random wavelength from
the latter is trivial , because all photons are emitted with the same wavelength
. when the skirt code is used to calculate the observed kinematics of a dusty galaxy , photons must also carry with them the kinematic signature of the star that has emitted them , in the form of a doppler shift . because this is a one - dimensional quantity , not all kinematical information of the star can be transmitted to the observer , and it is important to consider how the observed doppler shift relates to the velocity vectors of the star and the scattering dust grains .
this problem was discussed in detail in section 2.3 of baes & dejonghe ( 2002a ) , where it is shown that the general expression for the doppler shift after @xmath11 scattering events is @xmath12 where @xmath13 and @xmath14 are the velocity vectors of the emitting star and the @xmath15th dust grain respectively .
the initial value of the doppler shift attached to the photon is hence the component of the star in the direction of the original emission , @xmath16 . to complete the initialization of a photon
, we also need to generate a line - of - sight velocity from the appropriate probability distribution . in this case , this function is the spatial line - of - sight velocity distribution ( losvd ) @xmath17 , which describes the probability for a star at a position @xmath9 to have a velocity component @xmath6 in the direction @xmath8 . in general ,
the spatial losvd is a marginal probability distribution of the stellar distribution function @xmath18 , which describes the probability density of the stellar component in six - dimensional phase space , @xmath19 where @xmath20 is the stellar number density and the integration covers the entire velocity space perpendicular to the direction @xmath4 .
thus , in order to be of use for a kinematical monte carlo run , each stellar component must allow the generation of a velocity from its spatial losvds for arbitrary @xmath3 and @xmath4 .
the spatial losvds can be calculated analytically for a number of self - consistent spherical models .
apart from isotropic models ( e.g. plummer 1911 ; hnon 1959 ; jaffe 1983 ; hernquist 1990 ) , this is possible for the anisotropic families of plummer and hernquist models described by dejonghe ( 1987 ) and baes & dejonghe ( 2002b ) . for more complicated dynamical models ,
however , an analytical evaluation of the spatial losvds is not possible .
for such models , we approximate the velocity distribution at each position in the galaxy as a local gaussian distribution .
this means that at each position @xmath3 , the distribution function can be written as @xmath21 .
\label{gaussf}\end{gathered}\ ] ] here @xmath22 and @xmath23 represent the mean velocities and velocity dispersions in the directions @xmath24 , the principle axes of the velocity ellipsoid at the position @xmath3 .
the orientation of the velocity ellipsoid and the values of the mean velocities and velocity dispersions can vary from position to position , such that the distribution function ( [ gaussf ] ) represents a fairly general distribution function . but a distribution function of this form has the advantage that the spatial losvds can be calculated exactly for any direction @xmath4 by applying the formula ( [ defslosvd ] ) .
we find that the spatial losvd will also be a gaussian distribution , @xmath25 , \label{gaussianslosvd}\ ] ] with parameters @xmath26 where obviously @xmath27 are the components of the vector @xmath4 with respect to the orthonormal reference system @xmath24 .
if hence , for each stellar component , we specify the orientation and parameters of the velocity ellipsoid at each position @xmath3 , we can easily sample line - of - sight velocities from the spatial losvds in an arbitrary direction , and hence initialize the photons initial doppler shifts . the life of a single photon can be thought of as a loop , whereby at each iteration ( representing a physical process ) we must update its position @xmath3 , propagation direction @xmath4 , wavelength @xmath5 and doppler shift @xmath6 .
the initial values are determined randomly at the emission phase , and change at every scattering event . as we only consider coherent scattering , and we are not taking into account the re - emission of photons at longer wavelength , the only wavelength change of a photon along its path is due to the varying doppler shift .
these wavelength variations are so tiny that the optical properties of the dust ( extinction curve , dust albedo and asymmetry parameter ) do not vary .
the wavelength of a photon can hence be considered fixed throughout its lifetime .
we must therefore only update the three quantities @xmath3 , @xmath4 and @xmath6 at each iteration , from their old values ( @xmath28 ) to the new ones ( @xmath29 ) .
this proceeds in several steps .
the first step in the iteration consists of determining whether an interaction with a dust grain will take place or whether the photon will leave the galaxy . to do this , we sample an optical depth @xmath30 from an exponential distribution and compare it to @xmath31 , the total optical depth along the path .
when @xmath32 , the photon will leave the galaxy ; in the other case , the photon will interact with a dust grain .
this interaction can either be a scattering or an absorption event , which is determined by the scattering albedo .
in the former case , the photon will continue its journey through the galaxy , in the latter case , the loop is ended .
the second step in the loop , if the photon is scattered , is the determination of the position of the scattering .
therefore we have to translate the sampled optical depth @xmath30 to a physical path length @xmath33 . with this path length
known , the new position is @xmath34 .
the third step in the iteration is the determination of the new propagation direction @xmath35 of the photon .
it is found by sampling a direction from the probability density @xmath36 , which represents the scattering phase function .
various phase functions are built into the skirt code , including the isotropic phase function and the anisotropic henyey - greenstein phase function .
the sampling of a direction from these phase functions can be performed analytically .
the final step is to update the photon s doppler shift .
the relative orientation of the propagation directions before and after the scattering event cause a change in doppler shift [ from equation ( [ vp ] ) ] @xmath37 to calculate the updated doppler shift , we need to sample a dust grain velocity from the dust velocity distribution .
therefore , the dust components in the skirt library must not only be specified by a spatial distribution , but also by the entire phase space distribution @xmath38 .
analogous to the stellar distribution function , we assume that the dust velocity field can be described by a trivariate gaussian distribution . because the dispersions in the interstellar medium are fairly small compared to the rotational velocities however , we restrict ourselves to assuming an isotropic dispersion tensor , such that @xmath39,\ ] ] with @xmath40 and @xmath41 being the mean velocity vector and the velocity dispersion of the dust . similarly as in the stellar case , we do nt have to sample a full three - dimensional velocity vector for the dust , but only a component in the direction @xmath42 this means that we just have to sample a velocity from a one - dimensional gaussian distribution with mean velocity @xmath43 and dispersion @xmath44 .
the first step in the iteration described in the previous subsection requires the calculation of the total optical depth along a given path .
it is found through the integral @xmath45 where @xmath46 represents the total ( absorption plus scattering ) opacity of the dust at a position @xmath3 . to execute the second step in the iteration
, we need to convert an optical depth @xmath30 into a physical path length @xmath47 along a given path , which is done by solving the equation @xmath48 for @xmath49 . only for a few simple geometries , such as constant density or some spherical dust distributions ,
can these integrations be performed analytically . in general , the calculation of the two quantities @xmath50 and @xmath47 must be performed numerically .
these integrations are usually the most time - consuming part of monte carlo codes .
therefore , it is important to think about an efficient and flexible way to do this ; we have considered two approaches . in order to allow a completely arbitrary distribution of dust ( including clumpy dust distributions ) ,
most modern radiative transfer monte carlo codes adopt an approach that can be described as a uniform dust density ( udd ) grid .
it consists of dividing space into a number of cells , and attaching a uniform dust density to each cell , for example the value at the centre of the cell . this uniform density ( and hence opacity )
makes it very easy to calculate the optical depth [ equation ( [ task1 ] ) ] along a given path : simply calculate the distance that the photon runs through each cell it crosses , multiply this distance with the opacity in the cell , and add all the pieces together .
the inversion of the equation ( [ task2 ] ) is not much more difficult .
first , look for the cell in which the interaction will take place . since the opacity increases linearly with path length within a single cell ,
this inversion of the equation is then straightforward . in the skirt code
, a udd grid is included based on a three - dimensional cartesian grid .
the grid cells can be chosen arbitrarily , in order to achieve an efficient grid for a variety of dust distributions .
an algorithm is included to construct clumpy dust distributions from an underlying smooth dust distribution , as described in witt & gordon ( 1996 ) and bianchi et al .
( 2000 ) .
in addition to the udd grid , we explore a new approach to handle the integration through the dust : an interpolated dust density ( idd ) grid .
the basis is the same : we divide space into a number of cartesian grid cells , but , instead of attaching a uniform opacity to each cell , we use the correct values of the opacity at the eight border points of the cells , and we apply a simple trilinear interpolation routine to determine the opacity in each of the points within a grid cell .
consider a photon at the position @xmath51 , a position within the @xmath52th cell with border points @xmath53 , etc .
we define the dimensionless quantities @xmath54 , @xmath55 and @xmath56 as @xmath57 since @xmath54 increases linearly from 0 to 1 as @xmath58 moves linearly from @xmath59 to @xmath60 , and similarly for @xmath55 and @xmath56 , the trilinear interpolation rule for the opacity simply reads @xmath61 where @xmath62 , etc .
represent the opacity in the border points . now consider the fraction of the path that remains in this cell .
this path can be parametrized using the expression @xmath63 , with the path length @xmath33 as a free parameter . in each of the points along this path ( as long as we remain in the same cell )
, we can calculate the opacity by a similar formula as ( [ kappaintpol ] ) , @xmath64 \left[1-q(s)\right ] \left[1-r(s)\right ] \kappa_{i , j , k } \nonumber \\ & \quad+ \left[1-p(s)\right ] \left[1-q(s)\right ] r(s)\ , \kappa_{i , j , k+1 } \nonumber \\ & \quad+ \cdots \nonumber \\ & \quad+ p(s)\,q(s)\,r(s)\,\kappa_{i+1,j+1,k+1},\end{aligned}\ ] ] where the parameters @xmath54 , @xmath55 and @xmath56 are now linear functions of the path length @xmath33 , @xmath65 along a path within a single cell , the opacity is thus a cubic polynomial in @xmath33 , the distance traveled along the path , @xmath66 with constant coefficients @xmath67 . with this expression for the opacity ,
the calculation of the optical depth @xmath50 is very straightforward .
indeed , it is sufficient to calculate , for each cell the path passes through , the two boundary positions .
this enables us to calculate the path length @xmath33 within the cell and the coefficients @xmath67 .
given these coefficients , the portion of the optical depth in that cell is found by substitution into the quartic polynomial @xmath68 also , the calculation of the path length @xmath47 is straightforward .
first , we determine the cell in which the optical depths at the border points bracket the value of @xmath30 , in a similar way as described above to calculate the total optical depth along the path .
the determination of @xmath33 then comes down to finding the root of a quartic equation . the obvious way of finding this root is by means of a newton - raphson iteration , because we can easily compute both the function values and the derivatives of this function .
this routine has a very fast quadratic convergence ( the number of significant digits approximately doubles with each iteration ) .
usually , no more than two iterations are necessary .
to compare the accuracy and efficiency of the two approaches , we applied both techniques to a simple spherical dust geometry , characterized by an opacity @xmath69 the integrations through the dust , i.e. the calculation of the two functions @xmath50 and @xmath47 , can be performed exactly for such a dust component with @xmath70 , 3 or 5 [ see baes ( 2001 ) for details ] .
these exact results were compared with the results from the udd and idd grids . in figure [ grid.eps ]
we plot the relative error of the calculated values of the opacity , the total optical depth along a path and the path length corresponding to a given optical depth for 1000 randomly generated photons for such a model . for a given number of dust grid cells ,
the accuracy of the integrations in the idd grid is much better than those in the udd grid .
however , in terms of computational efficiency , comparing the two grids with the same numbers of cells is not fair , because the integrations in the udd grid are less complicated and hence faster .
we found that the idd grid integrations are a factor 3 slower in the mean . on the other hand , in order to reach a similar accuracy as the idd grid , many more cells are necessary in the udd grid .
apart from slowing down the integrations , this will also increase the memory requirements significantly .
after several tests , we found that the idd grid is computationally preferable above the udd grid , in particular when the system harbours a large dynamical range of dust densities ( such as an exponential profile which is appropriate in disc galaxies ) .
it should be noted however , that the udd grid still remains a very useful alternative , in particular to include clumpy dust distributions into the skirt code .
the procedure described at the beginning of the previous section is the most basic version of a monte carlo iteration step . when executed as such
, however , it would be very inefficient . throughout the years a number of intelligent tricks have been presented which increase the computational efficiency considerably .
most of them attach a weight to each photon , which can be altered during its lifetime .
we incorporate three such techniques into our skirt code .
the first optimization to the basic monte carlo scenario is to turn all interactions into scattering events . in reality , of course , the interaction between a photon and a dust grain can either be an absorption or a scattering event ; the nature of this event is determined by the albedo @xmath71 of the dust grains .
the problem with absorptions is that the photon is lost , and therefore does not contribute to the observed radiation field anymore . to overcome this problem
, we force each interaction to be a scattering event and alter the weight of the photon after each scattering by a factor @xmath71 in order to compensate for the fraction of absorbed photons .
the second technique is the peeling - off procedure introduced by yusef - zadeh et al .
in the basic monte carlo iteration , each photon leaves the galaxy at a certain stage , and it can be classified in bins according to its position and direction .
this approach has a number of disadvantages .
firstly , if we are interested in only one single observing position , most of the photons are not used . secondly ,
if the observer has the misfortunate of being located in a direction in which not many photons leave the galaxy , he might have to wait for a very long time to detect enough photons to obtain good statistics .
thirdly , if one wants to compare the view of a system from two observing positions close to each other , the direction bins must be chosen to be very small , which requires a very large number of photons .
the peeling - off technique can be used to tackle these problems .
it consists of creating a new photon during each scattering [ resp .
emission ] process , which is scattered [ resp .
emitted ] exactly in the direction of the observer .
this photon will be detected by the observer .
its weight is altered by a factor @xmath72 [ resp . 1 ] to compensate for the probability of scattering [ resp .
emission ] in the direction @xmath73 , and a factor @xmath74 $ ] to compensate for extinction along the line - of - sight .
the third optimization we apply is the principle of forced first scattering ( witt 1977 ) , which is useful in the relatively low optical depth regime .
a problem in such systems is that most of the photons leave the galaxy directly without a single interaction with a dust grain , and therefore many photons need to be generated in order to have a good statistics of the scattered radiation . to overcome this problem , we split the emitted photon into two parts .
the first part of the photon directly leaves the galaxy and obtains a weight @xmath75 ( because we adopt the peeling - off technique , we do nt have to consider this fraction anymore ) .
the other part of the photon with the remaining weight is forced to be scattered at least once , which is achieved by sampling an optical depth @xmath30 such that @xmath76 .
this technique is called the principle of forced first scattering , but nothing prevents us applying it only to freshly emitted photons : it can equally well be applied to scattered photons ( bianchi et al .
the number of forced scatterings we apply in our calculations is usually 3 , but this value can be set arbitrarily according to the studied system .
the last phase in the monte carlo cycle is the detection phase , whereby the information contained in the photon is transmitted to the observer . because of the peeling - off procedure , the number and position of the observers can be chosen randomly . at each observing position
, the user can choose from several simulated instruments to detect the photons : photometers , low - resolution and high - resolution spectrographs .
each instrument consists of a number of pixels on the plane of the sky and they come in two geometries : rectangular and circular . each pixel contains a detector , the nature of which depends on the kind of instrument .
the photometers are the most simple instruments , whereby the detector in each pixel is just a counter . every time a photon hits the pixel and its wavelength is within the detector s bandwidth
, its weight is added to the intensity . at the end of the simulation , the 2d surface brightness image of the system is directly obtained .
the photometers are designed such that each photon remembers the number of scatterings it has undergone , and this information is also used by the photometric detectors .
this allows us to construct images for each individual scattering ( direct emission , photons that have been scattered once , etc . ) , which can be useful to disentangle the effects of absorption and scattering . the low - resolution spectrographs work in a very similar way , except that there is not just one detector , but a range of detectors corresponding to different wavelength bins .
in this way a data cube is produced ( actually a set of data cubes , one for every scattering ) , very similar to the data cubes obtained by radio or fabry - perot observations .
the high - resolution spectrographs basically produce similar data cubes , with doppler shift bins replacing the wavelength bins . combining the different doppler shift bins at a given position @xmath77 on the sky
, we obtain the losvd or line profile @xmath78 , which describes the entire projected velocity distribution . from the losvds ,
the 2d mean projected velocity field @xmath79 and projected velocity dispersion field @xmath80 can be calculated . from a conceptual point of view ,
high - resolution and low - resolution spectrographs work in a very similar way .
both separate the incoming photons in different bins according to the extra information carried by the photon , which is the wavelength in the former and the doppler shift in the latter case . from a computational point of view
, however , there is an important difference between the two .
indeed , the wavelength of the photon is actively used throughout the monte carlo iteration , because the optical properties of the dust grains depend on the photon s wavelength . on the contrary , a photon s path does not depend on its doppler shift . moreover , the changes in doppler shift experienced by a photon when it moves through the dust are independent of the initial stellar doppler shift .
therefore , we can optimize our code in the following way . instead of initializing the photon s doppler shift by generating a @xmath81 from the stellar spatial losvd
, we postpone this contribution of the stellar velocity until later and initialize @xmath81 to zero .
when the photon leaves the galaxy after @xmath11 scattering events , it will then carry a doppler shift @xmath82 where we still have to add the stellar doppler shift in order to obtain the correct value . instead of generating a single initial stellar doppler shift @xmath81 from the spatial losvd @xmath17 , we split the photon over all velocity bins .
each fraction will then be weighted by a weight factor that represents the probability that the final doppler shift of the photon will fall within the corresponding bin . in order to fall within a bin with boundaries @xmath83 and @xmath84
, the photon must have an initial stellar doppler shift satisfying @xmath85 .
the weight factor corresponding to the @xmath86th bin will hence equal @xmath87 when we substitute the gaussian spatial losvd ( [ gaussianslosvd ] ) into this expression , we can evaluate this weight factor exactly @xmath88.\ ] ] this method is of the same kind as the optimization techniques described in section [ opt.sec ] and also increases the efficiency of the skirt code in a significant way .
the skirt code is implemented in the ansi standard c++ language and makes optimal use of the object - oriented structure of this language .
the output of the code consists of fits files , which are created by means of the fitsio and ccfits packages .
the run time of the skirt code with reliable output depends critically on the number of photons , the number of observing directions , the spatial and velocity resolution of the instruments and the number of cells in the dust grid .
for the simulations described in section [ results.sec ] , we found that about @xmath89 photons are necessary to obtain accurate results .
the code spends 60 per cent of the time moving through the dust grid , while most of the remaining time is spent in the calculation of the spatial losvds .
when run on a modern pentium iv pc , the run time of a single simulation ( with 7 observing positions ) is of the order of 50 to 100 hours .
the code was first run without dust included in order to test the projection algorithm and the calculation of the observed kinematics .
we used simple non - rotating spherical geometries which allow an analytical expression of the surface brightness and projected velocity dispersion profiles ( dejonghe 1987 ; hernquist 1990 ; jaffe 1983 , baes & dejonghe 2002b ) .
also the surface brightness profiles of face - on and edge - on exponential discs ( which can be calculated analytically ) are accurately recovered . moreover ,
when we attach a simple velocity structure to an exponential disc , with a flat rotation curve and exponentially decreasing velocity dispersions , the observed kinematics can be also calculated analytically in the face - on and edge - on directions .
these are recovered very well , proving the correct behaviour of the code .
next , we compared the results of the skirt code with our previous monte carlo calculations of the observed kinematics of elliptical galaxies ( baes & dejonghe 2001b , 2002a ) .
the monte carlo code used in these papers also allowed for the calculation of the observed kinematics , but only for spherical models , chosen such that the integrations through the dust could be performed analytically .
we found an excellent agreement between these results and those calculated with the new skirt code .
-band surface brightness distribution of an edge - on disc galaxy , for various values of the optical depth .
the model is similar to the bt0.3 model described by byun et al .
( 1994 ) , i.e. an exponential stellar disc with scale length @xmath90 kpc and scale height @xmath91 kpc , a de vaucouleurs bulge with @xmath92 kpc and a bulge - to - total luminosity ratio of 0.3 .
the dust is also exponential with scale length @xmath93 kpc and @xmath94 kpc , and the face - on central optical depths in the different plots are @xmath95 , 0.5 , 1 , 2 , 5 and 10 ( from top to bottom ) .
surface brightness contours are shown with a step of @xmath96 mag / kpc@xmath97 .
, scaledwidth=40.0% ] finally , we compared our ( photometric ) results with the radiative transfer calculations of other authors . in figure [ byun.eps ]
we plot the two - dimensional edge - on surface brightness distribution of an sb galaxy model in the @xmath98 band for various values of the optical depth .
the models are the same models as the bt0.3 models of byun et al .
( 1994 ) , and they consist of a stellar disc , a stellar bulge and a dust disc . comparing this figure with their figure 4b , we find an excellent agreement between the two results .
notice that byun et al .
( 1994 ) adopted a completely different approach to solve the radiative transfer problem .
we apply the skirt code to investigate the effect of dust absorption and scattering on the photometry and the stellar kinematics of normal galactic discs .
we limit our modelling only to a stellar disc and do not include a bulge , because we are primarily interested in the disc kinematics .
our model consists of a simple axisymmetric double exponential disc galaxy , with stellar emissivity @xmath99 the luminosity @xmath100 is a free scaling parameter in our models , as both the input emissivity and the output light profile scale with the luminosity . for the scale length and scale height
we chose the values adopted by byun et al .
( 1994 ) to describe the milky way , @xmath90 kpc and @xmath101 pc .
as we focus on the observed kinematics , which are always measured in a very narrow wavelength region , we chose a trivial monochromatic sed centered on the @xmath102 band . in order to run a kinematical simulation for such a disc galaxy , we also need to provide details on the kinematical structure of the disc .
we assume that the stellar velocities at each position in the disc can be represented by a trivariate gaussian distribution , @xmath103.\ ] ] such a velocity distribution was first proposed by schwarzschild ( 1907 ) to describe the distribution of stellar velocities in the solar neighbourhood . the mean azimuthal velocity and
the three velocity dispersion components can vary from position to position .
the entire kinematical structure of the disc will be completely determined if we have expressions for the four functions @xmath104 , @xmath105 , @xmath106 and @xmath107 .
since the system is axisymmetric , these quantities will depend only on the cylindrical radius @xmath108 and the height @xmath109 .
moreover , because stellar discs are thin and the contribution of photons high above or under the symmetry plane of the galaxy will be quite small due to the vertical exponential decrease , we will assume that these four functions are a function of @xmath108 only . for the @xmath110 profile
, we assume the arctan profile , a simple two - parameter fitting function that fits the observed rotation curves of disc galaxies fairly well ( e.g. courteau 1997 ) , @xmath111 we have adopted an amplitude @xmath112 km / s and a scale radius @xmath113 kpc , which results in a rotation curve that has a steep initial gradient and reaches its flat part fairly quickly . the internal velocity dispersion profiles in disc galaxies are not well constrained . for a thin self - gravitating disc ,
the jeans equations yield that the vertical velocity dispersion scales with the square root of the surface density .
if we assume a constant mass - to - light ratio , we obtain the following parametrization , @xmath114 as all dispersions in a stellar disc are thought to exist from the gradual stellar heating of an initial cold disc , it makes sense to assume that the velocity dispersions in the other direction are intimately coupled to the vertical dispersion .
therefore , we assume a constant axis ratio of the velocity ellipsoid in our model , such that @xmath115 and @xmath116 have the same spatial variation as @xmath117 . notice that such behaviour is found to be in agreement with the sparse existing data on the velocity dispersions in external galaxies ( gerssen et al .
1997 , 2000 ) .
the only remaining quantities that need to be set are normalizations of the dispersions .
we adopted the values @xmath118 km / s . at a galactocentric radius of 8 kpc ,
this gives velocity dispersions of ( 37,28,18 ) km / s , which roughly corresponds to the measured velocity dispersions in the solar neighbourhood .
the dust was chosen to be distributed in a similar way as the stellar distribution , @xmath119 whereby @xmath30 ( here ) represents the face - on optical depth through the centre of the galaxy , @xmath120 for the dust scale length and scale height , we again choose the values from byun et al .
( 1994 ) , @xmath93 kpc and @xmath121 pc .
notice that the dust is hence significantly more confined to the plane of the galaxy , a feature which gives rise to the well - known dust lanes in edge - on galaxies . for the velocity field of the dust grains
, we assumed an isotropic gaussian distribution , @xmath122.\ ] ] we assumed the same mean velocity as the stellar distribution , and a velocity dispersion @xmath123 km / s , which is a typical value for the dispersion of the cold gas in the interstellar medium ( sofue & rubin 2001 ) .
finally , for the optical properties of the dust we took the average galactic values ( gordon , calzetti & witt 1997 ) .
these functions and parameters completely determine our model , except for the one free parameter @xmath124 , the central face - on optical depth in the @xmath102 band ( the corresponding optical depth at other wavelengths is determined by the assumed optical properties of the dust ) .
as discussed in the introduction , the optical depth in spiral galaxies is still a matter of debate .
we ran our models for the optical depth values @xmath95 , 0.5 , 1 , 2 , 5 and 10 , in order to cover a wide range of possible scenarios .
in this section we present and describe the results of our monte carlo simulations .
we will not discuss the effects of dust on the photometry , since such results have been discussed at length by other authors , such as byun et al .
( 1994 ) , corradi et al .
( 1996 ) and bianchi et al .
instead we concentrate on the effects of dust on the observed kinematics , in particular on the mean projected velocity @xmath79 and the projected velocity dispersion @xmath80 . in figures
[ mw_vp.eps ] and [ mw_sigp.eps ] we compare , at various inclination angles , the 2d mean projected velocity and projected velocity dispersion fields of an optically thin galaxy ( left ) with the corresponding fields of a dusty galaxy with optical depth @xmath125 ( centre ) and @xmath126 ( right ) respectively .
notice that the mean projected velocity field is not plotted in the face - on direction , as there is no rotation perpendicular to the galactic plane . to facilitate the interpretation of these 2d kinematical fields , we plotted in figure [ mwkin.eps ] the major and minor axis kinematical profiles of our models at various inclination angles , explicitly as a function of the optical depth .
.the effect of dust attenuation on the slope of the mean projected velocity curve .
this slope is expressed through the major axis radius at which the rotation curve reaches half of the asymptotic value @xmath127
these radii are tabulated for the various inclinations and optical depths considered in our simulations . for a infinitely thin and dust - free galaxy
, the intrinsic value would be 0.5kpc , independent of the inclination angle . [
cols="<,>,>,>,>,>",options="header " , ] [ rot.tab ] for edge - on galaxies , the effects of dust on both the mean projected velocity and the projected velocity dispersion are very strong , as is obvious from the bottom row panels of figure [ mw_vp.eps ] and [ mw_sigp.eps ] , and the panels in the right column of figure [ mwkin.eps ] .
the projected mean velocity profile along the major axis rises more and more slowly as the optical depth increases , and tends towards a solid body rotation .
these trends were expected , and in agreement with the arguments of davies ( 1990 ) and the absorption - only calculations of bosma et al . ( 1992 ) . to quantify this change of slope , we have calculated , for each optical depth , the major axis radius at which the mean projected velocity reaches half of its maximum value @xmath127 kms@xmath128 .
these values are tabulated in the second column of table [ rot.tab ] , and demonstrate that the effects of dust attenuation are very severe , even for small optical depths .
the reason for this shallower slope of the rotation curve is that we mainly see the radiation from the optically thin part of the galaxy , which is more and more restricted to the outer part of the galaxy as the optical depth increases . in the same way
, we can understand the effects of dust attenuation on the projected velocity dispersion profile , where also a prominent dust lane is visible in the contour plot : we mainly see the light ( and hence the doppler shift ) of the stars from the outer part of the disc , where the intrinsic velocity dispersion is much smaller .
bosma et al . (
1992 ) found that the signature of dust absorption on the rotation curve depends strongly on the inclination angle . for inclinations which deviate as few as 5 degrees from purely edge - on
, they found that the effects of dust on the rotation curve is already severely reduced , even for optical depths up to @xmath129 .
this result remains the same if scattering is also taken into account , because the absorption effects of dust strongly dominate in highly inclined galaxies .
table [ rot.tab ] quantitatively demonstrates these results .
for the projected velocity dispersion , a similar effect is valid , at least for the major axis dispersion profile . when the entire projected velocity dispersion field ( or just the minor axis dispersion profile ) is taken into account , the asymmetrical signature of the dust
can still be recognized for inclinations down to 80 degrees . moving to intermediate inclinations
, the effects of dust attenuation on the observed kinematics quickly become very limited . for example
, there is hardly any effect noticeable for an inclination of 60 degrees , which can easily be seen by comparing the corresponding panels in figure [ mw_vp.eps ] and [ mw_sigp.eps ] . in the light of the previous results
, we would expect that the effects of dust attenuation on the observed kinematics of face - on galaxies are completely neglegible . and
indeed , on the mean projected velocity there is no effect at all : the projected mean velocity is zero for face - on galaxies , regardless of whether dust is present or not .
however , we do see a conspicuous effect of dust attenuation on the projected velocity dispersion : it increases with increasing optical depth
. this can not be an absorption effect , because all stars along a given line of sight in our face - on galaxy discs have the same velocity dispersion @xmath117 in the direction of the observer .
the effect arises as a result of photons emitted in the central regions in a direction parallel to the plane of the galaxy .
when these photons are scattered into a vertical direction , they have a large probability to leave the galaxy .
they will then contribute the high - velocity information of the stars in the central regions to the losvds , causing high - velocity wings and hence an increased velocity dispersion .
the effect is similar to what we found for the projected velocity dispersion profile of elliptical galaxies containing a diffuse dust component ( baes & dejonghe 2002a ) .
compared to these models , where the signature of the dust is very significant , it is fairly modest for the face - on discs .
the reason for this weaker impact is the fact that the velocity dispersion profile in our disc galaxy models have a fairly shallow slope , compared to the elliptical galaxy models of baes & dejonghe ( 2002a ) . in disc galaxies ,
the doppler shifts contributed by the scattered photons will in the mean differ less from the doppler shifts contributed by the direct photons as in elliptical galaxies .
we could ask ourselves the question whether this conclusion is biased by the chosen star - dust geometry of our disc galaxy models . indeed , baes & dejonghe ( 2002a ) found their results depended strongly on the relative star - dust geometry : the scattering effect was strong only for models in which the dust was more extended than the stars . in our disc galaxy models , the dust distribution has the same scale length as the stars , whereas the scale height is smaller , such that the dust is always sandwiched between the stars .
recently however , evidence has been found for extended distributions of cold dust in spiral galaxies , both from the radiative transfer modelling of the dust lanes in edge - on galaxies ( xilouris et al .
1997 , 1998 , 1999 ) and from iso and scuba dust emission ( alton et al .
1999 ; davies et al . 1999 ) . in order to investigate whether this would bias our results , we considered a new set of models with an extended dust distribution .
they were constructed by increasing the dust scale length or the dust scale height by factors up to three , while keeping the face - on optical depth fixed .
running new face - on simulations for these models , we found that the strength of the scattering effect on the projected velocity dispersion increases only minimally , with deviations of the order of a few per cent .
the modest effects found in figure [ mwkin.eps ] are hence representative for disc galaxies .
one of the most useful and promising applications of stellar kinematics in disc galaxies is their contribution to mass modelling . in its simplest form ,
this mass modelling consists of decomposing the observed rotation curve into contributions by the disc and the dark halo ( assuming the contributions of a bulge and molecular / atomic gas are negligible ) . even in this most simple scenario ,
there is still a virtually complete degeneracy : the mass - to - light ratio of the disc .
authors have often adopted a solution in which the contribution of the disc is maximized , leading to the so - called maximum disc hypothesis ( van albada & sancisi 1986 ) .
there are however , no convincing arguments to favour a particular value for the stellar mass - to - light ratio , and often scenarios in which the dark matter halo dominates the disc are equally possible ( e.g. lake & feinswog 1989 ) .
stellar kinematics can , in principle , provide the necessary additional constraint to break this degeneracy .
indeed , for a self - gravitating disc , the vertical velocity dispersion scales linearly with the square root of the surface density of the disc , and can therefore be used to constrain the mass - to - light ratio of the disc .
the most obvious candidates for such an analysis would be face - on galaxies , because their projected velocity dispersion directly reflects the vertical velocity dispersion in the disc .
we found that for a dusty disc galaxy , the observed velocity dispersion is affected by scattering .
the effects are modest , however , even for models with an extended dust distribution . alas
, face - on galaxies do have strong observational disadvantages .
foremost , their low surface brightness make it very hard to obtain the stellar kinematics out to sufficiently large distances to isolate the disc component .
moreover , the shape and amplitude of the rotation curve can not be measured directly .
the amplitude can be estimated fairly accurate using the tully - fisher relation , whereas the shape could be approximated through the ( controversial ) universal rotation curve formalism ( persic , salucci & stel 1996 ) .
this is unsatisfactory , however , because there is a large spread in universal rotation curves , and the details of mass modelling usually depend crucially on small scale features of the rotation curve which statistical methods can not incorporate .
edge - on galaxies do nt suffer from these observational problems : their surface brightness is much higher , and their apparent rotation curve can be determined directly .
however , the interpretation of the kinematical data is non - trivial .
firstly , the vertical velocity dispersion must be linked to the observed dispersion , which is a mixture of contributions from the radial and tangential velocity components .
based on the observed axis ratio of velocity ellipsoid in the solar neighborhood , bottema ( 1993 ) estimated the vertical velocity dispersion in edge - on galaxies by means of the radial velocity dispersion evaluated at one scale length .
it has been shown , however , that the velocity ellipsoid axis ratios in disc galaxies can vary substantially ( gerssen et al .
secondly , our modelling indicates that both the projected mean velocity and the projected velocity dispersion field in edge - on galaxies are strongly affected by interstellar dust .
galaxies of an intermediate inclination ( @xmath130 ) , however , are most suitable for such an analysis .
gerssen et al .
( 1997 ) have shown that for galaxies with an intermediate inclination , the three components of the velocity ellipsoid can , in principle , be determined by studying the kinematics along major and minor axes . also ,
because the rotation curve of inclined galaxies can directly be determined and their surface brightnesses are high enough to allow the measurements of their kinematics in a reasonable time , they form ideal targets for such a study . moreover ,
determining the three components of the velocity ellipsoid is very useful to constrain the dynamical history of disc galaxies , because the various proposed mechanisms leave a different imprint on the velocity ellipsoid .
our modelling demonstrates that the observed kinematics of inclined disc galaxies can be directly interpreted and are not biased by dust attenuation .
we argue that studying the observed kinematics of a substantial set of optically smooth inclined galaxies could thus greatly contribute to understanding the mass structure and dynamical history of spiral galaxies .
the results of our modelling are important concerning the derivation of the rotation curve of spiral galaxies . whereas the outer parts of the rotation curves of spirals were the preferred field of interest for nearly three decades , it is the inner slope of the rotation curve which nowadays stands in the spotlight . to measure the slope in this inner region accurately ,
a high spatial resolution is required , which makes the usual 21-cm measurements less suitable , except for the most nearby galaxies .
a useful alternative are co rotation curves , which can achieve a high resolution in both the spatial and velocity directions , but these observations are quite costly in observing time . the main alternative remains optical h@xmath2 observations , either long - slit or fabry - perot .
using h@xmath2 rotation curves , several authors ( blais - ouellette , amram & carignan 2001 ; borriello & salucci 2001 ; de blok , mcgaugh & rubin 2001 ; de blok & bosma 2002 ) have recently found shallow slopes for the inner rotation curve of a significant number of ( mainly lsb and dwarf ) galaxies .
such slopes are not in agreement with the results of high - resolution cosmological n - body simulations , where it is found that dark matter halos have a strong cusp , and hence that rotation curves of dark matter dominated galaxies should be steeply rising ( navarro , frenk & white 1997 ; moore et al .
a number of `` solutions '' have been proposed to explain this discrepancy , amongst them the suggestion that the internal extinction in the centre of galaxies could be a major source of uncertainty .
our models show that , indeed , an edge - on galaxy with an observed slowly rising rotation curve could in principle be a _
dusty _ galaxy with an intrinsically steep rotation curve .
we found , however , that this effect can not be obtained with realistic amounts of dust if the galaxies deviate more than a few degrees from exactly edge - on , confirming the results of bosma et al .
( 1992 ) and matthews & wood ( 2001 ) .
in addition , because lsb galaxies are assumed to be dust poor ( mcgaugh 1994 , tully et al . 1998 ;
matthews et al .
1999 ) , we have demonstrated that dust effects can not explain the observed discrepancies . we do however demonstrate that dust effects are important for galaxies within a few degrees from edge - on , and that dust attenuation must be seriously taken into account in such cases .
the calculation of the final doppler shift of a photon in the skirt code takes into account the velocity information of the star and the individual velocity vectors of each individual scattering dust grain .
this means that for every scattering event , we need to sample a velocity from the dust velocity field , and this is a costly operator .
it would be very convenient from a computational point of view , if we could neglect the dust grain velocities .
we have done this in our previous monte carlo simulations ( baes & dejonghe 2001b , 2002a ) , where we studied the observed stellar kinematics of elliptical galaxies . in these models , both stars and dust grains
are supported by random motions , and because the dust is colder than the stars , the extra doppler shifts due to scattering are generally smaller than the stellar doppler shift . in a spiral galaxy , however , this argument does not hold anymore , because the velocities of both stars and dust grains are dominated by the rotation and are therefore of the same magnitude . in order to investigate the importance of including the individual dust grain velocities into the code ,
we ran the same models again but without taking the dust velocities into account . in figure [ mwappkin.eps ]
we plot the major and minor axes mean projected velocity and projected velocity dispersion profiles of our models calculated this way . comparing this figure with the corresponding this figure with figure [ mwkin.eps ] , we find that the effects of dust on the mean projected velocity profiles are significantly overestimated when the dust grain velocities are not taken into account , in particular for inclined galaxies .
the differences between the projected velocity dispersion curves in both figures are even worse : when the dust grain velocities are neglected , the apparent effect of dust attenuation is to increase the dispersion at nearly all radii and all inclinations , by factors of up to 200 per cent or more for realistic optical depths . to understand this behaviour ,
consider a photon emitted by a star with velocity @xmath13 in the direction @xmath8 , and assume this photon is scattered by a dust grain with velocity @xmath131 into the direction @xmath132 , whereafter it manages to escape the galaxy .
when we do not take the dust grain velocity into account , the observed doppler shift of the photon will simply be @xmath133 .
on the contrary , the correct observed doppler shift of the photon reads @xmath134 in a disc galaxy , the motion of both stars and dust grains is dominated by the same rotation .
if the pathlength of the photon between its emission and scattering is short , the velocity vectors of the star and dust grain will therefore be very similar , @xmath135 , such that @xmath136 .
the new correct doppler shift is hence a typical line - of - sight velocity of a star in the new propagation direction @xmath132 of the photon .
therefore , taking the dust motion into account more or less forces the dust grain to adopt a doppler shift that is `` appropriate '' for the new propagation direction . along a given line - of - sight ,
the velocity information carried by scattered photons will , in the mean , only modestly deviate from the velocity information carried by photons directly emitted in same direction .
if we do nt take the dust grain velocities into account , the photon can contribute , to the observed kinematics in a given direction , a line - of - sight velocity typical for a completely different direction .
in the elliptical galaxy models from baes & dejonghe ( 2002a ) , this does not make much difference , because the velocities of the stars have similar magnitudes in the different directions . in a rotating disc galaxy however , this makes a huge difference because the stars have hugely different velocities in different directions : the vertical and radial motions are determined by the dispersion and are of the order of a few tens of km / s , whereas the azimuthal motion is dominated by the rotation and is of the order of 200 km / s .
if the dust grain velocities are not correctly taken into account , scattered photons can thus cause erroneous extremely - high - velocity wings in the losvds .
these give rise to modest but significant errors in the mean projected velocity , and very large errors in the projected velocity dispersion . as a remark
, we want to point out that matthews & wood ( 2001 ) found that `` also including the doppler shifts arising from the relative bulk motion of the scattering dust particles had a negligible effect on the final rotation curves '' . whereas this statement is indeed true for the mean projected velocity of highly inclined galaxies
, we nevertheless judge that this could easily be misinterpreted as a justification to neglect dust velocities altogether .
it must be realized that for the calculation of the entire losvds , and for the projected velocity dispersion in particular , the dust velocities do play a crucial role , and must be taken into account .
in this paper we have presented a novel monte carlo code that can take velocity information into account , and can hence be used for kinematical modelling of dusty objects .
we applied this code to calculate the observed kinematics of dusty disc galaxies .
the main results of this paper can be summarized as follows . 1 .
a correct inclusion of kinematical information into radiative transfer problems requires the inclusion of the velocities of both the emitting stars and the scattering dust grains .
we see no other way of tackling this problem except with monte carlo techniques .
a new approach is presented to optimize the integration through the dust in a monte carlo code , using a trilinear interpolation instead of a constant opacity within each cell .
we compared both kinds of grids , and find that , to obtain a similar accuracy , the new approach is more efficient in terms of computation time and memory .
the effects of dust attenuation on the kinematics of edge - on disc galaxies are severe .
both the mean projected velocity and the projected velocity dispersion are severely affected , even for modest optical depths .
therefore we strongly advise to always interprete the stellar kinematics and optical rotation curves of edge - on galaxies very cautiously .
4 . for galaxies which are more than a few degrees from strictly edge - on , the effects of interstellar dust on the observed kinematics are much weaker .
therefore , we argue that dust attenuation can not be invoked as a possible mechanism to reconcile the observed slope of lsb galaxies with the predicted cdm cosmological models . 5 .
dust attenuation does not affect the kinematics of intermediately inclined galaxies .
such galaxies hence form the ideal targets for stellar kinematical studies , in particular to constrain the mass structure and to study the kinematical history of disc galaxies .
. the projected velocity dispersion of face - on galaxies increases slightly due to scattering of dust grains into the line of sight .
the effects are relatively small , however , even for extended dust distributions .
7 . neglecting the extra doppler effect caused by the scattering medium results in incorrect projected kinematics , in particular if the velocity components of the stars ( and dust grains ) in the galaxy differ considerably , such as in a rotating disc .
if the dust grain velocities are neglected , the mean projected velocity is significantly underestimated , and the projected velocity dispersion field is completely overestimated .
m.b . gratefully thanks the fund for scientific research for financial support .
alton p. , bianchi s. , rand r. , xilouris e. , davies j. , trewhella m. , 1999 , apj , 507 , l125 baes m. , ph.d .
thesis , univ . gent , belgium baes m. , dejonghe h. , 2000 , mnras , 313 , 153 baes m. , dejonghe h. , 2001a , mnras , 326 , 722 baes m. , dejonghe h. , 2001b , apj , 563 , l19 baes m. , dejonghe h. , 2002a , mnras , 335 , 441 baes m. , dejonghe h. , 2002b , a&a , 393 , 485 baes m. , dejonghe h. , de rijcke s. , 2000 , mnras , 318 , 798 bianchi s. , ferrara a. , davies j.i . , alton p.b . , 2000 ,
mnras , 311 , 601 bianchi s. , ferrara a. , giovanardi c. , 1996 , apj , 465 , 127 blais - ouellette s. , amram p. , carignan c. , 2001 , aj , 121 , 1952 borriello a. , salucci p. , 2001 ,
mnras , 323 , 285 bosma a. , byun y.i . , freeman k.c . ,
athanassoula e. , 1992 , apj , 400 , l21 bottema r. , 1993 , a&a , 275 , 16 bruzual g.a . ,
magris g.c . , calvet n. , 1988 , apj , 333 , 673 byun y.i . , freeman k.c .
, kylafis n.d . , 1994 , apj , 432 , 114 calzetti d. , 2001 , pasp , 113 , 1449 cashwell e.d .
, everett c.j .
, 1959 , a practical manual on the monte carlo method for random walk problems , pergamom , new york corless r.m .
, gonnet g.h . , hare d.e.g . , jeffrey d.j . ,
knuth d.e . , 1996 , adv .
comp . math .
, 5 , 329 corradi r.l.m .
, beckman j.e . ,
simonneau e. , 1996 , mnras , 282 , 1005 courteau s. , 1997 , aj , 114 , 2402 davies j.i . , 1990 , mnras , 245 , 350 davies j.i .
, burstein d. , 1995 , the opacity of spiral disks , nato asi series c , dordrecht : kluwer academic publishers davies j. , alton p. , trewhella m. , evans r. , bianchi s. , 1999 , mnras , 304 , 495 de blok w.j.g . , bosma a. , 2001 , a&a , 385 , 816 de blok w.j.g .
, mcgaugh s.s . ,
rubin v.c . , 2001 , aj , 122 , 2396 de vaucouleurs g. , de vaucouleurs a. , corwin h.g . , 1976 ,
second reference catalogue of bright galaxies , univ .
texas press , austin dejonghe h. , 1987 , mnras , 224 , 13 disney m.j . , davies j.i . ,
phillipps s. , 1989 , mnras , 239 , 939 fischer o. , henning th .
, yorke h.w .
, 1994 , a&a , 284 , 187 gaffney n.i .
, lester d.f . , doppmann g. , 1995 , pasp , 107 , 68 gerssen j. , kuijken k. , merrifield m.r . , 1997 , mnras , 288 , 618 gerssen j. , kuijken k. , merrifield m.r . ,
2000 , mnras , 317 , 545 gordon k.d .
, calzetti d. , witt a.n . , 1997 ,
apj , 487 , 625 hnon m. , 1959 , ann .
, 22 , 126 hernquist l. , 1990 , apj , 356 , 359 holmberg e. , 1958 , lund medd .
obs . serr .
ii , 136 , 1 jaffe w. , 1983 , mnras , 202 , 995 kinney a. l. ; calzetti d. , bohlin r.c . , mcquade k. , storchi - bergmann t. , schmitt h.r . , 1996 ,
apj , 467 , 38 kuchinski l.e . ,
terndrup d.m . ,
gordon k.d . ,
witt a.n . , 1998 ,
aj , 115 , 1438 lake g. , feinswog l. , 1989 , aj , 98 , 166 matthews l.d . , wood k. , 2001 , apj , 548 , 150 matthews l.d . , gallagher j.s.iii , van drielw . , 1999 , aj , 118 , 2751 mattila k. , 1970 , a&a , 9 , 53 mcgaugh s.s . , 1994 ,
apj , 426 , 135 moore b. , governato f. , quinn t. , stadel j. , lake g. , 1997 , apj , 499 , l5 navarro j.f . ,
frenk c.s .
, white s.d.m . , 1997 , apj , 490 , 493 persic m. , salucci p. , stel f. , 1996 , mnras , 281 , 27 pickles a.j . , 1998 , pasp , 110 , 863 popescu c.c . , tuffs r.j . , 2002 , mnras , 335 , l41 sandage a. , tammann g.a . , 1981 , a revised shapley - ames catalog of bright galaxies , carnegie institution of washington , washington schwarzschild k. , 1907 , in gttingen nachr .
, p. 614 sofue y. , rubin v. , 2001 , araa , 39 , 137 tully r.b .
, pierce m.j .
, huang j. , saunders w. , verheijen m.a.w . , witchalls p.l .
, 1998 , aj , 115 , 2264 valentijn e.a . , 1990 ,
nature , 346 , 153 van albada t.s .
, sancisi r. , 1986 , phil .
london a , 320 , 447 wise m.w .
, silva d.r . , 1996 ,
apj , 461 , 155 witt a.n . , 1977 ,
apjs , 35 , 1 witt a.n . ,
gordon k.d . , 1996 ,
apj , 463 , 681 witt a.n .
, thronson h.a.jr .
, capuano j.m.jr . , 1992 ,
apj , 393 , 611 xilouris e.m . ,
kylafis n.d . , papamastorakis j. , paleologou e.v . , haerendel g. , 1997 , a&a , 325 , 135 xilouris e.m . , alton p. , davies j. , kylafis n.d .
, papamastorakis j. , trewhella m. , 1998 , a&a , 331 , 894 xilouris e.m . , byun y. , kylafis n.d .
, paleologou e.v .
, papamastorakis j. , 1999 , a&a , 344 , 868 yusef - zadeh f. , morris m. , white r.l . , 1984 ,
apj , 278 , 186
in this appendix , we show how the skirt code generates a random position from the exponential disc model , i.e. how a random position is drawn from the three - dimensional probability density @xmath137 as this probability density is a separable function of @xmath108 , @xmath138 and @xmath109 , we can generate a random position by independently generating each of these three coordinates , whereby the generation of an azimuth is trivial . to generate a random radius and height
, we use the transformation technique , which says that a random variable @xmath58 can be drawn from a probability density @xmath139 by generating a uniform deviate @xmath140 and solving the equation @xmath141 for @xmath58 .
the calculation of a random @xmath109 is easy using this principle and results in @xmath142 this procedure works equally well when we introduce a cut - off , or when a sech or isothermal profile is used instead of the exponential profile to describe the vertical distribution of the stars ( e.g. bianchi et al . 1996 ) . applying the same technique to generate a random @xmath108 , we can find @xmath108 by solving the equation @xmath143 bianchi et al .
( 1996 ) invert this transcendental equation numerically to find @xmath108 .
however , this equation can be solved exactly by means of the lambert function , yielding @xmath144.\ ] ] the lambert function , also known as the product log function , is generally defined as the inverse of the function @xmath145 , and @xmath146 represents the only real branch of this complex function besides the principle branch .
the lambert function can be computed in a very efficient way by means of halley iteration ; for more information on this function see corless et al . | we present skirt ( stellar kinematics including radiative transfer ) , a new monte carlo radiative transfer code that allows the calculation of the observed stellar kinematics of a dusty galaxy .
the code incorporates the effects of both absorption and scattering by interstellar dust grains , and calculates the doppler shift of the emerging radiation exactly by taking into account the velocities of the emitting stars and the individual scattering dust grains .
the code supports arbitrary distributions of dust through a cellular approach , whereby the integration through the dust is optimized by means of a novel efficient trilinear interpolation technique .
we apply our modelling technique to calculate the observed kinematics of realistic models for dusty disc galaxies .
we find that the effects of dust on the mean projected velocity and projected velocity dispersion are severe for edge - on galaxies . for galaxies which deviate more than a few degrees from exactly edge - on ,
the effects are already strongly reduced . as a consequence ,
dust attenuation can not serve as a possible way to reconcile the discrepancy between the observed shallow slopes of the inner rotation curves of lsb galaxies and the predictions of cdm cosmological models . for face - on galaxies ,
the velocity dispersion increases with increasing dust mass due to scattering , but the effects are limited , even for extended dust distributions .
finally , we show that serious errors can be made when the individual velocities of the dust grains are neglected in the calculations .
dust , extinction galaxies : kinematics and dynamics galaxies : spiral radiative transfer |
we would like to thank holger gies and alvar sanchez for very helpful and interesting discussions .
e. r. williams , j. e. faller and h. a. hill , phys .
* 26 * ( 1971 ) 721 .
d. f. bartlett and s. loegl , phys .
* 61 * ( 1988 ) 2285 .
v. v. popov and o. v. vasilev , europhys .
lett . * 15 * ( 1991 ) 7 .
v. popov , turk .
j. phys .
* 23 * ( 1999 ) 943 .
s. andriamonje _ et al . _
[ cast collaboration ] , jcap * 0704 * ( 2007 ) 010 [ hep - ex/0702006 ] .
j. redondo , 0801.1527 [ hep - ph ] .
r. cameron _ et al . _
[ bfrt collaboration ] , phys .
d * 47 * ( 1993 ) 3707 . c. robilliard , r. battesti , m. fouche , j. mauchain , a. m. sautivet , f. amiranoff and c. rizzo [ bmv collaboration ] , phys .
* 99 * ( 2007 ) 190403 [ arxiv:0707.1296 [ hep - ex ] ] .
a. s. chou _ et al .
_ [ gammev ( t-969 ) collaboration ] , arxiv:0710.3783 [ hep - ex ] . m. ahlers , h. gies , j. jaeckel , j. redondo and a. ringwald , phys .
d * 76 * ( 2007 ) 115005 [ arxiv:0706.2836 [ hep - ph ] ] .
m. ahlers , h. gies , j. jaeckel , j. redondo and a. ringwald , phys .
d * 77 * ( 2008 ) 095001 [ arxiv:0711.4991 [ hep - ph ] ] .
g. mangano , a. melchiorri , o. mena , g. miele and a. slosar , jcap * 0703 * ( 2007 ) 006 [ astro - ph/0612150 ] .
k. ichikawa , m. kawasaki and f. takahashi , jcap * 0705 * , 007 ( 2007 ) , [ astro - ph/0611784 ] .
e. komatsu _ et al .
_ [ wmap ] , 0803.0547 [ astro - ph ] . j. jaeckel , j. redondo and a. ringwald , arxiv:0804.4157 [ astro - ph ] .
s. n. gninenko and j. redondo , phys .
b * 664 * , 180 ( 2008 ) , [ arxiv:0804.3736 [ hep - ex ] ] . j. jaeckel and a. ringwald , phys .
b * 659 * , 509 ( 2008 ) , [ arxiv:0707.2063 [ hep - ph ] ] .
u. seljak , a. slosar and p. mcdonald , jcap * 0610 * ( 2006 ) 014 [ astro - ph/0604335 ] . m. cirelli and a. strumia , jcap * 0612 * , 013 ( 2006 ) , [ astro - ph/0607086 ] .
j. hamann , s. hannestad , g. g. raffelt and y. y. y. wong , jcap * 0708 * ( 2007 ) 021 [ arxiv:0705.0440 [ astro - ph ] ] .
k. r. dienes , c. f. kolda and j. march - russell , nucl .
b * 492 * ( 1997 ) 104 [ hep - ph/9610479 ] .
d. lst and s. stieberger , hep - th/0302221 . s. a. abel and b. w. schofield , nucl .
b * 685 * ( 2004 ) 150 [ hep - th/0311051 ] .
b. batell and t. gherghetta , phys .
d * 73 * ( 2006 ) 045016 [ hep - ph/0512356 ] .
r. blumenhagen , s. moster and t. weigand , nucl .
b * 751 * ( 2006 ) 186 [ hep - th/0603015 ] .
s. a. abel , j. jaeckel , v. v. khoze and a. ringwald , hep - ph/0608248 .
s. a. abel , m. d. goodsell , j. jaeckel , v. v. khoze and a. ringwald , arxiv:0803.1449 [ hep - ph ] .
e. c. g. stckelberg , helv .
acta * 11 * ( 1938 ) 225 .
j. w. rohlf , `` modern physics from @xmath9 to @xmath98 '' , john wiley & sons , inc , 1994 . | we propose an experiment to search for extra hidden - sector " u(1 ) gauge bosons with gauge kinetic mixing with the ordinary photon , predicted by many extensions of the standard model .
the setup consists of a highly sensitive magnetometer inside a superconducting shielding .
this is then placed inside a strong ( but sub - critical ) magnetic field . in ordinary electrodynamics
the magnetic field can not permeate the superconductor and no field should register on the magnetometer . however , photon hidden - sector photon
photon oscillations would allow to penetrate the superconductor and the magnetic field would `` leak '' into the shielded volume and register on the magnetometer .
although this setup resembles a classic `` light shining though a wall experiment '' there are two crucial differences .
first , the fields are ( nearly ) static and the photons involved are virtual .
second , the magnetometer directly measures the field - strength and not a probability .
this improves the dependence of the signal on the kinetic mixing @xmath0 to @xmath1 instead of @xmath2 . in the mass range
@xmath3 the projected sensitivity is in the @xmath4 to @xmath5 range .
this surpasses current astrophysical and laboratory limits by several orders of magnitude ample room to discover new physics . in the near future
the lhc will commence searching for new particles with masses of the order of a tev .
this will test many proposed extensions of the standard model as , e.g. , supersymmetry , large extra dimensions , technicolor to name only a few .
however , many extensions of the standard model contain additional hidden sectors that interact only very weakly with ordinary matter . due to their feeble interactions even light particles in such hidden sectors
may be missed in such a collider experiment . yet
, it may be exactly these hidden sectors that carry crucial information on how the standard model is embedded in a more fundamental theory .
this creates the need for complementary experiments .
one interesting class of hidden - sector particles is extra u(1 ) gauge bosons . indeed
, many string theory models contain hidden - sector u(1)s under which standard model particles are uncharged . at low energies ,
the only renormalizable interaction with the visible sector can occur via mixing @xcite of the photon @xmath6 with the hidden sector photon @xmath7 .
current experimental bounds are shown in fig .
[ exclu ] .
interestingly there is a dip in the sensitivity in the mass region around @xmath8 .
hidden - sector photons in this mass region could , e.g. , explain @xcite the slight excess in the number of relativistic species observed in combinations of cmb with lyman-@xmath9 data @xcite ( although this could also be due to systematics @xcite ) .
moreover , one might ( wildly ) speculate about possible connections to neutrino masses and dark energy which incidentally also have energy scales around mev .
the experiment proposed in this note is designed to search for hidden - sector photons exactly in this region . and the blackbody nature of the spectrum ( black ) @xcite .
improvements of the solar bounds can be achieved using the superkamiokande detector or upgrading the cast experiment @xcite .
the region @xmath10 could be tested by an experiment using microwave cavities @xcite .
the lightly shaded areas bounded by lines give the projected sensitivity for the experiment proposed in this note .
the blue area corresponds to a relatively conservative estimate for the magnetometer sensitivity @xmath11 , and a thickness of the shielding @xmath12 much greater than the theoretical minimum required to have sufficient an external field of @xmath13 is assumed .
the red area is an optimistic scenario , @xmath14 , @xmath15 and @xmath16 .
( for both scenarios we used @xmath17 for the distance from the magnetic field source to the shield and @xmath18 for the distance from the shield to the magnetometer . )
[ exclu],width=8 ] the basic idea of the proposed experiment is very similar to a classic `` light shining through a wall experiment '' .
however , instead of light it uses a static magnetic field and the wall is replaced by superconducting shielding ( cf .
fig .
[ magnet ] ) . outside the shielding
we have a strong magnetic field . upon entering the superconductor the ordinary electromagnetic field
is exponentially damped with a length scale given by the london penetration depth @xmath19 . yet ,
due to the photon
hidden photon mixing a small part of the magnetic field is converted into a hidden magnetic field . after the superconducting shield is crossed the mixing turns a small fraction of the hidden magnetic field back into an ordinary magnetic field that can be detected by a magnetometer . since the magnetometer measures directly the field ( and not some probability or power output ) the signal is proportional to the transition amplitude and therefore to the mixing squared , @xmath1 , instead of being proportional to @xmath2 .
, width=8 ] let us start from the lagrangian describing the photon hidden - sector photon system at low energies , @xmath20 where @xmath21 is the field strength tensor for the ordinary electromagnetic u(1)@xmath22 gauge field @xmath23 , @xmath24 is its associated current ( generated by electrons , etc . ) and @xmath25 is the field strength for the hidden - sector u(1)@xmath26 field @xmath27 .
the first two terms are the standard kinetic terms for the photon and hidden photon fields , respectively . because the field strength itself is gauge invariant for u(1 ) gauge fields , the third term is also allowed by gauge and lorentz symmetry .
this term corresponds to a non - diagonal kinetic term , a so - called kinetic mixing @xcite .
this term is a renormalizable dimension four term and does not suffer from mass suppressions .
it is therefore a sensitive probe for physics at very high energies .
kinetic mixing arises in field theoretic @xcite as well as in string theoretic setups @xcite .
typical predicted values for @xmath28 in realistic string compactifications range between @xmath29 and @xmath30 .
the second to last term is a mass term for the hidden photon .
this could either arise from a higgs mechanism or it could be a stckelberg mass term @xcite .
finally , the last term corresponds to the london mass @xmath31 the photon acquires inside a superconductor . in vacuum @xmath32 . from this lagrangian
we obtain the equations of motion , @xmath33 here , we have chosen lorentz gauge for the massless field ( if @xmath34 ) and used that the lorentz condition is automatically enforced for massive fields . due to the kinetic mixing , a source @xmath35 originating from purely standard model fields , i.e. typically electrons , will automatically generate also a small @xmath27 field : very close to a localized source the the field equations are dominated by the derivative terms and the mass terms can be neglected .
accordingly the second equation in can only be fulfilled if the generated field is proportional to @xmath36 .
having said this , we will drop the source from now on .
taking appropriate derivatives in we get the equations for the field strengths , @xmath37 the essential features can be studied by using an one - dimensional setup : the superconductor fills the 2 - 3-plane and has a thickness @xmath38 .
the distances from the source of the magnetic field to the superconducting plane and from there to the detector will be called @xmath39 and @xmath40 , respectively .
note that the different space - time components of eq . decouple and
fulfill the same equations so essentially we can drop the spacetime indices . as an example let us choose a static magnetic field in the 3 direction , i.e. only the 12 components are non - vanishing , @xmath41 and @xmath42 .
we find , @xmath43 outside the superconductor @xmath34 and to lowest non - trivial order in @xmath28 the non - growing solutions are , @xmath44 in the @xmath45 basis .
as discussed above , close to the source the magnetic field will be @xmath46 .
however , we can now see that the small @xmath47 component is exponentially damped .
therefore , after a distance @xmath48 , the field is @xmath49 and will stay that way until it reaches the superconductor .
inside the superconductor the solutions are , @xmath50 note that @xmath51 is damped with @xmath52 and not @xmath53 . for @xmath54 the mainly hidden photon like @xmath55 can therefore `` leak '' through the superconducting shielding
. therefore , for @xmath56 ( distant source ) and using @xmath57 for simplicity , we have @xmath58 for @xmath57 and @xmath59 only the component @xmath60 effectively penetrates the shielding and @xmath61 at a distance @xmath62 after the shielding only the @xmath63 component survives and the measurable ordinary magnetic field approaches @xmath64 . keeping somewhat more carefully track of the exponentials we find , @xmath65 \left[1-e^{-{m_{\gamma^{\prime}}}l_2}\right]\ .
\end{aligned}\ ] ] the first part is the standard `` leaking '' expected from the finite thickness of the superconducting shielding , and constitutes the standard model background .
the second part @xmath66 is the signal caused by the hidden photons .
the first exponential arises from the damping of the hidden photon field inside the superconducting shield whereas the terms in brackets account for the fact that the conversion of photons into hidden photons and vice versa takes place over length scales @xmath67 .
the evolution of @xmath68 and @xmath47 from the source to the detector is sketched in fig .
[ evo ] .
( blue ) and hidden magnetic field @xmath47 ( red ) as a function of the distance from the source ( @xmath69 ) in our 1-dimensional set up .
both quantities are normalized to the value @xmath70 .
the dotted line represents the evolution of @xmath68 in absence of the hidden field ( @xmath71 ) . near the external ( left ) surface of the superconducting shield
@xmath47 changes sign , from negative to positive . for better visibility
we have chosen not too extreme values @xmath72 , @xmath73 , @xmath74 and @xmath75 .
for @xmath76 the evolution approaches @xmath1 , but for the chosen parameters the result is a bit smaller because the hidden field is a bit damped inside the shielding due to @xmath74 .
[ evo],width=8 ] let us now turn to the sensitivity of such an experiment .
magnetic field strengths @xmath77 of the order @xmath78 can be reached in the laboratory .
however , we have to stay below the critical field strength of the superconductor . in most materials
this ranges between @xmath79 and @xmath80 @xcite although fields as high as @xmath81 t can be shielded in certain cases ( cf .
, e.g. @xcite ) .
modern magnetometers @xcite can can detect magnetic fields as low as @xmath82 and @xmath83 seems relatively conservative .
accordingly we can expect a sensitivity in the @xmath5 to @xmath84 range magnetic field down to @xmath85 might be an experimental challenge ( see , however , @xcite ) . ] .
concerning the use of squids as magnetometers a comment is in order .
squids can only measure _ changes _ in the magnetic flux .
two possibilities to solve this problem are to modulate the strength of the external magnetic field or to change its direction ( for example by rotating it ) .
as long as the timescale for the change in the magnetic field is long compared to @xmath86 the above considerations remain unchanged . in addition , time - modulated magnetic fields might also be helpful in reducing the background .
finally , we have to ask which mass scales we can probe in the experiment . from we can read off that optimal sensitivity requires @xmath87 .
typical london penetration depths @xmath88 are of the order of @xmath89 ( cf , .
e.g. , @xcite ) . to avoid fields leaking directly through the shielding ( without having to convert into hidden fields ) at the @xmath90 level we need @xmath91 .
the requirement @xmath92 then allows , in principle , to search for masses up to @xmath93 . with thicker than the minimal required shielding
the experiment will be sensitive only to smaller masses . finally , the maximal size of the magnetic field and the shielded volume will typically be of the order @xmath94 .
this suggests a lower limit @xmath95 .
overall we have , @xmath96 two examples for the expected sensitivity are shown in fig . [ exclu ] . _ in conclusion : _ we have proposed a simple experiment to search for massive hidden photons .
the experiment could improve upon existing bounds by several orders of magnitude in a mass range @xmath97 where existing bounds are relatively weak .
it therefore bears significant discovery potential for hidden sector physics at the mev scale .
in particular , it could test recent speculations @xcite that the excess in the observed number of relativistic degrees of freedom at decoupling of the cmb originates from a resonant production of hidden photons in the early universe . |
due to the interaction of diffusion and reaction mechanisms , reaction - diffusion systems in fluctuating environments may develop collective behaviors that are very different from those occurring under well mixed conditions .
smoluchowski s theory @xcite quantifies the interaction of diffusion and reaction for a fast bimolecular reaction through an effective rate that is proportional to the molecular diffusion coefficient .
this approach is valid under well - mixed conditions .
spatial and temporal fluctuations may lead to the segregation of the reactants @xcite characterized by non - poissonian encounter processes and broad first - passage time distributions @xcite , such that reaction kinetics on small and large scales may be very different @xcite .
the sound understanding and quantification of the interaction of the reaction and fluctuation mechanisms play a central role in applications as diverse as contaminant degradation and chemical transformations in geological media @xcite and chemical kinetics in crowded intracellular environments @xcite . in this paper , we investigate the effective reaction dynamics in an environment undergoing both physical and chemical fluctuations characterized by broad distributions of transport and reaction times scales .
the physical picture is a particle decaying at rates affected by environmental factors fluctuating on the same scale as the mechanisms that lead to heavy tailed transition time distributions and thus subdiffusion . for a single homogeneous decay rate
, the particle distribution or survival probability decays exponentially in time @xcite .
the reaction kinetics are not affected by physical fluctuations and follow poissonian dynamics . in the opposite case of instantaneous spatially distributed traps , the effective reaction behavior as quantified by the survival probability
is fully determined by the transport characteristics @xcite .
( see also ref .
@xcite . ) here we study the interaction of fluctuations in transition and reaction times , and the resulting reaction - diffusion dynamics , with specific emphasis on the reaction kinetics .
a number of approaches have been proposed to model reaction behaviors in heterogeneous environments , including fractional kinetic orders and time - dependent rate coefficients @xcite as well as delay reaction equations @xcite .
oftentimes , effective kinetic equations to explain non - exponential survival probabilities are phenomenology - based lumped parameter models @xcite .
in contrast , in this paper , starting from explicit specification of microscopic fluctuations , we derive , solve , and interpret exact mesoscopic reaction - diffusion equations showing anomalous kinetics arising from emergent coupling of transport and chemical fluctuation .
_ reaction - diffusion in fluctuating environments_ we consider non - interacting particles of a chemical species @xmath0 that perform an unbiased one - dimensional random walk and at the same time undergo the reaction @xmath1 with random reaction rate @xmath2 . during the @xmath3th step ,
a live particle undergoes a spatial translation of random length @xmath4 during a random time @xmath5 and decays with random rate @xmath6 .
in particular , the particle survives the @xmath3th transition with probability @xmath7 . the position @xmath8 and time @xmath9 are given by @xmath10 and @xmath11 .
the sequences @xmath12 , @xmath13 , have common probability density function ( pdf ) @xmath14 and the set of all elements are mutually independent .
we define @xmath15 for convenience .
we denote by angle brackets averages with respect to all sequences @xmath16 as well as single copies @xmath17 .
the corresponding reaction times @xmath18 are distributed as @xmath19 .
equation constitutes a continuous time random walk ( ctrw ) @xcite subject to random decay .
we denote by @xmath20 the particle density of surviving particles at spatial coordinate @xmath21 at time @xmath22 .
thus , the normalized mean square displacement @xmath23 of the surviving particles is given by the second cumulant @xmath24 where @xmath25 is the fourier transform of @xmath20 , with respect to the first argument . in the following
, fourier and laplace transforms are distinguished from their space and time equivalents by the arguments @xmath26 and @xmath27 .
the survival probability @xmath28 of a particle is equal to the spatially integrated particle density @xmath29 .
its evolution describes the effective chemical kinetics and reaction dynamics .
we refer to reaction kinetics as anomalous if @xmath28 does not decay exponentially .
and ( green solid ) mean squared displacement @xmath23 defined in for heavy tailed transition time pdf , and power - law reaction time pdf with @xmath30 , @xmath31 , and @xmath32 .
left and right ordinate axes differ in physical dimensions , but are numerically equal .
( dotted ) exponential short time behavior of the survival probability , which is characterized by the average rate @xmath33 .
( lower dashed ) asymptotic power - law decay @xmath34 .
( dash - dotted ) short - time power - law behavior @xmath35 .
( upper dashed ) asymptotically constant @xmath23 occurring on the localization time scale @xmath36 . ]
_ anomalous kinetics and localization_ figure [ fig : pandmsd ] illustrates the survival probability @xmath28 and the mean square displacement @xmath23 for a heavy tailed transition time pdf that behaves as @xmath37 with @xmath38 , for @xmath22 larger than a characteristic time @xmath39 , and a rate pdf that behaves as @xmath40 with @xmath41 for @xmath42 smaller than a characteristic rate @xmath43 .
we observe two remarkable behaviors .
firstly , the survival probability decays as a power - law @xmath44 , where @xmath45 , and secondly @xmath23 increases as for passive particles until a characteristic reaction time scale after which it decays towards a constant .
this indicates localization of the particle distribution . the power - law decay of the survival probability observed in fig .
[ fig : pandmsd ] can be modeled by a non - linear kinetic rate law as @xcite @xmath46 with @xmath47 an effective reaction rate .
while this equation gives the power - law decay @xmath48 , we will see in the following that it does not describe the correct reaction kinetics for the system under consideration here .
we derive the governing equation for the reaction - diffusion system and analyze in detail the anomalous reaction behavior and localization of the particle distribution . _ reaction - diffusion dynamics_ here we present and interpret the results , outlining the calculations .
details of the calculations are given in appendix [ supps : calcs ] .
details of the simulations appearing in the figures are given in appendix [ app : mc ] .
the particle density @xmath49 is determined by [ ctrwr ] @xmath50 \rangle \psi(t - t ' ) .
\ ] ] this equation expresses the fact that the probability @xmath49 that a particle is at @xmath21 at time @xmath22 is given by the probability per time @xmath51 to have just arrived at any previous time @xmath52 times the probability @xmath53 to remain at @xmath21 until time @xmath22 times the probability @xmath54 \rangle$ ] to survive from @xmath55 to @xmath22 . the @xmath56 satisfies the chapman - kolmogorov type equation @xmath57 \rangle { \psi}(x - x',t - t ' ) , \end{aligned}\ ] ] which expresses particle balance under reaction losses .
the system can be combined into the generalized reaction - diffusion master equation @xmath58 , \label{gmer}\end{aligned}\ ] ] where we define the reaction kernel @xmath59 and the diffusion kernel @xmath60 through their laplace transforms as @xmath61}{s + { \rho } } \right\rangle } { \left\langle \frac{1 - { \psi}(s + { \rho})}{s + { \rho } } \right\rangle } , & & { \varphi}(x , s ) = \frac{\left\langle { \psi}(x , s + { \rho } ) \right\rangle}{\left\langle \frac{1 - { \psi}(s + { \rho})}{s + { \rho } } \right\rangle}.\end{aligned}\ ] ] integration of over space reveals the dynamics that govern the reaction kinetics , @xmath62 the reaction kinetics are non - local in time instead of non - linear as assumed by kinetic rate laws such as , and they depend in general on the diffusion history via the reaction kernel .
we analyze , , and for a broad class of reaction rate and waiting time distributions @xmath63 and @xmath64 showing _ weak and strong coupling _ of transport and kinetics .
strong coupling describes scenarios of emerging mesoscopic chemical kinetics in which reaction and diffusion are fully coupled .
the term _ weak coupling _ on the other hand refers to situations in which mesoscopic reaction and diffusion can be decoupled . in particular , @xmath65 where @xmath66 is the density of non - reacting particles . note that weak or strong coupling emerges on the mesoscopic level from microscopic dynamics rather than by varying a coupling parameter , or introducing dependence in the probability densities .
we now show that weak coupling occurs if diffusion is asymptotically normal , which is always the case if the mean transition time @xmath67 .
for weak coupling , the reaction and diffusion kernels simplify for @xmath68 to @xmath69 in this limit , reaction and diffusion separate and the reaction kinetics is described simply by a single rate : the mean reaction rate @xmath70 .
the generalized master equation reduces to the reaction - diffusion master equation @xmath71 , \label{mer}\end{aligned}\ ] ] which leads directly to .
obviously , weak coupling also occurs if the rate is constant , that is @xmath72 in this case , the reaction and diffusion kernels are given by @xmath73 where @xmath74 this gives the generalized reaction - diffusion master equation @xmath75 \times \exp[- r_{0 } ( t - t^\prime ) ] \left[p(x^\prime , t ) - p(x , t ) \right ] .
\label{gmer0}\end{gathered}\ ] ] although the density @xmath49 takes the form in this case , a heavy tailed waiting time pdf still gives a non - local kernel in , as discussed in ref .
@xcite . in both these weak coupling scenarios ,
the particle density is of the form .
this means that there is no localization of the particle density and the reaction kinetics are not affected by diffusion in any way .
the survival probability @xmath76 satisfies the first - order rate equation @xmath77 the effective reaction kinetics are determined solely by the characteristic reaction rate . on the mesoscopic level , the kinetics are effectively homogeneous in space .
note that this behavior is also observed , in general , at times shorter than a characteristic reaction time scale @xmath78 . in this case
, the reaction kernel also reduces to @xmath79 .
this is obtained from by considering the limit @xmath80 .
for a heavy tailed @xmath64 as in and a reaction - time pdf as in with @xmath32 .
( a ) @xmath81 , and ( from uppermost to lowermost curve ) @xmath82 .
( dashed ) asymptotic form .
filled circles indicate the localization time @xmath83 .
( b ) @xmath84 , with ( from uppermost to lowermost curve ) @xmath85 .
( dashed ) @xmath86 .
for both ( a ) and ( b ) , solid line curves are numerical inversion of , symbols are monte carlo simulations of the microscopic model . [
fig : varyalphanu ] , title="fig : " ] for a heavy tailed @xmath64 as in and a reaction - time pdf as in with @xmath32 .
( a ) @xmath81 , and ( from uppermost to lowermost curve ) @xmath82 . ( dashed ) asymptotic form .
filled circles indicate the localization time @xmath83 .
( b ) @xmath84 , with ( from uppermost to lowermost curve ) @xmath85 .
( dashed ) @xmath86 . for both ( a ) and ( b ) , solid line curves are numerical inversion of , symbols are monte carlo simulations of the microscopic model .
[ fig : varyalphanu ] , title="fig : " ] _ strong coupling_ we now analyze the more interesting case of strong coupling between microscopic reaction and diffusion , from which anomalous mesoscopic kinetics emerges .
because a transition time pdf with a finite mean leads to weak coupling , we employ the heavy tailed transition time pdf @xmath64 that behaves as @xmath87 for times larger than the microscopic time scale @xmath39 .
we also assume that the variance of the step distribution @xmath88 is finite . under these conditions ,
the laplace transform of for @xmath89 is given by @xmath90 .
likewise , the fourier transform of @xmath91 can be approximated for @xmath92 as @xmath93 , so that the joint pdf is @xmath94 . in the scaling limit @xmath95 and @xmath96 such that @xmath97 remains finite , we obtain the non - local diffusion - reaction equation @xmath98 where the laplace transforms of the reaction and diffusion kernels are @xmath99 these expressions manifest the intimate coupling between reaction and diffusion . in order to study this coupling ,
we focus now on rate pdfs that decay as a power - law for @xmath100 much smaller than a characteristic rate @xmath101 @xmath102 where @xmath103 .
this implies a power - law pdf of the reaction times @xmath104 for @xmath105 .
strong coupling occurs for all @xmath106 .
however , if we impose the restriction @xmath107 , and consider the limit @xmath108 , then kernels are given by @xmath109 and @xmath110 . under these conditions , becomes the fractional reaction - diffusion equation @xmath111 where @xmath112 and @xmath113 .
. has also been derived in @xcite for a model that effectively has @xmath114 .
likewise , the nature of reaction kinetics is evident in the fractional evolution equation for @xmath76 @xmath115 note that the case @xmath116 corresponds to a density of reaction rates with a point mass at the origin .
_ anomalous kinetics_ solving the kinetic equation for the kernel in laplace space , we obtain the compact expression @xmath117 as mentioned earlier , at short times @xmath118 , the survival probability is approximately exponential , @xmath119 . for long times @xmath105 , and all @xmath106
, we obtain by inversion of @xmath120 the form @xmath121 thus , as anticipated in the definition of the fractional - order derivative of , the exponent observed in fig .
[ fig : pandmsd ] is given by @xmath122 , which manifests again the intimate coupling of diffusion and reaction mechanisms .
in the limit @xmath123 , the particles never move , so the survival probability is just given by @xmath124 , which , for @xmath63 in , gives the decay @xmath125 . on the other hand , as @xmath126 ,
all the weight of @xmath63 becomes concentrated on zero and the particle does not decay .
the decay approaches @xmath127 , but the timescale of the onset of decay also diverges .
[ fig : varyalphanu ] shows the dependence of @xmath76 on @xmath128 and @xmath129 , and the excellent agreement with simulations of the microscopic model . with @xmath130 and @xmath131 as in fig .
[ fig : pandmsd ] , and @xmath132 ( black ) , @xmath133 ( yellow ) , @xmath134 ( blue ) .
the curve for @xmath135 is indistinguishable in this plot from the asymptotic distribution .
the inset shows the same curves on a semi - log scale .
curves are numerical inversion of . ]
_ localization_ localization in the strong coupling scenario is evident in the normalized mean square displacement @xmath23 , defined in , and normalized particle density @xmath136 in particular , they approach constant values at long times , as illustrated in fig .
[ fig : pandmsd ] and fig .
[ fig : propagator ] .
in contrast , in the weakly coupled case , implies that @xmath137 .
thus , @xmath23 evolves exactly as in the non - reactive case , and there is no localization . for strong coupling ,
we obtain from the laplace space solution @xmath138\end{aligned}\ ] ] as outlined above , for @xmath118 , the survival probability is approximately exponential and thus the propagator @xmath139 .
the density @xmath66 of non - reactive particles is obtained from by setting @xmath140 .
thus @xmath141 as in the non - reactive case .
( see fig .
[ fig : pandmsd ] . )
however , for times @xmath142 , can be be approximated by @xmath143 upon performing laplace - inversion we see that the only time dependence is in the factor @xmath76 .
thus , referring to , we see that the normalized density @xmath144 localizes , that is @xmath145 where we define the localization time @xmath146 and length @xmath147 via @xmath148 in the limit @xmath142 , we obtain @xmath149 .
the localization time @xmath146 marks the scale at which the mean square displacement crosses over from the power - law behavior @xmath150 to the constant value , as illustrated in fig . [ fig : pandmsd ] .
figure [ fig : propagator ] shows how the normalized particle density @xmath144 varies in time .
we clearly observe not only localization as a result of the interaction of reaction and subdiffusion , but also that @xmath49 is non - monotonic in @xmath22 .
_ lower cut - off in rate pdf_ suppose we shift @xmath151 by @xmath152 so that the probability that @xmath153 is zero .
we show in appendix [ app : shift ] that the propagator @xmath154 for the shifted reaction rates is @xmath155 where @xmath49 is the solution for the unshifted density @xmath63 .
note that this leaves @xmath144 unchanged , so that @xmath154 shows the same localization as @xmath49 .
however , the asymptotic survival probability now decays exponentially fast with the _ smallest _ rate @xmath156 .
for instance , for the power law @xmath63 and @xmath157 the survival probability follows the truncated power law @xmath158 . _
conclusions_ in conclusion , we find that incomplete mixing of reacting particles in a fluctuating environment resulting from a broad distribution of transition times and heterogeneity in the chemical properties can lead to anomalous reaction due to memory effects and localization of the particle density . for strong coupling between the physical and chemical fluctuations ,
we observe anomalous reaction kinetics , which manifest here in a power - law decay of the survival probability .
the reaction kinetics can not be represented by a non - linear , or fractal effective rate law , but are intrinsically non - local , which means the reactions have memory .
the reaction memory kernel is determined by both the physical and chemical system parameters , which emphasizes the intrinsic coupling between subdiffusion and reaction mechanisms .
this is also reflected in the spatial particle density , which localizes .
the evolution of the particle density is described in general by a non - local subdiffusion - reaction equation which , for heavy tailed reaction time densities reduces to a fractional diffusion - reaction equation and gives rise to fractional reaction kinetics .
these results shed light on the possible origins of anomalous reaction behaviors and the microscopic mechanisms that lead to non - local reaction kinetics .
in the following , we present details of the calculation of the results presented in the main body of the text .
we derive the integral equations for the density @xmath49 .
we then derive an explicit solution of @xmath49 in fourier - laplace space from these integral equations .
we then turn to the scaling limit , deriving @xmath49 in fourier - laplace space , asymptotic forms of various quantities , and the fractional reaction - diffusion equation . here
we derive from the definition of the microscopic model .
the particle motion is constructed from three sequences of random variables @xmath4 , @xmath159 , and @xmath160 with @xmath161 .
each sequence is i.i.d .
, and the collection of all random variables is mutually independent .
the positions @xmath8 and times @xmath9 at the @xmath3th step are determined by the recurrence relation @xmath162 and the rule that a particle decays at random rate @xmath6 for times @xmath22 satisfying @xmath163 .
in particular , a particle survives a transition from @xmath164 to @xmath165 with the probability @xmath166 .
we stress that the decay rate is random from step to step , but constant during one step . to complete the specification of the model ,
we take as initial condition @xmath167 concentrated on @xmath168 .
that is , the particle begins at the origin at time @xmath169 .
the @xmath170 , @xmath171 and @xmath172 are distributed according to common pdfs @xmath173 , @xmath174 , and @xmath63 , respectively , where we write @xmath175 for convenience .
the pdf @xmath173 has support on the real line , and pdfs @xmath174 and @xmath63 have support on the non - negative real numbers .
we assume that @xmath176 has finite variance . from the transition rules it follows that , at a given time @xmath22 , a particle that is at position @xmath8 has survived until the turning time @xmath9 with probability @xmath177 , and has survived the last time interval @xmath178 from the last turning point to the present time with probability @xmath179 $ ] , because the reaction rate @xmath2 changes at the turning point .
this implies that the particle density at time @xmath22 is given by @xmath180\bigg\rangle,\end{aligned}\ ] ] where the random variable @xmath181 is the number of steps performed up to time @xmath22 .
the angle bracket denotes average ( expectation ) with respect to all the random variables @xmath182 .
we employ indicator functions to partition the probability space into disjoint sets , so that the expectation becomes a sum of expectations @xmath183
\mathbb i(0 \leq t - t_n < \tau_n)\big\rangle.\end{aligned}\ ] ] recall that the indicator function @xmath184 is @xmath185 if the argument is true and @xmath186 otherwise .
we now separate explicitly the contributions up to the last turning point at time @xmath55 and during the final resting interval @xmath187 @xmath188 \big\rangle \int\limits_{t - t'}^\infty d \tau \psi(\tau).\end{aligned}\ ] ] the only factor inside the angle brackets that depends on @xmath6 is the last one . and
the only random variable in this last factor is @xmath6 .
the last factor is thus independent of the other factors .
thus the expectation of the product can be written as the product of expectations .
furthermore , @xmath6 has the same distribution as the single copy @xmath2 .
so , we can write @xmath189 \right\rangle \int\limits_{t - t'}^\infty d \tau \psi(\tau ) . \\ \nonumber\end{aligned}\ ] ] we now define @xmath190 and @xmath191 eq . is the joint probability density for a particle to arrive position @xmath21 at time @xmath22 on the @xmath3th step . with these definitions ,
is rewritten @xmath192 \rangle \int\limits_{t - t'}^\infty d \tau \psi(\tau).\end{aligned}\ ] ] we analyze @xmath193 by writing @xmath194 in the following form @xmath195 that this is indeed the expression for @xmath194 can be seen by performing the integrals and eliminating either one of the delta functions for @xmath196 and either one for @xmath55 , and using .
note that the only random variables appearing in the last three factors in the expectation are @xmath170 , @xmath171 and @xmath172 , while the first three factors depend only on random variables for @xmath197 .
the last three factors are thus independent of the first three and we can again factor the expectation .
furthermore , we use the fact that per the dirac delta @xmath198 we have @xmath199 , and that @xmath172 is independent of @xmath170 and @xmath171 .
thus , we can write @xmath200 \right\rangle .
\end{gathered}\ ] ] now , referring to , we identify the first factor in angular brackets with @xmath201 , the second factor with @xmath202 and the third factor with @xmath203 \rangle$ ] .
thus , we obtain @xmath204\rangle \eta_{n}(x',t ' ) . \\
\nonumber\end{aligned}\ ] ] summation over @xmath3 from @xmath186 to infinity on both sides of gives for @xmath56 @xmath205 \rangle { \psi}(x - x',t - t ' ) , \\
\nonumber\end{aligned}\ ] ] where we have used @xmath206 and our chosen initial condition @xmath207 .
integral equations and , appear in the main body of the paper as .
we derive the generalized master equation . to this end , we laplace transform and , which gives @xmath208 we now solve for @xmath209 and insert it into to obtain @xmath210 we now rewrite the left side tautologically as @xmath211 we define here @xmath212}{s + \rho}\right\rangle , \\ \nonumber \phi(x ,
s ) & = \left\langle \frac{1 - { \psi}(s + \rho)}{s + \rho } \right\rangle^{-1 } \left\langle { \psi}(x , s+\rho)\right\rangle.\end{aligned}\ ] ] using in and isolation of @xmath213 on the left gives , @xmath214 .
\end{gathered}\ ] ] inverse laplace transform of the latter gives the generalized master equation .
here , we calculate the propagator and survival probability exactly in fourier - laplace space .
we use the laplace transform shift formula , the convolution formula , and linearity of the expectation to write the fourier - laplace transform of @xmath215 using both the fourier and laplace convolution formulas , we write the fourier - laplace transform of as @xmath216 solving these two equations for @xmath217 , we obtain a montroll - weiss equation generalized for random rates @xmath218 equation is the exact solution of our random rate model in fourier - laplace space for all @xmath21 and @xmath22 .
the survival probability is found by setting @xmath219 in @xmath220 we consider the effect of shifting the random reaction rate @xmath221 .
inserting this shift into and gives @xmath222 { p}_c(s ) & = \frac{\left\langle \frac{1 - { \psi}(s + { \rho}+ r_c)}{s + { \rho}+ r_c } \right\rangle}{1-\langle { \psi}(s + { \rho}+ r_c ) \rangle}. \end{aligned}\ ] ] inverse fourier - laplace and inverse laplace transform , respectively give by using the shift theorem of the laplace transform @xmath223 p_c(t ) & = \exp(-r_c t ) p(t),\end{aligned}\ ] ] where @xmath49 and @xmath76 are propagator and survival probability for @xmath224 .
this derives .
the normalized mean square displacement is given by @xmath225 \nonumber = & - { p}(k = 0,t)^{-1 } \left.\frac{\partial^2 { p}(k , t)}{\partial k^2}\right|_{k = 0}.\end{aligned}\ ] ] note that the ( un - normalized ) mean square displacement @xmath226 is given by @xmath227 we assume , as we have in all explicit examples , that we have an `` uncoupled '' walk ; ie @xmath228 , and @xmath173 has finite variance . under this assumption ,
we obtain from the explicit laplace space expression @xmath229 where @xmath230 .
the normalized mean square displacement @xmath23 can then be found by inverting and .
here we give details of the calculation of quantities in the scaling limit .
we employ the transition time distribution that behaves as @xmath231 for @xmath22 larger than the microscopic time scale @xmath39 . to leading order in @xmath27 ,
the laplace transform of is @xmath232 .
one readily obtains from the laplace - space expression for the survival probability in the scaling regime for the heavy tailed distribution @xmath233 the @xmath3th moment of the particle position @xmath234 is given by @xmath235 we obtain the laplace transform of @xmath236 in the scaling regime directly from .
@xmath237^{-\frac{n}{2}},\\ \quad n = 0,2,4,\ldots.\end{gathered}\ ] ] as mentioned in the text , the fourier transform of @xmath91 for for @xmath92 is @xmath93 .
together with @xmath232 , we find the joint pdf is @xmath94 . in the scaling limit
, the montroll - weiss equation becomes @xmath238 the non - local ( in time ) reaction - diffusion equation considered in the text is @xmath239 where the laplace transforms of the reaction and diffusion kernels are @xmath240 one can take the fourier - laplace transform of , insert the explicit forms of the kernels , and rearrange terms , in a manner similar to the derivation in sec .
[ sec : gme ] , and arrive at .
thus , @xmath49 in the scaling limit indeed solves .
we now show a specific rate pdf for which the operators in reduce to standard liouville fractional derivatives .
we consider the rate distribution @xmath241 where @xmath242 .
we shall consider the case @xmath243 , for which it is possible to write a reaction - diffusion equation using fractional derivatives . to leading order
, the numerator in @xmath244 is @xmath245 .
the fractional nature of the kernels comes from the factor in the denominator @xmath246 . using , we have @xmath247 for @xmath248 , the last integral converges while the factor @xmath249 remains near @xmath185 .
but , @xmath250 diverges . replacing @xmath249 by @xmath251 ,
the integral can be evaluated , and we find @xmath252 thus , the reaction kernel @xmath253 and @xmath254 .
these kernels provide the laplace space definitions of the fractional derivative of order @xmath255 in . to determine the asymptotic form of the survival probability for @xmath256 we consider the laplace space expression .
we treated the case where the numerator diverges above .
but , we show that , even in the case that neither numerator nor denominator diverge as @xmath257 , the behavior in this limit is determined by the non - analyticity in the numerator .
suppose @xmath258 , with @xmath259 .
it is easy to see that all derivatives of @xmath260 of order less than @xmath26 tend to a constant as @xmath261 , but the @xmath26th derivative diverges , @xmath262 & \times \int_0^\infty ( 1+z)^{\alpha - k-1 } z^{\nu-1 } { { p_{{\rho}}}^*}({s}z / r_0 ) { \ensuremath{\operatorname{d}\!{z}}},\end{aligned}\ ] ] where @xmath263 is the falling factorial .
again , for small @xmath264 , we have @xmath265 where we have used @xmath266 and @xmath267 , where @xmath268 is the rising factorial .
we then invert , using @xmath269 , for @xmath270 and @xmath271 , where @xmath272 is the @xmath26th derivative .
we find @xmath273 thus , we find the survival probability @xmath274 the laplace space expression for the particle density is given by @xmath275}{2 \sqrt{\frac{\mathcal d_\alpha { { \mathcal k}_d}(s)}{s^{2}+ { { \mathcal k}_r}(s)}}}.\end{aligned}\ ] ] using the kernels , we obtain @xmath276 } { 2\sqrt{\frac{\mathcal d_\alpha}{\langle ( s + { \rho})^\alpha \rangle}}}. \end{aligned}\ ] ] for @xmath248 , we find , following the argument in the previous section , that the divergence is in @xmath120 , and not in the remaining factors .
thus , we can set @xmath277 , which gives @xmath278}{2\sqrt{\frac{\mathcal d_\alpha}{\langle r^\alpha \rangle}}}. \end{aligned}\ ] ] all the time dependence is in the factor @xmath120 .
thus , the normalized particle density tends to a constant profile : it localizes .
we describe the mc algorithm in detail below . here
we discuss the mc data used in specific plots .
equation is the expression used for those curves in fig .
[ fig : varyalphanu ] that were obtained by numerical inversion .
the corresponding monte carlo ( mc ) curves in the same figure were obtained by averaging @xmath279 trials of the microscopic model .
the rates were sampled from a gamma distribution with mean @xmath185 , and a microscopic timescale @xmath280 .
thus , systematic deviations of the mc from the scaling limit are not visible on the plot . for @xmath281 , and @xmath282 .
( blue points ) numerical inversion of with @xmath283 .
( violet points ) numerical inversion of with @xmath284 .
( black crosses ) mc with @xmath283 .
( black squares ) mc with @xmath284 . ] as another example , fig .
[ fig : msdmc ] shows the normalized msd for the rate pdf @xmath281 , and a truncated power law for @xmath174 with @xmath285 .
the figure compares mc of the microscopic model with numerical inversion of expansions of and , showing excellent agreement .
in fact , in order for the mc and inversion to agree for these parameters , we inverted numerically an expansion of keeping one more term than we did in obtaining . here
we present the method and algorithm used to compute an mc estimate of the survival probability @xmath76 from simulations .
the survival probability can be written @xmath286 where @xmath287 is the probability density for the death time , and @xmath288 is the cumulative distribution function ( cdf ) of the death time .
to make clear the meaning of death time : the probability that a particle dies between time @xmath22 and @xmath289 is @xmath290 , given that is alive and untrapped at time @xmath169 .
we compute the empirical cdf corresponding to @xmath288 , which is an unbiased estimator that converges to @xmath288 @xcite .
the empirical cdf of the cdf @xmath291 of a random variable @xmath292 is computed as follows .
1 ) draw @xmath3 independent samples of @xmath292 , storing each one in an array @xmath0 in sampling order .
2 ) sort the array @xmath0 in increasing order . in particular ,
after sorting , the first element @xmath293 is the least sample and the last element @xmath294 is the greatest sample .
the empirical cdf is given by the points @xmath295 . to be clear ,
the coordinate is @xmath296 and the ordinate is @xmath297 .
referring to , the empirical survival probability is given by @xmath298 .
we now describe how the death time is sampled .
let @xmath299 , @xmath300 , be a sequence of rates and times sampled with a pseudo - random number generator ( rng ) from the pdfs @xmath301 and @xmath174 .
note that @xmath302 and @xmath303 are real numbers here , not random variables .
assume the particle is alive at time @xmath304 .
then it survives the interval @xmath305 with probability @xmath306 .
we sample a deviate @xmath307 uniform on @xmath308 $ ] with the rng . if @xmath309 , then the particle has survived the interval .
otherwise , the particle has died at some time in the interval @xmath305 .
if the particle survives , then we repeat this procedure for the next step numbered @xmath310 , advancing time to @xmath311 , etc .
suppose the latter case , i.e. the particle died in the interval @xmath305 .
we define @xmath312 .
we must sample the death time from the distribution for the the particle to die at time @xmath313 conditioned on it being alive at @xmath314 and dead at time @xmath315 .
the cdf for this conditional distribution is @xmath316 for @xmath317 and @xmath186 otherwise .
it is well known that a deviate @xmath21 from an invertible cdf @xmath291 may be sampled by generating a deviate @xmath307 of the uniform distribution on the unit interval and computing @xmath318 .
inverting , we sample this conditional death time @xmath313 by computing @xmath319),\end{aligned}\ ] ] where @xmath307 is a uniform deviate on the unit interval .
the death time for this particle is then @xmath320 .
this time is stored in the array @xmath0 when constructing the empirical cdf , i.e. the complement of @xmath76 .
the efficiency and statistical error may be improved by using the analytically determined average value of @xmath313 computed from , rather than sampling .
we compute mc estimates as follows .
the time increments of a particle s trajectory and its death time are generated as in the previous section .
however , we also track the position of the walker at each step , using normally distributed step displacements with unit variance . in order to perform an ensemble average over trajectories
, we must establish an array of fixed times @xmath321 at which to record the msd .
we also maintain an array @xmath322 of the same length .
the element @xmath323 contains the sum over particle trajectories of the squared displacement recorded at time @xmath324 .
we consider , as before , two cases .
either the walker dies during a step , or does not .
consider the second case .
the walker is at position @xmath325 during the time interval @xmath305 .
we maintain an index @xmath326 into the array @xmath321 corresponding the most recent time at which the squared displacement for this trajectory was recorded .
we then check which of the recording times @xmath327 lie in the interval @xmath305 . for each of the corresponding indices
@xmath328 , etc .
we add @xmath329 to the element @xmath330 , etc .
we advance @xmath326 to the last recorded time .
we then proceed to the next step .
now consider the first case , when the particle dies .
the relevant time interval is now @xmath331 , because the particle dies at @xmath332 .
we only record @xmath329 at recording times lying in this final interval . in both cases
, we also increment the number of particle trajectories @xmath333 contributing to the sum at each recording time @xmath324 .
this number of course decreases with increasing time because particles are dying .
the estimate of the msd normalized by the survival probability is then @xmath334 .
suppose the total number of trials is @xmath3 .
the estimate of the msd normalized by the total number of particles , live or dead is of course @xmath335 .
the relation to the analytic quantities is @xmath336 and @xmath337 as an example to understand the difference : suppose only live particles are detectable .
then @xmath23 describes the observed width of the cloud .
note that the corresponding estimate of @xmath76 is @xmath338 , so that we have , as expected @xmath339 . in practice ,
we instead estimate @xmath76 using the method described in the previous section , which is far more efficient .
25ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) link:\doibase 10.1016/0301 - 0104(78)85052 - 6 [ * * , ( ) ] @noop ( ) http://www.cambridge.org/us/academic/subjects/physics/statistical-physics/diffusion-and-reactions-fractals-and-disordered-systems[__ ] ( , , ) @noop * * , ( ) link:\doibase 10.1016/j.jconhyd.2010.05.002 [ * * , ( ) ] link:\doibase 10.1016/j.pbiomolbio.2004.01.012 [ * * , ( ) ] link:\doibase 10.1103/physreve.73.031102 [ * * , ( ) ] , link:\doibase 10.1103/physreve.74.031116 [ * * , ( ) ] link:\doibase 10.1103/physreve.77.041115 [ * * , ( ) ] link:\doibase 10.1063/1.443832 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.51.79 [ * * , ( ) ] @noop * * , ( ) `` , '' ( , ) chap . , pp .
link:\doibase 10.1007/bf01010846 [ * * , ( ) ] link:\doibase 10.1006/jtbi.1995.0181 [ * * , ( ) ] link:\doibase 10.1073/pnas.0503858102 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.250601 [ * * , ( ) ] , link:\doibase 10.1371/journal.pone.0052029 [ * * ( ) , 10.1371/journal.pone.0052029 ] link:\doibase 10.1016/0255 - 2701(89)87007 - 2 [ * * , ( ) ] link:\doibase 10.1063/1.1704269 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) | we study reaction - diffusion systems with broadly distributed , independent , physical and chemical fluctuations .
we find subdiffusion at short times , and localization at long times .
the total mass decays as a power , obeying a fractional reaction equation .
the density obeys a fractional reaction - diffusion equation .
the memory kernels decay as a power , with exponent depending on both transport and chemical parameters , manifesting emergent coupling between microscopic physical and chemical fluctuations .
these systems provide a mechanism for power - law kinetics in systems with disordered rates .
= 1 |
consider the cauchy problem for a scalar conservation law in one space variable @xmath1 where @xmath2 , @xmath3 smooth ( by _ smooth _ we mean at least of class @xmath4 ) .
it is well known @xcite that there exists a unique entropic solution @xmath5 satisfying @xmath6 in particular we can assume w.l.o.g . that for each @xmath7 , @xmath8 and @xmath9 uniformly bounded . the solution
to can be constructed in several ways , for example nonlinear semigroup theory @xcite , finite difference schemes @xcite , and in particular wavefront tracking @xcite and glimm scheme @xcite . in the scalar case ,
in fact , the above functionals are lyapunov functionals , i.e. functionals decreasing in time , ( or _ potential _ , as they are usually called in the literature ) for all the schemes listed , yielding compactness of the approximated solution ( even in the multidimensional case ) .
in @xcite it is shown that there exists an additional lyapunov functional @xmath10 $ ] to the ones of . in the case of wavefront tracking solution ( see the references above or the beginning of section [ sect_wavefront ] for a short presentation ) , outside the interaction - cancellation times @xmath11 , this functional takes the simple form @xmath12 = \sum_{w \not= w ' } |\sigma(w ) - \sigma(w')| |w| |w'|,\ ] ] where @xmath13 , @xmath14 are wavefronts of the approximate solution @xmath5 traveling with speed @xmath15 , @xmath16 respectively . to avoid confusion / misunderstanding ,
we will call _ wavefronts _ the shocks / rarefactions / contact discontinuities of the ( approximate ) solution , while the word _ wave _ will be reserved to subpartitions of wavefronts .
the precise definition is given at the beginning of section [ ss_main_results ] .
here and in the future , we call the collision between two wavefronts @xmath13 , @xmath14 an _ interaction _ if the wavefronts which collide have the same sign , namely @xmath17 , otherwise we call it a _ cancellation _ ( see definition [ w_int_canc_points ] ) . in @xcite a general form of @xmath18 valid
at all @xmath19 is given .
this functional is of fundamental importance when studying the existence of solutions to systems of conservation laws in one space variable , where the total variation of the ( approximate ) solution is not decreasing in time . in particular , when two wavefronts @xmath13 , @xmath14 interact in the point @xmath20 , it is possible to show that @xmath18 decreases of at least @xmath21 where the speeds are computed at @xmath22 . since it is well known that @xmath23 it is customary to say that @xmath18 is a _
cubic functional _ , referring precisely to the exponent of .
it follows in particular that the _ total amount of interaction _ is cubic , @xmath24 ( see @xcite for the general definition . ) in order to prove a convergence rate estimate for the glimm schemes , in @xcite , @xcite it is shown that if @xmath25 is the approximate solution constructed by the glimm scheme and @xmath26 is the entropic solution to a systems of conservation laws in one space dimension , then @xmath27 under the assumption that the following estimate holds : @xmath28 ( here and in the following @xmath29 is a constant which can depend on the @xmath30 norm of @xmath9 , but not on the initial datum @xmath31 . )
the above estimates is written in the case that only two wavefronts at a time interact for simplicity , the general form is presented in the statements of theorem [ main_thm ] ( for the glimm scheme ) and theorem [ w_main_thm ] ( for the wavefront tracking algorithm ) . the key idea to prove estimate is to introduce a suitable functional @xmath32 , depending on the time , which decreases in time and is of quadratic order with respect to the total variation of the solution , and then to use this functional to get estimate . in the case of
genuinely nonlinear or linearly degenerate systems , because of the particular structure of solutions to the riemann problem , the interaction functional @xmath33 introduced by glimm @xcite gives the estimate @xmath34 and by the lipschitz regularity of @xmath15 ( inequality ) it is immediate to deduce .
however the functional @xmath18 introduced above can not produce a quadratic estimate , being cubic as observed earlier .
since 2006 many attempts have been made in order to get a proof of estimate ( in addition to @xcite , @xcite , see also @xcite )
. however , the proofs presented in @xcite , @xcite ) have some problems , as it is shown by the counterexamples in @xcite .
on the other hand , we discover an incorrect estimate in the proof of @xcite , precisely in lemma 2 , pag .
614 , formulas ( 4.84 ) , ( 4.85 ) . an explicit counterexample is presented in appendix [ app_conter ] , here we just notice that the 2 cited formulas try to estimate terms which are linear in the total variation ( e.g. the difference in speed across interactions ) with the decrease of the cubic functional @xmath18 defined in .
the main result of this paper is a new and correct proof of the estimate for approximate solutions constructed by wavefront tracking or by the glimm scheme , in the case of scalar conservation laws .
our aim is to simplify as much as possible the technicalities in order to single out the ideas behind our approach . in a forthcoming paper
we will study the general vector case . in order to state
precisely the two main theorems of this paper ( one referring to the glimm scheme , the other one to the wavefront tracking ) , we need to introduce what we call an _ enumeration of waves _ in the same spirit as the _ partition of waves _ considered in @xcite , see also @xcite
. roughly speaking , we assign an index @xmath35 to each piece of wave , and construct two functions @xmath36 , @xmath37 which give the position and the speed of the wave @xmath35 at time @xmath19 , respectively .
more precisely , let @xmath25 be the glimm approximate solution , with grid points @xmath38 : for definiteness we assume @xmath25 to be right continuous in space .
consider the interval @xmath39 $ ] , which will be called the _ set of waves_. in section [ s_front_glimm ] we construct for the glimm scheme a function @xmath40 \\ & & ( t , s ) & \mapsto & \mathtt x(t , s ) \end{array}\ ] ] with the following properties : 1 .
the set @xmath41 is of the form @xmath42 with @xmath43 : define @xmath44 as the set @xmath45 2 .
the function @xmath46 is increasing , @xmath47-lipschitz and linear in each interval @xmath48 if @xmath49 ; 3 . if @xmath49 , then @xmath50 for some @xmath51 , i.e. it takes values in the grid points at each time step ; 4 .
for @xmath52 such that @xmath53 it holds @xmath54 5 .
there exists a time - independent function @xmath55 , the _ sign _ of the wave @xmath35 , such that @xmath56 for all @xmath49 .
the last formula means that for all test functions @xmath57 it holds @xmath58 the fact that @xmath59 means that the wave has been removed from the solution @xmath25 by a cancellation occurring at time @xmath60 .
formula and a fairly easy argument , based on the monotonicity properties of the riemann solver and used in the proof of lemma [ w_lemma_eow ] , yield that to each wave @xmath35 it is associated a unique value @xmath61 ( independent of @xmath19 ) by the formula @xmath62 } \mathcal s(s ) ds.\ ] ] we finally define the _ speed function _
@xmath63 \cup \{+\infty\}$ ] as follows : if @xmath64 , then @xmath65}f\big)\big(\hat u(s)\big ) & \text{if } \mathcal{s}(s ) = + 1 , \\
\big(\frac{d}{du}\operatorname{conc}_{[u_{\varepsilon}(t_n,\mathtt x(t_n , s)),u_{\varepsilon}(t_n,\mathtt x(t_n , s)-)]}f\big)\big(\hat u(s)\big ) & \text{if } \mathcal{s}(s ) = -1 . \end{array } \right.\ ] ] in other words , to the wave @xmath66 and for @xmath67 we assign the speed given by the riemann solver in @xmath68 to the wavefront containing the value @xmath61 .
we can now state our theorem for glimm approximate solutions .
[ main_thm ] the following estimate holds : @xmath69 in the case of wavefront tracking , since the waves @xmath35 have size @xmath70 , with @xmath71 the discretization parameter and @xmath72 , it is possible to choose @xmath73 , and in section [ front_waves ] it is shown that the function @xmath74 defined in satisfies slightly different properties : property ( 3 ) is meaningless , and property ( 5 ) holds for all @xmath75 . the speed @xmath76 is now defined as @xmath77 outside the interaction / cancellation points , and it is extended to @xmath78 by right - continuity .
notice that outside interaction - cancellation times , the strength of the wavefront @xmath13 at @xmath79 is given by @xmath80 i.e. the strength of the wavefront @xmath13 is the sum of the strength of all waves @xmath35 which are mapped by @xmath74 into @xmath81 .
the main estimate for the wavefront tracking solution is contained in the following result . [ w_main_thm ] the following holds : if @xmath11 are the interaction - cancellation times , then @xmath82 where @xmath83 is _ the strength of the wave @xmath35_. as it is shown in the proof of theorem [ w_decreasing ] , formula , the estimate yields immediately for wavefront tracking solution .
the corresponding computation for glimm scheme is given in the proof of theorem [ decrease_thm_enunciato ] , formula .
we observe here that for the interaction of two wavefronts @xmath13 , @xmath14 , the quantity @xmath84 has a nice geometric interpretation ( see figure [ fig : figura36 ] ) , in the same spirit as the area interpretation of the cubic functional @xmath18 [ see bianchini - bressan curve shortening ] .
in fact , if @xmath13 , @xmath14 correspond to the jumps @xmath85 $ ] , @xmath86 $ ] with @xmath87 , then it is fairly easy to see that is equal to the height of the triangle @xmath88 , @xmath89 , @xmath90 , more precisely @xmath91 for the glimm scheme , each interaction involves the several wavefronts ( not just two ) , and it corresponds to replacing two adjacent riemann problems , namely @xmath85 $ ] and @xmath86 $ ] , with the riemann problem @xmath92 $ ] : the corresponding quantity is then given by @xmath93 } f(u^m),\ ] ] assuming again @xmath87 for definiteness ( see figure [ fig : figura19 ] ) . in this way , one can rewrite also for the glimm scheme . in the wavefront tracking.,width=302,height=188 ] in the glimm scheme , width=377,height=264 ] our choice to present two separate theorems is motivated by the following facts . first of all , due to the number of papers present in the literature concerning this estimate , we believe necessary to give a correct proof in the most simple case , i.e. wavefront tracking for scalar conservation laws .
since however the estimate has been used mainly to prove the convergence rate of the glimm scheme , we felt necessary to offer also a direct proof for this approximation scheme .
it turns out that even if the fundamental ideas are the same , the two proofs are sufficiently different in some points to justify a separate analysis . in our opinion , in fact , it is not trivial to deduce one from the other . as observed in @xcite , one of the main problems in obtaining an estimate of the form , is that the study of wave interactions can not be local in time , but one has to take into accoun the whole sequence of interactions - cancellations for every couple of waves .
this is a striking difference with the glimm interaction potential @xmath94 , where the past history of the solution is irrelevant .
previous attempts tried to adapt glimm s idea of finding a _ quadratic potential @xmath95 _ which is decreasing in time and at every interaction has a jump of the amount .
apart technical variations , the idea is to transform the function @xmath18 of into @xmath96 : = \sum_{w \not= w ' } \frac{|\sigma(w ) -
\sigma(w')| |w| |w'|}{|w| + |w'|}.\ ] ] for monotone initial data @xmath31 , this functional is sufficient ; however in @xcite f. ancona and a. marson show that @xmath95 defined above may be not bounded , so that in @xcite they consider the functional @xmath97 where @xmath98 is the position of the wave @xmath13 at time @xmath19 .
notice that at the time of interaction of the wavefronts @xmath13 , @xmath14 one has @xmath99 ( there are no wavefronts between the two interacting ) , so that for the couple of waves @xmath13 , @xmath14 one has @xmath100 if the flux function @xmath101 has no finite inflection points , the waves @xmath102 can join in wavefronts ( because of an interaction ) and split again ( because of a cancellation ) an arbitrary large number of times .
this implies that the functional @xmath103 ( as well as the other quadratic functionals introduced in the literature ) does not decay in time , but can increase due to cancellations .
hence , instead of proving directly that the quadratic functional @xmath104 controls the interactions , in @xcite the authors consider a term @xmath105 which bounds the oscillations of @xmath104 for the waves not involved in an interaction , and prove that @xmath106 } \frac{|\sigma - \sigma'| s s'}{s + s ' } \leq ( q^\text{{am}}(t_1 ) + g(t_1 ) ) - \big ( q^\text{{am}}(t_2 ) + g(t_2 ) \big).\ ] ] since @xmath103 is quadratic by construction because of the lipschitz regularity of the speed @xmath15 ( inequality ) and @xmath107 , @xmath108 , they reduce the quadratic interaction estimate to the following estimate : @xmath109 our approach is slightly different : we construct a quadratic functional @xmath110 such that its total variation in time is bounded by @xmath111 , and at any interaction decays at least of the quantity ( or more precisely of the quantities in the l.h.s . of or concerning that interaction ) . the functional can increase due to cancellations , but in this case we show that its positive variation is controlled by the total variation of the solution times the amount of cancellation . being the total variation a lyapunov functional , it follows that @xmath112 so that , being @xmath113 , the functional @xmath114 has total variation of the order of @xmath115 . in particular , @xmath116 the estimates , concerning cancellations is much easier ( and already done in the literature , see @xcite ) , and we present it in propositions [ w_canc_3 ] , [ canc_3 ] , depending on the approximation scheme considered . in the case of cancellations , in fact , there is a first order functional decreasing , namely the total variation . for simplicity we present the sketch of the proof in the wavefront tracking case
we define again a functional @xmath0 of a form similar to , @xmath117 but with @xmath118 main differences . 1 .
first of all its definition involves the waves @xmath35 , not the wavefronts @xmath13 .
[ point_glimm ] if the waves @xmath35 , @xmath119 have not yet interacted , then the weight @xmath120 is a large constant . in our case
, it suffices @xmath121 .
+ if the waves have already interacted , then the weight @xmath122 has the form @xmath123 3 . in the above formula
, the quantity @xmath124 is the difference in speed given to the waves @xmath35 , @xmath119 by an artificial riemann problem , which roughly speaking collects all the previous common history of the waves @xmath35 , @xmath119 .
this makes @xmath0 not local in time and space .
the denominator of is not the total variation of @xmath26 between two wavefronts @xmath13 , @xmath14 ( containing @xmath35 , @xmath119 respectively ) as it is the case in , but only the total variation between the two waves @xmath35 , @xmath119 . observe that @xmath125 by the lipschitz regularity of the speed given to the waves by solving a riemann problem , it is fairly easy ( and proved in section [ w_functional_q ] ) that @xmath126 we observe first that the functional @xmath110 restricted to the couple of waves which have never interacted is exactly ( apart from the constant @xmath121 ) the original glimm functional @xmath127 : = \sum_{w \not= w ' } |w| |w'|.\ ] ] since the couple of waves which have never interacted is decreasing , this part of the functional is decreasing , and as observed before it is sufficient to control the quadratic estimates for couple of waves which have never interacted . the choice of the denominator in yields that our functional @xmath128 is not affected by the issue of large oscillations , as observed in @xcite , even if its form is similar to @xmath103 : indeed , in our case
, the denominator we choose does not depend on the shock component which the waves belong to , and thus cancellations do not affect it .
next , the riemann problem used to compute the quantity @xmath124 is made of all waves @xmath129 which have interacted with both waves @xmath35 and @xmath119 .
we now show how it evolves with time .
interaction : : if at @xmath130 an interaction occurs , and @xmath35 , @xmath119 are not involved in the interaction , the set of waves which have interacted with both is not changing ( since they are separated , at most one of them is involved in the interaction ! ) , which means that @xmath131 .
if @xmath35 , @xmath119 are involved in the interaction , then the couple disappears from the sum , because when two wavefronts with the same sign interact a single wavefront comes out .
cancellation : : if a cancellation occurs at @xmath130 , then one can check that again if both @xmath35 , @xmath119 are not involved in the wavefront collision then @xmath132 is constant .
otherwise the change in @xmath132 corresponds to the change in speed obtained by removing some waves in a riemann problem , and adding all these variations one obtains that the oscillation of @xmath110 can be estimated explicitly as in the case of a single cancellation ( i.e. the total variation of the solution times the amount of cancellation ) . the cancellation case , corresponding to the positive total variation in time of @xmath110 , is thus controlled by @xmath133 and since @xmath113 and @xmath134 it follows that @xmath135 when an interaction between @xmath13 , @xmath14 occurs , the discussion above shows that we can split the waves involved in the interaction into @xmath118 sets : 1 . a set @xmath136 of waves in @xmath13 which have never interacted with the waves in @xmath14 ; 2 . a set @xmath137 of waves in @xmath13 which have interacted with a set @xmath138 of waves in @xmath14 ; 3 . a set @xmath139 of waves in @xmath14 which have never interacted with the waves in @xmath13 .
the speed assigned by the artificial riemann problems for all couples of waves @xmath140 , @xmath141 yields that the decrease of @xmath110 at the interaction time is ( larger than the one ) given by the wave pattern depicted in figure [ fig : figura37 ] .
.,width=415,height=226 ] by an explicit computation one can check that @xmath142 thus obtaining the desired estimate . the proof for the glimm scheme follows the same philosophy , but , as we said , due to the different structure of the approximating scheme it present different technical aspects .
the paper is organized as follows .
section [ s_convex_env ] provides some useful results on convex envelopes .
part of these results are already present in the literature , others can be deduced with little effort .
we decided to collect them for reader s convenience .
two particular estimates play a key role in the main body of the paper : the regularity of the speed function ( theorem [ convex_fundamental_thm ] and proposition [ convex_fundamental_thm_affine ] ) and the behavior of the speed assigned to a wave by the solution to riemann problem @xmath92 $ ] when the left state @xmath143 or the right state @xmath144 are varied ( propositions [ tocca ] , [ vel_aumenta ] , [ differenza_vel ] and [ incastro ] ) .
the next two sections contain the main results of the paper . as we said , for technical reasons
the proofs differ depending on the approximation scheme considered , but , in order to simplify the exposition , we tried to keep a similar structure of both sections .
section [ sect_wavefront ] is devoted to the proof of theorem [ w_main_thm ] .
after recalling how a wavefront approximated solution @xmath25 is constructed , we begin with the construction of the wave map @xmath74 in section [ front_waves ] . as we said , due to the fixed size @xmath71 of the waves , the set of waves @xmath145 will be a finite subset of @xmath146 , and we replace the properties of @xmath74 given at the beginning of this introduction with the ( easier to work with ) definition of _ enumeration of waves _ , definition [ w_eow ] .
this is the triple @xmath147 , where @xmath74 is the position of the waves @xmath35 and @xmath148 is its right state .
the equivalence of the two definitions is straightforward . in section [ w_pswaves ]
we show that it is possible to construct a function @xmath36 such that at any time @xmath149 is an enumeration of waves , with @xmath148 independent on @xmath19 .
once we have an enumeration of waves , we can start the proof of theorem [ w_main_thm ] ( section [ section_w_main_thm ] ) .
first we study the estimate when a single cancellation occurs .
this estimate is standard , since the cancellation is controlled by the decay of a first order functional , namely @xmath150 .
the precise estimate is reported in proposition [ w_canc_3 ] , where the dependence w.r.t . @xmath150 and
@xmath121 is singled out .
corollary [ w_canc_4 ] completes the estimate for the case of cancellation points .
the rest of section [ sect_wavefront ] is the construction and analysis of the functional @xmath110 described above , in order to prove proposition [ w_thm_interaction ] .
this proposition proves for the case of interaction points , completing the proof of theorem [ w_main_thm ] .
as we said , one of the main features of the enumeration of waves is that we can speak of couple of waves which have already interacted . in section [ w_waves_collision ]
we prove some important properties of these couples of waves .
lemma [ w_interagite_stesso_segno ] shows that they must have the same sign , and , because in the scalar case no new waves are created , all the waves between @xmath35 and @xmath119 have interacted with @xmath35 and @xmath119 , lemma [ w_quelle_in_mezzo_hanno_int ] . in this section
it is also defined the interval of waves @xmath151 , which is the set of waves @xmath152 which have interacted with both @xmath35 and @xmath119 .
propositions [ w_unite_realta ] and [ w_divise_tocca ] prove that , even if the solution to the riemann problem generated by the waves @xmath151 assigns artificial speeds to @xmath35 , @xmath119 , the property of being separated or not in the real solution can be deduced from the solution to the riemann problem @xmath151 . from this fact
we can infer a lot of interesting properties of the riemann problem @xmath151 : the most important one is that we can know the values @xmath26 where @xmath153 . in section [ w_functional_q ]
we write down the functional @xmath110 and conclude the proof of theorem [ w_main_thm ] .
we study separately the behavior of @xmath110 at interactions and cancellations .
theorem [ w_decreasing ] proves that the functional @xmath110 decreases at least of the quantity at a single interaction point , while theorem [ w_increasing ] shows that the increase of @xmath110 at each cancellation point is controlled by the total variation of the solution times the cancellation .
these two facts conclude the proof of proposition [ w_thm_interaction ] , as shown in section [ sss_sketch_proof ] .
section [ s_glimm_scheme ] is devoted to the proof of theorem [ main_thm ] .
as we said , the ideas of the proof are similar , but from the technical point of view there are substantial differences , making this case slightly more complicated . in this introduction
we will underline these variations , so that the reader can easily pass from one proof to the other , also because we tried to keep the structure of the two main sections similar .
as we already said , the first main difference is in the definition of enumeration of waves , definition [ eow ] .
in fact , for the glimm scheme the set of waves @xmath145 is a subset of an interval in the real line ( namely @xmath154 $ ] ) , and moreover the map @xmath46 has to pass trough the grid points .
this forces us to define the speed @xmath37 of a wave @xmath35 at time @xmath19 not by just taking the time - derivative of @xmath36 , but by considering the real speed given to @xmath35 by the riemann problem at each grid point as in .
this analysis is done in section [ pswaves ] .
another difference is that the glimm scheme , due to the choice of the sampling points , `` interacts '' with the solution ( i.e. it may alter the curve @xmath46 ) even when no real interaction - cancellation occurs .
this is why we need to study an additional case , namely when no interaction / cancellation of waves occurs at a given grid point , and this is done in proposition [ p_no_interaction ] : the statement is that trivially nothing is going on in these points , but we felt the need of a proof .
proposition [ canc_3 ] , namely the case of cancellation points , is analog to the wavefront tracking case . also the structure and
properties of the functional @xmath110 we are going to construct in the glimm scheme case are similar to the wavefront approximation analysis , the main difference being that several interactions and cancellations occur at each time step .
we thus require that the set of pairs of waves present in the solution at time step @xmath155 ( i.e. not moved to @xmath156 ) , namely @xmath157 , can be split into two parts : one part concerns the interactions , and decreases of the right amount , the other one concerns cancellations and increases of a quantity controlled by the total variation of the solution times the amount of cancellation .
the remain part of the section is the proof of these two estimates , from which one deduces theorem [ main_thm ] along the same line outlined in section [ sss_sketch_proof ] .
first , we define the notion of waves which have already interacted , definition [ d_glimm_interacted ] , and waves which are separated ( or _ divided _ ) , definition [ waves_divided ] .
notice that even if they occupy the same grid position for some time step , they are considered divided in the real solution if the riemann problem at that grid point assigns different speeds to them .
the statement that the artificial riemann problems we consider separate waves as in the real solution is completely similar to the wavefront case , proposition [ unite_realta ] , but the proof is quite longer . in the last section , section [ ss_glimm_funct_q ]
we define the functional @xmath110 , and due to the continuity of the set of waves @xmath145 we show a regularity property of the weight @xmath158 so that no measurability issues arise .
the proof of the two estimates ( interactions and cancellations ) is somehow longer than in the wavefront tracking case but it is based on the same ideas , and this concludes the section . in appendix [ app_conter ] we present a counterexample to formula ( 4.84 ) in lemma 2 , pag .
614 , of @xcite , which justifies the need of the analysis in the one - dimensional case . for usefulness of the reader ,
we collect here some notations used in the subsequent sections .
* @xmath159 , @xmath160 ; * @xmath161 ( resp .
@xmath162 ) is the left ( resp .
right ) derivative of @xmath163 at point @xmath26 ; * if @xmath164 is a sequence of real numbers , we write @xmath165 ( resp
. @xmath166 ) if @xmath164 is increasing ( resp . decreasing ) and @xmath167 ; * given @xmath168 , we will denote equivalently by @xmath169 or by @xmath170 the @xmath171-dimensional lebesgue measure of @xmath172 . *
if @xmath173 \to { \mathbb{r}}$ ] , @xmath174 \to { \mathbb{r}}$ ] are two functions which coincide in @xmath175 , we define the function @xmath176 \to { \mathbb{r}}$ ] as @xmath177 , \\
h(x ) & \text{if } x \in [ b , c ] . \end{array } \right.\ ] ] * sometime we will write @xmath178 instead of @xmath179 ( resp . @xmath180 instead of @xmath78 ) to emphasize the symbol of the variables ( resp . @xmath81 or @xmath19 ) we refer to . * for any @xmath3 and for any @xmath181 , the _ piecewise affine interpolation of @xmath101 with grid size @xmath71 _ is the piecewise affine function @xmath182 which coincides with @xmath101 in the points of the form @xmath183 , @xmath51 . * by @xmath184 we mean a quantity which does not depend on the data of the problem , neither on @xmath101 nor on the initial datum @xmath185 . * for any @xmath186 , @xmath187^+$ ] denotes its positive part .
in this section we define the convex envelope of a continuous function @xmath3 in an interval @xmath188 $ ] and we prove some related results .
the first section provides some well - known results about convex envelopes , while in the second section we prove some propositions which will be frequently used in the paper .
the aim of this section is to collect the statements we will need in the main part of the paper . in particular , the most important results are theorem [ convex_fundamental_thm ] and proposition [ convex_fundamental_thm_affine ] , concerning the regularity of convex envelopes , and proposition [ tocca ] , proposition [ vel_aumenta ] , corollary [ stesso_shock ] , proposition [ incastro ] and proposition [ diff_vel_proporzionale_canc ] , referring to the behavior of convex envelopes when the interval @xmath188 $ ] is varied : these estimates will play a major role for the study of the riemann problems .
[ convex_fcn ] let @xmath3 be continuous and @xmath188 \subseteq { \mathbb{r}}$ ] .
we define _ the convex envelope of @xmath101 in the interval @xmath188 $ ] _ as @xmath189}f ( u ) : = \sup\bigg\{g(u ) \ \big| \
g : [ a , b ] \to { \mathbb{r}}\text { is convex and } g \leq f\bigg\}.\ ] ] a similar definition holds for _ the concave envelope of @xmath101 in the interval @xmath188 $ ] _ denoted by @xmath190}f$ ] .
all the results we present here for the convex envelope of a continuous function @xmath101 hold , with the necessary changes , for its concave envelope . in the same setting of definition [ convex_fcn ] , @xmath191}f$ ]
is a convex function and @xmath191}f(u ) \leq f(u)$ ] for each @xmath192 $ ] .
the proof is straightforward . adopting the language of hyperbolic conservation laws , we give the next definition .
let @xmath101 be a continuous function on @xmath179 , let @xmath188 \subseteq { \mathbb{r}}$ ] and consider @xmath191}f$ ] .
a _ shock interval _ of @xmath191}f$ ] is an open interval @xmath193 $ ] such that for each @xmath194 , @xmath191}f(u ) < f(u)$ ] .
a _ maximal shock interval _ is a shock interval which is maximal with respect to set inclusion .
notice that , if @xmath192 $ ] is a point such that @xmath191}f(u ) < f(u)$ ] , then , by continuity of @xmath101 and @xmath191}f$ ] , it is possible to find a maximal shock interval @xmath195 containing @xmath26 .
it is fairly easy to prove the following result .
[ shock ] let @xmath3 be continuous ; let @xmath188 \subseteq { \mathbb{r}}$ ] .
let @xmath195 be a shock interval for @xmath191}f$ ] .
then @xmath191}f$ ] is affine on @xmath195 .
the following theorem provides a description of the regularity of the convex envelope of a given function @xmath101 .
[ convex_fundamental_thm ] let @xmath101 be a @xmath196-function .
then : 1 .
[ convex_fundamental_thm_1 ] the convex envelope @xmath191 } f$ ] of @xmath101 in the interval @xmath188 $ ] is differentiable on @xmath188 $ ] ; 2 .
[ convex_fundamental_thm_2 ] for each @xmath197 , if @xmath198}f(u)$ ] , then @xmath199}f(u);\ ] ] 3 .
[ convex_fundamental_thm_3 ] @xmath200 } f$ ] is lipschtitz - continuous with lipschitz constant less or equal than @xmath201 . by ` differentiable on @xmath188 $ ] ' we mean that it is differentiable on @xmath202 in the classical sense and that in @xmath203 ( resp .
@xmath175 ) the right ( resp .
the left ) derivative exists .
while the proof is elementary , we give it for completeness .
( [ convex_fundamental_thm_1 ] ) and ( [ convex_fundamental_thm_2 ] ) .
let @xmath204 } f$ ] . since @xmath205 is convex
, then it admits left and right derivatives at each point @xmath206 .
moreover in @xmath203 ( resp .
@xmath175 ) it admits right ( resp
. left ) derivative . for each @xmath206
, it holds @xmath207 in order to prove that @xmath205 is differentiable at @xmath31 it is sufficent to prove that equality holds in .
if @xmath208 , then , by proposition [ shock ] , @xmath31 lies in a shock interval and so clearly @xmath205 is differentiable at @xmath31 and the derivative is locally constant . hence we assume that @xmath206 is a point such that @xmath209 .
we claim that @xmath210 and @xmath211 . by , this is sufficient to prove that @xmath205 is differentiable at @xmath31 and that @xmath212 .
assume by contradiction that @xmath213 .
then , by definition of left derivative , and by the fact that @xmath209 , there exist @xmath214 , such that for each @xmath215 , @xmath216 by taylor expansion , for each @xmath217 , @xmath218 where @xmath219 is any quantity which goes to zero faster than @xmath220
. there exists @xmath221 such that for each @xmath222 @xmath223 and so by , @xmath224 , a contradiction , since @xmath225}f(u ) \leq f(u)$ ] for each @xmath26 . in a similar way one can prove that @xmath211 and thus @xmath226 ( [ convex_fundamental_thm_3 ] ) .
let @xmath227 $ ] , @xmath228 . since @xmath205 is convex and differentiable , @xmath229 .
we want to estimate @xmath230 , so let us assume that @xmath231 .
then there exist @xmath232 $ ] such that @xmath233 and @xmath234 , @xmath235 . indeed , if @xmath236 , then you can choose @xmath237 ; if @xmath238 , you choose @xmath239 , where @xmath195 is the maximal shock interval which @xmath26 belongs to ( recall that we already know that @xmath205 is differentiable ) . in a similar way you choose @xmath240 and it holds @xmath241 .
hence @xmath242}\ ] ] and so @xmath243 is lipschitz continuous with lipschitz constant less or equal than @xmath244 .
a similar result holds for the piecewise affine interpolation of a smooth function @xmath101 .
[ convex_fundamental_thm_affine ] let @xmath181 be fixed .
assume @xmath245 .
let @xmath101 be a smooth function and let @xmath246 be its piecewise affine interpolation with grid size @xmath71 .
then the derivative @xmath200 } f_{\varepsilon}$ ] of its convex envelope is a piecewise constant increasing function defined on @xmath188 \setminus { \mathbb{z}}{\varepsilon}$ ] , which enjoys the following lipschtitz - like property : for any @xmath247 in @xmath248 such that @xmath249 $ ] @xmath250 } f_{\varepsilon}\bigg ) \big ( ( ( m'-1){\varepsilon } , m'{\varepsilon } ) \big ) - \bigg(\frac{d}{du}\operatorname{conv}_{[a , b ] } f_{\varepsilon}\bigg ) & \big ( ( ( m-1){\varepsilon } , m{\varepsilon } ) \big ) \\ & \leq \lvert f''\lvert_{l^\infty(a , b ) } \big ( m'{\varepsilon}- ( m-1){\varepsilon}\big ) .
\end{split}\ ] ] arguing as in point ( [ convex_fundamental_thm_3 ] ) of theorem [ convex_fundamental_thm ] , assume that the l.h.s . of is strictly positive .
let @xmath251 , where @xmath195 is the maximal shock interval which @xmath252 belongs to ( if such an interval @xmath195 does not exist , set @xmath253 ) .
by definition of convex envelope @xmath254 } f_{\varepsilon}\big((m_1 - 1){\varepsilon}\big)\ ] ] and by definition of maximal shock interval @xmath255 } f_{\varepsilon } ( m_1{\varepsilon}).\ ] ] hence @xmath256 } f_{\varepsilon}\bigg ) \big ( ( ( m-1){\varepsilon } , m{\varepsilon } ) \big ) \geq&~ \bigg(\frac{d}{du } f_{\varepsilon}\bigg)\big ( ( ( m_1 - 1){\varepsilon } , m_1{\varepsilon})\big ) \\
= & ~ \frac{f_{\varepsilon}(m_1{\varepsilon } ) - f_{\varepsilon}((m_1 - 1){\varepsilon})}{{\varepsilon } } \\
\text{(by definition of $ f_{\varepsilon}$ ) } = & ~ \frac{f(m_1{\varepsilon } ) - f((m_1 - 1){\varepsilon})}{{\varepsilon } } \\ = & ~ f'(\xi ) , \end{split}\ ] ] for some @xmath257 .
similarly you can find @xmath258 such that @xmath259 and @xmath260 } f_{\varepsilon}\bigg ) \big ( ( ( m'-1){\varepsilon } , m'{\varepsilon } ) \big ) \leq f'(\xi ' ) . \ ] ] hence @xmath261 } f_{\varepsilon}\bigg ) \big ( ( ( m'-1){\varepsilon } , m'{\varepsilon } ) \big ) - \bigg(\frac{d}{du}\operatorname{conv}_{[a , b ] } f_{\varepsilon}\bigg ) & \big ( ( ( m-1){\varepsilon } , m{\varepsilon } ) \big ) \\ \text{(by \eqref{sinistra } and \eqref{destra } ) } \leq&~ f'(\xi ' ) - f'(\xi ) \\
\leq&~ \lvert f''\lvert_{l^\infty(a , b ) } ( \xi ' - \xi ) \\
\leq&~ \lvert f''\lvert_{l^\infty(a , b ) } \big [ m'{\varepsilon}- ( m-1){\varepsilon}\big ] , \end{split}\ ] ] concluding the proof .
we are now able to state some useful results about convex envelopes , which we will frequently use in the following sections .
[ tocca ] let @xmath3 be continuous and let @xmath262 . if @xmath191}f(\bar{u } ) = f(\bar{u})$ ] , then @xmath189}f = \operatorname{conv}_{[a,\bar u]}f \cup \operatorname{conv}_{[\bar u , b]}f.\ ] ] we have to prove that @xmath189}f|_{[a , \bar u ] } = \operatorname{conv}_{[a , \bar u]}f\ ] ] and @xmath189}f|_{[\bar u , b ] } = \operatorname{conv}_{[\bar u , b]}f.\ ] ] let @xmath263}f|_{[a , \bar u]}$ ]
. by contradiction , assume @xmath264}f$ ] .
then there exists a function @xmath163 defined on @xmath265 $ ] , convex , such that @xmath266 on @xmath265 $ ] and such that @xmath267 for some @xmath268 .
then a direct verification yields that @xmath269}f|_{[\bar u , b]}$ ] is a convex function , and it is less or equal then @xmath101 on @xmath188 $ ] .
hence , by definition of convex envelope , @xmath270}f|_{[\bar u , b]})(\tilde u ) \leq \operatorname{conv}_{[a , b]}f(\tilde u ) = h(\tilde u )
< g(\tilde u),\ ] ] a contradiction . in a similar way one can prove that @xmath191}f|_{[\bar u , b ] } = \operatorname{conv}_{[\bar u , b]}f$ ] .
[ rp_ridotto ] let @xmath3 be continuous and let @xmath262 .
assume that @xmath271 belongs to a maximal shock interval @xmath272 with respect to @xmath191}f$ ] .
then @xmath273}f|_{[a , u_1 ] } = \operatorname{conv}_{[a , b]}f|_{[a , u_1]}$ ] .
it is an easy consequence of proposition [ tocca ] , just observing that by maximality of @xmath274 , @xmath191}f(u_1 ) = f(u_1)$ ] .
[ vel_aumenta ] let @xmath3 be continuous ; let @xmath262
. then 1 .
@xmath275}f\big)(u+ ) \geq \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u+)$ ] for each @xmath276 ; 2 .
@xmath275}f\big)(u- ) \geq \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u-)$ ] for each @xmath277 $ ] ; 3 .
@xmath278}f\big)(u+ ) \leq \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u+)$ ] for each @xmath279 ; 4 .
@xmath278}f\big)(u- ) \leq \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u-)$ ] for each @xmath280 $ ] .
we prove only the first point , the other ones being similar .
let @xmath281 .
if @xmath191}f(\bar u ) = f(\bar u)$ ] , then by proposition [ tocca ] , @xmath191}f|_{[a , \bar u ] } = \operatorname{conv}_{[a,\bar u]}f$ ] and then we have done . otherwise , if @xmath191}f(\bar u ) < f(\bar u)$ ] , denote by @xmath274 the maximal shock interval of @xmath191}f$ ] , such that @xmath282 .
hence , by corollary [ rp_ridotto ] , @xmath191}f|_{[a , u_1 ] } = \operatorname{conv}_{[a,\bar u]}f|_{[a , u_1]}$ ] .
thus if @xmath283 , we have completed . for @xmath284 ,
if @xmath285}f\big)(u_1 + ) < \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1+),\ ] ] since @xmath273 } f(u_1 ) = \operatorname{conv}_{[a , b]}f(u_1)$ ] , then @xmath273 } f < \operatorname{conv}_{[a , b]}f$ ] on @xmath286 for some @xmath287 sufficiently small , but this is a contradiction , by definition of convex envelope . finally , if @xmath288 , then @xmath289}f\big)(u+ ) \geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a,\bar{u}]}f\big)(u_1
\geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1 + ) \\
= & ~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u ) .
\end{split}\ ] ] we have used that @xmath290 } f$ ] is affine on @xmath291 $ ] , by proposition [ shock ] .
[ differenza_vel ] let @xmath3 be continuous ; let @xmath262
. then 1 . for each @xmath292 , @xmath293 , @xmath294}f\big)(u_2 + ) - \big ( & \frac{d}{du}\operatorname{conv}_{[a,\bar{u}]}f\big)(u_1 + ) \\
\geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_2 + ) - \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1 + ) ; \end{aligned}\ ] ] 2 . for each @xmath295
$ ] , @xmath293 , @xmath294}f\big)(u_2- ) - \big(&\frac{d}{du}\operatorname{conv}_{[a,\bar{u}]}f\big)(u_1- ) \\
\geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_2- ) - \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1-);\end{aligned}\ ] ] 3 . for each @xmath296 , @xmath293 , @xmath297}f\big)(u_2 + ) -\big(&\frac{d}{du}\operatorname{conv}_{[\bar{u},b]}f\big)(u_1 + ) \\
\geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_2 + ) - \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1+);\end{aligned}\ ] ] 4 . for each @xmath298
$ ] , @xmath293 , @xmath297}f\big)(u_2- ) -\big(&\frac{d}{du}\operatorname{conv}_{[\bar{u},b]}f\big)(u_1- ) \\
\geq&~ \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_2- ) - \big(\frac{d}{du}\operatorname{conv}_{[a , b]}f\big)(u_1-).\end{aligned}\ ] ] easy consequence of previous proposition .
[ stesso_shock ] let @xmath3 be continuous and let @xmath262 .
let @xmath299 $ ] , @xmath300 .
if @xmath301 belong to the same shock interval of @xmath273}f$ ] , then they belong to the same shock interval of @xmath191}f$ ] .
[ incastro ] let @xmath3 be smooth ; let @xmath302 be real numbers .
assume that 1 .
@xmath303}f(u_2 ) = f(u_2)$ ] ; 2 .
@xmath304}f(u_3 ) = f(u_3)$ ] . then @xmath305}f(u_2 ) = f(u_2)$ ] .
let us first prove the following lemma
. in the same setting of proposition [ incastro ] , let @xmath306 \to { \mathbb{r}}$ ] , @xmath307 \to { \mathbb{r}}$ ] be convex functions . if @xmath308 on @xmath309 $ ] , then @xmath310}$ ] is convex
. immediate consequence of the fact that ` being convex ' is a local property .
let us now prove the proposition . since @xmath303}f(u_2
) = f(u_2)$ ] , by corollary [ rp_ridotto ] , @xmath311}f(u_2 ) = f(u_2)$ ] ; hence @xmath312 } f = \operatorname{conv}_{[u_1,u_2 ] } f \cup \operatorname{conv}_{[u_2,u_3 ] } f.\ ] ] moreover , by proposition [ tocca ] , @xmath313 } f = \operatorname{conv}_{[u_2,u_3 ] } f \cup \operatorname{conv}_{[u_3,u_5 ] } f.\ ] ] hence , by previous lemma , @xmath314}$ ] is convex ; moreover , by definition , @xmath315 and , for this reason , @xmath316}f$ ] .
hence @xmath317}f(u_2 ) = g(u_2 ) = \varphi(u_2 ) \leq \operatorname{conv}_{[u_1,u_5]}f ( u_2)\leq f(u_2)\ ] ] and so the thesis follows .
[ convex_approximation ] let @xmath101 be a continuous function on @xmath179 .
let @xmath318 and let @xmath319 , @xmath320 , two sequences such that @xmath321 .
assume that for each @xmath320 there is @xmath322 $ ] such that @xmath323 } f(u_k ) = f(u_k).\ ] ] if @xmath324 , then @xmath189 } f(\bar u ) = f(\bar u).\ ] ] by simplicity , assume @xmath325 for each @xmath326 , the general case being entirely similar .
by contradiction , assume @xmath191 } f(\bar u ) < f(\bar u)$ ] .
there is a maximal shock interval @xmath327 containing @xmath31 and w.l.o.g .
we can assume @xmath328 .
if @xmath329 , for @xmath326 sufficiently large , @xmath330 . hence @xmath331 }
f(\tilde b ) \overset{\text{(cor .
\ref{rp_ridotto})}}{= } \operatorname{conv}_{[a , b_k ] } f(\tilde b),\ ] ] and thus @xmath332 } f(u_k ) \overset{\text{(prop .
\ref{tocca})}}{= } \operatorname{conv}_{[a,\tilde b ] } f(u_k ) \overset{\text{(prop .
\ref{tocca})}}{= } \operatorname{conv}_{[a , b ] } f(u_k).\ ] ] passing to the limit , we get a contradiction . on the other hand ,
if @xmath333 , one can find a spherical neighborhood @xmath334 of @xmath335 in @xmath336 and a spherical neighborhood @xmath337 of @xmath338 in @xmath336 with the following property : for any @xmath339 and for any @xmath340 , if @xmath341 is the line joining @xmath342 and @xmath343 , then @xmath344 .
now observe that by continuity of @xmath101 , for @xmath326 sufficiently large , @xmath345 and @xmath346 .
hence , denoting by @xmath347 the line joining @xmath342 and @xmath348 , we get @xmath349 } f(u_k ) = f(u_k ) > r_k(u_k),\ ] ] a contradiction , since @xmath350}$ ] is a convex function whose graph contains points @xmath351
. [ diff_vel_proporzionale_canc ] let @xmath101 be a smooth function , let @xmath352 .
then @xmath353 } f\bigg)(\bar u- ) - \bigg(\frac{d}{du } \operatorname{conv}_{[a , b]}f\bigg)(\bar u ) \leq \lvert f '' \lvert_{l^\infty(a , b ) } ( b - \bar u).\ ] ] moreover , if @xmath246 is the piecewise affine interpolation of @xmath101 with grid size @xmath71 , it holds @xmath353 } f_{\varepsilon}\bigg)(\bar u - ) - \bigg(\frac{d}{du } \operatorname{conv}_{[a , b]}f_{\varepsilon}\bigg)(\bar u- ) \leq \lvert f '' \lvert_{l^\infty(a , b ) } ( b - \bar u).\ ] ] let us first prove the inequality for @xmath101 smooth . set @xmath354}f$ ] , @xmath355}f$ ] .
let @xmath356 $ ] be the maximal shock interval of @xmath163 which @xmath31 belongs to ( if it does not exist , the proof is trivial , because of theorem [ convex_fundamental_thm ] ) .
let @xmath341 be the line passing through @xmath335 with slope @xmath357 and let @xmath358 be the first coordinate of the intersection point between @xmath359 and @xmath163 in the interval @xmath360 $ ] .
it holds @xmath361 and @xmath362 moreover define @xmath363 clearly @xmath364 and thus @xmath365 hence @xmath366 now observe that there must be @xmath367 $ ] such that @xmath368 .
indeed , if there exists a strictly increasing sequence @xmath369 converging to @xmath31 such that @xmath370 } f(u_n)$ ] , then by theorem [ convex_fundamental_thm ] , point [ convex_fundamental_thm_2 ] , @xmath371 for each @xmath171 ; passing to the limit one obtains @xmath372 .
otherwise , if such a sequence does not exist , then one can easily find @xmath373 such that @xmath374 is a maximal shock interval .
this means that @xmath375 for some @xmath376 .
moreover , since @xmath377 there must be @xmath378 such that @xmath379 .
hence , from , we obtain @xmath380 concerning the piecewise affine case , substituting @xmath101 with @xmath246 one obtains the same inequality .
now observe that as in the smooth case one can find @xmath378 such that @xmath379 .
moreover it is also easy to see that there must be @xmath381 $ ] such that @xmath382 . using this ,
one concludes the proof as in the smooth case .
in this section we prove the main interaction estimate for an approximate solution of the cauchy problem obtained by the wavefront tracking algorithm , thus giving a proof of theorem [ w_main_thm ] .
first we briefly recall how such an approximate solution is constructed , mainly with the aim to set the notations used later on . for
any @xmath383 let @xmath246 be the piecewise affine interpolation of @xmath101 with grid size @xmath71 ; let @xmath384 an approximation of the initial datum @xmath31 ( in the sense that @xmath385 in @xmath386-norm , as @xmath387 ) of the cauchy problem , such that @xmath384 has compact support , it takes values in the discrete set @xmath388 , and @xmath389 let @xmath390 be the points where @xmath384 has a jump .
at each @xmath391 , consider the left and the right limits @xmath392
. solving the corresponding riemann problems with flux function @xmath246 , we thus obtain a local solution @xmath393 , defined for @xmath19 sufficiently small . from each
@xmath391 some wavefronts supported on discontinuity lines of @xmath25 , referred also as shocks ( positive or negative , according to the sign of the jump ) or contact discontinuities , emerge .
when two ( or more ) discontinuity lines supporting wavefronts meet ( we will refer to this situation as an _ interaction - cancellation _ ) , we can again solve the new riemann problem generated by the interaction - cancellation , according to the above procedure , with flux @xmath246 , since the values of @xmath394 always remain within the set @xmath388 .
the solution is then prolonged up to a time @xmath395 where other wavefronts meet , and so on .
+ one can prove @xcite , @xcite that the total number of interaction - cancellation points is finite , and hence the solution can be prolonged for all @xmath396 , thus defining an approximate solution @xmath397 , piecewise constant , with values in the set @xmath388 .
let @xmath398 , @xmath399 , be the point in the @xmath400-plane where an interaction - cancellation between two ( or more ) wavefronts occurs in the approximate solution @xmath25 .
let us suppose that @xmath401 and for every @xmath402 exactly two discontinuities meet in @xmath403 .
this is a standard assumption , achieved by slightly perturbing the wavefront speed .
we also set @xmath404 . in this section
we define the notion of _ wave _ , the notion of _ position of a wave _ and the notion of _ speed of a wave_. by definition of wavefront solution , for each time @xmath396 , @xmath394 is a piecewise constant function , which takes values in the set @xmath388 .
hence @xmath405 is an integer multiple of @xmath71 . in this section
we define the notion of _ enumeration of waves _ related to a function @xmath406 of the single variable @xmath81 : in the following sections , @xmath26 will be the piecewise constant , @xmath71-approximate solution of the cauchy problem for fixed time @xmath19 , considered as a function of @xmath81 .
[ w_eow ] let @xmath407 , @xmath408 , be a piecewise constant , right continuous function , which takes values in the set @xmath388 . an _ enumeration of waves _ for the function @xmath26 is a 3-tuple @xmath409 , where @xmath410 & \text{is \emph{the position function } } , \\
\hat u : \mathcal{w } \to { \mathbb{z}}{\varepsilon } & \text{is \emph{the right state function } } , \\ \end{array}\ ] ] with the following properties : 1 .
the function @xmath74 takes values only in the set of discontinuity points of @xmath26 ; 2 .
the restriction of the function @xmath74 to the set of waves where it takes finite values is increasing ; 3 . for given @xmath411 , consider @xmath412 ; then it holds : 1 . if @xmath413 , then @xmath414 \cap { \mathbb{z}}{\varepsilon}$ ] is strictly increasing and bijective ; 2 . if @xmath415 , then @xmath416 is strictly decreasing and bijective ; 3 . if @xmath417 , then @xmath418 .
given an enumeration of waves as in definition [ w_eow ] , we define the _ sign of a wave @xmath102 _ with finite position ( i.e. such that @xmath419 ) as follows : @xmath420.\ ] ] we immediately present an example of enumeration of wave which will be fundamental in the sequel .
[ w_initial_eow ] fix @xmath181 and let @xmath421 be the approximate initial datum of the cauchy problem , with compact support and taking values in @xmath388 .
the total variation of @xmath384 is an integer multiple of @xmath71 .
let @xmath422 , \quad x \mapsto u(x ) : = { \text{\rm tot.var.}}(\bar u_{\varepsilon } ; ( -\infty , x]),\ ] ] be the total variation function .
then define : @xmath423 and @xmath424 , \quad s \mapsto \mathtt x_0(s ) : = \inf \big\{x \in ( -\infty , + \infty ] \
| \ { \varepsilon}s \leq u(x ) \big\}. \ ] ] moreover , recalling , we define @xmath425.\ ] ] it is fairly easy to verify that @xmath426 , @xmath148 are well defined and that they provide an enumeration of waves , in the sense of definition [ w_eow ] .
let us now give another definition .
[ w_speed_function ] consider a function @xmath26 as in definition [ w_eow ] and let @xmath409 be an enumeration of waves for @xmath26 . the _ speed function _
@xmath427 \cup \{+\infty\}$ ] is defined as follows : @xmath428}f_{\varepsilon}\big ) \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) & \text{if } \mathcal{s}(s ) = + 1 , \\ \big(\frac{d}{du}\operatorname{conc}_{[u(\mathtt
x(s)),u(\mathtt x(s)-)]}f_{\varepsilon}\big ) \big ( ( \hat u(s ) , \hat u(s ) + { \varepsilon } ) \big ) & \text{if } \mathcal{s}(s ) = -1 .
\end{array } \right.\ ] ] roughly speaking , @xmath429 is the speed given to the wave @xmath35 by the riemann problem located at @xmath430 . [ w_speed_increasing_wrt_waves ] notice that for each @xmath411 , @xmath76 restricted to @xmath431 is increasing by the monotonicity properties of derivatives the convex / concave envelopes ( or by the study of the riemann solver ) . consider the cauchy problem and fix @xmath181 ; let @xmath25 be piecewise constant wavefront solution .
for the initial datum @xmath432 , consider the enumeration of waves @xmath433 provided in example [ w_initial_eow ] ; let @xmath434 be the sign function defined in for this enumeration of waves .
now our aim is to define two functions @xmath435
\cup \{+\infty\ } , \ ] ] called _ the position at time @xmath436 of the wave @xmath437 _ and _ the speed at time @xmath436 of the wave @xmath437_. as one can imagine , we want to describe the path followed by a single wave @xmath437 as time goes on and the speed assigned to it by the riemann problems it meets along the way . even if there is a slight abuse of notation ( in this section @xmath74 depends also on time ) ,
we believe that the context will avoid any misunderstanding . the function @xmath74 is defined by induction , partitioning the time interval @xmath438 in the following way @xmath439 \cup \dots
\cup ( t_j , t_{j+1 } ] \cup \dots \cup ( t_{j-1 } , t_j ] \cup ( t_j , + \infty).\ ] ] first of all , for @xmath440 we set @xmath441 , where @xmath442 is the position function in the enumeration of waves of example [ w_initial_eow ] .
clearly @xmath443 is an enumeration of waves for the function @xmath444 as a function of @xmath81 ( @xmath148 being the right state function , as in the example above ) .
assume to have defined @xmath445 for every @xmath446 and let us define it for @xmath447 $ ] ( or @xmath448 ) .
the speed @xmath37 for @xmath449 $ ] is computed accordingly to . for @xmath450 ( or @xmath451 ) set @xmath452 for @xmath453 set @xmath454 if @xmath455 is not the point of interaction / cancellation @xmath456 ; otherwise for the waves @xmath35 such that @xmath457 and @xmath458 ( i.e. the ones surviving the possible cancellation in @xmath459 ) define @xmath460 where @xmath461 is defined in , using the enumeration of waves for the initial datum . to the waves
@xmath35 canceled by a possible cancellation in @xmath459 we assign @xmath462 .
the following lemma proves that the above procedure produces an enumeration of waves .
[ w_lemma_eow ] for any @xmath463 $ ] ( resp .
@xmath464 ) , the 3-tuple @xmath465 is an enumeration of waves for the piecewise constant function @xmath466 .
we prove separately that the properties ( 1 - 3 ) of definition [ w_eow ] are satisfied .
_ proof of property ( 1 ) .
_ by definition of wavefront solution , @xmath467 takes values only in the set of discontinuity points of @xmath468 .
_ proof of property ( 2 ) .
_ let @xmath52 be two waves and assume that @xmath469
. by contradiction , suppose that @xmath470 .
since by the inductive assumption at time @xmath130 , the 3-tuple @xmath471 is an enumeration of waves for the function @xmath472 , it holds @xmath473 .
two cases arise : * if @xmath474 , then it must hold @xmath475 , but this is impossible , due to remark [ w_speed_increasing_wrt_waves ] . *
if @xmath476 , then lines @xmath477 and @xmath478 must intersect at some time @xmath479 , but this is impossible , by definition of wavefront solution and times @xmath480 .
_ proof of property ( 3 ) . _ for @xmath450 or @xmath453 and for discontinuity points @xmath481 , the third property of an enumeration of waves is straightforward .
so let us check the third property only for time @xmath482 and for the discontinuity point @xmath456 .
fix any time @xmath483 ; according to assumption on binary intersections , you can find two points @xmath484 such that for any @xmath35 such that @xmath485 , either @xmath486 or @xmath487 and moreover @xmath488 , @xmath489 , @xmath490 .
we now just consider two main cases : the other ones can be treated similarly .
recall that at time @xmath491 , the @xmath492-tuple @xmath493 is an enumeration of waves for the piecewise constant function @xmath494 . if @xmath495 , then @xmath496 \cap { \mathbb{z}}{\varepsilon}\ ] ] and @xmath497 \cap { \mathbb{z}}{\varepsilon}\ ] ] are strictly increasing and bijective ; observing that in this case @xmath498 , one gets the thesis .
if @xmath499 , then @xmath496 \cap { \mathbb{z}}{\varepsilon}\ ] ] is strictly increasing and bijective ; observing that in this case @xmath500 \big\},\ ] ] one gets the thesis . for fixed wave @xmath35 ,
functions @xmath501 are right - continuous .
moreover @xmath502 is piecewise constant .
finally we introduce the following notation . given a time @xmath436 and a position @xmath503 $ ] , we set @xmath504 we will call @xmath505 the set of the _ real waves _ , while we will say that a wave @xmath35 is _ removed _ at time @xmath19 if @xmath506 .
it it natural to define the interval of existence of @xmath507 by @xmath508 in this section we define the interval of waves and we prove an important result about it .
the necessity of introducing this notion is due to the fact that we avoid relabeling the waves @xmath35 .
[ w_iow_def ] let @xmath509 be a fixed time and @xmath510 .
we say that @xmath511 is an _ interval of waves _ at time @xmath509 if for any given @xmath512 , with @xmath513 , and for any @xmath514 @xmath515 we say that an interval of waves @xmath511 is _ homogeneous _ if for each @xmath516 , @xmath517 . if waves in @xmath511 are positive ( resp .
negative ) , we say that @xmath511 is a _ positive _ ( resp .
_ negative _ ) interval of waves .
[ w_interval_waves ] let @xmath510 be a positive ( resp .
negative ) interval of waves .
then the restricion of @xmath148 to @xmath511 is strictly increasing ( resp .
decreasing ) and @xmath518 $ ] ( resp .
@xmath519 ) is an interval in @xmath179 .
assume @xmath511 is positive , the other case being similar .
first we prove that @xmath148 restricted to @xmath511 is increasing .
let @xmath520 , with @xmath52 .
let @xmath521 be the discontinuity points of @xmath468 between @xmath522 and @xmath523 .
by definition of ` interval of waves ' and by the fact that each wave in @xmath511 is positive , for any @xmath524 , @xmath525 contains only positive waves .
thus , by definition [ w_eow ] of enumeration of waves , and by the fact that for each @xmath526 , @xmath527 , the restriction @xmath528 \cap { \mathbb{z}}{\varepsilon}\ ] ] is strictly increasing and bijective , and so @xmath529 ; hence @xmath530 is strictly increasing . in order to prove that @xmath518 $ ] is a interval in @xmath179 , it is sufficient to prove the following : for any @xmath531 in @xmath532 and for any @xmath51 such that @xmath533 , there is @xmath534 , @xmath535 such that @xmath536 .
this follows immediately from the fact that the map in is bijective and strictly increasing .
[ w_artificial_speed ] let @xmath537 be an interval in @xmath179 , such that @xmath538 .
let @xmath35 be any positive ( resp .
negative ) wave such that @xmath539 ( resp .
@xmath540 ) .
the quantity @xmath541 ( resp .
@xmath542 ) is called _ the ( artificial ) speed given to the wave @xmath35 by the riemann problem @xmath195_. moreover we will say that _ the riemann problem @xmath195 divides @xmath543 _ if @xmath544 ( resp .
@xmath545 ) do not belong to the same shock component of @xmath546 ( resp .
@xmath547 ) . [ w_artificial_speed_remark ]
let @xmath511 be any positive ( resp .
negative ) interval of waves at fixed time @xmath509 . by proposition [ w_interval_waves ] , the set @xmath548 $ ] ( resp .
@xmath549 ) is an interval in @xmath179 .
hence , we will also write @xmath550 instead of @xmath551 and call it the speed given to the waves @xmath35 by the riemann problem @xmath511 . moreover , we will also say that the riemann problem @xmath511 divides @xmath543 if the riemann problem @xmath195 does .
now we state the main result for the wavefront tracking approximation , namely theorem [ w_main_thm ] . for easiness of the reader
we repeat the statement below . as in the previous section ,
let @xmath552 be an @xmath71-wavefront solution of the cauchy problem ; consider the enumeration of waves and the related position function @xmath553 and speed function @xmath554 constructed in previous section .
fix a wave @xmath507 and consider the function @xmath555 . by construction
it is finite valued until the time @xmath60 , after which its value becomes @xmath556 ; moreover it is piecewise constant , right continuous , with jumps possibly located at times @xmath557 .
the results we are going to prove is the following holds : @xmath558 where @xmath83 is _ the strength of the wave @xmath35_. we recall the following definition . [ w_int_canc_points ] for each @xmath559 , we will say that @xmath560 is an _ interaction point _ if the wavefronts which collide in @xmath560 have the same sign .
an interaction point will be called _ positive _ ( resp . _
negative _ ) if all the waves located in it are positive ( resp .
negative ) .
moreover we will say that @xmath560 is a _ cancellation point _ if the wavefronts which collide in @xmath560 have opposite sign .
the first step in order to prove theorem [ w_main_thm ] is to reduce the quantity we want to estimate , namely @xmath561 to a two separate estimates , according to @xmath560 being an interaction or a cancellation : @xmath562 the estimate on the cancellation points is fairly easy .
first of all define for each cancellation point @xmath560 the _ amount of cancellation _ as follows : @xmath563 [ w_canc_3 ] let @xmath560 be a cancellation point .
then @xmath564 let @xmath565 be respectively the left and the right state of the left wavefront involved in the collision at point @xmath560 and let @xmath566 be respectively the left and the right state of the right wavefront involved in the collision at point @xmath560 , so that @xmath567 and @xmath568
. without loss of generality , assume @xmath569
. then we have @xmath570 } f_{\varepsilon}\bigg ) \big ( ( \hat u(s ) -{\varepsilon } , \hat u(s))\big ) - \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_m ] } f_{\varepsilon}\bigg)\big ( ( \hat u(s)-{\varepsilon } , \hat u(s))\big ) \bigg]|s| \\ & \overset{\text{(prop .
\ref{differenza_vel})}}{\leq } \sum_{s \in { \mathcal{w}}(t_j ) } \bigg [ \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_r ] } f_{\varepsilon}\bigg)(u_r- ) - \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_m ] } f_{\varepsilon}\bigg)(u_r- ) \bigg]|s| \\ & \quad \ \ = \bigg [ \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_r ] } f_{\varepsilon}\bigg)(u_r- ) - \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_m ] } f_{\varepsilon}\bigg)(u_r- ) \bigg ] \sum_{s \in { \mathcal{w}}(t_j ) } |s| \\ & \quad \ \
\bigg [ \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_r ] } f_{\varepsilon}\bigg)(u_r- ) - \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_m ] } f_{\varepsilon}\bigg)(u_r- ) \bigg ] { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) .
\end{split}\ ] ] now observe that , by proposition [ diff_vel_proporzionale_canc ] , @xmath571 } f_{\varepsilon}\bigg)(u_r- ) - \bigg(\frac{d}{du } \operatorname{conv}_{[u_l , u_m ] } f_{\varepsilon}\bigg)(u_r- ) \leq&~ \lvert f '' \lvert_{l^\infty } ( u_m - u_r ) \\
\leq&~ \lvert f '' \lvert_{l^\infty } \mathcal{c}(t_j , x_j ) .
\end{split}\ ] ] hence , from and , we obtain @xmath572 together with , this concludes the proof . [ w_canc_4 ] it holds @xmath573 from , and proposition [ w_canc_3 ] we obtain @xmath574 \\ & \leq \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \big [ { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) - { \text{\rm tot.var.}}(u_{\varepsilon}(t_j , \cdot ) ) \big ] \\ & \leq \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))^2 , \end{split}\ ] ] thus concluding the proof of the corollary . from now on ,
our aim is to prove that @xmath575 as outlined in section [ sss_sketch_proof ] , the idea is the following : we define a positive valued functional @xmath576 , @xmath396 , such that @xmath577 is piecewise constant in time , right continuous , with jumps possibly located at times @xmath578 and such that @xmath579 such a functional will have two properties : 1 . for each @xmath402
such that @xmath560 is an interaction point , @xmath577 is decreasing at time @xmath130 and its decrease bounds the quantity we want to estimate at time @xmath130 as follows : @xmath580;\ ] ] this is proved in theorem [ w_decreasing ] ; 2 . for each @xmath402
such that @xmath403 is a cancellation point , @xmath577 can increase at most by @xmath581;\ ] ] this is proved in theorem [ w_increasing ] . using the two estimates above
, we obtain the following proposition , which completes the proof of theorem [ w_main_thm ] .
[ w_thm_interaction ] it holds @xmath582 by direct computation , @xmath583 \\ \leq & ~ 2 \bigg [ \sum_{\substack{(t_j , x_j ) \\ \text{interaction } } } \big [ { \mathfrak{q}}(t_{j-1 } ) - { \mathfrak{q}}(t_j ) \big ] + \sum_{\substack{(t_j , x_j ) \\ \text{cancellation } } } \big [ { \mathfrak{q}}(t_{j-1 } ) - { \mathfrak{q}}(t_j ) \big ] \\ &
\qquad \qquad - \sum_{\substack{(t_j , x_j ) \\ \text{cancellation } } } \big [ { \mathfrak{q}}(t_{j-1 } ) - { \mathfrak{q}}(t_j ) \big ] \bigg ] \\
= & ~ 2 \bigg [ \sum_{j=1}^j \big [ { \mathfrak{q}}(t_{j-1 } ) - { \mathfrak{q}}(t_j ) \big ] + \sum_{\substack{(t_j , x_j ) \\ \text{cancellation } } } \big [ { \mathfrak{q}}(t_j ) - { \mathfrak{q}}(t_{j-1 } ) \big ] \bigg ] \\
\text{(by \eqref{w_increase } ) } \leq & ~ 2 \bigg [ \sum_{j=1}^j \big [ { \mathfrak{q}}(t_{j-1 } ) - { \mathfrak{q}}(t_j ) \big ] + \log(2 ) \sum_{\substack{(t_j , x_j ) \\ \text{cancellation } } } \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \mathcal{c}(t_j , x_j ) \bigg ] \\ \text{(by \eqref{w_canc_0 } , \eqref{boundqzero } ) } \leq & ~ 2 \big [ { \mathfrak{q}}(0 ) + \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))^2 \big ] \\
\leq&~ 2 ( 1 + \log(2 ) ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))^2 . \end{aligned}\end{gathered}\ ] ] in the remaining part of this section we prove estimates and . in this section
we define the notion of pairs of waves which _ have never interacted before a fixed time @xmath19 _ and pairs of waves which _ have already interacted _ and , for any pair of waves which have already interacted , we associate an artificial speed difference , which is some sense summarize their past common history . [ interagite_non_interagite ] let @xmath509 be a fixed time and let @xmath584 .
we say that _ @xmath543 interact at time @xmath509 _ if @xmath585 .
we also say that _ they have already interacted at time @xmath509 _ if there is @xmath586 such that @xmath543 interact at time @xmath19 .
moreover we say that _ they have not yet interacted at time @xmath509 _ if for any @xmath586 , they do not interact at time @xmath19 .
[ w_interagite_stesso_segno ] assume that the waves @xmath587 interact at time @xmath509
. then they have the same sign .
easy consequence of definition of enumeration of waves and the fact that @xmath588 is independent of @xmath19 .
[ w_quelle_in_mezzo_hanno_int ] let @xmath509 be a fixed time , @xmath584 , @xmath52 .
assume that @xmath587 have already interacted at time @xmath509 .
if @xmath589 and @xmath590 , then @xmath591 have already interacted at time @xmath509 .
let @xmath19 be the time such that @xmath543 interact at time @xmath19 .
clearly @xmath592 .
since for @xmath19 fixed , @xmath74 is increasing on @xmath505 , it holds @xmath593 .
let @xmath594 . by lemmas [ w_interagite_stesso_segno ] and [ w_quelle_in_mezzo_hanno_int ] , the set @xmath595 is an homogeneous interval of waves .
moreover set @xmath596 let @xmath587 be two waves .
assume that @xmath52 and that they have already interacted at a fixed time @xmath509 .
consider now the set @xmath597 this is clearly an interval of waves .
observe that @xmath598 and @xmath599 .
a fairly easy argument based on lemma [ w_quelle_in_mezzo_hanno_int ] implies that @xmath600 is made of the waves @xmath601 which have interacted with both @xmath35 and @xmath119 .
[ w_waves_divided ] let @xmath584 be two waves which have already interacted at time @xmath509 .
we say that _ @xmath543 are divided in the real solution at time @xmath509 _ if @xmath602 i.e. if at time @xmath509 they have either different position , or the same position but different speed .
if they are not divided in the real solution , we say that _ they are joined in the real solution_. [ rem_divise_solo_in_cancellazioni ] it @xmath603 for each @xmath402 , then two waves are divided in the real solution if and only if they have different position .
the requirement to have different speed is needed at collision times , more precisely at cancellations .
[ w_unite_realta ] let @xmath509 be a fixed time .
let @xmath584 . if @xmath543 are not divided in the real solution at time @xmath509 , then the riemann problem @xmath604 does not divide them . for the definition of the riemann problem @xmath600 see definition [ w_artificial_speed ] and remark [ w_artificial_speed_remark ] .
let @xmath605 .
clearly @xmath606 .
observe that @xmath607 is an interval of waves and that by definition the real speed of the waves @xmath608 is the speed given @xmath543 by the riemann problem @xmath609 .
the conclusion is then a consequence of corollary [ stesso_shock ] .
the remaining part of this section is devoted to prove the following proposition , which is in some sense the converse of the previous one and is a key tool in order estimate the increase and the decrease of the functional @xmath577 . [ w_divise_tocca ] let @xmath509 be a fixed time .
let @xmath543 be two waves which have already interacted at time @xmath509 .
assume that @xmath543 are divided in the real solution .
let @xmath610 .
if @xmath611 are divided in the real solution at time @xmath509 , then the riemann problem @xmath612 divides them .
fix two waves @xmath52 .
it is sufficient to prove the proposition only for times @xmath613 .
we proceed by induction on @xmath402 . for @xmath614
the proof is obvious .
let us assume the lemma is true for @xmath615 and let us prove it for @xmath402 .
suppose @xmath543 to be divided in the real solution at time @xmath130 .
we can also assume w.l.o.g . that @xmath543 are both positive .
when @xmath560 is an interaction the analysis is quite simple , while the cancellation case requires more effort .
* @xmath560 interaction .
* let us distinguish two cases .
if @xmath616 , then waves @xmath543 must be involved in the interaction , i.e. @xmath617 . since @xmath560 is an interaction point , @xmath618 , and so @xmath543 are not divided at time @xmath130 , hence the statement of the proposition can not apply to @xmath543 . if @xmath619 , take @xmath620 , such that @xmath611 are divided in the real solution .
since an interaction does not divide waves which were joined before the interaction , @xmath611 were already divided at time @xmath621 and so by inductive assumption we have done . * @xmath560 cancellation . *
assume that @xmath543 are divided in the real solution after the cancellation at time @xmath130 .
moreover , w.l.o.g .
, assume that in @xmath560 two wavefronts collide , the one coming from the left is positive , the one coming from the right is negative ; assume also that waves in @xmath622 are positive , and that @xmath623 ( the proof in the case @xmath624 is similar , but easier ) .
set @xmath625 , @xmath626 .
if @xmath627 , then @xmath543 were already divided before the collision ( i.e. at time @xmath621 ) , @xmath628 and @xmath629 and so by inductive assumption we conclude .
hence assume @xmath630 $ ] .
if @xmath631 , then @xmath632 and so by corollary [ stesso_shock ] , we can again conclude ; hence let us assume @xmath633 .
finally observe that we can assume @xmath634 \cap { \mathcal{w}}(t_j ) = [ l(t_j , s'),r_2]$ ] , i.e. no wave in @xmath635 $ ] has been canceled up to time @xmath130 ( the general case can be treated in a similar way ) .
see figure [ fig : figura03 ] .
let us observe that @xmath636 now set @xmath637 we need now the following three claims .
[ w_s_tilde_sep ] waves @xmath638 are divided in the real solution at time @xmath621 .
if @xmath543 are divided in the real solution at time @xmath621 , then by definition @xmath639 and then we have done .
if @xmath543 are not divided , this means that @xmath617 , while by one of our assumption @xmath640 ; hence @xmath638 have different position at time @xmath130 .
[ w_riemann_prima ] waves @xmath641 and @xmath642 are divided by the riemann problem @xmath643 . by claim [ w_s_tilde_sep ] , @xmath638 are divided in the real solution at time @xmath621 .
moreover , by definition of @xmath642 , also @xmath641 and @xmath642 are divided in the real solution at time @xmath130 ; since @xmath644 , @xmath641 and @xmath642 are divided also at time @xmath621 .
hence by inductive assumption , the riemann problem @xmath643 divides @xmath641 and @xmath642 .
[ w_riemann_dopo ] waves @xmath641 and @xmath642 are divided by the riemann problem @xmath645 . assume first that @xmath543 are divided in the real solution at time @xmath621 .
in this case , by claim [ w_riemann_prima ] , @xmath646 and @xmath642 are divided by the riemann problem @xmath647 ; hence by corollary [ stesso_shock ] they are divided also by the riemann problem @xmath645 since @xmath648 .
now assume @xmath543 are joined in the real solution at time @xmath621 . in this case @xmath617 . by claim [ w_riemann_prima ] ,
the riemann problem @xmath643 divides @xmath641 and @xmath642 .
moreover , since @xmath543 are divided in the real solution at time @xmath130 , the riemann problem @xmath649 $ ] divides them .
hence , observing that @xmath650 , by proposition [ incastro ] , the riemann problem @xmath651 $ ] divides @xmath641 and @xmath642 .
one concludes the proof just observing that @xmath652 and then @xmath651 = \mathcal{i}(t_j , s , s')$ ] .
now we are able to conclude the proof of our proposition .
take @xmath653 and assume that @xmath654 are divided in the real solution at time @xmath130 . 1 .
if @xmath655 $ ] and @xmath656 $ ] , then by claim [ w_riemann_dopo ] @xmath611 are divided by the riemann problem @xmath645 .
if @xmath657 $ ] , then they were already divided at time @xmath621 hence , by inductive assumption and claim [ w_s_tilde_sep ] , the riemann problem @xmath643 divides @xmath611 .
thus by claim [ w_riemann_prima ] and proposition [ tocca ] also the riemann problem @xmath658 $ ] divides @xmath611 and then by claim [ w_riemann_dopo ] and proposition [ tocca ] , the riemann problem @xmath645 divides them .
3 . finally assume that @xmath659 $ ] .
since @xmath611 are divided in the real solution at time @xmath130 , one has that the riemann problem @xmath660 $ ] divides them , and also the riemann problem @xmath661 \subseteq [ r_1,r_2]$ ] divides them .
hence , by claim [ w_riemann_dopo ] and proposition [ tocca ] , the riemann problem @xmath645 divides @xmath611 .
this concludes the proof of proposition [ w_divise_tocca ] .
we can finally define the functional @xmath577 and prove that it satisfies inequalities and .
first of all for any pair of wave @xmath662 , @xmath531 , define _ the weight @xmath158 of the pair of waves @xmath35 ,
@xmath119 at time @xmath19 _ in the following way : @xmath663 recall that @xmath664 ( resp .
@xmath665 ) is the speed given to the wave @xmath35 ( resp .
@xmath119 ) by the riemann problem @xmath666 . as an easy consequence of proposition [ convex_fundamental_thm_affine ]
, we obtain that @xmath667 takes values in @xmath668 $ ] .
[ w_rem_unite_realta ] if @xmath543 are joined in the real solution , then by proposition [ w_unite_realta ] @xmath669 .
finally set @xmath670 ( recall that @xmath671 is the strength of the waves @xmath543 respectively . )
it is immediate to see that @xmath577 is positive , piecewise constant , right continuous , with jumps possibly located at times @xmath578 , and @xmath672 . in the next two sections we prove that it also satisfies inequality and
this completes the proof of proposition [ w_thm_interaction ] .
this section is devoted to prove inequality .
[ w_decreasing ] for any interaction point @xmath560 , it holds @xmath673.\ ] ] by direct inspection of the proof one can verify that the constant @xmath674 is sharp .
assume w.l.o.g .
that all the waves in @xmath622 are positive .
we partition @xmath622 through the equivalence relation @xmath675 by our assumption @xmath622 is decomposed into two disjoint intervals of waves @xmath676 such that for each @xmath677 and @xmath678 , it holds @xmath679 .
first of all observe that by formula and by proposition [ w_unite_realta ] , if @xmath531 , @xmath680 , but @xmath681 , then @xmath682 . indeed , if at least one between @xmath543 does not belong to @xmath683 , then @xmath684 ; on the other side , if @xmath685 ( or @xmath686 ) , then @xmath543 are joined both before and after the interaction and for this reason , by remark [ w_rem_unite_realta ] , @xmath687 . now
observe that , if @xmath688 , then , by remark [ w_rem_unite_realta ] , @xmath689 .
hence it is sufficient to prove that @xmath690 for any positive interval of waves @xmath511 , define _ the strength of the interval @xmath511 _ as @xmath691 and _ the mean speed of waves in @xmath511 _ as @xmath692 now observe that @xmath693 hence @xmath694 set @xmath695 , @xmath696 and define ( see fig .
[ fig : figura14 ] ) @xmath697 .,width=453,height=283 ] thus @xmath698 dividing by @xmath699 we get @xmath700 where the last inequality is a consequence of lagrange s theorem .
let us now concentrate our attention on the second term of the last summation .
observe that waves @xmath701 are divided in the real solution at time @xmath621 ; hence , by proposition [ w_divise_tocca ] , they are divided by the riemann problem @xmath702 .
hence it is not difficult to see ( it is the cubic estimate when the speeds are monotone ) that we can write @xmath703 , for any @xmath704 and @xmath705 , if @xmath543 have already interacted at time @xmath621 , then @xmath706 . together with proposition [ vel_aumenta ] and with the fact that @xmath701 are divided by the riemann problem @xmath707 ,
this yields @xmath708 and thus @xmath709 instead , if @xmath543 have not yet interacted at time @xmath621 , using lagrange s theorem , we get @xmath710 thus , by , , , @xmath711 let us now observe that if @xmath712 and @xmath678 , then by definition of @xmath713 , @xmath543 have not yet interacted at time @xmath621 and so @xmath714 . the same holds if @xmath704 and @xmath715 . hence , recalling and , we get @xmath716 \\ \leq&~ 2 \bigg [ \sum_{(s , s ' ) \in \mathcal{l}_1 \times \mathcal{r } } \mathfrak{q}(t_{j-1},s , s')|s||s'| + \sum_{(s , s ' ) \in \mathcal{l}_2 \times
\mathcal{r}_1}\mathfrak{q}(t_{j-1},s , s')|s||s'| \\ & \quad \ + \sum_{(s , s ' ) \in \mathcal{l}_2 \times \mathcal{r}_2}\mathfrak{q}(t_{j-1},s , s')|s||s'| \bigg ] \\ \leq&~ 2 \sum_{(s , s ' ) \in \mathcal{l } \times \mathcal{r } } \mathfrak{q}(t_{j-1},s , s')|s||s'| , \end{split}\ ] ] which is what we wanted to obtain
. this section is devoted to prove inequality , more precisely we will prove the following theorem . [ w_increasing ] if @xmath560 is a cancellation point , then @xmath717,\ ] ] where @xmath718 is the amount of cancellation at point @xmath560 , defined in . to simplify the notations , w.l.o.g
assume the following wave structure ( see also the proof of proposition [ w_divise_tocca ] and figure [ fig : figura03 ] ) : * in @xmath560 two wavefronts collide , the one coming from the left is positive and contains all and only the waves in @xmath719 \subseteq { \mathcal{w}}(t_{j-1})$ ] , the one coming from the right is negative ; * in @xmath560 all and only the waves in @xmath660 \subseteq { \mathcal{w}}(t_j)$ ] survive ; * @xmath720 .
it holds @xmath721|s||s'| \\
\leq&~ \sum_{\substack{s , s ' \in { \mathcal{w}}(t_j ) \\ s
< s ' } } [ \mathfrak{q}(t_j , s , s')-\mathfrak{q}(t_{j-1},s , s')]^+ \ |s||s'| .
\end{split}\ ] ] the proof will follow by the next three lemmas .
[ l_w_canc_lem_1 ] if @xmath52 are waves such that @xmath722^+ > 0 $ ] , then @xmath723 $ ] and @xmath724 $ ] .
first let us prove @xmath723 $ ] .
by contradiction , assume @xmath725 $ ] .
since @xmath722^+ > 0 $ ] , then @xmath726 ; hence , by remark [ w_rem_unite_realta ] , @xmath543 are divided in the real solution at time @xmath130 . since @xmath725 $ ] ,
@xmath543 are divided in the real solution also at time @xmath621 . moreover , by @xmath722^+ > 0 $ ] and because @xmath560 is a cancellation point , it must hold @xmath727 .
assume now @xmath728 , the case @xmath729 being similar .
by proposition [ w_divise_tocca ] the riemann problems @xmath647 and @xmath645 divide waves @xmath641 and @xmath642 . moreover observe that @xmath730 .
hence , by proposition [ tocca ] , @xmath731 in a similar way , @xmath732 ; thus @xmath682 , contradicting the assumption .
let us now prove the second part of the statement , namely @xmath724 $ ] . since @xmath733 , then @xmath734 .
assume by contradiction @xmath735 .
in this case @xmath736 , hence @xmath737 ; thus @xmath738 and so @xmath682 , a contradiction with the initial assumption .
consider the following values : @xmath739 define two functions @xmath740 \to { \mathbb{r}}$ ] , piecewise affine , with discontinuity points of the derivative in the set @xmath388 and such that @xmath741 } & = \operatorname{conv}_{[b_1,b_3 ] } f , & h|_{[b_1,b_2 ] } & = \operatorname{conv}_{[b_1,b_2 ] } f , & g & = h \ \text { on } ( -\infty , b_1].\end{aligned}\ ] ] clearly @xmath163 , @xmath205 with the above properties exist .
[ w_lemma_uno ] for any @xmath723 $ ] and @xmath724 $ ] , @xmath531 , it holds @xmath742^+ \\
\leq&~ \frac { \bigg| \big [ h ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - h'\big((\hat u(s)-{\varepsilon } , \hat u(s))\big ) \big ] - \big [ g'\big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - g'\big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \big ] \bigg| } { |\hat u(s ' ) - ( \hat u(s)-{\varepsilon})|}. \end{split}\ ] ] in other words , we are saying that the maximal variation of @xmath158 is controlled by the maximal variation of speed ( the numerator of the r.h.s .
of ) divided by the total variation between the two waves @xmath35 , @xmath119 .
fix @xmath543 as in the hypothesis and set for shortness @xmath743 by multiplication for @xmath744 , becomes @xmath745^+ \leq&~ \bigg| \bigg(h'\big ( \big(\hat u(s')-{\varepsilon } , \hat u(s')\big)\big ) - h'\big(\big(\hat u(s)-{\varepsilon } , \hat u(s)\big)\big)\bigg ) \\ & ~ - \bigg(g'\big(\big(\hat u(s')-{\varepsilon } , \hat u(s')\big)\big ) - g'\big(\big(\hat u(s)-{\varepsilon } , \hat u(s)\big)\big)\bigg ) \bigg|.\end{aligned}\ ] ] we consider two cases depending on the position of the wave @xmath35 .
assume first @xmath746 $ ] .
in this case @xmath543 are divided in the real solution both at time @xmath621 and at time @xmath130 .
hence , by proposition [ w_divise_tocca ] the riemann problems @xmath747 , @xmath748 divide the waves @xmath641 and @xmath642 .
thus , by proposition [ tocca ] , and by the fact that @xmath749 , @xmath750 whence @xmath751 ( recall that @xmath308 on @xmath752 $ ] ) .
now distinguish two subcases : 1 . if @xmath753 , then @xmath754 = \mathcal{i}_j \cap [ r_1,r_2]$ ] , and so , arguing as above , @xmath755 .
2 . if @xmath756 , then @xmath757 and so @xmath758,s ' ) = \sigma([r_1,r_2 ] , s ' ) = h'\big((\hat u(s')-{\varepsilon } , \hat u(s'))\big),\ ] ] while @xmath759,s ' ) \overset{\text{(prop . \ref{vel_aumenta})}}{\geq } \sigma([r_1,r_3],s ' ) = g'\big((\hat u(s')-{\varepsilon } , \hat u(s'))\big),\ ] ] where the inequality is an easy consequence of proposition [ vel_aumenta ] . in both case
( a ) and ( b ) , @xmath760 together with , this yields the thesis .
assume now @xmath761 $ ] .
in this case @xmath762 since @xmath543 are divided at time @xmath130 in the real solution , we can argue as in the previous case and use proposition [ w_divise_tocca ] to obtain @xmath763 observe that since @xmath543 are joined before the collision in the real solution , @xmath764 moreover by proposition [ w_unite_realta ] , @xmath765 , which , together with , yields the thesis .
[ w_lemma_due ] define @xmath766 .
then @xmath767 \\ s ' \in [ r_1,r_2 ] , \ s
< s ' } } \frac { \big|\varphi ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - \varphi ' \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \big| |s||s'| } { |\hat u(s ' ) - ( \hat u(s)-{\varepsilon})| } \\ \leq \log(2 ) { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) \big [ h'(b_2- ) - g'(b_2-)\big].\end{gathered}\ ] ] first of all , let us observe that , since @xmath768 is increasing and @xmath531 with positive sign @xmath769 , we can forget about the absolute value and get @xmath770 \\ s ' \in [ r_1,r_2 ] , \
s < s ' } } \frac { \bigg [ \varphi ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - \varphi ' \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \bigg ] |s||s'| } { \hat u(s ' ) - ( \hat u(s)-{\varepsilon } ) } \\ & \quad \leq \sum_{\substack{s \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s ' \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\
s < s ' } } \frac { \bigg[\varphi ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - \varphi ' \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \bigg ] |s||s'| } { \hat u(s ' ) - ( \hat u(s)-{\varepsilon } ) } \\ & \quad = \sum_{\substack{s \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s ' \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\
s < s ' } } \frac { \bigg [ \varphi ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - \varphi ' \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \bigg ] } { \hat u(s ' ) - ( \hat u(s)-{\varepsilon } ) } \int_{\hat u(s)-{\varepsilon}}^{\hat u(s ) } du \int_{\hat u(s')-{\varepsilon}}^{\hat u(s ' ) } du ' . \\ \ ] ] since @xmath771 for each @xmath772 , and similarly for @xmath119 , we can continue the chain of inequality in the following way : @xmath773 \\ s ' \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s < s ' } } & \frac { \bigg[\varphi ' \big ( ( \hat u(s')-{\varepsilon } , \hat u(s ' ) ) \big ) - \varphi ' \big ( ( \hat u(s)-{\varepsilon } , \hat u(s ) ) \big ) \bigg ] } { \hat u(s ' ) - ( \hat u(s)-{\varepsilon } ) } \int_{\hat u(s)-{\varepsilon}}^{\hat u(s ) } du \int_{\hat u(s')-{\varepsilon}}^{\hat u(s ' ) } du ' \\ = & ~ \sum_{\substack{s \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s ' \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\
s < s ' } } \frac{1}{\hat u(s ' ) - ( \hat u(s)-{\varepsilon } ) } \int_{\hat u(s)-{\varepsilon}}^{\hat u(s ) } \int_{\hat u(s')-{\varepsilon}}^{\hat u(s ' ) } ( \varphi'(u ' ) - \varphi'(u ) ) du'du \\
\leq&~ \sum_{\substack{s \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s ' \in [ l(t_{j-1},r_2 + 1 ) , r_2 ] \\ s < s ' } } \int_{\hat u(s)-{\varepsilon}}^{\hat u(s ) } \int_{\hat u(s')-{\varepsilon}}^{\hat u(s ' ) } \frac{\varphi'(u ' ) - \varphi'(u)}{u'-u } du'du \\
\leq&~ \int_{a}^{b_2 } \int_{u}^{b_2 } \frac{\varphi'(u ' ) - \varphi'(u)}{u'-u } du'du . \\ \ ] ] since @xmath774 is piecewise affine , we can write @xmath775du'du \\ = & ~ \sum_{\substack{m \in { \mathbb{z}}\\ a < m{\varepsilon } < b_2 } } [ \varphi'(m{\varepsilon}+ ) - \varphi'(m{\varepsilon}- ) ] \int_{a}^{m{\varepsilon } } \int_{m{\varepsilon}}^{b_2 } \frac{1}{u'-u } du'du .
\\ \end{aligned}\ ] ] an elementary computation shows that @xmath776 } \int_{a}^{\xi } \int_{\xi}^{b_2 } \frac{1}{u'-u } du'du = \log(2 ) ( b_2-a).\ ] ] hence @xmath777 \int_{a}^{m{\varepsilon } } \int_{m{\varepsilon}}^{b_2 } \frac{1}{u'-u } du'du \leq&~ \log(2 ) ( b_2 - a ) \sum_{\substack{m \in { \mathbb{z}}\\ a < m{\varepsilon } < b_2 } } [ \varphi'(m{\varepsilon}+ ) - \varphi'(m{\varepsilon}- ) ] \\
\leq&~ \log(2 ) { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) \big [ \varphi'(b_2- ) - \varphi'(a+ ) \big ] \\ \text{(since $ \varphi'(a+ ) \geq 0 $ by proposition \ref{vel_aumenta } ) } \ \leq&~ \log(2 ) { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) \varphi'(b_2- ) \\
\leq&~ \log(2 ) { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) \big [ h'(b_2- ) - g'(b_2- ) \big ] . \ ] ] this concludes the proof of lemma [ w_lemma_due ] . * conclusion of proof of theorem [ w_increasing]*. observe that , by proposition [ diff_vel_proporzionale_canc ] , @xmath778 putting together lemma [ w_lemma_uno ] , lemma [ w_lemma_due ] , inequality and inequality one easily concludes the proof of the theorem .
in this section we prove the main interaction estimate for an approximate solution of the cauchy problem obtained by glimm scheme .
the line of the proof is very similar to the proof for the wavefront tracking case , even if a relevant number of technicalities arises .
for this reason the structure of this section is equal to the structure of section [ sect_wavefront ] : throughout the remaining part of this paper , we will emphasize the differences in definitions and proofs between the wavefront algorithm and the glimm scheme .
first let us briefly recall how glimm scheme constructs an approximate solution .
fix @xmath181 . to construct an approximate solution @xmath397 to the cauchy problem
, we start with a grid in the @xmath79 plane having step size @xmath779 , with nodes at the points @xmath780 moreover we shall need a sequence of real numbers @xmath781 , uniformly distributed over the interval @xmath782 $ ] .
this means that , for every @xmath783 $ ] , the percentage of points @xmath784 , which fall inside @xmath785 $ ] should approach @xmath786 as @xmath787 : @xmath788\}}{n } = \lambda \hspace{1 cm } \text{for each } \lambda \in [ 0,1].\ ] ] at time @xmath789 , the glimm algorithm starts by taking an approximation @xmath384 of the initial datum @xmath31 , which is constant on each interval of the form @xmath790 , and has jumps only at the nodal points @xmath791
. we shall take ( remember that @xmath31 is right continuous ) @xmath792 notice that , as in the wavefront tracking case , estimate holds also in this case . for times @xmath793 sufficiently small , the solution @xmath397 is then obtained by solving the riemann problems corresponding to the jumps of the initial approximation @xmath444 at the nodes @xmath794 . since @xmath8 for any @xmath7 , the solutions to the riemann problems do not overlap and thus @xmath795 can be prolonged on the whole time interval @xmath796 . at time @xmath797 a restarting procedure is adopted : the function @xmath798 is approximate by a new function @xmath799 which is piecewise constant , having jumps exactly at the nodes @xmath800 .
our approximate solution @xmath25 can now be constructed on the further time interval @xmath801 , again by piecing together the solutions of the various riemann problems determined by the jumps at the nodal points @xmath794 . at time
@xmath802 , this solution is again approximated by a piecewise constant function , etc ... a key aspect of the construction is the restarting procedure . at each time
@xmath803 , we need to approximate @xmath804 with a piecewise constant function @xmath805 having jumps precisely at the nodal points @xmath794 .
this is achieved by a random sampling technique . more precisely , we look at the number @xmath806 in the uniformly distributed sequence . on each interval @xmath807 ,
the old value of our solution at the intermediate point @xmath808 becomes the new value over the whole interval : @xmath809 one can prove that , if the initial datum @xmath31 has bounded total variation , then an approximate solution can be constructed by the above algorithm for all times @xmath396 .
finally set , by simplicity , @xmath810 similarly to section [ front_waves ] , we define the notion of _ wave _ , the notion of _ position of a wave _ and the notion of _ speed of a wave_. first of all , as we did for the wavefront tracking scheme in section [ front_eow ] , we define the notion of _ enumeration of waves _ related to a bv function @xmath406 of the single variable @xmath81 .
the biggest differences with the wavefront tracking case are that here we allow the set of waves @xmath811 to be a subset of @xmath179 ( while in the wavefront case it is a subset of @xmath146 ) and we allow the state function @xmath148 to take values in @xmath179 ( while in the wavefront case it takes values in @xmath388 ) .
this is due to the fact that in an @xmath71-approximate solution constructed with the wavefront algorithm any wavefront has strength at least equal to @xmath71 , while in an approximate solution obtained by the glimm scheme rarefactions ( i.e. infinitesimal wavefronts ) are allowed .
[ eow ] let @xmath407 , @xmath408 , be a piecewise constant , right continuous function , with jumps located at points of the form @xmath812 , @xmath51 .
an _ enumeration of waves _ for the function @xmath26 is a 3-tuple @xmath409 , where @xmath813 , & \text{is \emph{the position function } } , \\ \hat u : \mathcal{w }
\to { \mathbb{r } } , & \text{is \emph{the state function } } , \\ \end{array}\ ] ] with the following properties : 1 .
the function @xmath74 takes values only in the set @xmath814 ; 2 .
the restriction of the function @xmath74 to the set of waves where it takes finite values is increasing ; 3 . for given @xmath411 , consider @xmath412 ; then @xmath431 is a finite union ( possibily empty ) of intervals of the form @xmath815 $ ] ; moreover it holds : 1 .
if @xmath413 , then @xmath414 $ ] is strictly increasing and bijective , with @xmath816 for each @xmath35 in the interior of @xmath431 ; 2 . if @xmath415 , then @xmath817 is strictly decreasing and bijective , with @xmath818 for each @xmath35 in the interior of @xmath431 ; 3 .
if @xmath417 , then @xmath418 . given an enumeration of waves as in definition [ eow ] , we define _ the sign _ of a wave @xmath819 with finite position ( i.e. such that @xmath419 ) as follows : @xmath820.\ ] ] the analog of example [ w_initial_eow ] is the next example .
[ initial_eow ] fix @xmath181 and let @xmath421 be the approximate initial datum of the cauchy problem constructed in .
let @xmath422\ ] ] be the total variation function , defined as @xmath821)$ ] .
then define : @xmath822\ ] ] and @xmath424 , \qquad \mathtt x_0(s ) : = \inf\big\{x \in ( -\infty , + \infty ] \ | \
s \leq u(x ) \big\}. \ ] ] moreover , recalling , we define @xmath823.\ ] ] one can easily prove that @xmath148 is continuous .
we also adapt definition [ w_speed_function ] in the following way .
[ speed_function ] consider a function @xmath26 as in definition [ eow ] and let @xmath409 be an enumeration of waves for @xmath26 .
the _ speed function _
@xmath427 \cup \{+\infty\}$ ] is defined as follows : @xmath824}f\big)\big(\hat u(s)\big ) & \text{if } \mathcal{s}(s ) = + 1 ,
x(s)),u(\mathtt x(s)-)]}f\big)\big(\hat u(s)\big ) & \text{if } \mathcal{s}(s ) = -1 .
\end{array } \right.\ ] ] consider the cauchy problem and fix @xmath181 ; let @xmath397 be the @xmath825-glimm solution , with sampling sequence @xmath826 .
for the ( piecewise constant ) initial datum @xmath432 , consider the enumeration of waves @xmath433 provided in example [ initial_eow ] ; let @xmath434 be the sign function defined in for this enumeration of waves . our aim is to define two functions @xmath435
\cup \{+\infty\ } , \ ] ] called _ the position at time @xmath436 of the wave @xmath437 _ and _ the speed at time @xmath436 of the wave @xmath437_. let us begin with the definition of @xmath74 .
we first define @xmath74 on the domain @xmath827 and then we extend it over all the domain @xmath828 .
the definition of @xmath829 is given by recursion on @xmath171 , i.e. for each @xmath830 we define a function @xmath831 $ ] such that the 3-tuple @xmath832 is an enumeration of waves for the piecewise constant function @xmath833 .
for @xmath834 we set @xmath835 , where @xmath836 is the position function in the enumeration of waves of example [ initial_eow ] .
clearly @xmath443 is an enumeration of waves for the function @xmath444 .
assume now to have obtained @xmath837 with the property stated above and define @xmath838 in the following way .
let @xmath839 be the speed function related to the piecewise constant function @xmath833 and the enumeration of waves @xmath832 .
then set @xmath840 define @xmath841 where @xmath461 was defined in , with respect to the enumeration of waves for the initial datum , and set @xmath842 observe that for each @xmath437 , @xmath843 is uniquely defined since , for @xmath844 , sets @xmath845 , @xmath846 , are disjoint .
a fairly easy extension of lemma [ w_lemma_eow ] shows that the recursive procedure generates an enumeration of waves at the times @xmath847 .
the only differences in the proof are that now the map @xmath148 takes continuous values and that at each time @xmath847 you have to consider countably many disjoint riemann problems , instead of a single one . we have just defined the position function @xmath74 on the domain @xmath827 .
let us now extend it over all the domain @xmath848 , by setting , for @xmath849 , @xmath850 see figure [ fig : figura32 ] .
the speed @xmath851 \cup \{+\infty\}$ ] is defined setting for @xmath852 and for @xmath102 , @xmath853 . as in section [ w_pswaves ]
, we set @xmath854 and we will call @xmath505 the set of the _ real waves _ at time @xmath19 , while we will say that a wave @xmath35 is _ removed _ at time @xmath19 if @xmath506 . observe that the definition of @xmath855 agrees with the definition of @xmath856 given before , for @xmath857 .
finally for any @xmath507 set @xmath858 the curve @xmath46 , for @xmath859 , as a curve in the @xmath79-plane , is piecewise linear with nodes in the grid points .
this section is analog of section [ w_iow ] .
let @xmath509 be a fixed time and @xmath510 .
we say that @xmath511 is an _ interval of waves _ at time @xmath509 if for any given @xmath512 , with @xmath513 , and for any @xmath514 @xmath860 we say that an interval of waves @xmath511 is _ homogeneous _
if for each @xmath516 , @xmath517 . if waves in @xmath511 are positive ( resp .
negative ) , we say that @xmath511 is a _ positive _ ( resp .
_ negative _ ) interval of waves .
the following proposition is the continuous version of proposition [ w_interval_waves ] .
its proof is very similar and for this reason we omit it .
[ interval_waves ] let @xmath510 be an homogeneous interval of waves and assume that waves in @xmath511 are positive ( resp .
negative ) .
then the restricion of @xmath148 to @xmath511 is strictly increasing ( resp .
decreasing ) and its image is an interval ( in @xmath179 ) . a priori an interval of waves could be a wild subset of @xmath179 : we show that it enjoys a particularly nice structure .
[ interval_countable_prop ]
let @xmath861 be an interval of waves .
then @xmath511 is an ( at most countable ) union of mutually disjoint intervals of real numbers .
write @xmath862 clearly @xmath863 .
if equality holds , then , by definition [ eow ] of enumeration of waves , @xmath864 is a finite union of intervals of real numbers .
otherwise , if @xmath865 , then define @xmath866 by definition of interval of waves , @xmath867 the r.h.s .
of is a finite union of mutually disjoint intervals of real numbers .
[ iow_are_measurable ] any interval of waves is lebesgue - measurable .
the next proposition is the area formula adapted to our setting .
[ cambio_di_variabile ] let @xmath868 be continuous , @xmath869 .
let @xmath870 be a homogeneous interval of waves and let @xmath871 .
then @xmath872 observe that the function @xmath873 is lebesgue - measurable on @xmath511 since it is composition of continuous functions .
definition [ w_artificial_speed ] and remark [ w_artificial_speed_remark ] in this context become respectively the following .
[ artificial_speed ] let @xmath537 be an interval in @xmath179 .
let @xmath35 be any positive ( resp .
negative ) wave such that @xmath874 .
the quantity @xmath875 ( resp .
@xmath876 ) is called _ the ( artificial ) speed given to the wave @xmath35 by the riemann problem @xmath195_. moreover , if @xmath543 are two waves such that @xmath877 , we say that _ the riemann problem @xmath195 divides @xmath543 _ if @xmath878 do not belong to the same wavefront of @xmath879 ( resp .
@xmath880 ) . [ artificial_speed_remark ] in the previous definition ,
if @xmath881 , where @xmath511 is any interval of waves at fixed time @xmath509 , we will also write @xmath550 instead of @xmath551 and call it the speed given to the waves @xmath35 by the riemann problem @xmath511 . moreover , we will also say that the riemann problem @xmath511 divides @xmath543 if the riemann problem @xmath195 does .
now we state the main interaction estimate for glimm scheme .
fix a wave @xmath507 and consider the function @xmath555 . by construction
it is finite valued until the time @xmath60 , after which its value becomes @xmath556 ; moreover it is piecewise constant , right continuous , with jumps located at times @xmath882 .
the following estimate holds : @xmath883 corollary [ iow_are_measurable ] , definition [ speed_function ] and proposition [ cambio_di_variabile ] gives that the integral is meaningful . to prove theorem [ main_thm ]
let us write @xmath884 we spit the grid points in interaction points , cancellation points and no - collision points .
we thus adapt definition [ w_int_canc_points ] as follows .
let @xmath885 be a grid point , @xmath886 .
we say that @xmath885 is an _ interaction point _ if @xmath887 and @xmath888 are both nonempty and have the same sign .
we say that @xmath885 is a _ cancellation point _ if @xmath887 and @xmath888 are both nonempty and have different sign .
we say that @xmath885 is a _ no - collision point _ if at least one among @xmath887 and @xmath888 is empty .
accordingly , we split into three terms : @xmath889 the three terms are estimated in the next three propositions .
[ p_no_interaction ] let @xmath885 be a no - collision point .
then @xmath890 assume first @xmath891 empty ( the case @xmath892 is similar ) .
see figure [ fig : figura33 ] . in this case
@xmath893 moreover , due to the restarting procedure of the glimm scheme , @xmath894}f(u_{n , m}).\ ] ] hence , according to definition [ speed_function ] , and by proposition [ tocca ] , for each @xmath895 , @xmath896}f(\hat u(s ) ) \\ \text{(by proposition \ref{tocca } and \eqref{f_=_inv } ) } = & ~ \frac{d}{du}\operatorname{conv}_{[u_{n-1,m-1},u_{n-1,m}]}f(\hat u ( s ) ) \\ = & ~ \sigma((n-1){\varepsilon } , s ) , \ ] ] concluding the proof .
.,width=377,height=188 ] for each grid point @xmath897 , the amount of cancellation is defined by : @xmath898 it is easy to see that for fixed @xmath899 , @xmath900 [ canc_3 ] let @xmath885 be a cancellation point .
then @xmath901 the proof is completely similar to proposition [ w_canc_3 ] . using previous proposition
one gets the same estimate as in corollary [ w_canc_4 ] .
it holds @xmath902 the last estimate to prove is @xmath903 as in the wavefront case , in the subsequent sections we will define a functional @xmath904 of the form @xmath905 where @xmath906 is _ the domain of integration _ of the functional @xmath577 at time @xmath847 and @xmath907\ ] ] is a measurable function on @xmath908 for each @xmath830 , called _ the weight of the pair of waves @xmath662 at time @xmath847_. such a functional will have two properties : for each time @xmath847 a. there is a measurable subset @xmath909 such that @xmath910 , \end{split}\ ] ] theorem [ decrease_thm_enunciato ] and corollary [ decrease_thm_corollary ] ; b. on the complement @xmath911 , @xmath912 , \end{split}\ ] ] theorem [ increase_thm_enunciato ] and corollary [ increase_thm_corollary ] .
these two properties are the analogs of and . as in section [ section_w_main_thm ] , and implies the following proposition , which is the analog of proposition [ w_thm_interaction ] .
it holds @xmath913 for fixed time @xmath847 , @xmath914 \\ = & ~ 2 \bigg [ \iint\limits_{\mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds ' \\ & ~ \qquad + \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds'\\ & ~ \qquad + \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds ' \bigg ] \\
= & ~ 2 \bigg [ \iint\limits_{\mathcal{d}(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds'\bigg ] \\ & ~ + 2 \bigg [ \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds ' \bigg ] \\ \leq&~ 2 \big [ { \mathfrak{q}}((n-1){\varepsilon } ) - { \mathfrak{q}}(n{\varepsilon } )
\big]\\ & ~ + 2 \bigg [ \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}(n{\varepsilon } , s , s')\ , dsds ' - \iint\limits_{\mathcal{d}(n{\varepsilon } ) \setminus \mathcal{d}^i(n{\varepsilon } ) } \mathfrak{q}((n-1){\varepsilon } , s , s')\ , dsds'\bigg ] \\ \text{(using \eqref{increase } ) } \leq & 2 \big [ { \mathfrak{q}}((n-1){\varepsilon } ) - { \mathfrak{q}}(n{\varepsilon } ) \big ] \\ & ~ + 2 \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \big[{\text{\rm tot.var.}}(u_{\varepsilon}((n-1){\varepsilon } ) ) - { \text{\rm tot.var.}}(u_{\varepsilon}(n{\varepsilon}))\big ] .
\end{aligned}\ ] ] hence , exactly as in proof of proposition [ w_thm_interaction ] , @xmath915 \\ & \quad + 2 \log(2 ) \sum_{n=1}^n \lvert f ''
\lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \big[{\text{\rm tot.var.}}(u_{\varepsilon}((n-1){\varepsilon } ) ) - { \text{\rm tot.var.}}(u_{\varepsilon}(n{\varepsilon}))\big ] \\ & = 2 \big[{\mathfrak{q}}(0 ) - { \mathfrak{q}}(n{\varepsilon } ) \big ] + 2 \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \big [ { \text{\rm tot.var.}}(u_{\varepsilon}(0,\cdot ) ) - { \text{\rm tot.var.}}(u_{\varepsilon}(n{\varepsilon } ) ) \big ] \\ & \leq 2 \big [ { \mathfrak{q}}(0 ) + \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))^2 \big ] \\ &
\leq 2 ( 1 + \log(2 ) ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))^2,\end{aligned}\ ] ] where the last inequality is justified by the fact that @xmath916 passing to the limit as @xmath917 , one completes the proof .
this section introduces the notion of pairs of waves which have _ already interacted / not yet interacted_. even if the definitions and the propositions are completely similar to the ones in section [ w_waves_collision ] , some technicalities arise , for example in the proof of proposition [ divise_tocca ] , which is substantially longer than the proof of the analogous proposition [ w_divise_tocca ] .
the following definition is the same as definition [ interagite_non_interagite ] .
[ d_glimm_interacted ] let @xmath509 be a fixed time and let @xmath584 .
we say that _ @xmath543 interact at time @xmath509 _ if @xmath585 .
we also say that _ they have already interacted at time @xmath509 _ if there is @xmath586 such that @xmath543 interact at time @xmath19 .
moreover we say that _ they have not yet interacted at time @xmath509 _ if for any @xmath586 , they do not interact at time @xmath19 .
lemma [ w_interagite_stesso_segno ] and lemma [ w_quelle_in_mezzo_hanno_int ] hold also in this case , namely : 1 .
if @xmath587 interact at time @xmath509 , then they have the same sign ( lemma [ w_interagite_stesso_segno ] ) ; 2 .
let @xmath509 be a fixed time , @xmath584 , @xmath52 ; assume that @xmath587 have already interacted at time @xmath509 ; if @xmath918 \cap { \mathcal{w}}(\bar t)$ ] , then @xmath591 have already interacted at time @xmath509 ( lemma [ w_quelle_in_mezzo_hanno_int ] ) . as in section [ w_waves_collision ] , for any fixed time @xmath919 and for any @xmath594 ,
define @xmath920 by lemmas [ w_interagite_stesso_segno ] and [ w_quelle_in_mezzo_hanno_int ] , this is an homogeneous interval of waves .
if @xmath52 have already interacted at a fixed time @xmath509 , set as before @xmath597 this is clearly an interval of waves and so by proposition [ interval_waves ] the image of @xmath921 is an interval in @xmath179 .
the following definition is the same as definition [ w_waves_divided ] .
[ waves_divided ] let @xmath584 be two waves which have already interacted at time @xmath509 .
we say that _ @xmath543 are divided in the real solution at time @xmath509 _ if @xmath602 i.e. if at time @xmath509 they have either different position , or the same position , but different speed .
if they are not divided in the real solution , we say that _ they are joined in the real solution_. as noted in remark [ rem_divise_solo_in_cancellazioni ] , in the wavefront tracking algorithm two waves can have same position but different speed only in cancellation points ; on the contrary , in the glimm scheme two waves can have same position but different speed at every time step .
the analog of proposition [ w_unite_realta ] is the following proposition .
we omit the proof .
[ unite_realta ] let @xmath922 be a fixed time .
let @xmath923 . if @xmath543 are not divided in the real solution at time @xmath847 , then the riemann problem @xmath924 does not divide them .
also proposition [ w_divise_tocca ] holds in this framework , but the proof is slightly different and more technical .
[ divise_tocca ] let @xmath922 be a fixed time .
let @xmath543 be two waves which have already interacted at time @xmath509 .
assume that @xmath543 are divided in the real solution , and let @xmath610 .
if @xmath611 are divided in the real solution at time @xmath509 , then the riemann problem @xmath612 divides them .
the proof is by induction on the time step @xmath171 . for @xmath834 ,
the statement is obvious .
let us assume the proposition is true for @xmath925 and let us prove it for @xmath926 .
let @xmath927 be two waves which have already interacted at time @xmath928 and assume @xmath543 to be divided in the real solution at time @xmath928 .
we can also assume w.l.o.g . that @xmath543 are both positive .
let us define @xmath929 it is easy to see that @xmath930 exist and @xmath931 .
assume first @xmath932 , which means that the @xmath933 and the @xmath934 above coincide .
see figure [ fig : figura34 ] .
.,width=415,height=226 ] in this case @xmath935 this implies @xmath936 and so the thesis is easily proved , because the artificial riemann problem coincides with the real one .
we can thus assume @xmath937 , i.e. @xmath938 .
see figure [ fig : figura35 ] . under this assumption , we first write down some useful claims . .,width=491,height=264 ] [ bounds_on_i ] it holds @xmath939.\ ] ] let @xmath940 . by and , @xmath941 , for some @xmath942 . hence , @xmath943 .
[ segno ] each wave in @xmath944 has positive sign .
use definition [ eow ] of enumeration of waves and lemma [ w_interagite_stesso_segno ] .
[ divise_dopo ] let @xmath945 .
assume @xmath611 are divided at time @xmath926 , but not divided at time @xmath925 .
then a. either @xmath946 and waves in @xmath947 are negative ; b. or @xmath948 and waves in @xmath949 are negative .
since @xmath611 are not divided at time @xmath925 , there is @xmath950 such that @xmath951 , for some @xmath952 .
assume @xmath953 , the other case is completely similar .
if @xmath954 or @xmath955 , this means that @xmath956 is either a no - collision point or an interaction point . by proposition [ differenza_vel ] , since @xmath611 are not divided at time @xmath847 , they can not be divided at time @xmath928 . hence @xmath957 and @xmath958 .
thus by claim [ segno ] , @xmath959 $ ] , i.e. @xmath960 $ ] ; on the other hand , by claim [ bounds_on_i ] , @xmath961 $ ] .
hence @xmath962 and the thesis is proved .
now for @xmath963 , set @xmath964 [ interazione ] the following holds : 1 . if @xmath965 , then @xmath966 and @xmath967 ; 2 . if @xmath968 , then @xmath969 and @xmath970 . immediate from definition of @xmath971 .
the following lemma provides some bounds on @xmath972 .
[ bdd ] the following hold : a. [ point_a_bdd ] @xmath973 ; b. [ point_b_bdd ] @xmath974 ; c. [ point_c_bdd ] @xmath975 .
we prove each point separately .
define @xmath976 clearly @xmath977 and @xmath978 . using the definition of @xmath979 and the fact that @xmath924 is an interval of waves , it is not difficult to see that @xmath980 hence , @xmath981 . if either @xmath982 is a no - collision point , or it is an interaction point , but @xmath983 , then clearly @xmath984 and then the thesis holds ; if @xmath982 is an interaction point and @xmath968 , then it must hold @xmath985 , and hence the thesis holds ; if @xmath982 is a cancellation point , then either @xmath986 and so @xmath987 , or @xmath988 and so @xmath989 , concluding the proof .
same proof as the previous point .
[ ultimo ] let @xmath990 two waves which have already interacted at time @xmath925 , but are divided at time @xmath847 . take any @xmath991 , with @xmath992 , @xmath993 .
if @xmath994 then @xmath995 } f = \operatorname{conv}_{[\inf \hat u(\mathcal{i}(n{\varepsilon},\tilde s , \tilde s ' ) ) , u_{j , m } ] } f \cup
\operatorname{conv}_{[u_{j , m } , \sup \hat u(\mathcal{i}(n{\varepsilon},\tilde s , \tilde s ' ) ) ] } f.\ ] ] we can assume that both open intervals @xmath996 are non - empty ( otherwise the proof is trivial ) .
take a sequence @xmath997 such that @xmath998 and @xmath999 and a sequence @xmath1000 such that @xmath1001 and @xmath1002 .
since @xmath1003 , then @xmath1004 are divided at time @xmath925 and then , since by inductive assumption the statement of the proposition holds at time @xmath847 , there is @xmath1005 such that @xmath995}f(u_k ) = f(u_k).\ ] ] passing to the limit as @xmath1006 and using proposition [ tocca ] , we get the thesis .
we now construct two waves @xmath1007 with the following two properties : a. [ a ] @xmath1007 are divided at time @xmath847 , b. [ b ] @xmath1008 .
notice that we do not require @xmath1007 to exist at time @xmath928 .
the procedure is : 1 . if at least one of the states @xmath878 belongs to @xmath1009 $ ] , then set @xmath1010 . by claim [ divise_dopo ] , property [ a ] ) holds . by lemma
[ bdd ] , also property [ b ] ) holds ; 2 . similarly ,
if @xmath1011 $ ] and @xmath1012 $ ] , set @xmath1010 ; 3 .
if @xmath1013 $ ] , then set @xmath1014 , and take as @xmath1015 any wave such that @xmath1016 by definition of @xmath1017 , and the assumption that @xmath938 , property [ a ] ) holds ; moreover observe that in this case @xmath1018 , while @xmath1019 and thus also property [ b ] ) holds ; 4 .
if @xmath1020 $ ] , then set @xmath1021 , and take as @xmath1022 any wave such that @xmath1023 as in the previous point , one can show that properties [ a ] ) and [ b ] ) holds .
now , by [ a ] ) , [ b ] ) and lemma [ ultimo ] one gets @xmath1024}f = \operatorname{conv}_{[\tilde l(n ) , u_{n+1,m_1 } ] } f \cup \operatorname{conv}_{[u_{n+1,m_1},u_{n+1,m_2 } ] } f \cup \operatorname{conv}_{[u_{n+1,m_2 } , \tilde r(n ) ] } f.\ ] ] [ lemma_riemann_dopo_sep ]
it holds @xmath1025 } f = \operatorname{conv}_{i_1 } f \cup \operatorname{conv}_{i_2 } f \cup \operatorname{conv}_{i_3 } f,\ ] ] where @xmath1026 , \qquad i_2 : = [ u_{n+1,m_1},u_{n+1,m_2 } ] , \qquad i_3 : = [ u_{n+1,m_2 } , r(n+1)].\end{aligned}\ ] ] we consider four cases .
if at least one among @xmath878 belongs to @xmath1009 $ ] , then by our definition @xmath1027 .
moreover , by claim [ interazione ] , @xmath1028 , @xmath1029 ; hence , by and corollary [ stesso_shock ] , we get the thesis .
if @xmath1011 $ ] and @xmath1012 $ ] , argue as in previous case .
if @xmath1013 $ ] , then by our definition @xmath1014 , and @xmath1015 is any wave such that @xmath1030 . observe that in this case the riemann problem @xmath1031 $ ] is not solved by a single wavefront ( since @xmath543 at time @xmath928 are divided ) .
thus , using this fact , and proposition [ incastro ] , we get @xmath1032 } f = \operatorname{conv}_{[\tilde l(n ) , u_{n+1,m_1 } ] } f \cup \operatorname{conv}_{i_2 } f \cup \operatorname{conv}_{[u_{n+1,m_2 } , u_{n+1 , m_2 + 1 } ] } f.\ ] ] since @xmath1033 and @xmath1034 , using corollary [ stesso_shock ] , we get the thesis .
the case @xmath1020 $ ] is similar to previous point .
_ conclusion of the proof of proposition [ divise_tocca ] .
_ take @xmath1035 , divided at time @xmath928 .
we have to prove that the riemann problem @xmath1036 $ ] divides them .
we can assume @xmath1037 for some @xmath1038 , where @xmath1039 are the intervals defined in lemma [ lemma_riemann_dopo_sep ] ( otherwise the proof is trivial ) .
if @xmath1040 $ ] , then by claim [ divise_dopo ] @xmath611 are divided at time @xmath847 and so by inductive assumption the riemann problem @xmath1041 $ ] divides them . by
, the riemann problem @xmath1042 $ ] divides them and so by , also the riemann problem @xmath1036 $ ] divides @xmath611 , which is what we wanted to prove .
assume now @xmath1043 $ ] , the case @xmath1044 being similar .
we know @xmath611 are divided at time @xmath928 .
this means that the riemann problem @xmath1031 $ ] divides them . by lemma [ bdd ] , @xmath1045 and so also the riemann problem @xmath1046 $ ] divides @xmath611 .
hence , by , the riemann problem @xmath1036 $ ] divides @xmath611 , which is what we wanted to prove . in this last section
we consider the same functional @xmath577 defined in section [ w_functional_q ] for the wavefront tracking algorithm , adapted to the glimm scheme and we prove inequalities and . differently from the wavefront tracking , where @xmath577 is defined as a finite sum of weights , here @xmath577
is defined as an integral and for this reason we have also to prove that it is well defined .
recall from the definition of the domain of integration : @xmath1047 next define the _ weight of the pair of waves @xmath662 at fixed time @xmath847 _ as @xmath1048 as an easy consequence of theorem [ convex_fundamental_thm ] , point ( [ convex_fundamental_thm_3 ] ) , we obtain that @xmath667 takes values in @xmath1049 $ ] .
finally set @xmath1050 first of all we have to prove that @xmath577 is well defined , i.e. @xmath122 is integrable .
actually in the following proposition we prove an additional regularity property on @xmath1051 .
[ w_partitioned ] it is possible to write @xmath1051 as an ( at most ) countable union of mutually disjoint interval of waves @xmath1052 , @xmath1053 such that for each @xmath320 and for each @xmath1054 , @xmath1055 .
since derivation of convex envelopes is a borel operation , it follows that @xmath1056 is borel . for each @xmath1057 ,
define @xmath1058 we claim that each @xmath1059 is an interval of waves and for fixed @xmath830 , the cardinality of @xmath1060 is at most countable .
this is proved using the following lemmas .
[ l_eh_1 ] @xmath1059 is an interval of waves at time @xmath847 .
let @xmath1061 and let @xmath1062 such that @xmath1063 .
we have to prove that @xmath1064 .
we have @xmath1065 by contradiction , assume @xmath1066 , i.e. @xmath1067 .
hence , there is @xmath1068 assume @xmath1069 . hence @xmath1070 .
thus , whatever the position of @xmath1071 is , since @xmath1072 , @xmath1071 must have already interacted with @xmath1062 , a contradiction . on the other hand , if @xmath1073 , whatever the position of @xmath1071 is , @xmath1071 must have already interacted either with @xmath601 or with @xmath1074 .
thus @xmath1075 , a contradiction .
[ l_open ] let @xmath1057 .
there exist @xmath1076 such that 1 .
@xmath1077 $ ] ; 2 . for each @xmath1078 $ ] , @xmath1079 .
the proof is by induction on the time step @xmath830 . for @xmath1080 , assume @xmath1081 ; it is sufficient to choose @xmath1082 : = { \mathcal{w}}(0,m{\varepsilon})$ ]
. now assume the lemma holds for @xmath171 and prove it for @xmath1083 .
assume @xmath1084 at time @xmath928 ( the case @xmath1085 is completely similar ) .
moreover assume @xmath1086 $ ] to be the interval at time @xmath847 with the properties 1 , 2 of the statement of the lemma .
for time @xmath928 define @xmath1087 : = ( s_1 , s_2 ] \cap { \mathcal{w}}^{(0)}((n+1){\varepsilon } , m{\varepsilon}).\ ] ] clearly property ( 1 ) holds .
moreover , if property ( 2 ) does not hold , there must be a wave @xmath1088 $ ] and another wave @xmath1062 which has interacted neither with @xmath35 nor with @xmath601 at time @xmath847 , but interacts with only one among @xmath1089 at time @xmath928 .
this is impossible , since at time @xmath928 , @xmath1090 have the same position .
[ l_eh_2 ] for fixed @xmath830 , for each @xmath1091 , either @xmath1092 or @xmath1093 .
assume there is @xmath1094 . by simmetry , it is sufficient to prove that @xmath1095 .
take @xmath1096 . then @xmath1097 .
hence @xmath1098 . for fixed @xmath830 ,
@xmath1099 is at most countable . by lemma [ l_open ] ,
the interior of @xmath1059 is not empty .
the conclusion follows from separability of @xmath179 .
_ conclusion of the proof of proposition [ w_partitioned ] .
_ as an immediate consequence , we have that @xmath1100 is a countable family of pairwise disjoint sets .
let us fix @xmath1101 and let us define @xmath1102 . for each @xmath51
such that @xmath1103 is an interaction point , let us define @xmath1104 and @xmath1105 then define @xmath1106 [ decrease_thm_enunciato ] for any interaction point @xmath1103 it holds @xmath1107 .
\end{split}\ ] ] for simplicity , let us set @xmath1108 , @xmath1109 .
we can assume @xmath1110 ( otherwise the statement is trivial ) and all the waves in @xmath1111 to be positive .
moreover define , as in proof of theorem [ w_decreasing ] , @xmath1112 then set @xmath1113 and @xmath1114 .,width=453,height=283 ] see figure [ fig : figura30 ] .
moreover , given any positive interval of waves with nonempty interior , define _ the mean speed of waves in @xmath511 at time @xmath847 _ as @xmath1115 observe that for each @xmath1116 , @xmath1117 is equal to some constant @xmath786 , while @xmath1118 } f(\hat u(s ) ) & \text{if } s \in \mathcal{l } , \\
\frac{d}{du}\operatorname{conv}_{[u_m , u_r ] } f(\hat u(s ) ) & \text{if } s \in \mathcal{r}. \end{cases}\ ] ] let us prove now the following [ um_tocca ] it holds @xmath1119 } f(u_m ) = f(u_m)$ ] .
assume @xmath1120 ( otherwise trivial ) .
take any sequence @xmath1121 in @xmath1122 such that @xmath1123 , @xmath1124 and @xmath1125 . in a similar way ,
take another sequence @xmath1126 in @xmath1127 such that @xmath1128 , @xmath1129 and @xmath1130 . for @xmath326 sufficiently large , @xmath1131 and so , by proposition [ divise_tocca ] , @xmath1132
are divided by the riemann problem @xmath1133 ; this means that there is some point @xmath1134 such that @xmath1135 finally , using proposition [ convex_approximation ] , one completes the proof .
we now complete the proof of the theorem .
the steps are the same as in proof of theorem [ w_decreasing ] .
_ by definition of @xmath1111 , for any @xmath688 , @xmath543 are not divided in the real solution at time @xmath928 ; then , by proposition [ unite_realta ] , they are not divided by riemann problem @xmath971 ; hence , @xmath1136 .
_ step 2 and 3 .
_ as in theorem [ w_decreasing ] , @xmath1137 \cdot \frac{(u_m - u_l)(u_r - u_m)}{u_r - u_l } \\ = & ~ 2 \cdot \frac{|\sigma_m(\mathcal{l } ) - \sigma_m(\mathcal{r})|}{|\mathcal{l}| + |\mathcal{r}| } \ |\mathcal{l}| |\mathcal{r}| \\
\leq & ~ 2 \bigg [ \frac{|\sigma_m(\mathcal{l}_1 ) - \sigma_m(\mathcal{r})|}{|\mathcal{l}| + |\mathcal{r}| } \ |\mathcal{l}_1| |\mathcal{r}| + \frac{|\sigma_m(\mathcal{l}_2 ) - \sigma_m(\mathcal{r}_1)|}{|\mathcal{l}| + |\mathcal{r}| }
\ |\mathcal{l}_2| |\mathcal{r}_1| \\ & ~ + \frac{|\sigma_m(\mathcal{l}_2 ) - \sigma_m(\mathcal{r}_2)|}{|\mathcal{l}| + |\mathcal{r}| } \ |\mathcal{l}_2| |\mathcal{r}_2| \bigg]\\ \leq & ~ 2\bigg [ \lvert f '' \lvert_{l^\infty } |\mathcal{l}_1| |\mathcal{r}| + \frac{|\sigma_m(\mathcal{l}_2 ) - \sigma_m(\mathcal{r}_1)|}{|\mathcal{l}| + |\mathcal{r}| } \
|\mathcal{l}_2| |\mathcal{r}_1| + \lvert f '' \lvert_{l^\infty } |\mathcal{l}_2| |\mathcal{r}_2| \bigg ] .
\end{split}\ ] ] _ step 4 .
_ let us now concentrate our attention on the second term of the last summation . as a consequence of lemma [ um_tocca ] , we get @xmath1138}f(u ' ) - \frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(u )
\bigg]dudu ' \bigg | \\ = & ~ \bigg | \int_{\mathcal{l}_2 } \int_{\mathcal{r}_1 } \bigg[\frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(\hat u(s ' ) ) - \frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(\hat u(s ) ) \bigg]\ , dsds ' \bigg | . \end{split}\ ] ] we observe that , by definition of @xmath1139 , for any @xmath704 and @xmath705 , if @xmath543 have already interacted at time @xmath847 , then @xmath1140 $ ] . together with lemma [ um_tocca ] and proposition [ vel_aumenta ] , this yields @xmath1141}f(\hat u(s ' ) ) - \frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(\hat u(s ) ) \leq \sigma(\mathcal{i}(n{\varepsilon},s , s'),s ' ) - \sigma(\mathcal{i}(n{\varepsilon},s , s'),s ) , \ ] ] and hence @xmath1142}f(\hat u(s ' ) ) - \frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(\hat u(s))}{|\mathcal{l}|+|\mathcal{r}| } \leq&~ \frac{\sigma(\mathcal{i}(n{\varepsilon},s , s'),s ' ) - \sigma(\mathcal{i}(n{\varepsilon},s , s'),s)}{|\mathcal{l}|+|\mathcal{r}| } \\
\leq&~ \frac{\sigma(\mathcal{i}(n{\varepsilon},s , s'),s ' ) - \sigma(\mathcal{i}(n{\varepsilon},s , s'),s)}{\hat u(s ' ) - \hat u(s ) } \\ = & ~ \mathfrak{q}(n{\varepsilon},s , s ' ) .
\end{split}\ ] ] instead , if @xmath543 have not yet interacted at time @xmath847 , using lagrange s theorem , we get @xmath1143}f(\hat u(s ' ) ) - \frac{d}{du}\operatorname{conv}_{[u_1,u_2]}f(\hat u(s))}{|\mathcal{l}|+|\mathcal{r}| } \leq \lvert f '' \lvert_{l^\infty } = \mathfrak{q}(n{\varepsilon},s , s').\ ] ] thus , by , , , @xmath1144 _ step 5 . _
finally observe that if @xmath712 and @xmath678 , then by definition of @xmath713 , @xmath543 have not yet interacted at time @xmath847 . the same holds if @xmath704 and @xmath715 .
hence , recalling , we get @xmath1145 , \\ & \leq 2 \bigg [ \iint_{\mathcal{l}_1 \times \mathcal{r } } \mathfrak{q}(n{\varepsilon},s , s')\ , dsds ' + \iint_{\mathcal{l}_2 \times \mathcal{r}_1 } \mathfrak{q}(n{\varepsilon},s , s')\ , dsds ' + \iint_{\mathcal{l}_2 \times \mathcal{r}_2 } \mathfrak{q}(n{\varepsilon},s , s')\ , dsds ' \bigg ] \\ & = 2\iint_{\mathcal{l } \times \mathcal{r } } \mathfrak{q}(n{\varepsilon},s , s')\ , dsds ' \\ & = 2 \bigg [ \iint_{\mathcal{l } \times \mathcal{r } } \mathfrak{q}(n{\varepsilon},s , s')\ , dsds ' - \iint_{\mathcal{l } \times \mathcal{r } } \mathfrak{q}((n+1){\varepsilon},s , s')\ , dsds'\bigg ] , \end{aligned } \end{gathered}\ ] ] where the last equality is a direct consequence of point 1 . as a corollary
, we immediately obtain inequality .
[ decrease_thm_corollary ] it holds @xmath1146.\end{aligned}\ ] ] immediate consequence of theorem [ decrease_thm_enunciato ] , definition of @xmath1102 in and the fact that for @xmath1147 , @xmath1148 and @xmath1149 are disjoint .
let @xmath847 be a fixed time .
first of all let us prove the following [ maximal_hom_iow ] it is possible to write @xmath1051 as a ( locally finite ) countable union of mutually disjoint , maximal ( with respect to set inclusion ) homogenous interval of waves @xmath1150 , @xmath1151 it is sufficient to partition @xmath1051 with respect to the equivalence relation @xmath1152 \cap { \mathcal{w}}(n{\varepsilon } ) \text { is an homogeneous interval of waves}.\ ] ] this partition is countable , since any two waves having the same position belong to the same equivalence class . for any interval @xmath1153 ,
let us define the following quantities : @xmath1154 then set @xmath1155 and , in a similar way , @xmath1156 finally define @xmath1157 and @xmath1158 clearly @xmath1159 .
[ increase_thm_enunciato ] for any maximal interval @xmath1153 , it holds @xmath1160^+ \ , dsds ' \\
\leq&~ \log(2 ) \lvert f '' \lvert_{l^\infty } \mathcal{l}^1(\mathcal{m}_k ) \big [ ( b_1^{(k)}-b_2^{(k ) } ) + ( a_2^{(k ) } -a_1^{(k)})\big ] .\end{aligned}\ ] ] we will not write the index @xmath326 and we assume that waves in @xmath1161 are positive , the other case being entirely similar .
first of all let @xmath1162 \to { \mathbb{r}}$ ] be two @xmath1163 function with the following properties : @xmath1164 } & = \operatorname{conv}_{[a_1,a_3]}f , & h_{|[a_2,a_3 ] } & = \operatorname{conv}_{[a_2,a_3]}f , \\
g_{|[b_3,b_1 ] } & = \operatorname{conv}_{[b_3,b_1]}f , & h_{|[b_3,b_2 ] } & = \operatorname{conv}_{[b_3,b_2]}f , \\\end{aligned}\ ] ] and set @xmath1165 } = h_{[a_3,b_3]}.\ ] ] see figure [ fig : figura31 ] .
it is easy to see that some @xmath1166 with the above properties exist . and
@xmath205.,width=453,height=283 ] [ lemma_uno ] for any @xmath1167 , @xmath1168^+ \leq \frac{\big| \big ( h'(\hat u(s ' ) ) - h'(\hat u(s ) ) \big ) - \big ( g'(\hat u(s ' ) ) - g'(\hat u(s ) ) \big ) \big|}{|\hat u(s')- \hat u(s)|}.\ ] ] if @xmath543 have not yet interacted at time @xmath847 , then the l.h.s . of is equal to zero
. thus we can assume @xmath543 have already interacted at time @xmath847 . in this case set for simplicity @xmath1169 the proof reduces to prove the following inequality : @xmath1170^+ \leq \big| \big ( h'(\hat u(s ' ) ) - h'(\hat u(s ) ) \big ) - \big ( g'(\hat u(s ' ) ) - g'(\hat u(s ) ) \big ) \big|.\ ] ] if @xmath543 are not divided in the real solution at time @xmath928 , then , by proposition [ unite_realta ] , @xmath1171 ; in this case the l.h.s . of is equal to zero .
let us then assume @xmath543 are divided at time @xmath928 .
we consider four main cases .
_ case 1 . _
assume @xmath1172 $ ] .
in this case let us observe what follows : 1 . the r.h.s . of is equal to zero ; 2 .
@xmath543 are divided also at time @xmath847 , due to their position ; 3 .
@xmath1173 \subseteq \mathcal{i}_{n+1 } \cap ( a_3,b_3]$ ] , because in @xmath1174 $ ] only interactions can occur ; 4 . for @xmath1175 and @xmath1176 , if @xmath1177 , then @xmath1178 .
indeed let us prove this equality for @xmath1179 , the other ones being similar . if @xmath1180 , we have done .
so let us assume @xmath1181 .
then one can find two sequences @xmath1182 in @xmath1183 such that @xmath1184 and @xmath1185 , while @xmath1186 and @xmath1187 and , for each @xmath326 , @xmath1188 have already interacted at time @xmath928 .
hence , since @xmath543 are divided at time @xmath928 , by proposition [ divise_tocca ] , @xmath1189 are divided by the riemann problem @xmath1183 , i.e. there is @xmath1005 such that @xmath1190 .
passing to the limit , we get the thesis .
thus for @xmath1175 , @xmath1191 } f$ ] on @xmath1192 $ ] .
hence , @xmath1193 } f \bigg ) \big(\hat u(s')\big ) - \bigg ( \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_n)\cap [ a_3,b_3 ] } f \bigg ) \big(\hat u(s)\big ) \\
( \text{by prop . \ref{vel_aumenta } } ) \ \geq & ~ \bigg ( \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n+1 } ) \cap [ a_3,b_3 ] } f \bigg ) \big(\hat u(s')\big ) - \bigg ( \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n+1})\cap [ a_3,b_3 ] } f \bigg ) \big(\hat u(s)\big ) \\
= & ~ \bigg ( \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n+1 } ) } f \bigg ) \big(\hat u(s')\big ) - \bigg ( \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n+1 } ) } f \bigg ) \big(\hat u(s)\big ) \\ = & ~ \sigma_{n+1}(s ' ) - \sigma_{n+1}(s).\end{aligned}\ ] ] thus , the l.h.s . of
is equal to zero
. assume now @xmath1194 $ ] .
in this case @xmath1195 since @xmath543 are divided at time @xmath928 in the real solution , we can argue as in the previous case and use proposition [ divise_tocca ] to obtain @xmath1196 } f = \operatorname{conv}_{[b_3,b_2 ] } f = h \hspace{1 cm } \text{on } [ b_3,b_2].\end{aligned}\ ] ] hence @xmath1197 now distinguish two possibilities : 1 .
@xmath1198 : in this case @xmath878 belong to the same wavefront interval of @xmath1199 } = \operatorname{conv}_{[b_3,b_1]}f$ ] ; thus they are not divided in the real solution at time @xmath847 and so by proposition [ unite_realta ] , @xmath1200 , which , together with , yields the thesis .
@xmath1201 : this means that @xmath543 are divided in the real solution at time @xmath847 and so , using the same argument as before , by proposition [ divise_tocca ] , @xmath1202 } f = \operatorname{conv}_{[b_3,b_1 ] } f = g \hspace{1 cm } \text{on } [ b_3,b_1],\ ] ] which , together with , yields the thesis .
assume now @xmath1203 $ ] and @xmath1204 $ ] .
in this case @xmath543 are divided in the real solution also at time @xmath847 .
hence , arguing as in the first point , @xmath1205 } = \operatorname{conv}_{\hat u(\mathcal{i}_n ) \cap [ a_3,b_3]}f \cup \operatorname{conv}_{\hat u(\mathcal{i}_n ) \cap [ b_3,b_1]}f,\ ] ] and @xmath1206 } = \operatorname{conv}_{\hat u(\mathcal{i}_{n+1 } ) \cap [ a_3,b_3]}f \cup \operatorname{conv}_{\hat u(\mathcal{i}_{n+1 } ) \cap [ b_3,b_2]}f.\ ] ] thus , observing that @xmath1207 = \hat u(\mathcal{i}_{n } ) \cap [ a_3,b_3]\ ] ] ( @xmath543 can not interact at time @xmath928 ) , @xmath1208 } f(\hat u(s ) ) \\ & \ , \overset{\eqref{b3}}{= } \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n } ) \cap [ a_3,b_3 ] } f(\hat u(s ) ) \overset{\eqref{b1}}{= } \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n } ) } f(\hat u(s ) ) = \sigma_n(s),\end{aligned}\ ] ] from which we deduce @xmath1209 for @xmath119 , distinguish two cases : 1 . if @xmath1210 , then @xmath1211 = \hat u(\mathcal{i}_{n } ) \cap [ b_3,b_1]$ ] , and so @xmath1212 } f(\hat u(s ' ) ) \\
= & ~ \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n } ) \cap [ b_3,b_1 ] } f(\hat u(s ' ) ) \overset{\eqref{b1}}{= } \frac{d}{du}\operatorname{conv}_{\hat u(\mathcal{i}_{n } ) } f(\hat u(s ' ) ) = \sigma_n(s ' ) .
\end{aligned}\ ] ] 2 .
if @xmath1213 , then @xmath1214 and so @xmath1212 } f(\hat u(s ' ) ) \\ = & ~ \frac{d}{du}\operatorname{conv}_{[b_3,b_2 ] } f(\hat u(s ' ) ) = h(\hat u(s ' ) ) , \end{aligned}\ ] ] while @xmath1215 } f(\hat u(s ' ) ) \\
\geq&~ \frac{d}{du}\operatorname{conv}_{[b_3,b_1 ] } f(\hat u(s ' ) ) = g(\hat u(s ' ) ) , \end{aligned}\ ] ] where the inequality is due to proposition [ vel_aumenta ] . in both cases ( 1 ) and ( 2 ) , @xmath1216 .
together with , this yields the thesis .
assume now @xmath1217 $ ] , @xmath1218 $ ] .
as in the previous point , @xmath1219 and so the thesis follows .
the proof of the remaining cases @xmath1217 $ ] , @xmath1220 $ ] and @xmath1217 $ ] , @xmath1221 $ ] is similar to case ( 3 ) and case ( 2 ) respectively .
[ lemma_due ] we have @xmath1222.\end{aligned}\ ] ] the proof is the analog of lemma [ w_lemma_due ] . in fact @xmath1223d\xi \bigg|du'du \\ & \leq \int_{a_2}^{b_2 } \int_{u}^{b_2 } \frac{1}{u'- u } \int_u^{u ' } |h''(\xi ) - g''(\xi)|d\xi du'du \\ & \leq \int_{a_2}^{b_2 } \big|h''(\xi ) - g''(\xi)\big| \bigg(\int_{a_2}^\xi \int_\xi^{b_2 } \frac{1}{u'- u } du'du \bigg ) d\xi . \end{aligned}\end{gathered}\ ] ] hence , by , @xmath1224 .
\end{aligned}\end{gathered}\ ] ] now remember that @xmath163 and @xmath101 coincide on @xmath1225 $ ] . moreover , by corollary [ rp_ridotto ] , there are @xmath1226 , \bar b \in [ b_3,b_2]$ ] such that @xmath308 on @xmath1227 \cup [ b_3 , \bar b]$ ] and @xmath163 is affine on @xmath1228 $ ] and @xmath1229 $ ] .
thus @xmath1230 \\ & = \log(2 ) \mathcal{l}^1(\mathcal{m } ) \bigg [ \int_{a_2}^{\bar a } h''(\xi ) d\xi + \int_{\bar b}^{b_2 } h''(\xi ) d\xi \bigg ] \\ & = \log(2 ) \mathcal{l}^1(\mathcal{m } ) \big [ \big(h'(\bar a ) - h'(a_2)\big ) + \big(h'(b_2 ) - h'(\bar b)\big ) \big ] \\ &
= \log(2 ) \mathcal{l}^1(\mathcal{m } ) \big [ \big(g'(a_2 ) - h'(a_2)\big ) + \big(h'(b_2 ) - g'(b_2)\big ) \big ] , \end{aligned}\end{gathered}\ ] ] concluding the proof of lemma [ lemma_due ] .
_ end of the proof of theorem [ decrease_thm_enunciato]_. let us observe that , by proposition [ diff_vel_proporzionale_canc ] , @xmath1231 putting together lemma [ lemma_uno ] , lemma [ lemma_due ] and inequalities , one easily concludes the proof of the theorem .
[ increase_thm_corollary ] it holds @xmath1232\ , dsds'\\ & \leq \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot ) ) \big[{\text{\rm tot.var.}}(u_{\varepsilon}(n{\varepsilon},\cdot ) - { \text{\rm tot.var.}}(u_{\varepsilon}((n+1){\varepsilon } , \cdot))\big ] .
\end{aligned}\ ] ] we first observe that @xmath1233\ , dsds'\\ \leq&~ \iint_{\mathcal{d}((n+1){\varepsilon } ) \setminus \mathcal{d}^i((n+1){\varepsilon } ) } \big [ \mathfrak{q}((n+1){\varepsilon } , s , s ' ) - \mathfrak{q}(n{\varepsilon } , s , s ' ) \big]^+\ , dsds'\\ \leq&~ \iint_{\mathcal{d}((n+1){\varepsilon } ) } \big [ \mathfrak{q}((n+1){\varepsilon } , s , s ' ) - \mathfrak{q}(n{\varepsilon } , s , s ' ) \big]^+\ , dsds ' .
\end{split}\ ] ] now , if @xmath1234 , @xmath1235 , with @xmath1236 , then @xmath543 have not yet interacted at time @xmath847 ; this means that @xmath1237 and hence @xmath1238^+ = 0.\ ] ] thus we can continue the chain of inequalities in as follows @xmath1239^+\ , dsds ' \\ & \leq \sum_{k \in { \mathbb{z } } } \iint_{(\mathcal{m}_k \times \mathcal{m}_k ) \cap \mathcal{d}((n+1){\varepsilon } ) } \big [ \mathfrak{q}((n+1){\varepsilon } , s , s ' ) - \mathfrak{q}(n{\varepsilon } , s , s ' ) \big]^+\ , dsds ' \\ & \leq \log(2 ) \lvert f '' \lvert_{l^\infty } { \text{\rm tot.var.}}(u(0,\cdot))\sum_{k \in { \mathbb{z } } } \big [ ( b_1^{(k)}-b_2^{(k ) } ) + ( a_2^{(k ) } -a_1^{(k ) } ) \big],\end{aligned}\ ] ] where the last inequality is an immediate consequence of theorem [ decrease_thm_enunciato ] . by the definition of @xmath1240 , @xmath1241 , @xmath1242 , @xmath1243 , @xmath1244 = & ~ \big| { \mathcal{w}}(n{\varepsilon } ) \setminus { \mathcal{w}}((n+1){\varepsilon } ) \big| \\ = & ~ { \text{\rm tot.var.}}(u_{\varepsilon}(n{\varepsilon},\cdot ) ) - { \text{\rm tot.var.}}(u_{\varepsilon}((n+1){\varepsilon } , \cdot ) ) . \ ] ] and this concludes the proof .
as we said in the introduction , in order to obtain the estimate and hence to prove theorem [ main_thm ] , the authors restrict the evaluation of the oscillation of @xmath1245 to regions belonging to a disjoint partition of the plane : each region @xmath1246 is called a _ tree _ in @xcite .
we give the definition of tree only for a wavefront solution to a scalar conservation law , because it is definitely simpler .
[ def_tree ] if at @xmath1247 a splitting occurs , then the _ tree starting at @xmath20 _ is the collection of all the backward trajectories @xmath1249 of the waves @xmath1250 surviving after the cancellation , such that either @xmath1251 or @xmath1252 is the last time when a splitting involving @xmath1250 occurs before @xmath509 .
the fundamental estimate is thus the one contained in @xcite , lemma 2 : the oscillation of @xmath105 restricted to a tree @xmath1253 is bounded by @xmath1254 in } \mathcal t \big\}.\ ] ] where the cubic interactions term corresponds by definition to the decreases of the functional @xmath18 defined in , on the tree @xmath1255 .
in fact in the scalar case no waves of other families are present .
they thus conclude that @xmath1256 in } \mathcal t \big\ } \crcr \leq&~ \mathcal o(\|f''\|_{l^\infty } ) \big\ { \text{cubic interaction estimates and [ cancellations$\times { \text{\rm tot.var.}}(u)$ ] in } { \mathbb{r}}^+ \times { \mathbb{r}}\big\ } \crcr \leq&~ \mathcal o(\|f''\|_{l^\infty } ) { \text{\rm tot.var.}}(\bar u)^2.\end{aligned}\ ] ] a key point in the proof is estimates ( 4.84 ) , ( 4.85 ) in @xcite , where the authors bounds the variation in time of the speeds of waves belonging to a given tree in terms of the cubic amount of interaction and the cancellation .
now we present an explicit counterexample of a tree in which this estimate fails .
the idea is that there can be a large interaction in a tree , and a vary small cancellation at its top , corresponding to a splitting of some waves .
then the quantity @xmath1257 estimated at the interaction time is of the order of @xmath1258 , and thus the variation of speed of the wavefronts is of order @xmath1259 ( see ) .
the cancellation can be made arbitrarily small , so that the estimate which should be proved becomes @xmath1260 \\ \leq & ~ \mathcal o(\|f''\|_{l^\infty } ) { \text{\rm tot.var.}}(u)^3.\end{aligned}\ ] ] the solution of the riemann problem at @xmath1272 is a single shock @xmath1273 of strength @xmath1274 traveling with speed @xmath1275 , while the solution of the riemann problem at @xmath1271 has a first shock @xmath1276 $ ] of strength @xmath1274 traveling with speed @xmath1277 a rarefaction @xmath1278 $ ] and a second shock @xmath1279 $ ] traveling with speed @xmath1280 . since every wavefront starting from @xmath1281 and connecting to a point @xmath1282 has positive speed ,
then after a time @xmath1283 all the waves have collided into a single wavefront with a speed @xmath1284 now consider a contact discontinuities with size @xmath1285 $ ] , located at @xmath1286 .
since it is traveling with speed @xmath1287 , this discontinuity meets the other waves at @xmath1288 , i.e. after the large wavefront @xmath1289 $ ] has formed , and it is fairly easy to see that this point corresponds to a splitting .
hence this wave configuration corresponds to a tree .
we now estimate the amount of interaction calculated as the decrease of the cubic functional @xmath18 of @xcite : since it is bounded by the area , then we can write @xmath1290 . \end{split}\ ] ] the amount of cancellation is @xmath71 .
for the couple of wavefronts @xmath1291 , formula ( 4.84 ) of @xcite bounds the variation of speed of @xmath1273 between @xmath789 and the splitting time @xmath1292 by @xmath1293 \\ = & ~ { \mathcal{o}(1)}\big [ \alpha ( l+{\varepsilon})^3 + { \varepsilon}\big].\end{aligned}\ ] ] we have used . on the other hand , an explicit computation gives @xmath1294 it is clear that since we can choose @xmath1295 and @xmath1296 , then we can not bound the difference in speed with the amount of interaction and cancellation . | we prove a quadratic interaction estimate for approximate solutions to scalar conservation laws obtained by the wavefront tracking approximation or the glimm scheme . this quadratic estimate has been used in the literature to prove the convergence rate of the glimm scheme .
the proof is based on the introduction of a quadratic functional @xmath0 , decreasing at every interaction , and such that its total variation in time is bounded . differently from other interaction potentials present in the literature , the form of this functional is the natural extension of the original glimm functional , and coincides with it in the genuinely nonlinear case . |
thermally activated escape over a potential barrier is ubiquitous in physics , chemistry and biology ( see @xcite , @xcite , @xcite , @xcite , @xcite and refs . therein ) .
this phenomenon is important for both quantum @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and classical @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite,@xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite systems .
it can proceed in strong friction ( overdamped ) , weak friction ( underdamped ) and needless to say intermediate regimes ( see refs . above ) .
the case of thermally activated escape unperturbed by additional external influences pioneered by kramers is exhaustively investigated and by now is a well understood phenomenon .
however in most physical realizations the thermally activated escape is modulated by some external driving . in this case
the stationary limit that may be a rather good approximation in the absence of external driving is already inapplicable . in the presence of the latter
the system becomes intrinsically non - equilibrium and the problem in known to be notoriously difficult for analytical treatment @xcite .
the particular case of periodically driven escape @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite is relevant among others for chemical physics @xcite , @xcite , @xcite ( where a chemical reaction can be influenced by , e.g. , laser electric field ) and enzymology @xcite ( where an enzymatic reaction can be influenced by an oscillating electric field produced by the dynamics of protein structure @xcite ) . according to @xcite the periodic driving force `` ' '
heats up `` the system by changing its effective temperature thus giving rise to lowering of the activation energy of escape which can be much bigger than the real temperature even for comparatively weak fields '' .
revealing the physical aspects of enzyme action may well become one of the most important grounds for application of the periodically modulated thermally activated escape theory . in support of this point of view
it is worthy to note that understanding the role of driving at activated escape in biological systems is considered by the authors of @xcite as `` a fundamentally important and most challenging open scientific problem '' .
the reasons for the above statement are as follows .
the problem of enzyme catalysis is the main unsolved interdisciplinary enigma of molecular biophysics , biochemistry and needless to say enzymology . up to now
there is no definite and commonly accepted understanding of `` how does an enzyme work ? ''
( see , e.g. , heat controversy at a recent conference in the subject issue of phil .
b ( 2006 ) 361 ) .
the idea that dynamical effects may play a crucial role at enzyme action is very popular at present and the concept of the so - called rate promoting vibration is a central one in modern enzymology .
however the practitioners engaged in chemical enzymology traditionally discuss the phenomenon of enzyme catalysis in terms and notions of the transition state theory ( as can be seen from the materials of the above mentioned conference ) .
the latter is essentially equilibrium one that is embedded into its cornestone postulate and is poorly suited for taking into account dynamical effects in a reaction rate .
the main tool to study such effects is the kramers theory @xcite , @xcite , @xcite , @xcite .
regretfully at present this theory is much less known and necessitated for applications in enzymology than its transition state theory counterpart .
the thermally activated escape problem at periodic driving in the overdamped classical regime was conceptually solved in the papers @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
two mutually complementary theories @xcite , @xcite , @xcite and @xcite , @xcite , @xcite , @xcite , @xcite , @xcite provide deep insights on the behavior of many physical values of interest .
they are based on a physical idea of the optimal path and are able to provide an imaginable picture of the process .
the papers @xcite , @xcite , @xcite provide description of the escape rate enhancement at weak modulation ( the so - called log - linear regime ) where the change of the activation energy is linear in the modulation amplitude .
most important of all , the escape rate enhancement exhibits the replacement of the log - linear regime by more moderate growth with the increase of modulation amplitude to noise intensity ratio .
such behavior for the intermediate regime of moderately strong and moderately fast driving is well described by the @xcite , @xcite theory that is corroborated by high - precision numerical results .
the scaling bahavior of the prefactor near the bifurcation point is investigated in details @xcite , @xcite , @xcite , @xcite .
in particular the so called adiabatic regime is most thorougly investigated @xcite , @xcite , @xcite , @xcite .
this regime is defined by the authors of @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite as the limit of slow modulation @xmath2 where `` the driving frequency is small compared to the relaxation rate in the absence of fluctuations and the system remains in quasi - equilibrium '' and by the authors of the @xcite , @xcite , @xcite theory as the one that `` goes up to driving frequencies of the order of the inverse instanton time which is related to the curvatures of the potential '' . in the adiabatic regime the scaling behavior of the prefactor for the cubic potential
is given by a simple analytical formula @xcite , @xcite .
however the existing literature leaves room for parallel activity for the following reasons . to attract attention of chemists and biochemists to the modulated thermally activated escape theory within the kramers approach it is necessary to present its final results in as simple and understandable form as is done , e.g. , in the transition state theory . on the contrary the results of both @xcite , @xcite , @xcite and @xcite , @xcite , @xcite , @xcite , @xcite ,
@xcite theories are presented via involved notions and values characterizing the optimal action corresponding to the minimizing path .
correct making use of these results requires profound comprehension of their physical content and mastering in depth the methods involved in their deriving . as a matter of fact people engaged in applications ( whom the author of the present manuscript belongs to ) as a rule
are concerned with much more modest objective : how at a given combination from the parameter space ( noise intensity @xmath3 , modulation amplitude @xmath4 , modulation frequency @xmath5 and characteristics of the static potential @xmath6 ) to evaluate the escape rate enhancement in the presence of driving at least for a simple analytically smooth ( i.e. , not piecewise ) metastable or bistable potential ?
that is why it is desirable to have the period - average escape rate in a closed analytical form and explicitly expressed only via the parameters @xmath3 , @xmath4 , @xmath5 and characteristics of the static potential @xmath6 including no other physical values . in other words a theory convinient for
applications should restrict the physical content of the resulting formula only by the notions used at initial setting the problem . besides both @xcite , @xcite , @xcite and @xcite , @xcite , @xcite , @xcite , @xcite , @xcite theories invoke to rather sophisticated methods such as , e.g. , path integrals technique .
the initial mathematical formulation of the problem is a partial differential equation and it seems interesting to see what results can be obtained among others by means of usual mathematics .
these reasons motivate the appearance of the present manuscript .
our aim is to derive by means of elementary methods a closed analytical form of the formula for the escape rate enhancement in the low frequency limit for arbitrary values of the modulation ampitude to noise intensity ratio @xmath1 except a narrow region near the bifurcation point .
our approach is not based on a physical idea a priory inserted into the theory but rather is a direct purely mathematical treatment of the problem .
we define the low frequency limit as the one where there is no essential dependence on frequency so that the formal limit @xmath0 in the appropriate equations can be taken . at the same time we observe the requirement @xmath7 ( where @xmath8 is the stationary kramers rate ) that is necessary for the efficient averaging over the period to be possible .
the latter means that all interesting phenomena related to the so called stochastic resonance ( taking place at @xmath9 ) are beyond the scope of the present theory .
as the value @xmath10 $ ] is usually vanishingly small there certainly should be a range ( perhaps @xmath11 ) where the contradictory requirements @xmath7 and @xmath0 can be reconciled . as this range turns out to be more restricted than that of slow modulation @xmath2 we use the term low frequency limit instead of adiabatic regime to avoid confusion .
the manuscript is organized as follows . in sec.2
the problem is formulated and its solution is argued to be sought by perturbation technique for the action in powers of the modulation amplitude .
sec.3 and sec.4 are devoted to the first and second order contributions into action respectively . in sec.5 the results are combined in a closed form for the escape rate enhancement in the low frequency limit . in sec.6
the formula is used to obtain plots . in sec . 7 a simple analytical formula for the limiting case of moderately strong modulation @xmath12
is obtained . in sec.8
the results are discussed and the conclusions are summarized . in the appendix
some technical details are presented .
in this preliminary sec . we pose the problem and remind some facts on the kramers theory to introduce designations and notions used further . in the kramers model a chemical reaction is considered as the escape of a brownian particle from the well of a potential @xmath6 along the reaction coordinate @xmath13 with @xmath14 being the point of the bottom of the well and @xmath15 being the point of its top .
the problem of interest is to take into account the presence of periodic driving with modulation amplitude @xmath4 and frequency @xmath5 . for the driving we adopt without any serious loss of generality the commonly used form @xmath16 the results obtained for this simplest case can be directly generalized to any arbitrary periodic driving because the latter can be expanded into a fourier series . in the overdamped limit ( strong friction case ) the fokker - planck equation ( fpe ) for the probability distribution function @xmath17 is @xmath18p'(x , t)+dp^{\prime\prime}(x , t)\ ] ] where the dot denotes a derivative in time
, the prime denotes a derivative in coordinate , @xmath19 is the time independent force field and @xmath3 is the so called noise intensity that is actually the ratio of temperature in energetic units to the barrier height . in the absence of periodic
driving ( @xmath20 ) the stationary limit of ( [ eq2 ] ) is @xmath21'\ ] ] which can be integrated to yield for the stationary ( kramers ) probability distribution function @xmath22 the equation @xmath23 where @xmath24 is the stationary flux .
if we adopt the boundary condition as the absorbtion @xmath25 at some point @xmath26 ( with @xmath27 where @xmath15 is the barrier top ) then we have @xmath28 for the kramers rate ( taking into account that @xmath29 ) we have @xmath30\ ] ] where @xmath31 and @xmath32 . in the presence of driving the position of the barrier top
becomes time dependent ( @xmath33 ) and the population of the well is @xmath34 a convenient operational definition of the reaction rate constant ( adopted , e.g. , in @xcite , @xcite and @xcite ) is @xmath35 it should be stressed that the behavior of this value is sensitive to the actual choice of the absorbtion boundary @xmath36 where we set @xmath37 .
the requirement we adopt further is that the absorbtion point should be sufficiently far from the barrier top @xmath38 this requirement is in accordance with that to measure the current well behind the boundary argued in @xcite . let us consider , e.g. , the the simplest cubic ( metastable ) potential @xmath39 ( cp ) .
in this case @xmath40 , @xmath41 , @xmath42 , @xmath43 and @xmath44 , @xmath45 . by analogy
we adopt for the general case @xmath46 for the quartic ( bistable ) potential @xmath47 ( qp ) we also have @xmath42 while @xmath48 , @xmath49 and @xmath50 . thus taking into account that @xmath51 and ( [ eq2 ] ) we obtain from ( [ eq8 ] ) @xmath52p(q_b(t),t)\ ] ] the rate constant averaged over the period of oscillations @xmath53 is @xmath54
our final aim is to calculate the escape rate enhancement @xmath55p(q_b(s),s)-\ ] ] @xmath56 to attain this goal we will also need the probability distribution function near the bottom of the well that is known to be @xcite , @xcite @xmath57 ^ 2}{2d\sigma_a^2(t)}\right\}\ ] ] here @xmath58 is the dispersion that can be identified with the inverse frequency of the well @xmath59 .
the latter can be evaluated , e.g. , for the cp @xmath60 . by analogy
we adopt for the general case @xmath61 substituting ( [ eq11 ] ) and ( [ eq15 ] ) into ( [ eq14 ] ) we obtain @xmath62 ^ 2}{2d}\right\}\ ] ] we seek the solution of ( [ eq2 ] ) in the form @xmath63 the form ( [ eq17 ] ) means that @xmath64 because @xmath25 , i.e. , adopting this form we have to neglect the possible dependence of the absorbtion point @xmath36 on time in the presence of driving and assume @xmath65 where @xmath26 is that in the absence of driving .
the latter may be justified by the requirement ( [ eq9 ] ) . in the present approach
we adopt this approximation without further discussing its validity .
substitution of ( [ eq17 ] ) into ( [ eq13 ] ) with taking into account ( [ eq5 ] ) yields for the value of interest @xmath66 @xmath67u(q_b(s),s)-du'(q_b(s),s)\bigr]\ ] ] we denote @xmath68 for the function @xmath69 we obtain the equation @xmath70 @xmath71u'(x , t)-f(t)\psi ( x ) u(x , t)\ ] ] at @xmath72 we must have @xmath73 for asymptotically large time that suggests to seek the solution of ( [ eq21 ] ) in the form @xmath74\ ] ] this form can not be exact because in the limit @xmath75 we obtain @xmath76 that can be valid only asymptotically at @xmath77 .
we denote @xmath78 for the function @xmath79 we obtain the equation @xmath80 ^ 2+\ ] ] @xmath81\alpha'(x , t)- \psi ( x ) sin(\omega t)\ ] ] in the present manuscript we argue the point of view that in the low frequency limit the function @xmath79 can be sought as a perturbation series in the modulation amplitude @xmath4 @xmath82 the function @xmath79 by its final result in ( [ eq13 ] ) plays the same role as the action from the theories @xcite , @xcite , @xcite , @xcite and @xcite , @xcite , @xcite .
that is why we will also use this name .
the results of the present manuscript testify that at least in the low frequency limit @xmath83 so that @xmath84 .
thus taking into account that @xmath85 and @xmath86 ( see below ) we generally have for the escape rate enhancement @xmath87\ ] ] though at large ratios @xmath1 the term @xmath88 we will not take it into account in the present paper despite of the fact that the method of calculation developed below enables us to treat it .
this term seems to give minor correction and its role will be considered elsewhere . substituting ( [ eq25 ] ) in ( [ eq22 ] ) , ( [ eq21 ] ) and collecting the terms at powers of @xmath4 we obtain the following system of equations @xmath89
@xmath90 ^ 2- sin(\omega t)\ \varphi'(x , t)\ ] ] @xmath91\chi'(x , t)\ ] ] etc . from ( [ eq2 ] ) , ( [ eq18 ] ) and ( [ eq21 ] ) we obtain the boundary conditions @xmath92 from ( [ eq16 ] ) we obtain @xmath93
we start from ( [ eq27 ] ) .
its solution for asymptotically large time can be sought in the form @xmath94 from ( [ eq30 ] ) and ( [ eq31 ] ) we obtain the boundary conditions @xmath95 @xmath96 substitution of ( [ eq32 ] ) into ( [ eq27 ] ) yields a system of coupled equations @xmath97 @xmath98
this system is rather difficult for analytical treatment .
however in the present manuscript we restrict ourselves by the low frequency limit @xmath0 . in this case the equations ( [ eq35 ] ) , ( [ eq36 ] ) ) are decoupled @xmath99 @xmath100 taking into account that @xmath101= \frac{p^2_k(s)}{p^2_k(z)}exp\biggl [ \frac{u(s)-u(z)}{d}\biggr]\ ] ] we obtain the solution of ( [ eq37 ] ) , ( [ eq38 ] ) satisfying ( [ eq33 ] ) , ( [ eq34 ] ) as @xmath102 @xmath103 we substitute @xmath22 from ( [ eq5 ] ) into ( [ eq40 ] ) and notice that @xmath104 ^ 2}= d \frac{1}{\int\limits_{y}^{x_c}ds\ exp\left(\frac{u(s)}{d}\right)}\ ] ] making use of integration by part and denoting @xmath105 @xmath106 ^ 2\ ] ] @xmath107 we obtain @xmath108 making use of ( [ eq20 ] ) and ( [ eq5 ] ) we obtain @xmath109 @xmath110 we recall ( [ eq11 ] ) and denote @xmath111 @xmath112 @xmath113 @xmath114(s - x)^n\ ] ] making use of n2.3.15.1 from @xcite we obtain @xmath115 @xmath116 d_{-(n+1)}\left[-\sqrt{2\lambda}\ a sin ( \omega t ) \right]\ ] ] where @xmath117 is a parabolic cylinder function . making use of its known properties we have @xmath118\ ] ]
@xmath119 @xmath120+\ ] ] @xmath121 where @xmath122 is an additional error function .
making use of ( [ eq53 ] ) and ( [ eq55 ] ) we obtain ( see appendix for details ) @xmath123 and @xmath124 the value @xmath125 is negligibly small compared with @xmath126 and can be omitted . thus finally we obtain the first order contribution into action as @xmath127 sin ( \omega t)\ ] ]
from ( [ eq28 ] ) and the results for the first order contribution into action we obtain @xmath128\ ] ] where we denote @xmath129}{2}\ ] ] we seek the solution of ( [ eq59 ] ) in the form @xmath130 where the new functions obey the equations @xmath131 @xmath132 from ( [ eq30 ] ) and ( [ eq31 ] ) we obtain the boundary conditions
@xmath133 @xmath134 the solution for the function @xmath135 is @xmath136 we denote @xmath137 after straightforward calculations we obtain @xmath138 the inner integrals for both @xmath139 and @xmath140 mave maximum at @xmath141 .
that is why we adopt the following approximations @xmath142 @xmath143 where @xmath144 @xmath145 denoting @xmath146 we obtain @xmath147 we seek the solution of ( [ eq63 ] ) in the form @xmath149 for the new functions we obtain the equations @xmath150 @xmath151 recalling that we are in the low frequency limit we decouple the equations ( the limit @xmath0 in ( [ eq76 ] ) , ( [ eq77 ] ) ) @xmath152 @xmath153 from ( [ eq64 ] ) , ( [ eq65 ] ) we obtain the boundary conditions for these equations @xmath154 @xmath155 the solutions satisfying these boundary conditions are @xmath156 @xmath157 thus we obtain @xmath158 combining the results we obtain the second order contribution into action @xmath160 to calculate @xmath161 from ( [ eq73 ] ) we need @xmath162 and @xmath163 . taking into account ( [ eq60 ] ) the latter means that we need @xmath164 and @xmath165 . from ( [ eq40 ] )
we have @xmath166 from ( [ eq47 ] ) and ( [ eq55 ] ) we obtain @xmath167 where @xmath168 as a result we have @xmath169 thus @xmath170 from ( [ eq54 ] ) and ( [ eq55 ] ) we obtain @xmath171 @xmath172\biggr \rbrace\ ] ] substitution of ( [ eq91 ] ) into ( [ eq60 ] ) yields @xmath163 that is not written out explicitly to save room .
let us evaluate the integral @xmath173 given by ( [ eq72 ] ) .
the maximum of the inner integrals takes place at @xmath141 . in this case
both inner integrals @xmath174 , otherwise only one of them is such and the whole integral should be smaller .
thus we approximate at @xmath141 @xmath175 @xmath176 as a result we obtain @xmath177 taking into account that @xmath178 we obtain @xmath179 now we start rather tedious evaluation of the function @xmath180 given by ( [ eq71 ] ) .
let us first calculate the auxiliary value @xmath181 @xmath182 the inner integrals have maximum at @xmath141 . in this case we expand the last integral into a taylor series @xmath183 then @xmath184 making use of ( [ eq55 ] ) we obtain @xmath185 to calculate @xmath180 we decompose it as follows @xmath186 where we denote @xmath187\int\limits_{x_b}^{x_c}dz\ exp\left(-\frac{u(z ) } { d}\right)\int\limits_{z}^{x_c}ds\ exp\left(\frac{u(s ) } { d}\right)\ ] ] @xmath188 @xmath189\int\limits_{q_b(t)}^{x_b}dz\ exp\left(-\frac{u(z ) } { d}\right)\int\limits_{z}^{x_c}ds\ exp\left(\frac{u(s ) } { d}\right)\ ] ] we begin with the itegral in the square brackets from @xmath190 and @xmath191 .
it can be evaluated as follows @xmath192 @xmath193 then @xmath190 can be approximated as @xmath194 altering the order of integration in the double integral we obtain @xmath195 @xmath196 making use of ( [ eq55 ] ) we obtain @xmath197 at evaluation of the inner integrals in @xmath198 we apply the same arguments as those used for deriving ( [ eq98 ] ) .
then we have @xmath199 after evaluation of the integral we obtain @xmath200\ ] ] for @xmath191 after taking into account ( [ eq106 ] ) we have the approximation @xmath201 for the double integral we have @xmath202 making use of n1.5.1.9 from @xcite we obtain @xmath203 @xmath204 \biggr\rbrace\ ] ] combining the results we finally obtain @xmath205 @xmath206
combining the results and taking into account that @xmath207 we finally obtain the escape rate enhancement in the low frequency limit @xmath208 sin ( \phi)-\ ] ] @xmath209 @xmath210 exp\left(\frac{u\left(x_b - a sin(\phi)\right)-u(x_b)}{d}\right)\times\ ] ] @xmath211 @xmath212\biggr \rbrace\times\ ] ] @xmath213 @xmath214/erfc^2\left(-\frac{\omega_b a sin(\phi ) } { \sqrt{2d}}\right)\biggr\rbrace\biggr\rbrace\ ] ] this formula is the first result of the present manuscript .
the formula ( [ eq115 ] ) notwithstanding it looks very cumbersom can be easily treated by , e.g. , _
mathematica_. prior doing it we recall that we want to stay within the so called subthreshold driving regime .
the latter provides that the potential surface always has a minimum and a maximum , i.e. , the oscillating field is small enough not to distort the physical picture of the chemical reaction as the brownian particle escape from the metastable state .
this regime is defined by the requirement @xmath215 where @xmath216 for the case of the cubic ( metastable ) potential @xmath39 ( cp ) and @xmath217 ( see , e.g. , @xcite ) for the case of the quartic ( bistable ) potential @xmath47 ( qp ) . in fig.1 and fig.2 the results for the cp and qp respectively at relatively large values of the noise intensity are depicted . at the value @xmath218
we have @xmath219 and @xmath220 for the case of the cp and qp respectively . in fig.3 and fig.4 the results for the cp and qp respectively at relatively small values of the noise intensity aredepicted . at the value @xmath221 we have @xmath222 and @xmath223 for the cases of the cp and qp respectively .
these values limit the horizontal coordinate in the plots .
let us consider the limiting case of moderately strong modulation @xmath225 .
we denote @xmath226\ ] ] taking into account the requirement @xmath227 and discarding in ( [ eq115 ] ) the terms @xmath228 we obtain @xmath229 \right \}\ ] ] taking into account @xmath230 we can evaluate the integral by the steepest descent method and obtain a simple formula @xmath231 } } exp\biggl \ { \frac{a}{d}\bigl[x_b - x_a - a p\bigr]\biggr \}\ ] ] where @xmath232 is the constant given by ( [ eq116 ] ) . for the case of cp
we have @xmath233 while for the qp we have @xmath234 . for both of them we have @xmath235 . in the considered range @xmath225 the expression under the square root in the denominator can not be zero at physically reasonable values of the noise intensity @xmath236 . the formulas ( [ eq116 ] ) and ( [ eq118 ] ) are the second result of the present manuscript .
the results obtained testify that the perturbation expansion for the action @xmath237 in the modulation amplitude @xmath4 yields a reasonable and convergent expression at least in the low frequency limit . in this case
we restrict ourselves by the second order term @xmath238 . in our opinion the corrections from the third and higher order contributions ( which are @xmath239 ) will not distort the results appreciably except in a narrow region near the bifurcation point @xmath240
. in this region our results diverge with the scaling laws obtained in @xcite , @xcite , @xcite , @xcite . for the regime of weak modulation
our results precisely coincide with those known from the literature .
the formula ( [ eq118 ] ) in the low limit of its validity range @xmath225 ( @xmath4 small enough for the terms @xmath241 to be neglected ) with taking into account that @xmath242 for both the cp and qp yields the formula ( 6.34 ) from @xcite obtained for the case of the qp in the limit of small driving frequencies . the value @xmath243 is a coefficient before the linear in @xmath1 term in the exponent of the escape rate enhancement and is actually the so called logarithmic susceptibility from the @xcite , @xcite , @xcite , @xcite , @xcite , @xcite theory .
the value @xmath244 is in agreement with @xmath245=1 $ ] obtained in @xcite for the case of the cp . from fig.1 , fig.2 , fig.3 and fig.4
we see that the present theory yields analytical description for the retardation of the exponential growth of the escape rate enhancement ( i.e. , transition from a log - linear regime to more moderate growth ) .
moreover fig.2 exhibits the examples of the reverse behavior when the escape rate enhancement attains a maximum at some value of @xmath1 and then becomes to decrease with the further increase of @xmath1 .
it is worthy to note that this phenomenon vividly manifests itself for the qp and is not noticed for the cp .
regretfully the @xcite , @xcite , @xcite and @xcite , @xcite , @xcite , @xcite , @xcite , @xcite theories were not exemplified by the case of the qp and our prediction can not be directly compared with the results of those theories . from the comparison of fig.1 , fig.2 and fig.3 , fig.4
respectively we see that at a given @xmath1 the periodic driving produces stronger escape rate enhancement for the case of cp than that of qp .
the latter certainly can not be explained by the fact that the barrier height for the cp ( @xmath246 ) is smaller than that for the qp ( @xmath247 ) because the barrier height does not enter the formulas ( [ eq115 ] ) and ( [ eq118 ] ) .
this phenomenon can be attributed to the only difference between these potentials manifested in the value of the frequency near the bottom of the well @xmath248 . for
the qp this value ( @xmath249 ) is higher than that for the cp ( @xmath244 ) .
the latter means that the qp goes steeper from the bottom of the well that hinders the escape rate enhancement by periodic driving .
thus the precise shape of the potential is of utmost importance for the phenomenon of interest .
we obtain two forms of the resulting formula valid for arbitrary potentials with an activation barrier .
the formula ( [ eq115 ] ) encompasses the case of arbitrary @xmath1 except a narrow region near the bifurcation point .
its drawback is that it is very cumbersome
. nevertheless it can be easily tackled by a computer with the help of , e.g. , _
mathematica_. its main merit is that it contains only the notions used at initial setting the problem and can be used by people engaged in applications of the theory without reading the rest parts of the present manuscript .
the formula ( [ eq118 ] ) is valid for the case of moderately strong modulation @xmath225 .
its merit is that it has closed and quite simple analytical form to be used by practitioners in chemistry and biochemistry for quick by hand estimates .
the exponent in this formula explicitly and vividly demonstrates how the linear term @xmath238 dominating at weak modulation ( comparatively small @xmath1 ) and responsible for the log - linear regime is replaced by that substracted by the term @xmath239 at further increase of @xmath1 providing the reardation of the exponential growth of the escape rate enhancement .
regretfully we can not directly compare our results with those of the @xcite , @xcite , @xcite theory because the latter is inapplicable for the case of slow modulation .
the formula ( [ eq118 ] ) is valid in a rather narrow range of the parameters .
however this range turns out to be relevant for applications in enzymology .
the typical value of noise intensity at enzymatic reactions is @xmath250 @xcite . in this case
we have the range of validity for ( [ eq118 ] ) as @xmath251 .
the typical values of @xmath1 for a particular model suggested in @xcite were estimated as @xmath252 and find themselves at some stretch within this range .
the stringent results for this case are given by the formula ( [ eq115 ] ) and can be seen in fig.3 and fig.4 .
we conclude that for the particular case of the low frequency limit the aim to obtain tractable and convenient formulas describing the escape rate enhancement by periodic driving seems to be attained .
the mathematical tools used at their deriving are within the scope of elementary methods .
we denote @xmath253 then @xmath254 our aim here to show that the term @xmath255 is @xmath256 then ( [ eq56 ] ) will be proven .
let us write @xmath257/ \int\limits_{x_a}^{x_c}ds\ exp\left[-\lambda ( s - x_b)^2\right ] \biggr \}\ ] ] making use of the substitution @xmath258 we obtain @xmath259 as a result of straightforward calculations we have @xmath260 taking into account ( [ eq53 ] ) we obtain the required result @xmath261 00 p. hnggi , p. talkner , m. borkovec , fifty years after kramers equation : reaction rate theory , rev.mod.phys .
62 ( 1990 ) 251 - 341 .
p. jung , periodically driven stochastic systems , phys.reports 234 ( 1993 ) 175 - 295 .
fleming and p. hnggi ( eds . ) , activated barrier crossing : applications in physics , chemistry and biology , world scientific , singapore 1993 .
p. talkner and p. hnggi ( eds . ) , new trends in kramers reaction rate theory , kluwer , dordrecht 1995 .
l. gammaitoni , p. hnggi , p. jung , f. marchesoni , stochastic resonance , rev.mod.phys .
70 ( 1998 ) 223 - 287 .
u. weiss , quantum dissipative systems , series in mod- ern condensed matter physics , vol . 2 ( world scientific , singapore , 1998 ) .
j. ankerhold , p. pechukas , h. grabert , strong friction limit in quantum mechanics : the quantum smoluchowski equation , phys .
87 ( 2001 ) 086802 .
j. ankerhold , quantum decay rates for driven barrier potentials in the strong friction limit , phys .
e 64 ( 2001 ) 060102 ( r ) .
j. ankerhold , h. grabert , p. pechukas , quantum brownian motion with large friction , chaos 15 ( 2005 ) 26106 .
p. reimann , p. hnggi in : stochastic dynamics , lecture notes in physics , vol .
484 , l. schimansky - geier and th .
pschel ( eds . ) , springer , berlin 1997 : pp .
127 - 139 .
j. lehmann , p. reimann , p. hnggi , surmounting oscillating barriers , phys .
84 ( 2000 ) 84 - 87 .
j. lehmann , p. reimann , p. hnggi , surmounting oscillating barriers : path - integral approach for weak noise , phys.rev .
e62 ( 2000 ) 6282 - 6294
. j. lehmann , p. reimann , p. hnggi , activated escape over oscillating barriers : the case of many dimensions , phys.stat.sol.(b ) 237 ( 2003 ) 53 - 64 .
smelyanskiy , m.i .
dykman , b. golding , time oscillations of escape rates in periodically driven systems , phys.rev.lett .
82 ( 1999 ) 3193 - 3197 .
smelyanskiy , m.i .
dykman , h. rabitz et al .
, nucliation in periodically driven electrochemical systems , j.chem.phys . 110
( 1999 ) 11488 - 11504 .
r. s. maier , d.l .
stein , noise - activated escape from a sloshing potential well , phys.rev.lett .
86 ( 2001 ) 3942 - 3946 .
dykman , b. golding , l.i .
mccann et al . , activated escape of periodically driven systems , chaos 11 ( 2001 ) 587 .
p. talkner , stochastic resonance in the semiadiabatic limit , new j.phys . 1 ( 1999 ) 4.1 - 4.25 .
p. talkner , j.luczka , rate description of fokker - planck processes with time dependent parameters , phys.rev.e 69 ( 2004 ) 046109 .
d. ryvkine , m.i .
dykman , b. golding , scaling and crossovers in activated escape near a bifurcation point , phys .
e 69 ( 2004 ) 061102 .
dykman , b. golding , d. ryvkine , critical exponent erossovers in escape near a bifurcation point , phys.rev.lett .
92 ( 2004 ) 080602 .
dykman , d. ryvkine , activated escape of periodically modulated systems , phys.rev.lett .
94 ( 2005 ) 070602 .
d. ryvkine , m.i .
dykman , noise - induced escape of periodically modulated systems : from weak to strong modulation , phys rev .
e 72 ( 2005 ) 011110 .
r. de vivie - riedle , h. rabitz , k .-
kompa , special issue : laser control of quantum dynamics , chem .
267 , nos . 1 - 3 ( 2001 ) .
m. shapiro , adv .
114 , 123 ( 2000 ) p. k. ghosh , d. barik , b. c. bag , d. s. ray , j. chem .
123 , 224104 ( 2005 ) .
a.e . sitnitsky , dynamical contribution into enzyme catalytic efficiency , physica a371 ( 2006 ) 481 - 491 ; los alamos preprint datebase , arxiv : cond - mat/0601165 .
sitnitsky , discrete breathers in protein secondary structure , in : soft condensed matter .
new research .
dillon , ch.5 , pp.157 - 172 , nova , 2007 ; los alamos preprint datebase , arxiv : cond - mat/0306135 .
d.s . ray , notes on brownian motion and related phenomena ; los alamos preprint datebase , arxiv : physics/9903033 .
d. ludwig , siam rev .
17 ( 1975 ) 605 .
prudnikov , a.p . ; brychkov , yu.a . ; marichev , o.i . integrals and series .
elementary functions ; nauka : moscow , 1981 .
prudnikov , a.p . ; brychkov , yu.a . ; marichev , o.i . integrals and
special functions ; nauka : moscow , 1983 . | the period - average rate in the low frequency limit for thermally activated escape with periodic driving is derived in a closed analytical form .
we define the low frequency limit as the one where there is no essential dependence on frequency so that the formal limit @xmath0 in the appropriate equations can be taken .
we develop a perturbation theory of the action in the modulation amplitude and obtain a cumbersom but closed and tractable formula for arbitrary values of the modulation ampitude to noise intensity ratio @xmath1 except a narrow region near the bifurcation point and a simple analytical formula for the limiting case of moderately strong modulation . the present theory yields analytical description for the retardation of the exponential growth of the escape rate enhancement ( i.e. , transition from a log - linear regime to more moderate growth and even reverse behavior ) .
the theory is developed for an arbitrary potential with an activation barrier but is exemplified by the cases of cubic ( metastable ) and quartic ( bistable ) potentials .
keywords : kramers theory , thermally activated escape , periodic driving .
+ , |
let @xmath4 be a surface .
two closed curves @xmath5 are , written @xmath6 , if there exists a continuous map @xmath7\times\r/\z$ ] such that @xmath8 and @xmath9 for all @xmath10 . assuming the curves in generic position , their number of intersections is @xmath11 their only depends on their free homotopy classes and is defined as @xmath12 likewise
, the number of self - intersections of @xmath13 is given by @xmath14 and its minimum over all the curves freely homotopic to @xmath13 is its @xmath15 .
note the one half factor that comes from the identification of @xmath16 with @xmath17 .
the geometric intersection number is an important parameter that allows to stratify the set of homotopy classes of curves on a surface .
the surface is usually endowed with a hyperbolic metric , implying that each homotopy class is identified by its unique geodesic representative . extending a former result by mirzakhani @xcite
, sapir @xcite has recently provided tight asymptotics for the number of closed geodesics with bounded length and bounded geometric intersection number . chas and
lalley @xcite also proved that the distribution of the geometric intersection number with respect to the length of geodesics approaches the gaussian distribution as the length grows to infinity .
other more experimental results were obtained with the help of a computer to show the existence of length - equivalent homotopy classes with distinct geometric intersection numbers @xcite .
hence , for both theoretical and practical reasons various aspects of the computation of geometric intersection numbers have been studied in the past including the algorithmic ones .
nonetheless , all the previous approaches rely on rather complex mathematical arguments and to our knowledge no exact complexity analysis has yet appeared . in this paper , we make our own the words of dehn who noted that the metric on words ( on some basis of the fundamental group of the surface ) can advantageously replace the hyperbolic metric @xcite .
we propose a combinatorial framework that leads to simple algorithms of low complexity to compute the geometric intersection number of curves or to test if this number is zero .
our approach is based on the computation of canonical forms as recently introduced in the purpose of testing whether two curves are homotopic @xcite .
canonical forms are instances of combinatorial geodesics who share nice properties with the geodesics of a hyperbolic surface . on such surfaces
each homotopy class contains a unique geodesic that moreover minimizes the number of self - intersections .
although a combinatorial geodesic is generally not unique in its homotopy class , it must stay at distance one from its canonical representative and a careful analysis of its structure leads to the first result of the paper .
[ th : main - result ] given two curves represented by closed walks of length at most @xmath0 on a combinatorial surface of complexity @xmath1 we can compute the geometric intersection number of each curve or of the two curves in @xmath2 time . as usual the complexity of a combinatorial surface stands for its total number of vertices , edges and faces .
a key point in our algorithm is the ability to compute the primitive root of a canonical curve in linear time .
we next provide an algorithm to compute an actual curve immersion that minimizes the number of self - interactions in its homotopy class .
[ th : compute - immersion ] let @xmath18 be a closed walk of length @xmath0 in canonical form .
we can compute a combinatorial immersion with @xmath19 crossings in @xmath20 time .
we also propose a nearly optimal algorithm that answers an old problem studied by poincar @xcite : decide if the geometric intersection number of a curve is null , that is if the curve is homotopic to a simple curve .
[ th : simple - curve ] given a curve represented by a closed walk of length @xmath0 on a combinatorial surface of complexity @xmath1 we can decide if the curve is homotopic to a simple curve in @xmath21 time . in the affirmative
we can construct an embedding of @xmath18 in the same amount of time . in the next section we review some of the previous relevant works .
we introduce our combinatorial framework in section [ sec : framework ] and discuss the structure of combinatorial geodesics in section [ sec : geodesics ] .
section [ sec : strategy ] presents our general simple strategy to compute the geometric intersection number .
the proof of theorem [ th : main - result ] is given in the next three sections where the case of non - primitive curves is also treated .
the computation of a minimally crossing immersion is presented in section [ sec : computing - immersions ] and [ sec : finding - bigons ] with the proof of theorem [ th : compute - immersion ] .
we finally propose a simple algorithm to detect and embed curves homotopic to simple curves ( theorem [ th : simple - curve ] ) in section [ sec : simple - curves ] .
in the fifth supplement to its _ analysis situs _ poincar @xcite describes a method to decide whether a given closed curve @xmath22 on a surface can be continuously deformed to a simple curve . for this
, he considers the surface as the quotient @xmath23 of the poincar disk @xmath24 by a ( fuchsian ) group @xmath25 of hyperbolic transformations .
the endpoints of a lift of @xmath22 in the poincar disk are related by a hyperbolic transformation whose axis is a hyperbolic line @xmath26 representing the unique geodesic homotopic to @xmath22 .
he concludes that @xmath22 is homotopic to a simple curve if and only if all the transforms of @xmath26 by @xmath25 are pairwise disjoint or equal .
this method was turned into an algorithm by reinhart @xcite who worked out the explicit computations in the poincar disk using the usual representation of hyperbolic transformations by two - by - two matrices .
the entries of the matrices being algebraic the computation could indeed be performed accurately on a computer .
the ability to recognize curves that are _ primitive _ , i.e. whose homotopy class can not be expressed as a proper power of another class , happens to be crucial in this algorithm though computationally expensive .
birman ans series @xcite subsequently proposed an algorithm for the case of surfaces with nonempty boundary that avoids manipulating algebraic numbers .
while their arguments appeal to a hyperbolic structure , their algorithm is purely combinatorial .
intuitively , a surface with boundary deform retracts onto a fat graph ( in fact a fat bouquet of circles ) whose universal covering space embeds as a fat tree in the poincar disk .
the successive lifts of a curve @xmath22 trace a bi - infinite path in this tree .
the limit points of this path belongs to the circle @xmath27 at infinity and coincide with the ideal endpoints of the axis of the hyperbolic transformation corresponding to any of the lifts of the curve in the path .
the question as to whether two lifts give rise to intersecting axes can thus be reduced to test if the corresponding bi - infinite paths have separating limit points on @xmath27 . as for reinhart , birman and series
assume that the homotopy class of @xmath22 is given by a word @xmath28 on some given set of generators of the fundamental group of the surface . in turn , the above test on bi - infinite paths boils down to consider the cyclic permutations of @xmath28 and @xmath29 in a cyclic lexicographic order and to check if this ordering is well - parenthesized with respect to some pairing of the words .
cohen and lustig @xcite further observed that the approach of birman and series could be extended to count the geometric intersection number of one or two curves . in a second paper lustig @xcite
tackles the case where the curves are taken on a surface without boundary .
he considers a closed surface with negative euler characteristic as the quotient of the poincar disk by a group of transformations isomorphic to the fundamental group of the surface .
the main contribution of the paper is to define a canonical representative for every free homotopy class @xmath13 given as a word in some fixed system of loops generating the fundamental group .
lustig first notes that there is no known way of choosing a unique word for @xmath13 .
in particular , the shortest words are far from unique .
he rather represents ( a lift of ) @xmath13 by a path in the union @xmath30 of three tessellations of @xmath24 , where the edges of @xmath31 are all the lifts of the generating loops , @xmath32 is the dual tessellation , and the edges of @xmath33 joins the vertices of @xmath31 with the vertices of @xmath32 .
( although the graphs of @xmath31 , @xmath32 and @xmath33 are embedded their union is not as every edge of @xmath31 crosses its dual edge in @xmath32 . )
lustig gives a purely combinatorial characterization of canonical paths and argues that the method in his first paper @xcite can be applied to the canonical representative of @xmath13 .
overall the two papers adds up to 60 pages with essential arguments from hyperbolic geometry and the complexity of the whole procedure remains to be done .
other approaches were developed without assuming any hyperbolic structure .
based on the notion of winding number , chillingsworth @xcite provides an algorithm to test whether a curve is homotopic ( this time with fixed basepoint ) to a simple curve on a surface with nonempty boundary .
he also proposed an algorithm for determining when a given set of _ simple _ closed curves can be made disjoint by ( free ) homotopy @xcite .
while the winding number relies on a differentiable structure , zieschang @xcite used the connection between the automorphisms of a topological surface and the automorphisms of its fundamental group in order to detect the homotopy classes of simple closed curves .
a related work by hass and scott @xcite is concerned with curves which have excess intersection , i.e. that can be homotoped so as to reduce their number of intersections .
hass and scott introduce various types of monogons and bigons that can be either embedded , singular or weak .
a _ singular monogon _ of a curve @xmath34 is a contractible subpath of @xmath22 whose endpoints define a self - intersection of @xmath22 .
a _ bigon _ of @xmath22 is defined by two self - intersections joined by two homotopic subpaths ( with fixed endpoints ) @xmath35 and @xmath36 with @xmath37 .
the bigon is said _ singular _ if the defining segments @xmath38 are disjoint and _ weak _ otherwise .
hass and scott prove that a curve with excess self - intersection on an orientable surface must have a singular monogon or a singular bigon .
their result directly suggests an algorithm to compute the geometric intersection number of a curve : iteratively remove monogons or untie bigons until there is no more .
the final configuration must have the minimal number of self - intersections .
designing an efficient procedure to find monogons and bigons remains the crux of this approach . in any case , the method can not be extended to compute the geometric intersection number of two curves since hass and scott give two counter - examples to the fact that two curves with excess intersections should have a singular monogon or bigon .
one of their counter - examples contains a curve with excess self - intersections and the other one contains a non - primitive curve .
our counter - example , figure [ fig : no - singular - bigon ] , shows that even assuming each curve to be primitive and in minimal configuration , we may have excess intersections without singular bigons .
nonetheless , it was proved @xcite that starting from any configuration of curves one may reach a configuration with a minimal number of intersections by applying a finite sequence of elementary moves involving monogons , bigons and trigons similar to the reidemeister moves in knot theory .
a surprising consequence was obtained by neumann - coto @xcite .
define a _ cut and paste _ on a family of curves by cutting the curves at some of their intersection points and glueing the resulting arcs in a different order .
neumann - coto proves that any set of primitive curves can be brought to a homotopic set with minimal ( self-)intersections by a set of cut and paste operations .
note that each intersection of two curve pieces can be re - arranged in three different manners , including the original one , by a cut and paste .
hence , if the curves have @xmath39 ( self-)intersections we may find a minimal configuration out of the @xmath40 possible re - configurations !
we also mention the algebraic approach of gonalves et al .
@xcite based on previous works by turaev @xcite who introduced intersection forms over the integral group ring of the fundamental group of the surface .
see @xcite for more historical notes .
we introduce our combinatorial framework and show that it faithfully represents its continuous analogue . [ [ combinatorial - surfaces . ] ] combinatorial surfaces .
+ + + + + + + + + + + + + + + + + + + + + + + as usual in computational topology , we model a surface by a cellular embedding of a graph @xmath31 in a compact topological surface @xmath4 .
such a cellular embedding can be encoded by a composed of the graph @xmath31 itself together with a rotation system @xcite that records for every vertex of the graph the clockwise cyclic order of the incident arcs .
the are obtained from the rotation system by the face traversal procedure as described in @xcite . in order to handle surfaces with boundaries we allow every face of @xmath31 in @xmath4 to be either an open disk or an annulus ( open on one side ) . in other words
@xmath31 is a cellular embedding in the closure of @xmath4 obtained by attaching a disk to every boundary of @xmath4 .
we record this information by storing a boolean for every facial walk of @xmath31 indicating whether the associated face is perforated or not .
all the considered graphs may have loop and multiple edges .
a directed edge will be called an and each edge corresponds to two opposite arcs .
we denote by @xmath41 the arc opposite to an arc @xmath42 . for simplicity , .
every combinatorial surface @xmath43 can be reduced by first contracting the edges of a spanning tree and then deleting edges incident to distinct faces .
the resulting has a single vertex and a single face . the combinatorial surface @xmath43 and its reduced version encodes different cellular embeddings on a same topological surface . [ [ combinatorial - curves . ] ] combinatorial curves .
+ + + + + + + + + + + + + + + + + + + + + consider a combinatorial surface with its graph @xmath31 . a combinatorial curve ( or path )
@xmath18 is a walk in @xmath31 , i.e. an alternating sequence of vertices and arcs , starting and ending with a vertex , such that each vertex in the sequence is the target vertex of the previous arc and the source vertex of the next arc .
we generally omit the vertices in the sequence .
a combinatorial curve is closed when additionally the first and last vertex are equal .
when no confusion is possible we shall drop the adjective combinatorial .
the of @xmath18 is its total number of arc occurrences , which we denote by @xmath44 .
if @xmath18 is closed , we write @xmath45 , @xmath46 , for the vertex of index @xmath47 of @xmath18 and @xmath48 for the arc joining @xmath45 to @xmath49 . for convenience we set @xmath50=c[\ibar,\ibar+1]{\ensuremath{^{-1}}}$ ] to allow the traversal of @xmath18 in reverse direction .
in order to differentiate the arcs with their occurrences we denote by @xmath51_c$ ] the corresponding occurrence of the arc @xmath52 $ ] in @xmath53 , where @xmath54 is obtained by traversing @xmath55 in the opposite direction .
more generally , for any non - negative integer @xmath0 and any sign @xmath56 , the sequence of indices @xmath57 is called an of @xmath18 of length @xmath0
. the index path can be ( @xmath58 ) or ( @xmath59 ) and can be longer than @xmath18 so that an index may appear more than once in the sequence .
we denote this path by @xmath60 .
its is given by the arc sequence @xmath61,c[\overline{\imath+\varepsilon},\overline{\imath+2\varepsilon}],\dots , c[\overline{\varepsilon(\ell-1)},\overline{\varepsilon\ell}]).\ ] ] the image path of a length zero index path is just a vertex .
a of @xmath18 is a subsequence of the form @xmath62 where @xmath42 is an arc .
a closed curve is if it is homotopic to a trivial curve ( i.e. , a curve reduced to a single vertex ) .
we will implicitly assume that a homotopy has fixed endpoints when applied to paths and is free when applied to closed curves .
[ [ combinatorial - immersions . ] ] combinatorial immersions .
+ + + + + + + + + + + + + + + + + + + + + + + + + following the notion of a combinatorial set of loops as in @xcite , we define a of a set @xmath63 of combinatorial curves as the data for each arc @xmath42 in @xmath31 of a left - to - right order @xmath64 over all the occurrences of @xmath42 or @xmath41 in the curves of @xmath63 .
the only requirement is that opposite arcs should be associated with inverse orders .
let @xmath65 be the set of occurrences of @xmath42 or @xmath41 in the curves of @xmath63 where @xmath42 runs over all arcs of @xmath31 with origin @xmath66 .
a combinatorial immersion induces for each vertex @xmath66 of @xmath31 a circular order @xmath67 over @xmath65 ; if @xmath68 is the clockwise - ordered list of arcs of @xmath31 with origin @xmath66 , then @xmath67 is the cyclic concatenation of the orders @xmath69 .
an immersion of @xmath63 restricts to an immersion of each of its curves .
conversely , given an immersion of each curve in @xmath63 we can built an immersion ( in general many ) of their union by merging for each arc @xmath42 the left - to - right orders @xmath64 of each curve immersion .
[ [ combinatorial - crossings . ] ] combinatorial crossings .
+ + + + + + + + + + + + + + + + + + + + + + + + given an immersion @xmath70 of two combinatorial closed curves @xmath18 and @xmath71 we define a of @xmath72 as a pair of indices @xmath73 such that @xmath74 .
likewise , a double point of @xmath18 is a pair @xmath75 with @xmath76 and @xmath74 the double point @xmath77 is a in @xmath70 if the pairs of arc occurrences @xmath78_c,[\ibar,\ibar+1]_c)$ ] and @xmath79_d,[\jbar,\jbar+1]_d)$ ] are linked in the @xmath80-order , i.e. if they appear in the cyclic order @xmath81_c\cdots [ \jbar,\jbar-1]_d\cdots [ \ibar,\ibar+1]_c\cdots [ \jbar,\jbar+1]_d\cdots,\ ] ] with respect to @xmath80 or the opposite order .
an analogous definition holds for a self - crossing of a single curve , taking @xmath82 in the above definition .
note that the notion of ( self-)crossing is independent of the traversal directions of @xmath18 and @xmath71 .
the number of crossings of @xmath18 and @xmath71 and of self - crossings of @xmath18 in @xmath70 is denoted respectively by @xmath83 and @xmath84 .
we define the of @xmath18 , denoted by @xmath19 , as the minimum of @xmath85 over all the combinatorial immersions @xmath86 of any combinatorial curve @xmath87 freely homotopic to @xmath18 .
the of two combinatorial curves @xmath18 and @xmath71 is defined the same way taking into account crossings between @xmath18 and @xmath71 only .
[ lem : geometric - vs - combinatorial ] the combinatorial ( self-)crossing number coincides with the geometric ( self-)intersection number , i.e. for every combinatorial closed curves @xmath18 and @xmath71 : @xmath88 every combinatorial immersion @xmath86 of @xmath18 can be realized by a continuous curve @xmath89 with the same number of self - intersections as the number of self - crossings of @xmath86 . to construct @xmath22
we consider small disjoint disks centered at the images of the vertices of @xmath31 in @xmath4 .
we connect those vertex disks by disjoint strips corresponding to the edges of @xmath31 . for every arc @xmath42 of @xmath31 we draw inside the corresponding edge strip parallel curve pieces labelled by the occurrences of @xmath42 or @xmath41 in @xmath18 in the left - to - right order @xmath64 .
the endpoints of those curve pieces appear on the boundary of the vertex disks in the circular vertex orders induced by @xmath70 .
it remains to connect those endpoints by straight line segments inside each disk ( via a parametrization over a euclidean disk ) according to the labels of their incident curve pieces .
the resulting curve @xmath22 can be homotoped to @xmath90 and its self - intersections may only appear inside the vertex disks .
clearly , an intersection of two segments of @xmath22 in a disk corresponds to linked pairs of arc occurrences , i.e. to a combinatorial crossing , and vice - versa .
it follows that @xmath91 . to prove the reverse inequality we show that for any continuous curve @xmath22 in generic position there exists a combinatorial immersion of a curve @xmath87 such that @xmath92 and the number of self - crossings @xmath87 is at most the number of self - intersections of @xmath22 . by an isotopy we can enforce the self - intersections of @xmath22 to lie inside the vertex disks . after removing the vertex disks from @xmath4 ,
we are left with a set of disjoint and simple pieces of @xmath22 that can be isotoped inside the edge strips and vertex disks without introducing any new intersections . if a piece of curve in a strip has its endpoints on the same end of the strip we can further isotope the piece inside the incident vertex disk .
this way all the curve pieces join the two ends of their strip and can be ordered from left - to - right inside each strip ( assuming a preferred direction of the strip ) . replacing each curve piece by an arc occurrence we thus define a combinatorial immersion of a curve @xmath87 with the required properties .
the same constructions apply to an immersion of two curves showing that @xmath93 .
[ [ reduction - to - a - system - of - quads . ] ] reduction to a system of quads .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath43 be a combinatorial surface with negative euler characteristic .
we describe the construction of a system of quads for a surface without boundary .
a similar construction applies when @xmath43 has perforated faces . following lazarus and rivaud @xcite
we start putting @xmath43 into a standard form called a by erickson and whittlesey @xcite . after reducing @xmath43 to a surface @xmath94 with a single vertex @xmath66 and a single face @xmath95 this system of quads
is obtained by adding a vertex @xmath96 at the center of @xmath95 , adding edges between @xmath96 and all occurrences of @xmath66 in the facial walk of @xmath95 , and finally deleting the edges of @xmath94 .
the graph of the resulting system of quads , called the @xcite , is bipartite .
it contains two vertices , namely @xmath66 and @xmath96 , and @xmath97 edges , where @xmath98 is the genus of @xmath43 .
all its faces are quadrilaterals .
note that the system of quads can be deduced from @xmath43 by a sequence of edge contractions , deletions and insertions .
every cycle @xmath18 of @xmath43 can thus be modified accordingly to give a cycle @xmath99 in the system of quads .
clearly , there exist cellular embeddings @xmath100 of the graphs of @xmath43 and of the system of quads in a same topological surface such that @xmath101 .
[ lem : systems - of - quads ] let @xmath1 be the number of edges of the combinatorial surface @xmath43 .
the above construction of a system of quads can be performed in @xmath102 time so that for every closed curve @xmath18 of length @xmath0 in @xmath43 , we can compute in @xmath103 time a curve @xmath87 of length at most @xmath104 in the system of quads such that @xmath105 . for the rest of the paper we shall assume that .
[ [ diagrams . ] ] diagrams .
+ + + + + + + + + a over the combinatorial surface @xmath43 is a combinatorial sphere @xmath106 with one perforated face together with a labelling of the arcs of @xmath106 by the arcs of @xmath43 such that 1 .
opposite arcs receive opposite labels , 2 .
the facial walk of each non - perforated face of @xmath106 is labelled by the facial walk of some non - perforated face of @xmath43 .
the diagram is when no edge of @xmath106 is incident to two non - perforated faces labelled by the same facial walk ( with opposite orientations ) of @xmath43 .
an is defined similarly by a combinatorial sphere with two distinct perforated faces .
a vertex of a diagram that is not incident to any perforated face is said .
[ lem : diagram ] a cycle of @xmath43 is contractible if and only if it is the label of the facial walk of the perforated face of a reduced disk diagram over @xmath43 .
two cycles are freely homotopic if and only if the facial walks of the two perforated faces of a reduced annular diagram over @xmath43 are labelled by these two cycles respectively .
note that two non - perforated faces that are adjacent and consistently oriented in a reduced diagram are labelled by adjacent faces that are consistently oriented in @xmath43 .
moreover , the degree of an interior vertex of the diagram is a multiple of the degree of the corresponding vertex in @xmath43 . in the sequel , . [ [ spurs - brackets - and - canonical - curves . ] ] spurs , brackets and canonical curves .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + thanks to lemma [ lem : systems - of - quads ] we may assume that our combinatorial surface @xmath43 is a system of quads . moreover , the construction of this system of quads with the assumption on the euler characteristic implies that . following the terminology of erickson and whittlesey @xcite
, we define the of a pair of arcs @xmath107 sharing their origin vertex @xmath66 as the number of face corners between @xmath108 and @xmath109 in clockwise order around @xmath66 . hence
, if @xmath66 is a vertex of degree @xmath71 in @xmath43 , the turn of @xmath107 is an integer modulo @xmath71 that is zero when @xmath110 .
the of a subpath @xmath111 of a closed curve of length @xmath0 is the sequence of @xmath112 turns of @xmath113 for @xmath114 , where indices are taken modulo @xmath0 .
the subpath may have length @xmath0 , thus leading to a sequence of @xmath115 turns .
note that the turn of @xmath113 is zero precisely when @xmath116 is a spur .
a is any subpath whose turn sequence has the form @xmath117 or @xmath118 where @xmath119 stands for a possibly empty sequence of turns @xmath120 and @xmath121 stands for @xmath122 .
it follows from lemma [ lem : diagram ] and a simple combinatorial gauss - bonnet theorem @xcite that [ th : four - vertex ] a nontrivial contractible closed curve on a system of quads must have either a spur or four brackets .
moreover , if the curve is the label of the boundary of a disk diagram with at least one interior vertex , then the curve must have either a spur or five brackets .
lazarus and rivaud @xcite have introduced a canonical form for every free homotopy class of closed curves in a system of quads . in particular , two curves
are freely homotopic if and only if their canonical forms are equal ( up to a circular shift of their vertex indices ) .
it was further characterized by erickson and whittlesey @xcite in terms of turns and brackets .
[ th : canonical ] the canonical form of a combinatorial closed curve of length @xmath0 on a system of quads can be computed in @xmath103 time .
it is the unique homotopic curve that contains no spurs or brackets and whose turning sequence contains no @xmath123 s and contains at least one turn that is not @xmath124 .
the canonical form is an instance of a , i.e. a curve that contains no spurs or brackets .
the canonical form is the rightmost homotopic geodesic .
the definitions of a geodesic and of a canonical form extend trivially to paths .
in particular , the canonical form of a path is its unique rightmost homotopic geodesic and is characterized as for closed curves .
[ th : canonical - path ] the canonical form of a combinatorial path is the unique homotopic path that contains no spurs or brackets and whose turning sequence contains no @xmath123 s .
although we can not claim in general the uniqueness of geodesics in a homotopy class , homotopic geodesics are almost equal and have the same length . specifically , define a as a planar sequence of quads obtained by stitching an alternating sequence of rows and columns of quads to form a staircase .
assuming that the staircase goes up from left to right , we define the of a quad staircase as the lower left vertex of the first quad in the sequence .
the is defined as the upper right vertex of the last quad .
a is obtained by identifying the two vertical arcs incident to the initial and final tips of a staircase .
[ th : geodesic ] let @xmath125 be two non - trivial homotopic combinatorial geodesics .
if @xmath125 are closed curves , then they label the two boundary cycles of an annular diagram composed of a unique closed staircase or of an alternating sequence of paths ( possibly reduced to a vertex ) and quad staircases connected through their tips . likewise , if @xmath125 are paths , then the closed curve @xmath126 labels the boundary of a disk diagram composed of an alternating sequence of paths ( possibly reduced to a vertex ) and quad staircases connected through their tips .
we only detail the proof when @xmath125 are paths .
see figure [ fig : lentille - plate ] .
the similar case of closed curves is covered in @xcite . by lemma [ lem :
diagram ] , @xmath126 is the label of the facial walk of the perforated face of a disk diagram @xmath106 .
this diagram has a cactus - like structure composed of 2-cells subdivided into quads and connected by trees .
a vertex @xmath66 such that @xmath127 is not connected is called a _
cut vertex_. we also consider as cut vertices the endpoints of @xmath18 in @xmath106 .
a 2-cell of @xmath106 must have more than one cut vertex on its boundary .
otherwise , this boundary is entirely labelled by a subpath of either @xmath18 or @xmath71 and theorem [ th : four - vertex ] implies the existence of four brackets , one of which ( in fact two ) must avoid the cut vertex , hence be contained in the interior of this subpath .
this would contradict that @xmath18 and @xmath71 are geodesic .
moreover , because @xmath18 and @xmath71 have no spur , @xmath106 can not have more than two degree one vertices .
it follows that @xmath106 is an alternating sequence of paths and 2-cells .
in particular , each 2-cell has exactly two cut vertices .
no 2-cell in this sequence has an interior vertex . for otherwise , by the second part of theorem [ th : four - vertex ]
, the boundary of this 2-cell would contain five brackets one of which would be contained in the interior of either @xmath18 or @xmath71 and this would again contradict that @xmath18 and @xmath71 are geodesic .
it follows that the dual of a 2-cell , viewed as an assembling of quads , is a tree .
we finally remark that this tree must be a path with a staircase shape .
indeed , any other shape would imply the existence of a bracket in either @xmath18 or @xmath71 .
[ cor : equal - length ] with the hypothesis of theorem [ th : geodesic ] , @xmath18 and @xmath71 have equal length which is minimal among homotopic curves . [ cor : no - monogon ] a combinatorial geodesic has no non - trivial index path whose image path is contractible .
the next two remarks follow directly from the characterization of geodesics and canonical forms in terms of spurs , brackets and turns .
[ rem : sub - geodesic ] the image path of any index path of a combinatorial geodesic is geodesic .
if the combinatorial geodesic is in canonical form , so is the image path .
[ rem : geodesic - power ] likewise , any power @xmath128 of a combinatorial closed geodesic @xmath18 is also a combinatorial geodesic .
moreover , if @xmath18 is in canonical form , so is @xmath128 .
let @xmath129 and @xmath130 be canonical paths such that @xmath131 . in other words
, @xmath132 is the leftmost geodesic homotopic to @xmath129 . by theorem
[ th : geodesic ] , there is a disk diagram @xmath106 composed of quad staircases and paths and whose left and right boundaries @xmath133 and @xmath134 are labelled by @xmath132 and @xmath129 respectively .
the next technical lemma will be used in proposition [ prop : counting - crossings ] to analyze the intersection of canonical curves .
a is a non - boundary edge of @xmath106 .
[ lem : double - path - struct ] let @xmath135 be two vertices , one on each side of @xmath106 . then @xmath106 contains a path @xmath136 from @xmath66 to @xmath96 labelled by a canonical path .
moreover , @xmath136 can be uniquely decomposed as either @xmath137 , @xmath138 , @xmath139 or @xmath140 , where 1 .
[ it : lambda ] @xmath141 is a subpath ( possibly reduced to a vertex ) of @xmath133 or @xmath142 , 2 .
@xmath143 is a subpath ( possibly reduced to a vertex ) of @xmath134 or @xmath144 , 3 .
@xmath145 is a spoke , 4 .
[ it : neg - l ] if @xmath141 is a subpath of @xmath142 of positive length then @xmath146 , 5 .
[ it : pos - r ] if @xmath143 is a subpath of @xmath134 of positive length then @xmath147 .
[ it : both - neg ] if @xmath143 is a subpath of @xmath144 and @xmath141 is a subpath of @xmath133 then either @xmath143 is reduced to a vertex and @xmath147 or @xmath141 is reduced to a vertex and @xmath146 .
we assume that @xmath66 is on the left side and @xmath96 on the right side of @xmath106 , the other case being symmetric .
let @xmath148 be such that @xmath149 and @xmath150 .
we first consider the case where @xmath151 .
if @xmath133 and @xmath134 coincide at @xmath66 , then we trivially obtain the desired decomposition as @xmath152 . otherwise , @xmath66 may be incident to @xmath153 or @xmath154 spokes . *
if @xmath66 is not incident to a spoke then @xmath155 is either the initial tip of a staircase or is incident to a spoke @xmath145 .
we set @xmath156 in the first case and @xmath157 otherwise .
the path @xmath158 has no spurs or @xmath123 turns but may start with a bracket .
if not , by the characterization of theorem [ th : canonical - path ] , @xmath158 is already canonical and we can set @xmath159 . otherwise , we short cut the bracket in @xmath158 to obtain a canonical path @xmath136 satisfying the above points [ it : lambda ] to [ it : pos - r ] .
see figure [ fig : structure - geodesic ] .
* if @xmath66 is incident to exactly one spoke @xmath145 , then @xmath145 connects @xmath66 to either @xmath160 or @xmath161 . in the former case we set @xmath162 . as above
, @xmath158 may be canonical or starts with a bracket and we easily obtain the path @xmath136 with the desired properties .
see figure [ fig : structure - geodesic - i ]
. + if @xmath145 is incident to @xmath161 and @xmath163 the path @xmath164 has the required properties .
when @xmath145 is incident to @xmath161 and @xmath165 , then either @xmath155 is incident to a spoke @xmath166 and we must have @xmath167 , or @xmath155 must be the initial tip of a staircase and we must have @xmath168 . *
if @xmath66 is incident to two spokes , then one of them , @xmath169 , connects @xmath66 to @xmath160 and the other @xmath170 connects @xmath66 to @xmath161 .
we can directly set @xmath171 if @xmath165 and @xmath172 if @xmath163 . for point
[ it : both - neg ] in the lemma , we note that we can not have @xmath173 or @xmath174 with both @xmath141 and @xmath143 being subpaths of positive length of @xmath133 and @xmath144 respectively .
indeed , @xmath136 would have a spur or a @xmath175 turn in the first case and a bracket in the other case .
moreover , if @xmath143 is reduced to a vertex and @xmath141 is a subpath of @xmath133 , we may assume that the intersection @xmath176 is reduced to @xmath143 .
otherwise , the canonical path between any other intersection point and @xmath143 would have to follow @xmath134 and we could express @xmath136 so that @xmath143 is a subpath of positive length of @xmath134 .
an analogous argument holds to show that we can assume @xmath177 if @xmath141 is reduced to a vertex .
we next consider the case @xmath178 .
when @xmath133 and @xmath134 coincide at @xmath96 , we obtain the desired decomposition as @xmath179 . otherwise , @xmath96 may be incident to @xmath153 or @xmath154 spokes .
similar arguments as in the case @xmath151 allow to conclude the proof .
following poincar s original approach we represent the surface @xmath4 as the hyperbolic quotient surface @xmath180 where @xmath25 is a discrete group of hyperbolic motions of the poincar disk @xmath181 .
we denote by @xmath182 the universal covering map .
any closed curve @xmath183 gives rise to its infinite power @xmath184 that wraps around @xmath13 infinitely many times .
a of @xmath13 is any curve @xmath185 such that @xmath186 where the parameter of @xmath187 is defined up to an integer translation ( we thus identify the curves @xmath188 , @xmath189 ) . note that @xmath190 is the union of all the images @xmath191 of @xmath187 by the motions in @xmath25 .
the curve @xmath187 has two limit points on the boundary of @xmath181 which can be joined by a unique hyperbolic line @xmath26 .
the projection @xmath192 wraps infinitely many times around the unique geodesic homotopic to @xmath13 .
in particular , the limit points of @xmath187 are independent of the chosen representative in the homotopy class of @xmath13 .
no two motions of @xmath25 have a limit point in common unless they are powers of the same motion .
this can be used to show that when @xmath13 is primitive , its lifts are uniquely identified by their limit points @xcite .
let @xmath13 and @xmath193 be two primitive curves .
we fix a lift @xmath187 of @xmath13 and denote by @xmath194 the hyperbolic motion sending @xmath195 to @xmath196 .
let @xmath197 be the set of lifts of @xmath193 .
we consider the subset of lifts @xmath198 [ lem : strategy ] @xmath199 where @xmath200 is the set of equivalence classes of lifts generated by the relations @xmath201
. put @xmath202 .
define a map @xmath203 as follows .
given @xmath204 there is , by the unique lifting property of coverings , a unique lift @xmath205 of @xmath193 that satisfies @xmath206 .
we set @xmath207 to the class of this lift .
note that changing @xmath208 to @xmath209 , @xmath189 , leads to the lift @xmath210 , so that @xmath211 is well - defined .
the map @xmath211 is onto . indeed ,
if @xmath212 then @xmath205 and @xmath187 must intersect at some point @xmath206 .
it follows that @xmath207 is the class of @xmath205 . as an immediate consequence , @xmath213 when @xmath13 and @xmath193 are geodesics all their lifts are hyperbolic lines and
@xmath211 is a bijection .
we conclude that @xmath214 is minimized among all homotopic curves so that @xmath215 . when @xmath13 and @xmath193 are hyperbolic geodesics , their lifts being hyperbolic lines have alternating limit points exactly when they have a non - empty intersection and they have a unique intersection point in that case .
this point projects to a crossing of @xmath13 and @xmath193 that actually identifies the corresponding element of @xmath200 .
when @xmath13 and @xmath193 are not geodesic the situation is more ambiguous and their lifts may have multiple intersection points . those intersection points project to crossings of @xmath13 and @xmath193 , so that the elements of @xmath200 are now identified with subsets of crossing points ( with odd cardinality ) rather than single crossing points
. the induced partition is generated by the following relation : two crossings are equivalent if they are connected by a pair of homotopic subpaths of @xmath13 and @xmath193 , namely one of the two subpaths of @xmath13 and one of the two subpaths of @xmath193 cut by the two crossings .
indeed , if the two crossings are projections of intersections of two lifts , then the paths between those intersections in each lift project to homotopic paths .
conversely , homotopic paths lift to paths with common endpoints that can be seen as subpaths of two intersecting lifts . in order to compute the above partition , we thus essentially need an efficient procedure for testing if two paths are homotopic .
this homotopy test can indeed be performed in linear time according to theorem [ th : canonical ] .
since a combinatorial curve of length @xmath0 may have @xmath216 crossings , we directly obtain an algorithm with time complexity @xmath217 to compute the above partition .
when dealing with combinatorial ( canonical ) geodesics , the situation is more constrained and somehow intermediate between the ideal hyperbolic case and the most general situation .
thanks to theorem [ th : geodesic ] , we know that homotopic paths must stay parallel and at distance at most one .
see figure [ fig : strategy ] .
this allows us to identify equivalent crossings more efficiently .
these ideas are formalized in the next sections .
let @xmath125 be two combinatorial closed curves on a combinatorial surface .
a of @xmath72 of length @xmath0 is a pair of forward index paths @xmath218 with the same image path @xmath219 . if @xmath220 then the double path is just a double point .
a double path of @xmath18 is defined similarly , taking @xmath82 and assuming @xmath221 .
the next lemma follows from remark [ rem : sub - geodesic ] .
[ lem : double - path ] let @xmath222 and @xmath223 be forward index paths of two canonical curves @xmath18 and @xmath71 such that the image paths @xmath224 and @xmath225 are homotopic .
then @xmath226 and @xmath218 is a double path .
a double path @xmath227 gives rise to a sequence of @xmath115 double points @xmath228 for @xmath229 $ ] .
a priori a double point could occur several times in this sequence .
the next two lemmas claim that this is not possible when the curves are primitive . recall that a curve is if its homotopy class can not be expressed as a proper power of another class .
[ lem : short - double - path - c ] a double path of a primitive combinatorial curve @xmath18 can not contain a double point more than once in its sequence .
in particular , a double path of @xmath18 must be strictly shorter than @xmath18 .
suppose that a double path @xmath230 of @xmath18 contains two occurrences of a double point @xmath231 . because the couples @xmath231 and @xmath232 represent the same double point there are two cases to consider .
* if @xmath230 contains the couple @xmath231 twice then it must contain a subsequence of length @xmath44 starting with @xmath231 .
we thus have @xmath233 .
this implies that @xmath18 is equal to some non - trivial circular permutation of itself .
it is a simple exercise to check that @xmath18 must then be a proper power of some other curve , contradicting that @xmath18 is primitive . * otherwise @xmath230 contains @xmath231 and @xmath232 .
let @xmath0 be the distance between these two occurrences in @xmath230 .
we thus have @xmath234 from which we deduce that @xmath18 is a square ( and @xmath235 ) , contradicting that @xmath18 is primitive .
[ lem : short - double - path - c - d ] let @xmath18 and @xmath71 be two non - homotopic primitive combinatorial curves .
a double path of @xmath72 can not contain a double point more than once in its sequence .
moreover , the length of a double path of @xmath72 must be less than @xmath236 .
suppose that a double path of @xmath72 contains two occurrences of a double point .
after shortening the double path if necessary , we may assume that these two occurrences are the first and the last double points of the double path .
its length must accordingly be a nonzero integer multiple @xmath136 of @xmath44 as well as a nonzero integer multiple @xmath158 of @xmath237 .
it follows that for some circular permutations @xmath87 of @xmath18 and @xmath238 of @xmath71 we have @xmath239 . by a classical result of combinatorics on words (
1.3.1 ) this implies that @xmath87 and @xmath238 are powers of a same curve , in contradiction with the hypotheses in the lemma .
in fact , by a refinement due to fine and wilf ( * ? ? ? * prop .
1.3.5 ) it suffices that @xmath240 and @xmath241 have a common prefix of length @xmath236 to conclude that @xmath87 and @xmath238 are powers of a same curve .
this proves the second part of the lemma .
a double path whose index paths can not be extended is said . as an immediate consequence of lemmas [
lem : short - double - path - c ] and [ lem : short - double - path - c - d ] we have : [ cor : partition - double - paths ] the maximal double paths of a primitive curve or of two primitive curves in canonical form induce a partition of the double points of the curves .
let @xmath231 and @xmath242 be the first and the last double points of a maximal double path of @xmath72 , possibly with @xmath82 .
when @xmath243 the arcs @xmath244 $ ] , @xmath245 $ ] , @xmath246 $ ] must be pairwise distinct because canonical curves have no spurs , and similarly for the three arcs @xmath247 $ ] , @xmath248 $ ] , @xmath249 $ ] .
we declare the maximal double path to be a if the circular ordering of the first three arcs at @xmath45 and the circular ordering of the last three arcs at @xmath250 are either both clockwise or both counterclockwise with respect to the rotation system of the system of quads .
when @xmath220 , that is when the maximal double path is reduced to the double point .
which notion of crossings is used should always be clear from the context . ]
@xmath251 , we require that the arcs @xmath252,d[j , j-1 ] , c[i , i+1 ] , d[j , j+1]$ ] are _ pairwise distinct _ and appear in this circular order , or its opposite , around the vertex @xmath253 .
let @xmath125 be primitive combinatorial curves such that @xmath71 is canonical and let @xmath129 and @xmath254 be the canonical curves homotopic to @xmath18 and @xmath54 respectively .
we denote by @xmath106 the annular diagram corresponding to @xmath129 and @xmath132 . when the two boundaries @xmath134 and @xmath133 of @xmath106 have a common vertex we implicitly assume that @xmath129 and @xmath132 are indexed so that this vertex corresponds to the same index along @xmath134 and @xmath133 .
we consider the following set of double paths : * @xmath255 is the set of crossing double paths of positive length of @xmath129 and @xmath71 , * @xmath256 is the set of crossing double paths @xmath251 of zero length of @xmath129 and @xmath71 such that either * * the two boundaries of @xmath106 coincide at @xmath257 and @xmath258 = c_l[i-1,i]$ ] or @xmath259 = c_l[i , i+1]$ ] , or * * one of @xmath260 $ ] or @xmath259 $ ] is the label of a spoke @xmath261 of @xmath106 and @xmath262 = c_l[i'-1,i']$ ] in the first case or @xmath263 = c_l[i',i'+1]$ ] in the other case .
* @xmath264 is the set of crossing double paths @xmath265 ( @xmath266 ) of @xmath254 and @xmath71 such that _ none _ of the following situations occurs : * * the two boundaries of @xmath106 coincide at @xmath267 and @xmath258 = c_r[i'-1,i']$ ] , * * the two boundaries of @xmath106 coincide at @xmath268 and @xmath269 , * * @xmath258 $ ] is the label of a spoke @xmath270 of @xmath106 and @xmath262 = c_r[i'-1,i']$ ] , * * @xmath271 $ ] is the label of a spoke @xmath272 of @xmath106 and @xmath273 = c_r[i',i'+1]$ ] .
those definitions allow the case @xmath274 , recalling that the index paths of a double path of @xmath18 must be distinct by definition . referring to section [ sec : strategy ]
, we view the underlying surface of the system of quads @xmath43 as a quotient @xmath180 of the poincar disk .
the system of quads lifts to a quadrangulation of @xmath181 and the lifts of a combinatorial curve in @xmath43 are combinatorial bi - infinite paths in this quadrangulation . by remark [ rem : geodesic - power ] , if the combinatorial curve is geodesic ( resp .
canonical ) so are its lifts . in this case , each lift is simple by corollary [ cor : no - monogon ] .
we fix a lift @xmath275 of @xmath129 and consider the set @xmath200 of lemma [ lem : strategy ] corresponding to the classes of lifts of @xmath71 whose limit points alternate with the limit points of @xmath275 along @xmath276 . [
prop : counting - crossings ] @xmath200 is in 1 - 1 correspondence with the disjoint union @xmath277 .
let @xmath278 be the lift of @xmath132 with the same limit points as @xmath275 .
these two lifts project onto the boundaries of the annular diagram @xmath106 and thus form an infinite strip @xmath279 of width at most 1 in @xmath181 composed of paths and quad staircases ( possibly a single infinite staircase ) .
we shall define a correspondence between @xmath200 and @xmath277 . to this end
we consider a lift @xmath280 of @xmath71 whose limit points alternate with those of @xmath275 .
in other words , @xmath281 . the lift @xmath280 must cross @xmath279 .
let @xmath282 and @xmath283 be respectively the smallest and largest index @xmath284 such that @xmath285 is in @xmath279 . by remark [ rem : sub - geodesic ] , the corresponding subpath @xmath286 $ ] of @xmath280 is canonical .
since @xmath181 is simply connected , @xmath286 $ ] is homotopic to any path joining the same extremities and we can apply lemma [ lem : double - path - struct ] to show that @xmath286 $ ] is actually contained in @xmath279 and that it can be decomposed as either @xmath137 , @xmath138 , @xmath139 or @xmath140 where @xmath145 is a spoke of @xmath279 , @xmath287 and @xmath288 for some @xmath289 . * if @xmath290 then by point [ it : pos - r ] of lemma [ lem : double - path - struct ] we have @xmath291 so that this intersection defines a maximal double path of @xmath280 and @xmath275 .
it must be crossing since @xmath281 .
its projection on @xmath43 is a crossing double path of length @xmath290 of @xmath71 and @xmath129 , hence in @xmath255 , to which we map @xmath280 . *
if @xmath292 and @xmath293 , then point [ it : both - neg ] of lemma [ lem : double - path - struct ] implies @xmath291 . as above , @xmath143 must define a crossing double path of length zero of @xmath280 and @xmath275 .
we map @xmath280 to the projection of this crossing double point on @xmath43 and remark that this projection is in @xmath256 .
* otherwise , we must have @xmath294 and @xmath295 by point [ it : both - neg ] of lemma [ lem : double - path - struct ] .
if @xmath296 then point [ it : neg - l ] of lemma [ lem : double - path - struct ] implies @xmath297 .
this intersection defines a crossing double path of @xmath280 and @xmath298 and we map @xmath280 to its projection on @xmath43 .
if @xmath220 and @xmath299 then point [ it : both - neg ] of lemma [ lem : double - path - struct ] implies @xmath300 , which also holds true if @xmath301 . in both cases
@xmath141 corresponds to a crossing double path of length zero of @xmath280 and @xmath302 and we map @xmath280 to its projection on @xmath43 .
we finally remark that this last projection or the above one belong to @xmath264 .
because @xmath279 is left globally invariant by @xmath303 ( the hyperbolic motion that sends @xmath304 to @xmath305 ) , we have @xmath306 .
it follows that @xmath280 and @xmath307 are mapped to the same crossing double path by the above rules .
we thus have a well defined map @xmath308 .
the uniqueness of the decomposition in lemma [ lem : double - path - struct ] implies that this map is 1 - 1 . in order to check that the map is onto
we consider a maximal crossing double path @xmath309 of @xmath129 and @xmath71 in @xmath255 . by the unique lift property of coverings
there is a unique lift @xmath280 of @xmath71 such that @xmath310 and @xmath230 lifts to a crossing double path of @xmath280 and @xmath275 . by lemma [ lem : double - path ]
this double path is the only intersection of @xmath280 and @xmath275 so these lifts must have alternating limit points .
in other words @xmath280 is in @xmath311 and is mapped to @xmath230 .
a similar argument applies to the crossing double paths of @xmath256 and @xmath264 .
this leads to a simple algorithm for computing combinatorial crossing numbers .
[ cor : crude - result ] let @xmath125 be primitive curves of length at most @xmath0 on a combinatorial surface with complexity @xmath1 .
the crossing numbers @xmath312 and @xmath19 can be computed in @xmath313 time . by lemma [ lem : systems - of - quads ]
we may assume that the surface is a system of quads . by theorem
[ th : canonical ] we may compute the canonical forms of @xmath314 and @xmath71 in @xmath103 time . according to proposition [
prop : counting - crossings ] , we have @xmath315 the set @xmath255 can be constructed in @xmath216 time . indeed , since the maximal double paths of @xmath18 and @xmath71 form disjoint sets of double points by corollary [ cor : partition - double - paths ] , we just need to traverse the grid @xmath316 and group the double points into maximal double paths .
those correspond to diagonal segments in the grid that can be computed in time proportional to the size of the grid .
we can also determine which double paths are crossing in the same amount of time .
likewise , we can construct the sets @xmath256 and @xmath264 in @xmath216 time .
we can also compute @xmath19 in quadratic time using that @xmath317 .
in order to finish the proof of theorem [ th : main - result ] , we need to tackle the case of non - primitive curves .
thanks to canonical forms , computing the primitive root of a curve becomes extremely simple .
[ lem : non - primitive ] let @xmath18 be a combinatorial curve of length @xmath293 in canonical form .
a primitive curve @xmath71 such that @xmath18 is homotopic to @xmath318 for some integer @xmath284 can be computed in @xmath103 time . by theorem
[ th : canonical ] , we may assume that @xmath18 and @xmath71 are in canonical form . by remark [ rem : geodesic - power
] , the curve @xmath318 is also in canonical form .
the uniqueness of the canonical form implies that @xmath319 , possibly after some circular shift of @xmath71 .
it follows that @xmath71 is the smallest prefix of @xmath18 such that @xmath18 is a power of this prefix .
it can be found in @xmath103 time using a variation of the knuth - morris - pratt algorithm to find the smallest period of a word @xcite . the geometric intersection number of non - primitive curves is related to the geometric intersection number of their primitive roots .
the next result is rather intuitive , see figure [ fig : non - primitive ] , and is part of the folklore although we could only find references in some relatively recent papers .
[ prop : non - primitive - formulas ] let @xmath18 and @xmath71 be primitive curves and let @xmath320 be integers . then , @xmath321 let @xmath125 and @xmath43 be the two combinatorial curves and the combinatorial surface as in the theorem .
we first assume that the @xmath43 has negative euler characteristic .
we can compute the canonical forms of @xmath18 and @xmath71 in @xmath103 time after @xmath102 time preprocessing by lemma [ lem : systems - of - quads ] .
thanks to lemma [ lem : non - primitive ] we can further determine primitive curves @xmath87 and
@xmath238 and integers @xmath320 such that @xmath322 and @xmath323 in @xmath103 time .
we can compute @xmath324 and @xmath325 in @xmath216 time according to corollary [ cor : crude - result ] .
we finally use the formulas in the previous proposition to deduce @xmath312 and @xmath19 from @xmath324 and @xmath325 .
if @xmath43 is a sphere or a disk , then every curve is contractible and @xmath326 .
if @xmath43 is a cylinder , then every two curves can be made non crossing so that @xmath327 while @xmath328 . finally ,
if @xmath43 is a torus , the radial graph of the system of quads can be decomposed into two loops @xmath329 such that @xmath330 and @xmath331 .
we may then use the classical formulas : @xmath332 and @xmath333 .
in the subsequent sections we only deal with the self - intersection number of a single curve . we thus drop the subscript @xmath18 to denote an index path @xmath334 or an arc occurrence @xmath335 .
we also implicitly consider intersections as self - intersections . by theorem
[ th : main - result ] we can compute the geometric intersection number of a curve efficiently . here , we describe a way to compute a minimal immersion , that is an actual immersion with the minimal number of intersections .
we refer to the combinatorial framework of section [ sec : framework ] to describe such an immersion combinatorially .
thanks to lemma [ lem : geometric - vs - combinatorial ] , we know that this framework faithfully encodes the topological configurations . [ [ bigons - and - monogons . ] ] bigons and monogons .
+ + + + + + + + + + + + + + + + + + + + a of an immersion @xmath70 of @xmath18 is a pair of index paths @xmath336 whose @xmath224 and @xmath337 have strictly positive lengths , are homotopic , and whose @xmath251 and @xmath338 are combinatorial crossings for @xmath70 .
a of @xmath70 is an index path @xmath334 of strictly positive length such that @xmath339 is a combinatorial crossing and the image path @xmath224 is contractible .
[ prop : bigon ] a combinatorial immersion of a primitive curve has excess self - crossing if and only if it contains a bigon or a monogon . we just need to prove the existence of a bigon or monogon for a continuous realization @xmath34 of the combinatorial immersion .
this bigon or monogon corresponds to a combinatorial bigon or monogon as claimed in the proposition . to prove the topological counterpart , we first note as in section [ sec : strategy ] that the minimal number of intersections is counted by the lifts of @xmath22 whose limit points alternate along @xmath27 .
it follows that @xmath22 is minimally crossing when its lifts are pairwise disjoint unless the corresponding limit points alternate in @xmath27 and they should intersect exactly once in that case .
when @xmath22 is not minimally crossing there must exist two lifts intersecting more than necessary , thus forming a bigon in @xmath181 .
the projection of such a bigon on @xmath4 gives the desired result .
note that a statement similar to the one in the proposition holds for an immersion of a pair of primitive curves @xmath18 and @xmath71 with excess crossing . in appendix [ app : bigon ] , we provide a purely combinatorial proof of the direct implication of the proposition in the case of two curves .
suppose that @xmath86 is a combinatorial immersion of a primitive geodesic @xmath18 .
it can not have a monogon by corollary [ cor : no - monogon ] .
hence , according to proposition [ prop : bigon ] the immersion @xmath86 is minimal unless it contains a bigon .
the two sides of such a bigon may share some common part that would prevent us from swapping these two sides . in agreement with the terminology of hass and scott @xcite ,
a bigon @xmath336 is said if 1 .
its two index paths have disjoint interiors , i.e. they do not share any arc occurrence ; 2 .
when @xmath340 the following arc occurrences @xmath341,\focc{\imath},[\jmath+k,\jmath+k+1],[j , j+1],[\jmath+k,\jmath+k-1]\ ] ] do not appear in this order or its opposite in the circular ordering induced by @xmath86 at @xmath342 ; 3 .
when @xmath343 the following arc occurrences @xmath344 do not appear in this order or its opposite in the circular ordering induced by @xmath86 at @xmath345 .
note that when @xmath18 is geodesic and @xmath346 there can not be any identification between @xmath347 and @xmath348 .
for instance , @xmath349 implies @xmath350 so that @xmath351 , while @xmath340 implies @xmath352 so that @xmath353 . in both cases
@xmath18 would have a monogon which would contradict lemma [ cor : no - monogon ] .
when @xmath18 is primitive and geodesic , we can not have @xmath354 and @xmath355 at the same time
. otherwise , we must have @xmath226 by the preceding paragraph , and @xmath18 would be a circular shift of @xmath356 and thus homotopic to a square . also remark that when one of the last two conditions in the definition is not satisfied , the bigon maps to a non - singular bigon in the continuous realization of @xmath70 as in lemma [ lem : geometric - vs - combinatorial ] .
see figure [ fig : singular - def ] .
when the bigon is singular we can swap its two sides by exchanging the two arc occurrences @xmath357 $ ] and @xmath358 $ ] , for @xmath359 and @xmath360 .
[ lem : swap ] swapping the two sides of a singular bigon of an immersion of a geodesic primitive curve decreases its number of crossings by at least two .
this is relatively obvious if one considers a continuous realization of the immersion , performs the swapping and comes back to a combinatorial immersion as in the proof of lemma [ lem : geometric - vs - combinatorial ] .
we nonetheless provide a purely combinatorial proof in appendix [ app : swap ] .
hence , by swapping singular bigons we may decrease the number of crossings until there is no more singular bigons .
it follows from the next lemma that the resulting immersion has no excess crossing .
[ th : singular - bigon ] an immersion of a primitive geodesic curve has excess crossing if and only if it contains a singular bigon .
we realize @xmath70 by a continuous curve @xmath22 with the same construction as in lemma [ lem : geometric - vs - combinatorial ] . by theorem 4.2 in @xcite the curve @xmath22 has a singular bigon . in turn
this bigon corresponds to a singular bigon in @xmath70 . in the next section
we describe how to detect bigons in practice .
if an immersion @xmath86 of a primitive geodesic @xmath18 has a bigon then its two sides have equal length according to corollary [ cor : equal - length ] and form a sequence of paths and staircases following the description of theorem [ th : geodesic ] . in particular , the vertices of the two sides can be put in 1 - 1 correspondence and corresponding vertices are at distance at most one . hence ,
when looking for bigons in @xmath86 , we can start from any crossing @xmath251 and walk in parallel along the two sides from @xmath282 and @xmath283 , checking that the two sides stay at distance one .
however , we do not know a priori if the two sides will cross again to form a bigon and the search may be unsuccessful .
in order to analyze the complexity of the bigon search we thus consider pair of paths that could potentially be part of a bigon , but that are not necessarily so . formally , a pair @xmath361 of distinct index paths such that @xmath224 and @xmath362 are corresponding subpaths of homotopic geodesic paths is called a of length @xmath0 . in other words ,
a thick double path is such that its image paths can be extended to form homotopic geodesic paths . as for a double path ,
a thick double path gives rise to a sequence of @xmath115 index pairs @xmath363 for @xmath229 $ ] .
let @xmath106 be a disk diagram whose boundary is labelled by homotopic geodesic paths extending the thick double path as illustrated on figure [ fig : thick ] . the restriction of @xmath106 to the part delimited by the thick double path is called a . in a partial diagram
each index pair @xmath363 may be in at most one of five : either the corresponding vertices coincide in the partial diagram , or they appear diagonally opposite in a quad ; in turn this quad may be labelled by one of the ( at most ) four quads incident to either @xmath364 $ ] or @xmath365 $ ] .
a partial diagram is if it can not be extended to form a partial diagram of a longer thick double path .
[ lem : configurations ] if @xmath18 is a primitive geodesic curve each configuration of an index pair may occur at most once in a partial diagram .
suppose for a contradiction that the index pair @xmath251 occurs twice in the same configuration in a partial diagram of a thick double path of @xmath18 .
exchanging the roles of @xmath282 and @xmath283 if necessary , we may assume that the indices @xmath148 are represented by integers such that @xmath366 .
the two occurrences of @xmath251 are separated by a path whose length is an integer multiple of @xmath44 , say @xmath367 .
we first assume that the two index paths of the thick double path are both forward . in the partial diagram , @xmath345 and @xmath342 label vertices that are either identical or opposite in a quad .
they are thus connected by a path @xmath136 of length zero or two .
we infer that @xmath368{\ensuremath{^{-1}}}\cdot c\ipath{\imath}{k|c|}\cdot c[i , j]\ ] ] this can be written as @xmath369 where @xmath370 is a cyclic permutation of @xmath18 and @xmath371{\ensuremath{^{-1}}}$ ] is a ( closed ) path of length at most @xmath372 .
since the homotopy classes of the loops @xmath373 and @xmath158 are commuting they must admit a common primitive root @xcite .
we thus have _ ( i ) _ @xmath374 and _ ( ii ) _
@xmath375 for some primitive curve @xmath208 and some @xmath376 .
( i ) _ implies that @xmath87 and @xmath377 are also commuting so that @xmath87 is homotopic to a power of the primitive curve @xmath208 . but
@xmath87 being also primitive we actually have @xmath378 . by remark [ rem : geodesic - power ] , @xmath379 is geodesic , hence has minimal length in its homotopy class . from _
( ii ) _ we obtain @xmath380 .
this is only possible if @xmath381 or if @xmath382 and @xmath383 ( indices in a pair have the same parity ) . *
if @xmath381 , then @xmath384 , so that @xmath385 $ ] . since @xmath18 is geodesic , we must have @xmath386 , hence @xmath387 and the corresponding vertices must be diagonally opposite in a quad .
two homotopic paths each of length two bound a disk diagram composed of either a length two path or a single quad .
in other words , @xmath388 $ ] and @xmath136 are either equal or bound the first quad in the partial diagram . in either case we infer that @xmath389= c[i , j].c[j , j+1]$ ] contains a spur or a bracket , leading to a contradiction . * if @xmath382 and @xmath383 , then @xmath390 ( recall that @xmath391 ) . a simple case study of the possible partial diagrams for such a @xmath18 lead to a contradiction in each case .
see the appendix [ sec : end - proof - lem - config ] for a proof .
we now assume that the thick double path is composed of a forward and a backward index paths . similarly to the previous case , we infer that @xmath392 , where @xmath370 and @xmath371{\ensuremath{^{-1}}}$ ] .
equivalently , @xmath18 and its inverse are freely homotopic .
this is however impossible unless @xmath18 is contractible .
[ cor : index - pairs ] each configuration of an index pair may occur at most once in the set of maximal partial diagrams . by the preceding lemma
, we only need to check that a configuration of an index pair can not occur twice in distinct partial diagrams of maximal thick double paths .
this is essentially the unique lift property of coverings . by theorem
[ th : canonical ] we can assume that @xmath18 is geodesic .
we first consider the case where @xmath18 is primitive .
let @xmath70 be a randomly chosen immersion of @xmath18 . we store the induced vertex orderings of the arc occurrences in arrays so that we can test if a double point is a crossing in constant time using pointer arithmetic . by theorem
[ th : singular - bigon ] , if @xmath70 has excess crossing it has a singular bigon .
this singular bigon defines a thick double path that must appear in some maximal partial diagram of @xmath70 as two boundary paths with common extremities .
hence , in order to find a singular bigon we just need to scan the index pairs in every maximal partial diagram and test whether the two sides between consecutive index pairs define a singular bigon . by the very definition of a singular bigon
, each test can be easily performed in constant time .
the number of tests is itself bounded by the number of index pairs in all the maximal partial diagrams .
this is @xmath216 by corollary [ cor : index - pairs ] .
once a singular bigon is found we swap its sides in @xmath103 time .
this preserves the geodesic character and , by lemma [ lem : swap ] , the number of crossings is reduced by at least two . since @xmath70 may have @xmath216 excess crossings , we need to repeat the above procedure @xmath216 times and we may conclude the theorem in the case of primitive curves . when @xmath393 , with @xmath71 primitive , we first find a minimally crossing immersion for @xmath71 by the above procedure .
we further traverse the immersion @xmath136 times duplicating @xmath71 as many times .
as we start each traversal we connect the last arc occurrence of the previously traversed copy with the first occurrence of the next copy .
we continue this copy by duplicating each traversed arc occurrence to its right .
it is easily seen that the number of crossings of the final immersion of @xmath18 , after connecting the last traversed arc with the first one , satisfies the formula in proposition [ prop : non - primitive - formulas ] .
we end this section with some refinements of lemma [ lem : configurations ] and its corollary .
we first highlight a simple property of the system of quads . [ lem : facial - walk ] the facial walk of a quad can not contain an arc twice , either with the same or the opposite orientations .
if the facial walk of a quad contains two occurrences of an arc with the same orientation , then their identification creates a mbius strip in the system of quads , in contradiction with its orientability .
if the facial walk contains two consecutive occurrences of an arc with opposite orientations , then their common endpoint must have degree one in the system of quads .
this contradicts the minimal degree 8 in our system of quads .
finally , if the facial walk contains non consecutive occurrences of an arc with opposite orientations , then their identification creates a cylinder bounded by loop edges .
however , the radial graph being bipartite , this can not occur .
[ lem : no - common - index ] let @xmath18 be a primitive geodesic curve . if a forward index path @xmath394 and a backward index path @xmath395 have homotopic image paths @xmath396 , then they can not share any index .
remark that we can not have @xmath397 for otherwise @xmath398 in contradiction with corollary [ cor : no - monogon ] .
likewise , we can not have @xmath399 . by way of contradiction , suppose that the index paths @xmath394 and @xmath395 have a common index @xmath400 for some integers @xmath401 . by the previous remark we have @xmath402 .
it follows that @xmath403 is an interior index pair of @xmath404 , where @xmath405 ( recall that @xmath282 and @xmath283 have the same parity ) .
consider a disk diagram for @xmath396 .
the index pair @xmath403 can not correspond to the same vertex in the diagram since the path @xmath388=c[i , m].c[m , j]$ ] would then label a closed path in the diagram , hence be contractible .
this would again contradict corollary [ cor : no - monogon ] .
it ensues that @xmath403 corresponds to opposite vertices in a quad of the diagram .
this quad is part of a staircase where the arc @xmath406 $ ] labels two arcs on either sides of the partial diagram .
figure [ fig : m - m - opposite-1 ] depicts the four possible configurations . in each case
we infer that the system of quads would contain a mbius strip , in contradiction with its orientability .
[ lem : ell - plus - un ] two distinct index paths of a non - trivial primitive geodesic curve @xmath18 having homotopic image paths have length distinct from @xmath44 and at most @xmath407 .
when the index paths have opposite orientations , the previous lemma directly implies the result .
let @xmath408 and let @xmath409 be two index paths such that @xmath410 . clearly , @xmath411 since otherwise @xmath18 would be homotopic to a conjugate of itself by the shorter curve @xmath388 $ ] , in contradiction with the primitivity of @xmath18 . for
the sake of a contradiction , suppose that @xmath412 .
consider a diagram @xmath106 with sides @xmath134 and @xmath133 for the homotopic paths @xmath410 . by lemma [ lem : configurations ] the vertices @xmath413 and @xmath414 must be diagonally opposite in a quad of @xmath106 . using that @xmath415 and @xmath416
are labelled by the same edges and similarly for @xmath417 and @xmath418 we easily deduce that the system of quads has a quad with two occurrences of a same arc in its facial walk , or has a vertex of degree at most @xmath419 , or is non - orientable .
this would contradict lemma [ lem : facial - walk ] or the hypotheses on the system of quads . in details ,
figure [ fig : lengthbigon ] depicts the five possible configurations ( a , b , c , d ) for @xmath415 and @xmath417 in the diagram @xmath106 and the eight possible configurations ( a1 - 2 , b1 - 3 , c1 - 3 ) for @xmath416 and @xmath418 .
remark that cases c1 - 3 are symmetric to cases b1 - 3 , so that we only need to consider the five configurations a1 - 2 , b1 - 3 .
we denote by @xmath420 the conjunction of configurations @xmath421 and @xmath422 in @xmath106 . the quads in configurations a , b , c , d are part of an initial staircase in @xmath106 that may have a positive or negative orientation according to whether or not its orientation is consistent with some default orientation of the quad system and similarly for the final staircase containing @xmath413 and @xmath414 . *
if the orientations of the two staircases have the same sign , then the face to the right of @xmath423 $ ] has the same facial walk as the face to the right of @xmath424 .
the induced identifications imply that for @xmath425 a quad is bounded twice by the same arc ( with or without the same orientation ) in contradiction with lemma [ lem : facial - walk ] .
if @xmath426 , we easily deduce that @xmath427 has degree four in the system of quads , again a contradiction .
* when the initial and final staircases have opposite orientations the faces to the right of @xmath423 $ ] and to the right of @xmath424 ( resp . to the left of @xmath428 $ ] and to the left @xmath429 $ ] ) correspond to the two faces incident to @xmath430 $ ] ( resp .
we further split case @xmath431 into two variants a@xmath432 and a@xmath433 according to whether the two quads ( see figure [ fig : lengthbigon ] ) have consistent orientations or not . by the induced identifications we derive the following forbidden situations : two occurrences of a same arc in a quad for @xmath434 , a degree two vertex ( namely @xmath435 ) for @xmath436 , a degree three vertex for @xmath437 , a degree four vertex for @xmath438 , a degree five vertex for @xmath439 , and finally a mbius band for @xmath440 .
when the initial and final staircases in the above proof have consistent orientations it is sufficient to use that @xmath441 and @xmath442 are labelled by the same edge , and similarly for @xmath443 and @xmath444 , in order to reach a contradiction for cases @xmath445 and @xmath431 . in other words ,
the two homotopic paths @xmath410 can not have length @xmath115 .
this leads to the following refinement .
[ lem : short - bigon ] a bigon @xmath446 of an immersion of a primitive geodesic curve @xmath18 without intermediate crossings ( i.e. , no index pair @xmath447 , @xmath448 , is a combinatorial crossing ) has length @xmath449 .
[ lem : flat - bigons ] if two forward index paths of a canonical curve have homotopic image paths then they actually have the same image paths , i.e. they form a double path . in particular , bigons composed of two forward paths must be flat . by theorem
[ th : geodesic ] the image paths bound a disk diagram composed of paths and staircase .
remark that a staircase with both sides directed forward must have a @xmath175 turn on its left side .
since the curve is canonical we infer that the diagram can not have any staircase , implying that the two sides coincide .
we now turn to the original problem of poincar @xcite , deciding whether a given curve @xmath18 is homotopic to a simple curve . in the affirmative we know by lemma [ lem : swap ] and theorem [ th : singular - bigon ] that some geodesic homotopic to @xmath18 must have a ( combinatorial ) , i.e. an immersion without crossings . rather than swapping the sides of a singular bigon as in lemma [
lem : swap ] we can choose to switch one side along the other side .
this will also decrease the number of crossings if the bigon contains no other interior bigons . by considering interior - most bigons only
, we can thus enforce a given edge of @xmath18 to stay fix as we remove crossings .
this suggests an incremental computation of an embedding in which the image of the first arc occurrence is left unchanged : we assume that @xmath18 is canonical and consider the trivial embedding of its first arc occurrence @xmath450 $ ] .
we next insert the successive arc occurrences incrementally to obtain an embedding of the path formed by the already inserted arcs .
when inserting the occurrence @xmath451 $ ] we need to compare its left - to - right order with each already inserted arc occurrence @xmath193 of its supporting arc .
if @xmath452 $ ] we can use the comparison of the occurrence @xmath453 $ ] with the occurrence @xmath22 preceding @xmath193 ( or succeeding @xmath193 if it is a backward occurrence ) .
if @xmath453 $ ] and @xmath22 have the same supporting arc , we just propagate their relative order to @xmath451 $ ] and @xmath193 .
otherwise , we use the circular ordering of the supporting arcs of @xmath453 $ ] , @xmath22 and @xmath451 $ ] in order to conclude .
when @xmath454 $ ] , we can not use the occurrence preceding @xmath450 $ ] as it is not yet inserted .
we rather compare @xmath451 $ ] and @xmath450 $ ] as follows . in the poincar disk , we consider two lifts @xmath455 and @xmath456 of @xmath18 such that @xmath457=\tilde{d}_0[0,1]$ ] . we decide to insert @xmath451 $ ] to the left ( right ) of @xmath450 $ ] if one of the limit points of @xmath455 lies to the left ( right ) of @xmath456 .
note that when @xmath18 is homotopic to a simple curve the two limit points of @xmath455 should lie on the same side of @xmath456 . after comparing @xmath451 $ ] with all the occurrences of its supporting arc , we can insert it in the correct place .
if no crossings were introduced this way , we proceed with the next occurrence @xmath458 $ ] .
it may happen , however , that no matter how we insert @xmath451 $ ] in the left - to - right order of its supporting arc , the resulting immersion of @xmath459 will have a combinatorial crossing . in order to handle this case
, we first check if @xmath451 $ ] is , i.e. if for some @xmath460 and some turns @xmath461 the subpath @xmath462 has turn sequence @xmath463 and the index path @xmath464 does not contain the arc occurrence @xmath450 $ ] . when @xmath451 $ ] is switchable we can switch @xmath136 to a new subpath @xmath465 with turn sequence @xmath466 such that @xmath136 and @xmath465 bound a diagram composed of a single horizontal staircase
. see figure [ fig : switch - path ] .
we indeed perform the switch if some intersection is actually avoided this way .
more formally , this happens when there is a crossing @xmath251 such that the turn of @xmath467 , c[i , i+1])$ ] or of @xmath468 , c[i , i+1])$ ] is one as illustrated on figure [ fig : switch - path-1 ] .
note that replacing @xmath136 by @xmath465 leads to a curve @xmath87 that is still geodesic and homotopic to @xmath18 . moreover , _
the part of @xmath87 that remains to be embedded is in canonical form _ as it contains no @xmath175 turns .
we then insert the arc occurrence @xmath451 $ ] and proceed with the algorithm using @xmath87 in place of @xmath18 .
the successive switches in the course of the computation untangle @xmath18 incrementally and we call our embedding procedure the . for further reference
we make some observations that easily follows from the above switch procedure and the fact that @xmath18 is initially canonical .
[ rem : switchable ] if @xmath451 $ ] is switchable but not switched at the time of being processed it remains true until the end of the algorithm , meaning that the vertex of index @xmath282 is directly followed by a subpath @xmath136 with a turn sequence of the form @xmath469 ( @xmath284 may get smaller if some other arc of @xmath136 is switched ) . similarly ,
if @xmath451 $ ] is not switchable because of an inappropriate turn sequence , it remains true until the end of the algorithm .
[ rem:01-right - side ] let @xmath87 be a curve homotopic to @xmath18 obtained after a certain number of switches during the execution of the algorithm .
then @xmath87 can not contain a subpath with turn sequence @xmath470 that ends at index @xmath471 , nor can it contain a subpath with turn sequence @xmath472 that starts at index @xmath471 . [ lem : zipper - complexity ] the algorithm applied to a canonical primitive curve @xmath18 of length @xmath0 can be implemented to run in @xmath473 time .
we first traverse @xmath18 in reverse direction to mark all the switchable arcs .
in the course of the algorithm , each time a switchable arc triggers the switch of a subpath @xmath136 we unmark all the arcs in @xmath136 except the last one that may become switchable ( depending on the last turn of @xmath136 and on the status of the arc following @xmath136 ) .
it easily follows that an arc can be switched at most twice and that the amortized cost for the switches is linear .
alternatively , we could use the run - length encoded turn sequence of @xmath18 ( as defined in @xcite ) to detect each switchable arc and update the turn sequence in constant time per switches . in a preprocessing phase
we also compute the relative order of @xmath450 $ ] with all the other occurrences of the same arc in @xmath18 as follows .
if @xmath364=c[0,1]$ ] , the corresponding arc occurrences form a double path of length one and we compute their relative order by extending maximally this double path in the backward direction .
looking at the tip of this double path , say @xmath474 , we can decide which side is to the left of the other . indeed
, the three arcs @xmath475 , @xmath476 and @xmath477 must be pairwise distinct and their circular order about their common origin vertex in the system of quads provides the necessary information as follows from lemma [ lem : flat - bigons ] . the computation of the maximal extensions in the backward direction amounts to evaluate the longest common prefix of @xmath54 with all its circular shifts .
overall , this can be done in @xmath103 time thanks to a simple variation of the knuth - morris - pratt algorithm .
we next initialize an empty ( balanced ) binary search tree for every edge of the system of quads . assuming a default orientation of each edge
, its search tree should eventually contain the set of occurrences of the corresponding arc using the left - to - right order induced by the embedding .
we now traverse @xmath18 in the forward direction starting with @xmath450 $ ] and insert each traversed arc occurrence @xmath478 $ ] in its tree .
if the search tree is empty , we just insert @xmath13 at the root .
otherwise , either @xmath13 is switchable and needs to be switched to a new arc @xmath42 , or it should be inserted into a non - empty search tree . the former case can be easily detected in @xmath479 time using a dichotomy over the occurrences of @xmath42 : _ @xmath13 needs to be switched when one of these occurrences defines a crossing with @xmath13 at their common endpoint_. in the latter case , we perform the usual tree insertion : each time @xmath13 must be compared with some already inserted arc occurrence @xmath452 $ ] we can use the comparison of the occurrence preceding @xmath13 with the occurrence preceding @xmath193 ( or succeeding @xmath193 if it is a backward occurrence ) .
this takes @xmath479 time .
when @xmath480 $ ] we can use our precomputed comparisons unless @xmath13 was previously switched , thus not compared with @xmath450 $ ] during the preprocessing phase . since @xmath18 is canonical , arcs may only be switched to their left in one of four possible configurations as on figure [ fig : under_a0 ] and we infer that @xmath13 must lie to the right of @xmath450 $ ] as justified by lemma [ lem : relative - pos ] below . in conclusion , each comparison costs @xmath479 time so that @xmath13 can be inserted in its search tree in @xmath481 time .
note that after @xmath13 is inserted we do not try to determine if the current immersion has a crossing or not .
this will be checked in a second step after the algorithm is completed .
[ lem : relative - pos ] let @xmath482 and @xmath280 be two lifts of a primitive canonical geodesic in the poincar disk @xmath181 .
the limit points of @xmath482 cut the boundary circle @xmath483 into two pieces . by the _
right piece _
, we mean the piece of @xmath483 that bounds the part of @xmath484 to the right of @xmath18 .
suppose that @xmath482 contains an arc @xmath42 and that @xmath280 contains an arc @xmath485 such that @xmath42 and @xmath485 are in one of the four relative positions depicted on figure [ fig : relative - pos ] .
then one of the limit points of @xmath280 is in the right piece of @xmath483 .
we first note that @xmath485 itself lies to the right of @xmath482 .
indeed , @xmath482 would have to use a @xmath175 turn to see @xmath485 on its left or to pass along @xmath485 , in contradiction with its canonicity .
if both limit points of @xmath280 were to the left of @xmath482 , then @xmath482 and @xmath280 would form a bigon with one staircase part containing @xmath485 .
the @xmath482 side of this staircase would thus see @xmath485 on its right , which is impossible since @xmath482 is canonical ( here , we have not used that @xmath280 is canonical but only geodesic ) .
[ prop : zipper ] if @xmath486 the algorithm returns an embedding of a geodesic homotopic to @xmath18 .
let @xmath70 be the immersion computed by the algorithm .
we denote by @xmath487 the geodesic homotopic to @xmath18 resulting from the first switches in the algorithm up to the insertion of the arc occurrence @xmath488 $ ] .
note that @xmath489 implies @xmath490 .
suppose that @xmath70 has a crossing .
for the rest of this proof , we denote by @xmath282 the smallest index such that the insertion of @xmath451 $ ] creates a crossing , i.e. the restriction of @xmath70 to @xmath459 has a crossing double point while its restriction to @xmath491 is an embedding . by convention we set @xmath492 if the crossing appeared after the last arc insertion .
we shall show that @xmath493 has two lifts whose limit points are alternating on the boundary of the poincar disk .
it will follow from section [ sec : strategy ] that @xmath494 thus proving the proposition .
we first establish some preparatory claims .
[ claim : branched - crossing ] if @xmath251 is a crossing of @xmath70 with @xmath495 , then the backward arc @xmath496 $ ] and the forward arc @xmath497 $ ] are distinct from the supporting arc @xmath498 $ ] of @xmath451 $ ] . moreover ,
if @xmath451 $ ] was switched just before its insertion , then @xmath496 $ ] and @xmath497 $ ] are also distinct from the supporting arc @xmath499 $ ] of @xmath451 $ ] _ just before its switch_. the first part of the claim follows directly from our insertion procedure and the fact that the restriction of @xmath70 to @xmath491 is an embedding .
moreover , assume that @xmath451 $ ] was switched just before its insertion . by the insertion procedure
, this means that the insertion of @xmath451 $ ] before the switch would have induced a crossing @xmath500 where @xmath501 and @xmath502 , c_{i-1}[k , k-1 ] \neq c_{i-1}[i , i+1]$ ] by the first part of the claim . by the same first part we have @xmath496 , c_i[j , j+1]\neq c_i[i , i+1]$ ] .
since the restriction of @xmath70 to @xmath503 has no crossing , the length 2 path @xmath504 separates @xmath505 from @xmath499 $ ] , implying @xmath496 , c_i[j , j+1]\neq c_{i-1}[i , i+1]$ ] as desired .
[ claim : parallel - right ] if @xmath506 has a @xmath175 turn at index @xmath507 , with @xmath508 , then there is an index @xmath509 with @xmath510 such that * @xmath511=c_{k}[r , r-1]$ ] , * the arc @xmath512 $ ] lies to the right of @xmath513 $ ] , * @xmath514\neq c_{k-1}[k , k+1]$ ] .
a @xmath175 turn can only occur at the destination of an arc that has been switched and such that the next arc was not switched ( otherwise , we would get a @xmath515 turn ) .
if @xmath506 has a @xmath175 turn at @xmath507 then so has @xmath516 since the algorithm never backtracks .
it follows that the arc @xmath511 $ ] must have been switched just before its insertion , thus witnessing the existence of an arc @xmath42 of the form @xmath517 $ ] or @xmath518 $ ] , with @xmath519 , that lies parallel to and to the right of @xmath511 $ ] ( with respect to @xmath70 ) .
moreover , according to claim [ claim : branched - crossing ] we can also assume @xmath517 $ ] and @xmath518 $ ] to be distinct from the supporting arc of @xmath488 $ ] just before it was switched , namely @xmath520 $ ] .
in the case @xmath521 $ ] , we may set @xmath522 and conclude .
otherwise , @xmath523 $ ] and , recalling that the restriction of @xmath70 to @xmath524 has no crossing , it must be that @xmath525 also lies parallel to and to the right of @xmath526 .
so @xmath506 has a @xmath175 turn at index @xmath527 , and we are back to the hypothesis of the claim , decreasing the value of @xmath284 to @xmath208 .
we can repeat the same arguments inductively , each time decreasing the value of @xmath284 in the claim .
since @xmath528 , the process must stop , implying that we have reached the former case .
recall that @xmath282 is the smallest index such that the restriction of @xmath70 to @xmath459 has a crossing .
we denote by @xmath529 the geodesic homotopic to @xmath18 resulting from all the switches in the course of the algorithm execution .
let @xmath251 be a crossing of @xmath70 with @xmath495 .
we consider two lifts @xmath530 and @xmath531 of @xmath87 in the poincar disk such that @xmath532 .
we first suppose @xmath533 . [
claim : no - crossing - after - i ] we can choose @xmath283 so that @xmath534 has no crossing with @xmath531 .
( crossings are defined with respect to the lift of @xmath70 in the poincar disk . )
fix any @xmath283 such that @xmath251 is a crossing of @xmath70 ( @xmath495 ) and suppose that @xmath534 and @xmath531 cross .
let @xmath535 be the smallest positive integer such that @xmath455 and @xmath531 crosses at @xmath536 .
we thus have a bigon of the form @xmath537 for some @xmath538 .
moreover , this bigon has no intermediate crossings by the choice of @xmath535 .
let @xmath106 be a disk diagram for this bigon , oriented consistently with the system of quads .
@xmath106 must start with a staircase part by claim [ claim : branched - crossing ] .
in particular , the turn @xmath120 between @xmath539 $ ] and @xmath540 $ ] should be @xmath541 . the algorithm may have run across four possible situations at step @xmath282 . 1 .
[ enum:1 ] either @xmath451 $ ] was switchable just before its insertion but was not switched , 2 .
[ enum:2 ] or it was switchable and switched , 3 .
[ enum:3 ] or it was not switchable because of an inappropriate turn sequence , 4 .
[ enum:4 ] or it was not switchable because the part to be switched contains @xmath450 $ ] .
in the first situation , we know by remark [ rem : switchable ] that @xmath542 is followed by a turn sequence of the form @xmath543 . hence , @xmath120 is exactly @xmath544 ; but this contradicts the fact that @xmath451 $ ] was not switched though switchable .
the third situation together with remark [ rem : switchable ] also lead to a contradiction as the inappropriate turn sequence prevents @xmath545 from being part of any staircase .
thanks to claim [ claim : branched - crossing ] , the second situation equally prevents @xmath545 from being part of any staircase . it remains to consider the fourth situation .
we first suppose @xmath546 , i.e. that the diagram @xmath106 corresponds to the bigon @xmath547 . by lemma [ lem : short - bigon ]
, we have @xmath548 and the fourth situation implies that the @xmath549 side of the bigon contains @xmath450 $ ] .
these two properties imply that @xmath550 .
it ensues that the @xmath551 side contains index @xmath282 , for otherwise @xmath552 would be a crossing occurring before step @xmath282 .
however , the occurrence of @xmath282 on both sides of the bigon contradicts lemma [ lem : no - common - index ] .
hence , it must be that @xmath553 , i.e. that @xmath106 is bounded by two forward paths . since the @xmath549 side of @xmath106 contains @xmath450 $ ] and since @xmath554 , it ensues from remark [ rem:01-right - side ] that @xmath70 has a @xmath175 turn at index @xmath112 .
claim [ claim : parallel - right ] ensures the existence of a smaller index @xmath509 such that @xmath555=c_{j}[r , r-1]$ ] and @xmath556\neq c_{j-1}[j , j+1]$ ] .
since @xmath555 $ ] makes a one turn with both @xmath557 $ ] and @xmath540 $ ] , these last arcs are equal and , since @xmath558 is not a crossing , @xmath559 must be a crossing of @xmath70 .
we replace @xmath283 by @xmath509 to obtain another lift @xmath560 that crosses @xmath455 at the same point @xmath561 with the additional property that @xmath562 $ ] and @xmath457 $ ] make a one turn .
we finally observe that @xmath560 satisfies the claimed property . indeed
, the first staircase of @xmath106 must lie to the left of its @xmath549 side by remark [ rem:01-right - side ] , hence the other side of @xmath106 must start with @xmath563 $ ] .
this is however impossible by the preceding argument for @xmath546 .
it follows that @xmath534 and @xmath560 have no crossing .
we next consider the smallest positive integer @xmath509 such that @xmath455 and @xmath531 crosses at @xmath564 .
if no such @xmath509 exists , then by the above claim [ claim : no - crossing - after - i ] , @xmath531 and @xmath455 have a unique intersection point and we may conclude that their limit points alternate
. we can thus assume the existence of a bigon @xmath565 with @xmath290 and @xmath56 , and without intermediate crossings .
we first examine the case @xmath546 where the bigon @xmath566 has oppositely oriented index paths .
we must have @xmath567 for otherwise the tip @xmath568 of the bigon would define a crossing occurring before step @xmath282 , contradicting the choice of @xmath282 .
hence , @xmath569 contains @xmath283 .
this is however impossible by lemma [ lem : no - common - index ] .
we now look at the case of two forwards index paths ( @xmath553 ) .
we must have @xmath570 for otherwise the tip @xmath571 of the bigon would define a crossing occurring before step @xmath282 , again contradicting the choice of @xmath282 .
it follows that @xmath572 contains @xmath450 $ ] and that the forward branches @xmath573 and @xmath574 have a unique intersection point .
the bigon labels a disk diagram @xmath106 composed of paths and staircases as described in theorem [ th : geodesic ] . *
if @xmath450 $ ] and @xmath575 $ ] label the same arc of a path part in @xmath106 there are two possibilities : either it holds initially that @xmath576=c(\alpha)$ ] or @xmath13 was switched in the course of the algorithm . in the former case , we know by the preprocessing phase and lemma [ lem : flat - bigons ] that the left - to - right order of @xmath450 $ ] and @xmath13 is coherent with its extension in the backward direction .
this implies that ultimately in the backward direction @xmath455 lies on the same side of @xmath531 as does @xmath13 . in the latter case , as described in the end of the proof of lemma [ lem : zipper - complexity ] , we know by the insertion procedure that at least one of the limit points of @xmath455 lies on the same side of @xmath531 as does @xmath13 . since @xmath251 is the only crossing in the forward direction , we conclude in both cases that @xmath531 and @xmath455 have alternating limit points .
* otherwise , @xmath450 $ ] and @xmath13 label two distinct arcs , say @xmath577 and @xmath578 , of a staircase part @xmath579 of @xmath106 .
let @xmath580 $ ] and @xmath581 $ ] , @xmath582 be the index paths corresponding to the sides @xmath583 and @xmath584 of @xmath579 . by remark [ rem:01-right - side ] ,
@xmath585 $ ] must label the right side @xmath584 while @xmath580 $ ] can not contain @xmath450 $ ] .
it follows that @xmath586 , whence @xmath580\subset [ 1,i]$ ] .
if @xmath578 belongs to a horizontal part of @xmath583 , the first vertex in this part has a @xmath175 turn and we let @xmath587 be the index of this vertex .
otherwise , @xmath578 belongs to a vertical part whose last vertex has a @xmath175 turn .
it follows that the vertex of index @xmath588 had a @xmath175 turn at step @xmath589 of the algorithm and we set @xmath590 . by claim [ claim : parallel - right ]
, there is an arc occurrence @xmath591 $ ] , with @xmath592 , that lies to the right of @xmath578 and such that the turn at @xmath593 is not one .
we view @xmath106 as a subset of the poincar disk so that @xmath133 and @xmath134 can be seen as portions of @xmath455 and @xmath531 respectively .
let @xmath594 be the lift of @xmath87 that extends the above occurrence @xmath591 $ ] in @xmath181 .
see figure [ fig : push-1 ] .
+ we denote by @xmath595 and @xmath596 the portion of @xmath594 respectively after and before its vertex with index @xmath421 .
+ we claim that @xmath597 can not cross @xmath455 .
otherwise , we would get a pair of homotopic paths , one piece of @xmath597 starting at index @xmath421 and one piece of @xmath455 ending at index @xmath587 , with opposite orientations . by lemma [ lem : no - common - index ] ,
the @xmath597 piece would not contain index @xmath587 and would thus have length at most @xmath598 . in turn , this would imply that the @xmath455 piece would only contain vertices with indices in @xmath599 $ ] .
the crossing of @xmath597 and @xmath455 would thus occur before step @xmath282 , a contradiction . +
we next claim that @xmath600 can not cross @xmath531 . to see this
, first note that the index path corresponding to the part of @xmath600 inside @xmath106 must contain @xmath471 .
indeed , @xmath600 must cross @xmath133 or the part of @xmath134 after @xmath577 and , in any case , the index along @xmath600 of the corresponding crossing should be at least @xmath282 , since no crossing occurs before step @xmath282 .
+ it follows from the above claims that @xmath594 and @xmath455 have alternating limit points .
see figure [ fig : push-2 ] . it remains to consider the case @xmath492 where the first crossing appears after the last arc insertion .
since all the arcs have been inserted without introducing crossings it means that the crossings of the computed immersion @xmath70 must have the form @xmath601 .
we first claim that each bigon of @xmath70 must have its two sides oriented the same way .
otherwise , the tips of the bigon must have the form @xmath601 and @xmath602 for some index @xmath603 , implying that @xmath604 is contractible .
this would contradict corollary [ cor : no - monogon ] .
let @xmath605 and @xmath531 be two intersecting lifts of @xmath87 in the poincar disk . if @xmath605 and @xmath531 intersect only once , then we are done as they must have alternating limit points .
we now suppose that @xmath605 and @xmath531 intersect at least twice and we consider the bigon @xmath106 between their last two intersections in the forward direction . by lemma [ lem : short - bigon ] , the length of @xmath106 is smaller than @xmath0
. let @xmath601 and @xmath606 be the tips of @xmath106 , so that @xmath607 .
* if @xmath450 $ ] and @xmath608 $ ] label the same arc @xmath609 = \tilde{d}_j(\alpha)$ ] then , by the insertion procedure , one of the limit points of @xmath605 is on the same side of @xmath531 as does @xmath13 . the same argument as in the general case @xmath533 allows to conclude that @xmath605 and @xmath531 have alternating limit points .
* we finally suppose that @xmath450 $ ] and @xmath13 label distinct arcs , say @xmath577 and @xmath610 , of a staircase part @xmath579 of @xmath106 . as in the general case @xmath533
, @xmath577 must see @xmath579 on its left , while @xmath610 must see @xmath579 on its right .
hence , the @xmath531 side of @xmath579 is canonical while the side along @xmath605 is not and must have been switched . by claim [ claim : parallel - right ] , there is an arc occurrence @xmath193 to the right of @xmath611 $ ] with the opposite orientation .
let @xmath594 be the lift of @xmath87 that extends @xmath193 .
as in the general case @xmath533 , we can show that the part of @xmath594 after @xmath193 can not cross @xmath605 , while the part before @xmath193 can not cross @xmath531 . using that @xmath601 is the last crossing along @xmath605 and @xmath531 ,
we equally conclude that @xmath605 and @xmath594 have alternating limit points .
see figure [ fig : push-2 ] with @xmath492 .
let @xmath18 be a combinatorial curve of length @xmath0 on a combinatorial surface of size @xmath1 .
we compute its canonical form in @xmath612 time and check in linear time that @xmath18 is primitive . in the negative , we conclude that either @xmath18 is contractible , hence reduced to a vertex , or that @xmath18 has no embedding by proposition [ prop : non - primitive - formulas ] . in the affirmative ,
we apply the algorithm to compute an immersion @xmath70 of some geodesic @xmath87 homotopic to @xmath18 . according to proposition [ prop : zipper ]
, we have @xmath486 if and only if @xmath70 has no crossings .
this is easily verified in @xmath612 time by checking for each vertex @xmath66 of the system of quads that the set of paired arc occurrences with @xmath66 as middle vertex form a well - parenthesized sequence with respect to the local ordering @xmath613 induced by @xmath70 .
we conclude the proof thanks to lemma [ lem : zipper - complexity ] .
a related problem was tackled by chang _
et al . _
8.2 ) who try to find an embedding of a given closed path on a combinatorial surface . in their formulation , though , the path is not authorized to be modified . in our terminology , they only look for the existence of a combinatorial immersion without crossings . they suggest a linear time complexity for this problem and it seems likely that we could also eliminate the @xmath614 factor in our complexity .
the existence of a singular bigon claimed in theorem [ th : singular - bigon ] relies on theorem 4.2 of hass and scott @xcite .
as noted by the authors themselves this result is `` surprisingly difficult to prove '' . except for this result and the recourse to some hyperbolic geometry in the general strategy of section [ sec : strategy ] our combinatorial framework allows to provide simple algorithms and to give simple proofs of results whose known demonstrations are rather involved . concerning proposition [ prop : bigon ] ,
the existence of an immersion without bigon could be achieved in our combinatorial viewpoint by showing that if an immersion has bigons , then one of them can be swapped to reduce the number of crossings .
( the example in figure [ fig : no - singular - bigon ] shows that such a bigon need not be singular . )
what is more , if those swappable bigons could be found easily this would provide an algorithm to compute a minimally crossing immersion of two curves by iteratively swapping bigons as in section [ sec : computing - immersions ] for the case of a single curve . however , we were unable to show the existence or even an appropriate definition of a swappable bigon .
note that in the analogous approaches using reidemeister - like moves by de graff and schrijver @xcite or by paterson @xcite , the number of moves required to reach a minimal configuration is unknown .
comparatively , the number of bigon swapping would be just half the excess crossing of a given immersion . although the geometric intersection number of a combinatorial curve of length @xmath0 may be @xmath615 , it is not clear that the complexity in theorem [ th : main - result ] is optimal .
in particular , it would be interesting to see if the algorithm of section [ sec : simple - curves ] yields minimally crossing curves even with curves that are not homotopic to simple curves , thus improving theorem [ th : compute - immersion ] .
this would lead to an algorithm that is faster to find an immersion with the minimal number of crossings than to actually count them !
it is also tempting to check whether the algorithm applies to compute the geometric intersection number of two curves rather than a single curve . another intriguing question related to the computation of minimally crossing immersions comes from the fact that they are not unique . given a combinatorial immersion we can construct a continuous realization as described in the proof of lemma [ lem : geometric - vs - combinatorial ] .
say that two realizations are in the if there is an ambient isotopy of the surface where they live that brings one realization to the other .
it was shown by neumann - coto @xcite that every minimally crossing immersion is in the configuration of shortest geodesics for some riemaniann metric @xmath616 , but hass and scott @xcite gave counterexamples to the fact that we could always choose @xmath616 to be hyperbolic .
+ _ is there an algorithm to construct or detect combinatorial immersions that have a realization in the configuration of geodesics for some hyperbolic metric ? _
+ cdvl05 joan s. birman and caroline series .
an algorithm for simple curves on surfaces .
, 29(2):331342 , 1984 .
ric colin de verdire and francis lazarus . optimal system of loops on an orientable surface .
, 33(3):507 534 , march 2005 .
hsien - chih chang , jeff erickson , and chao xu . detecting weakly simple polygons . in _ proc .
acm - siam symp
. discrete algorithms _ , pages 16551670 , 2015 .
moira chas .
self - intersection numbers of length - equivalent curves on surfaces . , 23(3):271276 , 2014 .
david r.j .
simple closed curves on surfaces .
, 1(3):310314 , 1969 .
david r.j .
an algorithm for families of disjoint simple closed curves on surfaces . , 3(1):2326 , 1971 .
david r.j .
chillingworth . winding numbers on surfaces .
ii . , 199(3):131153 , 1972 . marshall cohen and martin lustig .
paths of geodesics and geometric intersection numbers : i. in _ combinatorial group theory and topology _ , volume 111 of _ ann . of math .
_ , pages 479500 .
princeton univ . press , 1987 .
moira chas and steven p lalley .
self - intersections in combinatorial topology : statistical structure .
, 188(2):429463 , 2012 .
tamal k. dey and sumanta guha .
transforming curves on surfaces .
, 58(2):297325 , 1999 .
maurits de graaf and alexander schrijver .
making curves minimally crossing by reidemeister moves .
, 70(1):134156 , 1997 .
pierre de la harpe .
topologie , thorie des groupes et problmes de dcision : clbration dun article de max dehn de 1910 .
, 125:4175 , 2010 .
jeff erickson and kim whittelsey . transforming curves on surfaces redux . in _ proc .
acm - siam symp .
discrete algorithms _ , 2013 .
benson farb and dan margalit . .
princeton university press , 2012 .
daciberg l. gonalves , elena kudryavtseva , and heiner zieschang .
an algorithm for minimal number of ( self-)intersection points of curves on surfaces . in _ proceedings of the seminar on vector and tensor analysis _ , volume 26 , pages 139167 , 2005 .
steve m. gersten and hamish b. short .
small cancellation theory and automatic groups .
, 102:305334 , 1990 .
joel hass and peter scott .
intersections of curves on surfaces . , 51(1 - 2):90120 , 1985 .
joel hass and peter scott . shortening curves on surfaces . , 33(1):2543 , 1994 . joel hass and
peter scott .
configurations of curves and geodesics on surfaces . , 2:201213 , 1999 .
donald e. knuth , james h. morris , jr , and vaughan r. pratt .
fast pattern matching in strings .
, 6(2):323350 , 1977 .
m. lothaire . .
cambridge university press , 1997 .
francis lazarus and julien rivaud . on the homotopy test on surfaces . in _ proc .
ieee symp .
foundations of computer science _ , pages 440449 , 2012 .
martin lustig .
paths of geodesics and geometric intersection numbers : ii . in _
combinatorial group theory and topology _ , volume 111 of _ ann . of math .
_ , pages 501543 .
princeton univ . press , 1987 .
maryam mirzakhani .
growth of the number of simple closed geodesies on hyperbolic surfaces . , pages 97125 , 2008 .
maryam mirzakhani .
counting mapping class group orbits on hyperbolic surfaces .
preprint http://arxiv.org/pdf/1601.03342[arxiv:1601.03342 ] , january 2016 .
bojan mohar and carsten thomassen . .
studies in the mathematical sciences .
johns hopkins university press , 2001 .
max neumann - coto .
a characterization of shortest geodesics on surfaces .
, 1:349368 , 2001 .
a combinatorial algorithm for immersed loops in surfaces .
, 123(2):205234 , 2002 .
henri poincar .
cinquime complment lanalysis situs .
, 18(1):45 110 , 1904 .
bruce l. reinhart .
algorithms for jordan curves on compact surfaces .
, pages 209222 , 1962 .
jenya sapir .
bounds on the number of non - simple closed geodesics on a surface .
preprint http://arxiv.org/pdf/1505.07171[arxiv:1505.07171 ] , may 2015 .
vladimir g. turaev .
intersections of loops in two - dimensional manifolds . , 35(2):229 , 1979 .
heiner zieschang .
algorithmen fr einfache kurven auf flchen .
, 17:1740 , 1965 .
heiner zieschang .
algorithmen fr einfache kurven auf flchen ii .
, 25:4958 , 1969 .
* appendix *
if a combinatorial immersion of one or primitive curves has excess ( self-)crossing then it contains a bigon or a monogon .
we first introduce some terminology .
an on a combinatorial curve @xmath18 consists in adding or removing a spur , or replacing in @xmath18 a possibly empty part of a facial walk by its complementary part . for a free elementary homotopy
we can also apply a circular shift to the indices of the closed curve @xmath18 . the equivalence relation generated by ( free ) homotopies
is called combinatorial ( free ) homotopy .
let @xmath617 be a cellular embedding of @xmath31 in a topological surface @xmath4 corresponding to the combinatorial surface we are working with .
every combinatorial curve @xmath18 can be realized by a continuous curve on @xmath4 by replacing each arc in @xmath18 with a continuous arc deduced from the restriction of @xmath143 to the corresponding arc .
we denote this continuous realization by @xmath90 .
it is part of the folklore that ( free ) combinatorial homotopy coincides with continuous ( free ) homotopy , meaning that @xmath18 is ( freely ) homotopic to @xmath71 if an only if @xmath90 is ( freely ) homotopic to @xmath618 .
an of a combinatorial immersion consists either in an elementary homotopy or in an , i.e. in exchanging the left - to - right order of two occurrences in a same arc where one occurrence is next to the right of the other .
we further require before performing an elementary homotopy that the immersion is in .
this means , if the elementary homotopy applies to a nonempty part @xmath208 of a facial walk of some face , that each arc occurrence in @xmath208 should be the rightmost element of its arc , i.e. the most interior to the face . when removing a spur
, we just require that the two arc occurrences to be removed are adjacent in left - to - right order .
see figure [ fig : elementary - moves ] . the elementary homotopy does not modify the order of the remaining arc occurrences .
when inserting a spur we make the inserted arc occurrences adjacent and when inserting part of a facial walk we insert each arc occurrence after the rightmost element of its arc . remark that by a sequence of adjacent transpositions we can always enforce an immersion to be in good position .
we consider the case of an immersion @xmath70 of two curves @xmath18 and @xmath71 .
suppose that @xmath70 has @xmath136 excess crossings .
we shall prove the stronger claim that @xmath70 has at least @xmath619 bigons whose tips are pairwise distinct .
let @xmath620 be an immersion of two curves @xmath87 and @xmath238 respectively homotopic to @xmath18 and @xmath71 such that @xmath620 has no excess crossings .
consider a sequence of elementary homotopies from @xmath621 to @xmath72 . following the above remark
, we can insert adjacent transpositions between each elementary homotopy in order to obtain a sequence of elementary moves from @xmath620 to @xmath70 . see the above remark .
the claim is trivially verified by @xmath620 .
we now check that the claim remains true after each elementary move . if the move is an adjacent transposition we may assume that we exchange an arc occurrence of @xmath18 with an arc occurrence of @xmath71 since we only consider intersections between @xmath18 and @xmath71 . without loss of generality
we also assume exchanging the forward occurrences @xmath622_c$ ] and @xmath623_d$ ] .
there are three cases to consider . 1 .
if none of @xmath77 and @xmath624 is a crossing then the transposition adds two crossings that we may pair as the tips of a new bigon .
we now have @xmath625 excess crossings with at least @xmath626 bigons as required .
[ it : one - of - two ] if @xmath77 is a crossing but @xmath624 is not and if @xmath77 is paired to form the tips of a bigon , say @xmath627 , we may just replace this bigon by @xmath628 sliding its tip @xmath77 to @xmath624 .
a similar procedure applies when @xmath624 is a crossing but @xmath77 is not . in each case
the number of excess crossings and bigons is left unchanged .
[ it : two - of - two ] it remains the case where both @xmath77 and @xmath624 are crossings .
if none of the two is paired to form the tips of a bigon , then the transposition removes two crossings and no pairing , so that the claim remains trivially true . if exactly one of the two is paired or if the two are paired together , we loose one bigon and two crossings after the transposition so that the claim remains true . otherwise , there are two bigons of the form @xmath627 and @xmath629 that we can recombine to form the bigon @xmath630
. we again have one less bigon and two less crossings .
we now consider the application of an elementary homotopy as described before the proof .
there are again three possibilities . 1 .
if the homotopy replaces a nonempty subpath @xmath208 of a facial walk @xmath631 by the nonempty complementary part @xmath66 then no crossing may appear or disappear as we assume the immersion in good position .
in particular , no crossing may use an internal vertex of @xmath208 and @xmath208 is either entirely included in or excluded from any side of any bigon .
we can replace @xmath208 by @xmath66 in any bigon side where @xmath208 occurs to obtain valid bigons in the new immersion after the elementary homotopy is applied .
the number of excess crossings and bigons is left unchanged .
2 . when @xmath208 is empty in the above replacement , or when inserting a spur , we may only add crossings by pairs forming bigons with one zero - length side and the complement of @xmath208 or the spur as the other side .
a similar analysis as in case of a transposition applies to take care of each pair .
3 . when @xmath66 is empty in the above replacement , or when removing a spur , we may only remove crossings by pairs and a similar analysis as in case of a transposition applies to take care of each pair .
this ends the proof for the case of two curves .
a similar proof holds for the existence of a bigon or a monogon in an immersion with excess self - intersection .
this time the excess crossings are either paired to form bigons or left alone as the tips of monogons .
swapping the two sides of a singular bigon of an immersion of a geodesic primitive curve decreases its number of crossings by at least two .
consider a singular bigon @xmath632 of an immersion @xmath70 of a closed primitive canonical curve @xmath18 .
let @xmath620 be the immersion after the bigon has been swapped .
we shall partition the set of potential double points and show that the net change of the number of crossings with respect to @xmath70 and @xmath620 is non positive in each part .
as the tips of the bigon are not crossings in @xmath620 , this will prove the lemma .
we set for @xmath633 and @xmath634 : @xmath635 when @xmath636 or @xmath340 , the set @xmath637 consists of three double points only ( or zero if @xmath638 ) .
note that we can not have both equalities as this would imply @xmath639 , whence @xmath235 and @xmath18 would be a square , in contradiction with the hypothesis that @xmath18 is primitive .
for each double point in @xmath640 the four incident arc occurrences are left in place in @xmath70 and @xmath620 .
it ensues that @xmath640 has the same crossings in @xmath70 and @xmath620 .
the double points in @xmath641 are the tips of the bigon .
they are crossings in @xmath70 and not in @xmath620 , whence a net change of @xmath124 crossings . for each of @xmath642 or
@xmath643 , the first double point in their above definition is a crossing in @xmath70 ( resp .
@xmath620 ) if and only if the second one is a crossing in @xmath620 ( resp .
@xmath70 ) .
their net change of crossings is thus null . for @xmath644 , @xmath645 , @xmath646 and @xmath647
there are a few case analysis depending on the relative ordering of the two arc occurrences incident to @xmath648 ( resp .
@xmath649 , @xmath650 ) with respect to the crossing @xmath77 ( resp . @xmath242 . in each case
( see figure [ fig : dx1 ] ) the number of crossings can not increase from @xmath70 to @xmath620 . , @xmath645 , @xmath646 or @xmath647 has four distinct possible configurations .
the blue and red strands represent the crossing @xmath77 ( resp .
@xmath242 ) and each arrow links the left configuration before the bigon swap ( in @xmath70 ) with the right configuration after the swap is applied ( in @xmath620 ) . ] for @xmath637 there are three cases according to whether @xmath651 , @xmath652 , or none of the two identifications occurs .
recall from the paragraph before the lemma that we can not have both identifications .
the first two cases can be treated the same way .
see figure [ fig : d2special ] . ,
the end of @xmath334 overlaps with the beginning of @xmath653 ( in the purple strand ) .
the third configuration on the right is actually forbidden by the definition of a singular bigon !
it corresponds to the non - singular bigon on figure [ fig : singular - def ] . ]
the number of possible configurations is much larger in the last case , i.e. when @xmath654 and @xmath655 .
we essentially have to shuffle two circular orderings of length four corresponding to the two crossing tips . without loss of generality
we can assume that the path from @xmath47 to @xmath656 is the right side of the bigon .
we consider the following arc occurrences : @xmath657 since @xmath77 and @xmath242 are crossings we must see @xmath658 in this counterclockwise order around the vertex @xmath659 and similarly for @xmath660 .
we denote by @xmath661 and @xmath662 , respectively , these two circular sequences .
we need to consider the effect of the bigon swapping on all the possible shuffles of @xmath661 and @xmath662 .
the restriction of these shuffles to @xmath663 and @xmath664 gives the 6 possible shuffles of @xmath665 and @xmath666 . among them
the order @xmath667 can not occur .
indeed , since the bigon is a thick double path the arcs @xmath668 and @xmath669 either coincide or form a corner of a quad .
this would force @xmath670 and @xmath671 to lie in a similar configuration .
in turn , the constrained order @xmath662 would also enforce @xmath672 and @xmath670 to coincide or form a corner of a quad in contradiction with the hypothesis that @xmath18 has no spurs or brackets .
similar arguments show that the orders @xmath673 , @xmath674 , @xmath675 and @xmath676 can only occur as factors in the possible shuffles of @xmath661 and @xmath662 .
these 4 orders thus leads to 24 distinct shuffles of @xmath661 and @xmath662 by factoring with the 6 shuffles of @xmath677 with @xmath678 . by exchanging the roles of @xmath665 and @xmath666 and by turning clockwise instead of
counterclockwise we see that @xmath674 leads to the same orders as @xmath673 and a similar correspondence holds for @xmath675 and @xmath676 .
we thus only need to check the 12 configurations depicted on figure [ fig : d2quater ] .
on @xmath637 for all the orders including the factor @xmath673 or @xmath675 .
the blue strand represents @xmath679 , the red one @xmath680 , the orange one @xmath681 and the green one @xmath682 . ]
it remains to consider the order @xmath683 .
because of the constrained order @xmath662 , @xmath684 and @xmath685 can not lie between @xmath22 and @xmath664 .
this leaves out @xmath686 possible shuffles of @xmath678 and @xmath683 : @xmath687 we finally shuffle each of these orders with @xmath688 . since @xmath689 and @xmath690 can not lie between @xmath13 and @xmath193 , we obtain @xmath691 possible shuffles when considering either @xmath692 or @xmath693 , @xmath694 possible shuffles with @xmath695 or @xmath696 , and @xmath686 possible shuffles with @xmath697 . by swapping left and right and turning clockwise instead of counterclockwise , we remark that @xmath698 and @xmath693 lead to the same orders and similarly for @xmath695 and @xmath696 .
we thus only need to consider the shuffles of @xmath688 and @xmath699 or @xmath697 to complete the inspection of all the cases .
those shuffles are represented on figures [ fig : d2-s1 ] , [ fig : d2-s2 ] , [ fig : d2-s4 ] , [ fig : d2-s6 ] respectively . in each of the configurations
, we trivially check that the number of crossings is not increasing .
this allows to conclude the lemma when the two index paths of a singular bigon are directed the same way .
a similar analysis can be made when their directions are opposite , that is when the considered bigon has the form @xmath700 .
_ claim : let @xmath18 be a primitive geodesic of length four and let @xmath284 be a positive integer .
then @xmath701 and @xmath702 can not bound a partial diagram where the index pairs @xmath703 and @xmath704 are in the same configuration . _ consider a partial diagram @xmath106 for the thick double path @xmath705 and suppose that the index pairs @xmath703 and @xmath704 are in the same configuration . in order to reach a contradiction
we will use the fact that @xmath706 label the first arcs of the @xmath707 side of @xmath230 as well as the last arcs of the @xmath708 side , and similarly for @xmath709 .
if the first and last index pairs of @xmath230 correspond to the same vertex in @xmath106 , then there are five possible configurations for the first two arcs on each side of @xmath230 ( depicted on figure [ fig : length - partial - diag ] , b1 - 5 ) and five possible configurations for the last two arcs on each side of @xmath230 ( depicted on figure [ fig : length - partial - diag ] , e1 - 5 ) .
clearly , @xmath106 can not start with a path part of length two ( case b1 ) , as this would imply @xmath710 and in turn @xmath711 would be a square ; this would contradict the fact that @xmath18 is primitive .
case b2 leads to the same contradiction noting that @xmath712 .
cases e1 and e2 can be dealt with analogously . as in lemma
[ lem : ell - plus - un ] , we denote by @xmath713 the conjunction of configurations @xmath421 and @xmath714 . *
if the orientations of the first and last staircases in @xmath106 coincide , we easily obtain that @xmath715 has degree 2 for the conjunctions @xmath716 , @xmath717 has degree 2 for @xmath718 , @xmath715 has degree 3 for @xmath719 and @xmath720 and @xmath715 has degree 4 for @xmath721 , @xmath722 , @xmath723 and @xmath724 .
in case @xmath725 a quad has two sides labelled by the same arc , which is forbidden by lemma [ lem : facial - walk ] . *
if the orientations of the first and last staircases in @xmath106 are opposite then each conjunction of @xmath726 with @xmath727 implies that a quad has two sides labelled by the same arc , which is again forbidden .
when the first and last index pairs of @xmath230 correspond to diagonally opposite vertices in a quad , we get eight possible configurations for the first two arcs on each side of @xmath230 ( depicted on figure [ fig : length - partial - diag ] , b1 - 8 ) and eight possible configurations for the last two arcs on each side of @xmath230 ( depicted on figure [ fig : length - partial - diag ] , e1 - 8 ) .
all the pink quads in the figure denote the same quad with the same orientation , recalling that the configuration for the first and last index pairs of @xmath230 is the same .
we easily get in each case that a vertex has degree at most 5 , or that a quad has two sides labelled by the same arc , or that @xmath18 is a square ( case @xmath728 ) . in each case
, we have thus reached a contradiction . | the geometric intersection number of a curve on a surface is the minimal number of self - intersections of any homotopic curve , i.e. of any curve obtained by continuous deformation .
likewise , the geometric intersection number of a pair of curves is the minimal number of intersections of any homotopic pair . given two curves represented by closed walks of length at most @xmath0 on a combinatorial surface of complexity @xmath1 we describe simple algorithms to compute the geometric intersection number of each curve or of the two curves in @xmath2 time .
we also propose an algorithm of complexity @xmath3 to decide if the geometric intersection number of a curve is zero , i.e. if the curve is homotopic to a simple curve . |
-5 mm the nervous system creates many different rhythms , each associated with a range of behaviors and cognitive states .
the rhythms were first discovered from scalp recordings of humans , and the names by which they are known still come mainly from the electroencephalograph ( eeg ) literature , which pays attention to the frequency and behavioral context of those rhythms , but not to their mechanistic origins .
the rhythmic patterns include the alpha ( 9 - 11 hz ) , beta ( 12 - 30 hz ) , gamma ( 30 - 80 hz ) , theta ( 4 - 8 hz ) , delta ( 2 - 4 ) hz and slow wave ( .5 - 2 hz ) rhythms .
the boundaries of these ranges are rough .
more will be said about the circumstances in which some of these rhythms are displayed .
it is now possible to get far more information about the mechanisms behind the dynamics of the nervous system from other techniques , including electrophysiology .
the revolutions in experimental techniques , data acquisition and analysis , and fast computation have opened up a broad and deep avenue for mathematical analysis .
the general question addressed by those interested in rhythms is : how does the brain make use of these rhythms in sensory processing , sensory - motor coordination and cognition ?
the mathematical strategy , to be discussed below , is to investigate the `` dynamical structure '' of the different rhythms to get clues to function .
most of this talk is about dynamical structure , and the mathematical issues surrounding its investigation .
i ll return at the end to issues of function .
-5 mm the mathematical framework for the study of brain dynamics are the hodgkin - huxley ( hh ) equations .
these are partial differential equations describing the propagation of action potentials in neurons ( cells of the nervous system ) .
the equations , which play the same role in neural dynamics that navier - stokes does in fluid dynamics , are an elaborate analogy to a distributed electrical circuit . the central equation
, @xmath0 describes conservation of current across a piece of a cell membrane ; @xmath1 is the cross - membrane voltage and the left hand side is the capacitive current .
the first sum on the right hand side represents the intrinsically generated ionic currents across the membrane .
the term @xmath2 represents the spatial diffusion , and @xmath3 the current fed into the cell .
@xmath4 represents the currents introduced by coupling from other cells .
thus , these equations can also be used to model networks of interacting neurons , the focus of this talk .
each of the intrinsic currents @xmath5 is described by ohm s law : @xmath5 is electromotive force divided by resistance . in this context
, one usually uses the concept of `` conductance '' , which is the reciprocal of resistance .
the electromotive force depends on the type of charged ion ( e.g. , na , k , ca , cl ) and the voltage of the cell ; it has the form @xmath6 $ ] , where @xmath7 is the so - called `` reversal potential '' of that ion .
the dynamics of the conductances are what make the equations so mathematically interesting and rich . in a simple description of the conductance ,
each ionic current has up to two `` gates '' , which open or close at rates that are dependent on the voltage of the cell . for each such gate
, there is then a first order equation of the form @xmath8/\tau_{x}(v)\ ] ] where x denotes the fraction of channels of that type that are open at any given time , @xmath9 is the steady state value of @xmath10 for a fixed voltage , and @xmath11 is the rate constant of that gating variable . for example , the standard na current has the form @xmath12 $ ] , where @xmath13 is the maximal conductance , and @xmath14 and @xmath15 are gating variables satisfying the above equation for @xmath10 .
the dynamics for @xmath14 and @xmath15 differ because @xmath16 is an increasing function of @xmath1 , while @xmath17 is a decreasing function ; also @xmath18 is much smaller than @xmath19 .
the ( chemical ) synaptic currents have the same form as the intrinsic ones , with the difference that the dependence of the driving force on voltage uses that of the post - synaptic cell , while the conductance depends on the pre - synaptic voltage .
that is , @xmath20 has the form @xmath21,$ ] where @xmath22satisfies the equation for x above , with @xmath1 replaced by @xmath23 , the voltage of the cell sending the signal , and @xmath24 is the reversal potential of the synapse . the coupling is said to be excitatory if the current is inward ( increases voltage toward the threshold for firing an action potential ) or inhibitory if the current is outward ( moves voltage away from threshold for firing . ) for a simple version of the hh equations , there are three ionic currents ; one of these ( na ) creates an inward current leading to an action potential , one ( k ) an outward current helping to end the action potential , and a leak current ( mainly cl ) with no gating variable .
the hh equations are not one single set of equations , but a general ( and generalizable ) form for a family of equations , corresponding to different sets of intrinsic currents ( which can depend on position on the neuron ) , different neuron geometries , and different networks created by interactions of neurons , which may themselves be highly inhomogeneous .
numerical computation has become highly important for observing the behavior of these equations , but does not suffice to understand the behavior , especially to get insight into what the specific ionic currents contribute ; this is where the analysis , including simplification , comes in .
for an introduction to hh equations , some analysis and some of its uses in models , see [ 1 ] .
-5 mm it is not possible to analyze the full class of equations in all generality .
our strategy is to look for mathematical structures underlying some classes of behavior observed experimentally ; the emphasis is on the role of dynamical systems , as opposed to statistics , though probabilistic ideas enter the analysis .
our central scientific question here is how rhythms emerge from the `` wetware '' , as modeled by the hh equations .
as we will see , different rhythms can be based on different sets of intrinsic currents , different classes of neurons , and different ways of hooking up those cells .
there are some behaviors we can see by looking at small networks , and others that do not appear until the networks are large and somewhat heterogeneous . even in the small networks ,
there are a multiplicity of different building blocks for the rhythms , with excitation and inhibition playing different roles .
noise appears , and plays different roles from heterogeneity .
investigators often use simplifications of the hh equations .
for example , this talk deals only with `` space clamped '' cells in which the spatial distribution of each cell is ignored , and the equations become odes .
( there are circumstances under which this can be a bad approximation , as in [ 2 ] ) . under some circumstances ,
the 4-dimensional simplest space - clamped hh equations ( one current equation , three gating variables ) can be reduced to a one - dimensional equation ; thus , networks of neurons can be described by a fraction of the equations that one needs for the full hh network equations .
another kind of reduction replaces the full hh odes by maps that follow the times of the spikes . in both cases ,
there are at least heuristic explanations for why these reductions are often very successful , and hints about how and why the simplifications can be expected to break down .
-5 mm -5 mm some kinds of cells coupled by inhibition like to form rhythms and synchronize [ 3 - 5 ] .
this is unintuitive , because inhibition to cells can temporarily keep the latter from firing ( see below for important exceptions ) , but mutual inhibition can encourage cells to fire simultaneously .
there are various ways to see this , with methods that are valid in different contexts . for weak coupling
, it can be shown rigorously that the full equations reduce to interactions between phases of the oscillators [ 6 ] ; the particular coupling associated with inhibition can then be shown to be synchronizing ( though over many cycles ) [ 7 ] .
if the equations can be reduced to one - dimensional `` integrate and fire '' models , one can use `` spike - response methods '' to see the synchronizing effect of inhibitory synapses on timing of spikes . both of these are described in [ 6 ] along with more references .
another method , which i believe is most intuitive , looks at the ongoing effect of forced inhibition on the voltage of the cells , and how some of the processes are `` slaved '' to others .
this is seen most clearly in the context of another one - dimensional reduction that has become known as the `` theta '' model , because of the symbols used for the phase of the oscillations [ 8 ] .
the reduced equations have been shown to be a canonical reduction of equations that are near a saddle - node bifurcation on an invariant circle ( limit cycle ) .
many versions of hh - like models ( and some kinds of real neurons ) have this property for parameter values near onset of periodic spiking , and they are known as `` type 1 '' neurons .
the `` theta model '' has the form @xmath25 here the equation for the phase @xmath26 has periodic solutions if the parameter @xmath27 is positive , and two fixed points ( stable , saddle ) if @xmath27 is negative .
to understand the effects of forced inhibition , we replace @xmath27 by a time dependent inhibition given by @xmath28 , where @xmath29 for @xmath30 and zero otherwise . with the change of variables @xmath31
, this is a 2-d autonomous system .
figures [ 9 ] and analysis show that the system has two special orbits , known in the non - standard analysis literature as `` rivers '' [ 10 ] , and that almost all of the trajectories feed quickly into one of these , and are repelled from the other .
the essential effect is that initial conditions become irrelevant to the outcome of the trajectories . a similar effect works for mutually coupled systems of inhibitory neurons .
the rhythm formed in this way is highly dependent on the time scale of decay of the inhibition for the frequency of the network [ 11 , 12 ] .
these models , and the `` fast - firing '' inhibitory cells that they represent , can display a large range of frequencies depending on the bias ( @xmath3 in hh , the parameter @xmath27 in the theta model ) ; however , in the presence of a small amount of heterogeneity in parameters , the rhythm falls apart unless the frequency is in the gamma range ( 30 - 80 hz ) [ 4 , 13 ] .
this can be understood from spike response methods or in terms of rivers .
the above rhythm is known as ing or interneuron gamma [ 14 , 15 ] . a variation on this uses networks with fast - firing inhibitory cells ( interneurons or i cells ) and excitatory cells ( pyramidal cells or e - cells ) .
this is called ping ( pyramidal interneuron gamma ) [ 14 , 15 ] .
heuristically , it is easy to understand the rhythm : the inhibitory cells are set so they do not fire without input from the e - cells . when the e - cells fire , they cause the i - cells to cross firing threshold and inhibit the e - cells , which fire again when the inhibition wears off . this simple mechanism becomes much more subtle when there is heterogeneity and noise in large networks , which will be discussed later .
-5 mm the fast - firing cells described above are modeled using only the ionic currents needed to create a spike .
most other neurons have channels to express many other ionic currents as well , with channel kinetics that range over a large span of time constants .
these different currents change the dynamical behavior of the cells , and allow such cells to be `` type ii '' , which means that the onset of rhythmic spiking as bias is changed is accompanied by a hopf bifurcation instead of a saddle node .
the type of onset has important consequences for the ability of pair of such cells to synchronize .
e.g. , models of the fast - firing neuron , if connected by excitatory synapses , do not synchronize , as can be shown from weak coupling or other methods described above ( e.g. , [ 7 ] )
. however , if the cells are type ii , they do synchronize stably with excitation ( and not with inhibition ) .
this was shown by gutkin and ermentrout using weak coupling methods [ 16 ] .
a more specific case study was done by acker et al . [
17 ] , motivated by neurons in the part of the cortex that constitutes the input - output pathways to the hippocampus , a structure of the brain important to learning and recall .
these cells are excitatory and of type ii ( j.white , in prep . ) ; models of these cells , based on knowledge of the currents that they express , do synchronize with excitatory synapses , and do not with inhibitory synapses . the synchronization properties of the such cells can be understood from spike - timing functions and maps [ 17 ] . given the hh equations for the cell , one can introduce at any time in the cycle excitation or inhibition whose time course is similar to what the synapse would provide . from this , one can compute how much the next spike is advanced or delayed by this synapse .
from such a graph , one can compute a spike - time map which takes the difference in spike times in a single cycle to the difference in the next cycle .
the analysis of such a map is easy , but the process raises deeper mathematical issues .
one set of issues concerns what is happening at the biophysical level that gives rise to the type ii bifurcation , which is associated with a particular shape of the spike advance function [ 18 ] .
analysis shows that the type ii is associated with slow outward currents or certain slow inward currents that ( paradoxically ) turn on when the cell is inhibited [ 16 , 17 ] ; this shows how biophysical structure is connected with mathematical structure . a second set of questions concerns why the high - dimensional coupled hh equations can be well approximated by a 1-d map .
( in some parameter ranges , but not all , this is any excellent approximation ) .
the mathematical issues here concern how large subsets of high - dimensional phase space collapse onto what is essentially a one - dimensional space .
ideas similar to those in section 3.1 are relevant , but with different biophysics creating the collapse of the trajectories . in this case
( and others ) there are many different ionic currents , with many different time scales , so that a given current can be dominant in some portion of the trajectory and then decrease to zero while others take over ; this leads to structure that is more complex than that of the traditional `` fast - slow '' equations , and which is not nearly as understood . such reductions to 1-d maps
have been used in other investigations of synchrony [ 19 - 21 ] involving multiple cells and multiple kinds of currents .
-5 mm so far , i ve talked about networks containing fast - firing neurons ( inhibitory ) or excitatory cells .
but there are many different kinds of cells in the nervous system , with intrinsic and synaptic currents that make them dynamically very different from one another .
once there are more currents with more time scales , it is easier to create more rhythms with different frequency .
that is , the differences in frequencies often ( but not always ) come from some time scales in the interacting currents , and can not be scaled away .
the stellate cell of section 3.2 is an excellent example of currents creating frequencies ; in a wide range of parameters , these cells , even without coupling , form a theta rhythm .
indeed , they are believed to be one of the primary sources of that rhythm in the hippocampus , which is thought by many to use these rhythms in tasks involving learning and recall .
as described above , these cells are excitatory , and synchronize when coupled by excitation .
more puzzling are inhibitory cells in the hippocampus that are capable of forming theta rhythms as isolated cells with ionic currents similar to those in the stellate cells .
the puzzle is that these cells do not cohere ( in models ) using inhibitory coupling .
( the decay time of inhibition caused by these cells is roughly four times longer than the inhibition caused by the fast - firing cells , but neither fast nor this slower decaying inhibition creates synchrony in models . ) so what is providing the coherence seen in the theta rhythm ? ( the rhythm can be seen in small slices that do not have inputs from other parts of the brain producing theta , so in such a paradigm , the rhythm must be produced locally . )
one suggestion ( rotstein , kopell , whittington , in preparation ) is that the inhibitory rhythms seen in slice preparations with excitation blocked pharmacologically depend on both kinds of inhibitory cells discussed , the special ones ( called o - lm cells [ 22 ] ) and the others .
simulations show that networks of these cells can have the o - lm cells synchronize and i - cells synchronize at a different phase , to create an inhibitory network with considerably more complexity than interacting fast - firing cells involved in ing . again
, this can be reduced to a low - dimensional map for a minimal network ( two o - lm cells , one fast firing i - cell ) .
however , the reduction now requires properties of the currents involved in the o - lm model , including the kinetics of the gating variables .
-5 mm another set of mathematical issues is associated with transitions among rhythms . in general ,
rhythms slower than gamma ( e.g. , beta , theta and alpha ) make use of ionic currents that are active between spikes .
these currents are voltage - dependent , so that changes in voltage , in the sub- and super - threshold regimes , can turn on or off these currents .
thus , neuromodulators that change the voltage range of a neuron ( e.g. , by changing a leak current ) can change which other currents are actively expressed . in that way
, they can cause a switch from one rhythm to another .
for example , models of the alpha rhythm [ 20 ] suggest that this rhythm makes use the inhibition - activated `` h - current '' ; this current is effectively off line if the voltage is increased ( even below threshold level ) .
thus , a switch from alpha to a faster rhythm ( gamma or beta ) can be effected by simply making the e - cells operate in a moderately higher voltage regime . these switches can be seen in simulations ( pinto , jones , kaper , kopell , in prep . ) , but are still understood only heuristically .
the mathematical issues are associated with reduction of dimension methods . in the regime
in which the network is displaying alpha , there are many more variables that are actively changing , notably the gating variables of each of the currents that is important in this rhythm .
when there is a switch to gamma and those currents go off line , the phase space becomes effectively smaller .
the mathematics here involves understanding how that phase compression takes place . a related set of mathematical questions concerns rhythms that are `` nested '' , one within another .
for example , the theta rhythm often presents as the envelope of a series of faster gamma cycles , and the beta rhythm , at least in some manifestations , occur with the i - cells firing at a gamma rhythm and the e - cells firing at the slower beta rhythm , missing some cycles of the inhibitory rhythm .
the gamma / beta switch has been understood from a physiological point of view ( see [ 19 ] and its references ) and has been simulated .
the gamma / theta nesting is less understood , though new data and simulations are providing the physiological and heuristic basis for this [ 22 ; rotstein , kopell , whittington , in prep . ] .
-5 mm though there are many more examples of other building blocks , i m turning to issues that do not appear in small network analysis .
i m going to go back to a very simple building block , but now put many such together .
the simple building block is one e cell , one i - cell , which together can create a gamma rhythm .
-5 mm we now consider a network with n e - cells and m i - cells , with random coupling from the e - cells to the i - cells and vica versa .
suppose , for example , there is a fixed probability of connection in each direction between any pair of e and i cells .
then the number of inputs to any cell is distributed across the population , leading to heterogeneity of excitation and inhibition .
is it still possible to get coherent gamma rhythms ?
this can be answered with mathematical analysis using the `` theta neuron '' model described above [ 9 ] . to understand synchrony in e / i networks , it is helpful to understand what each pulse of inhibition does to the population of excitatory cells and vica versa .
the part in which both probability and dynamical systems play a large role is the effect of a pulse of inhibition on a population .
the `` rivers '' referred to above in section 3.1 create synchronization if the inputs to cells have no variance , but with variation in the size of the inputs , there is a spread in the times of the outputs .
this can be accurately computed using features of the dynamics and probability theory .
similarly , but with less accuracy , one can compute the the effect of variation of inputs on the spike times of the receiving population due to a pulse of excitation .
the results lead to unintutive conclusions , e.g. , that increasing the strength of the inhibition ( which strengthens the synchronizing effect of the rivers ) does not reduce the desynchronizing effects of random connectivity .
furthermore , tight synchrony can be obtained even with extremely sparse coupling provided that variance in the size of the inputs is small .
-5 mm the above analyses can be put together to understand synchrony of `` ping '' . however , they leave only partially answered many questions about larger networks .
one such question , which is central to understanding how the assemblies of neurons are created and destroyed , is the circumstances under which the synchrony falls apart , i.e. , what modulations of cells and/or synapses will lead to loss of coherence of the gamma rhythm .
the above analysis shows that too large a variation in size of inputs to different cells of the same population can be fatal .
similar phenomena occur with too much variation in drive or intrinsic currents .
there are less obvious constraints that are understood from working with smaller networks described above . from those ,
it is possible to see that ing and ping operate in different parameter regimes : the firing times of the population in ing are governed by the bias of the i - cells ( as well as the decay time of the inhibition ) ; in ping , the inhibitory cells are more passive until driven by the e - cells , and the timing comes from bias of the e - cells ( as well as decay of inhibition ) .
this means that the mechanism of coherence can switch between ing and ping by changing relative excitability of the two populations .
changing the strengths of the i - e and e - i synapses can also get the population ( large or small ) out of the regime in which the e - cells synchronize the i - cells , and vica versa .
a more mysterious issue that can not be addressed within minimal networks is how the size of the sub - populations responding on a given cycle affects the coherence on the next cycle and the numbers of cells participating , especially when there is some heterogeneity in the network .
e.g. , as the number of inhibitory neurons firing in a cycle changes , it changes the total inhibition to the e - cells , which changes the number of e - cells that are ready to fire when inhibition wears off , and before the next bout of inhibition .
if the amount of inhibition gets too small , or inhibition gets too dispersed , the coherence can rapidly die . without taking into account the trajectories of each of the large number of cells , it is likely that some possibly probabilistic account of the numbers of cells spiking per cycle can give some insight into the dynamical mechanisms surrounding failure of coherence .
such a reduction has been successfully used in a different setting , involving the long - distance coherence of two populations of heterogeneous cells . in this case , if the populations are each minimal ( one e / i pair ) for each site , there is 1-d map that describes the synchronization , with the variable the timing between the e and i sites [ 19 ] . for large and heterogeneous networks , the synchronization ( within some parameter regimes ) can be described by a 3-dimensional map , in which the first variable is the time between the first spikes of a cycle in the two e - cell populations , and the others are the fraction of i - cells firing on that cycle in each of the two i - cell populations ( mcmillen and kopell , in prep . ) .
related work has been done from a different perspective , starting with asynchronous networks and asking how the asynchrony can lose stability [ 23 - 25 ] .
work using multiple time scales to address the formation of `` clusters '' when synchrony fails is in [ 26 ] .
-5 mm one of the main differences between ing and ping is the difference in robustness .
small amounts of heterogeneity of any kind make ing coherence fall apart dramatically [ 4,13 ] . by contrast , ping is tolerant to large ranges of heterogeneity .
the `` ping - pong '' mechanism of ping is also able to produce frequencies that cover a much wider range than the ing mechanism , which is constrained by loss of coherence to lie in the gamma range of approximately 30 - 80 hz [ 4,13 ] .
since many versions of gamma seen in experiments are of the ping variety , this raises the question of what constrains the ping rhythms to stay in the gamma frequency range .
a possible answer to this comes from simulations .
c. borgers , d. mcmillen and i found that heterogeneity , unless extreme , would not disrupt the ping coherence .
however , a very small amount of noise ( with fixed amplitude and poisson - distributed times ) could entirely destroy coherence of the ping , provided the latter had a frequency below approximately 30 hz ; if the same noise is introduced when the network is in the gamma range , the behavior is only slightly perturbed .
furthermore , the ability to withstand the noise is related to adding some i - i connections , as in ing .
a heuristic explanation is that , at low frequencies , the inhibition to the i - cells ( which has a time constant around 10ms ) wears off before the excitation from the e - cells causes these cells to spike .
thus , those cells hang around the threshold for significant amounts of time , and are therefore vulnerable to being pushed over threshold by noise .
the mathematics has yet to be understood rigorously .
-5 mm the mathematical questions are themselves interesting , but the full richness of the scientific endeavor comes from the potential for understanding how the rhythms generated by the brain might be used in sensory processing , motor coordination and cognition .
we are still at the outer edges of such an investigation , but there are many clues from animal behavior , physiology and mathematics .
work done with eegs ( see , e.g. , reviews [ 27,28 ] ) has shown that many cognitive and motor tasks are associated with specific rhythms appearing in different parts of the tasks .
gamma is often associated with attention , awareness and perception , beta with preparation for motor activies and high - order cognitive tasks , theta with learning and recall and alpha with quiet awareness ( there are several different versions of alpha in different parts of the brain and found in different circumstances ) .
work done in whole animals and in slice preparations are giving clues to the underlying physiology of the rhythms , and how various neuromodulators change the rhythms , e.g. , [ 14 ] .
much of the math done so far has concerned how the networks produce their rhythms from their ionic currents and connectivity , and has not directly addressed function .
however , the issues of function are starting to be addressed in terms of how the dynamics of networks affects the computational properties of the latter .
one of the potential functions for these rhythms is the creation of `` cell assemblies '' , temporary sets of neurons that fire synchronously .
these assemblies are believed to be important in distributed processing ; they enhance the effect of the synchronized pulses downstream , and provide a substrate for changes in synapses that help to encode experience .
( `` cells that fire together wire together . '' ) simulations show , and help to explain , why gamma rhythms have especially good properties for creating cells assemblies , and repressing cells with lower excitatbility or input [ 29 ] .
furthermore , the changes in synapses known to occur during gamma can facilitate the creation of the beta rhythm ( see [ 19 ] for references ) , which appears in higher - order processing .
mathematical analysis shows that the beta rhythm is more effective for creating synchrony over distances where the conduction time is longer .
thus , we can understand the spontaneous gamma - beta switch seen in various circumstances ( see [ 19 ] and [ 29 ] ) as creating cell assemblies ( during the gamma portion ) , using the synaptic changes to get cell assemblies encoded in the beta rhythm , and then using the beta rhythm to form highly distributed cell assemblies .
the new flood of data , plus the new insights from the mathematics , are opening up many avenues for mathematical research related to rhythms and function . a large class of such questions concerns how networks that are displaying given rhythms filter inputs with spatio - temporal structure , and how this affects the changing cell assemblies .
this question is closely related to the central and controversial questions of what is the neural code and how does it operate .
these questions will likely require new techniques to combine dynamical systems and probability , new ways to reduce huge networks to ones amenable to analysis , and new ideas within dynamical systems itself , e.g. , to understand switches as global bifurcations ; these are large and exciting challenges to the mathematical community .
n. kopell and g.b .
ermentrout , mechanisms of phase - locking and frequency control in pairs of coupled neural oscillators , in _
handbook on dynamical systems _ ,
_ toward applications _ .
b. fiedler , elsevier ( 2002 ) , 354 .
whittington , r.d .
traub , n. kopell , g.b .
ermentrout and e.h .
buhl , inhibition - based rhythms : experimental and mathematical observation on network dynamics , _ int .
j. of psychophysiology _ * 38 * ( 2000 ) , 315336 . m.j .
gillies , r.d .
traub , f.e.n lebeau , c.h .
davies , t. gloveli , e.h .
buhl and m.a .
whittington , stratum oriens interneurons temporally coordinate atropine - resistant theta oscillations in hippocampal area ca1 , to appear in _
j. neurophysiol . | the nervous system displays a variety of rhythms in both waking and sleep .
these rhythms have been closely associated with different behavioral and cognitive states , but it is still unknown how the nervous system makes use of these rhythms to perform functionally important tasks . to address those questions ,
it is first useful to understood in a mechanistic way the origin of the rhythms , their interactions , the signals which create the transitions among rhythms , and the ways in which rhythms filter the signals to a network of neurons .
this talk discusses how dynamical systems have been used to investigate the origin , properties and interactions of rhythms in the nervous system .
it focuses on how the underlying physiology of the cells and synapses of the networks shape the dynamics of the network in different contexts , allowing the variety of dynamical behaviors to be displayed by the same network .
the work is presented using a series of related case studies on different rhythms .
these case studies are chosen to highlight mathematical issues , and suggest further mathematical work to be done .
the topics include : different roles of excitation and inhibition in creating synchronous assemblies of cells , different kinds of building blocks for neural oscillations , and transitions among rhythms .
the mathematical issues include reduction of large networks to low dimensional maps , role of noise , global bifurcations , use of probabilistic formulations .
4.5 mm * 2000 mathematics subject classification : * 37n25 , 92c20 . |
the formation of galaxies is a classic problem in astrophysics .
contemporary galaxy formation models combine the well - established lambda - cold dark matter ( lcdm ) cosmology , which describes behavior of dark matter on very large scales , with baryonic physics to model galaxy formation . in the early universe ,
pockets of dark matter decoupled from the hubble flow , collapsed into virialized halos , and then clustered hierarchically into larger structures .
meanwhile , gas aggregated in the interiors of the halos to form rotating disks , which are the building blocks of galaxies ( navarro & steinmetz , 2002 ; cole et al .
such disks were destroyed during mergers of their parent halos , leaving behind classical de vaucouleurs bulges .
spiral disk galaxies formed subsequently as gaseous disks accreted around spheroids ( burkert & naab , 2004 ) .
troubling inconsistencies exist between real galaxies and lcdm models of galaxy formation .
one issue is the angular momentum problem ; simulated galaxy disks have smaller scalelengths and , therefore , less specific angular momentum than their counterparts in nature ( donghia & burkert , 2006 ) .
a second problem is the severe under prediction in the fraction of galaxies with low bulge - to - total mass ratio ( @xmath00.2 ) and of so - called bulgeless galaxies , which lack a classical bulge .
simulated spiral galaxies feature prominent classical bulges in their cores .
such predictions are in glaring contradiction with emerging observations that suggest 15 - 20% of disk galaxies out to [email protected] are bulgeless ( kautsch et al .
2006 ; barazza et al .
2007 ) there are many unanswered questions about the assembly of bulges , the distribution of @xmath2 , and the properties of so - called bulgeless galaxies with low @xmath2 .
how do properties , such as disk scalelengths , mass , kinematics , colors , and star formation histories vary across galaxies of different @xmath2 , ranging from bulge - dominated systems to quasi - bulgeless systems ?
are quasi - bulgeless systems confined to low mass systems with high specific star formation rates , while classical bulges populate high mass systems ?
how do the fraction , mass function , and structural properties of galaxies with different @xmath2 vary across environments with different large - scale cosmological overdensities ?
if environment plays a central part in suppressing bulge formation , then differences would be expected in the properties of bulgeless galaxies in different environments , such as field versus dense galaxy clusters .
how does the frequency and properties of galaxies with low @xmath2 as a function of redshift over @xmath3 compare to the recently reported merger history of galaxies over this epoch ( jogee et al 2007 ) ? answering these questions will help us to understand the reasons behind the apparent failure of lcdm galaxy formation models , and shed light on how galaxies assemble .
progress is possible by observationally constraining properties of enigmatic bulgeless galaxies . a powerful technique for measuring the structural properties ( e.g. scalelengths , srsic indexes , _ b / t _ ) of galaxies is the decomposition of the 2d light distribution into separate structural components with galfit ( peng et al . 2002 ) .
most earlier work has only performed 2d bulge - disk decomposition , but because late - type spirals have been shown to have higher optical bar fractions than early - type galaxies ( barazza et al .
2007 ) , it is important to include the bar when analyzing disk - dominated systems .
bars can contain a significant fraction of light , so failure to account for bars could lead to inflated _ b / t _ ( laurikainen et al . 2006 ) .
we perform three - component decomposition of the 2d galaxy light distribution , while taking the psf into account , with galfit .
since galfit utilizes a non - linear least squares algorithm , initial guess parameters are required for each component galfit attempts to fit . while reasonable initial guesses can be generated by inspection in many cases with common tools ( e.g. iraf ) ,
this is time - consuming and inefficient for large samples . in practice , we break three - component decomposition into three separate invocations of galfit .
we first perform one and two - component fits to constrain the bulge and disk parameters .
the single - component fit models the entire galaxy with only a srsic bulge component .
in addition to constraining the bulge structural parameters , the total luminosity of the object is also determined . a two - component fit , consisting of a srsic bulge and exponential disk ,
is then made based on the output of the previous fit .
if galfit is allowed to do an unconstrained two - component bulge - disk fit in a strongly barred galaxy , it will often try to fit the bar by artificially stretching the disk along the bar pa . in order to get physically meaningful two - component fits
, we therefore constrain the fit by fixing the position angle and axis ratio ( b / a ) of the outer disk to values pre - determined by fitting an ellipse to the outermost disk isophote .
finally , a three - component bulge - bar - disk fit is performed , using the two - component fits as initial guesses for the bulge parameters , and fixing the disk b / a and pa as before .
bars are modeled with elongated , low - index srsic components using initial guesses for the size and position angle estimated from the images .
all objects are subjected to the three - component fits , regardless of whether they appear by eye to possess a bar .
if there is independent evidence for an agn or nuclear cluster , a point source is fitted as a fourth component . in order to decide which of the two or three - component fit is better ,
a number of criteria are used .
1 ) if the one or two - component residuals show a bar signature that is removed in the three - component residual , then the three component fit is favored ; 2 ) structural parameters ( scalelength , srsic index , b / a ) of the bar fit must well behaved ; 3 ) visual evidence of a strong bar in the input images favors the three - component fit .
weak bars are often not visually prominent , but for such bars , the changes in the disk or bulge parameters , between the two and three component fits , are small ; ( 4 ) in addition , we test the robustness of the three - component solution by varying the initial guesses to check that the same solution is converged upon . in order to address the questions outlined in 1 , we are applying the three - component decomposition to three complementary samples , which span different epochs and different environments ( field and cluster ) : ( 1 ) a @xmath4 0 sample of @xmath1 200 galaxies with hubble types s0 to sm drawn from the osu bright spiral galaxy survey ( osubsg ) ( eskridge et al .
2002 ) and ukidss ( mclure et al .
2006 ) ; ( 2 ) a sample of galaxies in the dense environment of the coma cluster from our acs treasury survey ( carter et al 2007 ) ; and a sample of early disk galaxies out @xmath4 2 with deep nicmos imaging ( 180 orbits ) .
for the two - component fits , we have performed consistency checks by testing our decomposition on samples of galaxies with published results from the millennium galaxy catalog ( driver et al .
2007 ) and the new york university value - added catalog ( blanton et al . 2005 ) for the sloan digital sky survey . for three - component fits ,
we have performed similar tests on a few galaxies drawn from small samples with published three - component bulge - bar - disk fits , drawn from laurikainen et al .
( 2006 ) and reese et al .
( 2007 ) . 1 .
luminosity is conserved between the two and three - component fits .
2 . modeling the bar in the three - component fits forces a reshuffling of luminosity .
generally , the bulge declines in luminosity , whereas light can be taken from , or added back , to the disk .
the reshuffling of light occurs because the two - component model adjusts the bulge and disk accordingly to compensate for the bar , which can include artificially elongating and brightening the bulge .
accounting for the bar returns the bulge and disk parameters closer to their true values .
inclusion of the bar can reduce bulge fractional luminosity _
b / t _ by a factor of two or more .
larger changes in bulge luminosities ( a factor of 10 or more ) occur in cases where a prominent bar influences the two - component fit to very much overstate the bulge luminosity .
the bulge - disk fits in such extreme cases underscore the importance of including the bar in 2d luminosity decomposition .
the scalelength of the disk is generally unchanged by including the bar .
however in a few cases , the two - component disk structure can be erroneous , as in the case of ngc 4643 , shown in figure 1 .
we have provided a proof of concept of our ongoing three - component bulge - disk - bar decomposition with galfit .
we are optimistic about our on - going work , which will be described in weinzirl et al .
2008 ( in prep ) . )
that does not match the galaxy .
due to its resulting low surface brightness , the fittted disk is hard to see , and ellipses are drawn to show its pa and b / a . in step 3 , the addition of the bar component restores the disk scalength to a reasonable value .
the fit parameters are presented in table 1 . ]
c c c c c c c fit & & @xmath5 or h ( @xmath6 ) &
n & b / a & position angle & fractional light + step 1&sersic & 27.90 & 4.44 & 0.80 & -51.08 & 100% + step 2&bulge & 23.86 & 4.16 & 0.80 & -51.08 & 34.6% + & disk & 335.88 & 1.0 & 0.84 & 66.94 & 65.4% + step 3&bulge & 5.43 & 2.53 & 0.90 & 60.52 & 25.0 % + & disk & 48.22 & 1.0 & 0.84 & 66.94 & 54.1 % + & bar & 21.30 & 0.62 & 0.37 & -45.84 & 20.9 % + barazza f. d. , jogee , s. , marinova , i. 2007 , apj , accepted , arxiv:0710.4674 blanton , m. r. , et al .
2005 , , 129 , 2562 burkert , a. , & naab , t. 2004 , coevolution of black holes and galaxies , 422 carter , d. et al .
2007 , apj , submitted cole , s. m. , baugh , c. , frenk , c. , lacey , c. , benson , a. 2000 , mnras , 319 , 168 donghia , e. & burkert , a. 2006 , mnras , 372 , 1525 driver , s. p. , liske , j. , graham , a. w. , 2007 , apjl , accepted ( astro - ph/0701728 ) eskridge , p. b. , et al .
2002 , , 143 , 73 jogee et al .
2007 , proceedings of `` formation and evolution of galaxy disks '' , held in rome , 1 - 5 october 2007 ( arxiv:0802.3901 ) kautsch , s. j. , grebel , e. k. , barazza , f. d. , & gallagher , j. s. , iii 2006 , a&a , 445 , 765 laurikainen , e. , salo , h. , buta , r. , knapen , j. , speltincx , t. , & block , d. 2006 , , 132 , 2634 mclure , r. j. , et al .
2006 , , 372 , 357 navarro , j.f . , steinmetz , m. 2000 , apj , 538 , 477 peng c. y. , ho l. c. , impey c. d. , rix h .- w .
, 2002 , aj , 124 , 266 reese , a. s. , williams , t. b. , sellwood , j. a. , barnes , e. i. , & powell , b. a. 2007 , , 133 , 2846 | structural decomposition of galaxies into bulge , disk , and bar components is important to address a number of scientific problems .
measuring bulge , disk , and bar structural parameters will set constraints on the violent and secular processes of galaxy assembly and recurrent bar formation and dissolution models .
it can also help to quantify the fraction and properties of bulgeless galaxies ( those systems having no bulge or only a relatively insignificant disky - pseudobulges ) , which defy galaxy formation paradigms requiring almost every disk galaxy to have a classical bulge at its core .
we demonstrate a proof of concept and show early results of our ongoing three - component bulge - disk - bar decomposition of nir images for a sample of three complementary samples spanning different epochs and different environments ( field and cluster ) .
in contrast to most early studies , which only attempt two - component bulge - disk decomposition , we fit three components using galfit : a bulge , a disk , and a bar .
we show that it is important to include the bar component , as this can significantly lower the bulge - to - total luminosity ratio ( _ b / t _ ) , in many cases by a factor of two or more , thus effectively changing the hubble type of a galaxy from early to late . |
relativistic flows are involved in many of the high - energy astrophysical phenomena , such as , for example , jets in extragalactic radio sources , accretion flows around compact objects , pulsar winds and @xmath0 ray bursts . in many instances the presence of a magnetic field is also an essential ingredient for explaining the physics of these objects and interpreting their observational appearance .
theoretical understanding of relativistic phenomena is subdue to the solution of the relativistic magnetohydrodynamics ( rmhd ) equations which , owing to their high degree of nonlinearity , can hardly be solved by analytical methods .
for this reason , the modeling of such phenomena has prompted the search for efficient and accurate numerical formulations . in this respect ,
godunov - type schemes @xcite have gained increasing popularity due to their ability and robustness in accurately describing sharp flow discontinuities such as shocks or tangential waves .
one of the fundamental ingredient of such schemes is the exact or approximate solution to the riemann problem , i.e. , the decay between two constant states separated by a discontinuity .
unfortunately the use of an exact riemann solver @xcite is prohibitive because of the huge computational cost related to the high degree of nonlinearities present in the equations . instead
, approximate methods of solution are preferred .
linearized solvers @xcite rely on the rather convoluted eigenvector decomposition of the underlying equations and may be prone to numerical pathologies leading to negative density or pressures inside the solution @xcite . characteristic - free algorithms based on the rusanov lax - friedrichs or the harten - lax - van leer ( hll , * ? ? ?
* ) formulations are sometime preferred due to their ease of implementation and positivity properties .
implementation of such algorithms can be found in the codes described by @xcite .
although simpler , the hll scheme approximates only two out of the seven waves by collapsing the full structure of the riemann fan into a single average state . these solvers , therefore , are not able to resolve intermediate waves such as alfvn , contact and slow discontinuities .
attempts to restore the middle contact ( or entropy ) wave ( hllc , initially devised for the euler equations by * ? ? ?
* ) have been proposed by @xcite in the case of purely transversal fields and by @xcite ( mb from now on ) , @xcite in the more general case .
these schemes provide a relativistic extension of the work proposed by @xcite and @xcite for the classical mhd equations .
hllc solvers for the equations of mhd and rmhd , however , still do not capture slow discontinuities and alfvn waves . besides
, direct application of the hllc solver of mb to genuinely 3d problems was shown to suffer from a ( potential ) pathological singularity when the component of magnetic field normal to a zone interface approaches zero .
a step forward in resolving intermediate wave structures was then performed by @xcite ( mk from now on ) who , in the context of newtonian mhd , introduced a four state solver ( hlld ) restoring the rotational ( alfvn ) discontinuities . in this paper
we propose a generalization of miyoshi & kusano approach to the equations of relativistic mhd .
as we shall see , this task is greatly entangled by the different nature of relativistic rotational waves across which the velocity component normal to the interface is no longer constant .
the proposed algorithm has been implemented in the pluto code for astrophysical fluid dynamics @xcite which embeds a computational infrastructure for the solution of different sets of equations ( e.g. , euler , mhd or relativistic mhd conservation laws ) in the finite volume formalism .
the paper is structured as follows : in [ sec : equations ] we briefly review the equations of relativistic mhd ( rmhd ) and formulate the problem . in [ sec : solver ] the new riemann solver is derived .
numerical tests and astrophysical applications are presented in [ sec : num ] and conclusions are drawn in [ sec : conclusions ] .
the equations of relativistic magnetohydrodynamics ( rmhd ) are derived under the physical assumptions of constant magnetic permeability and infinite conductivity , appropriate for a perfectly conducting fluid @xcite . in divergence form , they express particle number and energy - momentum conservation : @xmath1 & = & 0\ , , \\
\noalign{\medskip } \label{eq : clind } \partial_\mu\left(u^\mu b^\nu - u^\nu b^\mu\right ) & = & 0\,,\end{aligned}\ ] ] where @xmath2 is the rest mass density , @xmath3 is the four - velocity ( @xmath4 lorentz factor , @xmath5 three velocity ) , @xmath6 and @xmath7 are the gas enthalpy and thermal pressure , respectively .
the covariant magnetic field @xmath8 is orthogonal to the fluid four - velocity ( @xmath9 ) and is related to the local rest frame field @xmath10 by @xmath11\,.\ ] ] in eq .
( [ eq : clmomen ] ) , @xmath12 is the squared magnitude of the magnetic field .
the set of equations ( [ eq : clmass])([eq : clind ] ) must be complemented by an equation of state which may be taken as the constant @xmath13-law : @xmath14 where @xmath13 is the specific heat ratio .
alternative equations of state ( see , for example , * ? ? ?
* ) may be adopted . in the following we will be dealing with the one dimensional conservation law @xmath15 which follows directly from eq .
( [ eq : clmass])-([eq : clind ] ) by discarding contributions from @xmath16 and @xmath17 .
conserved variables and corresponding fluxes take the form : @xmath18 where @xmath19 , @xmath20 is the the density as seen from the observer s frame while , introducing @xmath21 ( total enthalpy ) and @xmath22 ( total pressure ) , @xmath23 are the momentum and energy densities , respectively .
@xmath24 is the kronecker delta symbol .
note that , since @xmath25 , the normal component of magnetic field ( @xmath26 ) does not change during the evolution and can be regarded as a parameter .
this is a direct consequence of the @xmath27 condition . a conservative discretization of eq . ( [ eq:1dconslaw ] ) over a time step @xmath28 yields @xmath29 where @xmath30 is the mesh spacing and @xmath31 is the upwind numerical flux computed at zone faces @xmath32 by solving , for @xmath33 , the initial value problem defined by eq . ( [ eq:1dconslaw ] ) together with the initial condition @xmath34 where @xmath35 and @xmath36 are discontinuous left and right constant states on either side of the interface
this is also known as the riemann problem . for a first order scheme , @xmath37 and @xmath38 . the decay of the initial discontinuity given by eq .
( [ eq : riemann ] ) leads to the formation of a self - similar wave pattern in the @xmath39 plane where fast , slow , alfvn and contact modes can develop . at the double end of the riemann fan , two fast magneto - sonic waves bound the emerging pattern enclosing two rotational ( alfvn ) discontinuities , two slow magneto - sonic waves and a contact surface in the middle .
the same patterns is also found in classical mhd .
fast and slow magneto - sonic disturbances can be either shocks or rarefaction waves , depending on the pressure jump and the norm of the magnetic field .
all variables ( i.e. density , velocity , magnetic field and pressure ) change discontinuously across a fast or a slow shock , whereas thermodynamic quantities such as thermal pressure and rest density remain continuous when crossing a relativistic alfvn wave .
contrary to its classical counterpart , however , the tangential components of magnetic field trace ellipses instead of circles and the normal component of the velocity is no longer continuous across a rotational discontinuity , @xcite .
finally , through the contact mode , only density exhibits a jump while thermal pressure , velocity and magnetic field remain continuous .
the complete analytical solution to the riemann problem in rmhd has been recently derived in closed form by @xcite and number of properties regarding simple waves are also well established , see @xcite .
for the special case in which the component of the magnetic field normal to a zone interface vanishes , a degeneracy occurs where tangential , alfvn and slow waves all propagate at the speed of the fluid and the solution simplifies to a three - wave pattern , see @xcite .
the high degree of nonlinearity inherent to the rmhd equations makes seeking for an exact solution prohibitive in terms of computational costs and efficiency . for this reasons ,
approximate methods of solution are preferred instead .
and @xmath36 are connected to each other through a set of five waves representing , clockwise , a fast shock @xmath40 , a rotational discontinuity @xmath41 , a contact wave @xmath42 , a rotational discontinuity @xmath43 and a fast shock @xmath44 .
the outermost states , @xmath35 and @xmath36 are given as input to the problem , whereas the others must be determined consistently solving the rankine - hugoniot jump conditions.,scaledwidth=50.0% ] without loss of generality , we place the initial discontinuity at @xmath45 and set @xmath46 .
following mk , we make the assumption that the riemann fan can be divided by @xmath47 waves : two outermost fast shocks , @xmath44 and @xmath40 , enclosing two rotational discontinuities , @xmath41 and @xmath43 , separated by the entropy ( or contact ) mode with speed @xmath42 .
note that slow modes are not considered in the solution .
the five waves divide the @xmath39 plane into the six regions shown in fig [ fig : fan ] , corresponding ( from left to right ) to the @xmath48 states @xmath49 with @xmath50 . the outermost states ( @xmath35 and @xmath36 ) are given as input to the problem , while the remaining ones have to be determined . in the typical approach used to construct hll - based solvers , the outermost velocities @xmath40 and @xmath44 are also provided as estimates from the input left and right states . as in mb
, we choose to use the simple davis estimate @xcite . across
any given wave @xmath51 , states and fluxes must satisfy the jump conditions @xmath52_{\lambda } \equiv \big(\lambda{\bmath{u } } - { \bmath{f}}\big)_+ - \big(\lambda{\bmath{u } } - { \bmath{f}}\big)_- = 0\,,\ ] ] where @xmath53 and @xmath54 identify , respectively , the state immediately ahead or behind the wave front .
note that for consistency with the integral form of the conservation law over the rectangle @xmath55\times[0 , \delta t]$ ] one has , in general , @xmath56 , except of course for @xmath57 or @xmath58 . across the fast waves , we will make frequent use of @xmath59 which are known vectors readily obtained from the left and right input states . a particular component of @xmath60 is selected by mean of a subscript , e.g. , @xmath61 is the density component of @xmath60 .
a consistent solution to the problem has to satisfy the @xmath62 nonlinear relations implied by eq .
( [ eq : jump ] ) for each of the @xmath47 waves considered , thus giving a total of @xmath63 equations . moreover
, physically relevant solutions must fulfill a number of requirements in order to reflect the characteristic nature of the considered waves .
for this reason , across the contact mode , we demand that velocity , magnetic field and total pressure be continuous : @xmath64_{\lambda_c } = \big[{\bmath{b}}\big]_{\lambda_c } = { \bmath{0 } } \,,\quad \big[p\big]_{\lambda_c } = 0 \,,\ ] ] and require that @xmath65 , i.e. , that the contact wave moves at the speed of the fluid . however , density , energy and total enthalpy may be discontinuous . on the other hand , through the rotational waves @xmath41 and @xmath43 , scalar quantities such as total pressure and enthalpy are invariant whereas all vector components ( except for @xmath26 ) experience jumps .
since slow magnetosonic waves are not considered , we naturally conclude that only the total pressure remains constant throughout the fan , contrary to newtonian mhd , where also the velocity normal to the interface ( @xmath66 ) is left unchanged across the waves .
this is an obvious consequence of the different nature of relativistic alfvn waves across which vector fields like @xmath67 and @xmath8 trace ellipses rather than circles . as a consequence ,
the normal component of the velocity , @xmath66 , is no longer invariant in rmhd but experiences a jump .
these considerations along with the higher level of complexity of the relativistic equations makes the extension of the multi - state hll solver to rmhd considerably more elaborate .
our strategy of solution is briefly summarized . for each state
we introduce a set of @xmath68 _ independent _ unknowns : @xmath69 and write conservative variables and fluxes given by eq .
( [ eq : uandf ] ) as @xmath70 @xmath71 where @xmath19 labels the vector component , @xmath72 is the state and @xmath8 is computed directly from ( [ eq : bmu ] ) .
we proceed by solving , as function of the total pressure @xmath73 , the jump conditions ( [ eq : jump ] ) across the outermost waves @xmath40 and @xmath44 . by requiring that total pressure and alfvn velocity do not change across each rotational modes , we find a set of invariant quantities across @xmath41 and @xmath43 .
using these invariants , we express states and fluxes on either side of the contact mode ( @xmath74 ) in terms of the total pressure unknown only .
imposing continuity of normal velocity , @xmath75 , leads to a nonlinear scalar equation in @xmath73 , whose zero gives the desired solution .
once @xmath73 has been found to some relative accuracy ( typically @xmath76 ) , the full solution to the problem can be written as @xmath77 where @xmath78 are computed in [ sec : fast ] , @xmath79 in [ sec : contact ] and @xmath80 ( for @xmath81 or @xmath82 ) follow from the jump conditions .
the wave speeds @xmath83 and @xmath42 are computed during the solution process .
here and in what follows we adopt the convention that single subscripts like @xmath84 ( or @xmath85 ) refers indifferently to @xmath86 ( or @xmath87 ) .
thus an expression like @xmath88 means @xmath89 and @xmath90 .
we start by explicitly writing the jump conditions across the outermost fast waves : @xmath91 where , to avoid cluttered notations , we omit in this section the @xmath92 ( when @xmath93 ) or @xmath94 ( when @xmath95 ) index from the quantities appearing on the left hand side .
similarly , the @xmath96 s appearing on the right hand sides of equations ( [ eq : jumpd])([eq : jumpbk ] ) are understood as the components of the vector @xmath97 ( when @xmath98 ) or @xmath99 ( when @xmath95 ) , defined by eq .
( [ eq : r ] ) .
the jump conditions of faraday s law allow to express the magnetic field as a function of velocities alone , @xmath100 the energy and momentum equations can be combined together to provide an explicit functional relation between the three components of velocity and the total pressure @xmath73 . to this purpose , we first multiply the energy equation ( [ eq : jumpe ] ) times @xmath101 and then subtract the resulting expression from the jump condition for the @xmath102-th component of momentum , eq .
( [ eq : jumpmk ] ) . using eq .
( [ eq : jumpbk ] ) to get rid of the @xmath103 term , one finds after some algebra : @xmath104 with @xmath105 defined by ( [ eq : bv ] ) . the system can be solved for @xmath101 giving @xmath106}{x}\ , , \\
\noalign{\medskip } \label{eq : vza } v^z & = & \frac{qr_{m^z } + r_{b^z}\left[c + b^x\left(\lambda r_{m^x } - r_e\right)\right]}{x}\,,\end{aligned}\ ] ] where @xmath107 once the velocity components are expressed as functions of @xmath73 , the magnetic field is readily found from ( [ eq : bv ] ) , while the total enthalpy can be found using its definition , @xmath108 , or by subtracting @xmath109 from the inner product @xmath110 , giving @xmath111 where @xmath112 .
although equivalent , we choose to use this second expression .
since the @xmath101 are functions of @xmath73 alone , the total enthalpy @xmath113 is also a function of the total pressure .
the remaining conserved quantities in the @xmath92 or @xmath94 regions can be computed once @xmath73 has been found : @xmath114 one can verify by direct substitution that the previous equations together with the corresponding fluxes , eq .
( [ eq : flux ] ) , satisfy the jump conditions given by ( [ eq : jumpd])([eq : jumpbk ] ) . across the rotational waves one could ,
in principle , proceed as for the outer waves , i.e. , by explicitly writing the jump conditions .
however , as we shall see , the treatment greatly simplifies if one introduces the four vector @xmath115 where , for reasons that will be clear later , we take the plus ( minus ) sign for the right ( left ) state . from @xmath116
we define the spatial vector @xmath117 with components given by @xmath118 the vector @xmath119 has some attractive properties , the most remarkable of which is that the @xmath120 component coincides with the propagation speed of the alfvn wave @xcite .
for this reason , we are motivated to define @xmath121 , where the subscript @xmath84 stands for either the left or right rotational wave ( i.e. @xmath122 or @xmath123 ) since we require that both @xmath124 and @xmath73 are invariant across the rotational discontinuity , i.e. , @xmath125 , a property certainly shared by the exact solution . as we will show
, this choice naturally reduces to the expressions found by mk in the non - relativistic limit . indeed , setting @xmath126 and using eq .
( [ eq : k ] ) to express @xmath101 as functions of @xmath127 , the jump conditions simplify to @xmath128_{\lambda_a } & = & 0 \\ \noalign{\medskip } \label{eq : alfm } \left[\frac{\eta\sigma^k b^x}{\sigma^0 } - p\delta_{kx}\right]_{\lambda_a } & = & 0 \\ \noalign{\medskip } \label{eq : alfe } \left[\eta b^x - \frac{\sigma^x}{\sigma^0}p\right]_{\lambda_a } & = & 0 \\ \noalign{\medskip } \label{eq : alfb } \left[\frac{b^x\sigma^k}{\sigma^0 } \right]_{\lambda_a } & = & 0 \,,\end{aligned}\ ] ] since also @xmath129_{\lambda_a } = 0 $ ] , the previous equations further imply that ( when @xmath130 ) also @xmath131 , @xmath113 , @xmath132 and @xmath133 do not change across @xmath134 : @xmath135 @xmath136 being invariant , @xmath119 can be computed from the state lying to the left ( for @xmath41 ) or to the right ( for @xmath43 ) of the discontinuity , thus being a function of the total pressure @xmath73 alone . instead of using eq .
( [ eq : k ] ) , an alternative and more convenient expression may be found by properly replacing @xmath101 with @xmath127 in eq .
( [ eq : jumpd])([eq : jumpbk ] ) .
after some algebra one finds the simpler expression @xmath137 still being a function of the total pressure @xmath73 .
note that , similarly to its non relativistic limit , we can not use the equations in ( [ eq : alfd])([eq : alfb ] ) to compute the solution across the rotational waves , since they do not provide enough independent relations .
instead , a solution may be found by considering the jump conditions across both rotational discontinuities and properly matching them using the conditions at the contact mode . at the contact discontinuity ( cd )
only density and total enthalpy can be discontinuous , while total pressure , normal and tangential fields are continuous as expressed by eq .
( [ eq : contact_conditions ] ) .
since the magnetic field is a conserved quantity , one can immediately use the consistency condition between the innermost waves @xmath41 and @xmath43 to find @xmath105 across the cd .
indeed , from @xmath138 one has @xmath139 , where @xmath140_{ar } - \left[b^k(\lambda - v^x ) + b^xv^k\right]_{al } } { \lambda_{ar } - \lambda_{al } } \ , . \ ] ] since quantities in the @xmath122 and @xmath123 regions are given in terms of the @xmath73 unknown , eq .
( [ eq : bc ] ) are also functions of @xmath73 alone . at this point , we take advantage of the fact that @xmath141 to replace @xmath142 with @xmath143 and then rewrite ( [ eq : k ] ) as @xmath144 the previous equations form a linear system in the velocity components @xmath101 and can be easily inverted to the left and to the right of the cd to yield @xmath145 which also depend on the total pressure variable only , with @xmath113 and @xmath127 being given by ( [ eq : fastw ] ) and ( [ eq : k_simple ] ) and the @xmath146 s being computed from eq .
( [ eq : bc ] ) . imposing continuity of the normal velocity across the cd , @xmath147 , results in @xmath148 = 0 \,,\ ] ] where @xmath149 is a function of @xmath73 only and
@xmath150 is the numerator of ( [ eq : bc ] ) and @xmath151 .
equation ( [ eq : final ] ) is a nonlinear function in @xmath73 and must be solved numerically .
once the iteration process has been completed and @xmath73 has been found to some level of accuracy , the remaining conserved variables to the left and to the right of the cd are computed from the jump conditions across @xmath41 and @xmath43 and the definition of the flux , eq .
( [ eq : flux ] ) .
specifically one has , for @xmath152 or @xmath153 , @xmath154 this concludes the derivation of our riemann solver . in the previous sections we have shown that the whole set of jump conditions can be brought down to the solution of a single nonlinear equation , given by ( [ eq : final ] ) , in the total pressure variable @xmath73 . in the particular case of vanishing normal component of the magnetic field ,
i.e. @xmath155 , this equation can be solved exactly as discussed in [ sec : bx0 ] . for the more general case
, the solution has to be found numerically using an iterative method where , starting from an initial guess @xmath156 , each iteration consists of the following steps : * given a new guess value @xmath157 to the total pressure , start from eq .
( [ eq : vxa])([eq : vza ] ) to express @xmath158 and @xmath159 as functions of the total pressure .
also , express magnetic fields @xmath160 , @xmath161 and total enthalpies @xmath162 , @xmath163 using eq .
( [ eq : bv ] ) and eq .
( [ eq : fastw ] ) , respectively . *
compute @xmath164 and @xmath165 using eq .
( [ eq : k_simple ] ) and the transverse components of @xmath166 using eq .
( [ eq : bc ] ) .
( [ eq : final ] ) to find the next improved iteration value . for the sake of assessing the validity of our new solver , we choose the secant method as our root - finding algorithm .
the initial guess is provided using the following prescription : @xmath167 where @xmath168 is the total pressure computed from the hll average state whereas @xmath169 is the solution in the @xmath170 limiting case .
extensive numerical testing has shown that the total pressure @xmath168 computed from the hll average state provides , in most cases , a sufficiently close guess to the correct physical solution , so that no more than @xmath171 iterations ( for zones with steep gradients ) were required to achieve a relative accuracy of @xmath76 .
the computational cost depends on the simulation setting since the average number of iterations can vary from one problem to another . however , based on the results presented in
[ sec : num ] , we have found that hlld was at most a factor of @xmath172 slower than hll . for a solution to be physically consistent and
well - behaved , we demand that @xmath173 hold simultaneously .
these conditions guarantee positivity of density and that the correct eigenvalue ordering is always respected .
we warn the reader that equation ( [ eq : final ] ) may have , in general , more than one solution and that the conditions given by ( [ eq : conditions ] ) may actually prove helpful in selecting the correct one .
however , the intrinsic nonlinear complexity of the rmhd equations makes rather arduous and challenging to prove , _ a priori _ , both the existence and the uniqueness of a physically relevant solution , in the sense provided by ( [ eq : conditions ] ) . on the contrary
, we encountered sporadic situations where none of the zeroes of eq .
( [ eq : final ] ) is physically admissible .
fortunately , these situations turn out to be rare eventualities caused either by a large jump between left and right states ( as at the beginning of integration ) or by under- or over- estimating the propagation speeds of the outermost fast waves , @xmath40 and @xmath44 .
the latter conclusion is supported by the fact that , enlarging one or both wave speeds , led to a perfectly smooth and unique solution
. therefore , we propose a safety mechanism whereby we switch to the simpler hll riemann solver whenever at least one or more of the conditions in ( [ eq : conditions ] ) is not fulfilled . from several numerical tests , including the ones shown here , we found the occurrence of these anomalies to be limited to few zones of the computational domain , usually less than @xmath174 in the tests presented here .
we conclude this section by noting that other more sophisticated algorithms may in principle be sought .
one could , for instance , provide a better guess to the outer wave - speeds @xmath40 and @xmath44 or even modify them accordingly until a solution is guaranteed to exist .
another , perhaps more useful , possibility is to bracket the solution inside a closed interval @xmath175 $ ] where @xmath176 and @xmath177 may be found from the conditions ( [ eq : conditions ] ) . using an alternative root finder , such as ridder @xcite , guarantees that the solution never jump outside the interval .
however , due to the small number of failures usually encountered , we do not think these alternatives could lead to a significant gain in accuracy . in the limit
@xmath178 a degeneracy occurs where the alfvn ( and slow ) waves propagate at the speed of the contact mode which thus becomes a tangential discontinuity . across this degenerate front ,
only normal velocity and total pressure remain continuous , whereas tangential vector fields are subject to jumps .
this case does not pose any serious difficulty in our derivation and can be solved exactly .
indeed , by setting @xmath170 in eq .
( [ eq : k_simple ] ) and ( [ eq : final ] ) , one immediately finds that @xmath179 leading to the following quadratic equation for @xmath73 : @xmath180 where the superscript hll " refers to the hll average state or flux given by eq .
( 28 ) or ( 31 ) of mb .
we note that equation ( [ eq : bx0 ] ) coincides with the derivation given by mb ( see also * ? ? ?
* ) in the same degenerate case and the positive root gives the correct physical solution .
the intermediate states , @xmath181 and @xmath182 , loose their physical meaning as @xmath178 but they never enter the solution since , as @xmath183 , only @xmath184 and @xmath185 will have a nonzero finite width , see fig .
[ fig : fan ] . given the initial guess , eq .
( [ eq : guess ] ) , our proposed approach does not have to deal separately with the @xmath186 and @xmath187 cases ( as in mb and * ? ? ? * ) and thus solves the issue raised by mb .
we now show that our derivation reduces to the hlld riemann solver found by mk under the appropriate non - relativistic limit .
we begin by noticing that , for @xmath188 , the velocity and induction four - vectors reduce to @xmath189 and @xmath190 , respectively .
also , note that @xmath191 in the non - relativistic limit so that @xmath192 and thus @xmath66 can not change across @xmath134 . replacing ( [ eq : jumpd])-([eq : jumpmk ] ) with their non - relativistic expressions and demanding @xmath193 gives , in our notations , the following expressions : @xmath194 which can be shown to be identical to eqns ( 38 ) and ( 41 ) of mk . with little algebra , one can also show that the remaining variables in the @xmath122 and @xmath123 regions reduce to the corresponding non - relativistic expressions of mk .
similarly , the jump across the rotational waves are solved exactly in the same fashion , that is , by solving the integral conservation laws over the riemann fan .
for instance , eq .
( [ eq : bc ] ) reduces to equation ( 61 ) and ( 62 ) of mk .
these results should not be surprising since , our set of parameters to write conserved variables and fluxes is identical to the one used by mk .
the only exception is the energy , which is actually written in terms of the total enthalpy .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] [ tab : ic ] we now evaluate , in [ sec : test1d ] , the accuracy of the proposed hlld riemann solver by means of selected one dimensional shock tube problems .
applications of the solver to multi - dimensional problems of astrophysical relevance are presented in [ sec : test23d ] .
the initial condition is given by eq .
( [ eq : riemann ] ) with left and right states defined by the primitive variables listed in table 1 .
the computational domain is chosen to be the interval @xmath195 $ ] and the discontinuity is placed at @xmath196 .
the resolution @xmath197 and final integration time can be found in the last two columns of table 1 .
unless otherwise stated , we employ the constant @xmath198 law with @xmath199 .
the rmhd equations are solved using the first - order accurate scheme ( [ eq:1st_ord ] ) with a cfl number of @xmath200 .
numerical results are compared to the hllc riemann solver of mb and to the simpler hll scheme and the accuracy is quantified by computing discrete errors in l-1 norm : @xmath201 where @xmath202 is the first - order numerical solution ( density or magnetic field ) , @xmath203 is the reference solution at @xmath204 and @xmath205 is the mesh spacing . for tests @xmath206 we obtained a reference solution using the second - order scheme of mb on @xmath207 zones and adaptive mesh refinement with @xmath48 levels of refinement ( equivalent resolution @xmath208 grid points ) .
grid adaptivity in one dimension has been incorporated in the pluto code using a block - structured grid approach following @xcite . for test @xmath209 ,
we use the exact numerical solution available from @xcite .
errors ( in percent ) are shown in fig . [
fig : error ] . .
density and @xmath16 component of magnetic field are shown , respectively .
the different symbols show results computed with the new hlld solver ( filled circles ) , the hllc solver ( crosses ) and the simpler hll solver ( plus signs ) .
note that only hlld is able to capture exactly both discontinuities by keeping them perfectly sharp without producing any grid diffusion effect .
hllc can capture the contact wave but not the rotational discontinuity , whereas hll spreads both of them on several grid zones.,scaledwidth=50.0% ] we now show that our hlld solver can capture _ exactly _ isolated contact and rotational discontinuities .
the initial conditions are listed at the beginning of table 1 . in the case of an isolated stationary contact wave ,
only density is discontinuous across the interface . the left panel in fig .
[ fig : isolated ] shows the results at @xmath210 computed with the hlld , hllc and hll solvers : as expected our hlld produces no smearing of the discontinuity ( as does hllc ) . on the contrary , the initial jump broadens over several grid zone when employing the hll scheme . across a rotational discontinuity ,
scalar quantities such as proper density , pressure and total enthalpy are invariant but vector fields experience jumps .
the left and right states on either side of an exact rotational discontinuity can be found using the procedure outlined in the appendix .
the right panel in fig .
[ fig : isolated ] shows that only hlld can successfully keep a sharp resolution of the discontinuity , whereas both hllc and hll spread the jump over several grid points because of the larger numerical viscosity . .
computations are carried on @xmath211 zones using the hlld ( solid line ) , hllc ( dashed line ) and hll ( dotted line ) riemann solver , respectively .
the top panel shows , from left to right , the rest mass density , gas pressure , total pressure .
the bottom panel shows the @xmath120 and @xmath16 components of velocity and the @xmath16 component of magnetic field.,scaledwidth=50.0% ] .
density and the two components of velocity are shown in the left , central and right panels , respectively .
diamonds , crosses and plus signs are used for the hlld , hllc and hll riemann solver , respectively.,scaledwidth=50.0% ] the first shock tube test is a relativistic extension of the brio wu magnetic shock tube @xcite and has also been considered by @xcite and in mb .
the specific heat ratio is @xmath212 .
the initial discontinuity breaks into a left - going fast rarefaction wave , a left - going compound wave , a contact discontinuity , a right - going slow shock and a right - going fast rarefaction wave .
rotational discontinuities are not generated . in figs .
[ fig : st1]-[fig : st1_close ] we plot the results obtained with the first - order scheme and compare them with the hllc riemann solver of mb and the hll scheme .
although the resolution across the continuous right - going rarefaction wave is essentially the same , the hlld solver offers a considerable improvement in accuracy in the structures located in the central region of the plots . indeed , fig .
[ fig : st1_close ] shows an enlargement of the central part of the domain , where the compound wave ( at @xmath213 ) , contact ( @xmath214 ) and slow shock ( @xmath215 ) are clearly visible . besides the steeper profiles of the contact and slow modes , it is interesting to notice that the compound wave , composed of a slow shock adjacent to a slow rarefaction wave , is noticeably better resolved with the hlld scheme than with the other two .
these results are supported by the convergence study shown in the top left panel of fig .
[ fig : error ] , demonstrating that the errors obtained with our new scheme are smaller than those obtained with the hllc and hll solvers ( respectively ) . at the largest resolution employed , for example ,
the l-1 norm errors become @xmath216 and @xmath217 smaller than the hll and hllc schemes , respectively . the cpu times required by the different riemann solvers on this particular test
were found to be scale as @xmath218 . on @xmath219 grid points . from left to right , the top panel shows density , gas and total pressure .
the middle panel shows the three components of velocity , whereas in the bottom panel we plot the lorentz factor and the transverse components of magnetic field .
solid , dashed and dotted lines are used to identify results computed with hlld , hllc and hll , respectively.,scaledwidth=50.0% ] around the contact wave .
middle and right panels : close - ups of the @xmath17 component of velocity and @xmath16 component of magnetic field around the right - going slow shock and alfvn discontinuity .
different symbols refer to different riemann solver , see the legend in the left panel.,scaledwidth=50.0% ] this test has also been considered in @xcite and in mb and the initial condition comes out as a non - planar riemann problem implying that the change in orientation of the transverse magnetic field across the discontinuity is @xmath220 ( thus different from zero or @xmath221 ) .
the emerging wave pattern consists of a contact wave ( at @xmath222 ) separating a left - going fast shock ( @xmath223 ) , alfvn wave ( @xmath224 ) and slow rarefaction ( @xmath225 ) from a slow shock ( @xmath226 ) , alfvn wave ( @xmath227 ) and fast shock ( @xmath228 ) heading to the right .
computations carried out with the @xmath229 order accurate scheme are shown in fig .
[ fig : st2 ] using the hlld ( solid line ) , hllc ( dashed line ) and hll ( dotted line ) .
the resolution across the outermost fast shocks is essentially the same for all riemann solvers . across the entropy mode
both hlld and hllc attain a sharper representation of the discontinuity albeit unphysical undershoots are visible immediately ahead of the contact mode .
this is best noticed in the the left panel of fig .
[ fig : st2_close ] , where an enlargement of the same region is displayed .
on the right side of the domain , the slow shock and the rotational wave propagate quite close to each other and the first - order scheme can barely distinguish them at a resolution of @xmath219 zones .
however , a close - up of the two waves ( middle and right panel in fig .
[ fig : st2_close ] ) shows that the proposed scheme is still more accurate than hllc in resolving both fronts . on the left hand side ,
the separation between the alfvn and slow rarefaction waves turns out to be even smaller and the two modes blur into a single wave because of the large numerical viscosity .
this result is not surprising since these features are , in fact , challenging even for a second - order scheme @xcite .
discrete l-1 errors computed using eq .
( [ eq : error ] ) are plotted as function of the resolution in the top right panel of fig .
[ fig : error ] . for this particular test ,
hlld and hllc produce comparable errors ( @xmath230 and @xmath231 at the highest resolution ) while hll performs worse on contact , slow and alfvn waves resulting in larger deviations from the reference solution .
the computational costs on @xmath219 grid zones has found to be @xmath232 . .
from top to bottom , left to right , the panels show density @xmath2 , gas pressure @xmath233 , total pressure @xmath73 , @xmath120 and @xmath16 components of velocity ( @xmath66 @xmath234 ) and @xmath16 component of magnetic field @xmath235 .
the @xmath17 components have been omitted since they are identical to the @xmath16 components .
solid , dashed and dotted lines refer to computations obtained with the hlld , hllc and hll solvers . @xmath211 computational zones were used in the computations.,scaledwidth=50.0% ] .
filled circles crosses and plus sign have the same meaning as in fig .
[ fig : st2_close ] .
note the wall heating problem evident in the density profile ( left panel ) .
central and right panels show the transverse field profiles . clearly
the resolution of the slow shocks ( @xmath236 ) improves from hll to hllc and more from hllc to hlld.,scaledwidth=50.0% ] in this test problem we consider the interaction of two oppositely colliding relativistic streams , see also @xcite and mb .
after the initial impact , two strong relativistic fast shocks propagate outwards symmetrically in opposite direction about the impact point , @xmath196 , see fig .
[ fig : st3 ] . being a co - planar problem ( i.e. the initial twist angle between magnetic fields is @xmath221 ) no rotational mode can actually appear .
two slow shocks delimiting a high pressure constant density region in the center follow behind .
although no contact wave forms , the resolution across the slow shocks noticeably improves changing from hll to hllc and from hllc to hlld , see fig .
[ fig : st3 ] or the enlargement of the central region shown in fig .
[ fig : st3_close ] . the resolution across the outermost fast shocks is essentially the same for all solvers .
the spurious density undershoot at the center of the grid is a notorious numerical pathology , known as the wall heating problem , often encountered in godunov - type schemes @xcite .
it consists of an undesired entropy buildup in a few zones around the point of symmetry .
our scheme is obviously no exception as it can be inferred by inspecting see fig .
[ fig : st3 ] .
surprisingly , we notice that error hlld performs slightly better than hllc .
the numerical undershoots in density , in fact , are found to be @xmath237 ( hlld ) and @xmath238 ( hllc ) .
the hll solver is less prone to this pathology most likely because of the larger numerical diffusion , see the left panel close - up of fig .
[ fig : st3_close ] . errors ( for @xmath235 ) are computed using the exact solution available from @xcite which is free from the pathology just discussed . as shown in the bottom left panel of fig .
[ fig : error ] , hlld performs as the best numerical scheme yielding , at the largest resolution employed ( @xmath207 zones ) , l-1 norm errors of @xmath239 to be compared to @xmath238 and @xmath240 of hllc and hll , respectively .
the cpu times for the different solvers on this problem follow the proportion @xmath241 . on @xmath219 computational zones .
the panels are structured in a way similar to fig .
[ fig : st2 ] .
top panel : density , gas pressure and total pressure .
mid panel : @xmath120 , @xmath16 and @xmath17 velocity components .
bottom panel : lorentz factor @xmath0 and transverse components of magnetic field.,scaledwidth=50.0% ] .
the left panel shows the density profile where the two slow shocks and the central contact wave are clearly visible .
central and right panels display the @xmath16 components of velocity and magnetic field .
rotational modes can be most clearly distinguished only with the hlld solver at @xmath242 and @xmath243.,scaledwidth=50.0% ] ) for the four shock tube problems presented in the text as function of the grid resolution .
the different combinations of lines and symbols refer to hlld ( solid , circles ) , hllc ( dashed , crosses ) and hll ( dotted , plus signs).,scaledwidth=50.0% ] the fourth shock tube test is taken from the generic alfvn " test in @xcite .
the breaking of the initial discontinuous states leads to the formation of seven waves . to the left of the contact discontinuity one has a fast rarefaction wave , followed by a rotational wave and a slow shock .
traveling to the right of the contact discontinuity , one can find a slow shock , an alfvn wave and a fast shock .
we plot , in fig .
[ fig : st4 ] , the results computed with the hlld , hllc and hll riemann solvers at @xmath244 , when the outermost waves have almost left the outer boundaries . the central structure ( @xmath245 ) is characterized by slowly moving fronts with the rotational discontinuities propagating very close to the slow shocks . at the resolution employed ( @xmath219 zones ) , the rotational and slow modes appear to be visible and distinct only with the hlld solver , whereas they become barely discernible with the hllc solver and completely blend into a single wave using the hll scheme .
this is better shown in the enlargement of @xmath234 and @xmath235 profiles shown in fig .
[ fig : st4_close ] : rotational modes are captured at @xmath242 and @xmath243 with the hlld solver and gradually disappear when switching to the hll scheme . at the contact wave hlld and hllc
behave similarly but the sharper resolution attained at the left - going slow shock allows to better capture the constant density shell between the two fronts .
our scheme results in the smallest errors and numerical dissipation and exhibits a slightly faster convergence rate , see the plots in the bottom right panel of fig .
[ fig : error ] . at low resolution
the errors obtained with hll , hllc and hlld are in the ratio @xmath246 while they become @xmath247 as the mesh thickens .
correspondingly , the cpu running times for the three solvers at the resolution shown in table [ tab : ic ] have found to scale as @xmath248 .
this example demonstrates the effectiveness and strength of adopting a more complete riemann solver when describing the rich and complex features arising in relativistic magnetized flows .
we have implemented our @xmath47 wave riemann solver into the framework provided by the pluto code @xcite . the constrained transport method is used to evolve the magnetic field .
we use the third - order , total variation diminishing runge kutta scheme together with piecewise linear reconstruction . .
panels on the left show the density map ( at @xmath249 ) in the @xmath250 plane at @xmath251 while panels to the right show the density in the @xmath252 plane at @xmath253.,scaledwidth=50.0% ] but showing the total pressure in the @xmath250 ( left ) and @xmath252 ( right ) panels for the hlld solver.,scaledwidth=50.0% ] ( left ) and @xmath17 ( right ) axis showing the density profiles at different resolutions ( @xmath254 and @xmath255 ) and with different solvers .
solid , dashed and dotted lines are used for the hlld solver whereas plus and star symbols for hll.,scaledwidth=50.0% ] we consider a three dimensional version of the standard rotor problem @xcite .
the initial condition consists of a sphere with radius @xmath256 centered at the origin of the domain taken to be the unit cube @xmath257 ^ 3 $ ] .
the sphere is heavier ( @xmath258 ) than the surrounding ( @xmath259 ) and rapidly spins around the @xmath17 axis with velocity components given by @xmath260 where @xmath261 is the angular frequency of rotation .
pressure and magnetic field are constant everywhere , @xmath262 , @xmath263 . exploiting the point symmetry , we carried computations until @xmath249 at resolutions of @xmath254 and @xmath255 using both the hlld and hll solvers .
we point out that the hllc of mb failed to pass this test , most likely because of the flux - singularity arising in 3d computations in the zero normal field limit . as the sphere starts rotating ,
torsional alfvn waves propagate outward carrying angular momentum to the surrounding medium .
the spherical structure gets squeezed into a disk configuration in the equatorial plane ( @xmath251 ) where the two collapsing poles collide generating reflected shocks propagating vertically in the upper and lower half - planes .
this is shown in the four panels in fig .
[ fig : rotor ] showing the density map in the @xmath250 and @xmath252 planes obtained with hlld and hll and in fig .
[ fig : rotor2 ] showing the total pressure . after the impact a hollow disk enclosed by a higher density shell at @xmath264 forms ( top right panels in fig [ fig : rotor ] ) . in the @xmath250 plane
, matter is pushed in a thin , octagonal - like shell enclosed by a tangential discontinuity and what seems to be a slow rarefaction .
the whole configuration is embedded in a spherical fast rarefaction front expanding almost radially .
flow distortions triggered by the discretization on a cartesian grid are more pronounced with hlld since we expect it to be more effective in the growth of small wavelength modes . in fig .
[ fig : rotor3 ] we compare the density profiles on the @xmath16 and @xmath17 axis for different resolutions and schemes . from both profiles
, one can see that the central region tends to become more depleted as the resolution increases
. inspecting the profiles in the @xmath16 direction ( left panel ) , we observe that hll and hlld tend to under- and over - estimate ( respectively ) the speed of the thin density shell when compared to the reference solution computed with the hlld solver at a resolution of @xmath255 .
the height of the shell peak is essentially the same for both solvers , regardless of the resolution . on the contrary , the right panel of fig .
[ fig : rotor3 ] shows a similar comparison along the vertical @xmath17 axis . at the same resolution , hll under - estimates the density peak located at @xmath265 and almost twice the number of grid zones is needed to match the results obtained with the hlld solver .
the location of the front is approximately the same regardless of the solver . in terms of computational cost ,
integration carried with the hlld solver took approximately @xmath266 that of hll .
this has to be compared with the cpu time required by hll to reach a comparable level of accuracy which , doubling the resolution , would result in a computation @xmath267 as long . in this respect ,
three dimensional problems like the one considered here may prove specially helpful in establishing the trade off between numerical efficiency and accuracy which , among other things , demand choosing between accurate ( but expensive ) solvers versus more diffusive ( cheap ) schemes . at different integration times , @xmath268 .
panels on top ( bottom ) refer to computations accomplished with hlld ( hll ) .
poloidal magnetic field lines overlap.,scaledwidth=50.0% ] the setup , taken from @xcite , consists of a 2d planar cartesian domain , @xmath269 , y\in[-1,1]$ ] with a shear velocity profile given by @xmath270 density and pressure are set constant everywhere and initialized to @xmath271 , @xmath272 , while magnetic field components are given in terms of the poloidal and toroidal magnetization parameters @xmath273 and @xmath274 as @xmath275 where we use @xmath276 , @xmath277 .
the shear layer is perturbed by a nonzero component of the velocity , @xmath278\,,\ ] ] with @xmath279 , while we set @xmath280 .
computations are carried at low ( l , @xmath281 zones ) , medium ( m , @xmath282 zones ) and high ( h , @xmath283 zones ) resolution . at low ( l ) , medium ( m ) and
high ( h ) resolutions .
solid , dashed and dotted lines show results pertaining to hlld , whereas symbols to hll .
bottom : small scale power as a function of time for the kelvin - helmholtz application test .
integrated power is given by @xmath284 where @xmath285 is the complex , discrete fourier transform of @xmath286 taken along the @xmath120 direction . here
@xmath287 is the nyquist critical frequency.,scaledwidth=50.0% ] for @xmath288 the perturbation follows a linear growth phase leading to the formation of a multiple vortex structure . in the high resolution ( h )
case , shown in fig [ fig : kh ] , we observe the formation of a central vortex and two neighbor , more stretched ones .
these elongated vortices are not seen in the computation of @xcite who employed the hll solver at our medium resolution .
as expected , small scale patterns are best spotted with the hlld solver , while tend to be more diffused using the two - wave hll scheme .
the growth rate ( computed as @xmath289 , see top panel in fig .
[ fig : kh2 ] ) , is closely related to the poloidal field amplification which in turn proceeds faster for smaller numerical resistivity ( see the small sub - plot in the same panel ) and thus for finer grids .
still , computations carried with the hlld solver at low ( l ) , medium ( m ) and high ( h ) resolutions reveal surprisingly similar growth rates and reach the saturation phase at essentially the same time ( @xmath290 ) . on the contrary , the saturation phase and the growth rate during the linear phase change with resolution when the hll scheme is employed .
field amplification is prevented by reconnection events during which the field wounds up and becomes twisted by turbulent dynamics . throughout the saturation phase ( mid and right panel in fig [ fig : kh ] )
the mixing layer enlarges and the field lines thicken into filamentary structures .
small scale structure can be quantified by considering the power residing at large wave numbers in the discrete fourier transform of any flow quantity ( we consider the @xmath16 component of velocity ) .
this is shown in the bottom panel of fig [ fig : kh2 ] where we plot the integrated power between @xmath291 and @xmath287 as function of time ( @xmath287 is the nyquist critical frequency ) .
indeed , during the statistically steady flow regime ( @xmath292 ) , the two solvers exhibits small scale power that differ by more than one order of magnitude , with hlld being in excess of @xmath293 ( at all resolutions ) whereas hll below @xmath76 . in terms of cpu time , computations carried out with hlld ( at medium resolution ) were @xmath294 slower than hll . at the resolution of @xmath295 points per beam radius . in clockwise direction , starting from the top right quadrant : density logarithm , gas pressure logarithm , thermal to total pressure ratio and @xmath296 component of magnetic field .
the color scale has been normalized such that the maximum and minimum values reported in each subplots correspond to @xmath297 and @xmath298.,scaledwidth=50.0% ] \times[10,18]$ ] at @xmath299 showing the poloidal magnetic field structure ( in log scale ) for the high and medium resolution runs ( @xmath295 and @xmath300 points per beam radius.,scaledwidth=50.0% ] as a function of time . here
@xmath301 is the poloidal magnetic field .
solid , dashed and dotted lines refers to computations carried out with hlld , whereas symbols give the corresponding results obtained with hll.,scaledwidth=50.0% ] as a final example , we consider the propagation of a relativistic magnetized jet . for illustrative purposes , we restrict our attention to axisymmetric coordinates with @xmath302 $ ] and @xmath303 $ ] .
the jet initially fills the region @xmath304 with density @xmath305 and longitudinal ( @xmath17 ) velocity specified by @xmath306 ( @xmath307 ) .
the magnetic field topology is described by a constant poloidal term , @xmath308 , threading both the jet and the ambient medium and by a toroidal component @xmath309 with @xmath310 where @xmath311 is the magnetization radius and @xmath312 is a constant and vanishes outside the nozzle .
the thermal pressure distribution inside the jet is set by the radial momentum balance , @xmath313 yielding @xmath314\,,\ ] ] where @xmath315 is the jet / ambient pressure at @xmath316 and is recovered from the definition of the mach number , @xmath317 , with @xmath318 and @xmath319 , although we evolve the equations using the approximated synge gas equation of state of @xcite .
the relative contribution of the two components is quantified by the two average magnetization parameters @xmath320 @xmath321 yielding @xmath322 where for any quantity @xmath323 , @xmath324 gives the average over the jet beam @xmath325 $ ] .
we choose @xmath326 , @xmath327 , thus corresponding to a jet close to equipartition .
the external environment is initially static ( @xmath328 ) , heavier with density @xmath329 and threaded only by the constant longitudinal field @xmath308 .
pressure is set everywhere to the constant value @xmath315 .
we carry out computations at the resolutions of @xmath330 and @xmath295 zones per beam radius ( @xmath316 ) and follow the evolution until @xmath299 . the snapshot in fig .
[ fig : jet ] shows the solution computed at @xmath299 at the highest resolution . the morphological structure is appreciably affected by the magnetic field topology and by the ratio of the magnetic energy density to the rest mass , @xmath331 .
the presence of a moderately larger poloidal component and a small poynting flux favor the formation of a hammer - like structure rather than a nose cone ( see * ? ? ?
* ; * ? ? ?
at the termination point , located at @xmath332 , the beam strongly decelerates and expands radially promoting vortex emission at the head of the jet . close to the axis
, the flow remains well collimated and undergoes a series of deceleration / acceleration events through a series of conical shocks , visible at @xmath333 . behind these recollimation shocks
, the beam strongly decelerates and magnetic tension promotes sideways deflection of shocked material into the cocoon .
the ratio @xmath334 ( bottom left quadrant in fig [ fig : jet ] ) clearly marks the kelvin - helmholtz unstable slip surface separating the backflowing , magnetized beam material from the high temperature ( thermally dominated ) shocked ambient medium . in the magnetically dominated region turbulence
dissipate magnetic energy down to smaller scales and mixing occurs .
the structure of the contact discontinuity observed in the figures does not show suppression of kh instability .
this is likely due to the larger growth of the toroidal field component over the poloidal one @xcite .
however we also think that the small density ratio ( @xmath335 ) may favor the growth of instability and momentum transfer through entrainment of the external medium @xcite . for the sake of comparison
, we also plot ( fig [ fig : jet2 ] ) the magnitude of the poloidal magnetic field in the region @xmath336 $ ] , @xmath337 $ ] where turbulent patterns have developed . at the resolution of @xmath295 points per beam radius
, hlld discloses the finest level of small scale structure , whereas hll needs approximately twice the resolution to produce similar patterns .
this behaviour is quantitatively expressed , in fig [ fig : jet3 ] , by averaging the gradient @xmath338 over the volume .
roughly speaking , hll requires a resolution @xmath339 that of hlld to produce pattern with similar results .
a five - wave hlld riemann solver for the equations of relativistic magnetohydrodynamics has been presented .
the solver approximates the structure of the riemann fan by including fast shocks , rotational modes and the contact discontinuity in the solution .
the gain in accuracy comes at the computational cost of solving a nonlinear scalar equation in the total pressure . as such
, it better approximates alfvn waves and we also found it to better capture slow shocks and compound waves .
the performance of the new solver has been tested against selected one dimensional problems , showing better accuracy and convergence properties than previously known schemes such as hll or hllc .
applications to multi - dimensional problems have been presented as well .
the selected tests disclose better resolution of small scale structures together with reduced dependency on grid resolution .
we argue that three dimensional computations may actually benefit from the application of the proposed solver which , albeit more computationally intensive than hll , still allows to recover comparable accuracy and resolution with a reduced number of grid zones . indeed , since a relative change @xmath340 in the mesh spacing results in a factor @xmath341 in terms of cpu time , this may largely favour a more sophisticated solver over an approximated one .
this issue , however , need to receive more attention in forthcoming studies .
anile , m. , & pennisi , s. 1987 , ann . inst .
henri poincar , 46 , 127 anile , a. m. 1989 , relativistic fluids and magneto - fluids ( cambridge : cambridge university press ) , 55 balsara , d. s. 2001 , apjs , 132 , 83 berger , m. j. and colella , p. j. comput .
phys . , 82 , pp .
64 - 84 , 1989 .
brio , m. , & wu , c .- c .
1988 , j. comput .
phys . , 75 , 400 bucciantini , n. , & del zanna , l. 2006 , astronomy & astrophysics , 454 , 393 s.f .
davis , siam j. sci .
statist . comput . 9 ( 1988 ) 445 .
einfeldt , b. , munz , c.d . , roe , p.l . , and sj " ogreen , b. 1991 , j. comput .
phys . , 92 , 273 del zanna , l. , bucciantini , n. , & londrillo , p. 2003 , astronomy & astrophysics , 400 , 397 ( dzbl ) del zanna , l. , zanotti , o. , bucciantini , n. , & londrillo , p. 2007 , astronomy & astrophysics , 473 , 11 gammie , c. f. , mckinney , j. c. , & tth , g. 2003 , the astrophysical journal , 589 , 444 gehmeyr , m. , cheng , b. , & mihalas , d. 1997 , shock waves , 7 , 255 giacomazzo , b. , & rezzolla , l. 2006 , journal of fluid mechanics , 562 , 223 gurski , k.f .
2004 , siam j. sci .
comput , 25 , 2165 harten , a. , lax , p.d . , and van leer , b. 1983 , siam review , 25(1):35,61 honkkila , v. , & janhunen , p. 2007
, journal of computational physics , 223 , 643 jeffrey a. , taniuti t. , 1964 , non - linear wave propagation . academic press , new york keppens , r. , meliani , z. , van der holst , b. , & casse , f. 2008 , astronomy & astrophysics , 486 , 663 koldoba , a. v. , kuznetsov , o. a. , & ustyugova , g. v. 2002 , mnras , 333 , 932 komissarov , s. s. 1997 , phys .
a , 232 , 435 komissarov , s. s. 1999 , mnras , 308 , 1069 leismann , t. , antn , l. , aloy , m. a. , mller , e. , mart , j. m. , miralles , j. a. , & ibez , j. m. 2005 , astronomy & astrophysics , 436 , 503 li s. , 2005 , j. comput .
344 - 357 lichnerowicz , a. 1976 , journal of mathematical physics , 17 , 2135 lichnerowicz , a. 1967 , relativistic hydrodynamics and magnetohydrodynamics , new york : benjamin , 1967 mignone , a. , & bodo , g. 2005 , mnras , 364 , 126 mignone , a. , massaglia , s. , & bodo , g. 2005 , space science reviews , 121 , 21 mignone , a. , & bodo , g. 2006 , mnras , 368 , 1040 ( mb ) mignone , a. , bodo , g. , massaglia , s. , matsakos , t. , tesileanu , o. , zanni , c. , & ferrari , a. 2007 , astrophysical journal supplement , 170 , 228 mignone , a. , & mckinney , j. c. 2007 , mnras , 378 , 1118 t. miyoshi , k. kusano , k. , j. comp . phys . 208 ( 2005 ) 315 ( mk ) noh , w.f .
1987 , j. comput .
phys . , 72,78 press , w. , s. teukolsky , w. vetterling , and b. flannery ( 1992 ) .
numerical recipes in c ( 2nd ed . ) .
cambridge , uk : cambridge university press .
romero , r. , mart , j. m. , pons , j. a. , ibez , j. m. , & miralles , j. a. 2005 , journal of fluid mechanics , 544 , 323 rossi , p. , mignone , a. , bodo , g. , massaglia , s. , & ferrari , a. 2008 , astronomy & astrophysics , 488 , 795 toro , e. f. , spruce , m. , and speares , w. 1994 , shock waves , 4 , 25 toro , e. f. 1997 , riemann solvers and numerical methods for fluid dynamics , springer - verlag , berlin van der holst , b. , keppens , r. , & meliani , z. 2008 , arxiv:0807.0713
left and right states across a rotational discontinuity can be found using the results outlined in [ sec : alfven ] .
more specifically , we construct a family of solutions parameterized by the speed of the discontinuity @xmath124 and one component of the tangential field on the right of the discontinuity . our procedure can be shown to be be equivalent to that of @xcite .
specifically , one starts by assigning @xmath342 on the left side of the front ( @xmath343 ) together with the speed of the front , @xmath124 .
note that @xmath26 can not be freely assigned but must be determined consistently from eq .
( [ eq : kvb ] ) . expressing @xmath127 ( @xmath344 ) in terms of @xmath345 and @xmath26 and substituting back in the @xmath346 component of ( [ eq : kvb ] )
, one finds that there are two possible values of @xmath26 satisfying the quadratic equation @xmath347 where the coefficients of the parabola are @xmath348 and @xmath349 with @xmath350 , @xmath351 and @xmath0 being the lorentz factor .
the transverse components of @xmath119 are computed as @xmath352 on the right side of the front , one has that @xmath2 , @xmath233 , @xmath113 , @xmath26 and @xmath119 are the same , see [ sec : alfven ] . since the transverse field
is elliptically polarized @xcite , there are in principle infinite many solutions and one has the freedom to specify , for instance , one component of the field ( say @xmath353 ) .
the velocity @xmath354 and the @xmath17 component of the field can be determined in the following way .
first , use equation ( [ eq : vc ] ) to express @xmath355 ( @xmath356 ) as function of @xmath357 for given @xmath358 and @xmath353 . using the jump condition for the density together with the fact that @xmath2 is invariant
, we solve the nonlinear equation @xmath359 whose roots gives the desired value of @xmath357 . | we present a five - wave riemann solver for the equations of ideal relativistic magnetohydrodynamics .
our solver can be regarded as a relativistic extension of the five - wave hlld riemann solver initially developed by miyoshi and kusano for the equations of ideal mhd .
the solution to the riemann problem is approximated by a five wave pattern , comprised of two outermost fast shocks , two rotational discontinuities and a contact surface in the middle .
the proposed scheme is considerably more elaborate than in the classical case since the normal velocity is no longer constant across the rotational modes .
still , proper closure to the rankine - hugoniot jump conditions can be attained by solving a nonlinear scalar equation in the total pressure variable which , for the chosen configuration , has to be constant over the whole riemann fan . the accuracy of the new riemann solver is validated against one dimensional tests and multidimensional applications . it is shown that our new solver considerably improves over the popular hll solver or the recently proposed hllc schemes . [ firstpage ] hydrodynamics - mhd - relativity - shock waves - methods : numerical |
since its inception 13 years ago , the gauge / gravity duality @xcite has engendered vigorous exploration of properties of non - abelian gauge theories by studying their gravitational dual .
one of the crucial attributes is the strong / weak coupling nature of the duality : in the limit of strong coupling and high rank of the gauge group , the theory is described in terms of classical gravity on a higher - dimensional asymptotically anti de sitter ( ads ) background .
such dual picture has offered invaluable insight into the behavior of the field theory in the regime of strong coupling where quantum effects dominate and traditional methods prove inadequate .
one of the most natural questions concerning properties of a strongly - coupled medium concerns its ability to transmit signals .
in particular , how does ` radiation ' propagate in such a medium ?
while this question has been well - studied for numerous weakly coupled systems , the strong coupling case is much more interesting , and of potential relevance to experimentally attainable settings . of course , to achieve closest contact with systems such as the quark gluon plasma produced at the relativistic heavy ion collider , one would like to study strongly coupled field theories at high temperature
. nevertheless , while perhaps less indicative of real - world systems , the zero - temperature setting is also of active interest , since it provides an excellent toy model for the more realistic cases while being amenable to exact calculations . from the gravitational dual standpoint , zero temperature constitutes a drastic simplification because the vacuum of the gauge theory corresponds to pure ads , whereas a thermal state corresponds to a schwarzschild - ads black hole , which is both geometrically and causally more complicated .
for example , while an exact solution for a string with its endpoint following a given trajectory is known in pure ads for arbitrary endpoint trajectories , in schwarzschild - ads an analogous solution is known only in special cases with high enough symmetry . nevertheless , once a given setup has been well - understood at zero temperature
, one can then use this as a crutch to extend the analysis to non - zero temperature .
a very interesting recent study of radiation in strongly coupled gauge theory at zero temperature was undertaken in @xcite for the case of a quark in uniform circular motion .
particle of an @xmath1 hypermultiplet ) , coupled to high gauge group rank ( @xmath2 ) limit of 3 + 1 dimensional super yang - mills , at finite but large t hooft coupling @xmath3 . ]
the gravitational dual consists of a fundamental string in ads spacetime whose endpoint is anchored at the quark on the boundary .
the quark s trajectory then determines the string s embedding in ads .
the string , having non - trivial stress tensor supported on its world - volume , backreacts on the bulk spacetime , which in turn induces a non - trivial stress tensor in the boundary field theory .
the authors of @xcite compute the corresponding boundary energy density by solving linearized einstein s equations in the bulk .
they find a rather surprising result : the radiation closely resembles synchrotron radiation exhibited in classical electrodynamics and weakly - coupled sym .
in particular , the radiation pulse does not broaden as it propagates outward , despite the field theory being strongly coupled .
indeed , this result is just as remarkable from the gravitational dual standpoint , since it implies that the backreaction effects from string bits deep in the bulk are somehow ` beamed ' towards the boundary , contradicting naive expectations from the scale / radius ( or uv / ir ) duality @xcite , according to which an excitation deeper in the bulk should be represented by a more dispersed signal in the boundary field theory .
this mysterious beaming effect in fact prompted the present study , which was recently summarized in @xcite .
in particular , our main goal is to attain better geometrical understanding of how the backreaction effects of a bulk string manage to produce a tightly collimated beam on the boundary .
as it turns out , this has the added benefit of simplifying the stress tensor calculation itself . while far simpler than any conceivable direct method within the field theory ,
the calculation of @xcite is still fairly involved . here
we propose a much easier method to obtain the boundary stress tensor , in a straightforward and computationally non - intensive manner which essentially reduces to finding geodesics in ads .
this simplification applies to the regime where the string is sufficiently relativistic , which , as we discuss below , includes the present case of interest , and indeed occurs quite generically for bulk dual describing radiation from an accelerating quark .
it rests on the observation that in such a regime , the backreaction of the string is captured by a superposition of gravitational shock waves sourced by individual string bits . justifying and testing this proposal against the results of @xcite will be the main focus of the present work .
once the string s backreaction is understood in terms of gravitational shock waves , it elucidates the mechanism producing the observed non - dispersive character of the boundary radiation .
this is due to the geometrical nature of the gravitational shock waves @xcite : each shock wave is supported on a null hypersurface in the spacetime , which means that the influence of the backreaction is not allowed to disperse as naively expected . as we will demonstrate
, it is the combined effect from the set of all such shock waves produced by the string that leads to the observed phenomenon of the radiation pulses having correct width and propagating at the speed of light without broadening .
this has further interesting implications for the scale / radius duality , as discussed below .
the plan of this paper is as follows . in
we introduce the circling quark configuration and the corresponding bulk dual in more detail .
this allows us to review the key observations of @xcite , which provide a sharp testing - ground for our proposal . in we detail our proposal for the beaming mechanism .
we first justify our expectations based on general principles and then explain how to calculate the backreaction explicitly .
the actual calculations are relegated to the appendices , in order to maintain the flow of the main text .
having outlined the construction , in we present the results of implementing this construction .
we first do so numerically , by plotting various aspects of our prediction for the energy density induced on the boundary .
we then determine the salient features of this energy density analytically in the asymptotic regime , which allows more detailed quantitative comparison with @xcite .
we summarize our findings in and end by discussing further applications and extensions of our construction .
let us start by reviewing the essential ingredients of the work of @xcite , which uses the ads / cft correspondence to study the radiation from a quark in uniform circular motion in a strongly coupled field theory at zero temperature
. in the gravitational dual , such set - up corresponds to a string in ads , ending on the quark .
as indicated above , the gravitational backreaction of the string deforms the bulk geometry away from ads , and this in turn induces a non - trivial stress tensor on the boundary , which captures both the motion of the quark , as well as the resulting radiation . to analyze the set - up more explicitly ,
let us first fix the notation .
we will consider the field theory to be formulated on minkowski spacetime .
] so the notion of radiation is best - defined in the asymptotic region @xmath4 .
the source of the radiation is a test quark moving on a circle of radius @xmath5 with constant angular velocity @xmath6 , so that its trajectory is given by @xmath7 and the quark moves with constant speed @xmath8 , with the corresponding lorentz factor @xmath9 . in the gravitational dual ,
this set - up corresponds to a trailing string moving in ads . in the poincare ( or fefferman - graham ) coordinates @xmath10 , where ads ( of unit size ) has the metric @xmath11 \ , \label{adsmet}\ ] ] @xmath12 corresponds to the bulk radial coordinate , with @xmath13 representing the ads boundary . parameterizing the string worldsheet by @xmath14 and @xmath12 , @xcite write the string embedding as @xmath15 and solve the nambu - goto action @xmath16 , which for the induced metric on the worldsheet @xmath17 with @xmath18 given by becomes @xmath19 to determine the profile functions @xmath20 and @xmath21 : @xmath22 from these we can see that the string has the shape of a rigidly rotating helix , flaring - out into the bulk at a rate @xmath23 .
spacetime corresponding to quark undergoing a circular motion .
several curves are drawn , corresponding to different quark velocities @xmath24 ; specifically , @xmath25 ( vertical red line ) , @xmath26 ( yellow ) , @xmath27 ( green ) , @xmath28 ( blue ) , and @xmath29 ( horizontal purple spiral ) .
the small black circle on the top indicates the quark s trajectory on the ads boundary ; the vertical direction corresponds to the bulk radial direction.,width=240 ] ( 0,0 ) ( -9,5.6)(1,0)1.44 ( -9,5.6)(-1,-1)1 ( -9,5.6)(0,-1)1.44 ( -9.1 , 3.7)@xmath12 ( -10.4 , 4.5)@xmath30 ( -7.4 , 5.5)@xmath31 this is illustrated in for several values of the quark velocity @xmath24 .
we see that in the limit @xmath32 , the string hugs the boundary , so that we expect the boundary stress tensor being localized just above the string , as is indeed confirmed by explicit calculation .
we now discuss several key aspects of the geometry in a bit more detail .
[ [ worldsheet - horizon ] ] worldsheet horizon : + + + + + + + + + + + + + + + + + + + let us first consider the causal structure of the worldsheet metric .
it is intriguing to note that despite the bulk having the causally trivial structure of pure ads , the worldsheet metric corresponds to a black hole geometry , with an event horizon at @xmath33 ( this is easy to see by finding the induced metric on the worldsheet @xmath34 explicitly , @xmath35 and noting that since the metric is static , the killing horizon at @xmath36 coincides with the event horizon . )
this suggests that beyond @xmath37 , the boundary stress tensor will describe radiation form the quark rather than the gluonic ` halo ' of the quark itself .
the total power radiated by the quark coincides with the energy flux down the string , which @xcite evaluate to be @xmath38 , where @xmath39 is the quark s proper acceleration .
[ [ transverse - velocity ] ] transverse velocity : + + + + + + + + + + + + + + + + + + + + an important quantity in what follows is the string s velocity .
since a fundamental string has no longitudinal excitations , a natural quantity to describe the string s motion is the transverse velocity of the string , @xmath40 . in the static gauge chosen above with @xmath41 ,
@xmath42 is a purely spatial vector , given by the component of @xmath43 perpendicular to the string , @xmath44 the time dependence is trivial since the string rotates rigidly , and using the parameterization , in the coordinates @xmath45 , we find that @xmath46 we see that the direction of the transverse velocity points towards larger @xmath47 , @xmath48 , and @xmath12 . in particular , note that @xmath42 points _ away _ from the ads boundary . in the interesting regime of large @xmath49 and @xmath50 , the leading @xmath45 components of @xmath51 approach @xmath52 . of greater interest
is the magnitude of the transverse velocity .
using we can evaluate , we use only the flat spatial metric without its conformal factor @xmath53 because the @xmath14 direction is likewise scaled by the same conformal factor .
] this to be @xmath54 it is then easy to see that the transverse speed of the string increases monotonically with @xmath12 , and asymptotes the speed of light deep in the bulk . for future use , also note that the product of this speed and the associated lorentz boost factor @xmath55 , which characterizes to what extend is the string moving relativistically , is given by @xmath56 with the worldsheet horizon @xmath57 defined in .
[ [ boundary - stress - tensor ] ] boundary stress tensor : + + + + + + + + + + + + + + + + + + + + + + + we now summarize the key results of @xcite . to find how the radiation induced by the quark s acceleration propagates through the strongly coupled medium ,
@xcite consider the energy density @xmath58 , where @xmath59 is the boundary stress tensor .
since the on - shell bulk gravitational action is the generating functional for the boundary stress tensor , @xmath59 can be read - off from the asymptotic form of the bulk metric deformation . in particular , using the bulk stress tensor produced by the string , @xcite solve the linearized einstein s equations near the boundary .
this yields an analytic few - line expression for @xmath60 ( see eq.(3.70 ) of @xcite , plotted in their fig.4 ) in terms of retarded time @xmath61 which is evaluated numerically .
as remarked above , this boundary energy density turns out to have rather surprising properties : it is mainly supported on a thin spiral rotating rigidly at constant angular velocity , with the energy pulses _ not _ widening in the boundary - radial direction . ; at non - zero temperature , these pulses would widen , slow down , and eventually thermalize . however , for @xmath62 , the synchrotron radiation persists on length scales smaller than the thermal scale @xmath63 . ]
the radiation pattern is in fact in close quantitative agreement with that of a corresponding radiation at weak coupling , which likewise mimics that of synchrotron radiation in classical electrodynamics .
more specifically , the energy density @xmath64 exhibits the following salient features . *
each pulse remains constant width as it propagates outward , i.e. it does not broaden .
* a pulse propagates outward at the speed of light , independently of the quark velocity @xmath24 .
if we change the quark velocity @xmath24 , the quatitative features of @xmath64 change in response : * the spatial separation between the spiral arms of @xmath64 at a fixed time narrows with increasing @xmath24 as @xmath65 , as implied by the previous point .
* pulse width decreases with increasing @xmath24 as @xmath66 .
* @xmath67 decreases off the orbital plane , with the characteristic width narrowing for increasing @xmath24 as @xmath68 .
as argued in @xcite , these features , most notably the first point , are quite surprising from the field theory point of view : since the radiation is strongly coupled , one might have expected it to diffuse and isotropize .
nevertheless , we will see that in fact all of these features are reproduced by our simple method of estimating the backreaction of a relativistic string , which we now discuss .
the curious features of the radiation propagating through a strongly coupled medium summarized above are no less surprising and fascinating from the bulk standpoint : why does the gravitational backreaction of the bulk string behave in such a sharply - localized fashion ? as pointed out in the introduction , this is at odds with the naive expectations from the scale / radius duality that the deeper parts of the string would produce more diffuse signals on the boundary . in this section
, we propose a general mechanism which leads to the observed localization ; we refer to this mechanism as ` beaming ' .
first of all , note that large part of the string is moving relativistically .
this is evident from , and more conveniently quantified by : as long as @xmath69 ( for any @xmath12 ) or @xmath70 ( for any @xmath24 ) , the string s transverse velocity is relativistic @xmath71 .
the object then is to understand how such a relativistic string backreacts on the spacetime .
we use a well - known fact that the backreaction of a massless particle moving at the speed of light is given by a _ gravitational shock wave _ @xcite .
such metric perturbation is supported on a null plane transverse to the particle s velocity , analogously to the behaviour of the electric field for an infinitely - boosted charge in classical electrodynamics .
accordingly , the backreaction of all ` string bits ' ( viewed as a continuum of particles distributed along the string ) , moving relativistically in the direction of @xmath42 , should be given by a superposition of such gravitational shock waves , one for each particle .
we therefore propose that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the backreaction of the full string is well - approximated by a linear superposition of gravitational shock waves normal to the string s transverse velocity .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ making this proposal more explicit and testing it will be the main focus of this work . for compactness of notation , we will henceforth use gsw to abbreviate gravitational shock wave " and denote the superposition of all gravitational shock waves sourced by the string as . in the remainder of this section ,
we first justify why is a good approximation to the string s backreaction , we then explain how to find the explicitly , and finally we indicate how to extract the boundary stress tensor induced by such . since the exact result for the boundary energy density @xmath64 is known from the calculation of @xcite , we can readily test our proposal by comparing our results with @xcite , which will be carried out in detail in . our proposal that the string s backreaction is well - approximated by superposition of gravitational shock waves , , rests on several important assumptions , which we will try to justify in turn . [ [ neglecting - string - tension ] ] neglecting string tension : + + + + + + + + + + + + + + + + + + + + + + + + + + firstly , we require that the interaction between the individual string bits does not contribute appreciably to the backreaction of the string
. admittedly such an assumption would fail for a slowly - moving string where longitudinal boost invariance of the string s stress tensor implies that the ( negative ) pressure is of the same magnitude as the energy density .
however , under transverse boosts with lorentz factor @xmath55 , energy density is enhanced by a factor of @xmath72 compared to the pressure which captures the effect of interactions .
in fact , this is not surprising , given the parton model of the string @xcite .
this means that our assumption of non - interaction becomes arbitrarily good for high enough string s transverse velocity , so that in the relativistic regime we can indeed treat the string as composed of relativistic particles .
[ [ gsw - from - each - string - bit ] ] gsw from each string bit : + + + + + + + + + + + + + + + + + + + + + + + + + the second assumption is that each string bit produces a backreaction in the form of gsw , which is valid as long as the string s transverse velocity is sufficiently relativistic . as observed above , guarantees @xmath73 whenever either the quark velocity is sufficiently relativistic ( @xmath69 ) or the string bit under consideration is sufficiently deep in the bulk ( @xmath70 ) , which quantifies the regime of validity of our second assumption .
we note that this regime of validity happily coincides with our regime of interest : in particular , we are interested in @xmath64 at @xmath74 where the notion of radiation is well defined , which indeed originates from the string bits at @xmath70 . [ [ linear - superposition - of - gsws ] ] linear superposition of gsws : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + finally , we assume that the individual gsws superpose linearly , despite the nonlinearities of general relativity .
although this may not hold for arbitrary gsws for example head - on collisions can produce black holes ( see e.g. @xcite , and more recent work reviewed in @xcite ) it does hold in our case : examining the geometry in detail , we can show that the individual gsws intersect mildly enough ( in the sense indicated in the next section ) , and their interaction is strongest at the string itself . because we started with a nearly - test string , the gsws produced by its backreaction will themselves backreact parametrically less
( by @xmath75 in the present context ) .
hence , given that @xcite justifiably used linearized gravity to calculate the string s backreaction , it is certainly valid to use linearized gravity for the gsws backreaction .
this justifies superposing individual gsws to estimate the backreaction of the string .
having justified why we may take the string to backreact in a manner which is captured by , we now proceed to construct such a superposition explicitly .
the key step is understanding how a single string bit , moving relativistically in the direction of @xmath42 , backreacts on ads .
such a metric is given by the ads analog of the aichelburg - sexl metric @xcite describing a gravitational shock wave in flat spacetime .
by @xcite and studied further in @xcite . ]
geometrically , a gsw in flat spacetime is supported on a ` transverse null plane ' ; in ads , which is conformally flat , there is a corresponding transverse null hypersurface .
although one could construct this explicitly by a similar procedure to @xcite , we will instead use a trick , detailed in .
we observe that a spatial slice of gsw at a fixed time forms a hypersurface which is generated by spacelike geodesics , emanating from its source in a direction normal to the source s velocity @xmath42 . while this is readily apparent in the coordinates used in @xcite simply by symmetry arguments , the statement as such is a geometric one , and therefore holds in any coordinate system .
this means that to find the support of a gsw in ads at constant @xmath14 , written in the form , we only need to find the constant-@xmath14 spacelike geodesics , whose initial position @xmath76 lies along the string and initial velocity points normally to @xmath42 . as we show explicitly in
, such geodesics form a ` hemisphere ' in the @xmath77 space , whose equator lies on the boundary @xmath13 and whose radius and position is determined by the geometry of the string .
the ` intersection ' of the gsw with the ads boundary determines the spatial support of the boundary energy density @xmath64 at a fixed time , which forms a sphere in @xmath78 space and therefore a circle on the orbital @xmath79 plane . apart from knowing the support of @xmath64 sourced by a given string bit
, we are interested in knowing its magnitude .
this will be discussed in , but as a zeroth order estimate , we can determine where @xmath64 dominates . at each bulk radius @xmath12 ,
the dominant contribution of the gsw lies along the direction nearest to the source ; geometrically this represents a ` steepest ' bulk spacelike geodesic .
the endpoint @xmath80 of this geodesic singles out a position at which @xmath64 peaks .
the construction indicated above and detailed in is illustrated in ( left ) , which shows a single gsw sourced by the string bit at @xmath76 , with the green geodesic and its endpoint @xmath80 corresponding to the dominant contribution of the gsw .
so far , we have indicated how to construct the gsw sourced by a particular string bit , but we can simply repeat the construction for all string bits , as indicated in the right panel of ( which is further tilted for greater ease of visualization ) for a representative set of gsws . ) .
we pick an initial point @xmath76 ( here we chose @xmath81 ) .
we then construct the transverse velocity ( red line segment ) and consider all geodesics emanating perpendicularly to it ( blue line segments ) .
this generates the full gsw ( surface ) , with the shortest geodesic to the boundary ( green curve ) ending on the ( blue point ) @xmath80 . *
right : * the dominant contribution @xmath80 from a set of representative points along the bulk string .
we can see the beginnings of the boundary spiral along which expect @xmath64 to peak ( blue curve ) .
, title="fig:",width=336 ] ) .
we pick an initial point @xmath76 ( here we chose @xmath81 ) .
we then construct the transverse velocity ( red line segment ) and consider all geodesics emanating perpendicularly to it ( blue line segments ) .
this generates the full gsw ( surface ) , with the shortest geodesic to the boundary ( green curve ) ending on the ( blue point ) @xmath80 . *
right : * the dominant contribution @xmath80 from a set of representative points along the bulk string .
we can see the beginnings of the boundary spiral along which expect @xmath64 to peak ( blue curve ) .
, title="fig:",width=307 ] we see that the ( blue ) points @xmath82 on the boundary , indicating the dominant contribution to @xmath64 , indeed lie on a spiral , as calculated in @xcite . in
we will examine this spiral in greater detail .
now that we have determined the bulk locus of a given gravitational shock wave , we can trivially extract the _ support _ of the boundary stress tensor induced by such a metric deformation .
the essential tool from ads / cft is that the asymptotic form of the bulk metric deformation @xmath83 away from ads determines the boundary stress tensor expectation value @xmath84 as we review in ( cf
. equations and ) .
in particular , we know that @xmath85 only where the gsw intersects the boundary . however , to determine the value of @xmath84 at this locus , we need to make additional use of the gsw s _
profile_. here we describe a way to estimate this .
the proper calculation would be to generate the explicit gsw metric corresponding to a source propagating along some arbitrary null direction in ads , for instance by boosting a schwarzschild - ads black hole to speed of light keeping the total energy fixed .
once obtained , we could calculate the boundary stress tensor induced by this metric , either directly via a brown - york type procedure using the covariant form of @xcite , or by first recasting the metric into the fefferman - graham form by a suitable change of coordinates and then reading off the stress tensor as in @xcite . instead
, here we take a shortcut , by using previous calculations @xcite as described in .
in particular , we recast the induced stress tensor on the boundary in terms of geometrical properties of the geodesics described above . to write this compactly , consider a gsw at a given time @xmath14 , sourced at the bulk point @xmath76 .
the boundary light - up ( i.e. induced energy density @xmath64 ) occurs at all points @xmath86 reachable from @xmath76 by a spacelike geodesic , normal to the source s velocity .
let the regularized proper length along such geodesics be denoted by @xmath87 .
in we argue that the profile for the boundary energy density @xmath88 is given by @xmath89 we will use this expression at the end of below .
we see that the profile of @xmath64 peaks sharply at a point @xmath80 on the gsw located closest to the source . for example , for a source moving parallel to ads boundary along a trajectory @xmath90 , @xmath91 , and @xmath92 , the induced energy density at time @xmath14 is @xmath93 once we have expressed the energy density due to a single string bit at @xmath92 , we integrate this over all values of @xmath70 .
in fact , due to the transverse localization of each gsw , only a small range of @xmath12 contributes to the energy density in given small neighbourhood .
in the preceding section we have explained how to obtain the boundary energy density @xmath64 sourced by corresponding to a relativistic string in the bulk .
here we carry out this construction in detail for the circling quark discussed in , and compare with the results of @xcite . we will first present our results graphically to confirm the qualitative features of @xmath64 .
although one could in principle obtain the corresponding expressions analytically , they would be far too lengthy and unilluminating to write explicitly .
however , since we are mainly interested in the behavior of @xmath64 in the asymptotic region ( where the radiation is cleanly separated from the background ) , we can simplify these expressions by focusing on the @xmath94 regime .
such asymptotic analysis will be carried out in , where we will extract the key features of @xmath64 quantitatively . since the time dependence is given by rigid rotation , it suffices to consider @xmath64 at @xmath95 . rather than estimating @xmath96 using the full profile , it is instructive to separate the analysis of @xmath64 into several parts .
we will first consider the dominant contribution to @xmath64 which will confirm the spiral shape and allow us to extract the spiral arm spacing .
we will then consider the full support of @xmath64 , which will reveal the @xmath24-dependence of the width of the spiral arm .
we will also briefly consider the spiral peak profile .
finally , we will examine the polar angle dependence of @xmath67 .
[ [ spiral - spacing ] ] spiral spacing : + + + + + + + + + + + + + + + let us first consider the boundary curve where @xmath64 is dominant , corresponding to the spiral indicated in the right panel of .
the bulk counterpart is given by maximizing the metric deformation @xmath97 at each @xmath12 . and
@xmath98 ) .
the surface plotted is the dominant part of , whose intersection with the boundary gives the expected spiral curve ( the thin red curve at the top is its approximation discussed in below ) .
* right : * spiral induced on boundary for two representative velocities : @xmath99 ( orange ) and @xmath100 ( purple).,width=268 ] for each string bit , parameterized by @xmath101 , this occurs along the steepest geodesic ( indicated by the green curve in ) .
the family of the dominant parts of gsw(@xmath101 ) is plotted in ( left ) .
its intersection with the boundary @xmath13 ( top of the plot ) generates the spiral formed by @xmath82 .
such a spiral is plotted in ( right ) for two distinct velocities , @xmath99 and @xmath100 .
both spirals have qualitatively correct behaviour , with the spiral arms growing linearly with angle , but this rate of growth scales inversely with @xmath24 . in particular , we find that the distance between the spiral arms for general parameters @xmath5 and @xmath24 is given by @xmath102 , in agreement with @xcite . this will be confirmed analytically in .
note that periodicity of the quark s motion implies that the spiral rotates rigidly with periodicity @xmath103 , which in turn implies that the spiral arms move outward at the speed of light , independently of the velocity @xmath24 .
this provides the most crucial check of the beaming mechanism .
[ [ spiral - width ] ] spiral width : + + + + + + + + + + + + + so far we have confirmed that the dominant contribution to the boundary energy density induced by indeed lies on a spiral with correct spacing of spiral arms and time dependence . in order to extract further features of the spiral , such as its width
, we need to consider more than just the dominant contribution of each gsw .
let us therefore focus on the full support of the gsw , but instead ignore the variations of its intensity .
in other words , the boundary light - up sourced by each string bit is a full sphere rather than a single point . on the equatorial @xmath79 plane of the boundary
such light - up is a circle , and the superposition of all circles indicates of the support of @xmath104 due to the entire string .
figure [ f : gswsuperpos ] shows the boundary light - up from , for quark velocity @xmath105 ( left ) and @xmath106 ( right ) . while only the individual gsws are plotted ( each one a circle with radius and origin approximated by , as discussed in ) , we see that an overall spiral pattern indeed emerges from the superposition of such circles .
this spiral of course agrees with that seen in ( approximately given by ) , but now we can also see that it has a sharp width @xmath107 , which gets smaller with increasing @xmath24 . numerically we confirm agreement with the scaling @xmath108 found in @xcite and familiar in weakly - coupled contexts ; in we will be able to go further in extracting both this scaling and its coefficient . ,
@xmath109 , and @xmath13 from gsws originating at various points along the string ( corresponding to quark velocity @xmath105 * ( left ) * and @xmath106 * ( right ) * and @xmath110 in increments of 0.3 ) .
each gsw forms a circle approximated by ; but from the superposition of all such circles , there clearly emerges an overall spiral pattern which matches that of , approximated by . , height=268 ]
the most striking feature of figure [ f : gswsuperpos ] is that the spiral width ( corresponding to the energy pulse discussed in section [ s : setup ] ) does _ not _ broaden with radius @xmath47 , but rather remains constant along the full spiral , despite the fact that the radius of a gsw sourced at @xmath101 grows with @xmath101 .
although a - priori one might worry that this feature is an outcome of considering the full support of the gsws , it is nevertheless possible to check that the spiral peak remains of constant width even with the actual profile of each gsw taken into account .
[ [ spiral - peak - profile ] ] spiral peak profile : + + + + + + + + + + + + + + + + + + + + we can now combine the calculations summarized in and , and consider the superposition , by taking into account the actual profile . since @xmath111 falls off with @xmath47 ( due to the corresponding source being further in the bulk ) , we can instead consider the rescaled energy density @xmath112 as done in @xcite . while conceptually straightforward , the determination of @xmath111 is still computationally rather involved , so we will resort to graphical means to estimate the profile . specifically , in we use color - coding and opacity ( modulated by the profile ) to indicate how the full superposition behaves .
we see that the resulting profile of @xmath64 is again in qualitative agreement with that calculated in @xcite
. it would be interesting to examine the peak profile more closely to see how precisely we can recover the quantitative shape @xmath111 computed in @xcite .
in particular , the latter has a curious feature of @xmath64 becoming slightly negative at the tails of the peak .
[ [ dependence - of - ce - on - the - polar - angle ] ] dependence of @xmath64 on the polar angle : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we have hitherto considered the energy density @xmath104 in the orbital plane , but we can use the same methods to find the full @xmath113 ( to be compared with the cutaway plot presented in @xcite s fig.4 ) . due to the orbital symmetry of the set - up , it suffices to focus on the @xmath114 plane .
this is presented in ( left ) ; as in , we plot the full support of a representative set of gsws , with color and opacity modulated by the profile @xmath67 ( though for simplicity we havent folded in the radial dependence ) . on the polar angle @xmath115 .
* left : * light - up @xmath116 for @xmath105 ; color - coding and opacity indicate the relative fall - off in @xmath115 . *
right : * plot of @xmath117 as a function of the quark velocity @xmath24 for various values of @xmath118 ; from top to bottom , @xmath119 ( blue ) , @xmath120 ( green ) , @xmath121 ( red ) , @xmath122 ( orange ) , and @xmath123 ( purple ) .
this indicates that @xmath124 .
, width=192 ] in particular , we find that the gsws superpose along crescent - shaped regions , for which @xmath67 decreases off the orbital plane ( horizontal axis ) as indicated , with characteristic width @xmath125 , which we approximate by the angular span of these crescents . in the right panel of
, we analyse the dependence of this width @xmath125 on the quark velocity @xmath24 , by plotting @xmath117 as a function of @xmath24 for several values of the exponent @xmath118 .
although for our definition of @xmath125 none of these curves is a constant , there is a clear indication that as @xmath32 , only the exponent @xmath126 gives an approximately constant behavior .
this confirms the result of @xcite that the characteristic spread @xmath125 decreases with increasing @xmath24 as @xmath127 .
although the results presented above in a graphical format can be obtained analytically , the exact expressions are far too lengthy and unilluminating . however , we are not so interested in the exact answer of @xmath96 for the full range of @xmath47 , as in a more restricted asymptotic regime @xmath94 where the notion of radiation can be easily disentagled from the quark .
correspondingly , in the bulk we likewise wish to focus on the region @xmath128 , which is simultaneously the region where the approximation is most accurate .
this allows us to simplify the exact expressions for the string embedding , and approximate ( which originates from the @xmath129 term in ) , since it introduces a phase shift in the quantities of interest and therefore is visible even in the asymptotic regime . ] the full string by a linearly - flaring - out helix , @xmath130 that in turn enables us to obtain compact closed - form expressions for the boundary light - up due to the resulting . in this subsection , we summarize such analysis , which complements the numerical results of . the approximation of the string profile simplifies the transverse velocity , from which we can extract the constants of motion describing the steepest geodesic plotted in .
having the constants of motion and the initial point , we can now construct the full geodesic , and thereby find the coordinates of its deepest point , @xmath131 , as well as the dominant light - up @xmath132 , using .
the boundary projection of @xmath133 gives the origin of the boundary light - up due to string bit at @xmath12 ( for compactness of notation , we will drop the subscript from @xmath101 ) , whereas @xmath80 corresponds to the peak of this light - up . specifically , for the string bit at @xmath12 , @xmath134 @xmath135 and @xmath136 , where @xmath137 note that the origin of the gsw from a point @xmath12 along the string describes a circle of constant radius @xmath138 in the @xmath79 plane on the boundary , with angular dependence parameterized by @xmath12 . to summarize , the boundary light - up from a gsw sourced by a string bit at @xmath12
is supported on a sphere of radius @xmath139 with origin at @xmath140 , given approximately by @xmath141 the amplitude of this gsw is strongest nearest the source ; within the boundary , the dominant light - up due to the string bit at @xmath12 occurs at @xmath142 note that this indeed gives a spiral curve on the orbital plane , parameterized by @xmath12 , whose radius @xmath47 grows linearly with angle @xmath143 .
the spacing @xmath144 between the spiral arms ( cf . ) can now be obtained by examining how @xmath145 increases under a single orbit @xmath146 : @xmath12 jumps by @xmath147 , which immediately gives the previously quoted result @xmath148 apart from verifying the expected spiral spacing and therefore the speed of propagation of radiation bursts , we can now also confirm the phase , which likewise agrees precisely with the results of @xcite and the weakly coupled synchrotron radiation .
this width @xmath107 of the spiral arm ( cf . ) is rather more involved , albeit conceptually straightforward to extract .
the calculation proceeds by finding intersections of ` adjoining ' gsws , at @xmath12 and @xmath149 in the limit as @xmath150 ( there are two solutions for each circle , and under variation of @xmath12 they form two nested spirals ) . these give the edges of the emergent spiral in .
we determine its width @xmath107 by taking the difference of the edges intersections with a given radial ray .
this yields @xmath151 while this expression applies for any @xmath24 in the asymptotic regime @xmath152 , it is more interesting to consider the scaling of the thickness @xmath107 at large velocities @xmath153 , i.e. as @xmath154 . in this limit
we find that @xmath155 this confirms the expected scaling @xmath156 , which was found in @xcite .
in fact , there the authors conjectured that the natural length scale would be given by @xmath157 ; here we confirm the @xmath5 scaling and the coefficient up to a @xmath158 factor .
in the preceding section we have described the important features of the boundary energy density @xmath159 induced by a relativistic coiling string in the bulk .
we proposed approximating this by a superposition of gravitational shock waves produced by each string bit we dubbed this the ` beaming mechanism ' and calculated the consequences of this proposal . in particular , we have obtained the characteristic spiral pattern of synchrotron radiation ( cf . e.g. ) .
we have calculated the spacing , phase , width , speed of propagation , and polar extent , of each energy pulse , and found each of these characteristic features in complete quantitative agreement with the computations of @xcite .
our results suggest that it is indeed valid to view the string s backreaction as given by superposition of gravitational shock waves . from a computational perspective , this presents a major simplification : instead of having to solve the full linearized einstein s equations in the bulk , we merely need to find geodesics . in fact , this provides a useful tool for understanding the radiation of a relativistic quark following an arbitrary trajectory , since the string tends to become more relativistic deeper in the bulk and develops a worldsheet horizon , making the approximation more accurate further away from the quark . since for any quark trajectory the corresponding string motion
is known explicitly @xcite , we can construct the just as easily as for the synchrotron case discussed above .
encouraged by the success in using in ads to reproduce the synchrotron radiation in strongly coupled medium at zero temperature calculated in @xcite , a natural question to ask is whether this method also applies in the case of non - zero temperature .
indeed , once one generalizes the gsw construction to schwarzschild - ads background , one should likewise be able to estimate at least some part of the quark s radiation in a thermal strongly - coupled plasma , which should suffice to reproduce , for instance , the diffusion wake and sonic boom generated by a moving quark @xcite .
however , it is important to realize that in the presence of a bulk ads black hole , the string no longer remains relativistic deep in the bulk , so the approximation will have much more constrained regime of validity in particular , it will no longer improve with distance from the quark .
zeroth order expectations from directly adapting our geodesic construction would suggest that the radiation which starts out collimated diffuses on a thermal scale ; as is indeed consistent with expectations form the field theory side .
moreover , the construction of a gravitational shock wave itself , in schwarzschild - ads background , is substantially more involved and is it no longer guaranteed that the metric perturbation from schwarzschild - ads due to a relativistic particle would remain similarly localized .
nevertheless , given some region where the approximation is accurate , it may be possible to go beyond the initial regime of validity by performing series expansion in @xmath160 and building up the backreaction effects order - by - order .
we however leave this for future work .
finally , we end with the issue which originally motivated the present work , namely how the observed beaming effect bears on the scale / radius duality .
recall that scale / radius ( or uv / ir ) duality @xcite , which has been one of the first and most important entries in the ads / cft dictionary , suggests that the closer a bulk excitation is to the ads boundary , the smaller is the characteristic size of the dual cft excitation .
this provides an important insight into the holographic nature of the ads / cft duality : the ` extra ' ( radial ) dimension in the bulk arises from a scale in the dual cft .
this intuition has bolstered our understanding of various dynamical processes .
for example , the field theory notion of color transparency , that different - scale excitations pass through each other without interacting , is beautifully explained by bulk locality : in the bulk , the corresponding excitations do not interact due to being separated in the radial direction . from this , one might have naively expected that same - scale excitations passing through each other would interact .
however , the present context emphasized that naive applications of the scale / radius duality should be treated with caution .
in particular , as we have seen , a source deep in the bulk of ads may nevertheless produce a sharply - localized effect on the boundary . by taking an ordinary source and a beamed source
, we can engineer a situation where two excitations with ( instantaneously ) similar characteristic scale do not interact ( because within the bulk they are well separated ) , as well the converse situation where two excitations with widely differing scale nevertheless do interact ( because the dual bulk excitations collide with each other ) .
the beaming mechanism , which underlies such constructions , is quite interesting .
the relativistic velocity of the string causes the backreaction due to each string bit to be localized on a co - dimension one surface of the gsw .
although the size of this surface grows with the bulk position of the source ( it has larger radius for the string bits deeper in bulk ) , the dominant effect is sharply peaked .
the final ingredient , however , is the superposition of all the gsws from individual string bits , as emphasized most clearly by .
it is due to this combined effect that the radiation pulse on the boundary does not broaden as it propagates outward .
it is a pleasure to thank paul chesler , roberto emparan , gary gibbons , gary horowitz , hong liu , shiraz minwalla , rob myers , joe polchinski , and mukund rangamani for valuable discussions .
i would also like to thank mit , ubc , galileo galilei institute , and imperial college london for their hospitality during this project .
this work is supported in part by a stfc rolling grant .
in this appendix , we describe how to obtain the metric due to a gravitational shock wave in ads , and a simplification for constructing the support of the gsw using geodesics . [ [ defining - gsw ] ] defining gsw : + + + + + + + + + + + + + although one could construct a gsw in ads directly in a manner similar to @xcite , we will instead simply use their results , adapted to the present context . the method employed by @xcite
involves boosting an ads black hole to the speed of light while keeping the total energy fixed .
conveniently , they use coordinates in which the ads@xmath161 metric is manifestly conformally flat and lorentz invariant ( though not static ) , @xmath162 and utilize the known the shock wave solution in minkowski space . without loss of generality , one can consider a massless particle moving in the @xmath163 direction , and write an ansatz for its backreaction as @xmath164 the function @xmath165 can then be obtained by solving einstein s equations for this ansatz . on the transverse space @xmath166 : @xmath167 , whose behavior near @xmath168 mimics that in flat spacetime . ]
however , we first consider the _ support _ of the gsw , for which the exact form of @xmath169 is immaterial , and only the @xmath170 term is relevant ( we will return to considering the profile in , where we will start with a slightly different form of the metric ) . to make contact with our set - up , we wish to recast this into the fefferman - graham form .
although a change of coordinates of the form @xmath171 considered by @xcite corresponds to a radial null geodesic @xmath172 at @xmath173 , we could generalize this to arbitrary direction , by simply making an extra change of the @xmath31 coordinates , @xmath174 where @xmath175 is the requisite lorentz transformation to rotate the direction of motion .
however , while straightforward , this is still a bit messy , so we will instead proceed using the following trick : we observe that the transverse null plane " defined by @xmath176 in the coordinates of has its spatial cross sections given by a co - dimension 2 hypersurface generated by ` zero - energy ' spacelike geodesics emanating transverse to the @xmath177 direction .
this is however a purely geometric statement , which therefore applies in any coordinate system .
since the original form of the metric does not single - out any preferred direction , we can use this as a general prescription for finding any gsw in ads . [
[ geodesics - in - ads ] ] geodesics in ads : + + + + + + + + + + + + + + + + + the actual construction of the gsw sourced by a given string bit moving relativistically in direction of @xmath42 therefore proceeds as follows .
, so that the ads metric is @xmath178 $ ] . ]
without loss of generality , let us fix @xmath95 and consider the support of the gsw from a string bit parameterized by @xmath12 .
a general zero - energy spatial geodesic in ads has a tangent vector @xmath179 , and the killing fields @xmath180 , and @xmath181 generate constants of motion @xmath182 , @xmath183 and @xmath184 . defining @xmath185
, we can write the radial part of the geodesic equation as @xmath186 from which we can readily extract the solution for @xmath187 along such a geodesic .
it is easy to verify that all geodesics will have the shape of a semi - circle in the @xmath77 space , intersecting the boundary @xmath13 orthogonally .
the position , orientation , and size of this semi - circle will be given by the initial position and velocity ( and the latter is given by the constants of motion @xmath188 ) .
in particular , let @xmath189 denote the initial position , @xmath190 denote the endpoint of the ( upward part of ) the geodesic at @xmath13 , and for completeness let @xmath191 denote the symmetric point along the geodesic , which has the largest value of @xmath12 .
we can then write the @xmath192 part of the geodesic compactly as @xmath193 and similarly for the @xmath194 and @xmath195 parts .
now that we have the general form of a zero - energy geodesic , we can find the specific set of geodesics pertaining to a given string bit by specifying the initial position and velocity .
the initial position @xmath196 is obtained from the string s profile in cartesian coordinates , @xmath197 $ ] , @xmath198 $ ] , and @xmath199 . from each initial position
@xmath76 we have a full two - parameter family of geodesics , constrained by the condition of emanating orthogonally to @xmath200 .
denoting the initial velocity @xmath201 , we can freely specify the components @xmath202 and @xmath203 , and impose the orthogonality condition by @xmath204 and then normalize this velocity by fixing @xmath205 in terms of @xmath202 and @xmath203 appropriately . as discussed in below
, the dominant contribution to @xmath64 arises from the point along the gsw which is ` closest ' to the source .
this will occur along the ` steepest ' geodesic towards the boundary from the point @xmath76 , emanating orthogonally to @xmath200 .
for such geodesic , it is simple to construct @xmath206 by taking any @xmath207 and letting @xmath208 finally , from the initial position specified by @xmath101 and a velocity @xmath209 we can extract the constants of motion , @xmath210 , @xmath211 , and @xmath212 , and relate the endpoint @xmath80 to the initial position @xmath76 via , @xmath213 and similarly for @xmath214 .
[ [ asymptotic - analysis ] ] asymptotic analysis : + + + + + + + + + + + + + + + + + + + + from the asymptotic analysis indicated in , we can extract these constants of motion explicitly as @xmath215 where @xmath216 from which it follows that @xmath217 [ [ regularized - proper - length ] ] regularized proper length + + + + + + + + + + + + + + + + + + + + + + + + + finally , let us find the ( regularized ) proper length along such a geodesic .
the direction of the geodesic is unimportant , only the starting point @xmath76 and the total momentum @xmath218 matters ; these determine how far from @xmath76 is the final point @xmath80 . since the proper length along such a geodesic , between @xmath101 and some final value @xmath219 is @xmath220
we see that the @xmath221 limit generates a universal piece @xmath222 which can be regulated away , by the now standard procedure . rewriting @xmath218 in terms of the geodesic endpoint on the boundary
we obtain the regularized proper length @xmath223
in this appendix we wish to determine the actual profile of the boundary stress tensor induced by a given gsw .
we make use of the calculation performed already in @xcite .
a gsw sourced by a point particle propagating _ parallel _ to the boundary in the @xmath30 direction at @xmath224 , with @xmath91 and @xmath225 has the following metric : @xmath226 + \frac{1}{u } \ , \phi(y , z , u ) \
, \delta(t - x ) \ , ( dt - dx)^2 \label{pargswmet}\ ] ] where , up to an overall normalization factor , @xmath227 note that the source of the gsw is at @xmath228 .
we can easily generalize this set - up to a particle propagating parallel to the boundary at any other bulk radial distance @xmath229 by using symmetry of ads .
rescaling coordinates @xmath230 , @xmath231 , @xmath232 , @xmath233 , and @xmath234 , leaves the form of the metric invariant , and @xmath235 likewise has the same functional dependence on @xmath236 in , but this function is now different : @xmath237 in particular , the source of the gsw at @xmath228 propagates at @xmath238 .
now that we have the exact metric of a gsw for a particle propagating parallel to the boundary at any given @xmath239 in fefferman - graham coordinates , it is straightforward to read off the induced stress tensor on the boundary in the standard manner @xcite . since for a metric of the form @xmath240 the boundary stress tensor expectation value is given by @xmath241 we find that in the present case the boundary energy density @xmath242 sourced by a null particle at @xmath92 is given by @xmath243 using the results in , we can recast this expression in terms of a regularized proper length along a geodesic from the source towards the boundary along the gsw ( in this case along constant @xmath30 ) .
in particular , writing this more generally in terms of a geodesic between the source at @xmath76 and a boundary point @xmath80 , with regularized proper length @xmath244 , the profile of energy density ( which is supported along the boundary value of the gsw ) is @xmath245 in we will take this expression to be valid in full generality , even when the source is not propagating parallel to the boundary .
note , however , that in the interesting case of a relativistic quark for which @xmath246 , the source indeed does propagate approximately parallel to the boundary .
j. m. maldacena , `` the large n limit of superconformal field theories and supergravity , '' _ adv .
* 2 * ( 1998 ) 231252 , http://arxiv.org/abs/hep-th/9711200[arxiv:hep-th/9711200 ] .
s. s. gubser , i. r. klebanov , and a. m. polyakov , `` gauge theory correlators from non - critical string theory , '' http://dx.doi.org/10.1016/s0370-2693(98)00377-3[_phys .
* b428 * ( 1998 ) 105114 ] , http://arxiv.org/abs/hep-th/9802109[arxiv:hep-th/9802109 ] .
e. witten , `` anti - de sitter space and holography , '' _ adv .
* 2 * ( 1998 ) 253291 , http://arxiv.org/abs/hep-th/9802150[arxiv:hep-th/9802150 ] . c. athanasiou , p. m. chesler , h. liu , d. nickel , and k. rajagopal , `` synchrotron radiation in strongly coupled conformal field theories , '' http://dx.doi.org/10.1103/physrevd.81.126001[_phys.rev._ * d81 * ( 2010 ) 126001 ] , http://arxiv.org/abs/1001.3880[arxiv:1001.3880 [ hep - th ] ] .
a. mikhailov , `` nonlinear waves in ads / cft correspondence , '' http://arxiv.org/abs/hep-th/0305196[hep-th/0305196 ] .
y. hatta , e. iancu , a. h. mueller , and d. n. triantafyllopoulos , `` aspects of the uv / ir correspondence : energy broadening and string fluctuations , '' http://arxiv.org/abs/1011.3763[1011.3763 ] . l. susskind and e. witten , `` the holographic bound in anti - de sitter space , '' http://arxiv.org/abs/hep-th/9805114[arxiv:hep-th/9805114 ] . v. e. hubeny , `` relativistic beaming in ads / cft , '' http://arxiv.org/abs/1011.1270[1011.1270 ] .
p. aichelburg and r. sexl , `` on the gravitational field of a massless particle , '' http://dx.doi.org/10.1007/bf00758149[_gen.rel.grav._ * 2 * ( 1971 ) 303312 ] .
t. dray and g. t hooft , `` the gravitational shock wave of a massless particle , '' http://dx.doi.org/10.1016/0550-3213(85)90525-5[_nucl .
* b253 * ( 1985 ) 173 ] .
m. chernicoff and a. guijosa , `` acceleration , energy loss and screening in strongly- coupled gauge theories , '' _ jhep _ * 06 * ( 2008 ) 005 , http://arxiv.org/abs/0803.3070[0803.3070 ] .
xiao , `` on the exact solution of the accelerating string in @xmath247 space , '' http://dx.doi.org/10.1016/j.physletb.2008.06.017[_phys .
_ * b665 * ( 2008 ) 173177 ] , http://arxiv.org/abs/0804.1343[arxiv:0804.1343 [ hep - th ] ] .
m. chernicoff , j. a. garcia , and a. guijosa , `` generalized lorentz - dirac equation for a strongly - coupled gauge theory , '' http://dx.doi.org/10.1103/physrevlett.102.241601[_phys . rev
* 102 * ( 2009 ) 241601 ] , http://arxiv.org/abs/0903.2047[arxiv:0903.2047 [ hep - th ] ] .
m. chernicoff and a. paredes , `` accelerated detectors and worldsheet horizons in ads / cft , '' http://arxiv.org/abs/1011.4206[1011.4206 ] .
m. hotta and m. tanaka , `` shock wave geometry with nonvanishing cosmological constant , '' http://dx.doi.org/10.1088/0264-9381/10/2/012[_class .
_ * 10 * ( 1993 ) 307314 ] .
g. t. horowitz and n. itzhaki , `` black holes , shock waves , and causality in the ads / cft correspondence , '' _ jhep _ * 9902 * ( 1999 ) 010 , http://arxiv.org/abs/hep-th/9901012[arxiv:hep-th/9901012 [ hep - th ] ] .
j. b. kogut and l. susskind , `` the parton picture of elementary particles , '' http://dx.doi.org/10.1016/0370-1573(73)90009-4[_phys .
_ * 8 * ( 1973 ) 75172 ] .
d. grumiller and p. romatschke , `` on the collision of two shock waves in ads(5 ) , '' http://dx.doi.org/10.1088/1126-6708/2008/08/027[_jhep_ * 0808 * ( 2008 ) 027 ] , http://arxiv.org/abs/0803.3226 [ arxiv:0803.3226 [ hep - th ] ] .
s. s. gubser , s. s. pufu , and a. yarom , `` entropy production in collisions of gravitational shock waves and of heavy ions , '' http://arxiv.org/abs/0805.1551[arxiv:0805.1551 [ hep - th ] ] .
y. v. kovchegov , `` shock wave collisions and thermalization in ads5 , '' http://arxiv.org/abs/1011.0711[1011.0711 ] .
k. sfetsos , `` on gravitational shock waves in curved space - times , '' _ nucl .
* b436 * ( 1995 ) 721746 , http://arxiv.org/abs/hep-th/9408169[hep-th/9408169 ] .
j. podolsky and j. b. griffiths , `` impulsive waves in de sitter and anti - de sitter space- times generated by null particles with an arbitrary multipole structure , '' http://dx.doi.org/10.1088/0264-9381/15/2/018[_class .
* 15 * ( 1998 ) 453463 ] , http://arxiv.org/abs/gr-qc/9710049[arxiv:gr-qc/9710049 ] .
v. balasubramanian and p. kraus , `` a stress tensor for anti - de sitter gravity , '' _ commun
* 208 * ( 1999 ) 413428 , http://arxiv.org/abs/hep-th/9902121[hep-th/9902121 ] .
s. de haro , s. n. solodukhin , and k. skenderis , `` holographic reconstruction of spacetime and renormalization in the ads / cft correspondence , '' http://dx.doi.org/10.1007/s002200100381[_commun .
_ * 217 * ( 2001 ) 595622 ] , http://arxiv.org/abs/hep-th/0002230[arxiv:hep-th/0002230 ] . c. athanasiou , p. m. chesler ,
h. liu , d. nickel , and k. rajagopal , `` radiation of a circulating quark in strongly coupled n=4 super yang - mills theory , '' http://arxiv.org/abs/1009.1885[1009.1885 ] .
j. j. friess , s. s. gubser , g. michalogiorgakis , and s. s. pufu , `` the stress tensor of a quark moving through n = 4 thermal plasma , '' http://dx.doi.org/10.1103/physrevd.75.106003[_phys .
_ * d75 * ( 2007 ) 106003 ] , http://arxiv.org/abs/hep-th/0607022[arxiv:hep-th/0607022 ] .
t. damour and a. vilenkin , `` gravitational wave bursts from cusps and kinks on cosmic strings , '' _ phys
* d64 * ( 2001 ) 064008 , http://arxiv.org/abs/gr-qc/0104026[gr-qc/0104026 ] .
h. nastase , `` on high energy scattering inside gravitational backgrounds , '' http://arxiv.org/abs/hep-th/0410124[hep-th/0410124 ] . | we propose a new and simple method of estimating the radiation due to an accelerated quark in a strongly coupled medium , within the framework of the ads / cft correspondence .
in particular , we offer a heuristic explanation of the collimated nature of synchrotron radiation produced by a circling quark , which was recently studied in phys.rev.*d81 * ( 2010 ) 126001 .
the gravitational dual of such quark is a coiling string in ads , whose backreaction on the spacetime geometry remains tightly confined , as if ` beamed ' towards the boundary . while this appears to contradict conventional expectations from the scale / radius duality , we resolve the issue by observing that the backreaction of a relativistic string is reproduced by a superposition of gravitational shock waves .
we further demonstrate that this proposal allows us to reduce the problem of computing the boundary stress tensor to merely calculating geodesics in ads , as opposed to solving linearized einstein s equations .
dcpt-10/75 * holographic dual of collimated radiation * veronika e. hubeny@xmath0 + _ science laboratories , south road , durham dh1 3le , united kingdom _ + |
the phase behavior of fluids confined in narrow regions has been studied extensively @xcite .
it strongly depends on the geometry of the walls and on the molecular interactions between the fluid and the walls .
its understanding is crucial in the physics of fluids in porous media .
it is also needed to study the dynamics of confined fluids . in particular , the liquid phase is usually favored by the walls in one - component fluids , while one component is preferentially attracted to the walls in binary mixtures @xcite . in such situations , narrow regions may be filled with the phase favored by the walls or may hold some fraction of the disfavored phase . hence
, in the film geometry , there appears a first - order phase transition between these states , which forms a line ( ccl ) ending at a film critical point outside the bulk coexistence curve in the @xmath3-@xmath6 plane @xcite , where @xmath6 is the reservoir chemical potential @xcite .
we call it the capillary condensation transition even for binary mixtures , though this name has been used for the gas - liquid phase transition in porous media @xcite . around ccl ,
the reservoir is rich in the component disfavored by the walls for binary mixtures . with increasing the wall separation @xmath7 , the film critical point approaches the bulk critical point .
crossover then occurs between two - dimensional ( 2d ) and three - dimensional ( 3d ) phase transition behaviors . for ising films near the bulk criticality , fisher and nakanishi @xcite presented the scaling theory of ccl in the @xmath3-@xmath8 plane , where @xmath8 represents applied magnetic field .
they also calculated ccl in the mean - field @xmath9 theory .
et al_. used the density functional theory to calculate the inhomogeneous structures in pores @xcite . for a lennard - jones fluid in cylindrical pores , peterson _
et al._@xcite obtained steady gas - liquid two - phase patterns . for a lattice gas model , binder and
landau @xcite studied the capillary condensation transition using a monte carlo method . for a microscopic model of 2d ising stripes , macioek _
et al _ @xcite found a ( pseudo ) ccl using a density - matrix renormalization - group method .
for square well fluids in slit pores , singh _
et al._@xcite numerically examined the crossover from 3d to 2d .
recently , two of the present authors @xcite calculated ccl near the bulk critical point using the local functional theory @xcite , which accounts for the renormalization effect due to the critical fluctuations .
the lowering of the film critical temperature @xmath4 from the bulk critical temperature @xmath10 was shown to be proportional to @xmath11 ( where @xmath12 ) in accord with the scaling theory @xcite .
along ccl , our calculations @xcite and those by macioek _ et al _ @xcite showed strong enhancement of the so - called casimir amplitudes @xcite .
similar first - order transitions were found between plates @xcite and colloids @xcite in binary mixtures containing salt .
the aim of this paper is to investigate the phase separation in near - critical binary mixtures between parallel plates using model h and model b @xcite .
here , phase separation takes place around ccl and the hydrodynamic interaction is crucial in the late - stage phase separation .
it is worth noting that near - critical fluids in porous media exhibit history - dependent frozen domains and activated dynamics with non - exponential relaxations @xcite . to gain insight into such complicated effects
, we may start with near - critical fluids in the film geometry .
treating near - critical fluids , we may construct a universal theory with a few materials - independent parameters , where @xmath7 much exceeds microscopic spatial scales . in the literature ,
much attention has been paid to the interplay of wetting and phase separation @xcite , which is referred to as surface - directed phase separation .
however , simulations including the hydrodynamic interaction have not been abundant @xcite .
we mention that tanaka and araki @xcite integrated the model h equations in the semi - infinite situation and jaiswal _ et al . _
@xcite performed molecular dynamics simulation to investigate the hydrodynamic flow effect between parallel plates . in our simulation , the order parameter @xmath5 changes in the direction perpendicular to the walls in the strong adsorption regime .
then , the dynamics is one - dimensional in an early stage but the fluid flow in the lateral directions accelerates the late - stage coarsening even under the no - slip boundary condition on the walls @xcite .
on the other hand , porcheron and monson @xcite numerically studied the dynamics of extrusion and intrusion of liquid mercury between a cylindrical pore and a reservoir .
such a process is crucial in experiments of adsorption and desorption between a porous material and a surrounding fluid @xcite . in our simulation
we assume no mass exchange imposing the periodic boundary condition in the lateral directions , as in the previous simulations of surface - directed phase separation .
the organization of this paper is as follows . in sec.ii
, we will summarize the results of the local functional theory of near - critical binary mixtures in the film geometry .
we will newly present some results on the phase behavior , which will facilitate understanding the phase separation near ccl . in sec.iii
, we will present our simulation results of the phase separation with the velocity field ( model h ) and without it ( model b ) .
this section provides the theoretical background of our simulation on the basis of our previous paper @xcite .
the boltzmann constant @xmath13 will be set equal to unity .
we suppose near - critical binary mixtures with an upper critical solution temperature ( ucst ) @xmath10 at a constant pressure .
the order parameter @xmath5 is proportional to @xmath14 , where @xmath15 is the composition and @xmath16 is its critical value .
the reduced temperature is written as = ( t - t_c)/t_c . in our numerical analysis ,
the usual critical exponents take the following values@xcite : & & = 0.110 , = 0.325 , = 1.240 , + & & = 0.630 , = 0.0317,=4.815 . at the critical composition with @xmath17 ,
the correlation length is written as @xmath18 , where @xmath19 is a microscopic length .
the coexistence curve in the region @xmath20 is denoted by cx .
the correlation length on cx is written as @xmath21 , where @xmath22 is another microscopic length with the ratio @xmath23 being a universal number .
we write @xmath5 in the coexisting two phases as @xmath24 with _ cx = b_cx||^ , where @xmath25 is a constant .
we assume that the bulk free energy @xmath26 including the gradient part is of the local functional form @xcite , f = dr[f + t_c c||^2 ] . in the following
, we give a simple form for the free energy density @xmath27 . in our theory , the critical fluctuations with sizes smaller than the correlation length @xmath28 have already been coarse - grained at the starting point . outside cx , @xmath29 is of the ginzburg - landau form ,
/t_c= ^2 + ^4 . here
, we have omitted the free energy contribution for @xmath30 , whose singular part is proportional to @xmath31 yielding the specific heat singularity .
the coefficients @xmath32 and @xmath33 in @xmath29 and @xmath34 in @xmath26 are renormalized ones in three dimensions . as in the linear parametric model @xcite
, we use a nonnegative parameter @xmath35 representing the distance from the critical point in the @xmath36-@xmath5 plane to obtain r/&= & c_1 _ 0 ^ -2w^-1 , + u / u^*&= & c_1 ^ 2 _ 0 ^ -1w^(1 - 2 ) , + c & = & c_1 w^- , where @xmath37 and @xmath38 are constants
. we may set @xmath39 by rescaling @xmath40 without loss of generality . in the present case , @xmath41 is dimensionless .
the constant @xmath38 is a universal number and we set @xmath42 .
the fractional powers of @xmath43 in eqs.(2.6)-(2.8 ) arise from the renormalization of the critical fluctuations with wavenumbers larger than the inverse correlation length @xmath44 .
we determine @xmath43 as a function of @xmath36 and @xmath5 by w= + ( 3u^ * c_1_0 ) w^1 - 2 ^ 2 , which is equivalent to @xmath45 .
thus , @xmath46 for @xmath30 and @xmath47 , while @xmath48 for @xmath49 .
the derivative @xmath50 at fixed @xmath36 denotes the chemical potential difference between the two components@xcite , but it will be simply called the chemical potential . in terms of the ratio @xmath51 , it reads = c_1 w^. on cx , we require @xmath52 , which yields the equation @xmath53 for @xmath54 . on cx , this gives @xmath55 or @xmath56 with @xmath57 .
together with eqs.(2.3 ) and ( 2.9 ) , we obtain b_cx^2= ( 1+)^2 - 1/3u^*c_1_0 . w introduce the susceptibility @xmath58 defined by t_c/= /= ^2 f/^2 . for @xmath59 and @xmath17 ,
we simply obtain @xmath60 . on cx
, we write @xmath61 . in terms of the critical amplitude ratio @xmath62 for @xmath20 ,
the susceptibility on cx reads _
cx= r_^-1 c_1 ^ -1 _ 0 ^ 2 ||^- .
some calculations give @xmath63 @xcite . in terms of @xmath64 ,
the correlation length is expressed as @xmath65 , which yields the critical amplitude ratio @xmath66 @xcite . for @xmath49 , we have @xmath67 .
the interior of cx is given by @xmath68 and @xmath20 , where we need to define the free energy density @xmath29 to examine two - phase coexistence .
we assume a @xmath69-theory with coefficients depending only on @xmath36 , where @xmath70 and @xmath71 are continuous across the coexistence curve .
we then obtain ( f - f_cx)/ t_c= ( _ cx^2 /8 _ cx ) ( ^2/_cx^2 - 1)^2 , where @xmath72 is the free energy density on cx and @xmath73 is defined by eq.(2.13 )
. we also set c = c_cx= c_1 ||^- , which is the value of @xmath34 on cx .
the renormalization effect inside cx is assumed to be unchanged from that on cx with the same @xmath36 .
the @xmath74 inside cx then reads /t_c=
( ^2/_cx^2 - 1)/2 _ cx
the surface tension @xmath75 between coexisting bulk two phases is given by the standard expression , & = & 2t_c ( c_cx/_cx)^1/2_cx^2/3 + & = & a_s t_c /^2 , where @xmath76 is the correlation length on cx .
the universal number @xmath77 is estimated to be @xmath78 in our model , while its reliable value is about @xmath79 @xcite .
( left ) and @xmath80 ( right ) vs @xmath81 for @xmath82 ( top ) , @xmath83 ( middle ) , and @xmath84 ( bottom ) .
top : adsorption - dominated profile a with @xmath85 .
middle : two profiles b and c on the capillary condensation line with the same grand potential @xmath86 . in ( b ) , the two curves of @xmath87 enclose two regions with the same area close to the surface tension @xmath75 .
bottom : three profiles d , e , and f with the same @xmath36 and @xmath6 ( see fig.3 ) . in ( c ) , @xmath86 increases in the order of d , e , and f. ] - @xmath88 plane .
the bulk coexistence line is given by @xmath20 and @xmath89 .
on its left , there appears a first - order capillary condensation line ( red bold line ) ending at a film critical point , which is calculated from 1d profiles .
displayed also are values of @xmath90 in steady two - phase coexistence in our simulation of a @xmath91 system ( @xmath92 ) .
starting point of our simulation ( @xmath93 ) is also shown ( @xmath94 ) . ]
-@xmath95 plane , which are calculated for 1d profiles at @xmath96 , -10 , and -20 . for @xmath36 less than its film critical value ( @xmath97 )
, a van der waals loop appears .
dotted parts of the curves for @xmath98 and -20 are not stable in contact with a reservoir .
points a , b , c , d , e , and f correspond to the curves in fig.1 . ]
we suppose a near - critical fluid between parallel symmetric walls in the region @xmath99 , where @xmath7 is much longer than any microscopic lengths .
to avoid the discussion of the edge effect , the lateral plate dimension @xmath100 is supposed to much exceed @xmath7 .
the fluid is close to the bulk criticality and above the prewetting transition line @xcite .
we use our local functional theory , neglecting the two - dimensional thermal fluctuations with sizes exceeding @xmath7 in the @xmath101 plane .
we scale @xmath36 and @xmath5 in units of @xmath102 and @xmath103 , respectively , defined by _
d & = & ( _ 0/d)^1/ , + _ d & = & ( 24^/ /3u^ * c_1 _ 0)^1/2 _ d^. in equilibrium theory , it is convenient to assume that the fluid between the walls is in contact with a large reservoir containing the same binary mixture , where the order parameter is @xmath104 and the chemical potential is _ =( _ , ) .
here , @xmath6 corresponds to magnetic field @xmath8 for films of ising spin systems .
we are interested in the case @xmath105 ( or @xmath106 ) and @xmath107 , where @xmath108 is the value of @xmath5 at the walls .
if equilibrium is attained in the total system including the reservoir , we should minimize the film grand potential @xmath86 . including the surface free energy
, we assume the form , = dr _ loc - t_c ds h_1 , where the space integral @xmath109 is within the film , the surface integral @xmath110 is on the walls at @xmath111 and @xmath7 , and @xmath112 is a surface field symmetrically given on the two walls .
in addition , we neglect the surface free energy of the form @xmath113 assumed in the literature@xcite ( or we consider the limit @xmath114 ) . in eq.(2.21 )
@xmath115 is the local grand potential density including the gradient part , _
loc= _ s + t_cc ||^2 , where @xmath116 is the excess grand potential density written as _ s = f()-f ( _ ) - _ ( - _ ) . now minimization of @xmath86 yields the bulk equation , = - c||^2 -t_c c ^2= _ , where @xmath117 .
the boundary conditions at @xmath111 and @xmath7 are given by (x , y,0)=- (x , y , d)= -h_1/c .
where @xmath118 .
the role of @xmath112 in this paper is simply to assure the strong adsorption regime @xmath119 @xcite , where @xmath108 is the boundary value of @xmath5 .
this regime is eventually realized on approaching the criticality ( however small @xmath112 is ) . in our simulation ,
the profile of @xmath5 in the region @xmath120 is nearly one - dimensional depending only on @xmath121 even in two phase states ( see figs.5 and 6 ) .
it decays slowly as @xmath122 for @xmath120 @xcite , where @xmath123 is a short microscopic length introduced by rudnick and jasnow @xcite . with the gradient free energy in the form of eq.(2.22 ) , @xmath123 is expressed as _ 0 ~_0 ( c_1_0)^-/2 _ 0 ^ -/ ~d ( _ d/_0)^-/ , in terms of @xmath108 . the excess surface adsorption of @xmath5 in the region @xmath124 is of order @xmath125 and is negligible for large @xmath108 from @xmath126 , while that in the region @xmath127 is of order @xmath128 for @xmath129 @xcite . in the strong - adsorption regime
we calculate the average of @xmath5 along the @xmath121 axis , = _ 0^d dz /d . in two - phase states , @xmath2 depends on @xmath130 . from eq.(2.25 )
it follows the estimation @xmath131 .
as @xmath132 , we find h_1 = _ 0 ^ -/(b_1 + b_2 _ 0 ^ -1/+ ) , where @xmath133 and @xmath134 are positive constants .
this is the expression for @xmath135 . in this regime ,
the surface free energy in eq.(2.21 ) is given by @xmath136 , where @xmath137 is the surface area . in our previous paper@xcite
, we examined the film phase behavior at fixed large @xmath108 , treating the surface free energy as a constant .
in our simulation , we assume the boundary condition ( 2.25 ) with @xmath138 to obtain @xmath139 .
we consider the capillary condensation transition on the basis of one - dimensional ( 1d ) profiles @xmath140 . from eq.(2.25 ) , we have the symmetry @xmath141 . in the region
@xmath142 , eq.(2.24 ) is integrated to give z=
^_0_d ^1/2 . here
, @xmath143 is the osmotic pressure .
it is the force density per unit area exerted by the fluid to the plates . in our case , @xmath144 , indicating attractive inter - wall interaction . in the 1d case ,
it is also written as & = & f(_)-f(_m)- _ ( _ -_m ) + & = & -_s(_m ) . at the midpoint @xmath145 , we set @xmath146
. the fluid at the midpoint can be in the phase favored by the walls with @xmath147 due to the strong adsorption on the walls or in the disfavored phase with @xmath148 .
equation ( 2.30 ) indicates @xmath149 in the former case and @xmath150 in the latter case , so @xmath151 can be very different in these two cases @xcite .
-@xmath152 plane in ( a ) and in the @xmath88-@xmath95 plane in ( b ) , where @xmath153 is the midpoint value . in ( a ) and ( b ) , the capillary condensation curves ( bold lines )
are calculated from 1d profiles , where points for five @xmath88 are those along the @xmath121 axis with @xmath154 and @xmath155 in the final two - phase states in our simulation ( see figs.4 and 5 ) .
bulk coexistence curve is in broken line . in ( b )
our phase separation process is illustrated by arrows , where the total order parameter is conserved . ]
figure 1 displays typical 1d profiles of @xmath156 from eq.(2.29 ) and @xmath87 in eq.(2.22 ) in the range @xmath142 , which will be needed to explain our simulation results .
here we set @xmath157 ( top ) , @xmath158 ( middle ) , and @xmath159 ( bottom ) , where @xmath6 is measured in units of _
/ d^3_dd^/-3 .
salient features in fig.1 are as follows .
( i ) in fig.1(a ) , @xmath5 is positive in the whole region with @xmath160 .
( ii ) in figs.1(b ) and ( b ) , the fluid is on the capillary condensation line , where we give two equilibrium profiles b and c with the same @xmath86 . here , b represents an adsorption - dominated state with @xmath147 and @xmath161 , while for c the film center is occupied by the disfavored phase with @xmath162 and @xmath163 . in the right panel ( b ) , the integral of @xmath87 in the region @xmath142 is the same for b and c. the enclosed two regions have the same area @xmath164 in units of @xmath165 , which is close to the surface tension @xmath166 at this @xmath36 .
in addition , @xmath167 for b and @xmath168 for c @xcite .
( iii ) in figs.1(c ) and ( c ) , the parameters are those slightly below the capillary condensation line ( in fig.2 below ) . here , there are three solutions with the common @xmath36 and @xmath6 , but @xmath169 is @xmath170 for ( d ) , 0.028 for ( e ) , and 1.20 for ( f ) ( see fig.3 below ) .
if we perform simulation in contact with a reservoir with these @xmath36 and @xmath6 , the profile d is realized at long times . in the very early stage of our simulation ,
the dynamics is one - dimensional and the profile f is approached after quenching from a in fig.1(a ) ( see fig.7 below ) . in fig.2
, we show the capillary condensation line ( ccl ) from 1d profiles located on the left of the bulk coexistence line in the @xmath36-@xmath6 plane . in our previous paper @xcite
, the corresponding phase diagram was displayed in the @xmath36-@xmath104 plane .
the discontinuities of the physical quantities across ccl increase with increasing @xmath171 vanishing at a film critical point . at this
film criticality , @xmath36 , @xmath104 , and @xmath6 are calculated as ( , , ) = ( -3.14 , -1.27 , -16.3 ) , where we also have @xmath172 and @xmath173 .
hereafter , the chemical potential @xmath6 on this ccl will be written as @xmath174 .
our numerically calculated ccl is well fitted to the linear form , _
cx^ca ( ) /_d+ 16.3 0.86 ( /_d + 3.14 ) . in fig.1(b ) , the two areas enclosed by the two curves of @xmath175 are the same ( @xmath176 ) . thus , for @xmath177 , the surface tension @xmath75 and the free energy difference per unit area @xmath178 are of the same order .
see the sentences below eq.(2.30 ) and the explanation of fig.1(b ) . for @xmath179
, it follows the relation , _
cx^ca ~- /_cxd ~- |/_d|^2- _ d. since @xmath180 , the theoretical formula ( 2.34 ) is consistent with the numerical formula ( 2.33 ) .
note that eq.(2.34 ) is equivalent to the kelvin equation known for the gas - liquid transition in pores @xcite .
we have already presented a special case of three 1d profiles in fig.1(c ) for @xmath181 . in fig.3
, we show isothermal curves in the @xmath6-@xmath2 plane , which are calculated from 1d profiles with @xmath182 , @xmath183 , and @xmath184 .
the relation between @xmath6 and @xmath2 is monotonic for @xmath185 ( above the film critical temperature ) , while it exhibits a van der waals loop for @xmath186 with three 1d states in a window range @xmath187 @xcite . here ,
@xmath188 and @xmath189 coincide at the film criticality .
the isothems consist of stable and unstable parts characterized by the sign of the film susceptibility defined by _
film= t_c ( /_)_. in fig,3 , points a , b , c , d , e , and f correspond to the curves in fig.1 .
dotted parts of the two curves of @xmath190 and @xmath184 are not stable in the presence of a mass current from a reservoir with common @xmath6 . previously , some authors @xcite calculated the stable parts of isotherms of the average density in the film versus the chemical potential . in our local functional theory ,
the three 1d profiles can be calculated since a unique profile follows for any given set of @xmath36 and @xmath2 . in equilibrium fluctuation theory of films
@xcite , @xmath191 is proportional to the variance of the order parameter fluctuations , so its negativity indicates thermodynamic instability .
furthermore , fig.4 gives the phase diagrams in the @xmath36-@xmath192 and @xmath36-@xmath193 planes .
bold lines represent the capillary condensation curve from 1d profiles as in fig.2 . in steady two - phase states in our simulation , @xmath153 and @xmath2
depend on @xmath194 and @xmath195 , so points for five @xmath36 represent @xmath196 and @xmath197 with @xmath154 and @xmath155 ( see figs.5 and 6 ) .
the former line passes through a domain of the phase disfavored by the walls and the latter through the favored phase only .
phase diagrams similar to fig.4(b ) have been obtained in experiments of the capillary condensation in porous media @xcite . .
values of @xmath198 from our simulation in a finite @xmath91 system for five @xmath88 .
the corresponding values of @xmath199 on the ccl from 1d profiles are also shown . [ cols="^,^,^,^,^,^",options="header " , ]
we performed simulation of phase separation in a @xmath200 cell with @xmath201 imposing the periodic boundary condition along the @xmath194 and @xmath195 axes . in this section , we describe phase separation realized for deep quenching .
however , it was not realized for shallow quenching @xmath202 ) , for which @xmath191 in eq.(2.35 ) is positive . in a @xmath203 system : @xmath204 ( top ) and @xmath175 ( middle ) in the @xmath101 plane at @xmath145 ( left ) and in the @xmath205 plane at @xmath206 ( right ) .
bottom : @xmath204 ( left ) and @xmath175 ( right ) along the @xmath121 axis for @xmath154 ( blue bold line ) and @xmath155 ( red bold line ) , while dotted lines represent 1d profiles from eq.(2.29 ) . ] in a @xmath203 system : @xmath204 ( top ) and @xmath175 ( middle ) in the @xmath101 plane at @xmath145 ( left ) and in the @xmath205 plane at @xmath206 ( right ) .
bottom : @xmath204 ( left ) and @xmath175 ( right ) along the @xmath121 axis for @xmath207 ( blue bold line ) and @xmath155 ( red bold line ) , which are closer to the 1d profiles ( dotted lines ) from eq.(2.29 ) than in fig.5 . ] supposing an incompressible fluid binary mixture with a homogeneous temperature , we use the model h equations @xcite .
the order parameter @xmath5 is a conserved variable governed by = - ( ) + ^2 , where @xmath208 is the kinetic coefficient and the functional derivative @xmath209 may be calculated from eq.(2.4 ) with the aid of eqs.(2.10 ) , ( 2.15 ) , and ( 2.16 ) outside and inside cx .
we neglect the random source term originally present in critical dynamics @xcite , because we treat the deviations much larger than the thermal fluctuations .
the velocity field @xmath210 satisfies @xmath211 and vanishes at @xmath111 and @xmath7 . in the stokes approximation@xcite , @xmath210 is determined by | ^2v = p_0 + ( f/ ) , where @xmath212 is the shear viscosity and the role of a pressure @xmath213 is to ensure @xmath211 . see appendix for the expression of the stress tensor in near - critical fluids and the derivation of eq.(3.2 ) .
the kinetic coefficients @xmath208 and @xmath212 should be treated as renormalized ones @xcite ( see the last sentence of subsec.iia ) . in the vicinity of the bulk coexistence curve , @xmath208
may be approximated by = _
cxd_/t_c , where @xmath214 is the susceptibility on cx in eq.(2.13 ) and @xmath215 is the mutual diffusion constant of the stokes form , d_= t_c /6| , with @xmath216 being the correlation length on cx . in our simulation , @xmath217 is of order @xmath218 at @xmath145 ( see figs.5 and 6 below ) , which supports eqs.(3.3 ) and ( 3.4 ) .
the viscosity @xmath219 exhibits a very weak critical singularity and may be treated as a constant independent of @xmath36 . in this paper
, we also performed simulation for model b without the hydrodynamic interaction @xcite , where @xmath5 obeys the diffusive equation , = ^2 .
the kinetic coefficient @xmath208 is assumed to be given by eqs.(3.3 ) and ( 3.4 ) as in the model h case .
then , comparing the results from the two models , we can examine the role of the hydrodynamic interaction in phase separation .
model b has been used to investigate surface - directed phase separation in binary alloys @xcite . in integrating eqs.(3.1 ) and ( 3.5 ) , the mesh length was @xmath220 and the time interval width was @xmath221 .
the initial state was the 1d profile a in fig.1(a ) at @xmath182 with small random numbers ( @xmath222 ) superimposed at the mesh points . at @xmath223 , we decreased @xmath36 to a final reduced temperature . for @xmath224 , there was no mass exchange between the film and the reservoir so that the total order parameter @xmath225 was fixed at @xmath226 .
we will measure time after quenching in units of t_0= d^2/d _ , which is the mutual diffusion time in the film assumed to be much longer than the thermal diffusion time .
we note that the natural time unit in bulk phase separation has been the order parameter relaxation time @xmath227 @xcite . here
, @xmath228 for @xmath229 . for model h ( left ) and model b ( right )
after quenching from the profile a in fig.1(a ) to @xmath230 . shown
are the profiles along the @xmath121 axis for @xmath154 at @xmath231 , 0.1 , 0.2 , and 0.3 ( top ) and along the @xmath194 axis for @xmath232 ( bottom ) at later times . in the top panels ,
the profile f in fig.1(c ) is approached without noticeable differences between the two models . in the bottom panels ,
fluctuations in the @xmath101 plane appear and coarsening is much quickened for model h than for model b , where @xmath233 0.5 , 0.6 , and 0.7 for model h and 1.0 , 1.1 , and 1.2 for model b. ] for model h and model b ( left ) and @xmath234 for model h ( right ) after quenching from the profile a in fig.1(a ) . at points ( a ) , ( b ) , ( c ) , and ( d ) ( right ) snapshots of @xmath5 and @xmath210 will be given in fig.9 . ] for sufficiently deep quenching , we realized phase separation to find a steady two - phase state at long times both for model h and model b. we could also calculate this final state more accurately from the following relaxation - type equation , = -l_0 , where @xmath235 is a constant and @xmath236 is the space average of @xmath237 at time @xmath238 .
because of its simplicity , we integrated eq.(3.7 ) with a fine mesh length of @xmath239 .
the data points in figs.2 and 4 and the snapshots in figs.5 and 6 are those from the steady states of eq.(3.7 ) . in fig.4 , the deviations of @xmath153 and @xmath2 from those on ccl are surprisingly small , though the lateral dimension @xmath100 is only @xmath240
however , in table 1 , the final two - phase values of @xmath6 considerably deviate from those on ccl .
this may be ascribed to the relatively small size of the susceptibility @xmath73 for these cases .
that is , if we set @xmath241 on cx , we obtain @xmath242 and 283 for @xmath98 and @xmath184 , respectively . here , if we multiply the deviation of @xmath152 or @xmath95 by of @xmath243 , we obtain that of @xmath90 . in figs.5 and 6 , we display the final profiles of @xmath5 and @xmath244 in the @xmath101 plane at @xmath145 ( left ) and in the @xmath245 plane at @xmath206 ( right ) for @xmath98 and @xmath184 . in these cases ,
@xmath28 on cx is @xmath246 and @xmath247 , respectively , which is of the order of the interface thickness . displayed in the bottom panels
are 1d profiles of @xmath5 and @xmath244 along the @xmath121 axis for the two lateral points @xmath154 and @xmath155 .
these profiles are rather close to the 1d profiles from eq.(2.29 ) in accord with fig.4 . both for model h and model b ,
early - stage time - evolution proceeds as follows .
just after quenching , @xmath5 changes only along the @xmath121 axis to approach the 1d profile at the final @xmath36 with fixed @xmath2 ( see fig.3 ) .
if this 1d profile satisfies the instability condition @xmath248 , it follows 3d spinodal decomposition . on the other hand , if it satisfies the stability condition @xmath249 , it remains stationary in simulation without thermal noise .
figure 7 displays @xmath5 after quenching to @xmath230 . in the top panels ,
it is plotted along the @xmath121 axis with @xmath154 at @xmath250 and 0.3 .
the velocity field nearly vanishes for model h , so there is almost no difference between the results of these models .
however , in the bottom panels , the 1d profile becomes unstable with respect to the fluctuations varying in the @xmath101 plane for @xmath251 .
the velocity field grows gradually for model h. in this second time range , the domain formation is much quicker for model h than for model b. in the left panel of fig.8 , we show the free energy decrease @xmath252 at @xmath230 as a function of @xmath238 for model h and model b. here , @xmath253 is the total bulk free energy in eq.(2.4 ) with @xmath254 being its value just after quenching .
its decrease is accelerated with development of the fluctuations in the @xmath101 plane .
the coarsening is slower for model b than for model h by about 5 times . in the right panel of fig.8
, we show the characteristic velocity amplitude @xmath255 for model h , which is defined by ( t)^2 = dr |v|^2 / dl^2 . for @xmath230 ,
@xmath255 is equal to 0.021 , 0.106 , and 0.485 at @xmath256 , 0.6 , and 0.7 , respectively , increasing up to @xmath257 , in units of @xmath258 . in fig.9 ,
we show late stage snapshots of @xmath5 and @xmath210 in the @xmath259 plane ( top ) and in the @xmath101 plane ( middle ) for @xmath230 for model h , where @xmath260 is equal to ( a ) 0.7 , ( b ) 0.9 , ( c ) 1.1 , and ( d ) 1.3 after quenching . in ( a ) we can see a network - like domain of the disfavored phase . in ( b ) three domains can be seen , where the middle one is being absorbed into the bottom one , soon resulting in two domains at @xmath261 .
this process gives rise to a dip in @xmath255 in fig.8(b ) since these two domains are considerably apart . in ( c ) and
( d ) , furthermore , coalescence of these two domains is taking place .
the arrows below the panels indicate the typical velocity , @xmath262 or @xmath263 , where @xmath75 is the surface tension and @xmath219 is the shear viscosity .
note that the typical velocity in the late - stage bulk spinodal decomposition is given by @xmath264 @xcite , which follows from the stress balance @xmath265 with @xmath266 being the typical domain length . in our case , from eq.(2.17 ) these velocities are related as d / t_0=t_c/6d= ( /6a_d)(/| ) .
thus , @xmath267 for @xmath230 in fig.9 .
in summary , we have examined the phase separation in a near - critical binary mixture between symmetric parallel plates in the strong adsorption regime around the capillary condensation line ( ccl ) . using model h and model b ,
simulation has been performed in a @xmath91 cell .
we summarize our main results .
+ ( i ) in sec.ii , we have presented the singular free energy with the gradient part outside and inside the bulk coexistence curve . applying it to near - critical fluids between parallel plates , typical 1d profiles
have been given in fig.1 .
ccl has been plotted in the @xmath36-@xmath6 plane in fig.2 .
the points for the steady two - phase states from our simulation are located in the left side of ccl . in fig.3 , we have also found the van der waals loop of isothermal curves in the @xmath2-@xmath6 plane , where @xmath36 is smaller than the film critical value .
the phase diagrams have been plotted in the @xmath36-@xmath153 and @xmath36-@xmath2 planes in 3d ( bulk ) and 2d ( film ) in fig.4 . the kelvin relation ( 2.34 ) has also been obtained , since the osmotic pressure @xmath151 is of order @xmath268 right below ccl @xcite .
+ ( ii ) in sec.iii , we have first displayed the cross - sectional profiles of @xmath5 and @xmath244 in steady two - phase states in figs.5 and 6 .
the profiles along @xmath121 axis for @xmath154 and @xmath155 closely resemble the corresponding 1d profiles . for quenching to @xmath230 , we have examined time - evolution of @xmath5 . it occurs only along the @xmath121 axis in the very early stage in the top panels of fig.7 , where there is no difference between the results of model h and model b. subsequently , inhomogeneities appear in the @xmath101 plane .
the free energy decrease @xmath269 and the typical velocity amplitude @xmath255 defined in eq.(3.8 ) have been plotted in fig.8 .
the velocity field considerably quicken the interface formation and the coarsening for model h than for model b. profiles of @xmath5 and @xmath210 in the late stage coarsening due to the flow have been presented in fig.9 , where the domain coalescence can be seen and the maximum velocity is of order @xmath270 as in bulk spinodal decomposition @xcite .
we make some remarks .
+ 1 ) in the static part of our theory , we neglect the thermal fluctuations varying in the lateral directions with wavelengths longer than @xmath7 .
thus this 2d transition exhibits mean - field behavior .
in fact , the curves of @xmath153 vs @xmath36 and @xmath2 vs @xmath36 are parabolic near the film criticality in fig.4 . + 2 ) in our simulation , we soon have only two or three domains in the cell as in the bottom panels of fig.9 .
the lateral dimension @xmath271 in this paper is too short to investigate the domain growth law in the @xmath101 plane .
simulation with larger @xmath272 should be performed in future .
+ 3 ) from the van der waals loop of the isothermal curves in the @xmath2-@xmath6 plane in fig.3 , we may predict how phase separation proceeds after quenching .
we have examined phase separation via spinodal decomposition . however , in real experiments , phase separation may occur via nucleation for metastable 1d profiles .
note that hysteretic behavior has been observed in phase - separating fluids in pores and has not been well explained @xcite .
+ 4 ) in a number of experiments and simulations of surface - directed phase separation @xcite , composition waves along the @xmath121 axis have been observed near the wall in the early stage . in these cases ,
the degree of adsorption has changed appreciably upon quenching . in the strong adsorption regime in this paper ,
1d dynamics occurs in the initial stage , but there are no composition waves as in the top panels of fig.7 .
+ 5 ) the static part of this work is applicable to any ising - like near - critical systems and can readily be generalized to @xmath273-component spin systems . in the dynamics , we have used model h with a homogeneous temperature and incompressible flows . on the other hand , in one - component near - critical fluids ,
the latent heat released or absorbed at the interfaces gives rise to significant hydrodynamic flow because of the enhanced isobaric thermal expansion @xcite .
also promising in future should be extension of this work to near - critical fluids in porous media .
+ this work was supported by grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology of japan .
s. y. was supported by the japan society for promotion of science .
would like to thank sanjay puri for informative correspondence . in near - critical fluids
, we treat slow flows with low reynolds numbers .
the total ( reversible ) stress tensor is given by @xmath274 , where @xmath213 is nearly homogeneous throughout the film and the reservoir .
the @xmath275 is the stress tensor due to the composition deviation@xcite , _
ij = p__ij + c ( _ i ) ( _ j ) , where @xmath276 . the diagonal part @xmath277 is written as p_&= & - f - ( c+c)||^2 -c ^2 + & = & ( f/)- f- c ||^2/2 , where @xmath278 . the second part in eq.(a1 ) contains off - diagonal components relevant for curved interfaces . in deriving eq.(3.2 ) , we use the relation , _ j _ j _ ij = _ i ( f/ ) .
k. binder , d. landau , and m. m@xmath279ller , j. stat .
phys . * 110 * , 1411 ( 2003 ) ; m. m@xmath279ller and k. binder , j. phys . : condens .
matter 17 , s333 ( 2005 ) ; k.binder , j. horbach , r. vink , and a. de virgiliis , soft matter , * 4 * , 1555 ( 2008 ) . for binary mixtures ,
we use the chemical potential difference @xmath280 betwen the two components . to be precise ,
@xmath6 in the text is the deviation of the chemical potential difference @xmath281 around the critical composition @xmath16 .
a. onuki , _ phase transition dynamics _
( cambridge university press , cambridge , 2002 ) .
m. c. goh , w. i. goldburg , and c. m. knobler , phys .
* 58 * , 1008 ( 1987 ) ; a. p. y. wong , s. b. kim , w. i. goldburg , and m. h. w. chan , _ ibid .
_ * 70 * , 954 ( 1993 ) .
s. b. dierker and p. wiltzius , phys .
lett . * 58 * , 1865 ( 1987 ) .
the scaled quantity @xmath284 is one of the casimir amplitudes depending on @xmath88 and @xmath285 @xcite .
its value at the bulk criticality is @xmath286 . for @xmath287
, we obtain @xmath288 with @xmath289 just below ccl ( for ( b ) in fig.1 ) and @xmath290 just above ccl ( for ( c ) in fig.1 ) .
in fig.13 of ref.@xcite , we already found a cubic relation between @xmath153 ( instead of @xmath2 ) and @xmath104 ( instead of @xmath6 ) near the film criticality for @xmath291 , which is a mean - field result due to neglect of 2d thermal fluctuations with sizes longer than @xmath7 . | we investigate phase separation of near - critical binary mixtures between parallel symmetric walls in the strong adsorption regime .
we take into account the renormalization effect due to the critical fluctuations using the recent local functional theory @xmath0j .
chem .
phys . * 136 * , 114704 ( 2012)@xmath1 $ ] . in statics ,
a van der waals loop is obtained in the relation between the average order parameter @xmath2 in the film and the chemical potential when the temperature @xmath3 is lower than the film critical temperature @xmath4 ( in the case of an upper critical solution temperature ) . in dynamics , we lower @xmath3 below the capillary condensation line from above @xmath4 .
we calculate the subsequent time - development assuming no mass exchange between the film and the reservoir . in the early stage ,
the order parameter @xmath5 changes only in the direction perpendicular to the walls . for sufficiently deep quenching ,
such one - dimensional profiles become unstable with respect to the fluctuations varying in the lateral directions .
the late - stage coarsening is then accelerated by the hydrodynamic interaction .
a pancake domain of the phase disfavored by the walls finally appears in the middle of the film . |
the system of falling balls was introduced by wojtkowski @xcite .
it describes the motion of @xmath0 , @xmath1 , point masses , with positions @xmath2 , momenta @xmath3 and masses @xmath4 , moving up and down a vertical line and colliding elastically with each other .
the bottom particle collides elastically with a rigid floor placed at position @xmath5 . for convenience
, we will refer to the point particles as balls @xcite . ] .
the system is an autonomous hamiltonian system , with hamiltonian given by the sum of the kinetic and potential energy of each ball .
it possesses a smooth invariant measure with respect to the hamiltonian flow and with respect to a suitable poincar map , describing the movement of the balls from one collision to the next .
one aspect that makes the description of the dynamics cumbersome is the presence of singularities .
these are codimension one manifolds in the phase space , on which the dynamics are not well - defined , in particular the orbit bifurcates .
a point belongs to the singularity manifold , if its next collision is either between three balls or two balls with the floor . for three falling balls
, wojtkowski proved @xcite , that for upward strictly decreasing masses @xmath6 , every point in phase space has non - zero lyapunov exponents .
the present work revolves around three falling balls with this particular mass configuration . since the system of three falling balls has non - zero lyapunov exponents everywhere , the theory of katok - strelcyn @xcite yields , that the phase space partitions into at most countably many components on which the conditional smooth measure is ergodic . a reliable method to check the ergodicity of such systems is the local ergodic theorem @xcite .
it consists of five verifiable conditions , which , if true , give the existence of an open neighbourhood , around a _
least expanding ) . ]
_ point , that lies ( @xmath7 ) in one of the ergodic components . to prove , that there is only one ergodic component
needs the validity of a transitivity argument , in the literature often referred to as global ergodicity .
it states , that one can build a chain of the aforementioned open neighbourhoods from any least expanding point to another .
these neighbourhoods intersect pairwise on a subset of positive measure and , hence , there can only be one ergodic component . for three falling balls only one of the five conditions , as well as the transitivity argument , have been verified @xcite .
we give proofs for two of the remaining four conditions , namely , the chernov - sinai ansatz and the non - contraction property . in terms of @xcite ,
the ansatz will follow as a simple corollary once we prove , that every orbit is _ strictly unbounded _
( see definition [ unboundedness ] ) . in order to prove the non - contraction property we will only need to use the invariance of the contracting cone field .
denote by @xmath8 the phase space , which is partitioned ( @xmath7 ) into subsets @xmath9 , @xmath10 , where each subset describes the moment right after collision of balls @xmath11 and @xmath12 . for @xmath13
, we have a collision with the floor , i.e. @xmath14 .
let @xmath15 be the poincar map , describing the movement from one collision to the next .
after making wojtkowski s convenient coordinate transformation @xmath16 , we get a contracting cone field @xmath17 where @xmath18 denote the coordinates in tangent space .
the cone field is defined by the quadratic form @xmath19 denote by @xmath20 the closure of the cone @xmath21 .
the sequence @xmath22 is called strictly unbounded , if @xmath23 our main result states , that every orbit is strictly unbounded . for every @xmath24
, we have @xmath25 for all @xmath26 .
the singularity manifold on which @xmath27 resp .
@xmath28 is not well - defined is given by @xmath29 resp .
let @xmath31 resp .
@xmath32 be the restriction to @xmath29 resp .
@xmath30 of the smooth @xmath27-invariant measure @xmath33 .
the chernov - sinai ansatz holds once @xmath31 resp .
point is strictly unbounded . for @xmath34 -a.e .
@xmath35 , we have @xmath36 for all @xmath26 . in view of the main theorem ,
it follows immediately , that the ansatz holds , once every orbit is strictly unbounded .
the last result we will prove is the non - contraction property there exists @xmath37 , such that for every @xmath24 , @xmath38 and @xmath26 , we have is an arbitrary norm on @xmath39 . ]
@xmath40 _ outline of the paper _
+ in section 1 we give a formal introduction of the system of three falling balls and state the facts about the existence of lyapunov exponents and ergodicity . at the end of the section
there is a sketch of the proof for strict unboundedness .
section 2 marks the core of this work , where we prove the existence of constant positive lower bounds on the expansion for the iterated boundary of the contracting cone field .
section 3 completes the proofs of the main results by using the constant positive lower bounds obtained in section 2 .
+ in every section , the reader will find a more precise and thorough explanation of the above . + * acknowledgments * i want to cordially thank my advisor henk bruin and pter blint , nandor simnyi , domokos szsz , imre pter tth and maciej wojtkowski for many helpful discussions .
let @xmath41 be the position , @xmath42 the momentum and @xmath43 the velocity of the @xmath12-th ball .
the momenta and the velocities are related by @xmath44 .
we assume that the masses @xmath45 satisfy @xmath46 .
the movements of the balls are a result of a linear potential field and their kinetic energies .
the total energy of the system is given by the hamiltonian function @xmath47 the hamiltonian equations are @xmath48 \dot{p_{i } } & = & -m_{i}. \end{array}\end{aligned}\ ] ] the dots indicate differentiation with respect to time @xmath49 and the hamiltonian vector field on the right hand side will be denoted as @xmath50 .
the solutions to these equations are @xmath51 p_{i}(t ) & = & -tm_{i } + p_{i}(0 ) , \end{array}\end{aligned}\ ] ] which form parabolas in @xmath52 .
it is clear from the choice of the linear potential field , that the acceleration of each ball points downwards and , thus , these parabolas can not escape to infinity .
hence , for every initial condition @xmath53 the balls go through every collision in finite time and , thus , every collision happens infinitely often .
the energy manifold @xmath54 and its tangent space @xmath55 are given by @xmath56 the hamiltonian vector field ( [ equations ] ) gives rise to the hamiltonian flow @xmath57 for convenience , the image will also be written with the time variable as superscript , i.e. @xmath58 .
+ the standard symplectic form @xmath59 induces the symplectic volume element @xmath60 .
we restrict it to @xmath61 , where @xmath62 is any vector satisfying @xmath63 and @xmath64 is the interior derivative .
since the flow preserves the standard symplectic form , it preserves the volume element and , hence , the liouville measure @xmath65 on @xmath54 obtained from it .
we define the poincar section , which describes the states right after a collision as @xmath66 , with @xmath67 in the same way we define the set of states right before collision @xmath68 , by @xmath69 the + resp .
- superscript refer to the states right after resp . before collision .
the system of falling balls is considered as a hard ball system with fully elastic collisions . during a collision of the balls @xmath12 and @xmath70 the momenta resp .
velocities change according to @xmath71 v_{i}^{+ } & = & \gamma_{i } v_{i}^{- } + ( 1 - \gamma_{i})v_{i+1}^{-},\\ v_{i+1}^{+ } & = & ( 1 + \gamma_{i})v_{i}^{- } - \gamma_{i } v_{i+1}^{- } , \end{array}\end{aligned}\ ] ] where @xmath72 , @xmath73 , and when the bottom particle collides with the floor the sign of its momentum is simply reversed @xmath74 these collision laws are described by the linear collision map @xmath75 & ( q , p^{- } ) \mapsto ( q , p^{+}).\nonumber \end{array}\end{aligned}\ ] ] we will write @xmath76 if we do not want to refer to any specific collision . note , that at a collision the flow is discontinuous in the coordinates of the momenta .
we get around this problem by identifying the points @xmath77 and @xmath78 via the collision map @xmath76 .
let @xmath79 be the first return time to @xmath80 .
we define the poincar map as @xmath81 @xmath27 is the collision map , that maps from one collision to the next . by restricting the volume form on @xmath54 with respect to the direction of the flow
we obtain the volume form @xmath82 .
this volume form defines a smooth measure @xmath33 on @xmath8 , which is @xmath27-invariant .
our dynamical system can be stated as the triple @xmath83 .
each @xmath84 and @xmath85 , @xmath86 , further partitions ( @xmath7 ) into @xmath87 it can be calculated , that @xmath88 .
define @xmath89 to be the first return map to @xmath90 .
the system of falling balls possesses codimension one singularity manifolds @xmath91 the states in @xmath92 face a triple collision next , while the states in @xmath93 experience a collision of the lower two balls with the floor next .
the map @xmath27 is not well - defined on these sets , i.e. it bifurcates and , thus , has two different futures .
this happens because we can approximate , say , a triple collision in two ways : once by letting ball one and two collide an instant before ball two and three and vice versa .
we abbreviate @xmath94 .
+ * collision facts * + here are some well - known facts regarding the combinatorial aspect of collisions .
the first two points follow immediately from the hamiltonian equations .
the third point follows from ( * ? ? ?
* proposition 3 ) .
+ [ cols="^ , < , > " , ] since , for every orbit @xmath95 , @xmath24 , every collision happens infinitely often , at least one of the cases ( [ e1 ] ) - ( [ eeee4 ] ) happens infinitely often .
this is equivalent to saying , that at least one collision pattern happens infinitely often .
the exception here is collision pattern 1a , because it does not include a @xmath96 collision .
therefore , if an orbit experiences collision pattern 1a infinitely often , then it must also experience at least one of the other collision patterns infinitely often .
we will first show , that each point @xmath97 experiencing a collision pattern of cases ( [ e1 ] ) and ( [ ee2 ] ) , except for collision pattern 1a , is not truant .
we deal first with case ( [ e1 ] ) .
[ thm1 ] given an energy value @xmath98 , there exists a @xmath99 , such that for all @xmath100 for all @xmath101 and collision patterns 1b , 1c , 1d , we have @xmath102
. _ outline of the proof_. assume on the contrary , that @xmath101 and @xmath103 .
we will show , that in the limiting case @xmath104 , only @xmath105 satisfies this statement .
since the hamiltonian @xmath106 is a continuous function , we have @xmath107 .
hence , for a given energy value @xmath98 , there exists @xmath108 , such that @xmath109 , for all @xmath110 , which contradicts our assumption of constant energy and therefore @xmath102 . in this text , we will call this _ the argument of energy loss_. we will give a proof by contradiction : assume , that @xmath111 satisfies @xmath112 without loss of generality let @xmath113 happen at time @xmath114 .
+ since @xmath111 , we have in each case @xmath115 and , by our assumption @xmath116 , for each case separately ( see figures in appendix p. 46 ) @xmath117 & 0 \leq v_1^+(t_4 ) - v_2^+(t_4 ) \leq c_1,&\nonumber\\[0.1 cm ] & 0 \leq v_1^+(t_5 ) - v_2^+(t_5 ) \leq c_1.&\nonumber\\[0.1 cm ] \end{array}\end{aligned}\ ] ] we are first going to express the above inequalities in terms of the starting velocities @xmath118 . using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , we calculate for each case separately the identities @xmath119 using again ( [ equation2 ] ) , ( [ collisionqp ] ) , ( [ v1p1 ] ) and the above identities , a somewhat lengthy calculation gives @xmath120 & & + 2\sqrt{(v_{1}^{+}(t_1))^{2 } + 2q_{1}(t_1)}\label{e2}\\[0.3 cm ] & v_1^+(t_4 ) - v_2^+(t_4 ) & = ( -\gamma_1 + \gamma_1\gamma_2 + \gamma_2)d_1 + ( \gamma_2 - 1)(v_2^+(0 ) - v_3^+(0))\nonumber\\[0.1 cm ] & & + 2\sqrt{(v_{1}^{+}(t_3))^{2 } + 2q_{1}(t_3)},\label{e3}\\[0.3 cm ] & v_1^+(t_5 ) - v_2^+(t_5 ) & = ( \gamma_2 - \gamma_1\gamma_2 + \gamma_1\gamma_2 ^ 2 + \gamma_2 ^ 2 + 1)d_1 \nonumber\\[0.1 cm ] & & + ( -\gamma_1\gamma_2 + \gamma_1\gamma_2 ^ 2 + \gamma_1 + \gamma_2 ^ 2 + 1)(v_2^+(0 ) - v_3^+(0 ) ) \nonumber\\[0.1 cm ] & & + 2\sqrt{(v_{1}^{+}(t_3))^{2 } + 2q_{1}(t_3)},\label{e4}\end{aligned}\ ] ] the discriminant under the square roots explicitly calculated gives @xmath121 & ( v_{1}^{+}(t_3))^{2 } + 2q_{1}(t_3 ) = d + ( \gamma_2 - \gamma_1\gamma_2 + \gamma_1)^2 ( v_2^+(0 ) - v_3^+(0))^2 \nonumber\\[0.1 cm ] & + 2(\gamma_2 - \gamma_1\gamma_2 + \gamma_1)(v_2^+(0 ) - v_3^+(0))v_3^+(0)\nonumber\\[0.1 cm ] & + 2(1 - \gamma_1)(1 - \gamma_2)(v_2^+(0 ) - v_3^+(0))t_3 + ( v_3^+(0))^2,\nonumber\end{aligned}\ ] ] where @xmath122 is a term depending multiplicatively on @xmath123 . for collision patterns 1b@xmath124 and 1c@xmath124 we have to add @xmath125 to each expression .
+ the travel times @xmath126 , @xmath127 and @xmath128 between collisions are explicity given by @xmath129 & t_3 - t_2 & = \frac{(t_2 - t_1)d_1}{(1 - ( 1 - \gamma_2)(1 + \gamma_1))d_1 + ( 1 - \gamma_2)(v_2^+(0 ) - v_3^+(0))}\label{eq12 } , \end{aligned}\ ] ] where @xmath130 if and only if @xmath131 we are now interested , which points satisfy our assumption , when @xmath104 .
+ * collision pattern 1b . *
+ letting @xmath104 in ( [ d_1 ] ) gives @xmath132 if @xmath133 , then by ( [ eq2 ] ) @xmath134 & \lim_{c_1 \searrow 0 } v_2^+(0 ) - v_3^+(0 ) = 0 .
\end{array}\end{aligned}\ ] ] thus , letting @xmath104 in ( [ e2 ] ) , gives @xmath135 the only starting configurations , that satisfy ( [ e5 mm ] ) are @xmath136 . by ( [ d_11 ] ) and ( [ v23 ] )
, this also implies @xmath137 . in order to keep the total energy constant we always need to have bounded travel times in between collisions . letting @xmath104 in @xmath127 ( [ eq11 ] ) , gives @xmath138 & + \frac{q_3(0)}{d_1(1 + \gamma_1 ) + v_2^+(0 ) - v_3^+(0 ) } - \frac{q_2(0)}{d_1(1 + \gamma_1 ) + v_2^+(0 ) - v_3^+(0)}\bigr).\nonumber\end{aligned}\ ] ] thus , for @xmath139 , we have @xmath140 finally , the above yields , that @xmath141 if @xmath142 , then letting @xmath104 in ( [ e2 ] ) , leads to @xmath143 a sum of two positive summands is equal to zero if and only if both summands are equal to zero .
therefore , the only initial configurations satisfying this equation are @xmath144 . by @xmath145 ,
this also gives @xmath146 . since @xmath147 , we also know that ( [ q_3 ] ) holds .
it follows , that for @xmath142 , we also obtain ( [ zero ] ) .
since the energy function @xmath106 is continuous , this gives @xmath148 hence , for a given energy value @xmath98 , there exists @xmath149 , such that @xmath109 , for all @xmath150 .
since the dynamics are confined to a constant energy level , this yields a contradiction and , hence , @xmath151 . +
* collision pattern 1c .
* + from case 1b we have the validity of ( [ d_11 ] ) , ( [ v23 ] ) and ( [ q_3 ] ) .
additionally , we see from ( [ eq10 ] ) , that in order for @xmath152 to be bounded , we need to have @xmath153 , @xmath154 : it is straightforward , that we can not have @xmath153 , @xmath155 , because then @xmath156 .
hence , @xmath157 if @xmath133 and using the aforementioned facts , we let @xmath104 in ( [ e3 ] ) and obtain @xmath158 by ( [ d_11 ] ) and ( [ v23 ] ) , this gives @xmath159 . combining this with ( [ q_2 ] ) and ( [ q_3 ] )
, we see , that in the limiting case @xmath104 , ( [ zero ] ) holds and we can use the argument of energy loss as in case 1b . if @xmath142 , we use ( [ d_11 ] ) , ( [ q_3 ] ) , ( [ q_2 ] ) and the travel times ( [ eq10 ] ) - ( [ eq12 ] ) in order to calculate @xmath160 .
thus , letting @xmath104 in ( [ e3 ] ) , we obtain @xmath161 & + 2(\gamma_2 - \gamma_1\gamma_2 + \gamma_1)(v_2^+(0 ) - v_3^+(0))v_3^+(0)\nonumber\\[0.1 cm ] & + 2(1 - \gamma_1)(1 - \gamma_2)q_3(0 ) + ( v_3^+(0))^2 \bigr)^{\frac{1}{2 } } = 0.\label{eq13}\end{aligned}\ ] ] contrary to the case before , where all the initial states were converging to the trivial fixed point @xmath105 for @xmath104 , the solutions of ( [ eq13 ] ) form , a priori , a two dimensional surface in phase space , because of the opposite sign the above terms possess .
further , we get from ( [ eq13 ] ) , that if @xmath162 , then @xmath163 as well and , thus , from ( [ d_11 ] ) , @xmath164 . therefore , if @xmath165 goes to zero , the initial states satisfying ( [ eq13 ] ) approach the trivial fixed point and , hence , we could carry out the same argument of energy loss as in the previous case .
it is left to prove , that @xmath162 .
observe , that @xmath166 which implies @xmath167 , @xmath154 .
this means , that @xmath168 since the last collision , prior to the floor collision at time @xmath169 , is between balls one and two , we see , that the position @xmath170 is arbitrarily close to zero if and only if the difference of the velocities @xmath171 is arbitrarily close to zero .
we calculate , that @xmath172 . this implies , that @xmath170 is arbitrarily close to zero if and only if @xmath165 is arbitrarily close to zero . hence , @xmath173 follows from ( [ eq14 ] ) . + * collision pattern 1d . *
+ from the previous cases we have the validity of ( [ d_11 ] ) , ( [ v23 ] ) , ( [ q_2 ] ) and ( [ q_3 ] ) .
using this and assuming @xmath133 , we let @xmath104 in ( [ e4 ] ) and obtain @xmath174 exactly as in case 1c , we get , that in the limiting case @xmath104 , ( [ zero ] ) holds and we can apply the argument of energy loss again . if @xmath142 , we saw in case 1c , that @xmath160 .
letting @xmath104 in ( [ e4 ] ) and utilizing the latter information , gives @xmath175 & + 2 \bigl((\gamma_2 - \gamma_1\gamma_2 + \gamma_1)^2 ( v_2^+(0 ) - v_3^+(0))^2 \nonumber\\[0.1 cm ] & + 2(\gamma_2 - \gamma_1\gamma_2 + \gamma_1)(v_2^+(0 ) - v_3^+(0))v_3^+(0)\nonumber\\[0.1 cm ] & + 2(1 - \gamma_1)(1 - \gamma_2)q_3(0 ) + ( v_3^+(0))^2 \bigr)^{\frac{1}{2 } } = 0.\label{eq15}\end{aligned}\ ] ] the signs of both terms above are positive , hence , this equation is zero only for the trivial fixed point @xmath105 .
like before , we can apply the argument of energy loss .
+ setting @xmath176 to be the minimum of all the three constants obtained through the argument of energy loss from collision patterns 1b , 1c , 1d , yields the statement .
the same holds for ( [ ee2 ] ) .
[ thm2 ] given an energy value @xmath98 , there exists a @xmath177 , such that for all @xmath178 for all @xmath179 and collision patterns 2a , 2b , 2c , 2d , we have @xmath180 . the proof is carried out again by applying the argument of energy loss similar to theorem [ thm1 ] . in order not to disturb the reading flow we put it in the appendix .
[ rem0.5 ] the results of theorem [ thm1 ] and [ thm2 ] remain the same if we substitute the single @xmath181 collision happening at point @xmath182 with @xmath183 , @xmath184 , @xmath181 collisions . a reasonably quick calculation yields , that these @xmath183 floor collisions show their influence only by scaling the square root terms in ( [ e2 ] ) , ( [ e3 ] ) , ( [ e4 ] ) , by a factor @xmath183 .
[ rem1 ] theorem [ thm1 ] and [ thm2 ] also hold in backward time for the map @xmath28 and the sets @xmath185 , @xmath186 , with the respective collision patterns appropriately inverted .
for instance , the inverted or _ inversed collision pattern _
1b is depicted in figure [ fig:1binv ] .
the proof of this fact follows almost verbatim from the proof of theorem [ thm1 ] and [ thm2 ] . if @xmath187 for @xmath188 resp .
@xmath189 in theorem [ thm1 ] resp .
[ thm2 ] , then @xmath190 lies in a subset of codimension two . indeed , for @xmath187 the velocities @xmath191 resp .
@xmath192 are equal , after the lowest ball leaves the floor . since @xmath193 resp .
@xmath194 must be bounded , we saw in the proof of theorem [ thm1 ] resp .
[ thm2 ] , that @xmath195 resp .
@xmath196 , when @xmath187 .
hence , for @xmath188 and @xmath187 , we have the restrictions @xmath197 , @xmath198 , while for @xmath189 and @xmath187 , we have the restrictions @xmath199 , @xmath196 , @xmath200 , forming a codimension two subset in @xmath8
. the argument of energy loss does not give a satisfactory result when applied to the collision patterns of cases ( [ eee3 ] ) and ( [ eeee4 ] ) .
there are additional configurations to the trivial fixed point @xmath105 , which satisfy the limiting case @xmath104 .
however , by using point ( [ cf.2 ] ) of the collision facts , it is possible to extend those collision patterns and apply the argument of energy loss to the extended collision patterns .
we will make use of this in the proof of the next corollary , when proving , that there are no truant points .
set @xmath201 .
[ cor1 ] there are no truant points , i.e. given an energy value @xmath98 , there exists a @xmath202 , such that for all @xmath203 for all @xmath204 , there exists @xmath205 , such that @xmath206 .
every orbit experiences at least one collision pattern , except for 1a , of the cases ( [ e1 ] ) - ( [ eeee4 ] ) infinitely often .
if this collision pattern belongs to ( [ e1 ] ) resp .
( [ ee2 ] ) , then the claim follows from theorems [ thm1 ] resp .
[ thm2 ] . using ( [ cf.2 ] )
we can extend collision pattern 3a of case ( [ eee3 ] ) in the following four ways depicted in figures [ fig:3a1 ] - [ fig:3a4 ] below .
we see , that collision pattern 2a is part of collision pattern 3a1 ( partially blue colored balls in fig .
[ fig:3a1 ] ) .
now the claim for collision pattern 3a1 follows from theorem [ thm2 ] . for collision patterns
3a2 , 3a3 and 3a4 , we detect the inversed collision patterns 1b , 2c and 1d , inside of them ( partially blue colored balls in fig .
[ fig:3a2 ] - [ fig:3a4 ] ) .
the claim now follows from theorem [ thm1 ] , [ thm2 ] and remark [ rem1 ] .
the statement for collision pattern 3b follows in exactly the same way .
as before , we use ( [ cf.2 ] ) again to extend collision patterns 4a resp .
4b of case ( [ eeee4 ] ) in the following way depicted in figures [ fig:4aext ] and [ fig:4bext ] below .
collision pattern 4a resp .
4b partially consist of the inversed collision patterns 1b resp .
the inversed collision pattern 1b ranges from time @xmath207 to @xmath169 in fig .
[ fig:4aext ] and the inversed collision pattern 1d ranges from time @xmath207 to @xmath208 in fig .
[ fig:4bext ] .
thus , the statement follows from theorem [ thm1 ] , [ thm2 ] and remark [ rem1 ] .
a @xmath124 symbol added to the collision pattern , e.g. @xmath209 , describes the same collision pattern except for the first @xmath181 collision being changed to a @xmath210 or @xmath96 collision depending on the first ball to ball collision in the collision pattern .
for instance , the @xmath181 collision in collision pattern @xmath209 is changed to a @xmath96 collision , since the first ball to ball collision in collision pattern 1b is @xmath210 . in case of @xmath211 the @xmath181 collision is changed to a @xmath210 collision , since the first ball to ball collision in the pattern 2b is @xmath96 . for considerations in the next subsection
, we will need the following two results .
[ cor2 ] given an energy value @xmath98 , there exists a @xmath212 , such that for all @xmath213 for all @xmath214 and collision patterns 1b@xmath124 , 1c@xmath124 , 2b@xmath124 , 2c@xmath124 , we have @xmath215 . the proof of this statement follows along the same lines as the proofs of theorems [ thm1 ] and [ thm2 ] .
[ prop1 ] given an energy value @xmath98 , there exists a @xmath216 , such that for all @xmath217 for all @xmath218 and collision patterns 1 .
3a1 , 3a1@xmath124 and 2 . 3a2 , 3a2@xmath124 , we have 1 .
@xmath219 and 2 .
@xmath220 .
the proof is carried out again by applying the argument of energy loss similar to theorem [ thm1 ] and [ thm2 ] . in order not to disturb the reading flow we put it in the appendix . in this subsection
we will prove point 2 of our two step approach ( see p. 15 ) .
for @xmath222 , @xmath223 , we will equivalently say , that the collision @xmath224 emerges from @xmath221 . in order to prove , that every orbit @xmath95 , @xmath24 , experiences infinitely many @xmath210 and @xmath96 collisions emerging from @xmath221 , which are at most separated by multiple @xmath181 collisions , we will subdivide point 2 as follows : c p10 cm + + i. & we will first prove , that for every collision pattern , for which every @xmath96 collision does not emerge from @xmath221 , the next @xmath96 collision ( in positive time ) emerges from @xmath221 .
& the final step consists , in proving , that for every @xmath96 collision emerging from @xmath221 , we can find a @xmath210 collision emerging from @xmath221 , such that both collisions are at most separated by multiple @xmath181 collisions .
+ [ tab:1 ] let @xmath225 be the set of all collision patterns except 1a and 1b . for technical reasons , in order to prove i , we need to split @xmath8 , into @xmath226 & \mathcal{r}_2 = \mathcal{m}^+ \setminus \mathcal{r}_1.\nonumber\end{aligned}\ ] ] \1 .
let @xmath227 denote the number of returns to @xmath8 for @xmath190 until the first @xmath228 collision occurs .
let @xmath229 , @xmath230 , be the collision time of the @xmath183-th collision prior to the next collision @xmath228 of point @xmath190 . in the next two results we prove
case i for all points @xmath231 .
[ thm3 ] let @xmath232 mark the final collision prior to the @xmath96 collision in collision patterns 1c , 2a , 2b .
given an energy value @xmath98 , there exists a @xmath233 , such that for all @xmath234 for all @xmath235 , and collision patterns 1c , 2a , 2b , we have @xmath236 @xmath237see fig .
[ fig : thm2]@xmath238 . + with collision pattern 1c in the beginning . ]
the proof of the theorem will once more use the argument of energy loss .
we will assume on the contrary @xmath239 and show , that for @xmath104 , ( [ another ] ) only holds for the trivial fixed point @xmath240 .
the crucial step in the proof is , that even though for point @xmath241 of the collision patterns 1c , 2a , 2b , we have @xmath242 ( [ another ] ) will imply @xmath243 and therefore , since @xmath244 , we can deduce the result from theorem [ thm1 ] , [ thm2 ] and corollary [ cor2 ]
. we will give a proof by contradiction : assume on the contrary , that @xmath244 , satisfies @xmath245 . since we want to apply the argument of energy loss , the strategy consists again in expressing @xmath246 in terms of the velocities @xmath118 , at time zero . without loss of generality , we can assume , when calculating it back to @xmath118 , remains in both cases @xmath247 resp . @xmath248 the same . ] that @xmath247 .
since @xmath249 , we have @xmath250 and , by our assumption @xmath245 @xmath251 when calculating @xmath246 back into velocities @xmath118 , we have to distinguish between the collision patterns 1c , 2a , 2b , marking the beginning of our orbit segment at @xmath190 .
further , we assume for each collision pattern to have @xmath183 , @xmath252 , ( 0,1 ) @xmath253 ( 1,2 ) collisions after point @xmath182 to point @xmath254 ( see again fig . [
fig : thm2 ] ) . using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , ( [ v1p1 ] )
, a somewhat lengthy calculation gives for each collision pattern @xmath255 for collision pattern 2a this simplifies to @xmath256 and for collision patterns 1c , 2b we obtain @xmath257 where @xmath258 for collision pattern 2a and @xmath259 for collision patterns 1c , 2b .
+ we are now interested , which points satisfy our assumption , when @xmath104 . + * collision pattern 2a . *
+ letting @xmath104 in ( [ d_222 ] ) gives @xmath260 using this , we also let @xmath104 in ( [ ( 2,3)1 ] ) and receive , @xmath261 since the summands @xmath262 , for all @xmath263 , in the above sum are non - negative , equality ( [ g1 ] ) holds if and only if @xmath264 combining @xmath265 with ( [ d_22 mm ] ) , we can apply theorem [ thm2 ] to collision pattern 2a .
+ * collision patterns 1c , 2b . *
+ from the previous collision patterns we know , that ( [ d_22 mm ] ) holds . using this
, we let @xmath104 in ( [ ( 2,3)2 ] ) and receive , @xmath266 as for collision patterns 2a , we see , that this is a sum with non - negative summands . in particular , the limit in ( [ g2 ] ) implies , that @xmath267 .
observe , that collision pattern 1c is shorten by one collision since @xmath268 marks the first ( 1,2 ) collision , instead of the first ( 0,1 ) collision .
therefore , the collisions from @xmath190 to @xmath269 constitute collision pattern 2b@xmath124 ( partially blue colored balls in fig .
[ fig:1c2bast ] ) . inside collision pattern 1c . ] as before , we combine @xmath270 , with ( [ d_22 mm ] ) and apply corollary [ cor2 ] to collision pattern 2b@xmath124 and theorem [ thm2 ] to collision pattern 2b . the next corollary deals with the remaining collision patterns 1d , 2c and 2d .
[ cor3 ] let @xmath232 mark the final collision prior to the first @xmath96 collision and @xmath271 mark the last collision prior to the second @xmath96 collision in collision patterns 1d , 2c , 2d .
given an energy value @xmath98 , there exists a @xmath272 , such that for all @xmath273 for all @xmath235 , @xmath274 and collision patterns 1d , 2c , 2d , we have @xmath275 .
the proof follows almost exactly like the proof of theorem [ thm3 ] so we omit it .
+ theorem [ thm3 ] also holds for collision pattern 1b , if between collision pattern 1b and the next @xmath96 collision are at least @xmath183 , @xmath276 , @xmath277 collisions .
however , in the special case when @xmath278 , it does not hold . indeed ,
if after collision pattern 1b we immediately face a @xmath96 collision , additional points , except @xmath105 , satisfy the assumptions of theorem [ thm3 ] in the limiting case @xmath104 and , thus , the argument of energy loss does not apply . in order to deal with the special case @xmath278
, we look at three @xmath279 collisions separated only by a @xmath181 collision ( see fig .
[ fig:31b ] ) .
this proves case i for all orbits @xmath95 , with @xmath280 .
note , that this special case , when @xmath278 for collision pattern 1b , is why we had to split @xmath8 into @xmath281 and @xmath282 in the first place .
[ lem10 ] let @xmath283 mark the first @xmath210 collision in the collision pattern depicted in figure [ fig:31b ] .
given an energy value @xmath98 , there exists a @xmath284 , such that for all @xmath285 for all @xmath286 and @xmath287 , we have @xmath288 .
the proof is carried out again by applying the argument of energy loss . in order not to disturb the reading flow we put it in the appendix . for collision patterns
3a , 3b , 4a , 4b we already know from the proof of corollary [ cor1 ] , that they extend to one of the inversed collision patterns 1b , 1d , 2c or to collision pattern 2a . therefore , theorem [ thm3 ] and corollary [ cor3 ] also apply to them .
this concludes the proof of point i ( see p. 23 ) .
+ for point ii , we first need the following proposition [ prop2 ] 1 .
let @xmath289 , @xmath290 , @xmath291 . then @xmath292 .
let @xmath289 and @xmath293 .
then @xmath292 .
let @xmath268 , @xmath293 and @xmath294
. then @xmath292 .
the illustration of statements 1 .
, 2 . and 3 .
is given in figures [ fig : ill1 ] , [ fig : ill2 ] .
for @xmath295 , we have @xmath296 the term @xmath165 is negative , because @xmath297
. choose @xmath298 , such that @xmath299 .
then the above inequality holds if and only if @xmath300 this implies , that @xmath292 .
+ statements 2 . and 3 .
follow in the same way as statement 1 .
we are now in the position to prove point ii .
let @xmath301 , where @xmath302 are the constants appearing in theorem [ thm1 ] , [ thm2 ] , [ thm3 ] , corollary [ cor1 ] , [ cor2 ] , [ cor3 ] , lemma [ lem10 ] and proposition [ prop1 ] , [ prop3 ] , [ prop4 ] .
[ thm4 ] for every @xmath303 , we either have @xmath304 , @xmath305 or @xmath306 resp . @xmath307 , with @xmath308 resp .
@xmath309 .
we distinguish between collision patterns with one resp .
two ( 2,3 ) collisions . in the former case , these are 1b , 1c , 2a , 2b and in the latter case , we have 1d , 2c , 2d . for collision patterns
3a , 3b , 4a , 4b , we obtain the statement of the theorem by extending them again to the inversed collision patterns 1b , 1d , 2c and the collision pattern 2a , as we did in corollary [ cor1 ] .
we begin with proving the claim for collision patterns with only one ( 2,3 ) collision . for technical reasons
we postpone collision pattern 1c to the end .
+ note , that @xmath310 , due to the definition of @xmath311 . + * 1 .
collision patterns 1b , 2a , 2b . *
+ _ a. collision pattern 1b . _
+ due to theorem [ thm1 ] , we know , that either @xmath312 or @xmath102 . in either case the statement follows . + _
b. collision pattern 2a .
_ + using ( [ cf.2 ] ) , we extend collision pattern 2a backwards and obtain either collision pattern 3a1 or 3a1@xmath124 ( see fig .
[ fig:3a1ext ] , [ fig:3a1ast ] ) . .
] . ]
due to proposition [ prop1 ] , in both cases , either @xmath313 or @xmath314 , which yields the assertion . + _ c. collision pattern 2b . _
+ using ( [ cf.2 ] ) , we extend collision pattern 2b backwards ( see fig .
[ fig:2bext ] ) . the partially blue colored balls in the figure mark the inversed collision pattern 1b . due to theorem 1 and remark [ rem0.5
] , either @xmath315 or @xmath316 . in either case
the theorem follows .
collision patterns 1d , 2c , 2d . *
+ _ a. collision pattern 1d .
_ + figures [ fig:1d1 ] and [ fig:1d2 ] depict the possible placement of @xmath303 . .
] . ] in figure [ fig:1d1 ] , we have @xmath305 by proposition [ prop2 ] point 2 and the statement follows . in figure [ fig:1d2 ]
, we apply proposition [ prop2 ] point 3 , and obtain @xmath317 , which yields the claim . + _
b. collision pattern 2c .
_ + we have the following two possibilities for the placement of @xmath303 ( see fig . [ fig:2cleft ] , [ fig:2cright ] ) . .
] . ] in figure [ fig:2cleft ] , we have @xmath305 , which follows from proposition [ prop2 ] point 2 . using ( [ cf.2 ] ) , we extend figure [ fig:2cright ] backwards and obtain either to 3a1 or 3a1@xmath124 ( partially blue colored balls in fig .
[ fig:2cext ] ) .
we know from proposition [ prop1 ] , that either @xmath313 or @xmath318 . for both possibilities the assertion follows . + _ c. collision pattern 2d . _
+ the statement follows in the same way as collision pattern 2c above .
+ * 3 . collision pattern 1c .
* + we extend this collision pattern backward in the following two ways ( see fig .
[ fig:1cext1 ] and [ fig:1cext2 ] ) . in figure [ fig:1cext1 ] , we either have @xmath312 or @xmath102 , due to theorem [ thm1 ] . if @xmath312 , we are done .
assume therefore , that @xmath319 .
using theorem [ thm1 ] again for the reversed collision pattern 1c , we obtain that either @xmath320 or @xmath316 . in the latter case , we are done again .
thus , the only case left to treat is @xmath321 and @xmath319 .
unfortunately , it does not follow as immediatly as the other cases above : we have to extend the collision pattern backwards until we encounter the first collision pattern in the past with a ( 2,3 ) collision and then exercise a case by case analysis in order to derive , that @xmath322 . the proof of this fact follows from proposition [ prop3 ] and [ prop4 ] , which are stated and proven in the appendix .
both propositions are carried out similarly to the proof of theorem [ thm3 ] .
+ in figure [ fig:1cext2 ] , we have @xmath305 , which is a direct consequence of proposition [ prop2 ] point 1 . theorem [ thm4 ] yields ,
that @xmath323 in lemma [ lem4 ] can be replaced with the fixed positive constant @xmath324 . combining this with the result of point
i and setting @xmath325 , we obtain [ cor4 ] there exists a positive constant @xmath326 , such that for all @xmath327 , there exists @xmath328 and for all @xmath24 , there exists a sequence of strictly positive increasing integers @xmath329 : @xmath330
it remains to prove the divergence of the least expansion coefficient and conditions [ condi1 ] - [ condi2 ] .
the divergence of the least expansion coefficient will follow , once we prove , that the sequence @xmath331 is strictly unbounded .
this is the claim of theorem [ thm ] . we know from wojtkowski s results , that @xmath336 is invariant for every @xmath24 .
therefore , it is enough to prove ( [ ass ] ) along a subsequence @xmath337 .
set @xmath338 where @xmath329 is the sequence of strictly positive integers from corollary [ cor4 ] . according to corollary [ cor4 ] and lemma [ lem4 ]
, we can estimate for all @xmath339 , @xmath340 & \geq & q\bigl ( \begin{pmatrix } r_1r_2 & r_1a_2 + a_1r_2^t\\ 0 & r_1^t r_2^t \end{pmatrix } \begin{pmatrix } \delta \xi_{n_{k-1}}\\ \delta \eta_{n_{k-1 } } \end{pmatrix } \bigr)\nonumber \\[0.2 cm ] & = & \langle r_1r_2 \delta \xi_{n_{k-1 } } + ( r_1a_2 + a_1r_2^t)\delta \eta_{n_{k-1}},\ r_1^t r_2^t \delta \eta_{n_{k-1 } } \rangle \\[0.2 cm ] & = & \langle r_1r_2 \delta \xi_{n_{k-1}},\ r_1^t r_2^t \delta \eta_{n_{k-1 } } \rangle + \langle ( r_1a_2 + a_1r_2^t)\delta \eta_{n_{k-1}},\ r_1^t r_2^t \delta \eta_{n_{k-1 } } \rangle \nonumber \\[0.2 cm ] & \geq & \langle \delta \xi_{n_{k-1}},\ \delta \eta_{n_{k-1}}\rangle + \frac{\lambda}{\lambda_2 ^ 2 } = q_{n_{k-1 } } + \frac{\lambda}{\lambda_2 ^ 2}. \nonumber \end{aligned}\ ] ] denote by @xmath348 the euclidean norm .
observe , that ( [ est1 ] ) holds if and only if @xmath349 , where @xmath350 is a constant , different to @xmath351 in ( [ est1 ] ) , obtained due to equivalence of norms in @xmath352 . using the notation in the proof of lemma [ cond1 ] , we obtain @xmath353 observe , that @xmath354 , for all @xmath38 .
it is clear , that for every @xmath355 , @xmath356 .
since @xmath336 is invariant , @xmath357 is a positive monotone increasing sequence for every @xmath24
. therefore @xmath358 , for every @xmath38 and @xmath26 .
the constant @xmath351 in ( [ est1 ] ) is obtained due to the equivalence of norms in @xmath352 . since @xmath359
, we have @xmath361 and , by our assumption @xmath362 , @xmath363 & 0 \leq v_1^+(t_3 ) - v_2^+(t_3 ) \leq c_1 , & \text{(case 2b}),\\[0.1 cm ] & 0 \leq v_1^+(t_4 ) - v_2^+(t_4 ) \leq c_1 , & \text{(case 2c}),\\[0.1 cm ] & 0 \leq v_1^+(t_5 ) - v_2^+(t_5 ) \leq c_1 , & \text{(case 2d } ) .
\end{array}\end{aligned}\ ] ] we are first going to express the above inequalities in terms of the starting velocities @xmath118 . using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] )
, we calculate the identity @xmath364
@xmath365 & & + ( v_1^+(0 ) - v_2+(0))\label{e4.5}\\[0.3 cm ] & v_1^+(t_3 ) - v_2^+(t_3 ) & = ( \gamma_2 - 1)d_2 - ( v_1^+(0 ) - v_2^+(0))\nonumber\\[0.1 cm ] & & + 2\sqrt{(v_{1}^{+}(t_2))^{2 } + 2q_{1}(t_2)},\label{e55}\\[0.3 cm ] & v_1^+(t_4 ) - v_2^+(t_4 ) & = ( \gamma_1 + 2\gamma_1\gamma_2 + \gamma_1\gamma_2 ^ 2 + \gamma_2 ^ 2 -1)d_2 \nonumber\\[0.1 cm ] & & + ( \gamma_1 - \gamma_2 -\gamma_1\gamma_2)(v_1^+(0 ) - v_2^+(0))\nonumber\\[0.1 cm ] & & + 2\sqrt{(v_{1}^{+}(t_2))^{2 } + 2q_{1}(t_2)},\label{e6}\\[0.3 cm ] & v_1^+(t_5 ) - v_2^+(t_5 ) & = ( -\gamma_1 + 2\gamma_1\gamma_2 - \gamma_1\gamma_2 ^ 2 + \gamma_2 ^ 2 + 1)d_2 \nonumber\\[0.1 cm ] & & + ( -\gamma_1 + \gamma_2 + \gamma_1\gamma_2)(v_1^+(0 ) - v_2^+(0 ) ) \nonumber\\[0.1 cm ] & & + 2\sqrt{(v_{1}^{+}(t_4))^{2 } + 2q_{1}(t_4)}.\label{e7}\end{aligned}\ ] ] the discriminant under the square roots explicitly calculated gives @xmath366 & + 2\gamma_1(v_1^+(0 ) - v_2^+(0))v_3^+(0 ) + ( v_3^+(0))^2 , \nonumber\\[0.2 cm ] & ( v_{1}^{+}(t_4))^{2 } + 2q_{1}(t_4 ) = g_2 + \bigl(((1 - \gamma_1)\gamma_2 + \gamma_1)(1 + \gamma_1 ) - \gamma_1 \bigr)^2 ( v_1^+(0 ) - v_2^+(0))^2 \nonumber\\[0.2 cm ] & + 2\bigl(((1 - \gamma_1)\gamma_2 + \gamma_1)(1 + \gamma_1 ) - \gamma_1\bigr)(v_1^+(0 ) - v_2^+(0))v_3^+(0)+ ( v_3^+(0))^2\nonumber\\[0.2 cm ] & + 2t_4(\gamma_1 - \gamma_2 - \gamma_1 ^ 2(\gamma_2 -1))(v_1^+(0 ) - v_2^+(0 ) ) + 2t_2(1 - \gamma_1)(v_1^+(0 ) - v_2^+(0 ) ) , \nonumber\end{aligned}\ ] ] where @xmath367 are terms depending multiplicatively on @xmath368 , with @xmath369 .
the travel times @xmath126 , @xmath127 and @xmath128 between collisions are explicitly given by letting @xmath104 in the latter equation and using ( [ 3v_1 ] ) , ( [ q_3q_2 ] ) and ( [ q_2t_2 ] ) , gives @xmath382 which holds if and only if @xmath383 . from ( [ 3v_1 ] ) and ( [ q_3q_2 ] ) , we deduce , that in the limiting case @xmath104 , @xmath384 hence , for a given energy value @xmath98 , there exists @xmath385 , such that @xmath109 , for all @xmath386 .
since the dynamics are confined to a constant energy level , this yields a contradiction and , hence , @xmath180 . +
* collision pattern 2b . *
+ from case 2a we have the validity of ( [ d_22 ] ) and ( [ q_3q_2 ] ) . if @xmath387 , then by ( [ eq0 ] ) @xmath388 & \text{which implies}\ \lim_{c_1 \searrow 0 } v_1^+(0 ) - v_2^+(0 ) = 0 \end{array}\end{aligned}\ ] ] if and only if @xmath391 & \text{which implies}\ \lim_{c_1 \searrow 0 } q_2(0 ) = 0 , \end{array}\end{aligned}\ ] ] and , therefore , in the limiting case @xmath104 , we have again ( [ zero0 ] ) . now
the argument of energy loss applies .
if @xmath392 , we calculate from ( [ d_22 ] ) and ( [ eq17 ] ) , that @xmath393 using the latter , we also let @xmath104 in ( [ e55 ] ) and obtain @xmath394 as in the proof of collision pattern 1c , we see , that the solutions of the above equation form , a priori , a two dimensional surface in phase space , since the summands have opposite sign . the remainder of the proof of collision pattern 2b will now be carried out in exactly the same way as the one of collision pattern 1c . from ( [ eq18 ] )
we see , that if @xmath395 , then the only possible solutions are @xmath396 . by ( [ q_3q_2 ] ) , this gives @xmath397 and by ( [ d_22 ] ) it follows , that @xmath398 .
therefore , if @xmath399 then only the trivial fixed point @xmath105 satisfies ( [ eq18 ] ) and , thus , we could carry out the argument of energy loss again .
it is left to prove ( [ eq19 ] ) .
we have , that @xmath400 since the last collision , prior to the floor collision at time @xmath403 , is between balls one and two , we see , that the position @xmath404 is arbitrarily close to zero if and only if the difference of the velocities @xmath405 is arbitrarily close to zero .
we calculate , that @xmath406 .
this implies , that @xmath404 is arbitrarily close to zero if and only if @xmath407 is arbitrarily close to zero .
hence , @xmath408 follows from ( [ eq20 ] ) . + * collision pattern 2c . *
+ from the previous cases we have the validity of ( [ d_22 ] ) , ( [ q_3q_2 ] ) and ( [ v12 ] ) . if @xmath387 , we let @xmath104 in ( [ e6 ] ) and obtain @xmath409 by ( [ d_22 ] ) and ( [ v12 ] ) , this implies @xmath159 .
now , the rest of the proof for @xmath387 is executed in exactly the same way as the one in case 2b , therefore , we can apply the argument of energy loss again in the end .
if @xmath410 , we calculate using ( [ d_22 ] ) and ( [ eq17 ] ) , that @xmath393 . utilizing these properties and letting @xmath104 in ( [ e6 ] ) , we receive @xmath411 we see again , that if @xmath395 , then the trivial fixed point @xmath105 is the only solution of ( [ eq21 ] ) and , thus , we could carry out the same argument of energy loss as in the previous case .
it remains to prove , that @xmath395 .
we see , from @xmath166 that @xmath167 , @xmath154 and , hence , ( [ eq14 ] ) holds .
we distinguish between two different cases . if @xmath412 , then @xmath170 is arbitrarily close to zero if and only if @xmath413 and @xmath414 are arbitrarily close to zero .
this holds , because otherwise the second ball can not reach the first ball arbitrarily close when it hits the floor at time @xmath169 .
thus , it follows from ( [ eq14 ] ) , that @xmath270 , which explicitly calculated , reduces to @xmath415 & = \lim_{c_1 \searrow 0 } ( \gamma_1 - \gamma_2 - \gamma_1\gamma_2)(v_1^+(0 ) - v_2^+(0 ) ) = 0.\label{eq22}\end{aligned}\ ] ] by assumption @xmath416 , equation ( [ eq23 ] ) is a sum of two positive summands , which , for @xmath104 , is arbitrarily close to zero if and only if both summands are arbitrarily close to zero . for @xmath104
, it follows from ( [ eq22 ] ) , that @xmath413 is arbitrarily close to zero if and only if @xmath407 is arbitrarily close to zero .
this settles the prove of collision pattern 2c .
+ * collision pattern 2d .
* + from the previous cases we have the validity of ( [ d_22 ] ) , ( [ q_3q_2 ] ) and ( [ v12 ] ) .
if @xmath387 , then by ( [ d_22 ] ) and ( [ v12 ] ) , this implies @xmath159 .
we let @xmath104 in ( [ e7 ] ) and obtain @xmath418 additionally , we see from ( [ eq17 ] ) , that @xmath390 if and only if @xmath419 & \text{which implies}\ \lim_{c_1 \searrow 0 } q_2(0 ) = 0 , \end{array}\end{aligned}\ ] ] and , therefore , in the limiting case @xmath104 , we have again @xmath420 , @xmath86 .
now the argument of energy loss applies .
+ if @xmath410 , we let @xmath104 in ( [ e7 ] ) and receive @xmath421 & + 2\bigl(((1 - \gamma_1)\gamma_2 + \gamma_1)(1 + \gamma_1 ) - \gamma_1\bigr)(v_1^+(0 ) - v_2^+(0))v_3^+(0)+ ( v_3^+(0))^2\nonumber\\[0.2 cm ] & + 2t_4(\gamma_1 - \gamma_2 - \gamma_1 ^ 2(\gamma_2 -1))(v_1^+(0 ) - v_2^+(0 ) ) + 2t_2(1 - \gamma_1)(v_1^+(0 ) - v_2^+(0))\bigr)^{\frac{1}{2 } } = 0.\nonumber\end{aligned}\ ] ] since the last collision , prior to the floor collision at time @xmath425 , is between balls one and two , we see , that the position @xmath426 is arbitrarily close to zero if and only if the difference of the velocities @xmath427 is arbitrarily close to zero .
we calculate , that @xmath428 this implies , that @xmath426 is arbitrarily close to zero if and only if @xmath407 is arbitrarily close to zero .
hence , @xmath408 follows from ( [ eq144 ] ) . + setting @xmath429 to be the minimum of all the four constants obtained through the argument of energy loss from collision patterns 2a , 2b , 2c , 2d , yields the statement
. we will give a proof by contradiction : assume , that @xmath430 satisfies 1 .
@xmath431 , or 2 . @xmath432 . without loss of generality
let @xmath113 happen at time @xmath114 .
+ since @xmath430 , we have @xmath433 and , by our assumption 1 .
@xmath434 , or 2 .
@xmath432 @xmath435 we are first going to express the above inequality in terms of the starting velocities @xmath118 . using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , ( [ v1p1 ] ) , we calculate @xmath436 & v_1^+(t_3 ) - v_2^+(t_3 ) & = 2\sqrt{(v_1^+(t_1))^2 + 2q_1(t_1 ) } - d_1\nonumber\\ & & + ( 1-\gamma_2)(v_2^+(t_2 ) - v_3^+(t_2)),\ ( \text{for}\ 3a1,\ 3a1^{\ast}),\label{e222}\\[0.3 cm ] & v_1^+(t_4 ) - v_2^+(t_4 ) & = ( ( 1-\gamma_2)(1+\gamma_1 ) -1)d_1 + ( 1 -\gamma_2)(v_2^+(0 ) - v_3^+(0 ) ) \nonumber\\[0.1 cm ] & & + 3\sqrt{(v_{1}^{+}(t_1))^{2 } + 2q_{1}(t_1)},\ ( \text{for}\ 3a2,\ 3a2^{\ast}).\label{e4.4}\end{aligned}\ ] ] in the first case , we have due to ( [ d_1111 ] ) , the equality of all velocities @xmath447 . but
this contradicts @xmath448 from ( [ doh ] ) , unless all the velocities are zero .
we further get from the second fraction in ( [ eq101 ] ) , that @xmath378 , only if @xmath397 . combining this with ( [ q_222 ] )
we see , that in case 1 .
only the trivial fixed point @xmath105 satisfies our assumption and , thus , we can apply the argument of energy loss .
+ in the second case an easy calculation shows , that @xmath449 however , due to ( [ doh ] ) we have @xmath450 , since the first ball bounces off the floor . if @xmath451 , this gives a contradiction to @xmath452 .
if @xmath146 we can apply the argument of energy loss as in the first case . +
* collision pattern 3a2@xmath124 * + due to ( [ q_222 ] ) and @xmath453 , we have @xmath454 , @xmath455 .
this implies , that letting @xmath104 in ( [ e4.4 ] ) gives the same value as in collision pattern above .
therefore , it assertion follows from collision pattern 3a2 .
+ * collision pattern 3a1 * + from collision pattern 3a2 we have the validity of ( [ d_1111 ] ) and ( [ q_222 ] ) . letting @xmath104 in ( [ e222 ] ) , we obtain @xmath456 finally , we let @xmath104 in @xmath128 ( [ eq101 ] ) . for the third summand in this expression we have @xmath459 if and only if @xmath397 , because from the above , we know , that all the velocities converge to zero for @xmath104 .
now we can apply the argument of energy loss . + * collision pattern 3a1@xmath124 * + due to ( [ q_222 ] ) and @xmath453 , we have @xmath454 , @xmath455 .
this implies , that letting @xmath104 in ( [ e222 ] ) gives the same value as in collision pattern above .
therefore , it assertion follows from collision pattern 3a1 .
+ setting @xmath460 to be the minimum of all the four constants obtained through the argument of energy loss from collision patterns 3a1 , 3a1@xmath124 , 3a2 , 3a2@xmath124 , yields the statement
. we will give a proof by contradiction : assume , that @xmath461 , @xmath462 satisfies @xmath463 . without loss of generality
let @xmath113 happen at time @xmath114 .
+ since @xmath461 , we have @xmath464 and , by our assumptions @xmath462 satisfies @xmath463 @xmath465 & 0 \leq v_2^+(t_6 ) - v_3^+(t_6 ) \leq
c_1.&\nonumber \end{array}\end{aligned}\ ] ] we are first going to express the above inequalities in terms of the starting velocities @xmath118 . using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , ( [ v1p1 ] ) , we calculate @xmath466 & & + ( 1 - \gamma_2)d_2 + v_1^+(0 ) - v_2^+(0)\label{ok1}\\[0.3 cm ] & v_2^+(t_3 ) - v_3^+(t_3 ) & = ( 1+\gamma_1)(v_1^+(t_2 ) - v_2^+(t_2))\nonumber\\[0.2 cm ] & & - d_2 + v_2^+(t_2 ) - v_3^+(t_2)\label{ok2}\\[0.3 cm ] & v_2^+(t_6 ) - v_2^+(t_6 ) & = ( 1 + \gamma_1)\bigl(2\sqrt{(v_{1}^{+}(t_4))^{2 } + 2q_{1}(t_4 ) } \nonumber\\[0.2 cm ] & & + ( ( 1 - \gamma_2)(1+\gamma_1 ) - 1)(v_1^+(t_2 ) - v_2^+(t_2))\nonumber\\[0.2 cm ] & & - ( 1-\gamma_2)d_2\bigr ) - ( v_2^+(t_3 ) - v_3^+(t_3))\label{ok3}\end{aligned}\ ] ] the discriminant under the square roots explicitly calculated gives @xmath467 & -2\gamma_1(v_1^+(t_2 ) - v_2^+(t_2))v_2^+(t_2 ) + ( v_2^+(t_2))^2 , \nonumber\\[0.2 cm ] ( v_{1}^{+}(t_1))^{2 } + 2q_{1}(t_1 ) & = ( v_1^+(0))^2 + 2q_1(0).\nonumber\end{aligned}\ ] ] the travel times @xmath126 and @xmath128 between collisions are explicity given by for @xmath471 ( [ ok5 ] ) to be bounded , it is necessary , that @xmath472 . this implies @xmath454 , @xmath86 . since @xmath473 , it follows from ( [ ok1 ] ) , that @xmath265 .
we see from ( [ ok1 ] ) , that this holds if and only if @xmath474 in the case when @xmath480 , we have @xmath481 , @xmath86 , and the argument of energy loss applies . if @xmath451 , we know from the last limit and ( [ ok6 ] ) , that @xmath482 , when @xmath187 . on the other hand , we have from case 1 . of @xmath265 above , that @xmath483 , when @xmath187 .
we see , that both velocity equalities only hold if and only if @xmath146 .
hence , @xmath481 , @xmath86 , and the argument of energy loss applies .
the following two propositions settle the special case of collision pattern 1c in theorem [ thm4 ] .
we assume that we have a collision pattern 1c , where the ( 2,3 ) collision at point @xmath190 emerges from @xmath221 but the collisions ( 1,2 ) to the left and to the right of it do not .
the idea is to extend collision pattern 1c in the past until the first collision pattern , which has a ( 2,3 ) collision and then apply the argument of energy loss .
the extensions only have to be made for collision patterns 1b - 1d and 2a - 2d , since collision patterns 3a , 3b , 4a and 4b extend to one of them in the future .
the proof is very similar to the one of theorem [ thm3 ] .
[ prop3 ] let after any collision pattern 1c , 1d , 2a - 2d , the next collision pattern incoorporating a ( 2,3 ) collision be 1c .
let @xmath484 , be a point in one of the collision patterns 1c , 1d , 2a - 2d and let @xmath485 mark the ( 2,3 ) collision in the latter collision pattern 1c
. further assume , that @xmath486 and @xmath487 . given an energy value @xmath98 , there exists a @xmath488 , such that for all @xmath489 for all @xmath244 , and collision patterns 1c , 1d , 2a - 2d , we have @xmath490 .
the proof of this proposition is very similar to the proof of theorem [ thm3 ] , therefore we will only sketch it and omit the redudant repetition of long calculations .
assume , on the contrary , that @xmath244 satisfies @xmath491 .
without loss of generality let @xmath113 happen at time @xmath114 . between collision patterns 1c , 1d , 2a - 2d , we experience @xmath183 , @xmath252 , @xmath277 .
calculating @xmath492 back gives @xmath493 & = [ ( 1 - \gamma_2)(1 + \gamma_1 ) - 1](v_2^+(t_{(2,3 ) - 2}(t_rx ) ) - v_3^+(t_{(2,3 ) - 2}(t_rx)))\\[0.2 cm ] & + ( 1 - \gamma_2)(1 + \gamma_1 ) \sum_{i = 2}^{k-1 } ( v_1^+(t_{(2,3 ) - 2i}(t_rx ) ) - v_2^+(t_{(2,3 ) - 2i}(t_rx))\\[0.2 cm ] & + ( 1 - \gamma_2)(1 + \gamma_1 ) ( v_1^+(t_j ) - v_2^+(t_j))\\[0.2 cm ] & + ( 1 - \gamma_2)(v_2^+(t_j ) - v_3^+(t_j ) ) \leq c_1,\ j = 2,3,4,5 , \end{array}\end{aligned}\ ] ] with @xmath494 , where @xmath495 changes according to which collision pattern ( 1c , 1d , 2a - 2d ) we have at the beginning ( see fig .
[ fig:2a-1c ] ) . for collision patterns 1d , 2a - 2d
, we assume , that every ( 2,3 ) collision in it does not emerge from @xmath498 , otherwise theorem [ thm4 ] applies to these collision patterns .
it follows from an easy calculation using the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , ( [ v1p1 ] ) , that @xmath499 this is a sum with non - negative summands , thus , every summand must converge to zero as @xmath104 .
the argument of energy loss now follows , as in theorem [ thm3 ] , by applying theorem [ thm1 ] , [ thm2 ] and corollary [ cor2 ] . for collision pattern 1c , note
, that the assumption @xmath501 , is possible , and theorem [ thm4 ] would not apply , if we additionally assume , that @xmath502
. however , by using again the integrated hamiltonian equations ( [ equation2 ] ) and the law of elastic collisions ( [ collisionqp ] ) , ( [ v1p1 ] ) , the latter assumption still gives @xmath503 [ prop4 ] let after any collision pattern 1b , 1d the next collision pattern incoorporating a ( 2,3 ) collision be 1c .
let @xmath504 , be a point in one of the collision patterns 1b , 1d and let @xmath485 mark the ( 2,3 ) collision in the latter collision pattern 1c
. further assume , that @xmath486 and @xmath487 . given an energy value @xmath98
, there exists a @xmath505 , such that for all @xmath506 for all @xmath507 , and collision patterns 1b , 1d , we have @xmath289 . | the ergodicity hypothesis for the system of three ( or more ) falling balls , with upward decreasing masses , is known for almost three decades . for three falling balls , the model has non - zero lyapunov exponents for every point in phase space .
the usual approach for proving ergodicity for such systems is to verify the five conditions of the local ergodic theorem and then use a transitivity argument to obtain the existence of only one ergodic component . in the case of three ( or more ) falling balls
only one of those conditions has been proven . in this paper , we prove two out of the remaining four conditions for three falling balls with upward strictly decreasing masses .
these conditions are the chernov - sinai ansatz and the non - contraction property .
the former will follow from proving , that every orbit is _ strictly unbounded _ , a term originating from the liverani - wojtkowski version of the local ergodic theorem , that requires the rate of expansion for every vector in the contracting cone field to be infinite . |
quasars can be highly variable .
since this variability arises in the quasar continuum source itself , it appears in all the images of a gravitationally lensed quasar , but at different times in each image due to differences in the gravitational potential of the lens galaxy and the geometric length of the paths of each image @xcite . as first pointed out by @xcite
, the stars near each lensed image also cause temporal variations in the magnification of each image , leading to additional variability .
this `` microlensing '' variability is not correlated between the images , so it functions as both a difficult source of noise in time delay measurements and as a probe of quasar structure because the size of the emission region ultimately controls the amplitude of the microlensing variability .
traditional time delay measurements require either that the microlensing variability is of sufficiently small amplitude so as to be insignificant in the analysis or that its timescale is significantly longer than that of the quasar s intrinsic variability , allowing for some sort of parametric modeling and subtraction ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
attempting to measure time delays in systems with significant short - timescale microlensing is both difficult and error prone ( see * ? ? ?
* for quantitative characterizations of the influence of microlensing on time delay estimates ) . in @xcite
, we applied a new approach to analyzing the light curves of two lensed quasars with significant microlensing and time variability , in which we attempted to simultaneously model the microlensing variability using the bayesian monte carlo technique of @xcite . in our new approach
, we marginalized over the physically realistic models of the microlensing signal to yield a measurement of the most likely time delay .
we successfully confirmed the time delay of he 11041805 but failed to measure the delay in q j01584325 . in the present paper ,
we supplement our previously published light curves with 4 new seasons of monitoring data from two observatories and make a second attempt . in the models of @xcite , we could also marginalize over the time delay uncertainties and use the microlensing to study the structure of the quasar .
the amplitudes and the timescales of the variability due to quasar microlensing are a function of a number of physical variables such as the continuum source size , the mean properties of the lens in the vicinity of the lensed images , and the relative velocities of the source , the lens and the observer . during the last decade
a variety of techniques have been developed to analyze the effects of microlensing on lensed quasar images .
single - epoch photometry in which one or more of the quasar image fluxes are anomalous ( i.e. deviates significantly from the predictions of the macroscopic lens model or shows significant wavelength dependencies ) can provide ( prior dependent ) estimates of source sizes and constrain temperature profiles ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these studies must also assume a mean mass for the microlensing stars .
stronger and less prior - dependent constraints can be obtained from analyses of quasar microlensing variability .
we have employed the technique of @xcite to analyze uncorrelated variability in quasar optical lightcurves to yield a correlation between the sizes of quasar accretion disks and the masses of their black holes @xcite , measurements of the structure of quasar accretion disks ( e.g. * ? ? ?
* ) , and even a constraint on the orientation of a quasar s accretion disk ( in q2237 + 0305 , * ? ? ?
we have also modified the @xcite technique to simultaneously analyze optical and x - ray lightcurves , yielding measurements of the size of the x - ray emitting regions in four systems , pg1115 + 080 , he 11041805 ,
rxj 11311231 and he 04351223 ( * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* respectively ) . in those papers , we found that the x - ray continuum emission emerges from region @xmath4 in radius , where @xmath5 is the black hole s gravitational radius @xmath6 .
this is at least an order of magnitude smaller than sizes measured in the uv / optical , consistent with the arguments of @xcite .
these results seem to disfavor x - ray continuum emission from an extended disk - corona model ( e.g. * ? ? ?
* ; * ? ? ?
* ) , but more measurements are needed both to confirm these results and to begin searching for correlations between the x - ray emitting regions and other physical properties of the quasars . in this paper , we add to the sample of quasars studied at x - ray wavelengths by applying our monte carlo microlensing analysis technique to the combined optical and x - ray lightcurves of qj01584325 to yield size estimates for the hard and soft x - ray continuum emission regions . making use of our new constraints on the time delay , we are also able to significantly refine our estimate of the optical accretion disk size . in [ sec :
obs ] we begin by describing our _ chandra _ x - ray observations @xcite , followed by a presentation of our new optical monitoring data and a brief review of our data reduction process . in [ sec : td ] , we describe our time delay measurement attempts and in [ sec : results ] we present the results from our monte carlo microlensing simulations . in
[ sec : conclusions ] , we conclude with a discussion of the implications for the physics of the optical and x - ray continuum sources in this quasar . all calculations in this paper assume a flat @xmath7cdm cosmology with @xmath8 , @xmath9 and @xmath10 .
we imaged the doubly lensed quasar q j01584325 @xcite in x - rays using the acis imaging spectrometer @xcite on the _ chandra x - ray observatory _
chandra _ , * ? ? ?
* ) on six occasions between 2009
november 6 and 2010 october 6 .
these observations were a component of a larger _ chandra _ cycle 11 monitoring program , the details of which are published in @xcite .
we divided the full observed frame x - ray spectral range ( 0.48.0 kev ) into soft ( 0.41.3 kev ) and hard ( 1.38.0 kev ) energy bands .
we analyze the soft and hard x - ray variability separately in order to constrain the energy structure of the x - ray continuum source .
ccccc[h ] @xmath11 & @xmath12 & @xmath13 & @xmath14 & euler + @xmath15 & @xmath16 & @xmath17 & @xmath18 & smarts + @xmath19 & @xmath20 & @xmath21 & @xmath22 & euler + @xmath23 & @xmath24 & @xmath25 & @xmath26 & smarts + @xmath27 & @xmath28 & @xmath29 & @xmath30 & euler + @xmath31 & @xmath32 & @xmath33 & @xmath34 & euler + @xmath35 & @xmath36 & @xmath37 & @xmath38 & smarts + @xmath39 & @xmath40 & @xmath41 & @xmath42 & euler + @xmath43 & @xmath44 & @xmath45 & @xmath46 & smarts + [ tab : lightcurves ] we have monitored q j01584325 for eight seasons in the @xmath2-band using the smarts 1.3 m telescope with the andicam optical / infrared camera @xcite and using the 1.2 m euler swiss telescope as a part of the cosmograil project .
we analyzed the optical images from both observatories simultaneously , employing the deconvolution algorithm of @xcite . in this technique
, we measure the point spread function ( psf ) in each frame using the same 5 reference stars in the field . the astrometry of the lens components is held fixed to the values measured in hubble space telescope @xmath47-band images from the cfa - arizona space telescope lens survey ( castles ) , the details of which were published in @xcite . using the corresponding psfs , we simultaneously fit the photometry of the point sources and a regularized image of the lensing galaxy to all the frames .
the resulting qso image fluxes are calibrated relative to the deconvolved field stars .
finally , we apply a small multiplicative and additive correction to minimize the dispersion in the overlapping sections of the euler and smarts lightcurves .
this empirical adjustment is required to compensate for small color terms introduced by differences in the filters and quantum efficiency curves of the ccds .
the new photometric data are presented in table [ tab : lightcurves ] , and our composite eight - season light curves are displayed in figure [ fig : lightcurves ] .
we began our analysis by shifting the image a light curve by a reasonable set of trial time delays ( @xmath48 days @xmath49 days ) separated by one day intervals , where a leads b for @xmath50 .
the model prediction for a singular isothermal sphere ( sis ) lens model is that a leads b by @xmath51 days .
we chose to shift image a because it is less variable than image b. our analysis algorithm requires contemporaneous pairs of a / b data points , in most cases necessitating linear interpolation between bracketing epochs . for shifted points that fell into the inter - season gaps , we permitted 5 days of linear extrapolation , and we increased the size of the measurement uncertainties on the extrapolated points using our estimate of the quasar s intrinsic variability structure function . since the bayesian integral shown in eq .
[ eq : bayes ] is actually a discrete sum in which the statistical weight of each trial time delay must be held fixed , all trial lightcurves must have the same number of epochs , forcing us to discard some data points for the shorter trial delays .
the details of this selection technique are described in @xcite .
we also generated a second set of trial light curves , permitting unlimited extrapolation in the inter - season gaps using the damped random walk ( drw ) models of @xcite and @xcite . using the implementation of @xcite
, we first modeled the raw lightcurves , yielding a long term variability structure function measurement @xmath52 mags and timescale @xmath53 days for image a and @xmath54 mags and timescale @xmath55 days for image b. these models also provide the structure function estimate used in our first approach .
the obvious advantage of this treatment is that no data must be discarded , however the drw process models the intrinsic variability in the light curves of only one image at a time such that the variability in the drw points is not correlated with the contemporary points in the other image . in the absence of the strong microlensing signal , we would model the light curves jointly , following the procedure of @xcite .
the introduction of additional uncorrelated variability can be significant , particularly for the longer trial delays , forcing the models to fit the differences in the drw interpolations as microlensing .
this further blurred any effects of the time delay , so we do not report on these results . because we have only 6 epochs of x - ray observations , we are unable to shift the x - ray measurements by the trial time delay , since doing so would eliminate most of our data points .
both our macroscopic mass models and this present work predict a relatively short time delay @xmath56 days , but it is likely that some intrinsic variability also contaminates the x - ray flux ratios . to account for this ,
we estimated the variability structure function of the ( sparsely sampled ) x - ray light curves and then conservatively assumed a 30 day time delay , yielding an additional systematic uncertainty of 0.1 magnitudes in the x - ray flux ratios . this additional systematic uncertainty limited the precision of our monte carlo analysis , but it can not be avoided until the time delay is measured . using the same technique that we described in detail in @xcite , we applied the monte carlo microlensing analysis method of @xcite to the light curves from each trial time delay , in essence attempting to reproduce the observed lightcurves by random combinations of the unknown variables : the lens galaxy stellar mass fraction , the median microlens mass @xmath57 and the effective velocity between source , lens and observer @xmath58 .
we used the recently measured lens redshift @xmath59 @xcite to update our set of ten lens models , each consisting of concentric nfw @xcite and de vaucouleurs mass profiles .
we parametrize the stellar mass fraction @xmath60 in our model sequence by comparing the fractional mass of the de vaucouleurs component to the mass of a constant m / l model ( @xmath61 ) , varying @xmath62 between 0.1 and 1.0 in uniform steps . for each model , we generate four unique sets of microlensing magnification patterns using the @xmath63 method described in @xcite .
the magnification patterns are @xmath64 pixel arrays with an outer scale radius of @xmath65 , where @xmath66 is the einstein radius , yielding a source plane pixel scale of @xmath67 cm .
we performed four separate monte carlo simulations , one per set of magnification patterns . in each simulation , we attempted @xmath68 trials per mass model , and in each trial we attempt to fit all 61 trial time delays , for a grand total of @xmath69 trials .
we discard all solutions in which @xmath70 , since they contribute negligibly to the bayesian integrals and consume unnecessary disk space .
an example of a set of good fits to the observed flux ratios is shown in figure [ fig : fits ] .
we use bayes theorem to marginalize over the results from our monte carlo simulation such that the probability of a given time delay @xmath71 @xmath72 where @xmath73 is the probability of fitting the observations in a given trial , @xmath74 sets the priors on the microlensing variables ( see * ? ? ?
* ; * ? ? ?
* ) and @xmath75 is the ( uniform ) prior on the time delay .
the probability density for the time delay is displayed in figure [ fig : td ] . to first order
, our analysis appears to have failed , since it did not converge on a robust , single - valued estimate for the time delay .
we did succeed , however , in effectively eliminating the possibility of positive values for the delay @xmath76 , which is an improvement over our earlier results in @xcite . nonetheless , the present probability density for the delay is ( at least ) double - valued .
we also attempted a traditional polynomial fit time delay analysis ( e.g. * ? ? ?
* ) , in which we fit the quasar s intrinsic and microlensing variability using legendre polynomials .
the best - fit results from this attempt , based on a 30@xmath77 order fit to the intrinsic and 3@xmath78 order fit to the microlensing variability , are also displayed in figure [ fig : td ] .
again , as was the case with the monte carlo technique , our polynomial fits did not converge on a single value for the delay , in fact the ( inferior ) polynomial fitting technique still allows for some significant probability of a positive ( b leading a ) time delay .
nevertheless , the monte carlo procedure allowed us to study the structure of the quasar with proper statistical weighting of the uncertainties created by the time delay , ultimately enabling a successful joint fit to the observed x - ray and optical lightcurves using our monte carlo method . for the simultaneous optical and x - ray analysis , we employed the technique detailed in @xcite , in which a large number of good fits to the optical light curves are generated first . to enable better resolution at smaller angular scales
, we used a set of @xmath79 magnification patterns with an outer scale of @xmath80 , providing a source plane pixel scale of @xmath81 cm , @xmath82 times larger than the black hole s gravitational radius at @xmath83 cm .
we discuss the limitations of this pixel scale in
[ sec : results ] .
we also tested the monte carlo simulation using the original @xmath84 magnification patterns to confirm that the @xmath85 patterns from the low @xmath62 mass models did not introduce any significant systematic errors .
we saved all physical variables from the good optical fits ( e.g. trajectory , effective transverse velocity @xmath86 ) and then fit the x - ray light curves using the same path across the magnification pattern but a different range of source sizes .
an example of a good fit from this analysis is also displayed in fig .
[ fig : fits ] .
we then analyze the combined x - ray and optical solutions to yield a set of joint probability densities for the variables of interest , namely the x - ray and optical source sizes .
since the time delay was unknown , we performed this analysis for a range of assumed delays , but the x - ray size was insensitive to the resulting minor changes in the optical solutions .
we also tested our optical solutions for the possibility of @xmath87 line emission contamination from the quasar s un - microlensed broad line region , but this led to only modest changes in the size estimates ( see * ? ? ?
* ; * ? ? ?
similar to our analysis in @xcite , the unknown time delay introduces short - timescale noise in the optical difference lightcurve that the monte carlo code can successfully reproduce using non - physical transverse velocities . to eliminate this non - physicality in our models , we applied a prior on the effective transverse velocity @xmath88 where @xmath89 is in einstein units . to convert all variables from einstein units of @xmath90 into physical units , we combined the resulting probability density for the effective transverse velocity @xmath91 with the distribution for the true transverse velocity @xmath92 .
we generated the physical model for @xmath92 following the method of @xcite , where we estimated the peculiar velocities of the source and lens using the power - law fits of .
we found the transverse velocity of the observer by projection onto to the cmb dipole @xmath93 , and we used @xmath94 and @xmath95 for the peculiar velocities of lens and source , respectively .
we estimated the velocity dispersion in the lens galaxy @xmath96 using the monopole moment of a singular isothermal ellipsoid ( sie ) lens model .
in figure [ fig : rs ] , we display the probability densities for the size of the continuum emission region at optical and x - ray wavelengths . in our microlensing simulations , we assume a simple thin accretion disk model where the disk radiates locally as a blackbody ( e.g. * ? ? ? * ) .
since we expect the optical accretion disk to be many times larger than the black hole s gravitational radius @xmath97 , we assume that we can safely ignore the inner edge of the accretion disk and the associated relativistic corrections ( see * ? ? ?
* ) . with these assumptions , the surface brightness at rest wavelength @xmath98 is @xmath99^{-1 }
\label{eqn : disk1}\ ] ] where the scale length @xmath100^{1/3 } \\ & = & 9.7 \times 10^{15 } \left ( { \lambda_{rest } \over { \rm \micron } } \right)^{4/3 } \left ( { m_{bh } \over 10 ^ 9 { \rm m_{\sun } } } \right)^{2/3 } \left({l \over \eta
l_e } \right)^{1/3 } { \rm cm } \nonumber \label{eqn : thindisk}\end{aligned}\ ] ] is the radius at which the disk temperature matches the wavelength , @xmath101 , @xmath102 is the planck constant , @xmath103 is the boltzmann constant , @xmath104 is the black hole mass , @xmath105 is the mass accretion rate , @xmath106 is the luminosity in units of the eddington luminosity , and @xmath107 is the accretion efficiency . in previous investigations ( e.g. * ? ? ?
* ; * ? ? ?
* ) we found accretion disk temperature profiles consistent with the @xmath108 predictions of thin disk theory , so we report the size of the optical continuum emission region in terms of this model .
microlensing observations can estimate the temperature slope , since for @xmath109 the disk size should scale as @xmath110 for standard thin disk theory .
although there is general agreement that hotter regions ( i.e. shorter wavelengths ) are more compact , observational estimates of the slope @xmath111 have significant uncertainties and dispersion between studies ( e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
fortunately , @xcite determined that microlensing statistics are insensitive to the choice of model surface brightness profile and depend primarily on the half - light radius .
thus , the size we report here can easily be compared to other estimates using the half - light radius @xmath112 .
also , since microlensing statistics are actually sensitive to the projected area of the quasar s accretion disk , we must account for the unknown inclination angle @xmath113 .
we assume that the optical emission comes from a geometrically thin disk so that the de - projected estimate of the size is @xmath114 for random orientations . for simplicity
, we use the same surface brightness distribution in x - rays but assume that the emission is geometrically thick , so we report the half - light radius @xmath115 . in @xcite we demonstrated that in the context of the thin accretion disk model , a quasar s specific luminosity @xmath116 can also be used to estimate the size of its accretion disk at that rest - frame wavelength . in the observed - frame @xmath117-band , for example
, the so - called `` flux - based '' thin disk scale radius is @xmath118 there is now significant evidence in the literature that flux(luminosity)-based size estimates are consistently smaller than microlensing - based measurements ( e.g. * ? ? ?
? * ; * ? ? ?
* ) , and in @xcite , we found this to be true in 10 out of the 11 quasars in our sample .
the one outlier in that sample was q j01584325 , whose flux- and microlensing - based sizes were consistent .
integration of our updated @xmath119 probability density and correction for an average inclination of @xmath120 yields an optical accretion disk size estimate of @xmath121 = 15.6 \pm 0.3 $ ] at @xmath122 , the rest - frame center of the @xmath2-band .
so in our new , more precise measurement , the flux - based size ( @xmath123 ) is now @xmath124 dex smaller than the microlensing - based size , consistent with the discrepancies in the other systems .
probability densities for the half - light radii of the full , hard and soft x - ray continuum emission regions are also displayed in figure [ fig : rs ] . since there is no simple spatial model for the x - ray continuum source , we report the x - ray sizes in terms of the half - light radius @xmath115 for x - ray emission at rest - frame energies 0.9 - 18.4 kev ( full band ) , @xmath125 kev ( soft band ) and 3.0 - 18.4 kev ( hard band ) .
integration of the probability density for the combined hard and soft x - ray flux yields an estimate for the full band half light radius @xmath126 . the soft and hard size estimates are consistent with that of the full band and each other , where the half light radius for soft x - ray emission is @xmath0 and the upper limit on the half - light radius for hard x - ray emission is @xmath1 . because the pixel scale in our magnification patterns is @xmath127 cm for a 1 @xmath128 star , the lower limit on our x - ray size estimates is not fully resolved .
in essence , once a trial source becomes much smaller than the pixel scale , all solutions become equally likely . in the case of the hard x - ray source , this prevented any convergence on a lower limit , so we also compiled a probability density for the ratio between the soft and hard x - ray half - light radii @xmath129 .
this probability density is plotted in figure [ fig : sizeratios ] along with the ratio of the optical half - light radius to that in the full x - ray band , which peaks at @xmath130 . in other systems , @xcite and @xcite
have all found x - ray continuum regions that seem to be concentrated in the vicinity of the black hole s last stable orbit . in q
j01584325 , the black hole mass @xmath131 @xcite implies a gravitational radius @xmath132 and last stable orbit @xmath133 at @xmath134 in the schwarzschild metric .
consistent with our results in he 04351223 , rxj 11311231 and pg1115 + 080 , we find that the majority of the x - ray emission in q j01584325 emerges from a region in the vicinity of the last stable orbit
. we also find that the soft and hard x - ray continuum emission regions seem to be the same size , consistent with the findings of @xcite for he 04351223 . in figure
[ fig : fml ] , we display the results of marginalizing the fits from our monte carlo simulation over all variables other than the stellar mass fraction in the macroscopic mass model .
the resulting constraint on @xmath62 favors a dark - matter dominated halo , which seems to be marginally consistent with our expectations for a time delay of approximately @xmath135 days .
to demonstrate the improvement in accuracy gained by adding four seasons of new optical monitoring data and simultaneous analysis with x - ray lightcurves , we also show the size distribution from our first analysis of this system @xcite in fig .
[ fig : rs ] .
while our new measurement is statistically consistent with our previous estimate , the new expectation value is nearly 0.6 dex larger than our original analysis and has nearly twice the precision .
we attribute this difference to two effects .
first , while we failed to measure the time delay , the new data substantially reduced the allowed range to @xmath48 days @xmath136 days , leading to a significant narrowing of the source size probability density .
flux ratios estimated for an incorrect delay have extra power on short time scales which the model tries to fit using a more compact source size , leading to a broadening of the source size distribution toward smaller sizes . since the time delay signal with our four new seasons of monitoring data has effectively ruled out positive delays , there are fewer cases in which the monte carlo code successfully reproduced the noise from an incorrect delay using a small accretion disk size .
second , the simultaneous analysis of the ( admittedly ) sparse x - ray light curves provides a very strong constraint on the effective velocity distribution @xmath86 , which , in turn , constrains @xmath137 .
@xcite gained similar statistical leverage from the use of x - ray photometry in their analysis .
we have made significant improvements in our understanding of the qj01584325 continuum source , but we are still unable to measure the time delay
. this may be due to inadequacies in our data or techniques , such as uncertainties in our photometric measurements and/or insufficient signal from the time delay compared to the signal from microlensing , or it could possibly be due to our oversimplification of the accretion disk model .
for example , @xcite demonstrate that general relativistic effects become important during caustic crossing events and can explain high - magnification features in the light curves of sbs 1520 + 530 and q 2237 + 0305 .
several such features are present in the light curves of q j01584325 and are not perfectly fit by our present analysis . to be fair , the results of @xcite require disks with large inclination angles , and we see no evidence for this in q j01584325 .
nevertheless , we are left to conclude that the modeling of high - amplitude microlensing events may be improved by the inclusion of a fully relativistic disk model or a more sophisticated model for the motions of stars in the lens galaxy ( e.g. * ? ? ?
* ) . at x - ray wavelengths , a consistent picture of the structure of the quasar continuum emission region is beginning to emerge .
in the four systems with quantitatively measured continuum emission regions , he 04351223 @xcite , rxj 11311231 @xcite , pg 1115 + 080 @xcite and now qj 01584325 , the x - ray continuum emission source is located at @xmath138 . improving on our previous work
, we performed the second complete analysis of the hard and soft x - ray variability , leading to independent estimates for the sizes of the soft and hard x - ray sources . while our hard and soft size estimates are formally consistent with each other , there are some indications that the hard x - ray source is smaller than the soft source .
as we discussed in @xcite , these data were not designed to produce good statistical estimates for the hard and soft band separately , so we expect that adding more x - ray epochs with better photon statistics should lead to markedly improved x - ray size estimates .
this material is based upon work supported by the national science foundation under grant no . ast-0907848 . c.w.m .
also gratefully acknowledges support from the research corporation for science advancement .
, a.m.m . , x.d . and g.c .
are supported by chandra x - ray center award 11700501 .
, f.c . and g.m .
acknowledge support from the swiss national science foundation .
c.s.k . , j.a.b . and a.m.m .
are supported by nsf grant ast-1009756 .
kochanek , c.s . ,
dai , x. , morgan . c.w .
, morgan , n.d . , poindexter , s.a . , chartas , g. 2007 , in asp conf .
ser . 371 , statistical challenges in astronomy iv , ed .
babu & e.d .
feigelson , ( san francisco , ca : asp ) , 43 | we present four new seasons of optical monitoring data and six epochs of x - ray photometry for the doubly - imaged lensed quasar q j01584325 . the high - amplitude , short - period microlensing variability for which
this system is known has historically precluded a time delay measurement by conventional methods .
we attempt to circumvent this limitation by application of a monte carlo microlensing analysis technique , but we are only able to prove that the delay must have the expected sign ( image a leads image b ) . despite our failure to robustly measure the time delay , we successfully model the microlensing at optical and x - ray wavelengths to find a half light radius for soft x - ray emission @xmath0 , an upper limit on the half - light radius for hard x - ray emission @xmath1 and a refined estimate of the inclination - corrected scale radius of the optical @xmath2-band ( rest frame 3100 ) continuum emission region of @xmath3 . |
the most general linear ( bi - anisotropic ) electromagnetic medium can be represented in terms of 36 scalar medium parameters in different representations . in engineering form applying gibbsian vector fields and 3d dyadics it is typical to apply the form @xcite # d + # b= = & = + = & = .#e + # h. dbanother form favored by the physicists is @xcite # d + # h= = & =
+ = ^-1 & = .#b + # e. dha more compact form is obtained by applying the 4d formalism in terms of differential forms @xcite where the electromagnetic fields are characterized by two six - dimensional quantities , the two - forms @xmath0 and @xmath1 , % = # d - # h_4 , % = # b + # e_4 . in a linear medium
they are related by the medium bidyadic @xmath2 as % = m|% .
the medium bidyadic can be expanded in terms of 3d medium dyadics as m = = + = #e_4 + _ 4=^-1 + _ 4=#e_4 . for details in the present notation @xcite or @xcite should be consulted .
it is not easy to get a grasp of the most general medium defined by the bidyadic @xmath2 .
this is why many classes of media with @xmath2 restricted by special forms involving less than 36 parameters , and bearing strange names , have been defined and studied . because @xmath2 corresponds to a @xmath3 matrix ,
some classes have been based on expressing the bidyadic @xmath2 in terms of a dyadic corresponding to a @xmath4 matrix involving only 16 parameters .
as the most obvious one , the class of p media @xcite has been defined in terms of a dyadic @xmath5 as the double - wedge square m= p^(2)t = p^tp^t . as another obvious example , the class of q media @xcite has been defined through the modified medium bidyadic m_m = # e_nm = # e_nq^(2)_2_2 , in terms of a ( quasi- ) metric dyadic @xmath6 .
properties of both medium classes have been recently studied . for a plane wave , a field of the dependence @xmath7 of the spacetime variable @xmath8 ,
the wave one - form @xmath9 is normally restricted by a dispersion equation d(%)=0 , dnuwhich is a quartic equation in general @xcite and its form depends on the medium bidyadic @xmath2 . for example , for the q medium
the dispersion equation is of the form d(%)= _ q(q||%%)^2=0 , dqwhich means that , for a dyadic of full rank satisfying @xmath10 , the quartic equation reduces to a quadratic equation . on the other hand
, one can show that , for a p medium , the dispersion equation is actually an identity which is satisfied by any one - form @xmath9 @xcite .
also , if @xmath11 is of rank lower than 4 , we have @xmath12 and is satisfied identically . media of this kind have previously been called nde media ( media with no dispersion equation ) @xcite .
as simple generalizations of the classes of p and q media we can define m = p^(2)t + _ n#d#c , pextand m_m = q^(2 ) + # a#b , qextwhich have respectively been called extended ( or generalized ) p @xcite and q @xcite media . here ,
@xmath13 and @xmath14 are any bivectors .
the number of medium parameters in both cases has been increased from 16 to 23 .
plane waves propagating in extended p and q media have been previously studied and the dispersion equations have been shown to take the respective form @xcite d(% ) & = & _ p((#dp^t)||%%)((#cp^-1t)||%%)=0 , dpext + d(% ) & = & _ q(q||%%)(q + # a#bq^-1t)||%%=0 .
dqexthere we denote @xmath15 . in both cases , for full - rank dyadics @xmath16
, the dispersion equations can be decomposed in two quadratic equations .
such media make two examples of what have been called decomposable media @xcite . for dyadics @xmath17 and @xmath11 of rank less than 3 ,
the two media again fall in the class of nde media with dispersion equations satisfied identically for any @xmath9 @xcite . for ranks equaling 3
, some quadratic functions of @xmath18 may decompose to products of linear functions of @xmath9 .
in the present study we make a further generalization of both p and q media by considering medium bidyadics of the composite form m = p^(2)t + _ nq^(2 ) .
pqmedia defined by bidyadics of this form will be called pq media .
it is easy to see that medium dyadics of this form do not cover all possible linear media .
in fact , the number of parameters must certainly be less than @xmath19 , which falls short of the 36 parameters of the most general medium . to find a 3d representation of the pq medium
bidyadic the dyadics @xmath17 and @xmath11 can be expanded in terms of spatial vectors @xmath20 , a spatial one - form @xmath21 and spatial dyadics @xmath22 as q & = & q_s + # e_4#a_s + # b_s#e_4 + c#e_4#e_4 , + p & = & p_s + # e_4%_s + # p_s_4 + p#e_4_4 .
the spatial medium dyadics of the q medium can be expressed as @xcite = & = & _ 123q_s#a_s , aq + = & = & _ 123(cq_s -#b_s#a_s ) , + = ^-1 & = & -_123q_s^(2 ) , mq + = & = & _ 123(#b_sq_s ) , bq while those of the p medium have the form @xcite = & = & p_s^(2)t , ap + = & = & -%_sp_s^t , + = ^-1&= & -p_s^t#p_s , mp + = & = & % _
s#p_s -pp_s^t .
pbthe medium dyadics of the pq medium are obtained by summing up the corresponding expressions . to write the medium equations in the gibbsian form
requires that the dyadic @xmath23 exist . denoting the gibbsian dyadics by the subscript @xmath24
, they are obtained by inserting the previous expansions in the expressions @xcite = _ g & = & # e_123(=-=|=|= ) , + = _ g & = & # e_123=|= , + = _ g & = & -#e_123=|= , + = _ g & = & # e_123=. the dyadic denoted here by @xmath23 stands for the inverse of the sum of the two @xmath25 dyadics of and .
the full analytic form of the gibbsian medium dyadics of the pq medium would have quite an extensive form .
the main task of this study is to find properties of a plane wave propagating in the general pq medium .
expressing the field two - form of a plane wave in terms of a potential one - form as @xmath26 , the potential satisfies % % & = & % m|(%% ) = 0 + & = & % ( p^(2)t + _ nq^(2))|(%% ) + & = & % ( p^t|%)(p^t|% ) + _ n(q^(2)%%)|% . operating this by @xmath27 yields the equation # e_n(%% ) & = & # e_n(%(%|p)p^t)|%+ ( q^(2)%%)|% + & = & ( # fp^t + q^(2)%%)|%=0 .
eqhere we have introduced the bivector # f = # f(%)= # e_n(%(%|p ) ) , which is simple since it satisfies @xcite # f#f=0 , ( # f^t)^(2)=#f#f , ( # f^t)^(3)=0 , and it can be expressed in the form @xmath28 in terms of two vectors .
the equation for the potential , d(%)|%=0 , dnuphiis defined by the dispersion dyadic d(% ) = # fp^t + q^(2)%% . because of and @xmath29 , the rank of @xmath30 must be less than three , provided @xmath31 and @xmath9 are linearly independent , i.e. , when @xmath32 , which is assumed here . as a consequence ,
@xmath9 is restricted by the dyadic dispersion equation @xcite d^(3)(% ) & = & ( # fp^t + q^(2)%%)^(3 ) + & = & ( # f^t)^(3)|p^(3)t + ( ( # f^t)^(2)|p^(2)t)(q^(2)%% ) + & & + ( # fp^t)(q^(2)%%)^(2 ) + ( q^(2)%%)^(3 ) + & = & ( # f#f|p^(2)t)(q^(2)%% ) + ( # fp^t)(q^(2)%%)^(2 ) + ( q^(2)%%)^(3 ) + & = & 0 . disp3applying the expansion rules @xcite q^(2)%%&= & ( q||%%)q - ( q|%)(%|q ) + ( q^(2)%%)^(2 ) & = & ( q||%%)((q||%%)q^(2 ) - q(q|%)(%|q ) ) + & = & ( q||%%)(q^(3)%% ) + ( q^(2)%%)^(3 ) & = & ( q||%%)^2((q||%%)q^(3 ) - q^(2)(q|%)(%|q ) + & = & ( q||%%)^2(q^(4)%%)= _ q(q||%%)^2(#e_n#e_n%% ) , with @xmath33
, the dispersion equation can be written as d^(3)(% ) = c_1(% ) + c_2(% ) + c_3(% ) = 0 , d3nuwhere we denote c_1(% ) & = & ( # f#f|p^(2)t)(q^(2)%% ) + c_2(% ) & = & ( q||%%)(#fp^t)(q^(3)%% ) + c_3(% ) & = & _ q(q||%%)^2(#e_n#e_n%%).now one can show that is equivalent to a scalar dispersion equation . for that we expand the three dyadics @xmath34 as follows . * for the dyadic @xmath35
we apply the identity # a_i(#b_i% ) & = & ( # a_i#b_i)%- # b_i(#a_i% ) + & = & _ n|(#a_i#b_i)(#e_n% ) - # b_i(#a_i% ) , valid for any bivectors @xmath36 and one - form @xmath37 . assuming that @xmath38 and @xmath37 satisfy @xmath39
, we can construct the dyadic rule # a_1#a_2(#b_1#b_2%% ) = _ n_n||(#a_1#a_2#b_1#b_2)(#e_n#e_n%% ) . because of @xmath40 and @xmath41 , we can set @xmath42 , @xmath43 and @xmath44 .
since the rule is linear in the dyadic @xmath45 , we can set @xmath46 and apply the rule as c_1(% ) = ( # f#f|p^(2)t)(q^(2)%% ) = d_1(%)(#e_n#e_n%% ) , with d_1(% ) & = & _ n_n||((#f#f|p^(2)t)(q^(2 ) ) ) = ( _ n_nq^(2))||(#f#f|p^(2)t ) + & = & _ q q^(-2)t||(#f#f|p^(2)t ) = _ q#f#f||(p^t|q^-1)^(2 ) .
d1 in the last expression we have assumed that @xmath11 is of full rank , @xmath47 . * to expand the dyadic @xmath48 we proceed as c_2(% ) & = & ( q||%%)(q^(3)%%)(#fp^t ) + & = & _ q(q||%%)((#e_n#e_n(q^-1t%%))(#fp^t ) + & = & _ q(q||%%)#e_n#e_n((q^-1t%%)(#fp^t ) ) + & = & _ q(q||%%)(#e_n#e_n%%)(q^-1t||(#fp^t ) ) + & = & _ q(q||%%)(#e_n#e_n%%)(q^-1t||(#fp^t ) ) + & = & d_2(%)(#e_n#e_n%% ) , with d_2(% ) = _ q(q||%%)(#fp^t|q^-1 ) . *
finally , we have c_3(% ) = d_3(%)(_n_n%% ) , with d_3(% ) = _ q ( q||%%)^2 . d3 because each of the dyadics @xmath34 is a scalar multiple of @xmath49 , the dyadic dispersion equation equals the scalar dispersion equation as d(% ) & = & d_1(% ) + d_2(% ) + d_3(% ) + & = & _ q#f#f||(p^t|q^-1)^(2 ) + _ q(q||%%)(#fp^t|q^-1 ) + & & + _ q ( q||%%)^2 = 0 . dispsubstituting @xmath50 , the quartic form of the dispersion equation
this is the main result of the present paper . for @xmath51 we must replace _ q q^(-2 ) _
n_nq^(2)t , _
_ n_nq^(3)t , deqin the expression .
let us consider the expression for a few special cases of the pq medium for which we know the dispersion equation . 1 . for the pure p medium case @xmath52 , after inserting we
obtain the identity @xmath53 for all @xmath9 .
this proves that a pure p medium belongs to the class of nde media @xcite .
2 . for the pure q medium with @xmath54
is reduced to the known quadratic dispersion equation .
the case @xmath55 corresponds to a q - axion medium .
since we now have @xmath56 , only the last term of the expression in survives .
this , again , yields the dispersion equation of the q medium .
in fact , it is well known that adding an axion term @xmath57 to the medium bidyadic does not change the dispersion equation . here
one should note that the p - axion medium @xmath58 with @xmath59 is not a special case of the pq medium .
choosing @xmath60 we obtain m = p^(2)t + _
n(#a_1#a_2)(#b_1#b_2 ) , which yields a special case of the extended p medium with @xmath61 . applying for @xmath51 in yields the dispersion equation
whose form can be expanded as d(% ) & = & # f|(p^(2)t|q^(2)t)|#f + & = & # f|p^(2)t|(_n(#b_1#b_2))((#a_1#a_2)_n)|#f + & = & ( % ( % |p))#e_n|p^(2)t|_n(#b_1#b_2))((#a_1#a_2)|(%(%|p ) ) + & = & _ p(%(%|p)|p^(-2)|(#b_1#b_2))((#a_1#a_2)|(%(%|p ) ) + & = & _ p((%|p^-1)%)|(#b_1#b_2))((#a_1#a_2)|(%(%|p ) ) = 0.assuming @xmath62 the equation is split in two quadratic equations as % |(p%)|(#a_1#a_2 ) & = & % |(p(#a_1#a_2))|%=0 , + % |(p^-1%)|(#b_1#b_2 ) & = & % |(p^-1(#b_1#b_2))|%=0 , nup-1which coincide with for @xmath61 . in the case
@xmath63 we must replace @xmath64 by @xmath65 in .
choosing @xmath66 we have @xmath67 and m= _ nq^(2)t + ( % _ 1%_2)(#a_1#a_2 ) , which corresponds to a special case of the extended q medium with @xmath68 .
expanding # f|p^(2)t = # e_n(%(p^t|%))|p^(2)t= # e_n|(%(p^(3)t% ) ) = 0 , the first term of obviously vanishes .
expanding further # fp^t & = & # e_n(%(p^t|%)p^t ) = # e_n(%(p^(2)t% ) ) + & = & -%(#e_np^(2)t)%= ( # e_np^(2)t)%% , we have ( # fp^t|q^-1 ) = ( # e_np^(2)t)%%)||q^-1 t = ( ( # e_np^(2)t)q^-1t)||%% , whence the dispersion equation can be written as d(% ) = _ q(q||%%)(q + ( # e_np^(2)t)q^-1t)||%%=0 , which coincides with .
finally , let us assume that @xmath11 is an antisymmetric dyadic , which can be expressed in terms of some bivector @xmath69 in the form q = # a^t .
this implies @xmath70 , whence the dispersion equation reduces to d(% ) = _ q#f|p^(2)t|q^(-2)|#f=0 .
disp1 applying the expansion @xcite ( # a^t)^(2 ) = # a#a + # e_n^(2)t , = -_n|(#a#a ) , the pq medium bidyadic has the form of an extended p - axion medium bidyadic @xcite m = p^(2)t +
_ n#a#a + ^(2)t . applying the rule @xcite q^(-2 )
= _ n_nq^(2)t = ( _ n_n#a#a + _ n^(2 ) ) , one can show that its last term has no effect on the dispersion equation .
in fact , inserting # f|p^(2)t & = & # e_n|(%(p^t|%)p^(2)t ) = % |(#e_np^(3)t)% + & = & _ p%|(p^-1#e_n)% , in we have # f|p^(2)t|(_n^(2))|#f & = & _ p(%|(p^-1#e_n)%)|(_n(#e_n(%(p^t|% ) ) ) + & = & _ p(%|(p^-1#e_n)|(%%(p^t|%))=0 .
thus , is reduced to d(% ) & = & _ q#f|p^(2)t|q^(-2)|#f + & = & _ p%|(p^-1#e_n)%|(_n_n#a#a)|(#e_n(%(p^t|% ) ) + & = & _ p%|(p^-1#e_n)%|(_n#a)(#a|(%(p^t|% ) ) ) + & = & _ p((%|p^-1%)|#a)(#a|(%(p^t|% ) ) ) + & = & _ p((p^-1#a)||%%)((p#a)||%%)=0 .
since the axion component does not affect the dispersion equation , the result coincides with that of the extended p medium for the special case @xmath71 .
as a numerical example of a pq medium let us consider a special one by restricting the dyadic @xmath17 as p = p( + p_o ) , p_o=#a% , # a|%=0,and assuming that @xmath11 be a symmetric dyadic , q = s , s^t = s.in the corresponding medium bidyadic m = p^2^(2)t + p^2%#a^t + _ ns^(2 ) ms2the three terms can actually be recognized as components of the hehl - obukhov decomposition , respectively called as the axion , skewon and principal components @xcite .
in fact , the axion part is a multiple of the unit bidyadic @xmath72 while the skewon and principal parts are trace free .
also , any skewon bidyadic is known to be of the form @xmath73 with a skewon - free dyadic @xmath74 while any bidyadic @xmath75 satisfying @xmath76 is a principal bidyadic .
actually , one can show that @xmath77 vanishes for any symmetric dyadic @xmath78 , @xcite . to find the dispersion equation for the present pq medium from ,
let us first expand # f|p^(2)t & = & # e_n|(%(p^(3)t% ) ) + & = & p^3#e_n|(%(^(3)t+ % # a^(2)t))% + & = & p^3#e_n(%%^(2)t ) + p^3 # e_n|(%(%(#a|%)^(2)t + & & + % ( % # a)^t ) + & = & p^3(#a|%)#e_n(%% ) , fp2 t + # fp^t&= & # e_n(%(p^(2)t))% ) + & = & p^2#e_n(%(^(2)t+%#a^t)% ) + & = & p^2#e_n(%%^t+%(%(#a|%)^t+ % % # a ) ) + & = & p^2(#a|%)(#e_n(%%))^t .
since the last expression is an antisymmetric dyadic and @xmath79 is a symmetric dyadic or zero , we have _
q(#fp^t|q^-1 ) = _ s(#fp^t)||s^-1=0 .
thus , becomes d(% ) = _
s#f|(p^t|s^-1)^(2)|#f + _ s ( s||%%)^2 . finally , applying , we can expand _ s#f|(p^t|s^-1)^(2)|#f&= & p^4(#a|%)(#e_n(%%))|s^(2)|(#e_n(%(^t+%#a)|% ) ) ) + & = & p^4(#a|%)^2(%%)|s^(2)|(%% ) ) + & = & p^4(#a|%)^2(%%s^(2))||%% , whence the dispersion equation for the special pq medium has the form d(% ) & = & p^4(#a|%)^2(%%s^(2))||%%+_s ( s||%%)^2 + & = & p^4(#a|%)^2(%%||s)(s||%% ) + p^4(#a|%)^2(%|s|%)^2 + & & + _ s ( s||%%)^2=0 .
dispspecunlike for all the special cases considered above , the quartic dispersion equation corresponding to the medium defined by does not necessarily decompose in two quadratic equations . in the special case when the dyadic @xmath78 satisfies @xmath80 , whence we have @xmath81
, the dispersion equation is decomposable . setting @xmath82 in we obtain @xmath53 for any @xmath9 , a property shared by all skewon - axion media .
for @xmath83 we are left with @xmath84 , valid for the q - axion medium . to be able to work on this example numerically ,
let us choose s = sg_s - s#e_4#e_4 , where @xmath85 denotes the metric dyadic g_s = # e_1#e_1 + # e_2#e_2 + # e_3#e_3 .
thus , we have _ s = s^(4 ) = -s^3s .
let us further choose % = _ 2 , # a=#e_3 . inserting these in , the dispersion equation becomes d(% ) & = & p^4_3 ^ 2s(s(_1 ^ 2+_2 ^ 2+_3 ^ 2)-sk^2 ) + p^4_3 ^ 2s^2_2 ^ 2 + & & -s^3s ( s(_1 ^ 2+_2 ^ 2+_3 ^ 2)-sk^2)^2=0 .
dispspec1for a given value of @xmath86 , this corresponds to a quartic dispersion surface in the three - dimensional space spanned by @xmath87 .
let us assume that @xmath88 and @xmath89 are real and positive quantities and simplify the expressions by denoting x_i=_i , k = k , = , with @xmath90 .
the dispersion equation then takes the form ( x_1 ^ 2+x_2 ^ 2+(1-)x_3 ^ 2-k^2)(x_1 ^ 2+x_2 ^ 2+x_3 ^ 2-k^2 ) -x_3 ^ 2x_2 ^ 2 = 0 .
dispspec2 an idea of the dispersion surface can be obtained by considering the three main sections . * assuming @xmath91 , leads to x_1 ^ 2+x_2 ^ 2-k^2=0,which corresponds to a circle of unit radius . * assuming @xmath92 , yields ( x_1 ^ 2+(1-)x_3 ^ 2-k^2)(x_1 ^ 2+x_3 ^ 2-k^2)=0 .
this splits in two separate curves , one of which is a circle of unit radius x_1 ^ 2+x_3 ^ 2-k^2=0 , and , the other one , a quadratic curve x_1 ^ 2+(1-)x_3 ^ 2-k^2=0 . for @xmath93
the latter defines an ellipse with axial ratio @xmath94 . for @xmath95
the curve is a hyperbola . * finally ,
assuming @xmath96 , yields ( x_2 ^ 2+(1-)x_3 ^ 2-k^2)(x_2 ^ 2+x_3 ^ 2-k^2 ) -x_3 ^ 2x_2 ^ 2 = 0 .
dx1this corresponds to a curve of the fourth order , the form of which depends on the parameter @xmath97 . for @xmath93 the curve is closed and for @xmath98 it is open . for @xmath99
the curve approaches a circle of unit radius , in which case the pq medium approaches a q medium .
the cross sections @xmath96 and @xmath92 are depicted for the parameter value @xmath100 in fig .
[ fig : em26a08 ] in terms of normalized axis parameters @xmath101 .
it is seen that , for this particular pq medium , there is no birefringence for waves whose wave one - form satisfies @xmath102 . and @xmath103 . for @xmath104
the cross - section reduces to a circle and an ellipse . for @xmath105
the dispersion surface reduces to a single circle of unit radius . here the parameter value @xmath106 has been assumed . ]
a novel class of electromagnetic media , called that of pq media , was introduced as a generalization of the previously studied classes of p media and q media . plane - wave propagating in the general pq medium
was studied and the quartic dispersion equation was derived in analytic form .
the equation was verified for six special cases of pq media for which the analytic form has been found from previous studies . in all of these special cases the quartic equation either reduces to two quadratic equations or becomes an identity . as an example of a medium yielding a more general quartic dispersion equation , another special case of the pq medium was considered .
discussion with dr .
alberto favaro on the topic of this paper is acknowledged . | two basic classes of electromagnetic media , recently defined and labeled as those of p media and q media , are generalized to define the class of pq media .
plane wave propagation in the general pq medium is studied and the quartic dispersion equation is derived in analytic form applying four - dimensional dyadic formalism .
the result is verified by considering various special cases of pq media for which the dispersion equation is known to decompose to two quadratic equations or be identically satisfied ( media with no dispersion equation ) . as a numerical example , the dispersion surface of a pq medium with non - decomposable dispersion equation is considered . |
beginning with the early considerations of bevir @xcite , pierson @xcite , and steenbeck @xcite , there has always been a tight connection between experimental dynamo studies and research related to liquid metal fast reactors ( lmfr ) , in particular to sodium fast rectors ( sfr ) . for example , a precursor of the riga dynamo experiment @xcite had been carried out in 1987 at a test facility for sfr pumps in leningrad @xcite .
later , alemany et al .
@xcite studied the possibility of self - excitation in the pumps and cores of the french phnix and superphnix reactors . with the dresdyn project
we intend to resume this tradition by setting - up a new research infrastructure for liquid sodium experiments related both to the origin and action of cosmic magnetic fields ( with possible applications for the construction of large - scale liquid metal batteries ) as well as to safety aspects of lmfr s .
first , we present the plans for a precession driven dynamo experiment and a taylor - couette type experiment for the combined investigation of the magnetorotational instability ( mri ) and the tayler instability ( ti ) . the destructive effect of the ti on the layered stratification in envisioned large - scale liquid metal batteries , and a possible provision to avoid this effect , will then be discussed .
the paper closes with delineating further experiments related to in - service - inspection problems for lmfr s .
the most ambitious project within the framework of dresdyn is a large scale precession dynamo experiment .
precession has been discussed since long as an , at least complementary , energy source of the geodynamo @xcite .
this idea is supported by paleomagnetic measurements that have revealed a modulation of the geomagnetic field intensity by the 100 kyr milankovi cycle of the earth s orbit eccentricity and by the corresponding 41 kyr cycle of the earth s axis obliquity @xcite .
the 100 kyr cycle is also known to influence the reversal statistics of the geomagnetic field , an effect that is very probably being enhanced by stochastic resonance @xcite . in this context
there is an interesting correlation of geomagnetic field variations with climate changes , clearly demonstrated for the last 5000 years @xcite , and perhaps also existing for the sequence of ice ages @xcite , although the causal mechanism of how exactly the milankovi cycles affect both the climate and the magnetic field is far from being settled . in an early paper ,
doake had speculated that the increasing ice sheets would change the moment of the inertia of the earth , thereby influencing the geodynamo by a modified rotation period @xcite .
however it could be worthwhile to investigate also the reverse causal chain that changing earth s orbit parameters lead , in the first instance , to a modified geodynamo field which , in turn , could then influence the climate by modifying the shield against cosmic ray flux ( which was recently discussed as a key climate driver through cloud formation @xcite ) .
= apart from this general geophysical motivation , a precession driven dynamo experiment is also interesting from the narrower magnetohydrodynamic point of view . in comparison with the previous experiments in riga @xcite , karlsruhe @xcite and cadarache @xcite
, a precession experiment would represent a _ homogeneous _ dynamo par excellence . containing only a homogeneous fluid rotating around two axes
, it would neither contain any propeller , as in riga , nor any assembly of guiding tubes , as in karlsruhe , nor any soft - iron material ( which is crucial for the low critical magnetic reynolds number and the close to axisymmetric eigenmode in the cadarache experiment @xcite ) .
= the central part of the envisioned precession dynamo experiment ( see fig . 1 for a preliminary draft ) will be a cylindrical vessel of approximately 2 m diameter and length , rotating with up to 10 hz around its symmetry axis , and with up to 1 hz around another axis whose angle to the first axis can be varied between 90@xmath0 and 45@xmath0 .
the inner cylindrical shell ( made of copper ) is immersed into a larger cylindrical stainless container with conical end parts that can later also be used to house alternative inner shells , e.g. in the form of ellipsoids .
the mechanical and safety demands for such a large scale sodium experiment are tremendous . with a total mass of approximately 20 tons ( including the sodium in the conical end parts and the stainless steel walls )
the sodium filled cylinder will have a moment of inertia of around 10@xmath1 kg m@xmath2 . with a rotation rate of 10 hz and a precession rate of 1 hz
, this amounts to a gyroscopic moment of 5x10@xmath3 nm which requires an extremely massive basement and a careful avoidance of resonances .
further , the experiment must be fenced by a containment which can be quickly flooded with argon in case of a sodium accident . in order to determine various crucial parameters for the design of this large - scale sodium experiment ,
we have started a series of experiments at a smaller ( scale 1:6 ) water precession experiment which is shown in fig .
this small water experiment is similar to the ater experiment guided by j. lorat @xcite , but comprises some special measuring devices .
most important for the later sodium experiment is the determination of the torques and motor powers needed to drive the rotation of the cylinder and the turntable , and of the gyroscopic torques acting on the basement .
those measurements have already confirmed the expected sharp transition between a laminar flow and a turbulent flow which in our case occurs at a precession ratio of around 0.07 . at this point , the needed motor power increases sharply , as indicated in fig . 3a . concerning the flow field determination , up to present we have only installed a number of ultrasonic sensors for the determination of the axial velocity component .
for a ( rather low ) rotation rate of 0.2 hz and a precession rate of 0.01 hz , fig .
3b shows first results of the axial velocity measured by ultrasonic doppler velocimetry ( udv ) .
= the second liquid sodium experiment with geo- and astrophysical motivation will be a large - scale taylor - couette - experiment ( fig .
4 ) with a diameter of approximately 1 m and a height of 2.5 m , wrapped by a coil that produces a strong vertical magnetic field , and supplemented by technical means to guide independent electrical currents through a bore in the center and through the liquid sodium between the inner and outer cylinder . in the following this experiment
will be called the mri / ti experiment , indicating its astrophysical background in terms of the magnetorotational instability ( mri ) and the tayler instability ( ti ) .
the mri had been discussed , for the first time , by velikhov in 1959 @xcite , and was then rediscovered and applied to accretion disk physics by balbus and hawley in 1991 @xcite . nowadays , mri is thought to play a crucial role in enabling angular momentum transport by destabilizing the keplerian rotation profiles of accretion disks that would otherwise be hydrodynamically stable . with this large - scale mri
/ ti experiment we plan to extend our previous investigations @xcite on the helical version of mri ( hmri ) to the realm of the standard mri ( smri )
. a particular focus will be on the interesting mode transitions between the hmri and the smri , as discussed recently in @xcite .
in addition to this , the experiment will also allow to study the kink - type tayler instability ( ti ) @xcite and its transitions to the azimuthal mri ( amri ) @xcite .
recent experimental work on ti with the eutectic alloy gainsn has revealed a complicated interplay of the very ti with large scale convection effects due to joule heating @xcite . with the envisioned sodium experiment
we intend to suppress those convection effects significantly . on this basis , we see a much better chance to study the numerically predicted intrinsic saturation effects of ti by means of an increased turbulent resistivity @xcite . =
tightly connected with the research on ti , we will also investigate liquid metal batteries that have recently been proposed as cheap devices for the storage of the highly fluctuating renewable energies .
typically , such a battery would consist of a self - assembling stratification of a heavy liquid half - metal ( e.g. bi , sb ) at the bottom , an appropriate molten salt as electrolyte in the middle , and a light alkaline or earth alkaline metal ( e.g. na , mg ) at the top . the functioning of a small version of this type of battery has already been verified @xcite .
however , for large - scale batteries ( which are the only interesting ones in terms of economic competitiveness ) the occurrence of ti can be easily imagined to represent a serious problem for the integrity of the stratification ( see fig .
5a ) . in our experiments
, we will focus on various ways to avoid ti in such configurations , e.g. by using a return current through the center of the battery ( see fig .
5b ) , as it was proposed in a recent paper @xcite .
the safe and reliable operation of liquid metal systems in innovative reactor concepts like sodium fast reactors ( sfr ) or transmutation systems based on lead - bismuth cooled reactors ( lbfr ) requires appropriate measuring systems and control units , both for the liquid metal single - phase flow as well as for gas bubble liquid metal two - phase flows .
hence , there is a growing need for small and medium sized liquid sodium experiments to study various thermo - hydraulic and safety aspects of sfr s , comprising sodium boiling , argon entrainment , bubble detection , sodium flow metering and many more @xcite .
= a significant portion of the lmfr - related experiments in the framework of dresdyn will be conducted at an in - service - inspection ( isi ) facility the principle sketch of which is shown in fig 6 .
basically , it will consist of a heated stainless steel vessel with a diameter of approximately 1 m that is filled with liquid sodium covered by argon .
the internal components will comprise a simple mock - up of a reactor core and a primary pump .
the main goal of this facility is to test a variety of measurement techniques for the position of internal components , for flow velocities and argon bubble detection .
the latter will also include experiments on argon entrainment on the free surface .
particular attention will be paid to the application of the contactless inductive flow tomography ( cift ) for the flow reconstruction in lmfr s .
this technique has been developed at hzdr during the last decade @xcite , and it was recently deployed for visualizing the flow structure in a physical model of the continuous casting of steel @xcite .
cift could be of particular value for lead - bismuth cooled transmutation systems , such as myrrha @xcite .
the lower cold pool of myrrha is characterized by a rather free velocity field which is very likely prone to flow instabilities that could easily be detected by cift .
dresdyn will also comprise a liquid sodium loop which will replace the presently existing one .
this loop will contain various test sections , among them one section for the study of smart heat exchangers with intermediate heat transfer media . the deployment of such smart heat exchangers in future sfr s could reduce significantly the risk of energetic sodium - water reactions .
another suite of experiments will be devoted to the important problem of sodium boiling which is a key safety issue for sfr s due to their positive reactivity coefficient .
starting with very small experiments at a flat wall , we plan to go over later to boiling experiments at rods or rod bundles .
for the visualization of the boiling we plan to use the mutual inductance tomography @xcite , as well as x - ray radiography @xcite and , in collaboration with another group at hzdr , the ultrafast x - ray tomography @xcite . finally , with the advent of oxide dispersion strengthened ( ods ) steels as new promising nuclear reactor materials ( because of their superior swelling resistance and excellent high temperature strength ) @xcite , we come back to our starting point of considering the possibility of magnetic - field self - excitations in sfr s .
it has been shown that the use of ferritic or martensitic steels in the core of large sfr s could indeed foster magnetic field self - excitation @xcite .
this is consistent with the key role of high magnetic permeability material for the dynamo process in the cadarache experiment @xcite .
hence , it is certainly necessary to reconsider this point before new steel sorts can be utilized in large scale sfr s .
related experiments on this topic are also envisaged in the framework of dresdyn .
in this paper , we have discussed the motivation behind , and the concrete plans for a number of experiments to be set - up in the framework of dresdyn .
the new building and the essential parts of the experiments are expected to be ready in 2015 .
apart from hosting the discussed experiments , dresdyn is also meant as a general platform for further large - scale experiments , basically but not exclusively with liquid sodium .
proposals for such experiments are , therefore , highly welcome .
this research was supported by deutsche forschungsgemeinschaft ( dfg ) under grant ste 991/1 - 1 and in frame of the sfb 609 `` electromagnetic flow control in metallurgy , crystal growth and electrochemistry '' .
we thank jacques lorat for his proposal and his insistence to build a precession driven dynamo , and canlong ma for assistence in the processing of data from the water precession experiment . | the dresden sodium facility for dynamo and thermohydraulic studies ( dresdyn ) is intended as a platform both for large scale experiments related to geo- and astrophysics as well as for experiments related to thermohydraulic and safety aspects of liquid metal batteries and liquid metal fast reactors .
the most ambitious projects in the framework of dresdyn are a homogeneous hydromagnetic dynamo driven solely by precession and a large taylor - couette type experiment for the combined investigation of the magnetorotational instability and the tayler instability . in this paper
we give a short summary about the ongoing preparations and delineate the next steps for the realization of dresdyn . |
the main theoretical motivation for beyond the standard model physics around tev energies is provided by the hierarchy problem , inexplicable quantum stability of the weak interaction scale with respect to the ultraviolet cutoff .
this problem suggests the existence of some new regulating physics not far above the weak scale .
needless to say , understanding experimental consequences of the latter is of fundamental importance .
recently @xcite , it was pointed out that possible existence of massive vector fields ( @xmath5 ) with the internal quantum numbers identical to the standard model higgs ( or the lepton ) doublet , can result in some interesting phenomenological consequences . due to their quantum numbers , to the leading order such vectors can only have magnetic type interactions with the standard model fermions , @xmath6}^{\rm c } \ , \left ( g^d_{lr } \ , \bar{q}_l \sigma^{\mu\nu } d_r \ , + \ ,
g^e_{lr } \
, \bar{l}\sigma^{\mu\nu } e_r \right ) \\ % \nonumber + { g^u_{lr } \over m } d_{[\mu } v_{\nu ] } \ , \bar{q}_l \sigma^{\mu\nu } u_r \ , + { \rm h.c . } , \label{maincoupling}\end{aligned}\ ] ] where @xmath7 is the charge - conjugated doublet ; @xmath8 and @xmath9 are the left - handed quark and lepton doublets respectively .
@xmath10 are the usual @xmath11 covariant derivatives , and the obvious group and family indexes are suppressed .
@xmath12 is the scale of the new physics and @xmath13 are dimensionless constants .
up until now , no theoretical motivation for the existence of such states was given .
it is the purpose of this paper to provide such a motivation from the hierarchy problem point of view .
we shall show that such states are predicted by three different classes of theories that represent different approaches for explaining the relative lightness of the higgs doublets .
the crucial common feature of all three approaches is , that they are based on the existence of @xmath14 gauge extension of the @xmath11 electroweak group , which is spontaneously broken down to the latter at scale @xmath12 .
the weak doublet vectors @xmath15 are then identified with the @xmath1 doublet components of the @xmath16-dimensional gauge multiplet of the @xmath0 group , which under @xmath1 subgroup decomposes as @xmath17 where numbers refer to the dimensionality of the corresponding @xmath1 representations .
the vector fields @xmath15 ( @xmath18 ) obviously belong to fragments @xmath19 ( @xmath20 ) , and become massive during the spontaneous symmetry breaking @xmath21 .
the lightness of the higgs doublets is guaranteed , because they are related to @xmath15 vectors by symmetry .
this relation in three different approaches is established as follows . in the first approach the lightness of the standard model higgs doublet
is achieved because it is a pseudo - goldstone boson of a spontaneously broken @xmath22 symmetry of the scalar potential , whereas only the diagonal @xmath23 part of it is gauged .
this idea was originally proposed in the context of @xmath24 grand unification @xcite , but our current focus will be the @xmath25 realization of it @xcite . in the latter realization @xmath11
is embedded into @xmath26 group as a maximal subgroup . as explained above
, the @xmath16-dimensional gauge multiplet of @xmath0 , on top of the @xmath27 electroweak gauge bosons and an extra singlet , contains a complex doublet @xmath15 with the quantum numbers of the standard model higgs .
the spontaneous breaking of the local symmetry @xmath28 is triggered by the two independent higgs triplets , @xmath29 and @xmath30 , which under the @xmath1 subgroup decompose as @xmath31 and @xmath32 respectively .
the fragments @xmath33 and @xmath34 are doublets of @xmath1 and have the quantum numbers of the electroweak doublets .
the non - zero vacuum expectation values ( vevs ) are developed by the singlet components , @xmath35 and @xmath36 . as a result of this breaking ,
gauge bosons @xmath15 become massive . the following combination , @xmath37 is eaten - up and becomes a longitudinal components of @xmath15 .
whereas the orthogonal state , @xmath38 is a pseudo - nambu - goldstone , which is massless at the tree - level and gets the suppressed mass only at the loop level , and therefore remains lighter than the symmetry breaking scale .
this pseudo - goldstone plays the role of the standard model higgs doublet . in the second approach @xcite , the higgs mass
is protected because it is related by symmetry to an _ exact _ goldstone boson that becomes a longitudinal component of @xmath15 .
we shall refer to this scenario as `` goldstone - sister higgs '' .
the gauge symmetry structure of the simplest model is identical to the previous case .
there is an exact @xmath39 gauge symmetry that incorporates the standard model group as its maximal subgroup .
again , the spontaneous breaking is triggered by two higgs triplets , @xmath29 and @xmath40 .
however , no approximate global symmetry is required . instead
, the two higgs triplets are related by an exact custodial symmetry , such as , the permutation or an @xmath41 symmetry that transforms the triplets into each other @xmath42 .
the lightness of the higgs doublet is then guaranteed by the following effect .
breaking of @xmath43 symmetry is triggered by the vev of the singlet component of the @xmath30-triplet . during this breaking
, @xmath15 becomes massive and eats up the doublet @xmath34 .
thus @xmath34 becomes a longitudinal polarization of a massive gauge field @xmath44 where @xmath45 is the mass of @xmath15 .
since @xmath34 is a true eaten - up goldstone , it can not have any contribution to its mass from the scalar potential , but only from the kinetic mixing with the gauge field .
but , since the un - eaten doublet @xmath33 is related to @xmath34 by the custodial symmetry , the former also stays massless at the three - level . in this way the mass of the physical higgs doublet is protected by its sister doublet becoming a goldstone particle .
we shall consider this scenario in more details below .
finally , the third class of theories in which appearance of the doublet gauge fields is the must , is the one in which the standard model higgs doublet @xmath46 is an extra dimensional component of a high - dimensional gauge field @xcite . for understanding the key idea of this approach
, it suffices to consider a simplest case of a vector field in five dimensional minkowski space , @xmath47 , where @xmath48 is the five dimensional lorentz index . in the approach of @xcite ,
the higgs is identified with the fifth component of the gauge field , which is a four - dimensional lorentz - scalar , @xmath49 .
obviously , since @xmath46 and @xmath15 are the components of the same high - dimensional gauge field , their internal quantum numbers must be identical .
thus , the existence of the weak - doublet vector particles is reinforced by the high - dimensional gauge symmetry .
this symmetry is spontaneously broken by compactification . in this way the mass of the higgs doublet is controlled by the compactification scale , as opposed to the high - dimensional cutoff of the theory .
the realistic model building in this class of theories is much more involved than in the previous two cases . for us
the only important aspect is the model - independent property of the existence of the massive vector doublet @xmath15 .
this property is guaranteed by the symmetry and is insensitive to the concrete model building .
having specified the class of the theories of our interest , let us turn to the interaction between @xmath15 and the standard model fermions .
in all three classes of theories , the coupling ( [ maincoupling ] ) even if not present at the tree - level can ( and in general will ) be generated by the loop corrections , as it is permitted by all the symmetries of the low energy theory .
we shall first demonstrate how this generation happens by considering a toy model reduced to its bare essentials , and later illustrate it on a detailed example of goldstone - sister higgs @xcite .
in this section we shall discuss a generation of coupling ( [ maincoupling ] ) in a simple toy model .
the latter consists of two sectors : hypothetical heavy particles and chiral massless fermions of ordinary matter .
the corresponding lagrangian reads , @xmath50 } d^{[\mu } v^{\nu]}\ , + \ , m^2_v v^\dagger_\mu v^\mu + \partial_\mu\varphi^*\partial^\mu\varphi - m^2\varphi^*\varphi\\ \nonumber & + & \sum_{k=1,2}\bar{\psi}'_k \left(i d\!\bk - m\right)\psi'_k + g \bar{\psi}'_2\gamma^\mu\psi'_1 v_\mu + g v^\dagger_\mu\bar{\psi}'_1\gamma^\mu\psi'_2\\ \nonumber & + & \sum_{k=1,2}\bar{\psi}_k i d\!\bk \,\psi_k + \frac{h}{2}\bar{\psi}_2(1+\gamma^5)\psi'_2\varphi + \frac{h}{2}\bar{\psi}'_1(1+\gamma^5)\psi_1\varphi^*\\ % \nonumber & + & \frac{h}{2}\bar{\psi}'_2(1-\gamma^5)\psi_2\varphi^ * + \frac{h}{2}\bar{\psi}_1(1-\gamma^5)\psi'_1\varphi,\end{aligned}\ ] ] where the first line represents the bilinear lagrangian of the @xmath1-doublet vector fields , @xmath51 , and of a complex singlet scalar field , @xmath52 .
the second line describes the kinetic and mass terms of a singlet ( @xmath53 ) and a doublet ( @xmath54 ) heavy fermions and their interactions with the doublet vector fields .
the primed fermions are non - chiral , meaning that each of them comes in both left and right chiralities .
the last two lines include the kinetic terms of the ordinary chiral massless fermions , the left - handed doublets , @xmath55 , and the right - handed singlets , @xmath56 , and their interactions with the heavy fields . the one - loop diagram in fig .
[ fig:1 ] leads to new anomalous coupling of the vector fields with the ordinary fermions @xmath57},\end{aligned}\ ] ] where @xmath58 is a slow varied function at @xmath59 .
in this section we shall discuss an explicit example of the realistic model .
as such we shall choose a model based on the idea of goldstone - sister higgs @xcite . as discussed above , in this theory the electroweak @xmath11-symmetry is enhanced to @xmath39 . obviously the gauge sector contains an additional electroweak - doublet vector field ( @xmath15 ) , which after spontaneous breaking @xmath60 becomes massive .
this breaking is realized by the vev of a higgs triplet @xmath30 , the weak - doublet part of which ( @xmath34 ) becomes a longitudinal component of @xmath15 . in order to trigger the second standard model stage of symmetry breaking @xmath61
, the theory contains a second higgs triplet @xmath29 .
the key point is that the two triplets are related by an additional custodial @xmath41 symmetry .
the motivation for such symmetry is , that the physical higgs doublet ( @xmath33 ) is a partner of the goldstone boson @xmath34 that is eaten up via the higgs effect of @xmath60 breaking . due to this symmetry relation ,
the sm higgs doublet remains naturally light .
it is the simplest to discuss this mechanism directly in supersymmetric grand unified theory ( gut ) context , in which the @xmath62 group together with the color @xmath63 is embedded into the @xmath64 group as a maximal subgroup , @xmath65 .
notice that this embedding allows to treat color and weak @xmath66-groups in a completely democratic way .
notice , that for our present purposes having the full @xmath64-symmetry group is completely unessential .
the latter is anyway broken down to its @xmath67 subgroup at scales much above the energies of our present interest .
so we could have equally well limit ourselves by @xmath67 symmetry .
however , the analysis is much more convenient in terms of @xmath64 representations , rather than in terms of its subgroups .
it also allows us to understand the particle content in term of representations of a more familiar @xmath68 subgroup .
due to this advantages , we shall use @xmath64-classification for the particles .
if needed , the reader can easily perform decompositions into the @xmath67-reduced representations instead .
the full symmetry of the model is thus @xmath69 , where @xmath70 is an additional _ custodial _ symmetry that relates the sm higgs doublet to an _ eaten - up goldstone boson_. the chiral superfield content is : 1 .
higgs sector : + 35-plet @xmath71 and + @xmath72 , @xmath73 , where @xmath74 are @xmath64 and @xmath75 are @xmath70 indexes respectively .
the sm fermions are embedded in the following anomaly - free set ( per generation ) : + @xmath76}$ ] , @xmath77 and a singlet @xmath78 .
we shall denote the superfields by the same symbols as their components . in each case
it will be clear from the context which component we are referring to .
the symmetry breaking is achieved by the higgs part of the superpotential , which has the following form , @xmath79 the vacuum of the theory is : @xmath80 which leads to the symmetry breaking @xmath81 .
the value of @xmath82 is undetermined in susy limit .
we are interested in the situation when the scale @xmath82 is not very high , around tev or so , whereas the scale @xmath83 is much larger . in this way , below the latter scale
the low energy symmetry group is @xmath67 .
the further breaking of this symmetry to the standard model group , is triggered at much lower scale @xmath82 by the vev of the @xmath84-plets . as a result of this breaking ,
the higgs doublets in @xmath85 and @xmath86 are eaten up by the @xmath1-doublet gauge fields @xmath87 that reside in the adjoint representation of @xmath0 .
the doublets in @xmath88 and @xmath89 are physical and at the tree level ( in susy limit ) have very small masses @xmath90 .
an additional contribution comes from loop corrections after supersymmetry breaking ( see @xcite for details ) .
notice , that as a result of the symmetry structure of the theory , the color triplet partners of the higgs doublets are automatically super - heavy , with the masses @xmath91 , and decouple from the low energy spectrum .
this solves the doublet - triplet splitting problem in susy guts .
the fermion masses are generated from the following interactions .
the up - type quark masses come from @xmath92}{\bf 15}_{[nl]}~\varepsilon^{ijkmnl},\ ] ] and the masses of down - type quarks , charged leptons and heavy states come from @xmath93}{\bf{\bar{6}}}^{bj } + h_{aj}{\bf{\bar{6}}}^{aj } { \bf{1}}.\ ] ] from the above couplings the following interactions between the doublet vector and light fermions are generated .
the superdiagram containing external light up - type quarks is given in fig .
[ fig : s1 ] .
the corresponding effective operator of interest is , @xmath94}\sigma^{\mu\nu } { \bf 15}_{[ln ] } \ , \varepsilon^{ba}\varepsilon^{jikmln},\ ] ] where @xmath95 is the @xmath64 gauge field strength , the fermionic components are taken from @xmath96-plets and bosonic components from the rest . after substituting the vevs of @xmath84-plet higgses , this operator reduces to the magnetic coupling between @xmath15 and the up - type quarks given by the last term in ( [ maincoupling ] ) . coupling to down - type quarks and leptons
is generated through the diagram of fig .
[ fig : s2 ] , which leads to the following effective operator , @xmath97 } \sigma^{\mu\nu } { \bf \bar{6}}^{aj}\varepsilon_{ba}.\ ] ] after substitution of the vevs , the above interaction reduces to the magnetic coupling of @xmath15 with @xmath98-quark and charged leptons , given by the first two terms in ( [ maincoupling ] ) .
notice , that due to supersymmetry violating insertions in the vertexes , the flavor structure of the operators ( [ kup ] ) and ( [ kdown ] ) is not necessarily aligned with the flavor structure of quark and lepton mass matrixes .
this means , that the exchange by @xmath15 could potentially contribute into new flavor and cp violating interactions .
notice also , that the same operator ( [ kdown ] ) could potentially give a contribution to @xmath99 of leptons .
indeed , if we instead of inserting the @xmath1-doublet vev of @xmath86 , insert the vev of a physical higgs doublet living in @xmath89 , the operator ( [ kdown ] ) will reduce to the magnetic moment coupling of photon . however , this magnetic coupling will mix the light left - handed lepton residing in @xmath96-plet with the heavy right - handed lepton from @xmath100 , as opposed to the standard model right - handed lepton residing in @xmath101 , thus , giving no contribution to @xmath99 . however
, a non - zero contribution could arise in case of a small mixing between the doublets from @xmath102 and @xmath103 .
such mixing could arise from the calculable radiative corrections , and thus , could relate the phenomenological signatures of @xmath15 with the value of @xmath99 .
this study will not be attempted in the present work .
instead , we shall focus on characteristic signatures of resonance production and decay of @xmath15-bosons .
in paper @xcite it has been shown that tensor current@xmath104current interaction leads to a new angular distribution in comparison with well - known vector interactions .
it was realized later @xcite that this property ensures distinctive signature for their detection at the hadron colliders .
the hadron colliders , due to their biggest center - of - momentum ( cm ) energy @xmath105 several tevs and their relatively compact sizes , still remain the main tools for discoveries of very heavy particles .
the presence of partons with a broad range of different momenta allows to flush the entire energetically accessible region , roughly , up to @xmath106 .
the production mechanism for new heavy bosons at a hadron collider is the @xmath107 resonance fusion . in this paper
we will consider the resonance production and decay of the above - introduced heavy spin-1 gauge bosons into the light lepton pairs , electrons or muons .
for such high energies it is convenient to use the helicity formalism , since the helicity is a good quantum number for massless particles .
this , in a way , fixes the dominant production and decay mechanisms .
for example , the decay angular distribution in the cm frame of a particle with spin-@xmath108 and helicity @xmath109 with @xmath110 decaying into two massless particles with helicities @xmath111 and @xmath112 can be written as @xcite @xmath113 where the helicity amplitude @xmath114 is expressed through the difference @xmath115 and the reduced decay amplitude @xmath116 , which is a function of @xmath108 and the outgoing helicities , but is independent of the polar ( @xmath117 ) and the azimuthal ( @xmath118 ) angles .
up until now only resonance production and decay of spin-1 bosons with maximal helicities @xmath119 have been considered . they are associated with additional @xmath120 gauge symmetries and are usually called @xmath3 particles . the lorentz structure of its couplings to each fermion flavor
is characterized by two generally independent constants @xmath121 and @xmath122 @xmath123 experimental determination of these coupling constants or disentangling among the different models is a rather hard task and can not be fulfilled , for example , with the first lhc data . in the best case only the specific symmetric angular distribution over the polar angle @xmath117 , @xmath124 allows to distinguish its production from distributions of spin-0 , @xmath125 and spin-2 @xmath126 resonances .
another possibility is the resonance production and decay of longitudinal spin-1 bosons with @xmath127 , but this possibility is not widely discussed .
while the @xmath3 bosons with helicity @xmath119 are produced in left(right)-handed quark and right(left)-handed antiquark fusion , the longitudinal @xmath128 bosons can be produced through the new chiral couplings ( [ maincoupling ] ) , @xmath129 } + \frac{g^f_{rl}}{m}\overline{\psi^f_r}\sigma^{\mu\nu } \psi^f_l\partial_{[\mu } z^*_{\nu]}\right)\!,\ ] ] with the complex constants @xmath130 in left - handed or right - handed quark - antiquark fusion @xcite .
the new couplings lead to a different angular distribution @xmath131 than the previously considered ones . at first sight
, the small difference between the distributions ( [ gll ] ) and ( [ glr ] ) seems unimportant .
however , the absence of the constant term in the latter case results at least in two potential experimental signatures .
first of all , the known angular distributions for scalar ( [ g0 ] ) , vector ( [ gll ] ) and spin-2 ( [ g2qq],[g2gg ] ) bosons include a nonzero constant term , which leads to the kinematic singularity in @xmath132 distribution of the final lepton @xmath133 in the narrow width approximation @xmath134 @xmath135 this singularity is transformed into a well known jacobian peak due to a finite width of the resonance .
in contrast to this , the pole in the decay distribution of the @xmath128 bosons is canceled out and the lepton @xmath132 distribution even reaches zero at the kinematical endpoint @xmath136 .
therefore , the @xmath128 boson decay distribution has a broad smooth hump with the maximum below the kinematical endpoint , instead of a sharp jacobian peak ( fig .
[ fig : pt ] ) .
another striking feature of the distribution ( [ glr ] ) is the forbidden decay direction perpendicular to the boost of the excited boson in the rest frame of the latter ( the collins soper frame @xcite ) .
it leads to a peculiar `` swallowtail '' shape of the angular distribution with a profound dip at @xmath137 in the collins soper frame ( fig .
[ fig : cos ] ) @xcite . in conclusion
, we would like to emphasize the difference between group properties of the gauge spin-1 @xmath3 and @xmath128 bosons . while the @xmath3 bosons are described by real representations transforming as triplets or singlets of @xmath11 group , the @xmath128 bosons , like the higgs fields ,
are assigned to the complex representation and transform as doublets .
this results in another important experimental consequence . together with the neutral @xmath138 bosons
the doublets always contain also the charged bosons , @xmath139 , which decay into a charged lepton and an undetected neutrino .
therefore , the angular distribution in fig .
[ fig : cos ] is experimentally unaccessible for them in the lepton channel .
only the @xmath132-distribution of the charged lepton can be measured .
however , the distribution of the @xmath140 bosons differs drastically from the distribution of the @xmath4 bosons ( fig .
[ fig : pt ] ) .
hence , even relatively small decay width of the @xmath140 bosons leads to a wide hump without the jacobian peak , that obscures their identification as resonances at the hadron colliders .
the only way to access the angular distribution for @xmath140 bosons like in fig .
[ fig : cos ] is kinematical reconstruction of their decays into heavy quarks , @xmath141 , which , in spite of the strong qcd background , can be identified via @xmath142-tagging .
the presence of the heaviest @xmath143 quark in the final state will lead to an additional contribution , @xmath144 to the angular distribution ( [ glr ] ) proportional to the ratio @xmath145 due to helicity flip of the @xmath143 quark .
we are grateful to c. wetterich for discussions .
the work of mc was partially supported by grant - in - aid for scientific research 104/15.05.2009 from the sofia university .
the research of gd is supported in part by european commission under the erc advanced grant 226371 , by david and lucile packard foundation fellowship for science and engineering and by the nsf grant phy-0758032 .
chizhov , hep - ph/0609141 .
z.g . berezhiani and g.dvali , bull .
. inst . * 5 * ( 1989 ) 55 ; kratk .
soobshch . fiz . * 5 * ( 1989 ) 42 ; r. barbieri , g. dvali and m. moretti , phys .
b * 312 * ( 1993 ) 137 ; z. berezhiani , c. csaki and l. randall , nucl . phys .
b * 444 * ( 1995 ) 61 , hep - ph/9501336 . g.r .
dvali , phys .
b * 324 * ( 1994 ) 59 .
manton , nucl .
b * 158 * ( 1979 ) 141 ; d.b .
fairlie , j. phys .
g * 5 * ( 1979 ) l55 ; phys .
b * 82 * ( 1979 ) 97 . g.r .
dvali , s. randjbar - daemi and r. tabbash , phys .
d * 65 * ( 2002 ) 064021 , hep - ph/0102307 ; n. arkani - hamed , a.g .
cohen and h. georgi , phys .
b * 513 * ( 2001 ) 232 , hep - ph/0105239 ; l.j . hall , y. nomura and d.r .
smith , nucl . phys .
b * 639 * ( 2002 ) 307 , hep - ph/0107331 ; i. antoniadis , k. benakli and m. quiros , new j. phys .
* 3 * ( 2001 ) 20 , hep - th/0108005 ; c. csaki , c. grojean and h. murayama , phys .
d * 67 * ( 2003 ) 085012 , hep - ph/0210133 .
chizhov , hep - ph/0008187 .
haber , hep - ph/9405376 .
chizhov , v.a .
bednyakov and j.a .
budagov , phys . atom .
* 71 * ( 2008 ) 2096 , arxiv:0801.4235 [ hep - ph ] .
collins and d.e .
soper , phys .
d * 16 * ( 1997 ) 2219 . | we study phenomenological consequences of the standard model extension by the new spin-1 fields with the internal quantum numbers of the electroweak higgs doublets .
we show , that there are at least three different classes of theories , all motivated by the hierarchy problem , which predict appearance of such vector weak - doublets not far from the weak scale .
the common feature for all the models is the existence of an @xmath0 gauge extension of the weak @xmath1 group , which is broken down to the latter at some energy scale around tev .
the higgs doublet then emerges as either a pseudo - nambu - goldstone boson of a global remnant of @xmath0 , or as a symmetry partner of the true eaten - up goldstone boson . in the third class ,
the higgs is a scalar component of a high - dimensional @xmath0 gauge field .
the common phenomenological feature of these theories is the existence of the electroweak doublet vectors @xmath2 , which in contrast to well - known @xmath3 and @xmath4 bosons posses only anomalous ( magnetic moment type ) couplings with ordinary light fermions .
this fact leads to some unique signatures for their detection at the hadron colliders . |
in spite of many attempts , the evolution of @xmath0li in the galaxy is not completely understood .
lithium has the unique property of being produced by , at least , three different processes : big bang nucleosynthesis , spallation of galactic cosmic - ray ( gcr ) particles on interstellar matter ( ism ) nuclei , and stellar nucleosynthesis . at the same time , @xmath0li is easily destroyed by proton captures in stellar interiors at relatively low temperatures ( @xmath1 k ) .
observationally , population i ( hereafter pop .
i ) dwarf stars show a large dispersion in @xmath0li abundance attributed to different amounts of depletion . on the contrary
the abundance of @xmath0li in the atmosphere of old , warm ( @xmath2 k ) population ii ( hereafter pop .
ii ) dwarf stars is surprisingly uniform : @xmath0li / h @xmath3 ( spite & spite 1982 ) , a value @xmath4 times lower than the maximum abundance observed in pop .
several studies have subsequently confirmed the existence of the so - called `` spite plateau '' in pop .
ii stars , although discrepancies on the actual value of the plateau still exist at the level of @xmath5 dex ( see e.g. , spite & spite 1986 ; hobbs & duncan 1987 ; rebolo , beckman , & molaro 1988 ; thorburn 1994 ; spite et al . 1996 ; bonifacio & molaro 1997 ) .
@xmath6 + 12 $ ] .
the spite plateau corresponds to @xmath7 @xmath8 . ]
there is an ongoing debate about whether the spite plateau corresponds to the ( almost ) undepleted primordial abundance of @xmath0li , or whether the primordial value is closer to @xmath7 @xmath93.3 ( the value measured in the youngest stars and solar system meteorites ) and has been significantly but uniformly depleted in pop .
ii stars ( see e.g. deliyannis & ryan 1997 ; bonifacio & molaro 1997 ) .
on one hand , the former interpretation requires processes that increase the galactic @xmath0li abundance by a factor @xmath4 in the first few gyr . on the other hand _ standard models _ of stellar evolution which incorporate
only convective mixing predict little ( @xmath5 dex ) @xmath0li depletion in pop .
ii stars ( e.g. , deliyannis , demarque & kawaler 1990 ) .
stellar models including the effects of non - standard processes ( like e.g. diffusion , winds , turbulence induced by rotational instabilities , slow mixing driven by angular momentum loss , etc . ) predict substantial @xmath0li destruction in pop .
ii stars ( see pinsonneault 1997 for a review ) .
these non - standard models predict observable features , like a dispersion in @xmath0li among halo dwarfs , or correlations between @xmath0li and t@xmath10 or [ fe / h ] , which have not been confirmed by observations , suggesting that @xmath0li destruction must have been minimal in these stars ( see e.g. , boesgaard & steigman 1985 , bonifacio & molaro 1997 , ryan , norris , & beers 1999 , ryan et al .
it is true , however , that the presence of a few metal poor stars with @xmath0li abundances somewhat above the plateau , as well as other observational evidences in pop .
i stars ( deliyannis et al . 1998 , deliyannis 2000 ) support the idea that halo dwarfs may have suffered a certain amount of @xmath0li depletion
. indication of a primordial abundance of @xmath0li in excess of the spite plateau value is suggested also by ( _ i _ ) recent determinations of deuterium abundances in lyman-@xmath11 clouds ( see e.g. pettini & bowen 2001 and references therein ) , and ( _ ii _ ) observations of the spectrum of fluctuations in the cosmic background radiation ( de bernardis et .
al . 2000 , tegmark & zaldarriaga 2000 ) .
in both cases the value of the baryon - to - photon ratio @xmath12 results in the range 67 , rather than 45 as usually assumed . this implies a cosmological abundance of @xmath0li in excess of the spite plateau value by a factor 23 , still compatible with the observationally allowed range of primordial @xmath13he .
however , one should be very careful in interpreting these indications since possible sources of systematic errors have still to be completely understood and eliminated from the observational data .
the issue remains therefore open . assuming that the spite plateau corresponds to the primordial @xmath0li abundance , the difference between the plateau abundance and the maximum @xmath0li content in pop .
i stars underscores the need for one or more sites of @xmath0li production .
it is well known that @xmath14be , @xmath15b , and @xmath16li can be produced via spallation and fusion reactions between gcr particles and ism nuclei ( fowler , reeves , & silk 1970 ; meneguzzi , audouze & reeves 1971 ) .
whereas theoretical calculations of the amounts of @xmath14be , @xmath15b and @xmath17li produced by gcr ( see e.g. lemoine , vangioni - flam , & cass 1998 ; fields & olive 1999 ; ramaty et al . 2000 , and references therein ) can roughly reproduce the observed values , the amount of @xmath0li produced by spallation processes at the time of formation of the sun is a factor 510 lower than the measured meteoritic value ( see also nissen et al .
1999 and references therein ) [ fe / h ] @xmath18 , the predictions of gcr spallation models indicate a possible overproduction of @xmath16li due to @xmath11-@xmath11 reactions ( prantzos et al .
1993 , valle et al .
2001 in preparation ) that may suggest the occurence of a depletion process in halo stars at least in this metallicity range . ] .
additional sites of @xmath0li production are therefore required to match the observational data .
for example , dantona & matteucci ( 1991 ) proposed that novae and agb stars may represent the main sources of galactic @xmath0li , whereas matteucci , dantona , & timmes ( 1995 ) favoured a model where the main contributors to @xmath0li enrichment are snii and agb stars .
abia , isern , & canal ( 1995 ) considered the production of @xmath0li by low - mass agb stars , whereas romano et al .
( 1999 ) showed that novae are required in order to reproduce the growth of @xmath0li with metallicity , and that the contribution from snii should be lowered by at least a factor of two .
finally , romano et al . ( 2001 ) , using recent agb models by ventura , dantona , & mazzitelli ( 2000 ) , concluded that agb stars are not substantial @xmath0li producers in the galaxy . in this paper
we reanalyze the problem afresh with the help of a numerical model of galactic chemical evolution and a set of new models of agb stars of various masses and metallicities .
we consider various stellar sources of @xmath0li : novae , type ii supernovae , low and intermediate mass stars in the red giant branch ( rgb ) and agb phase .
all these sources can produce @xmath0li , as shown by theoretical models , and , in some cases , also confirmed by spectroscopic data .
however , individual @xmath0li yields are rather uncertain and strongly dependent on model parameters and assumptions .
the paper is organized as follows . in 2
we discuss our sample of observational data for field stars and open clusters . in
3 we briefly describe the adopted model for galactic chemical evolution . in the following sections we then focus on the analysis of the @xmath0li production by different stellar sources . in particular in
4 we consider the role of intermediate - mass agb stars ( hereafter ims - agb ) , presenting new @xmath0li yields obtained from the agb models by frost ( 1997 ) .
we reanalyze the production of @xmath0li by novae ( 5 ) , snii ( 6 ) , and low - mass rgb stars ( 7 ) , on the basis of recent model calculations and observations . finally , in 8 we discuss and summarize our results .
the main source of our compilation is the recent study by fulbright ( 2000 ) who determined @xmath0li , fe , and other element abundances for 168 halo and disk stars .
both stellar parameters and abundances of @xmath0li and fe were inferred by carrying out a self - consistent lte analysis , providing a homogeneous sample of both pop .
ii and pop .
the survey of fulbright does not extend to very low or very high metallicities ; therefore we complemented it for [ fe / h ] @xmath19 with the compilation of bonifacio & molaro ( 1997 ) . at [ fe / h ] @xmath20 we included a sample of warm ( @xmath21 k ) , unevolved f - type stars taken from balachandran ( 1990 ) and lambert , health , & edvardsson ( 1991 ) .
we note that , whereas cooler stars have certainly undergone li depletion ( unless very young ) , considering only stars warmer than 5700 k allows having a subsample of minimally depleted stars which trace the upper envelope of the @xmath0li vs. [ fe / h ] distribution .
surveys of @xmath0li among pop .
ii stars have also been made by e.g. rebolo et al .
( 1988 ) , thorburn ( 1994 ) , ryan , norris , & beers ( 1996 ) , spite et al .
( 1996 ) , ryan et al .
different datasets and/or analysis methods ( in particular the use of different effective temperature calibrations ) lead in general to a different abundance scale : therefore , including @xmath0li abundances from many different sources in the same compilation may introduce a spurious ( i.e. not real ) dispersion in the @xmath0li vs [ fe / h ] distribution . for this reason , we decided to base our compilation of @xmath0li abundances for pop .
ii stars on the minimum number of studies allowing the full coverage of the entire metallicity range .
we mention that fulbright ( 2000 ) derived an average offset between his and bonifacio & molaro s ( 1997 ) effective temperatures @xmath22 k , which should not introduce major systematic differences in the inferred @xmath0li abundances .
there are several stars in common between the two studies : for all of them the agreement in @xmath0li abundances is indeed very good ( within @xmath5 dex ) .
we did not include in our list stars evolved off the main sequence ( hereafter ms ) and/or stars cooler than 5700 k ; both groups of stars most likely did not preserve their original @xmath0li content due to either post - ms dilution or to ms @xmath0li depletion and thus are not adequate to trace the upper envelope of the @xmath0li vs. metallicity distribution .
we also did not include @xmath0li abundances for t tauri stars , since they may be affected by several problems such as spectral veiling , nlte , or , more generally , by the effects of a disturbed stellar photosphere " ( see e.g. duncan 1991 ) .
instead , we included in our compilation the average abundance of warm ( i.e. f type ) stars in young open clusters with @xmath23[fe / h]@xmath24 .
these stars have undergone minimal pre - ms or ms @xmath0li depletion and thus their abundance can be considered as representative of the present ism @xmath0li content .
several surveys of @xmath0li among open cluster stars ( see e.g. jeffries 2000 for a recent review ) have shown that li - undepleted stars in all the investigated clusters share the same abundance @xmath25 ( e.g. , randich et al .
1997)li abundances as high as @xmath26 have been derived for some t tauri stars , magazz , rebolo , & pavlenko ( 1992 ) and martn et al .
( 1994 ) concluded that abundances of pre - ms stars ( including both classical and weak - lines t tauri stars ) as a whole indicate an initial @xmath0li abundance for pop .
i stars @xmath27 . ] . in the upper panel of figure 1
we plot the @xmath0li abundance vs. metallicity for the stars of our sample .
a few features in this figure warrant detailed comments . as far as the @xmath0li
plateau is concerned , recent studies have inferred slightly different values for the plateau value , due as mentioned above to different types of analysis and , in particular , to possible offsets in the zero - point of the effective temperature scale ( see e.g. , spite et al .
in our model of galactic evolution we have generally adopted as initial @xmath0li abundance the value determined by bonifacio & molaro ( 1997 ) for the spite plateau , i.e. @xmath28 , under the implicit assumption that the spite plateau represents the primordial ( big - bang ) abundance .
we mention in passing that the mean abundance of the stars in the compilation of fulbright ( 2000 ) with [ fe / h]@xmath29 is @xmath30 in excellent agreement with the estimate of bonifacio & molaro ( 1997 ) .
we also ran our model starting with an initial @xmath0li abundance higher by a factor 2 than the spite plateau value ( see below ) , following the idea that pop .
ii stars may have depleted a certain amount of li in the course of their evolution .
figure 1 also suggests that the rise of @xmath0li abundance from the spite plateau may be smoother than claimed in previous studies .
in particular , a few stars are present in the survey of fulbright ( 2000 ) which have intermediate metallicities ( @xmath31 [ fe / h ] @xmath32 ) and @xmath0li abundances 0.20.3 dex ( i.e. @xmath33 ) above the plateau but not as high as @xmath34 . however , the caveat is that only a few stars define the upper envelope at intermediate metallicity and we can not exclude that these stars have suffered some li depletion . a sharp increase to abundances as high as the present ism abundance may occur at metallicities between [ fe / h]@xmath35 and @xmath36 , if the abundances for the few stars which trace the upper envelope of the distribution at those metallicities are correct .
these stars are from the survey of balachandran ( 1990 ) , but we mention that li abundances derived by balachandran are systematically higher than those of boesgaard & tripicco ( 1986 ) , with differences in the range 0.2 0.6 dex . for the case of hr 8315 ( hd 206901 ) lbre et al .
( 1999 ) inferred a metallicity [ fe / h]@xmath37 and a @xmath0li abundance @xmath38 to be compared with [ fe / h]@xmath39 and @xmath40 determined by balachandran ( 1990 ) .
we also notice that photometric iron abundances for some of the intermediate metallicity and li - rich stars observed by balachandran ( 1990 ) are at least 0.2 dex higher than the spectroscopic values inferred by balachandran ( see the last two columns of table 1 in balachandran 1990 ) , suggesting again that @xmath0li abundance @xmath41 is reached only at higher metallicities ( i.e. [ fe / h]@xmath42 ) .
determinations of @xmath0li abundances in this critical metallicity range are certainly needed in order to better constrain the @xmath0li vs. [ fe / h ] upper envelope . in the lower panel of figure 1
we show the abundance of @xmath0li vs. galactic age using @xmath0li data for stars in galactic open and globular clusters .
we also show the meteoritic @xmath0li abundance ( @xmath43 , anders & grevesse 1989 ) , the interstellar @xmath0li abundances determined in the line - of - sight toward @xmath44 oph ( @xmath0li /h @xmath45 , lemoine et al .
1993 ) , and toward @xmath46 and @xmath47 per ( @xmath0li /h @xmath48 and @xmath49 , knauth et al .
2000 ) . for young open clusters ( ages @xmath50 gyr )
the average abundance measured in warm ( @xmath51 k ) , undepleted cluster ms stars , is plotted in the figure .
the data were taken from the following sources : soderblom et al .
( 1999 ) for ngc 2264 ; randich et al .
( 1997 ) for ic 2602 ; randich et al .
( 2001 ) for ic 2391 ; martn & montes ( 1997 ) for ic 4665 ; randich et al .
( 1998 ) for alpha persei ; soderblom et al .
( 1993 ) for the pleiades ; thorburn et al .
( 1993 ) for the hyades . on the other hand ,
the initial abundance for older clusters was estimated from stars at ( or just evolved off ) the turn - off in the 2 gyr clusters ngc 752 ( hobbs & pilachowski 1986 ) , ngc 3680 ( randich , pasquini , & pallavicini 2000 ) and from the abundances of the two components of the tidally locked binary s1045 in the solar - aged m 67 cluster ( deliyannis et al .
1994 ; pasquini , randich , & pallavicini 1997 ) .
finally , globular clusters ( gcs ) are represented by a box that shows the ranges in @xmath0li and age estimates .
more specifically , gcs absolute ages are the subject of a long standing debate ; in the lower panel of figure 1 we used the recent determination by carretta et al .
( 2000 ) , who inferred an average age of @xmath52 gyr based on _ hipparcos _ parallaxes for local subdwarfs .
we mention , however , that ages as large as 1617 gyr can not be ruled out .
abundances of @xmath0li were derived for unevolved stars at the turn - off in three clusters only , namely , ngc 6397 ( pasquini & molaro 1996 ) , 47 tuc ( pasquini & molaro 1997 ) , and m 92 ( boesgaard et al .
1998 ) ; whereas the average abundances for the three clusters are close to the value of the spite plateau , stars in the same cluster show a significant dispersion in @xmath0li ( see discussion in boesgaard et al . 1998 ) and , in particular , a star is found in m 92 with a @xmath0li content significantly higher than the plateau and the other cluster stars .
consequently , we think that the use of a box is more appropriate to indicate the position of gcs in figure 1 .
the model of galactic chemical evolution adopted in this work is described in detail by ferrini & galli ( 1988 ) , galli & ferrini ( 1989 ) and ferrini et al .
the same model was adopted by galli et al .
( 1995 ) , travaglio et al .
( 1999 ) and travaglio et al .
( 2001 ) , to study the evolution of the light elements d and @xmath53he , and the evolution of the heavy elements from ba to pb , respectively .
we briefly recall here the basic features of the model .
the galaxy is divided into three zones , halo , thick disk , and thin disk , whose composition of stars , gas ( atomic and molecular ) and stellar remnants is computed as functions of time up to the present epoch @xmath54 gyr .
the thin disk is formed from material infalling from the thick disk and the halo .
the formation of the sun in the thin disk takes place 4.5 gyr ago , i.e. at epoch @xmath55 gyr . the star formation rate in the three zones is not assumed _ a priori _ , but is obtained as the result of self - regulating processes occurring in the molecular gas phase , either spontaneously or stimulated by the presence of other stars . in models of galactic chemical evolution
it is customary to separate the contribution of stars in different mass ranges to the enrichment of the ism .
our model follows the evolution of ( _ i _ ) single low- and intermediate - mass stars ( @xmath56 ) ending their life as he or c - o white dwarf , ( _ ii _ ) binary systems able to produce type i supernovae , and ( _ iii _ ) single massive stars ( @xmath57 ) , the progenitors of type ii supernovae .
the value of @xmath58 depends on metallicity ; we assume @xmath59 @xmath60 for [ fe / h]@xmath61 , and @xmath62 @xmath60 otherwise ( see for references tornamb & chieffi 1986 ) . the adopted initial mass function ( hereafter imf )
is discussed in ferrini et al .
stellar nucleosynthesis is treated according to the matrix formalism introduced by talbot & arnett ( 1973 ) .
specific yields are taken from calculations by woosley & weaver ( 1995 ) and thielemann , nomoto , & hashimoto ( 1996 ) , for type ii and type i supernovae , respectively .
hot bottom burning ( hereafter hbb ) occurs in intermediate mass agb stars when the bottom of the convective envelope of the star reaches the top layers of the hydrogen burning shell .
sackmann & boothroyd ( 1992 ) showed that the cameron - fowler mechanism ( cameron 1955 ; cameron & fowler 1971 ) is very effective at producing @xmath0li , and can quantitatively account for the @xmath0li abundance measured by smith & lambert ( 1990 ) in the super - li rich agb stars in the magellanic clouds .
we present here a summary of results for @xmath0li production in agb stars of various masses and metallicities , based on the evolutionary models of frost ( 1997 ) and synthetic models using the technique of forestini & charbonnel ( 1997 ) .
we then discuss the sensitivity of @xmath0li enrichment to the mass - loss prescriptions during the latest phases of agb evolution .
frost ( 1997 ) computed the evolution of 4 , 5 , and 6 @xmath60 models each with metal content @xmath64 , 0.008 , and 0.004 , appropriate to the solar neighborhood , and to the large and small magellanic clouds , spanning the evolution from the ms through the end of the agb phase .
each case has been the subject of a detailed nucleosynthesis study using a modified post - processing code ( cannon 1993 ) including 74 species and over 500 reactions .
some of the results of these models have been already published ( e.g. frost et al .
1998 , lattanzio & forestini 1999 ) ; in this paper we present in detail an analysis of the production of @xmath0li .
our present approach is to use the stellar models of frost ( 1997 ) as the basis of synthetic models computed with the code of forestini & charbonnel ( 1997 ) . in this way we force the synthetic code to follow the results of the stellar models (
accuracy is within 10% ) , yet we are free to vary those parameters which have no significant feedback on the evolution . in particular , for the aim of this paper , we investigated the mass - loss rate at the end of the agb phase ( see next section for detailed discussion ) .
we do not expect modest changes in the mass - loss rate during the latest agb phase to have a significant effect on the structural evolution of the star , since the variations in the mass - loss rate only affect the very late phases of stellar evolution .
nevertheless , we caution against placing too much belief in the quantitative results for the extreme cases discussed below .
the surface @xmath0li abundances during the agb lifetime in the stellar envelope for the nine models considered here is presented in figure 2 .
note that the abundance of @xmath0li at the beginning of the agb phase varies from one model to another as a result of previous evolution .
2 is a @xmath65 grid figure with mass increasing along the @xmath66-axis and @xmath67 decreasing along the @xmath68-axis : in this way we expect hbb to increase for increasing @xmath66-axis ( mass ) or increasing @xmath68-axis ( decreasing @xmath67 ) .
the most extreme hbb is for the top right plot , i.e. @xmath69 @xmath60 model with @xmath70 , where the maximum @xmath0li production lasts for the longest time ( @xmath71 yr ) .
the maximum @xmath0li abundances we obtain range around @xmath7 @xmath72 , independent of stellar mass and metallicity , and only depending on the occurrence of hbb . the models with @xmath73 @xmath60 and @xmath64
, 0.008 show no hbb and consequently no @xmath0li production .
in fact , @xmath0li destruction in these two models proves that their convective bottom temperatures are high enough to burn @xmath0li through @xmath0li@xmath74he channel , but not enough to produce @xmath0be ( that decays to @xmath0li , cameron - fowler mechanism ) by @xmath53he burning .
conversely the 6 @xmath60 models , with their substantial envelopes , begin hbb immediately on the agb and hence they produce a large amount of @xmath0li , but the high envelope temperature also assists in the destruction of @xmath0li . finally , figure 2 shows that in all models , as the evolution of the star proceeds , the surface @xmath0li abundance decreases again more or less rapidly , depending on the mass and metallicity .
this indicates that the initial reserves of @xmath53he have been consumed in the production of @xmath0li , and this @xmath0li is now itself being destroyed ( without replenishment ) by proton captures .
there are observational indications that some galactic c - stars ( stars with c / o @xmath75 ) are also super - li rich stars ( @xmath7 @xmath76 , see e.g. abia et al . 1991 , abia , pavlenko , & de laverny 1999 , abia & isern 2000 ) .
therefore , we also examined the @xmath0li vs. c / o trends for the nine cases presented above .
the models of frost ( 1997 ) use the algorithm for the third dredge - up ( i.e. the penetration of the convective envelope into the partially he - burnt zone after each thermal pulse ) , as described by frost & lattanzio ( 1996 ) , and include the entropy adjustment by wood ( 1981 ) . after each third dredge - up episode
the @xmath77c abundance is increased .
hence third dredge - up increases the ratio c / o ( by adding c to the stellar envelope ) , and hbb ( where it is active ) decreases it again as it transforms c into n. figure 3 shows the c / o ratio vs. @xmath0li for the nine models under consideration . in some cases hbb prevents the formation of a c star , while in other cases it merely delays the formation to a time when hbb has stopped and third dredge - up continues ( see frost et al .
1998 for more details ) .
the condition c / o @xmath75 together with high @xmath0li values are obtained only for the case @xmath78 0.004 and @xmath79 4 @xmath60 and for a very short phase for the case with the same metallicities and @xmath79 5 @xmath60 . in particular ,
the 4 @xmath60 model is a li - rich star ( with @xmath7 @xmath80 4 ) and c - rich star for a period of about 10@xmath81 yr .
moreover , all the other cases show high @xmath0li abundances together with o - rich envelopes , in agreement with observations of agb stars in the magellanic clouds ( see e.g. smith et al .
1995 ) and in the galaxy ( see e.g. the recent work from arellano ferro , giridhar , & mathias 2001 ) .
the range of @xmath82 covered by the nine models presented here ( @xmath83 ) during the li - rich phase is in agreement with bolometric magnitudes of the super - li rich stars observed in the magellanic clouds ( smith et al .
1995 ) . concerning galactic super - li rich c stars , only few estimates of their @xmath82 are available .
abia et al .
( 1991 ) , based on the distance determinations by claussen et al .
( 1987 ) , estimated @xmath84 for wzcas and wxcyg ( with @xmath7 @xmath85 and 4.7 , respectively ) .
more recently , abia & isern ( 2000 ) , based on _ hipparcos _ parallaxes , inferred @xmath866.44 and @xmath7 @xmath874.8 for wzcas , and @xmath884.35 and @xmath7 @xmath89 for wxcyg .
however , a caveat is that the latter is based on a parallax @xmath90 , which is unreliable .
two additional galactic c - rich super - li rich stars have been presented by abia et al .
( 1991 ) : iyhya ( @xmath91 and @xmath7 @xmath92 ) , and tsgr ( @xmath93 and @xmath7 @xmath94 ) .
given the uncertainties , and excluding the not reliable measurement of @xmath82 for wzcas by abia & isern ( 2000 ) , we conclude that the @xmath82 determinations for the galactic super - li rich stars are still consistent with the @xmath95 reached by our agb models during the super - li rich phase .
we also notice that the highest abundance reached by our models for @xmath7 is @xmath96 , while higher values are inferred from observations .
there are observational indications that toward the end of their evolution , agb stars ( both of low- and intermediate - mass ) can experience a short phase of extremely rapid mass - loss , called _ superwind _ ( van der veen , habing , & geballe 1989 ; schrder , winters , & sedlmayr 1999 ) . in this phase periods of heavy mass - loss
may be interspersed with much longer periods of lower mass - loss rates , possibly powered by he shell flashes .
these stars are often surrounded by a dense circumstellar envelope and some of them are no longer visible at optical wavelengths .
evidence for a superwind has been found in co observations of c stars ( e.g. olofsson 1993 ) , of k giants ( de la reza et al .
1997 ; castilho et al .
1998 ) , and in post - agb objects that show concentric rings around the central object ( see e.g. crabtree & rogers 1993 ; latter et al . 1993 ;
riera et al . 1995 ; klochkova et al . 1999 ; blcker et al .
2001 ) . from the theoretical point of view
, the term of superwind has been coined ( renzini 1981 ) to describe the heavy , final tip - agb mass - loss ( @xmath97 @xmath63 yr@xmath98 ) , which is required to form a planetary nebula of typically tenths of a solar mass within several 10@xmath13 yr .
the idea of a star terminating its agb life with a superwind has been also elaborated by bowen & willson ( 1991 ) .
these authors showed that all stars undergo the superwind phase as a result of an increase in scale height and density at the condensation radius as the star evolves toward the tip of the agb .
analytical expressions to describe this phase of high mass - loss have been presented by different authors , e.g. by vassiliadis & wood ( 1993 , hereafter vw93 ) , blcker ( 1995 ) , salasnich , bressan , & chiosi ( 1999 ) .
vw93 computed the effects of thermal pulses on mass - loss rates and suggested that a star may undergo several superwind phases and that more massive stars will stay longer in the superwind phase than lower mass stars .
for the agb models discussed in this section the superwind prescription follows the formula of vw93 ( without the correction for masses above 2.5 @xmath60 ) @xmath99 where the mass - loss rate is in @xmath63 yr@xmath98 and @xmath100 is the pulsation period in days .
note that the mass - loss rate is truncated at @xmath101 where @xmath102 is the stellar luminosity , corresponding to a radiation - pressure driven wind . again from vw93 we take the wind expansion velocity ( in km s@xmath98 ) @xmath103 ranging between 3.0 and 15.0 km s@xmath98 .
the @xmath0li yields are very sensitive not only on the extent of the hbb , but also on the mass - loss rate prescriptions .
as first qualitatively pointed by abia , isen , & canal ( 1993 ) , it is not clear if during the super - li rich phase agb stars are able to inject @xmath0li into the interstellar medium by the high mass - loss rate before @xmath0li is depleted in the atmosphere .
we analyze this point with our models , and we plot in figure 4 the surface @xmath0li abundance as a function of the mass - loss rate for the nine models described in the previous section .
the figure shows that only in a few cases does the star achieve very high mass - loss rates ( @xmath104 @xmath60 yr@xmath98 ) when the surface @xmath0li abundances is still high ( @xmath7 @xmath105 ) .
when this happens the overall @xmath0li yield is positive , or only slightly negative . in the first three columns of table 1 we report the @xmath0li yields ( in @xmath60 ) obtained with these models , showing that only in the case of 4 @xmath60 and @xmath70 the @xmath0li yield is positive .
table 1 + @xmath0li yields ( in solar masses ) from intermediate - mass agb stars [ cols="^,^,^,^,^,^,^",options="header " , ] ( _ a _ ) sb99 models .
+ ( _ b _ ) our extrapolations from ( _ a _ )
. + pop .
i low mass giant stars ( @xmath106 4 @xmath63 ) are thought to produce li both in the rgb and agb phases .
more specifically , low mass stars , below 2.5 @xmath63 , produce @xmath0li in the rgb and in the agb by a deep mixing process such as explored by wasserburg , boothroyd , & sackmann 1995 , and sackmann & boothroyd 1999 , hereafter sb99 .
this process is sometimes called cool bottom processing . we prefer to refer to this as `` deep mixing '' because it is more indicative of the physics involved : the cool bottom of the envelope does not play any role .
the nuclear processing occurs at the top of the h - shell and this is facilitated by mixing from the bottom of the convective envelope down to the deeper layers of the h - shell .
this mechanism needs to be further explored to see if it can be efficient also in the mass range 2.5 @xmath107 4.0 . according to stellar evolution models ,
rgb stars should be characterized by a relatively low @xmath0li content : stars that leave the ms undepleted in @xmath0li , after the first - dredge up dilution has occurred , are expected to have @xmath0li abundances of the order of @xmath7 @xmath108 ( iben 1965 , 1967a , b ) .
since most stars destroy @xmath0li on the ms , their @xmath0li content when they reach the rgb and in immediately subsequent phases should be much lower than the above values .
most field and cluster rgb stars show in fact very low @xmath0li abundances ( @xmath109 @xmath7 @xmath110 ) , even lower than predicted by the models .
an extra mixing mechanism has been proposed to explain those low abundances ( see e.g. charbonnel , brown , & wallerstein 1998 ; charbonnel & balachandran 2000 , and references therein ) .
however , after the first discovery by wallerstein & sneden ( 1982 ) of a li - rich rgb ( i.e. a rgb star with a @xmath0li abundance higher than model predictions ) , several other li - rich giants have been found ( for recent surveys , see castilho et al .
1998 ; jasniewicz et al .
note that , although the number of presently known li - rich rgb stars is relatively high , they represent only 12 % of all the giants with @xmath0li measurements ( wallerstein & sneden 1982 ; gratton & dantona 1989 ; pilachowski , hudek , & sneden 1990 ; pallavicini et al .
1990 ; fekel & balachandran 1993 ) .
some of these li - rich rgb stars have abundances even higher than the present ism value ( de la reza & da silva 1995 ; balachandran 2000 ) .
various suggestions have been made to explain the li - rich giants with @xmath7 @xmath111 ( for stars with @xmath112 @xmath7 @xmath113 a fresh @xmath0li production is not necessary since their @xmath0li abundance is consistent with the first dredge - up values ) .
the high @xmath0li content of these giants may be related to external processes ( alexander 1967 ; gratton & dantona 1989 ; siess & livio 1999 ) or to internal processes such as the production of fresh @xmath0li ( fekel & balachandran 1993 ; de la reza , drake , & da silva 1996 ; sb99 ) . as shown by sb99
, @xmath0li can be created in low - mass rgb stars via the cameron - fowler mechanism associated with a deep - mixing below the convective envelope .
this internal circulation ( possibly driven by stellar rotation ) transports envelope material into the outer wing of the h - burning shell , where it undergoes partial nuclear processing , and then back to the envelope .
sb99 choosed the free parameters of their model to match the low value of @xmath77c/@xmath114c observed in rgb stars .
in order to produce the required additional @xmath114c , the advected material must reach temperatures high enough that @xmath53he is burned , resulting in the creation of @xmath0be via @xmath53he(@xmath11,@xmath115)@xmath0be .
if extra - mixing is slow , @xmath0be is destroyed via @xmath0be(@xmath116,@xmath115)@xmath117b(@xmath118,@xmath119)@xmath117be or @xmath0b(@xmath120,@xmath119)@xmath0li , and any @xmath0li produced from @xmath0be electron captures is immediately burned up via @xmath0li(@xmath116,@xmath11)@xmath13he .
however , for higher mixing speed @xmath0be can be transported out to cooler regions before electron capture takes place and the stellar envelope becomes enriched in @xmath0li .
the production of @xmath0li in rgb stars is dependent on the mixing - speed , i.e. the stream mass flow rate .
sb99 discussed the range of values of this mixing - speed and argued that it must be slower than the velocity of convection in rgb or agb ( @xmath80 1 @xmath60 yr@xmath98 ) , while the streams must move faster than the speed with which the h - shell burns its way outward .
we show in table 3 how the different values for the mixing - speed can influence the @xmath0li production ( see also sb99 for more details on the 1 @xmath60 model ) .
it is important to notice that , as discussed by e.g. charbonnel ( 1994 ) and boothroyd & sackmann ( 1999 ) , the mass range in which the deep mixing is active is @xmath80 1.02.5 @xmath60 .
this is due to the fact that for low mass stars ( @xmath121 @xmath60 ) , the h - burning shell catches up to and erases the discontinuity while the star is still on the rgb ; for higher masses ( @xmath122 2.5 @xmath60 ) the star leaves the rgb before this can take place .
we use the models by sb99 to estimate the contribution of low - mass rgb stars to the chemical evolution of the galaxy .
the @xmath0li production phase on the rgb is short ( @xmath71 yr ) compared to the total red giant lifetime ( @xmath123 yr ) , and since the typical rgb mass - loss rate is rather low , they are not expected to contribute significantly to the @xmath0li enrichment in the ism . on the other hand , during the tp - agb phase , because of the higher mass - loss rate , the contribution to the @xmath0li ism enrichment can be substantial .
for these reasons , and since sb99 do not present specific @xmath0li predictions for the agb phase , in this work we make the following assumptions .
we consider the sb99 @xmath0li predictions at the end of the rgb phase ( when @xmath124 ) for 1 @xmath60 at two different metallicities ( @xmath125 and @xmath126 ) and for three different cases of mixing - speed ( @xmath127 , @xmath128 , @xmath129 @xmath60 yr@xmath98 ) . to obtain the yield of @xmath0li from these models we multiply the @xmath0li abundances by the total mass ejected , e.g. for 1 @xmath60 model we used @xmath130 @xmath60 ( see e.g. weidemann 1984 for the initial - final mass relationship ) .
we assume a mass range of 12.5 @xmath60 , in agreement with the above discussion , and since the authors showed only the results for the 1 @xmath60 model , we interpolate the @xmath0li production by deep mixing in the mass range 1 @xmath60 2.5 @xmath60 . to derive the 2.5 @xmath60 @xmath0li yields we follow sb99 assuming that all @xmath53he is converted in @xmath0li , and that the abundance of @xmath53he in stars scales as @xmath131 ( schatzman 1987 ) .
the @xmath0li yields obtained for 1 @xmath60 and 2.5 @xmath60 , with different metallicities and different mixing speed , are shown in table 3 . in our calculations , in order to derive an upper limit to the @xmath0li production , we adopt the highest @xmath0li abundance predicted by the sb99 models , i.e. that obtained with the model with 10@xmath132 @xmath60 yr@xmath98 mixing speed . the contribution of low - mass giant stars to the galactic evolution of @xmath0li is shown in figure 8 .
these stars enrich @xmath0li up to @xmath133 at the epoch of formation of the sun , i.e. @xmath134% of the solar value .
again , another more efficient source is required to explain the ism @xmath0li abundance . as in the case of snii ( fig .
7 ) we show for comparison the contribution from these stars to @xmath0li enrichment starting with a zero initial @xmath0li composition .
we have addressed the long - standing problem of the galactic chemical evolution of @xmath0li with the aim of clarifying the role of the different stellar contributions . we have analyzed four possible stellar sources of @xmath0li : snii , novae , low - mass giants and ims - agb stars . for low - mass giants , novae , and snii we have critically examined the available @xmath0li yields and discussed the possible extrapolations when no model predictions were available . in the case of snii and novae
we re - examined the @xmath0li yields in the light of recent hydrodynamical computations . for ims - agb stars we presented and used here new results for nucleosynthesis calculations based on evolutionary agb models by frost ( 1997 ) . in particular
, we discussed how the interplay between the hbb process and a phase of high mass - loss before the evolution off the agb may constitute a key process for @xmath0li enrichment in the galaxy . although we are not yet able to quantify the contribution of ims - agb stars , we have explored different realistic possibilities in terms of mass - loss rate and mass range for the hbb process .
the work presented here is summarized in figure 9 , where we compare the results of our model of chemical evolution with the sample of observational data presented in fig . 1 .
the predicted @xmath0li abundance resulting from all stellar sources considered in this paper is plotted in fig .
9 vs. metallicity ( upper panel ) and vs. time ( lower panel ) . for ims - agb we used a vw93 mass - loss
increased by a factor of 50 ( as discussed in
4.2 ) , and we have also taken into account the results obtained for the 3 @xmath60 agb model presented in
4.3 . as we discussed in sect . 2.1
, we also ran our model with an initial @xmath0li abundance higher by a factor 2 than the spite plateau value .
the result is shown for comparison in fig .
notice that the predicted @xmath0li abundance at the time of formation of the sun is virtually the same in both cases . in the lower panel of fig .
9 we plot the individual contributions from the different stellar sources , with the exception of the contribution from snii since we believe that the inclusion of rotation and mixing in the supernova models leads to a drastic reduction of the @xmath0li yields ( see 6 ) .
note that , from an observational point of view , it is critical to determine the metallicity where the rise of the @xmath0li abundance from the spite plateau occurs ; the available data indicate that the rise from the plateau occurs at metallicities [ fe / h ] between @xmath135 and @xmath136 , but additional data in the metallicity range @xmath137 [ fe / h ] @xmath138 are needed in order to better constrain this value .
( _ i _ ) in the light of the available stellar models , small contributions to the meteoritic @xmath0li abundance come from novae ( @xmath139% or less , see 5 ) and low - mass giant stars ( @xmath140% ,
see 7 ) . as for snii
, we predict a contribution to the solar system @xmath0li abundance of @xmath141% with the standard yields by woosley & weaver ( 1995 ) , and a contribution less than 10% if the results of the latest hydrodynamical simulations of the supernova explosion are taken into account ( see 6 ) .
spallation reactions between ism and cosmic - ray nuclei can provide an additional at least @xmath420% of the @xmath0li abundance in the solar system ( see e.g. lemoine et al .
1998 ) .
( _ ii _ ) figure 9 shows that even with our best - fit model we are not able to reproduce the meteoritic abundance , although both cluster and ism abundances are fitted fairly well .
our best - fit model is in good agreement with the observed @xmath7 vs. [ fe / h ] distribution up to [ fe / h ] @xmath142 .
this indicates that the metallicity at which the rise from the spite plateau occurs is consistent with the [ fe / h ] values at which the ims - agb stars contribute to the chemical enrichment of the ism .
we remind that the observational data for [ fe / h ] @xmath143 are mostly taken from balachandran ( 1990 ) ( see discussion in
2 ) ; nevertheless , as the comparison between photometric and spectroscopic metallicities shows , part of these stars might have higher [ fe / h ] values . ( _ iii _ ) our best - fit model and ,
more specifically , the contribution from ims - agb , is based on two major assumptions : first the high mass - loss phase at the end of the evolution of these stars must start _ earlier _ on the agb than the standard predictions by vw93 .
this is supported by recent ir observations , discussed in 4 .
second , we also need a contribution to @xmath0li through hbb production from stars with 3 @xmath60 @xmath144 @xmath60 and low metallicities ( see preliminary model results for @xmath145 @xmath60 in lattanzio et al .
should these two assumptions be not valid , the contribution from intermediate - mass agb would be much lower .
we stress however that , since neither novae , nor snii , nor low - mass giant stars seem to produce enough @xmath0li to account for the present - day abundance , we believe that intermediate - mass agb stars remain at present the best candidates as @xmath0li factories . our results and conclusions are strongly based on the agb models discussed in 4 ( see also frost 1997 ) .
a different view has been recently expressed by ventura et al .
they presented a grid of ims - agb models of different masses and tested the sensitivity of @xmath0li yields to different mass - loss rate prescriptions on the basis of a comparison with agb stars in the magellanic clouds .
ventura et al . ( 2000 ) concluded that , with their mass - loss calibration , ims - agb stars do not contribute significantly to the @xmath0li enrichment of the ism .
as they properly noticed , details of @xmath0li production depend on the input parameters of the stellar model , mainly the treatment of convection .
in addition , ventura et al . ( 2000 ) focused their work on the analysis of the strength of the mass - loss rate during the agb phase .
we instead followed a different approach , testing different times during the agb phase at which the high mass - loss rate starts , and looking for the consequences on @xmath0li yields . as we demonstrated in 4 ,
anticipating by few thermal pulses the beginning of the superwind phase can have significant impact in the mass of @xmath0li ejected .
work in progress includes an extension of the grid of agb models to lower masses ( @xmath146 @xmath60 ) and lower metallicities ( @xmath147 ) ( see preliminary results in lattanzio et al .
2001 ) , in order to analyze the possibility of @xmath0li production via hbb in these stars .
another interesting point that we just mentioned in this paper , is a re - analysis of the contribution of gcr to the production of @xmath0li ( valle et al .
, in preparation ) .
finally , new @xmath0li data for a statistically significant sample of stars in the critical metallicity range @xmath148 [ fe / h ] @xmath149 are being analyzed ; the data will provide stringent observational constraints on the galactic @xmath0li abundance in a metallicity range where ims - agb stars are expected , on the basis of this work , to give their main contribution .
we thank s. shore for interesting discussions and comments on the manuscript .
also thanks r. gallino for useful suggestions and encouragements during the development of this work .
we also thanks the anonymous referee for very usefull comments that improved our paper .
the research of c.t .
, s.r . and d.g .
is partially supported by grants cofin98 and cofin2000 . | in this paper we study the evolution of @xmath0li in the galaxy considering the contributions of various stellar sources : type ii supernovae , novae , red giant stars , and asymptotic giant branch ( agb ) stars .
we present new results for the production of @xmath0li in agb stars via the hot bottom burning process , based on stellar evolutionary models by frost ( 1997 ) . in the light of recent observations of dense circumstellar shells around evolved stars in the galaxy and in the magellanic clouds , we also consider the impact of a very high mass - loss rate episode ( superwind ) before the evolution off the agb phase on the @xmath0li enrichment in the interstellar medium .
we compare the galactic evolution of @xmath0li obtained with these new @xmath0li yields ( complemented with a critical re - analysis of the role of supernovae , novae and giant stars ) with a selected compilation of spectroscopic observations including halo and disk field stars as well as young stellar clusters .
we conclude that even allowing for the large uncertainties in the theoretical calculation of mass - loss rates at the end of the agb phase , the superwind phase has a significant effect on the @xmath0li enrichment of the galaxy . |
after building lep and lhc , and planning ilc the size and cost of classical high energy particle accelerators are reaching their limits around 50 mev / m . in their historical article @xcite
in 1979 t. tajima and j. m. dawson predicted : `` glass lasers of power density @xmath0 shone on plasmas of densities @xmath1 can yield gigaelectronvolts of electron energy per centimeter of acceleration distance . '' with the availability of high power short pulse lasers their ideas are becoming reality .
it turned out that it is possible to build not only laser wake field accelarator ( lwfa ) but also particel driven wake field accelaratior ( pwfa ) . in the laboratories 100
gv / m accelerating gradient was demonstrated with laser driven systems @xcite and 50 gv / m with electron driven case in slac @xcite . at cern
the awake collaboration @xcite has been formed in order to demonstrate proton driven plasma wake field acceleration for the first time , where the wigner team is interested in the creation of high uniformity plasma which is required to reach the plasma frequency stability at the percent level .
the sps proton beam in the cngs facility will be injected into a 10 m plasma cell where the long proton bunches will be modulated into significantly shorter micro - bunches .
these micro - bunches will then initiate a strong wakefield in the plasma with peak fields above 1 gv / m .
though this peak field is much less than the one achieved by laser or electron driven systems , the future accelerator based on this technology can have a much reduced length compared to proposed linear accelerators . in the proposal @xcite a uniform plasma
was assumed to be created by high enough laser power with over - barrier - ionization ( obi ) , achieving saturation along the whole length of the 10 meter plasma . in this paper
we propose a much softer method by applying the so - called resonance enhanced multi photon ionization ( rempi ) scheme .
it is a three photon process which requires significantly lower laser power than the non - resonant obi process . from this three photon ionization process
, we investigate the two - photon resonant excitation process with theoretical and experimental method .
in order to produce a desirable homogeneous laser plasma for the wake field acceleration of particles the ionization process by high intensity laser pulses has to be studied .
therefore the aim of the first experiment is the investigation of high field ionization gain from rubidium atomic beam by using the commercial regenerative ti : sapphire based chirped - pulse amplifier laser system ( coherent legend elite he - usp ) .
the amplifier delivers 35 fs pulses at 800 nm central wavelength with 1 khz repetition rate and with 4 mj pulse energy . the characteristic noise ratio of the average power is smaller than 0.5 % .
the maximum achievable focused peak intensity is around @xmath2 depending on the focusing optical element .
( the applied seed oscillator for the amplifier is a standard home - build ti : sapphire oscillator delivering 20 fs long pulses with 80 mhz repetition rate and with 7 nj pulse energy . )
the intensity dependence and the presumable saturation of the ionization probability of the free atoms will be measured by detecting the ions and/or electrons .
the experimental setup can be seen in fig .
1 . in order to prevent the effects of plasma screening ,
positive ions are detected throughout the experiments .
single ionized @xmath3 ions are collected by the negatively charged cathode of the mcp .
it is important that the interaction range must be isolated from all other constant voltages to prevent escaping of the ions . alternatively to ion detection
, electrons can be detected as well . in this case
a positive charge can be applied to the electrode which is grounded in fig . 1 in order to check that the atoms are single ionized only and also to measure the density , simultaneous detection of electrons at the position of the grounded electrode is possible as well . in this case
the mcp is connected so that even its cathode is positively charged whereas the anode is connected to -2.7 kv higher voltage . using this arrangement in the case of low densities ( in which case
the interaction range is within the debye length of the plasma ) equal number of ions and electrons are expected in the case of single ionization . for higher rb densities
the difference of ion - caused current and electron - caused current gives a possibility to determine the atomic density .
the electrons are produced by the 35 fs laser pulses in the high field ionization of the rb atomic beam of slab like shape from the electrically heated disperser and accelerated by an electrical field of 1000 v on the electrodes a and b into the direction of the cathode ( d ) of a double chevron michrochannel plate ( mcp ) trough the drift tube ( c ) .
the resulting impulse is observed on the anode ( e ) of the mcp by an oscilloscope . a simple rubidium disperser manufactured by ` saes getters ' serves as the first version of the source of the atomic beam ( see fig .
2 ) . it contains a suitable compound ( chromate ) of rubidium which , under appropriate heating by current , emits neutral rb atoms , while the rest of the components are absorbed by the reducing agent of the dispenser , thus producing a clean source of rb atoms .
alkali metal dispensers are commonly used as controlled sources of rubidium atoms e.g. for loading magneto - optical traps in quantum optics experiments , e.g. @xcite , and several constructions have been developed to produce directed beams from metal dispensers in this field @xcite .
the shape of the dispenser ( a ) is slab like of about 30 times 1 mm .
the metallic slab containing the rubidium compound is heated by electric current .
the rubidium atoms released after the dissociation of the compound disperse from the source of temperature of some hundred degrees and distributed in almost @xmath4 solid angle . for the aim of the experiments the atoms restricted to a narrow solid angle by two slit apertures positioned to about one centimeter distance .
the construction of the rubidium source box can be seen in figure 2 ( except for the side openings which are not indicated here ) .
the dispenser and one of the two slit apertures placed in a closed metallic box ( b ) the top of which contains the second aperture ( c ) .
the two slits collimate the emission of the atoms in a narrow region of the vacuum space .
the side walls of the two regions of the box have openings in order to avoid the buildup of dense gas phases , atoms emitted by the dispenser wire directly towards the slits can only be emitted .
closely spaced cover plates gather the atoms escaping the box through the side openings .
the advantage of using disperser as the source of atomic beam is its flexibility in switching on or off securing its economic long time use .
we have produced well collimated atomic emission profiles with this source with low background densities of rubidium .
the atomic density was monitored by measuring the fluorescence excited by resonant laser radiation .
the beam of an external cavity diode laser , frequency stabilized close to the strongest resonance line of the 85rb atoms , was sent through the atomic beam parallel to the slits , and the fluorescence intensity was measured at different positions of the beam .
the accumulated profile of the fluorescence can be seen in figure 3 , and a full width at half maximum of about 1.8 mm has been obtained a few mm above the source surface . this experimental arrangement placed into a stainless steel vacuum chamber with the ports for entry and exit of the laser pulses , for the optical observations and for the electrical feed troughs .
the vessel is pumped by turbo molecular pump to better than @xmath5 torr vacuum .
the field ionization measurements , the next step are now under way .
when considering generation of laser plasma with high homogeneity using multi - photon or over - the - barrier ionization ( obi ) , it is advantageous to find ways for decreasing the threshold conditions for the ionization , especially when generation of a spatially extended plasma is the main goal .
one of the ways to decrease the threshold of the laser multi - photon ionization ( mpi ) or obi is to transfer atomic population from its ground state to one of the excited states before the ionization process , thus significantly decreasing the ionization potential .
of course , such pre - excitation of the atoms must be performed by methods that are robust and are not too sensitive to the laser pulse parameters including duration , fluctuations in pulse intensity , phase , etc .
one of the most efficient ways to transfer the atomic population to a target state with probability close to unity , is to utilize the method of rapid adiabatic passage ( rap ) .
the word rapid relates to the interaction time that must be shorter than all relaxation times in the system . on the other hand ,
the variation of the laser parameters must be adiabatically slow to maintain the atom in a given dressed state , @xcite .
an approximate condition foradiabaticity is @xmath6 , where @xmath7 is the peak rabi frequency , and @xmath8 is the duration of the laser pulse . while the relaxation times in gases are in the nano and pico - second time range , the relaxation processes may be avoided already for laser pulses in the femto- ( tens of femto- ) second durations range .
rap may be efficiently realized using phase - modulated ( frequency - chirped ( fc ) ) laser pulses , which frequency varies monotonically in time during the laser pulse .
an instructive example of the population transfer produced by fc pulse is the action of such a pulse on a model two - level atom .
while action of a laser pulse with constant frequency results in the well - known rabi oscillations with in average 50 % population transfer , all the atomic population is transferred from one state to another one when a fc laser pulse is applied , the frequency of which is swept through the resonance with the atomic transition .
an important feature of the interaction is its robustness concerning variations of the laser parameters @xcite .
the fc pulses were successfully utilized also in more complicated multilevel atomic or molecular structures @xcite . in ref .
@xcite , for example , fc pulses from a diode laser having a central frequency resonant with the d2 line of rb85 produced a complete population transfer from a ground state to manifold of excited states and back , each time transferring a twice of photon momentum to the atom by a pair of counter - propagating pulses @xcite .
the fc pulses were used in @xcite to perform resonance enhanced multi - photon ionization in sodium atoms . in this communication
, we show on the example of rubidium atom , see fig.4 , that pre - excitation of the multilevel atom may be efficiently and robustly performed by application of fc laser pulses .
as it is seen from fig.4 , absorption of three photons of radiation of 800 nm wavelength ( typical for the ti : sa laser ) provides ionization of rb atom from the ground state 5s .
the efficiency of the ionization process may be increased in some extent using intermediate resonances providing resonance enhanced multi - photon ionization , ( see e.g. @xcite ) .
however , the threshold of the ionization processes may be drastically decreased if one could transfer the atomic population from the ground level 5s to excited states 5d or 7s , from which only one - photon transition to the vacuum - state provides ionization of the atom .
we have analyzed the possibilities of excitation of the rb atom by laser pulses having a gaussian shape with duration equal to 30 fsec , peak intensity in the range of a few @xmath9 with a linear frequency chirp having speed in the range of a few fs@xmath10 . since the duration of the laser pulse is much shorter than the atomic relaxation times , the schrdinger equation for the probability amplitudes of the working levels of rb atom in fig.4 ,
is solved numerically and the dynamics of the atomic populations in the field of the fc pulses is determined , see fig.5 . as it is clearly seen in fig.5 , a complete population transfer from the ground state 5s to the excited states 5d and 7s
is performed by a fc laser pulse .
the robustness of the excitation process is well seen from the color map in fig.6 : the population of the state 5d after the interaction with the laser pulse remains nearly the same value ( close to unity ) in a broad range of variation of the peak rabi frequency and the speed of the positive chirp . in conclusion
, we have shown that it is possible to efficiently transfer all the population of rb atom from its ground state to the excited states using frequency chirped laser pulses .
the presented results of the numerical simulations demonstrate high robustness of the excitation process against variations of the laser pulse parameters that may play a crucial role in generation of a highly homogeneous spatially extended laser plasma
. in the same awake collaboration one of us ( s. varr ) investigated a different question , namely the dirac equation of a charged particle interacting with an electromagnetic wave in a medium and found a new exact solution @xcite .
this problem is significant for the final realization of the awake experiment . | the key element in the proton - driven - plasma - wake - field - accelerator ( awake ) project is the generation of highly uniform plasma from rubidium vapor .
the standard way to achieve full ionization is to use high power laser which can assure the over - barrier - ionization ( obi ) along the 10 meters long active region .
the wigner - team in budapest is investigating an alternative way of uniform plasma generation .
the proposed resonance enhanced multi photon ionization ( rempi ) scheme probably can be realized by much less laser power . in the following the resonant pre - excitations of the rb atoms are investigated , theoretically and the status report about the preparatory work on the experiment are presented .
proton - driven - plasma - wake - field - accelerator , resonance enhanced multi photon ionization |
the time evolution of the gravitational collapse of perturbations in decelerating , isothermal shocked layers has been examined numerically and analytically using linearized equations by elmegreen ( 1994 ) . in this paper
, we continue with nonlinear analysis of the gravitational instability of spherically symmetric shells expanding into the stationary homogeneous medium .
we modify the approach adopted by fuchs ( 1996 ) who described the fragmentation of uniformly rotating self - gravitating disks .
the inclusion of higher order terms helps to determine with better accuracy when , where and how quickly the fragmentation happens .
formation rates and the mass distribution function of fragments are derived from simulations and compared to the observed mass spectrum of molecular clouds .
equations as derived by w " unsch & palou ( 2001 ) are @xmath0 @xmath1 @xmath2 where @xmath3 is the expansion velocity of the shell , @xmath4 is mass in the area @xmath5 on the surface of the shell , and @xmath6 denotes a two dimensional velocity of surface flows .
@xmath7 is the surface density , @xmath8 is the constant isothermal sound speed inside the cold shell , @xmath9 is the gravitational potential generated by the mass distribution in the shell , g is the gravitational constant and @xmath10 is a delta function of the space coordinate @xmath11 perpendicular to the surface of the shell . the solution of linearized eq .
( 1 - 3 ) may be written in the form of exponential functions for the perturbations : @xmath12 ( perturbed surface density ) and @xmath13 . the shell starts to be gravitationally unstable when @xmath14 is purely imaginary and negative .
@xmath15 denotes the radius and @xmath16 the unperturbed surface density of the shell .
@xmath17 is the angular frequency , in the case of instability @xmath18 gives the perturbation growth rate , and @xmath19 the wave number of perturbations .
the time evolution of @xmath16 and of @xmath20 , is shown in fig .
1 for shell expanding in the homogeneous medium of the constant volume density 1 @xmath21 and average atomic weight 1.3 .
the total energy released is @xmath22 erg and @xmath8 = 1 km s@xmath23 .
w " unsch & palou ( 2001 ) include quadratic terms into equations for the evolution of perturbations . the time evolution of the perturbed maximum surface density including nonlinear terms , @xmath24 , is given in fig .
1 . as another result of the nonlinear analysis fig . 2 shows the spatial distribution of the surface density and velocity vectors of the surface flows resulting from mode interactions .
after the time when the shell becomes unstable , the mass of the fragment decreases as its size decreases . later
, the inflow of mass begins to dominate ( surface flows are shown in fig .
2 ) and the fragment mass grows .
masses of individual fragments are between a few times @xmath25 and @xmath26 with the highest frequency around @xmath27 .
another consequence of the dispersion relation ( 4 ) is that the most massive fragments , corresponding to the smallest unstable wave number @xmath19 , form very slowly .
therefore , the most massive clouds are created at the latest stages of the shell evolution .
the mass spectrum of fragments is shown in fig . 1 .
its decreasing part can be approximated with a power law @xmath28 where @xmath29 .
this is quite close to the observed mass spectrum of gmc in the milky way .
combes ( 1991 ) gives @xmath30 .
nanten survey of the co emission of the lmc ( fukui , 2001 ) gives steeper slope : @xmath31 , which may be connected to higher level of random motions in the lmc compared to the milky way , which restricts the formation of late time massive fragments in the lmc and steepens their mass spectrum .
the authors gratefully acknowledge financial support by the grant agency of the academy of sciences of the czech republic under the grant no . a 3003705 and support by the grant project of the academy of sciences of the czech republic no . k1048102 . | the gravitational instability of expanding shells is discussed . linear and nonlinear terms
are included in an analytical solution in the static and homogeneous medium .
we discuss the interaction of modes and give the time needed for fragmentation .
masses of individual fragments are also estimated and their formation rates and the initial mass function are derived .
results of simulations are compared to observation . |
the spectral fluctuations in a rich variety of different physical systems show , if measured on the scale of the local mean level spacing , very similar features .
this high degree of universality makes it possible to describe these fluctuations with random matrices .
random matrix theory ( rmt ) is a simple , schematic model in which the matrix elements of the hamiltonian in some basis are replaced with random numbers .
apart from randomness , the only further input are the symmetries and invariances of the system , in particular time reversal invariance .
it turns out that this assumption of full ergodicity or `` chaoticity '' leads , in many cases , to a complete and parameter free description of the spectral fluctuations .
such universal cases are said to be of wigner dyson type .
we refer the reader to the reviews in refs .
originally , wigner had developed this approach in nuclear physics where it continues to find new applications . in recent years
, interest has been focussed on deviations from the universal , parameter free wigner
dyson fluctuations .
the deviations can have different reasons , such as regular effects competing with full chaoticity or the breaking of time reversal invariance , see ref .
@xcite . here
, we wish to discuss deviations attributed to the breaking of isospin symmetry .
we recall that symmetries such as isospin or parity are , in contrast to time reversal invariance , associated with quantum numbers . in the context of spectral fluctuations and rmt ,
symmetry breaking already received broad interest , see the compilation in ref .
as early as 1960 , rosenzweig and porter @xcite analyzed atomic spectra by setting up a general random matrix model which describes crossover transitions between different angular momentum coupling schemes . in the late eighties , mitchell et al .
@xcite measured and analyzed about 100 low lying states with known values of the isospin quantum number in the nucleus @xmath0al . in ref .
@xcite , this was discussed in the framework of a random matrix model which is a special case of the rosenzweig porter model .
an estimate for the statistical coulomb matrix element , i.e. a measure for the degree of isospin breaking , could be obtained .
motivated by similar questions in molecular physics , leitner and coworkers @xcite performed a perturbative calculation of the spectral fluctuations in the random matrix model .
more recently , new data were obtained in two statistically highly significant experiments on the breaking of a point group symmetry in a resonating quartz block @xcite , and the coupling of two chaotic microwave billiards @xcite . both cases , although physically very different , are statistically fully equivalent to symmetry breaking in quantum mechanics .
importantly , there is only one parameter entering the random matrix model .
it is a unique measure e.g. for the root mean square statistical coulomb matrix element .
this illustrates that the random matrix model is the ideal tool to extract a root mean square symmetry breaking matrix element from the data .
all these studies addressed the _ spectral _ fluctuation properties .
symmetry breaking , however , will also have an impact on the statistics of the _ wave functions _ and observables sensitive to them .
recently , adams , mitchell and shriner @xcite collected reduced @xmath1 ray transition probabilities from different experiments on @xmath0al . as mentioned above ,
this nucleus had already shown a strong deviation of the _ spectral _ fluctuations from the universal wigner
dyson result due to isospin breaking @xcite .
the new results @xcite show that the distribution of the _ transition _ probabilities also considerably deviates from the wigner
dyson statistics , i.e. from the distribution that corresponds to full , parameter free `` chaos '' . in the present contribution
, we wish to discuss these results . to this end , we extend the random matrix model of ref .
@xcite to discuss transition probabilities .
similar investigations were performed simultaneously and independently by andersen , ellegaard , jackson and schaadt @xcite for symmetry breaking in acoustic and elastomechanical systems .
the article is organized as follows . in sec .
[ ii ] we briefly review the experimental results on the reduced transition probabilities in @xmath0al collected by adams , mitchell and shriner @xcite . in sec .
[ iii ] , we discuss the random matrix model in the case that no symmetry is present , in particular its predictions for transition probabilities . the random matrix model for isospin breaking
is numerically studied in sec .
[ iv ] . in sec .
[ v ] , the numerical simulation is approximated by a qualitative albeit analytical model .
the data analysis is performed in sec .
summary and conclusion are presented in sec .
experimental reduced electromagnetic transition strengths between the excited states of the nucleus @xmath0al have been collected by adams et al .
see ref .
their data involve levels between the ground state and the excitation energy of 8.067 mev . in this region , states with isospin @xmath2 and @xmath3 are found and isospin is known to be approximately conserved .
the probability @xmath4 of a transition from the initial configuration @xmath5 to the final configuration @xmath6 is the square @xmath7 of the matrix element @xmath8 of the relevant transition operator @xmath9 in a special basis .
approximately 180 levels and 1500 electromagnetic transitions are known .
the fluctuations of the @xmath4 values shall be studied .
to this end , their systematic dependence on the quantum numbers of the initial and final states must be removed .
this has been done in ref .
@xcite in the following way .
the states in @xmath0al have been characterized by their excitation energy @xmath10 , spin @xmath11 , parity @xmath12 , and isospin @xmath13 .
transitions are characterized by their electromagnetic character @xmath14 which may be @xmath15 or @xmath16 , their multipolarity @xmath17 and a label @xmath18 which becomes isoscalar @xmath19 if @xmath20 and isovectorial @xmath21 if @xmath22 .
hence , both the transition operator @xmath23 and the transition probabilities @xmath24 are functions of @xmath14 , @xmath17 and @xmath18 .
we shall , however , not always write all these arguments
. a _ transition sequence _ is defined as a set of reduced transition probabilities where the initial states have a common assignment @xmath25 , @xmath13 , the final states have a common assignment @xmath26 , @xmath27 , and the transitions have their three defining characteristics all the same .
thus the reduced transition probabilities @xmath28 of a given transition sequence can be labeled by the energies @xmath29 and @xmath30 of the initial and final states , respectively .
the aforementioned secular variations of the @xmath31 values were removed by normalizing them to the local average value @xmath32 of @xmath33 so that the statistical variable used further on is @xmath34 the local average is defined by help of weighting factors that are gaussian functions of the excitation energies . in doing so one must , however , remove the systematic dependence of the local level distance @xmath35 on the excitation energy ; i.e. one measures the energy in units of @xmath35 and works with a dimensionless energy @xmath36 the details are given in ref .
the local average of the @xmath31 values of a given transition sequence is then @xmath37 the fact that the variance of the gaussian functions has been chosen equal to 4 is discussed in ref .
@xcite .
it is clear that the definition of the quantities @xmath38 requires the necessary spectroscopic information for the relevant states in @xmath0al .
furthermore , the local averages @xmath35 and @xmath39 require that there is a minimum number of members in the sequences from which these averages are derived .
these restrictions have finally led in ref .
@xcite to an ensemble of 873 values of @xmath40 .
the authors of ref .
@xcite have found it convenient to transform @xmath40 to the logarithmic variable @xmath41 the experimental distribution of @xmath42 is given by the histogram on fig .
[ fig26al ] .
although the widths @xmath43 of the bins @xmath44 in fig . [ fig26al ] vary , the ordinate is the probability density with respect to @xmath42 : the height @xmath45 of the @xmath46-th bin is @xmath47 where @xmath48 is the number of cases falling into the @xmath49th bin . since the states @xmath50 and @xmath51 are believed to be very complicated " , it is natural to assume that the matrix elements ( [ eqii.4 ] ) have a gaussian distribution ; see the discussion in sec .
this entails that @xmath40 has a porter
thomas distribution with unit average value , i.e. @xmath52 transforming this to the variable @xmath42 yields @xmath53 a comment on the notation is in order : for every probability density we shall use the symbol @xmath54 .
the fact that we deal with different functions will be clearly indicated by the argument of the probability density in question . the distribution ( [ eqii.6 ] ) is given on fig .
[ fig26al ] for comparison with the data .
the distribution ( [ eqii.6 ] ) is normalized to unity ; the histogram of the data , however , has been normalized to @xmath55 by the authors of ref .
this is motivated @xcite by the fact that @xmath56 over the range of the data amounts to 0.83 and approximately takes care of the upper and lower detection thresholds .
the point of ref .
@xcite is the shift of the experimental distribution with respect to the porter
thomas distribution @xmath57 .
the maximum of the histogram occurs around @xmath58 .
the maximum of @xmath59 occurs at @xmath60 .
the authors of ref .
@xcite have conjectured that the discrepancy between the experimental data and the porter
thomas distribution may be a consequence of isospin breaking in @xmath0al .
the following sections of the present paper focus on this conjecture . in the next section , a random matrix model of isospin violation
and more generally of symmetry breaking is presented .
in sec . [ iiia ] , we briefly review the distributions of wave functions and widths for the standard wigner dyson case in which no symmetry is present .
a more detailed presentation with further references can be found in ref .
@xcite . in sec .
[ iiib ] , we work out the distribution of the transition matrix elements for the same case . if the system is invariant under time reversal , the wave functions can be chosen real and the @xmath61 random matrix @xmath62 modeling the hamiltonian is real and symmetric .
the matrix elements are gaussian distributed random numbers and @xmath62 is said to be in the gaussian orthogonal ensemble ( goe ) . in the physically relevant limit of infinitely many states , @xmath63 , the fluctuation properties are of wigner dyson type .
a wave function is modeled by an eigenvector @xmath64 of @xmath62 , i.e. we have @xmath65 where @xmath66 is the eigenenergy .
we are interested in the probability density @xmath67 of finding the value @xmath68 with @xmath69 for a component @xmath70 of the eigenvector @xmath71 . for finite @xmath72
one finds @xmath73 the second moment of this distribution is @xmath74 . for a large number of levels one obtains a gaussian with variance @xmath75 , i.e. @xmath76 here , we have introduced the notation @xmath77 for a gaussian with variance @xmath78 .
the wave functions are rarely accessible in an experiment .
however , other observables , such as scattering matrix elements or partial widths , sensitively depend on them .
consider the scattering from a state @xmath79 with wave function @xmath71 into a channel @xmath80 with channel wave function @xmath81 .
the corresponding reduced partial width amplitude @xmath82 can be written as @xmath83 where @xmath84 is the overlap integral between the @xmath85th canonical basis vector and @xmath81 .
again , the probability density can be worked out @xmath86 where we suppress the indices @xmath79 and @xmath80 .
the second moment reads @xmath87 .
we notice that the functional form of this distribution agrees with that of eq .
( [ eqiii.0 ] ) .
thus , in the limit of large n , one again finds a gaussian .
usually , one introduces the partial width @xmath88 which can be measured . the relative partial width @xmath89 with @xmath90
is distributed according to the porter
thomas law ( [ eqii.5 ] ) .
we notice that the porter thomas law or , equivalently , the gaussian for the partial width amplitudes @xmath1 , results from a large @xmath72 expansion of the distribution ( [ eqiii.2 ] ) .
alternatively , one may derive these large @xmath72 results by using the central limit theorem : the partial width amplitudes are , according to eq .
( [ extraiii.1 ] ) , given as a linear combination of @xmath72 components @xmath70 .
if as assumed in ( [ eqiii.2 ] ) all these components are independently distributed , the distribution of the partial width amplitudes approaches a gaussian for large @xmath72 .
this line of arguing does not use the fact that the distribution of every single component @xmath70 is gaussian for large @xmath72 .
it would apply for any smooth distribution of @xmath70 , provided it does not sensitively depend on @xmath72 .
this subtle point will be important in sec .
[ iva ] .
the results compiled in sec . [ iiia ]
apply to the partial widths and to their amplitudes .
however , in the experiments on @xmath0al , electromagnetic transition probabilities @xmath91 were measured which are squares of transition matrix elements according to eqs .
( [ eqii.3 ] ) and ( [ eqii.4 ] ) . in ref .
@xcite it is argued that the distribution of the transition matrix elements @xmath92 is , once more , of the form ( [ eqiii.0 ] ) and ( [ eqiii.2 ] ) . in the present section , however , we give a derivation , valid for large @xmath72 , which is well suited for the discussion of isospin breaking in sec .
[ iv ] . in our model , the initial state @xmath50 and the final state @xmath51 ,
are represented by the eigenvectors @xmath71 and @xmath93 , respectively .
the electromagnetic transition operator @xmath94 is modeled by one fixed random matrix .
it is not necessary to consider an ensemble of such operators .
thus , the transition matrix elements @xmath95 are given by @xmath96 the transition probabilities read @xmath97 .
although these quantities differ from the partial widths and their amplitudes , it is intuitively obvious that , for large @xmath72 , the @xmath95 are gaussian distributed and that the @xmath91 obey a porter
thomas law : since we always consider @xmath98 , the @xmath95 are linear combinations of ( products of two ) independent variables and , therefore , the central limit theorem applies .
more precisely , the distribution of the matrix elements @xmath92 reads @xmath99 p(u_i ) \int d[u_f ] p(u_f ) \nonumber \\ & & \qquad\qquad \delta\left(w_{if } - u_f^t{\cal o}u_i\right ) \ , \label{a1}\end{aligned}\ ] ] where we suppress the argument @xmath100 of @xmath92 and @xmath101 .
since we may assume that @xmath72 is large , the distributions of the components @xmath70 and @xmath102 take the form @xmath103 with @xmath104 standing for either @xmath71 or @xmath93 .
we notice that the integration domain for each of the eigenvector components is the interval @xmath105 $ ] .
however , since we consider the large @xmath72 limit , the distributions ( [ a2 ] ) are so sharply peaked at the origin that we may extend the domain of integration to the entire real axis .
using the fourier transform of the @xmath106 function in eq .
( [ a1 ] ) we have @xmath107 the gaussian integrals absorbed in @xmath108 can be done easily , @xmath109
\int d[u_f ] \nonumber \\ & & \exp\left(-\frac{n}{2}[u_i^t \
u_f^t ] \left [ \begin{array}{cc } \openone _ n & it{\cal o}/n \\ it{\cal o}/n & \openone _
n \end{array } \right ] \left [ \begin{array}{c } u_i
\\ u_f \end{array } \right ] \right ) \nonumber \\ & = & { \rm det}^{-1/2}\left(\openone _
n+t^2{\cal o}^2/n^2\right ) \nonumber \\ & = & \exp\left(-\frac{1}{2 } { \rm tr}\ln\left(\openone _
n+t^2{\cal o}^2/n^2\right)\right ) \ . \label{a4}\end{aligned}\
] ] we are interested in large @xmath72 .
this allows us to expand the logarithm and to keep only the first term which is @xmath110 .
thus we have for large @xmath72 , @xmath111 where the second moment of the operator is defined by @xmath112 we assume that the matrix elements @xmath113 do not depend on @xmath72 .
hence , collecting everything , the fourier transform in eq .
( [ a3 ] ) yields @xmath114 which is the expected result .
in sec . [ iva ] , the model is set up for the case that isospin is partially conserved . in sec .
[ ivb ] , numerical simulations are described .
note that we speak of isospin breaking because we have in mind the experiment of sec .
the model , however , applies to any other quantum number that is partially conserved .
we consider two isospin values @xmath2 and @xmath3 .
if isospin were fully conserved , the hamiltonian @xmath62 would be block diagonal .
the coulomb interaction , however , destroys this symmetry . in ref .
@xcite , this was modeled by using random matrices of the form @xmath115 + \alpha \left [ \begin{array}{cc } 0 & h_{c } \\ h_{c}^{t } & 0 \end{array}\right ] , \label{hamiltonian}\end{aligned}\ ] ] where @xmath116 are independent goe matrices with dimension @xmath117 .
the total level number is @xmath118 .
the @xmath119 matrix @xmath120 accounts for the coulomb interaction .
it is real without further symmetries and has gaussian distributed entries . in this model ,
the parameter @xmath121 is proportional to the root mean square coulomb matrix element .
since @xmath122 and @xmath123 have positive and negative entries with equal weight , all observables can only depend on the modulus of @xmath121 , but not on its sign .
thus , in the sequel , we restrict ourselves to @xmath124 .
we recall that the _ spectral _ fluctuations are measured on the scale of the local mean level spacing @xmath35 .
thus , the relevant parameter governing the spectral fluctuations is @xmath125 for @xmath126 we have a non interacting superposition of two independent goe s . in this case , the distribution of the eigenvector components of the _ full _ matrix @xmath62 , i.e. including the many exact zeros , is given by @xmath127 where we have introduced the fractional level numbers @xmath128 the distributions @xmath129 have to be taken as either eq .
( [ eqiii.0 ] ) or ( [ eqiii.1 ] ) .
obviously , the total distribution @xmath130 is properly normalized . as discussed in ref .
@xcite , the variances of the distributions for the matrix elements are chosen in such a way that the spectra of @xmath131 and @xmath132 have the same length and that @xmath62 becomes a full @xmath61 goe matrix @xcite for @xmath133 if @xmath134 .
this also means that @xmath135 for @xmath133 but @xmath136 there are some deviations from the pure goe results . for arbitrary @xmath121 ,
the distribution @xmath137 is not known analytically .
a qualitative model is presented in sec .
[ v ] ( see eq .
( [ rat1 ] ) ) .
the transition matrix elements @xmath95 and the corresponding transition probabilities @xmath91 depend not only on the eigenstates @xmath71 of @xmath62 , but also on the transition operator @xmath138 .
since the latter contains the effective charges or the magnetic @xmath139 factors of proton and neutron , it causes isospin breaking in addition to the isospin breaking built into ( [ hamiltonian ] ) .
thus , in the same isospin basis that was used in eq .
( [ hamiltonian ] ) , we may model the transition operator by a matrix of the form @xmath140 + \beta_c \left [ \begin{array}{cc } 0 & { \cal o}_{c}\\ { \cal o}_{c}^{t } & 0 \end{array}\right ] \ . \label{operator}\end{aligned}\ ] ] here , each of the matrices @xmath141 , @xmath142 will be modeled by one fixed random matrix as described in sec .
[ ivb ] . for later convenience ,
@xmath101 has been written as a function of the two parameters @xmath143 and @xmath144 . however , only the ratio @xmath145 is important since the total strength of @xmath101 drops out of the observable @xmath40 of eq .
( [ eqii.1 ] ) . for @xmath146 ,
configurations with different isospin values are not mixed by @xmath101 while for @xmath147 the mixing is maximal . for @xmath148 the operator @xmath101 couples configurations with different isospin only .
we are interested in the distribution of the @xmath95 and the @xmath91 . according to the discussion in the previous sec .
[ iii ] , one is tempted to argue as follows : since eq .
( [ eqiiib.1 ] ) is general and also holds in the presence of conserved or broken isospin , the transition matrix elements @xmath95 are always a linear combination of products of independently distributed sets of variables @xmath70 and @xmath149 .
thus , the central limit theorem should apply and we should obtain a gaussian distribution for the @xmath95 .
this should even be true for any smooth distribution for the wave function components @xmath70 and @xmath149 .
in other words , the specific form of these latter distributions , which are non trivial functions of the mixing parameter @xmath150 , is not important and we would always expect a gaussian distribution for the transition matrix elements @xmath95 .
however , this reasoning is incorrect , because it does not make use of a specific basis . in other words : the result of this reasoning is an orthogonally invariant distribution . within the random matrix model ( [ hamiltonian ] )
one sees that this can not be true .
the statistical properties of the hamiltonian @xmath62 are not orthogonally invariant except for the special case of no isospin symmetry at all .
one sees the lack of orthogonal invariance very clearly in the distribution ( [ dist0 ] ) of the eigenvector components in the case @xmath126 : the @xmath151-distribution will appear only if the basis vectors have well defined isospin .
in general , the wave function components are functions of @xmath150 .
these functions parametrically depend on @xmath72 .
this @xmath72-dependence competes with a large @xmath72 expansion needed in the derivation of the central limit theorem whose premises are therefore violated .
we illustrate this in sec .
[ ivb ] by numerical examples .
the distributions of the reduced transition amplitudes and reduced transition probabilities were numerically investigated . to this end ,
random hamiltonians with the structure of eq .
( [ hamiltonian ] ) have been constructed .
the dimensions @xmath152 and @xmath153 were chosen to be 100 so that @xmath62 has the dimension @xmath154 .
the elements of @xmath122 , @xmath155 , were selected by a generator of gaussian random numbers such that the second moments were @xmath156 this ensures that @xmath62 is a full goe matrix for @xmath157 .
the mean spacing @xmath158 of the eigenvalues of @xmath122 is then ( in the center of the spectrum ) @xmath159 the mean level spacing @xmath35 of @xmath62 is @xmath160 this value changes little , when @xmath150 is varied between 0 and 0.157 .
this range of values was considered in the present numerical simulations .
the parameter @xmath161 of eq .
( [ 23 ] ) then covers the range of @xmath162 .
ten hamiltonians were constructed and diagonalized for each value of @xmath161 .
the transition operator @xmath101 was constructed very much as the hamiltonians ; i.e. its elements were selected by the random number generator .
the variances were chosen in complete analogy with eqs .
( [ 30 ] ) .
however , for each value of @xmath163 , one operator has been generated .
the range of @xmath164 was covered . for given parameters @xmath161 , @xmath163
the transition amplitudes ( [ eqiiib.1 ] ) have been calculated with the indices @xmath165 running over the eigenvectors of the ten hamiltonians .
the set of @xmath166 numbers @xmath92 with @xmath167 forms the numerically generated data . in a first step ,
we have checked whether these data follow a gaussian distribution . to this end
one can bin the set of @xmath92 and compare the histogram to a gaussian ; similarly one can compare the distribution of the variable @xmath40 of eq .
( [ eqii.1 ] ) to the distribution ( [ eqii.5 ] ) or the distribution of @xmath42 of eq . ( [ eqii.2a ] ) to @xmath59 given in ( [ eqii.6 ] ) .
in order to make the comparison as much as possible independent of the parametrization of the statistical variable , the generalized entropy @xmath168 has been used to express the difference between the distribution @xmath169 of the data and the ansatz @xmath54 .
this expression is independent of the parametrization @xmath170 . since both , @xmath169 and @xmath54 , are normalized , @xmath171 is never positive
. it vanishes if and only if @xmath172 . the data do of course not provide a continuous probability density @xmath173 but rather a histogram where the @xmath49th bin is centered at @xmath174 and contributes the probability @xmath175 .
therefore they provide the approximation @xmath176 to expression ( [ eqiva.3 ] ) . to the extent
that ( [ 34 ] ) is a good approximation to ( [ eqiva.3 ] ) , expression ( [ 34 ] ) is independent of the parametrization .
this holds in the present case to within a few times @xmath177 due to the large ensemble of numerical data . in tab .
[ entropy ] , the entropy ( [ 34 ] ) is given for a few values of @xmath163 and @xmath161 in the range of @xmath178 , @xmath179 .
one finds that the distribution of the data does depend on the parameters @xmath161 and @xmath163 which govern the symmetry breaking .
the entropy approaches zero , when @xmath163 or @xmath161 approach unity , i.e. when the symmetry is strongly broken .
hence , in this case the transition amplitudes @xmath92 have a gaussian distribution .
this does not hold if @xmath163 and @xmath161 have values much smaller than 1 , in which case neither the hamiltonian ( [ hamiltonian ] ) nor the operator ( [ operator ] ) significantly breaks the symmetry . .entropy for different values of the parameters @xmath163 and @xmath161 .
the parameter @xmath150 is also shown .
[ cols="^,^,^,^,^,^,^,^ " , ] [ entropy ] the figures [ fig1 ] and [ fig3 ] illustrate in which way the distribution of @xmath92 deviates from a gaussian .
consider the case of @xmath180 , i.e. the absence of symmetry breaking , displayed in fig .
there is no transition between different symmetry classes and therefore about one half of the matrix elements @xmath92 vanishes .
this leads to the narrow peak in fig .
[ fig1]a comprising one bin centered at zero .
this peak is superimposed over the distribution of the matrix elements connecting states in one and the same symmetry class .
[ fig1]a thus corresponds to eq .
( [ dist0 ] ) with @xmath181 .
when the symmetry is completely broken in fig .
[ fig1]d the matrix elements @xmath92 have a gaussian distribution .
intermediate situations are displayed in the parts b and c of fig .
[ fig1 ] .
the intermediate situation is especially well visualized on fig .
[ fig3 ] , where the distributions of fig .
[ fig1 ] have been transformed to the logarithmic variable @xmath182 which was introduced in eq .
( [ eqii.2a ] ) . [ t ! ] 0.3 cm 2 here , the distribution displays two peaks .
one of them centered at @xmath60 stems from the matrix elements connecting states within one symmetry class .
the other one centered at negative values of @xmath42 is due to the matrix elements connecting states from different symmetry classes . in case of complete symmetry breaking
the two peaks merge as in fig .
this distribution corresponds to eq .
( [ eqii.6 ] ) . in the present parametrization
the case of no symmetry breaking ( fig .
[ fig3]a ) can not be distinguished from the case of complete symmetry breaking ( fig .
[ fig3]d ) .
this occurs because @xmath183 when @xmath184 ; hence , the matrix elements with the value @xmath185 are found nowhere on the @xmath186axis . in sec .
[ ii ] , it has been described how the secular variations of the statistical properties of the experimental data of ref .
@xcite have been removed by help of the local average in eq .
( [ eqii.1 ] ) .
we have applied the same procedure to some of the present numerically generated data sets . introducing a local average @xmath187 instead of the global average
@xmath188 used in figs .
[ fig1],[fig3 ] , does not change the present results ; i.e. along the spectrum of eigenstates there are no secular variations of the statistical properties of the eigenfunctions .
the numerical simulations clearly demonstrate that there indeed is an effect of symmetry breaking on the distributions .
hence , it is shown that the complexity of the wave functions alone is not enough to ensure gaussian distributions of wave function , width and transition amplitudes , as a naive application of the central limit theorem would imply .
in the present section , we discuss the random matrix model in a heuristic fashion .
this is an extension of the sec .
[ iii ] to the case of broken symmetry .
although we believe that the model of eqs .
( [ hamiltonian],[operator ] ) can be completely solved analytically , we refrain from tackling that task .
as we shall see , qualitative reasoning will provide sufficient understanding of the statistical behavior of the observables .
thus , our analytical derivations will be approximate and only exact in some limiting cases . of course , we shall compare our approximate results with the above numerical simulations of the full model . to obtain the distribution
@xmath189 we use the ansatz ( [ operator ] ) for the transition operator and we need an ansatz for the distributions of the wave functions . the latter ones in our model of the eigenvector components are the entries of the orthogonal matrix @xmath190 which diagonalizes the hamiltonian , @xmath191 \ , \label{umatrix}\end{aligned}\ ] ] where the matrices @xmath192 are @xmath193 orthogonal and @xmath194 and @xmath195 are @xmath196 and @xmath197 matrices , respectively , with the condition that the total @xmath190 is orthogonal .
motivated by the numerical simulations in the previous section and by the general experience with wave function distributions , we assume that the distributions of the entries can be approximated by gaussians . more precisely , we estimate that the distributions of the @xmath198 matrix elements are roughly given by @xmath199 , while those of the @xmath194 and @xmath195 matrix elements are roughly given by @xmath200 .
thus , we reduce the problem to the determination of the three parameters @xmath201 , @xmath202 and @xmath203 . it is reasonable to assume that the former two are functions of the latter one , i.e. we set @xmath204 .
this is motivated by the fact that @xmath205 can replace the mixing parameter @xmath150 ( or @xmath161 ) and that apart from the level numbers @xmath152 , @xmath153 the distribution of the elements of @xmath190 is a function of the symmetry breaking .
the final task is then to find the relation between the purely phenomenological parameter @xmath203 and the parameter @xmath121 which measures the root mean square symmetry breaking matrix element and thus has a direct physical interpretation . in other words
, we eventually have to specify the function @xmath206 .
-0.4 cm [ t ! ] 0.3 cm 2 to construct a consistent guess for the functions @xmath204 , we notice that , in line with our basic assumptions , the combined distribution for the matrix elements of the total @xmath190 reads @xmath207 this form is consistent with the exact results ( [ dist0 ] ) for arbitrary @xmath208 and @xmath209 and ( [ dist1 ] ) for @xmath134 .
we mention in passing that we always deal with gaussian approximations of distributions which , in some limiting cases , are for finite level number given by eq .
( [ eqiii.0 ] ) .
let us consider the case @xmath210 which yields the distribution ( [ dist0 ] ) .
comparison with eq .
( [ rat1 ] ) shows that @xmath211 and @xmath212 in the case of @xmath133 and @xmath134 , the exact result ( [ dist1 ] ) leads to @xmath213 and @xmath214 moreover , the functions @xmath204 ought to be even in @xmath203 .
this suggests the ansatz @xmath215 it is reassuring that similar combinations of parameters do also show up in exact calculations of the spectral correlators .
a reasonable form for the function @xmath206 remains to be given .
the function will parametrically depend on @xmath216 and @xmath217 .
we construct it by comparing the ansatz ( [ rat1 ] ) with the distribution of the coefficients @xmath70 of the eigenfunctions @xmath218 from the numerical simulation of sec .
[ ivb ] . for each value of @xmath219 ,
the optimum @xmath205 was determined by maximizing the entropy ( [ 34 ] ) .
we have not attempted to assign errors to the values of @xmath205 thus found .
instead we have convinced ourselves by inspection that the ansatz ( [ rat1 ] ) reproduces the numerical distribution reasonably well .
a typical example is given by fig .
[ fig4 ] . in this way
the function @xmath220 was generated numerically .
for the case of @xmath181 is is displayed on fig .
[ fig5 ] together with the analytical expression @xmath221 that well approximates it .
the numerical constant in the exponential is here equal to the level distance of eq .
( [ 32 ] ) .
we emphasize , however , that the relation ( [ 42a ] ) is purely phenomenological . in sec .
[ vi ] , we shall see that ( [ 42a ] ) can successfully be applied to the case of @xmath0al , where @xmath35 is different from ( [ 32 ] ) , see eq .
( [ 73 ] ) below , but the difference is not too large .
thus it seems that the numerical constant is indeed related to the level distance .
this quantity shows a significant @xmath72 dependence , see eq .
( [ 31 ] ) .
it is at this point where the premises of the central limit theorem are violated for the calculations of the distributions of the transition matrix elements .
they can not be a single gaussian for all values of the parameters .
[ t ! ] 0.3 cm having fixed all parameters in our qualitative model for the distribution of the eigenvector components , we can now calculate the distributions for the transition matrix elements @xmath222 .
we emphasize that this involves , apart from the large @xmath72 limit , no further approximation .
we consider @xmath223 , such that the initial and final states @xmath71 and @xmath93 are always different columns of the matrix @xmath190 in eq .
( [ umatrix ] ) . due to isospin breaking
, we have to distinguish four structurally different possibilities : in case ( m00 ) both , @xmath71 and @xmath93 , are among the first @xmath208 columns .
we write @xmath224 \quad { \rm and } \quad u_f = \left [ \begin{array}{c } u_f(0 ) \\
u_f(10 ) \end{array}\right ] \ .
\qquad\quad { \rm ( m00)}\ ] ] in case ( m11 ) , both are from the last @xmath209 columns , @xmath225 \quad { \rm and } \quad u_f = \left [ \begin{array}{c } u_f(01 ) \\
u_f(1 ) \end{array}\right ] \ .
\qquad\quad { \rm ( m11)}\ ] ] in the cases ( m01 ) and ( m10 ) , @xmath71 and @xmath93 are from different parts in @xmath190 , we have @xmath224 \quad { \rm and } \quad u_f = \left [ \begin{array}{c } u_f(01 ) \\
u_f(1 ) \end{array}\right ] \ , \qquad\quad { \rm ( m01)}\ ] ] and @xmath225 \quad { \rm and } \quad
u_f = \left [ \begin{array}{c } u_f(0 ) \\
u_f(10 ) \end{array}\right ] \ .
\qquad\quad { \rm ( m10)}\ ] ] thus , for the case ( m@xmath226 ) , @xmath227 we will obtain a distribution @xmath228 . because of the symmetries @xmath229 and @xmath230 , the cases ( m01 ) and ( m10 ) must yield identical distributions . [ t ! ] 0.2 cm we proceed by generalizing the evaluation of the integral ( [ a1 ] ) in sec .
again , the distribution @xmath231 p(u_i ) \int d[u_f ] p(u_f ) \nonumber \\ & & \qquad\qquad \delta\left(w_{if } - u_f^t{\cal o}u_i\right ) \label{iso1}\end{aligned}\ ] ] is rewritten in the form @xmath232 where the main difficulty is the calculation of the function @xmath233 p(u_i ) \int d[u_f ] p(u_f ) \nonumber\\ & & \qquad\qquad \exp\left(-itu_f^t{\cal o}u_i\right ) .
\label{iso3}\end{aligned}\ ] ] in the case ( m00 ) , the distributions of the eigenvectors are according to our ansatz @xmath234 for the initial state and @xmath235 for the final state .
the case ( m11 ) is treated similarly .
the case ( m01 ) has a different structure : the initial state distribution is as in eq .
( [ iso4 ] ) , but the final state distribution reads @xmath236 however , one sees that in every case we may cast eq .
( [ iso3 ] ) into the form @xmath237 \int d[u_f ] \nonumber\\ & & \qquad \exp\left(-\frac{n}{2}[u_i^t \ u_f^t ] \ , { \cal k}_{jj^\prime } \left [ \begin{array}{c } u_i \\
u_f \end{array } \right ] \right ) \nonumber \\ & = & c_{jj^\prime}^\prime \ , { \rm det}^{-1/2}{\cal k}_{jj^\prime } .
\label{iso7}\end{aligned}\ ] ] here , @xmath238 and @xmath239 collect all normalization constants .
it is straightforward to write down the matrices @xmath240 .
we define the diagonal matrices @xmath241 and find @xmath242 \label{iso9}\end{aligned}\ ] ] in the four cases .
an easy calculation yields @xmath243 which leads in the large @xmath72 limit to @xmath244 the effective second moment @xmath245 of the operator @xmath101 is defined by @xmath246 hence , by performing the fourier transform in eq .
( [ iso2 ] ) , we arrive at @xmath247 i.e. at a gaussian in all four cases
. the gaussian form does not come as a surprise .
the crucial and non trivial result of this calculation is the explicit form of the effective second moments ( [ iso12 ] ) .
denoting the second moments of the blocks in the operator ( [ operator ] ) by @xmath248 we find after a simple computation in the four cases @xmath249 the distributions @xmath228 of eq .
( [ iso13 ] ) describe the cases ( m@xmath226 ) .
these were defined by assuming that the structure of the initial and final states , i.e. their place of origin in the matrix ( [ umatrix ] ) is known . in a set of experimental data ,
this information is usually lacking .
thus , the full distribution must be a proper superposition of the functions @xmath228 .
the weights in this superposition can be easily constructed : in the case ( m00 ) , there are @xmath208 states and therefore @xmath250 transitions .
we recall that we exclude recoupling transitions with @xmath251 . since , altogether , there are @xmath72 states and @xmath252 transitions , the weight of the case ( m00 ) is @xmath253 , which yields @xmath254 for large level numbers . in the case
( m11 ) , one obviously finds @xmath255 .
similarly , we find @xmath256 , i.e. @xmath257 for large level numbers , in both of the cases ( m01 ) and ( m10 ) .
thus , collecting everything we eventually end up with @xmath258 where the individual variances are explicitly given in eq .
( [ iso15 ] ) . the choice of the transition operator @xmath101 in sec .
[ ivb ] implies that the second moments of the blocks in ( [ operator ] ) are ( for large @xmath117 ) @xmath259 and @xmath260 inserting this and the definition ( [ kappa ] ) of the functions @xmath261 into ( [ iso15 ] ) leads to @xmath262 \beta _ { i } ^{2 } + 2\eta ^{2 } ( 1-g_{1 } \eta ^{2 } ) g_{1 } \beta _ { c}^{2 } \nonumber\\ \overline{{\cal o}_{11}^2 } & = & \left [ ( 1-g_{0 } \eta ^{2 } ) ^{2 } + \eta ^{4 } g_{0}^{2 } \right ] \beta _ { i } ^{2 } + 2\eta ^{2 } ( 1-g_{0 } \eta ^{2 } ) g_{0 } \beta _ { c}^{2 } \nonumber\\ \overline{{\cal o}_{01}^2 } & = & ( 1 - 2g_{0 } g_{1 } \eta ^{2})\eta ^{2 } \beta _ { i}^{2 } + ( 1- \eta ^{2 } + 2 g_{0}g_{1 } \eta ^{4})\beta _ { c}^{2 } \ . \nonumber\\ \label{a}\end{aligned}\ ] ] let us specialize these results to the case of the numerical simulation in sec .
[ ivb ] and see whether both are in qualitative agreement .
the numerical simulation has been performed with @xmath181 , whence , @xmath263 since the total strength of the operator @xmath101 is immaterial see the discussion of eq .
( [ 28 ] ) one can set @xmath264 which makes @xmath265 .
all this turns eqs .
( [ a ] ) into the more transparent expressions @xmath266 if @xmath147 , the operator ( [ operator ] ) completely mixes configurations with different isospin . in this case ,
( [ 62 ] ) yield @xmath267 the distribution ( [ rat2 ] ) of @xmath92 then becomes @xmath268 i.e. a gaussian for all @xmath205 and , hence , for all @xmath150 .
this qualitatively agrees with the results of the numerical simulation , see tab .
[ entropy ] .
if @xmath157 ( and @xmath181 ) , the hamiltonian ( [ hamiltonian ] ) is a full goe matrix and its eigenstates have no isospin symmetry .
according to eq .
( [ eta1 ] ) , one then has @xmath269 , eqs .
( [ 62 ] ) yield @xmath270 and the distribution ( [ rat2 ] ) becomes @xmath271 ^{1/2}\right ) , \label{67}\end{aligned}\ ] ] i.e. a gaussian for all @xmath163 .
again this qualitatively agrees with the results of the numerical simulation , see tab .
[ entropy ] .
if neither the hamiltonian ( [ hamiltonian ] ) nor the operator ( [ operator ] ) mixes isospin , i.e. if @xmath163 and @xmath205 are zero , eqs . ( [ 62 ] ) give @xmath272 in this limit , the distribution ( [ rat2 ] ) takes the form @xmath273 this says that half of the transition matrix elements is zero namely the ones , where @xmath79 and @xmath274 have different isospin .
the other half is gaussian distributed .
this result is expected and agrees with the numerical simulation , see tab .
[ entropy ] and fig .
[ fig1]a . in fig .
[ fig6 ] , the distribution of @xmath92 is displayed for the mixing parameters @xmath275 and @xmath276 . the histogram is the result of the numerical simulation in sec .
the curve is the qualitative model ( [ rat2 ] ) .
the parameters follow from eq .
( [ 42a ] ) which yields @xmath277 and from eqs .
( [ 62 ] ) which give @xmath278
this illustrates that the qualitative model reproduces the essential features of the numerical data although it does not describe them quantitatively .
[ t ! ] 0.2 cm
in sec.[via ] , the experimental data described in sec .
[ ii ] are compared to the prediction of the qualitative solution ( [ iso15 ] ) of the model of sec .
[ iva ] . in sec .
[ vib ] , we discuss the effect that symmetries other than isospin have on the data .
parity conservation is used as an example .
the results of secs .
[ iv ] and [ v ] suggest that the distribution of the reduced electromagnetic transition probabilities displayed in fig .
[ fig26al ] does carry a signature of isospin breaking . in order to verify this , we compare in the present section the random matrix model to the experimental data .
the parameters of the random matrix model are chosen as follows .
the dimensions @xmath152 , @xmath153 of the blocks in the hamiltonian ( [ hamiltonian ] ) are not equal .
rather one has @xmath279 which approximately corresponds to the ratio of the densities of states with @xmath2 and @xmath3 in @xmath0al .
this entails @xmath280 the parameter which quantifies isospin breaking in the hamiltonian was set to @xmath281 .
this value has been taken from the analysis of the eigenvalue statistics in ref .
we must convert it to the parameter @xmath205 of the qualitative model of sec .
however , it is so far not clear whether eq .
( [ 42a ] ) applies because that function is based on a numerical simulation with @xmath181 which is different from ( [ 72 ] ) .
we have therefore performed another numerical simulation of the distribution of the eigenvector coefficients @xmath68 .
this simulation was done with @xmath282 and @xmath283 . for each @xmath150 , the function ( [ rat1 ] )
was again compared to the numerical distribution of @xmath68 and the optimum @xmath205 was determined .
this yielded the crosses on fig .
it turns out that the function ( [ 42a ] ) again reproduces the numerical data if one considers only the range of @xmath284 .
note that in the case at hand , the level distance is @xmath285 for @xmath286 the function ( [ 42a ] ) does not reproduce the data of fig .
[ fig7 ] because the hamiltonian ( [ hamiltonian ] ) does not become a goe matrix for any value of @xmath150 if @xmath287 . due to eq .
( [ 73 ] ) the numerical constant in ( [ 42a ] ) is no longer equal to the level distance .
one finds @xmath288 for the present case of @xmath0al . due to the fact that @xmath0al is a mirror nucleus ,
i.e. it has the isospin projection @xmath289 , the transition operator is uniquely determined by its isospin structure . [ hbt ] if @xmath101 is isoscalar it can by definition not connect states with different isospin .
the parameter @xmath144 in ( [ operator ] ) is then zero and , of course , @xmath146 . if the operator is isovectorial it can not connect states with the same isospin .
this is obvious for states having @xmath2 . for states having @xmath3
, this follows from the symmetry properties of @xmath101 in isospin space .
see chap .
9 of ref .
@xcite , specially the discussion following eq .
( 9.107 ) .
the parameter @xmath143 in ( [ operator ] ) is then zero .
we evaluate the distribution ( [ rat2 ] ) for these two cases .
let the operator @xmath101 be isoscalar , whence @xmath290 .
we set @xmath291 , make use of eqs .
( [ 77],[74 ] ) , and obtain from eqs .
( [ a ] ) @xmath292 this gives the distribution ( [ rat2 ] ) of the isoscalar matrix elements @xmath92 .
it is transformed to the variable @xmath293 the result is the dot - dashed curve on fig .
[ fig8 ] .
now let the operator @xmath101 be isovectorial , whence @xmath294 .
we choose @xmath295 , use eqs .
( [ 74],[77 ] ) and obtain from eq .
( [ a ] ) @xmath296 this gives the distribution ( [ rat2 ] ) of the isovectorial matrix elements @xmath92 .
again this distribution is transformed to the variable ( [ 79 ] ) .
the result is the dashed curve on fig .
[ fig8 ] .
one sees that both , the distribution with @xmath297 and the one with @xmath298 , are shifted with respect to the porter - thomas distribution ( [ eqii.6 ] ) .
this is a consequence of the appearance of gaussian distributions with different widths in eq .
( [ rat2 ] ) .
the shoulder on the r.h.s . of the isovector distribution in fig .
[ fig8 ] even hints to the appearance of two peaks as in the numerical simulations on fig .
[ fig3]b and c. however , in @xmath0al , isospin mixing is too large ( namely @xmath299 ) as to let the peaks separate
. remember that it is the class of small matrix elements that produces the peak which is centered at negative values of @xmath42 .
the class of large matrix elements leads to a peak which is centered close to @xmath60
. the smaller the relative strength of the small matrix elements is , the more clearly the peaks separate .
[ hbt ] there is a factor other than @xmath150 ( or @xmath300 ) which determines the shape of the distributions on fig .
[ fig8 ] . for the isoscalar problem
the class of the large matrix elements is the one with variance @xmath301 , see eqs .
( [ 78 ] ) .
it has the largest weight in the distribution , namely @xmath302 , see ( [ rat2],[77 ] ) .
therefore the distribution is shifted only slightly to negative values of @xmath42 .
for the isovector problem the class of the small matrix elements is the one with variance @xmath303 , see eqs .
( [ 82 ] ) , which has the largest weight .
therefore the distribution is shifted more strongly to negative values of @xmath42 .
this makes the isovector matrix elements more sensitive to isospin breaking than the isoscalar matrix elements .
the distributions for the isovector and isoscalar cases have been summed with the weights 530 and 343 , respectively .
these are the numbers of observed transitions , see ref .
the result is the dotted curve on fig .
it is shifted with respect to the porter - thomas distribution by about the same amount as the experimental data . however , the data are more strongly peaked around @xmath304 than the present theory .
this discrepancy must be attributed at least partially to the way the detection thresholds have been taken into account , see eq .
( [ 8a ] ) : the percentage of undetected transitions may have been underestimated because its estimation was based on the porter - thomas distribution .
the distributions that include isospin violation predict a larger probability of events outside the range of the experimental data than does the porter - thomas distribution .
we conclude that the present investigation supports the conjecture of ref .
@xcite which attributes the discrepancy between the porter - thomas distribution and the experiment to partial isospin breaking in @xmath0al .
the nucleus @xmath0al has more than only isospin symmetry .
especially there are quantum numbers which are exactly conserved or nearly so . for definiteness
let us discuss parity conservation .
should it be included into all the above considerations that have focussed on isospin conservation alone ? the matrix model of sec .
[ iva ] can be extended to take parity into account .
if the system has two possible values of isospin @xmath305 ( isospin being broken to some extent ) and parity @xmath306 ( parity being a very well conserved symmetry ) , the hamiltonian @xmath62 as well as the operator @xmath101 can be written in the following block form @xmath307{1.13 cm } { 0.7 cm } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.5){\framebox(1.1,0.67 ) { } } \put(0.14,0.73 ) { .. . .
} \put(0.34,0.78){. . }
\put(0.49,0.88){. . .
} \put(0.44,0.63 ) { . .
} \put(0.74,0.58){$\ddots $ } \put(0.79,0.88 ) { . }
\put(1.04,0.83){. } \put(0.14,0.90 ) { .. . .
} \put(0.14,0.58){. . . .
} \put(0.14,0.98){. . .
.. }
\put(0.19,1.08){. . .. . }
\put(0.34,0.66){. } \end{picture } \\
\setlength{\unitlength}{1 cm } \begin{picture}(0.3,0.5 ) \put(0.0,0.2){\makebox(0.5,0.5){1 } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.1){\framebox(1.1,0.67 ) { } } \put(0.14,0.33 ) { .. . .
} \put(0.34,0.38){. . }
\put(0.49,0.48){. . .
} \put(0.44,0.23 ) { . .
} \put(0.74,0.18){$\ddots $ } \put(0.79,0.48 ) { . }
\put(1.04,0.43){. } \put(0.14,0.50 ) { .. . .
} \put(0.14,0.18){. . . .
} \put(0.14,0.58){. . .
.. } \put(0.19,0.68){. . .. . }
\put(0.34,0.26){. } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.35){\rule[-2.6mm]{1.13 cm } { 0.7 cm } } \end{picture } \end{array } $ & $ \begin{array}{cc } \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.3 ) \put(0.4,0.5){\makebox(0.5,0.5){0 } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.3 ) \put(0.4,0.5){\makebox(0.5,0.5){1 } } \end{picture } \\ \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.55){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.55){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } \\
\setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.15){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.5 ) \put(0.1,0.15){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } \\ & \end{array } $ \\ \hline % & % & % \\ \setlength{\unitlength}{1 cm } \begin{picture}(0.2,1.6 ) \put(-0.1,0.15){\makebox(0.5,0.5){$-$ } } \end{picture } & $ \begin{array}{ccc } \setlength{\unitlength}{1 cm } \begin{picture}(0.3,0.8 ) \put(0.0,0.35){\makebox(0.5,0.5){0 } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.55){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.55){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } \\ \setlength{\unitlength}{1 cm } \begin{picture}(0.3,0.8 ) \put(0.0,0.35){\makebox(0.5,0.5){1 } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.5){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.5){\framebox[1.1cm]{\rule[-1.5mm]{0 cm } { 0.45 cm } } } \end{picture } \end{array } $ & $ \begin{array}{cc } \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.53){\rule[-2.6mm]{1.13 cm } { 0.7 cm } } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.27){\framebox(1.1,0.67)[c ] { } } \put(0.14,0.50 ) { .. . . }
\put(0.34,0.55){. . }
\put(0.49,0.65){. . .
} \put(0.44,0.40 ) { . .
} \put(0.74,0.35){$\ddots $ } \put(0.79,0.65 ) { . }
\put(1.04,0.60){. } \put(0.14,0.67 ) { .. . .
} \put(0.19,0.35){. . . .
} \put(0.14,0.75){. . .
.. } \put(0.19,0.85){. .
} \put(0.34,0.43){. } \end{picture } \\
\setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.23){\framebox(1.1,0.67)[c ] { } } \put(0.14,0.46 ) { .. . .
} \put(0.34,0.51){. . }
\put(0.49,0.61){. . .
} \put(0.44,0.36 ) { . .
} \put(0.74,0.31){$\ddots $ } \put(0.79,0.61 ) { . }
\put(1.04,0.56){. } \put(0.14,0.63 ) { .. . .
} \put(0.19,0.31){. . . .
} \put(0.14,0.71){. . .
.. } \put(0.19,0.81){. . .. . }
\put(0.34,0.39){. } \end{picture } & \setlength{\unitlength}{1 cm } \begin{picture}(1.3,0.8 ) \put(0.1,0.49){\rule[-2.6mm]{1.13 cm } { 0.7 cm } } \end{picture } \end{array } $ \end{tabular } } .
\label{fim}\end{aligned}\ ] ] the four larger blocks are related to parity symmetry : the two blocks on the diagonal correspond to transitions between positive and negative parity configurations , respectively .
the off - diagonal blocks correspond to parity violating transitions .
they are zero in the present context .
each of the larger blocks is broken down into four smaller blocks related to isospin symmetry .
they correspond in a obvious way to isospin conserving and isospin violating transitions .
the color " of the blocks schematically shows the strength of the matrix elements .
the black blocks contain large matrix elements allowed by both isospin and parity conservation .
the dotted blocks contain small matrix elements allowed by parity conservation but forbidden by isospin conservation .
the white blocks contain zero matrix elements forbidden by parity conservation .
since in the present context parity is considered an exact symmetry , the experimenters do not try to measure parity violating transitions .
they assume them to be zero and this is the important point they do not even list them on their records .
hence , the experimental data set consists of the matrix elements in the two larger diagonal blocks .
they are thus equivalent to the data that we have assumed in all foregoing considerations .
hence , the existence of exactly conserved quantities in addition to the partially conserved isospin quantum number is irrelevant if they are not measured .
the existence of partially conserved quantum numbers other than isospin would , however , alter the above results .
in the present study , the effect of a partially broken symmetry has been investigated for the distributions of wave function components and transition probabilities .
this has been done by means of a random matrix model similar to the one introduced by rosenzweig and porter @xcite . however , in the present model not only the hamiltonian but also the transition operator may break the symmetry .
this was dictated by the experimental material at hand and the symmetry that we had in mind .
the data consist of a set of reduced electromagnetic transition probabilities in the nucleus @xmath0al .
the symmetry is isospin conservation .
eletromagnetic transition operators are in general not isospin conserving .
a numerical simulation has shown that neither the components of the eigenfunctions nor the transition matrix elements follow gaussian distributions .
this is counterintuitive since the central limit theorem seems to require gaussian distributions if the wave functions are sufficiently complicated " .
instead , the numerical simulation suggests that the ensembles of both , the components of the eigenfunctions and the transition matrix elements , comprise at least one class of small elements and one class of large elements .
the physical origin of these classes is clear : the small elements are isospin violating ; the large elements are isospin allowed .
at best , each of the classes has a gaussian distribution .
the probability density of the whole ensemble is then a superposition of gaussians with different parameters .
this idea has been worked out in detail and results in a qualitative analytical evaluation of the random matrix model .
it yields simple formulae for the parameters of the gaussians .
its practical use is to circumvent easily the demanding numerical simulations .
conceptually , it offers to see in detail , why the premises of the central limit theorem are not given : the elements in the different classes depend in different ways on the symmetry breaking parameter @xmath150 as well as the dimension @xmath72 of the space .
it is well known that in the limit of @xmath308 an ever smaller value of @xmath150 suffices to make the above classes indistinguishable and thus to produce simple gaussian distributions of the eigenvector components and the transition matrix elements . however , this is what happens if @xmath72 is taken to infinity such that the parameter @xmath309 goes to infinity .
if @xmath72 is finite or if it is taken to infinity together with @xmath310 such that @xmath161 remains finite , then the above mentioned classes conserve their identity .
it is this situation which the present paper deals with .
the present results thus support a conjecture formulated in ref .
@xcite on the basis of the data collected from @xmath0al : the authors conjectured that the distribution of the electromagnetic transition probabilities carries the signature of partial isospin breaking .
the present theory offers a qualitative understanding of the data .
whether it can quantitatively reproduce the data remains to be seen .
we have not attempted to fit the theory to the data .
a caveat seems necessary : the random matrix model ( [ hamiltonian ] ) applies to any partially broken quantum number it is not specific for isospin . certainly , the available information on isospin breaking in @xmath0al requires that the distribution of the transition probabilities differs from a porter - thomas distribution to the extent given by the dotted curve on fig .
[ fig8 ] but the data may in principle and in addition carry the signature of other partially conserved quantum numbers
. these may not be as fundamental as isospin but rather given by the specific form of the effective nucleon - nucleon interaction in the sd shell .
very recently , an analytical calculation , based on the supersymmetry method , of the wave function statistics in coupled chaotic systems was presented in ref .
these results are valid for the case of broken time reversal invariance and therefore , unfortunately , are not applicable to the case of conserved time reversal invariance which is discussed in the present paper .
we thank h. a. weidenmller , a. richter and v.k.b .
kota for fruitful discussions .
we are indebted to g.e .
mitchell and j.f .
shriner , jr . for supplying the experimental data of @xmath0al and for suggestions and criticisms during the work .
c.i.b . acknowledges the financial support granted by the fritz thyssen stiftung , deutsche akademische austauschdienst ( daad ) and conselho nacional de desenvolvimento e pesquisa ( cnpq ) at different periods during this work .
t.g . acknowledges support from the heisenberg foundation .
wigner , in _ statistical theories of spectra : fluctuations _ , edited by c.e .
porter ( academic press , new york , 1965 ) , p.199 .
t. guhr , a. mller groeling , and h.a .
weidenmller , phys .
rep . * 299 * , 189 ( 1998 ) .
n. rosenzweig and c.e .
porter , phys .
* 120 * , 1698 ( 1960 ) .
mitchell , e.g. bilpuch , p.m. endt , and j.f .
shriner , jr . ,
lett . * 61 * , 1473 ( 1988 ) . t. guhr and h.a .
weidenmller , ann .
( n.y . ) * 199 * , 412 ( 1990 ) .
leitner , phys .
e * 48 * , 2536 ( 1993 ) ; d.m .
leitner , h. kppel , and l.s .
cederbaum , phys .
* 73 * , 2970 ( 1994 ) . c. ellegaard , t. guhr , k. lindemann , j. nygrd , and m. oxborrow , phys .
lett . * 77 * , 4918 ( 1996 ) .
h. alt , c.i .
barbosa , h .-
grf , t. guhr , h.l .
harney , r. hofferbert , h. rehfeld , and a. richter , phys .
* 81 * , 4847 ( 1998 ) .
adams , g.e .
mitchell , and j.f .
shriner , jr .
b * 422 * , 13 ( 1998 ) .
a. andersen , c. ellegaard , a.d .
jackson , and k. schaadt ( to be published ) .
adams , g.e .
mitchell , w.e .
ormand , and j.f .
shriner , jr .
b * 392 * , 1 ( 1997 ) .
shriner , jr .
( private communication 1999 ) .
g. auger , o. bohigas , j. flores , notas de fisica * 3 * , 1 ( 1980 ) .
brody , j. flores , j.b .
french , p.a .
mello , a. pandey , and s.s.m .
wong , rev .
phys . * 53 * , 385 ( 1981 ) .
note that there is a lack of precision in the text of @xcite below eq.(1 ) there : the condition of equal dimensions of the goe blocks is missing .
brussaard and p.w.m .
glaudemans , _ shell - model applications in nuclear spectroscopy _
( north - holland , 1977 ) .
a. tschersich and k.b .
efetov , cond - mat/9911284 . | in the present paper we investigate the effect of symmetry breaking in the statistical distributions of reduced transition amplitudes and reduced transition probabilities .
these quantities are easier to access experimentally than the components of the eigenvectors and were measured by adams et al . for the electromagnetic transitions in @xmath0al .
we focus on isospin symmetry breaking described by a matrix model where both , the hamiltonian and the electromagnetic operator , break the symmetry .
the results show that for partial isospin conservation , the statistical distribution of the reduced transition probability can considerably deviate from the porter thomas distribution . 2 |
one of the powerful techniques for probing the dynamic state of driven vortex matter in type ii superconductors is the mode locking ( ml ) experiment .
ml is a dynamic resonance between a superimposed rf drive and the dynamic lattice mode ( as expected to appear when vortices flow coherently in a ( random ) pinning environment ) excited collectively at the internal ( washboard ) frequency @xmath0 with @xmath1 integer , @xmath2 the average dc velocity and @xmath3 the lattice spacing of the array along the flow direction .
when @xmath4 and the rf drive frequency @xmath5 are harmonically related , i.e. , @xmath6 with @xmath7 integer , the resonance appears as steps in the dc current - voltage ( @xmath8 ) characteristics and also emerges more sensitively as jumps ( dips ) in a real ( imaginary ) part of rf impedance at low rf drives @xmath9 @xcite@xcite@xcite .
so far , some features of the ml phenomenon have been well studied in sliding charge density waves in quasi one dimensional conductors @xcite .
it turns out that complete ml features characterized by `` perfect '' steps in dc @xmath10 curve appear when the flow is coherent enough , whereas incomplete features like rounded and broadened steps in the @xmath10 curve appear if the flow becomes more incoherent .
thus , the ml feature provides information on the amount of coherency in the flow . in this study , we have employed the ml technique to study flow of vortices driven through disordered mesoscopic channels @xcite .
the channel device consists of a strong pinning nbn film ( thickness @xmath11=50 nm ) on top of a weakly pinning amorphous nb@xmath12ge@xmath13 film ( @xmath11=550 nm and @xmath140.3 ) . by reactive ion etching and proper masking ,
straight channels were etched leaving a 300 nm thick nbge layer .
the width and length of each channel is 230 nm and 300@xmath15 m , respectively .
the pinning potential for vortices inside the channel is dominated by the interaction with strongly pinned edge vortices which form a disordered configuration ( see the inset to fig.1 ) .
the transport current is applied perpendicular to the channel and drives the vortices along the channel . for ml experiments
, we sweep the dc current with a constant superimposed rf current and record both the in- and out - of - phase part of the rf - voltage as well as the dc voltage .
the sample was immersed in superfluid @xmath16he at a temperature of 1.9k .
because of the confinement of the vortex array in the channels , an integer number @xmath17 of rows is expected to form inside each channel . in a previous study
, we demonstrated that the flow configuration can be related to the voltage @xmath18 at which the fundamental ml ( @xmath19 ) occurs @xcite .
namely , @xmath20 with @xmath21 the flux quantum and @xmath22 ) the number of channels .
in fig.1 we plot the real part of the rf - impedance @xmath23 versus the dc voltage @xmath24 taken at several magnitudes of rf currents @xmath25 of 6mhz .
we choose a field of 170mt where moving configurations of @xmath17=4 rows form in each channel .
around @xmath26v we observe rounded jumps , while below and above @xmath18 @xmath23 increases monotonically with @xmath24 .
it is clear that the height and the voltage broadening of the rounded jump are enhanced when @xmath25 is reduced . for definition of the jump height @xmath27 , we take the derivative of @xmath23-@xmath24 curve and integrate over the corresponding peak in a differential plot of @xmath28 versus @xmath24 .
the inset to fig.2 shows a plot of @xmath27 as a function of @xmath25 . by decreasing @xmath25 , @xmath29 first increases and
then levels off at low @xmath25 .
thus , @xmath29 is `` ohmic '' at low @xmath25 , consistent with the fiory experiment @xcite and theory @xcite .
next we turn to the voltage broadening @xmath30 of the rounded jump .
figure 2 shows the rf current dependence of @xmath30 .
we determine @xmath30 by taking the full width at half maximum of the peak in the differential plot .
it is clear that , with reducing @xmath25 , @xmath30 increases monotonically and seems to diverge at @xmath310 .
since @xmath29 is constant , this indicates that the jump is smeared out toward @xmath32 .
such divergence of @xmath30 implies a pronounced spread in the internal frequency , i.e. , a wide distribution of local velocities and/or lattice spacings .
in fact , we find that no narrow band noise at @xmath4 is detectable in the dc vortex flow
. with increasing @xmath25 , however , the spread in @xmath4 gradually decreases .
this can be explained by the frequency pulling effect where the distribution of @xmath4 narrows and becomes locked into @xmath5 @xcite .
thus , with increasing rf drive , the ml gradually enhances the order of the moving vortex lattice . at high rf drives ,
the ml jumps in the @xmath33 curves disappear .
therefore , we determined @xmath30 from the measured current steps in the dc @xmath8 curves .
the open symbols in fig.2 show that this @xmath30 first decreases monotonically and then levels off at large @xmath25 .
this indicates that complete ( characterized by @xmath30=0 ) ml does not occur , even at high rf drives .
since @xmath30 is more than one decade larger than what is expected for purely elastic deformations @xcite , we conclude that slip between moving rows and possibly remaining plastic regions in the channel partially influence the ml characteristics .
this plasticity is caused by the disordered vortex configurations in the channel edges .
fiory , phys .
lett * 27 * , 501 ( 1971 ) .
a. schmid and w. hauger , j. low .
11 * , 667 ( 1973 ) .
n. kokubo _
lett . * 88 * , 147004 ( 2002 ) .
g. grner , rev .
phys . * 60 * , 1129 ( 1988 ) ; r. e. thorne , _ et al .
b. * 35 * , 6360 ( 1987 ) .
s. bhattacharya _
et al . _ ,
rev . lett . * 59 * , 1849 ( 1987 ) . | we have studied the flow properties of vortices driven through easy flow mesoscopic channels by means of the mode locking ( ml ) technique .
we observe a ml jump with large voltage broadening in the real part of the rf - impedance . upon approaching the pure dc flow by reducing the rf amplitude ,
the ml jump is smeared out via a divergence of the voltage width .
this indicates a large spread in internal frequencies and lack of temporal coherence in the dc - driven state .
@twocolumnfalse 2 |
ls i + 61@xmath0303 is an x - ray binary formed by a compact object and a massive star with an optical spectrum typical for a rapidly rotating b0 v star @xcite .
the nature of the compact object , if neutron star or black hole , is still unknown due to the large uncertainty in the inclination of the object @xcite .
it travels around its companion star on an eccentric orbit with a period of 26.5 days @xcite .
radio spectral index analysis by @xcite have found two peaks along the orbit of ls i + 61@xmath0303 .
each peak shows the microquasar characteristic of a switch from a steady ( optically thick ) to a transient ( optically thin ) jet .
furthermore , each peak in the radio spectral index is accompanied by two distinguishable peaks in the radio flux .
this confirms that there are really two different outbursts ( an optically thick and an optically thin one ) .
also high energy observations with egret @xcite and _
fermi_-lat @xcite indicate two peaks along the orbit .
the high energy peaks are supposed to be due to inverse compton upscattering of the uv photons from the donor star by the relativistic electrons of the jet , which in turn strongly attenuates the radio peak around periastron ( see discussion in @xcite and @xcite ) .
303 from the green bank interferometer vs. orbital phase @xcite .
the large outburst towards apastron is clearly observable .
the broadness of the curve stems from the 4.6 yr period which modulates the orbital occurrence of the peak . ]
it has been shown from theory that for a microquasar with an eccentric orbit ( ls i + 61@xmath0303 : e=0.54 - 0.7 @xcite ) , indeed , the different relationship between the accretion rate for density and velocity described by @xcite , creates two peaks in the accretion rate curve , one at periastron and a second one towards apastron @xcite .
the clear periodicity in ls i + 61@xmath0303 is amongst its most peculiar characteristics .
the orbital period ( @xmath1 ) modulates the flux at radio , h - alpha , x - rays and also in gamma - rays ( high and very high energy ) @xcite .
but ls i + 61@xmath0303 holds another peculiarity : a second period , which modulates the radio flux over a long period of @xmath2=4.6 yr @xcite .
this modulation in radio is shown in fig .
[ fig1 ] ( radio flux density vs. @xmath1 ) , which shows how the peak flux shifts around apastron , and in fig [ fig3 ] , where the radio flux density is given vs. @xmath2 . in addition , this long period has also been shown in h - alpha ( see fig . 2 in @xcite ) .
it is suggested that this modulation could be due to periodic shell ejections from the circumstellar disk of the be star @xcite .
= 0.788 - 0.927 from gbi @xcite .
bottom : _ fermi_-lat lightcurve obtained by @xcite .
the two vertical lines indicate the peaks around periastron and around apoastron .
they both correspond to a negative spectral index in the radio data , which corresponds to optically thin emission in both cases . ]
the radio spectral index analysis by @xcite have shown the importance of the large phase @xmath2 for the analysis of radio data from ls i + 61@xmath0303 .
it has thereby proven that the radio outbursts really consist of two consecutive outbursts .
these outbursts have completely different characteristics . in the microquasar model ,
an optically thick outburst in radio is associated with a steady jet , centered on the compact object .
the optically thin outburst comes from a transient jet , detached from the central engine . in the unified model of x - ray states with radio jets ,
the radio states in microquasars are clearly associated with two x - ray states .
a steady jet corresponds to the low hard state , where the x - ray spectrum is characterized by @xmath31.5 , whereas a transient jet corresponds to a transitional state , e.g. the steep power law state with @xmath42.4 @xcite .
furthermore , high energy and very high energy emission are directly connected to the transitional state ( steep power law state ) as the power law is without cut - off and extends into the gamma - ray regime . as a matter of fact ,
when ls i + 61@xmath0303 was detected by magic and veritas the spectrum was always fitted with a power law with an index @xmath52.4 @xcite . moreover , as discussed in the next section , integral observations seem to indicate a change of the photon index consistent with a change from the low hard to the transitional state in agreement with the radio spectral index .
if the radio spectral index now tells us about the nature of the outburst , then the high and very high energy observations should corroborate this nature and might give additional information about the emission processes .
but in order to get the complete information from the radio spectral index for ls i + 61@xmath0303 , it was neccessary to take into account the long period @xmath2 .
therefore , to compare high energy data with the radio spectral index , it is neccessary as well to take only data of the same @xmath2 phase .
@xcite has done so e.g. for the first eight months of _
fermi_-lat observations and the results ( see fig . [ fig4 ] ) show that in this comparison the high energy peaks established by _ fermi_-lat both correspond to the optically thin outbursts of ls i + 61@xmath0303 .
one could then draw the conclusion that the population of relativistic particles , producing the optically thin outburst ( and therefore the transient jet ) , are also responsible for the production of the high energy emission . furthermore , in fig .
[ fig5 ] are shown radio spectral index curves vs. the orbital phase for two different @xmath2 intervals .
in particular the orbital occurrence of the second peak is different in the different intervals .
it is evident that folding data only on the orbital period , without respect to the long period , might result in mixing up different ejection processes , because the orbital phase of the peaks and therefore of the switch from optically thick to thin emission does not stay the same over the 4.6 yr period .
this point will be discussed in the next section .
303 by egret , magic , _
fermi_-lat , veritas and integral shown in the context of the 4.6 yr radio period ( @xmath2 ) @xcite .
the underlying light curve gives the radio flux at 8.3 ghz against @xmath2 @xcite .
the minimum lies around @xmath60.5 and the maximum around @xmath70.97 .
the three largest @xmath2 intervals where the source was observed with integral are named here i1 , i2 and i3 . ] in fig .
[ fig3 ] , high energy and very high energy observations of ls i + 61@xmath0303 with different instruments are shown with respect to the @xmath2 intervals in which they have been carried out . both , magic and veritas probe the minimum and the maximum parts of @xmath2 @xcite . due to the long integration times for the data ,
the sampling within these intervals is not very strong , although egret observed one complete orbital cycle at each of the two given intervals @xcite .
the latest published coverage by _ fermi _ is of 2.5 years @xcite . for integral ,
the long - term monitoring of ls i + 61@xmath0303 covers vast parts of the 4.6 yr period .
we have denoted the three big integral observations with respect to @xmath2 by i1 , i2 and i3 , as noted in fig .
[ fig3 ] . folding the data by orbital phase , of course , increases the sampling . by doing so
, it was established that the emission between 10 - 100 kev is clearly modulated by the orbital phase @xcite . following the radio spectral index though ( see fig .
[ fig5 ] ) , folding over almost a complete @xmath2 cycle would imply that even though the resulting light curves could show the overall periodicity of the source at these energies , a deeper insight about the ejections can be obscured by mixing different ejection processes ( namely optically thick and thin ejections ) .
the spectral analyses can then get corrupted , because of the same mixing of ejection processes .
i1 and i3 are covering most of the maximum of @xmath2 ( see fig .
[ fig3 ] ) .
an interpretation of these data with respect to the flux and the spectrum should be less corrupted than a mix of data from the maximum and the minimum .
in fact , @xcite , who used most of i3 and parts of i1 , have found not only the modulation of the lightcurve with the orbital period , but also found that along the orbit the spectral index @xmath8 changed from @xmath71.5 around periastron to @xmath73.2 around apastron .
this result strengthens the two - peak accretion model , as for the @xmath2-intervals covered by these observations an optically thin ejection ( a transient jet ) is expected around apastron ( see @xcite and discussions therein ) as mentioned above , in the unified x - ray states model with radio jets , a transient is associated with a transitional state , e.g. the steep power law , characterized by a spectral index @xmath42.4 @xcite .
in the high mass x - ray binary ls i + 61@xmath0303 , two clear radio periodicities are present , one coincident with its orbital period ( see fig . [ fig1 ] ) and the other modulating the strength of the large radio outburst over a period of 4.6 yr @xcite .
both periods have also been observed in h - alpha emission .
this system is amongst a few to have been detected not only in radio and x - rays , but also at high and very high energies .
it is from these observations that , together with radio spectral index analysis , important insights into the nature of the system and its emission processes can be deduced .
the radio spectral index tells us about the nature of the observed outburst .
there are two different kinds of outburst : optically thick ( spectral index @xmath9 0 ) and optically thin ( @xmath10 0 ) .
the first is attributed in the microquasar model to a slow outflow , while the latter is associated with an ultrarelativistic transient jet .
recent spectral index analysis of ls i + 61@xmath0303 show that a subsequent realization of these two types of outburst takes places not only once , but twice along the orbit ( see fig
. 2 top ) . this can be explained by the two peak accretion / ejection microquasar model for systems with an eccentric orbit like ls i + 61@xmath0303 ( e=0.54 - 0.7 ) ( @xcite and references within ) . in this model
an expected radio peak around periastron is attenuated by inverse compton losses due to the dense stellar uv field . in the first eight months observations with _ fermi_-lat , a related peak at high energies
is observed ( see fig . 2 at @xmath1=0.3 - 0.4 ) .
a second ( but smaller ) high energy peak is observed coincident with the large radio outburst around apastron ( @xmath1=0.65 - 0.75 ) . furthermore , as discussed in @xcite , the two peak shape of the @xmath11 vs. @xmath1 curve varies with @xmath2 , as does the distance of the two peaks ( see also fig . 2 here ) .
the two - peak microquasar model of @xcite predicts these variations in the accretion curve by incorporating changing wind velocities for the be star ( see their fig.6 ) .
this behaviour should be seen at other wavelengths as well , due to the connected emission processes .
in fact , after the first eight months , _ fermi_-lat detected an increase in the overall flux level ( @xmath6 0.92 ) and a broadening of the peak shape @xcite .
these variations are consistent with the observation that @xmath11 varies with @xmath2 .
variations are also seen with veritas and magic .
ls i + 61@xmath0303 was detected around apastron until 2008 and then became quiescent between 2008 - 2010 , where no detection was reported by veritas , while magic reported only weak detection around apastron .
then in october 2010 , veritas detected the source again , this time around periastron @xcite .
the strong connection between radio and he / vhe emission underlines therefore the importance of the radio periods for the analysis of e.g. integral data and can and should in principle be extended to other high energy instruments . | the most peculiar radio characteristics of the tev emitting high - mass x - ray binary ls i + 61@xmath0303 are two periodicities : a large periodic outburst which exhibits the same period as the orbit ( phase @xmath1 ) and a second periodicity of 1667 days ( phase @xmath2 ) which modulates the orbital phase and amplitude of the large outburst .
recent analysis of the radio spectral index present strong evidence for the presence of the critical transition from optically thick emission ( related to a steady jet ) to an optically thin outburst ( related to a transient jet ) as in other microquasars . in parallel , a switch from a low / hard x - ray state to a transitional state ( e.g. steep power law state ) would be expected .
we show how the critical transition from optically thick emission to an optically thin outburst is modulated by @xmath2 .
folding over a too large @xmath2 interval mixes up important information about the outbursts and can yield a false picture of the emission behaviour of the source along the orbit .
we therefore analyse the implications of this long period for treatment of hard x - ray / high energy data obtained from ls i + 61@xmath0303 , e.g. with _ fermi_-lat or integral , taking into account this long - term periodicity . |
in this appendix we give a derivation of the four fermi neutralino - quark effective lagrangian with cp violating phases given in the text . from the microscopic lagrangian of quark - squark - neutralino@xcite
= |q[c_qlp_l+c_qrp_r]+ |q[c^_qlp_l+c^_qrp_r]+h.c .
the effective lagrangian for @xmath58 scattering via the exchange of squarks is given by _
eff= |[c^*_qlp_r+c^*_qrp_l]q|q[c_qlp_l+c_qrp_r ] + + |[c^*_qlp_r+c^*_qrp_l]q |q[c^_qlp_l+c^_qrp_r ] we use fierz reordering to write the lagrangian in terms of the combinations @xmath59 , @xmath60 , @xmath61,@xmath62 , @xmath63 and @xmath64 . for this purpose , we define the 16 matrices ^a=\{1,^0,i^i , i^0_5 , ^i_5,_5,i^0i,^ij } : i , j=1 - 3 with the normalization tr(^a^b)=4^ab the fierz rearrangement formula with the above definitions and normalizations is ( |u_1^au_2)(|u_3^bu_4)=_c , df^ab_cd ( |u_1^cu_4)(|u_3^du_2 ) where @xmath65 are dirac or majorana spinors and f^ab_cd= -(+)tr(^c^a^d^b ) and where the + ve sign is for commuting u spinors and the -ve sign is for the anticommuting u fields . in our case
we have to use the -ve sign since we are dealing with quantum majorana and dirac fields in the lagrangian .
we give below the fierz rearrangement of the four combinations that appear in eq.(31 ) above : |q|q=-||qq- |_5|q_5q+|^ _ 5|q__5q + metric @xmath37 , and we also used the fact that @xmath66 is a majorana field so that @xmath67 and @xmath68 . by rearranging the terms to be in the form of @xmath69 of eq.(6 ) we can find out directly the squark contributions to @xmath70 and @xmath71 terms as given in the text of the paper . from the @xmath72 and @xmath73 interactions in eqs .
( c62 ) and ( c87a ) of ref.@xcite , we obtain the following effective lagrangian for the @xmath58 scattering via z - exchange : _ eff= [ |x_30|^2-|x_40|^2]|q^[d_qlp_l+ d_qrp_r]q|__5where @xmath74 and @xmath75 . from eq.(37 ) we can read off the contribution to a and b from the z exchange .
these contributions are given in eqs.(7 ) and ( 8) .
higgs exchange will contribute to @xmath76 and @xmath71 terms .
from the interaction lagrangian of @xmath77 and @xmath78 in eqs .
( 4.47 ) and ( 4.10 ) respectively of ref.@xcite , one can get the effective lagrangian for @xmath58 scattering via @xmath47 , @xmath48 and @xmath52 exchanges . in our formalism
we use @xmath47 , @xmath48 and @xmath52 for the light , heavy and cp - odd neutral higgs .
there are six contributions : three higgs exchange terms for the up flavor and three for the down flavor .
to illustrate we choose the up quark scattering with @xmath66 via the exchange of the heavy higgs @xmath48 ( @xmath79 in the notation of ref.@xcite ) : _ eff=(j^1_h0+j^2_h0)i^u_h0 where j^1_h0=-| ( q^"*_00p_l+q^"_00p_r ) + j^2_h0=| ( s^"*_00p_l+s^"_00p_r ) + i^u_h0=-|uu where @xmath80 are as defined in ref.@xcite . defining @xmath51 by
= q^"_00-s^"_00we get the @xmath48 contribution to @xmath69 : _ eff = re ( ) ||uu + + i m ( ) |_5|uu from eq.(41 ) we can read off directly the contributions @xmath81 and @xmath82 as given by eqs.(22 ) and ( 26 ) .
here we compare our results with those of ref.@xcite which is in the limit of no cp violation , no sfermion mixing and no heavy higgs . in the limit of no cp violation and no sfermion mixing @xmath83 given by eqs.(9 - 12 ) in the text
reduce to the following : to express the above in the notation of ref.@xcite we set @xmath85 , @xmath86 , @xmath87 , and @xmath88 keeping in mind that the @xmath18 and @xmath17 of ref.@xcite are defined oppositely to our notation . using the above along with eqs.(13 - 16 ) we find for @xmath89 to compare our @xmath96 term with that of ref.@xcite we again go to the limit of vanishing cp violating phases , assume no sfermion mixing , and in addition
ignore the heavy higgs exchange contribution ( i.e. , the term @xmath97 of eq.(22 ) in the text ) .
then using similar notational changes as above we find that our @xmath96 under the approximations made in ref.@xcite is given by @xmath98 comparing our results for a , b and c with those of ref.@xcite we find that our z , sfermion and higgs exchange terms have an overall minus sign relative to those of ref.@xcite . for a review
see , p. nath , r. arnowitt and a.h .
chamseddine , applied n=1 supergravity , trieste lectures , 1983(world scientific , singapore,1984 ) ; h. p. nilles , phys .
* 110 * , 1(1984 ) ; h.e .
haber , g.l.kane .
phys.rep . * 117 * , 195(1985 ) .
m. w. goodman and e. witten , phys .
d31 , 3095 ( 1985 ) ; i. wasserman , ibid 33 , 2071(1986 ) ; k. freese , j. frieman , and a. gould , ibid .
37 , 3388(1988 ) ; k. griest , ibid .
37 , 2703(1988 ) ; m. srednicki and r. watkins , phys .
b225 * , 140(1989 ) ; m. kamionkowski , phys . rev . * d44 * , 3021(1991 ) .
g. gelmini , p. gondolo , and e. roulet , nucl . phys . *
b351 * , 623(1991 ) ; j. engel , s. pittel and p. vogel , int . j. mod .
phys.e1 , 1(1992 ) ; t. ressel et.al .
, phys . rev . * d48 * , 5519(1993 ) ; v. bednyakov , h.v .
klapdor - kleingrothaus and s. kovalenko , phys . rev .
* d50 * , 7128(1994 ) ; a. bottino et.al .
, astropart . phys .
* 2 * , 77(1994 ) .
nath and r.arnowitt , phys .
lett . * 74 * , 4592(1995 ) ; phys . rev . *
d56 * , 2820(1997 ) ; r. arnowitt and p. nath phys .
rev.*d54 * , 2374 ( 1996 ) ; e. diehl , g. kane , c. kolda and j. wells , phys .
* d52 * , 4223(1995 ) ; l. bergstrom and p. gondolo , astropart .
phys . * 5 * , 263(1996 ) ; h. baer and m. brhlik , phys . rev . *
d57 * , 567(1998 ) .
the following corrections should be included in refs.@xcite : in the first paper of ref.@xcite a minus sign should be inserted on the right hand side of eq.(16 ) . in the expressions of @xmath99 ( or @xmath100 ) in refs.@xcite the first @xmath101 ( or @xmath102 ) should be preceded by a minus sign .
j. gunion and h. haber , nucl .
b272,1(1986 ) ; e : nucl . phys .
b402,567(1993 ) .
we thank toby falk for a communication drawing our attention to the erratum on eq.(5.5 ) of ref.@xcite which leads to the negative sign on the right hand side in eq.(16 ) . | a full analytic analysis of the effects of cp violating phases on the event rates in the direct detection of dark matter in the scattering of neutralinos from nuclear targets is given .
the analysis includes cp violating phases in softly broken supersymmetry in the framework of the minimal supersymmetric standard model ( mssm ) when generational mixings are ignored .
a numerical analysis shows that large cp violating phases including the constraints from the experimental limits on the neutron and the electron electric dipole moment ( edm ) can produce substantial effects on the event rates in dark matter detectors .
22 2 ^ 2_w 2m_z^2 2b2 v#1v_#1 # 1m_#1 # 1m_#1 # 1m_^0_#1 # 1m_^+_#1 2f_2(q^2 ) * effects of cp violation on event rates in the direct detection of dark matter * + .25 in utpal chattopadhyay@xmath0 , tarek ibrahim@xmath1 , & pran nath@xmath2 _ a department of physics , indiana university , bloomington , in 47405 , usa + b department of physics , faculty of science , university of alexandria , alexandria , egypt + c department of physics , northeastern university , boston , ma 02115 , usa + _ supersymmetric theories with r parity conservation imply the existence of a lowest mass supersymmetric particle ( lsp ) which is absolutely stable@xcite .
in supergravity models@xcite with r parity invariance one finds that over a majority of the parameter space the lsp is the lightest neutralino , and thus the neutralino becomes a candidate for cold dark matter ( cdm ) in the universe .
a great deal of dark matter exists in the halo of our galaxy and estimates of the density of dark matter in our solar neighborhood give densities of 0.3gev/@xmath3 and particle velocities of @xmath4 .
one of the interesting suggestions regarding the detection of dark matter is that of direct detection via scattering of neutralinos from target nuclei in terrestrial detectors@xcite .
considerable progress has been made in both technology of detection@xcite as well as in accurate theoretical predictions of the expected event rates in detectors such as ge , nai , caf@xmath5 , and xe@xcite . in this paper
we discuss the effects of cp violating phases in supersymmetric theories on event rates in the scattering of neutralinos off nuclei in terrestrial detectors .
such effects are negligible if the cp violating phases are small .
indeed the stringent experimental constraints on the edm of the neutron@xcite and of the electron @xcite would seem to require either small cp violating phases@xcite or a heavy supersymmetric particle spectrum@xcite , in the range of several tev , to satisfy the experimental limits on the edms .
however , a heavy sparticle spectrum also constitutes fine tuning@xcite and further a heavy spectrum in the range of several tev will put the supersymmetric particles beyond the reach of even the lhc .
recently a new possibility was proposed@xcite , i.e. , that of internal cancellations among the various components contributing to the edms which allows for the existence of large cp violating phases , with a susy spectrum which is not excessively heavy and is thus accessible at colliders in the near future .
cp violating phases o(1 ) are attractive because they circumvent the naturalness problem associated with small phases or a heavy susy spectrum .
the edm analysis of ref.@xcite was for the minimal supergravity model which has only two cp violating phases .
the analysis of ref.@xcite was extended in ref.@xcite to take account of all allowed cp violating phases in the minimal supersymmetric standard model ( mssm ) with no generational mixing .
this extension gives the possibility of the cancellation mechanism to occur over a much larger region of the parameter space allowing for large cp violating phases over this region .
indeed a general numerical analysis bears this out@xcite .
large cp violating phases can affect significantly dark matter analyses and other phenomena at low energy@xcite .
a detailed analysis in ref.@xcite shows that large cp violating phases consistent with the cosmological relic density constraints and the edm constraints using the cancellation mechanism are indeed possible .
cp violating phases affect event rates in dark matter detectors and a partial analysis of these effects with two cp phases and without the edm constraint was given in ref.@xcite .
thus , currently there are no analyses where the effect of large cp violating phases on event rates and the simultaneous satisfaction of the edm constraints via the cancellation mechansim are discussed .
further , the previous analyses are all limited to two cp violating phases while supergravity models with non - universalities and mssm can possess many more phases@xcite . in this paper
we give the first complete analytic analysis of the effects of cp violation on event rates with the inclusion of all cp violating phases allowed in the minimal supersymmetric standard model ( mssm ) when intergenerational mixings are ignored .
we then give a numerical analysis of the cp violating effects on event rates with the inclusion of the edm constraints .
it is shown that while the effects of cp violating phases on event rates are significant with the inclusion of the edm constraints , they are not enormous as for the case when the edm constraints are ignored .
we discuss now the details of the analysis in mssm .
the nature of the lsp at the electro - weak scale is determined by the neutralino mass matrix which in the basis ( @xmath6 , @xmath7 , @xmath8 , @xmath9 ) , where @xmath6 and @xmath7 are the u(1 ) and the neutral su(2 ) gauginos , is given by ( ) here @xmath10 is the z boson mass , @xmath11 is the weak angle , @xmath12=@xmath13 where @xmath14 = @xmath15 @xmath16 ( i=1,2 ) where @xmath17 is the higgs that gives mass to the up quarks and @xmath18 is the higgs that gives mass to the down quarks and the leptons , @xmath19 is the higgs mixing parameter ( i.e. it appears in the superpotential as the term @xmath20 ) , @xmath21 and @xmath22 are the masses of the u(1 ) and su(2 ) gauginos at the electro - weak scale with @xmath23 and @xmath24 being their phases .
the neutralino mass matrix of eq.(1 ) contains several phases .
however , it can be shown @xcite that the eigenvalues and the eigenvectors of the neutralino mass matrix depend on only two combinations : @xmath25= @xmath26+@xmath27 + @xmath28+@xmath29 and @xmath30=@xmath31
. the neutralino matrix can be diagonalized by a unitary matrix x such that x^t m_^0 x = diag(1 , 2 , 3 , 4 ) we shall denote the lsp by the index 0 so that ^0=x_10^*b+x_20^*w+ x_30^*h_1 + x_40^*h_2 the basic interactions that enter in the scattering of the lsp from nuclei are the neutralino - squark - quark interactions in the s channel and the neutralino - neutralino - z(higgs ) interactions in the cross channel .
the squark mass matrix @xmath32 involves both the phases of @xmath19 and of the trilinear couplings as given below ( ) here @xmath33 for q = u(d ) and @xmath34 for q = u(d ) , and @xmath35 is the quark mass .
the squark matrix is diagonalized by @xmath36 such that d_q^m_^2 d_q = diag(m_1 ^ 2 , m_2 ^ 2 ) as mentioned in the introduction the relative velocities of the lsp hitting the targets are small , and consequently we can approximate the effective interaction governing the neutralino - quark scattering by an effective four - fermi interaction .
we give now the result of our analysis including all the relevant cp violating effects in a softly broken mssm .
the effective four fermi interaction is given by _
eff=| _ _ 5 |q ^ ( a p_l + b p_r ) q+
c|m_q |q q + d |_5m_q |q_5 q + + e|i_5m_q |q q + f|m_q |qi_5 q where our choice of the metric is @xmath37 .
the deduction of eq.(6 ) starting from the microscopic supergravity lagrangian is given in appendix a. here we give the results .
the first two terms ( @xmath38 , @xmath39 ) are spin - dependent interactions and arise from the @xmath40 boson and the sfermion exchanges . for these
our analysis gives a=[|x_30|^2-|x_40|^2][t_3q-
e_q sin^2_w ] - - b=-[|x_30|^2-|x_40|^2 ] e_q sin^2_w + + where c_ql= ( _ q0 d_q11 -_q0 d_q21 ) c_qr= ( _ q0 d_q11 -_q0 d_q21 ) c^_ql= ( _ q0 d_q12 -_q0 d_q22 ) c^_qr= ( _ q0 d_q12 -_q0 d_q22 ) and where @xmath41 , @xmath42 , @xmath43 , and @xmath44 are given by@xcite _
u(d)j= _ u(d)j = eq_u(d)jx^*_1j+x^*_2j ( t_3u(d)-q_u(d)^2_w ) _ u(d)j = eq_u(d)jx_1j- x_2j _ u(d)j= here g is the @xmath45 gauge coupling and x_1j = x_1j_w+x_2j_w x_2j =- x_1j_w+x_2j_w the effect of the cp violating phases enter via the neutralino eigen - vector components @xmath46 and via the matrix @xmath36 that diagonalizes the squark mass matrix . the c term in eq.(6 )
represents the scalar interaction which gives rise to coherent scattering .. it receives contributions from the sfermion exchange , from the cp even light higgs ( @xmath47 ) exchange , and from the cp even heavy higgs ( @xmath48 ) exchange .
thus @xmath49 where c_(u , d)= - re[c_qlc^*_qr ] - re[c^_qlc^*_qr ] c_h^0(u , d)=-(+ ) re c_h^0(u , d)= re here @xmath41 is the higgs mixing angle , ( u , d ) indicate the flavor of the quark involved in the scattering , and @xmath50 and @xmath51 are defined by = x_40^*(x_20^*-_w x_10^*)+x_30^*(x_20^*-_w x_10^ * ) = - x_40^*(x_20^*-_w x_10^*)+x_30^*(x_20^*-_w x_10^*)the last three terms d , e and f in eq.(6 ) are given by d(u , d)= c_(u , d ) + ree(u , d)=t_(u , d)+ [ -(+ ) + ] f(u , d)=t_(u , d ) + imwhere @xmath52 is the cp odd higgs and where @xmath53 is defined by = -x_40^*(x_20^*-_w
x_10^*)+x_30^*(x_20^*-_w x_10^*)and t_(q)= im[c_qlc^*_qr ] + im[c^_qlc^*_qr ] in the limit of vanishing cp violating phases our results of a , b and c limit to the result of reference@xcite . in that limit
we have a difference of a minus sign in the z - exchange terms of ref .
@xcite in their equations ( 2a , 2b , a1 ) .
further , in the same limit of vanishing cp violating phases our results go to those of ref .
@xcite with an overall minus sign difference in the three exchange terms , i.e. , the z , the sfermions and the higgs terms .
( see appendix b for details ) . to compare our results to those of reference@xcite we note that the analysis of reference@xcite was limited to the case of two cp violating phases and it gave the analytic results for only some of the co - efficients in the low energy expansion of eq.(6 ) .
the terms e and f given by eqs.(26 ) and ( 27 ) are new and vanish in the limit when cp is conserved .
the term d given by eq.(25 ) is non - vanishing in the limit when cp phases vanish .
however , this term is mostly ignored in the literature as its contribution is suppressed because of the small velocity of the relic neutralinos .
in fact the contributions of d , e and f are expected to be relatively small and we ignore them in our numerical analysis here . for the computation of the event rates from nuclear targets in the direct detection of dark matter
we follow the techniques discussed in ref.@xcite and we refer the reader to it for details .
we give now the numerical estimates of the cp violating effects on event rates .
first we consider the case when the edm constraint is not imposed .
in fig.1 we exhibit the ratio @xmath54 for ge where r is the event rate with cp violation arising from a non - vanishing @xmath29 and r(0 ) is the event rate in the absence of cp violation .
the figure illustrates that the effect of the cp violating phase can be very large .
in fact , as can be seen from fig.1 the variations can be as large as 2 - 3 orders of magnitude .
however , as noted above the edm constraint was not imposed here .
next we give the analysis with inclusion of the edm constraints .
for this purpose we work in the parameter space where the cosmological relic density constraint and the edm constraints are simultaneously satisfied and we compute the ratio r / r(0 ) for ge in this region .
specifically the satisfaction of the relic density and the edm constraints is achieved by varying the parameters of the theory .
the satisfaction of the edm constraint is achieved through the cancellation mechanism discussed in refs.@xcite .
the result of this analysis is exhibited in fig.2 .
here we find the range of variation of r / r(0 ) with @xmath29 to be much smaller although still substantial .
thus from fig.2 we find that the variation of r / r(0 ) has a range of up to a factor of 2 over most of the allowed parameter space satisfying the relic density and the edm constraints in the regions of the parameter investigated .
this variation is substantially smaller than the one observed in fig.1 when the edm constraints were not imposed . in conclusion ,
we have given in this paper the first complete analytic analysis of the effects of cp violating phases on the event rates in the direct detection of dark matter within the framework of mssm with no generational mixing .
we find that the cp violating effects can generate variations in the event rates up to 2 - 3 orders of magnitude .
however , with the inclusion of the edm constraints the effects are much reduced although still significant in that the variations could be up to a factor @xmath55 as seen from the analysis over the region of the parameter space investigated .
of course the parameter space of mssm is quite large and there may exist other regions of the parameter space of mssm where the cp violating effects on event rates consistent with the edm constraints are even larger .
this research was supported in part by nsf grant phy-96020274 . in the mssm case
when tan@xmath42=2 , @xmath56=100 gev , @xmath57=1 for the cases when the gluino mass is 500 gev ( solid curve ) , 600 gev ( dotted curve ) , and 700 gev ( dashed curve),width=240 ] for the parameter space discussed in the text .
the parameter space spans regions satisfying the relic density and the edm constraints obtained by varying other parameters in the theory .
, width=240 ] |
the rare earth intermediate valence ( iv ) compounds@xcite are moderately heavy fermion compounds where the characteristic ( kondo ) energy is large compared to the crystal field splitting ( @xmath6 ) .
unlike the truly heavy fermion ( hf ) compounds@xcite ( where @xmath7 ) the iv compounds do not reside close to a quantum critical point for a transition to antiferromagnetism , so non - fermi liquid behavior is neither expected nor observed and the scaling behavior that is observed does not reflect proximity to such a phase transition.@xcite furthermore most of the iv compounds are cubic , so that anisotropy and low - dimensionality are not issues .
hence the iv compounds are physical realizations of the isotropic orbitally degenerate ( @xmath8 for yb ) anderson lattice model@xcite for three spatial dimensions .
this is an archetypal problem that is both simple and elegant for exploring the physics of electronic correlations in solids .
a key issue for the anderson lattice is the role of lattice coherence , which can be thought of as dispersive or band - like behavior of the 4@xmath2 electrons or alternatively as correlations between the 4@xmath2 electrons on different lattice sites .
this contrasts the anderson lattice ( al ) with the anderson impurity model ( aim ) where no such coherence is present . for the hf and iv compounds
the transport behavior , which depends crucially on the periodicity of the scattering potential , clearly manifests lattice coherence : the low temperature resistivity ( which is finite for the aim ) vanishes for the periodic compounds , as expected for a system obeying bloch s law;@xcite the de haas - van alphen signals are characteristic of renormalized @xmath2 bands;@xcite and the optical conductivity exhibits a drude response at low temperatures that also reflects renormalized masses .
@xcite on the other hand , the dynamic susceptibility @xmath9 shows a lorentzian power spectrum with very little @xmath10-dependence,@xcite suggesting that the spin / valence fluctuations are very local and uncorrelated , as expected for the aim ; and thermodynamic properties that are dominated by the spin / valence fluctuations such as the temperature dependence of the susceptibility @xmath11 , specific heat @xmath12 and 4@xmath2 hole occupation number @xmath3 ( the yb valence is @xmath13 = 2+@xmath14 ) seem to follow the predictions of the aim , at least qualitatively.@xcite recent theory@xcite of the anderson lattice suggests , however , that there should be observable differences between the behavior of the al and the aim for these quantities
. in particular , `` protracted screening '' can occur in the al , which means that the crossover from the low temperature fermi liquid state to the high temperature local moment state is slower for the lattice case than for the impurity case .
protracted screening has been invoked@xcite to explain photoemission results@xcite which show far less temperature dependence than expected based on the aim .
such results are controversial.@xcite to search for such effects in the bulk thermodynamic behavior we herein compare our measurements@xcite of @xmath11 , @xmath12 and @xmath15 for the series of related cubic ( c15b ) compounds yb@xmath4cu@xmath16 ( @xmath4 = ag , cd , in , mg , tl and zn ) to the predictions of the anderson impurity model , calculated within a single approximation scheme , the non - crossing approximation ( nca ) . by making this comparison for several measurements for several related compounds , we put strong constraints on the applicability of the model .
in addition , we include measurements of the dynamic susceptibility @xmath17 for @xmath4 = ag , mg , tl and zn ; these allow us to determine whether the deviations from aim behavior could arise from crystal field effects and also give a fourth experiment for comparison to the predictions of the model . finally , we include measurements of the x - ray absorption fine structure ( xafs ) from the @xmath4-atom @xmath18 edges for yb@xmath4cu@xmath16 and @xmath4 = ag , cd and in and the l@xmath19 edge for @xmath20 tl to determine whether deviations from aim behavior could arise from local lattice disorder in the samples .
the data for the 4_f _ occupation number @xmath3 ( fig .
1 ) , the susceptibility @xmath11 ( fig .
2 ) and the linear coefficient of specific heat @xmath21 ( table 1 ) are taken from sarrao _
et al_.@xcite as explained in that paper , the samples were small , high quality single crystals grown in @xmath4cu flux .
the specific heat was measured using a thermal relaxation technique .
the susceptibility was measured using a squid magnetometer . in this work
we have subtracted a small `` curie tail '' ( with curie constant typically of order 10@xmath22emu - k / mol ) from the susceptibility .
the occupation number @xmath3 was determined from the near - edge structure in yb @xmath23 x - ray absorption measurements .
the neutron scattering measurements were performed in the time - of - flight mode using the lrmecs spectrometer at ipns ( argonne national laboratory ) .
the experimental conditions were similar to those discussed in a recent study @xcite of ybincu@xmath16 ; we refer the reader to that publication for a more detailed account .
the temperature of the measurement was 10k , which is considerably smaller than @xmath24 in each case .
the samples utilized in these experiments were powders , typically of 50 g mass , grown from the melt inside evacuated tantalum tubes for @xmath4 = ag , mg and zn ; for @xmath4 = in and tl , a large number of single crystals grown as described in sarrao _
et al._@xcite were powdered to form the sample .
the susceptibilities of these samples were identical to those shown in fig .
[ chi_fig ] except for @xmath4 = ag , where the temperature @xmath25 of the maximum in the susceptibility was 10% smaller for the neutron sample , and for @xmath4 = tl where a fraction of the crystals used to create the powder had large curie tails indicative of disorder or impurities . for each sample ,
given that the elastic energy resolution ( fwhm ) varies as @xmath26 @xmath27 0.07 - 0.1 _ e_@xmath28 , we used several incident neutron energies @xmath29 ( e.g. 35 , 80 and 150mev ) to give greater dynamic range in energy transfer @xmath30 for the measurement . to improve statistics we took advantage of the lack of @xmath10-dependence of the magnetic scattering@xcite and grouped detectors into three bins , with average scattering angle 20@xmath31 ( low _ q _ ) , 60@xmath31 and 100@xmath32 ( high _ q _ ) .
the scattering was put on an absolute scale by comparison to the measured scattering of a vanadium sample .
we also measured the scattering of yagcu@xmath16 , lumgcu@xmath16 , ymgcu@xmath16 , ytlcu@xmath33 and luzncu@xmath16 in order to help determine the nonmagnetic scattering .
the xafs experiments were performed on beam line 4 - 3 at the stanford synchrotron radiation laboratory ( ssrl ) using a half - tuned si(220 ) double - crystal monochromator .
samples were ground , passed through a 30 @xmath34 m sieve and brushed onto scotch tape .
pieces of tape were stacked such that the change in the absorption at the yb @xmath35 edge was @xmath36 unity .
the samples were the same flux - grown crystals as used in sarrao _
et al_.@xcite and whose susceptibilities are shown in fig .
[ chi_fig ] . for temperature control ,
we utilized a lhe flow cryostat .
the predictions of the anderson impurity model were calculated in the non - crossing approximation ( nca).@xcite this approximation allows for calculation of all relevant experimental quantities , both static ( @xmath37 , @xmath38 , @xmath21 ) and dynamic ( @xmath39 ) , for realistic orbital degeneracy and for realistic spin - orbit and crystal - field splitting .
it agrees well@xcite over a broad temperature range with calculations using the bethe ansatz , but does produce spurious and unphysical non - fermi liquid artifacts in the low frequency density of states and in the dynamic susceptibility at very low temperatures.@xcite these artifacts are not significant for the temperatures and frequencies studied here .
the conduction band was assumed to have a gaussian density of states of width @xmath40 and centered at the fermi energy , i.e. @xmath41 . the hybridization matrix elements were assumed to be @xmath42-independent , i.e. @xmath43 , which may be a good approximation if the thermodynamic and magnetic properties are predominately determined by the conduction band states within @xmath44 of the fermi energy .
the finite temperature nca self - consistency equations @xcite were solved using three overlapping linear meshes . the thermodynamic and spectroscopic results for ce impurities with spin - orbit splitting and in various crystal fields were compared with those published in bickers _
et al_@xcite and good agreement was found despite the difference in procedure and the different choice of conduction band density of states . for yb iv compounds where
crystal fields can be ignored the magnetic orbital degeneracy is @xmath45 8 and there are essentially four input parameters for the aim calculation : the spin - orbit splitting ( which we fixed at @xmath46 1.3ev , the value typically observed in photoemission experiments@xcite ) , the width @xmath40 of the gaussian conduction band , the hybridization constant @xmath47 and the @xmath2-level energy ( relative the fermi level ) @xmath48 . since it is only the conduction states within @xmath49 of the fermi level that contribute significantly to the experimental quantities discussed here , the bandwidth parameter @xmath40 must be chosen to give the correct value of @xmath50 for the background conduction band . assuming the background bandstates in yb@xmath4cu@xmath16 are similar to those in lu@xmath4cu@xmath16 , we choose @xmath40 to reproduce the linear specific heat coefficient @xmath51lu@xmath52 of the corresponding lu@xmath4cu@xmath16 compounds,@xcite using the free electron approximation to convert the measured value of @xmath51lu@xmath52 to @xmath50 .
this is an important constraint since ( as we found in the earlier paper@xcite ) it is possible to force fits to @xmath3 over the whole temperature range if unrealistically small values of the bandwidth are used in the model . in yb@xmath4cu@xmath16 compounds ,
the valence bands have a width of order 10 ev;@xcite since the 2@xmath53 full - width of the gaussian band is @xmath54 , the values of @xmath40 ( @xmath55 , see table [ table1 ] ) obtained from @xmath51lu@xmath52 are quite reasonable .
once @xmath56 and @xmath40 are fixed , @xmath47 and @xmath48 can be uniquely determined by fitting to the ground state values of the susceptibility and occupation number .
the values of these parameters are given in table [ table1 ] ; the kondo temperature is calculated from the formula @xmath57 which includes the effect of spin orbit splitting but ignores crystal field splitting .
for these values of input parameters , we calculated the frequency dependence of the dynamic susceptibility at 10k and the temperature dependence of the static susceptibility , the 4f occupation number and the free energy . from the latter we determined the linear coefficient of specific heat by fitting to the formula @xmath58 in the temperature range @xmath59 ; the values of @xmath60 so determined have a systematic error of order 5 - 10% .
the primary goal of the neutron scattering measurements was to determine whether crystal field excitations can be resolved in these compounds .
determination of the magnetic scattering in polycrystals of iv compounds requires correct subtraction of the nonmagnetic scattering .
we have adopted a variant on the conventional procedure@xcite for accomplishing this where the y@xmath4cu@xmath16 ( and/or lu@xmath4cu@xmath16 ) nonmagnetic counterpart to the corresponding yb@xmath4cu@xmath16 sample is measured to determine the factor @xmath61 = @xmath62(high @xmath10 ; @xmath30;y)/@xmath62(low @xmath10 ; @xmath63y ) by which the high-@xmath10 scattering ( which is almost totally nonmagnetic scattering ) scales to the low @xmath10 values where the magnetic scattering is strongest .
this factor varies smoothly from a value @xmath274 - 5 , at @xmath64 to a value of 1 at large @xmath30 where @xmath10-independent multiple scattering dominates the nonmagnetic scattering .
( when this decrease in @xmath61 is neglected the nonmagnetic scattering will be underestimated at large energy transfer . ) in addition we have included an energy - independent multiplicative factor @xmath65 which is needed to account for the fact that the multiple scattering , which dominates the low-@xmath66 nonmagnetic scattering , has different strengths for yb than for y or lu .
( when this factor is neglected the magnetic scattering is overestimated . )
that is , we determine the magnetic scattering from the formula @xmath67 the magnetic scattering is related to the dynamic susceptibility through the formula @xmath68 where @xmath69 is the yb 4@xmath2 form factor .
for each @xmath65 we fit the dynamic susceptibility assuming that it is @xmath10-independent and assuming a lorentzian power spectrum : @xmath70})+(% \frac{1}{[(\delta e+e_{0})^{2}+\gamma ^{2}]})\}\ ] ] the prediction of the anderson impurity model for @xmath71 at @xmath72 also can be fit to this formula ; we have included in table 1 the values of @xmath73 and @xmath74 deduced from the fits to the nca results .
the fits to the experimental data include corrections for absorption and for instrumental resolution . the data for every @xmath75 is included in the fit .
we then find the value of @xmath65 that gives the smallest reduced @xmath76 for the fit .
the best - fit values of @xmath65 were consistently in the range 0.65 - 0.75 ; furthermore , the best - fit values of @xmath65 gave better agreement ( 10 - 20% ) between the fit value of @xmath77 and the value shown in fig .
[ chi_fig ] than was seen for @xmath78 .
use of this multiplicative factor also leads to excellent agreement between the values of magnetic scattering estimated at different incident energies . in fig .
[ neut_fig ] we show the data ( symbols ) for the magnetic scattering at low @xmath10 ( fixed average scattering angle @xmath7920@xmath80 ) and the fits ( solid lines ) .
( plots of the data for ybincu@xmath16 are given in the earlier publication.@xcite ) we note that in the time - of - flight experiment @xmath10 varies with both @xmath75 and @xmath30 at fixed @xmath81 and consequently the form factor , the absorption and the instrumental resolution all depend on @xmath75 ; this explains why the data and the fits for different @xmath75 do nt overlap .
the parameters @xmath73 and @xmath74 for the best fits are given in table 1 , where they are compared to the values predicted by the aim calculation .
we note that our results agree well with an earlier measurement of ybagcu@xmath16.@xcite although the fits are of reasonable statistical quality with reduced @xmath82 , the degree of systematic uncertainty due to the assumptions made about the nonmagnetic scattering is unknown .
the problem is especially serious for iv compounds where @xmath24 is large so that large incident energies are required , with the result that multiple scattering is large . to be cautious
, we expect a 10% error in the determination of @xmath73 and @xmath74 .
the primary goal of the xafs measurements was to determine whether @xmath4/cu site interchange is significant in these compounds .
such disorder is a real possiblity given that ybcu@xmath83 can grow in the c15b structure,@xcite and given that updcu@xmath16 , which also grows in the c15b struucture , has been shown to have significant pd / cu site interchange.@xcite our earlier neutron diffraction measurements@xcite of flux - grown crystals of ybincu@xmath16 , coupled with rietveld refinement , suggest the average crystal structure is very well ordered .
we employ the xafs technique here both because the measurement does not depend on lattice periodicity ( i.e. it is a local probe and is thus complimentary to diffraction ) and because it is atomic - species specific . in such a measurement
, a fine - structure function @xmath84 is extracted from the absorption data and related to the radial bond length distribution around the absorbing @xmath4 atomic species .
this measurement is particularly sensitive to @xmath4/cu site interchange because if any @xmath4 atoms rest on the cu sites , a short @xmath4-cu bond length at about 2.55 will appear that corresponds to the cu - cu nearest neighbor bond length in the nominal structure .
this bond length is significantly shorter than the nearest neighbor @xmath4-cu pairs at @xmath27 2.93 or the @xmath4-yb pairs at @xmath27 3.06 , and so is easy to resolve if the site interchange is frequent .
in addition to searching for @xmath4/cu site interchange , we also look for more generic disorder in the nominal @xmath4-cu and @xmath4-yb pairs by analyzing the distribution widths extracted from the @xmath85-space fits as a function of temperature between 15 - 300 k. bond length distribution widths should follow a correlated - debye model@xcite as a function of temperature if the structure does not have any static ( positional ) disorder .
this model differs from the usual debye model because it includes the correlations in the motions between the atoms in a given pair .
it has been shown to be accurate to within @xmath275% for metals.@xcite any static disorder should manifest as a constant offset , as is clearly shown , for example , for la@xmath86ca@xmath87mno@xmath88.@xcite the xafs data were fit in @xmath85-space , following procedures in ref .
initial fits included a short @xmath4-cu pair at @xmath892.55 .
the relative amplitude of this peak compared to the amplitude of the nominal @xmath4-cu at @xmath272.93 gives the percentage of @xmath90 atoms on cu sites .
errors in this measurement were determined by assuming that the fitted statistical @xmath76 was equal to the degrees of freedom in the fit as given by stern@xcite and then increasing the amount of site interchange until @xmath76 increased by one , while allowing all the other parameters in the fit to vary .
these results are presented in table [ xafs_table ] . for @xmath20 in , cd and ag
the site interchange has the same value as the error , and hence the results are consistent with the zero site interchange . for ybtlcu@xmath16 the error is smaller than the site interchange but it is also smaller than the error in the other cases .
furthermore , the fitted tl - cu bond length was very low ( 2.40 ) even though tl has the largest atomic radius of all the @xmath4 s measured .
we note that sytematic errors in the fit can arise from use of an incorrect theoretic backscattering amplitude @xcite . given all this ,
we think that the error is underestimated for ybtlcu@xmath16 .
therefore , we conclude that , within @xmath273% , none of the materials measured have any @xmath4/cu site interchange .
once the lack of site interchange was determined , we fit the data assuming no site interchange and determined the pair distribution widths to search for any other disorder .
examples of two of these fits are shown in fig .
[ xafs_fig ] .
the bond lengths of the nominal @xmath4-cu and @xmath4-yb pairs do not deviate from the average c15b structure assuming lattice parameters from ref . by more than 0.02 .
the distribution widths are shown in fig .
[ debye_fig ] , as are the fits to the correlated - debye model .
parameters of the correlated - debye fits are given in table [ xafs_table ] .
the fits are quite good and the observed values of the static disorder @xmath91 are typical of values seen in well - ordered compounds . the results of our calculations for @xmath11 and @xmath3 for the different yb@xmath4cu@xmath16 compounds are shown in figs .
[ nf_fig ] and [ chi_fig ] where they are compared to the experimental behavior.@xcite the most noteable feature is that for @xmath4 = cd , mg and zn , the experimentally determined occupation number @xmath3 and the effective moment @xmath92 ( where @xmath93 is the @xmath94 = 7/2 curie constant ) rise towards the high temperature limit more slowly than predicted by the impurity theory . the effect is somewhat weaker but still apparent for @xmath4 = ag . for @xmath4 = tl the experimental occupation number shows a weak retardation compared to theory but the experimental susceptibility tracks the impurity theory up to room temperature .
the linear coefficients of specific heat determined from our calculation are given in table 1 where they are compared to the experimental values ; the experimental and calculated values of the wilson ratio @xmath95 are also compared in the table .
figures [ nf_fig ] and [ chi_fig ] give our basic result : the crossover from the ground state fermi liquid to the high temperature local moment state is slower for the real materials than predicted by the anderson impurity model .
there are several reasons why this comparison to theory might be incorrect .
the first is that crystal fields may be large in these compounds .
crystal fields can cause the ground state multiplet to have a lower degeneracy than the value @xmath96 expected for @xmath97 .
since the aim results depend on @xmath98 this could lead to a discrepancy between theory and experiment if @xmath98 were incorrectly assigned in the calculation . indeed , in the earlier paper@xcite we found that for some cases we could fit the data assuming a lower degeneracy , e.g. @xmath99 for ybmgcu@xmath16 .
our inelastic magnetic neutron scattering experiments were designed to test for this possibility . as mentioned above
, there is a systematic error of order 10 - 20% in the determination of the peak position and width of the magnetic scattering . for our present purposes ,
however , the basic point is that the data can be well - fit by a single broadened lorentzian whose peak energy @xmath100 is in reasonable accord with the predicted value ( i.e. it is of order @xmath24 ) ; there are no sharp crystal field levels in evidence .
this also is in accord with the fact that the crystal fields seen in trivalent compounds such as ybinni@xmath16@xcite or ybaucu@xmath16@xcite are of order 2 - 4mev which is significantly smaller than @xmath24 for most of these compounds .
a second potential problem for our analysis is that if yb atoms are in a disordered environment , there can be a distribution of kondo temperatures . in this case
, the measured susceptibility involves an integral over this distribution of kondo temperatures , which would spread the crossover over a broader range of temperature.@xcite this effect may be relevant in updcu@xmath16 , where there is significant ( @xmath101 ) pd / cu site exchange.@xcite our xafs results for the yb@xmath4cu@xmath16 compounds show that within @xmath1023% there is no @xmath4/cu site interchange ; furthermore , the observed values of the static disorder @xmath103 are typical of values seen in well - ordered compounds
. basically , the xafs results are consistent with a very high degree of order for the flux - grown crystals .
we thus believe that kondo disorder is not a significant factor for these compounds .
( a possible exception is ybzncu@xmath16 for which we could not analyse the zn @xmath18-edge xafs because it is superimposed on the cu @xmath18-edge xafs . given the chemical similarity of zn and cu , zn / cu site disorder seems a real possibility . )
a third problem for this analysis concerns the interpretation of the yb @xmath104 x - ray absorption used to determine the occupation number n@xmath105(t ) .
the analysis@xcite relies on rather simple assumptions : the divalent and trivalent absorption edges are assumed to have the same shape and the same absorption matrix element , so that n@xmath105 is determined from the fractional weights of the two features .
similar assumptions are made for other methods ( xps core levels ; valence band ups ; mossbauer isomer shift ) for determining valence . while the resulting systematic error is not well - understood , it has been argued@xcite that the @xmath23 method is superior to use of isomer shift or lattice constant anomalies for the determination of valence .
it has been pointed out@xcite that the values of @xmath14 estimated from the @xmath23 experiments are consistently and significantly larger than those estimated from valence band photoemission .
the latter estimate also requires the assumption of equal absorption matirix elements for the di- and tri - valent configurations and is extremely sensitive to choice of background .
a point in favor of the use of x - ray absorption is that , unlike photoemission it truly samples the bulk of the compound ; indeed it has been argued@xcite that near - surface reconstruction gives rise to differences between the @xmath23 and ups estimates of @xmath14 . whether this is so and what the sytematic errors are in both measurements are , in our opinion , topics for future research . in any case , the deviations from the aim that we observe for @xmath3 using the @xmath23 measurement ( fig . 1 )
are qualitatively very similar to those observed for the effective moment ( fig .
2 ) , the measurement of which is not subject to significant systematic error .
( in our earlier paper@xcite we showed that at room temperature for all cases except @xmath20 ag , @xmath14 is nearly equal to the effective moment . )
the photoemission analysis yields an extremely slow crossover , to the extent that the hole occupancy is found to be @xmath106 in the high temperature state of ybincu@xmath16 which is much smaller than the value ( @xmath107 ) that we have observed@xcite using @xmath23 .
the latter value is , in our opinion , more realistic given the essentially trivalent behavior of the susceptibility of ybincu@xmath16 at high temperature .
we therefore argue that we have ruled out most of the major sources of error in our analysis , so that we can have a reasonable degree of confidence that the slow crossover is a real effect .
as mentioned in the introduction , recent theory@xcite for the ( @xmath62 = 1/2 ) anderson lattice , treated in the 1/@xmath108 approximation ( where @xmath108 is the spatial dimension of the system ) predicts such a slow crossover , referring to it as protracted screening .
the degree of slowness is sensitive to the filling of the background conduction band ; the crossover is slowest for partial filling of the background band . at present
it is unclear the extent to which the effect depends on the orbital degeneracy .
the deviations of the data ( figs .
1 and 2 ) for yb@xmath4cu@xmath16 from the predictions of the aim are strongest for @xmath20 mg , zn and cd , weaker for @xmath20 ag and weakest for @xmath20 tl .
we note that there is no correlation between this trend and the magnitude of the kondo temperature ( table 1 ) nor of the ground state occupation number ( fig .
there _ does _ appear to be a correlation with the magnitude of the hall coefficient for lu@xmath4cu@xmath16 : @xmath109 and @xmath110 for @xmath20 mg , zn , cd;@xcite @xmath111 for @xmath20 ag;@xcite and @xmath112 for @xmath20 tl.@xcite this quantity can be taken as a measure of the conduction electron density in the background band . in a one - band model , where @xmath113 and where there are 24 atoms in a unit cell of side 7.1 these values of @xmath114 correspond to the values 0.52 , 0.93 , 1.16 and 1.55 electrons per atom for the sequence @xmath115 mg , zn , cd , ag@xmath116 .
while the one - band assumption is unrealistic , nevertheless these results suggest that the deviations from the aim become weaker as the carrier density becomes larger .
this accords with theory@xcite which predicts no deviation for half filling of the background band and increasing deviation as the conduction electron density decreases . we should point out that a slow crossover is expected on rather general grounds . in the single impurity calculations
the position of the chemical potential @xmath117 is set by the host metal conduction band density of states , and the chemical potential is independent of temperature to order @xmath118 . in the hypothetical case of a periodic array of impurities that are noninteracting and incoherent the appreciable @xmath2-weight and
its asymmetric temperature dependence will lead to a significant temperature dependence of @xmath119 and hence lead to a deviation of @xmath3 from the behavior found in the solution to the single impurity model .
it is not clear to us the extent to which this effect contributes to the predicted protracted screening .
in addition to our main conclusion , we would like to point to two more features of the analysis .
the first is that the experimentally determined linear coefficients @xmath120 of specific heat for @xmath4 = mg , cd , ag and zn are 20 - 30% larger than the values predicted by the aim .
this means that the experimental value of the wilson ratio @xmath121 is smaller than predicted by the aim .
this has been noticed before.@xcite for ybincu@xmath16 there is excellent agreement between the experimental and calculated values ; for ybtlcu@xmath16 the calculated value of @xmath120 is 30% _ smaller _ than the experimental value .
the second feature is that there is considerable disagreement between the measured and predicted values of the neutron lineshape parameters @xmath73 and @xmath74 given in table 1 .
we remind the reader that the degree of systematic error for this determination is poorly understood , due to the difficulty of determining the nonmagnetic scattering for large @xmath24 compounds . in addition , all @xmath122 dependence of the lineshape in polycrystals is automatically averaged out . in the case of ybincu@xmath16 , we have confidence in the results , since an alternate means of subtracting the nonmagnetic scattering@xcite agrees with the method used here and since a neutron scattering study of a single crystal@xcite showed that the @xmath122-dependence of the lineshape is quite small . for that case
we note that the experimental peak energy @xmath73 agrees well with the prediction of the aim but the measured linewidth @xmath74 is a factor of two smaller than predicted , as though the spin fluctations have a longer lifetime than predicted by the aim . on the other hand for ybmgcu@xmath16 , ybtlcu@xmath16 and ybzncu@xmath16 the experimental values of peak energy differ significantly from the calculated values and the linewidths are _ larger _ than the predicted values .
such a large linewidth could be a consequence of a @xmath10-dependence of the peak parameter @xmath73(@xmath10 ) ; studies of single crystals are necessary to establish whether such a @xmath10-dependence is present .
finally we note that if the lineshape parameters for ybagcu@xmath16 are increased by 10% to account for the 10% smaller value of @xmath25 observed for the neutron sample , they are in excellent agreement with the aim predictions . in conclusion , we have given evidence that the crossover from low temperature fermi liquid behavior to high temperature local moment behavior is slower for periodic intermediate valence compounds than predicted by the anderson impurity model .
we have included supporting experiments to rule out the possibility that the disagreements between the data and the aim predictions are due to large crystal fields or to kondo disorder .
we feel that the results are fairly robust
. it will be interesting to see whether similar protracted screening is observed in other compounds , especially in those based on ce where the orbital degeneracy is smaller ( @xmath123 ) and where the size of the 4@xmath2 orbital is considerably larger .
work at uc irvine was supported by ucdrd funds provided by the university of california for the conduct of discretionary research by the los alamos national laboratory .
work at polytechnic was supported by doe fg02er84 - 45127 .
work at lawrence berkeley national laboratory was supported by the office of basic energy sciences ( obes ) , chemical sciences division of the department of energy ( doe ) , contract no
. de - ac03 - 76sf00098 .
work at los alamos was performed under the auspices of the doe .
work at argonne was supported by the doe office of science under contract no .
w-31 - 109-eng-38 .
the xafs experiments were performed at ssrl , which is operated by the doe / obes .
j. l. sarrao , c. d. immer , z. fisk , c. h. booth , e. figueroa , j. m. lawrence , r. modler , a. l. cornelius , m. f. hundley , g. h. kwei , r. movshovich , j. d. thompson and f. bridges , phys .
b * 59 * , 6855 ( 1999 ) . .
input parameters @xmath40 , @xmath48 and @xmath47 for the aim calculation ; the calculated kondo temperature @xmath24 ; and the theoretical and experimental values of the specific heat coefficient @xmath124 , the wilson ratio @xmath121 and the neutron lineshape parameters @xmath73 and @xmath125 .
[ cols= " < , < , < , < , < , < , < , < , < " , ] | we compare the results of measurements of the magnetic susceptibility @xmath0 , the linear coefficient of specific heat @xmath1 and 4@xmath2 occupation number @xmath3 for the intermediate valence compounds yb@xmath4cu@xmath5 ( @xmath4 = ag , cd , in , mg , tl , zn ) to the predictions of the anderson impurity model , calculated in the non - crossing approximation ( nca ) . the crossover from the low temperature fermi liquid state to the high temperature local moment state is substantially slower in the compounds than predicted by the nca ; this corresponds to the `` protracted screening '' recently predicted for the anderson lattice .
we present results for the dynamic susceptibility , measured through neutron scattering experiments , to show that the deviations between theory and experiment are not due to crystal field effects , and we present x - ray - absorption fine - structure ( xafs ) results that show the local crystal structure around the @xmath4 atoms is well ordered , so that the deviations probably do not arise from kondo disorder .
the deviations may correlate with the background conduction electron density , as predicted for protracted screening . |
a pin photodiode is a very useful device .
it is used not just for a photosensor but also for a detector of charged particles , x - rays and gamma - rays by a lot of experiments in high energy physics .
a pin photodiode is an improved version of the low - capacitance planar silicon photodiode and makes use of an extra high resistivity intrinsic silicon layer between a p- and an n@xmath1-layer to improve response time .
reversed bias enhances response time of the pin photodiode .
it has high resistance to breakdown and low leakage current .
we use pin photodiodes for an x - ray detector in the search experiment for solar axions ( the tokyo axion helioscope ( fig . [ sumico ] ) @xcite @xcite ) , because of a low sensitivity to magnetic fields , tolerance to low - temperature , and high efficiency for low energy x - rays @xcite . however , it has a slight disadvantage .
there is non - active area so - called ` a dead layer ' which consists of the sio@xmath2 layer formed on the detector surface as a protective film and the p - layer on a pin photodiode .
but these layers do not perform as depleted layers .
they have no sensitivity .
it could worsen the sensitivity to low energy x - rays in our experiment . in @xcite
, we reported that the thickness was @xmath3 at 95% confidence level .
but we thought the true value was much less than this one . in this work
, we determined the thickness of the insensitive surface layer of the pin photodiode to recalculate the upper limit of axion - photon coupling constant using the data obtained in our earlier experiments @xcite @xcite .
to measure the thickness of the insensitive surface layer , we made alpha - particles from an @xmath4am source impinge on a pin photodiode at various incidence angles .
the range for the 5.486 mev alpha - particle in silicon is @xmath5 and much less than the thickness of the expected depleted layer .
respective path lengths of alpha - particles in the insensitive and sensitive layers depend on their incidence angle .
therefore , it enables us to estimate the thickness of the insensitive layer measuring pulse heights of alpha - particles in the pin photodiode as a function of the incidence angle .
the experimental setup is shown in figs .
@xmath6 , @xmath7 and @xmath8 .
the experimental apparatus were put in a vacuum ( pressure less than 10@xmath9 pa ) .
we could rotate the pin photodiode from the outside of the vacuum to change the incidence angle .
the pin photodiode was exposed to 5.486 mev alpha - particles from the @xmath4am source installed in a brass holder .
the brass holder had a stainless steel pipe of @xmath104.3 mm inner diameter and 50 mm length as a collimator .
the distance between the source and the detector is 60 mm .
the pin photodiode we used is hamamatsu s3590 - 06 .
it is 11@xmath1111 mm@xmath12 in size and 500 @xmath13 m in thickness .
this pin photodiode is the same type as ones used in the tokyo axion helioscope @xcite@xcite .
it has high efficiency to x - rays of energies up to 10 kev , energy range of x - rays to which solar axions convert in magnetic fields .
we measured spectra for 8 incidence angles , @xmath14 , and @xmath15 ( fig .
[ spectrum ] ) .
each spectrum in fig .
[ spectrum ] exhibits peaks of 5.486 mev around 2900 ch .
the alpha - particle peak channels shifted lower with increasing incidence angles , suggesting an existence of the insensitive surface layer .
we determined the peak channel @xmath16 by fitting a gaussian function to spectrum of the incidence angle @xmath17 . in the fitting
, we use the upper half region of each peak , because the lower half contained the effect of the thickness of the @xmath4am source and the spread of incidence angles of alpha - particles .
@xmath18 shows the relation between the peak channels @xmath16 and incidence angles @xmath17 together with the best fit to the data by the equation , @xmath19 where @xmath20 is the thickness of the insensitive layer of the pin photodiode , @xmath21 is the energy of alpha - particles of @xmath4am , @xmath22 is the stopping power , the insensitive layer for the alpha - particles , and @xmath23 and @xmath24 are parameters of the relation between channels and energies ; @xmath25}= \alpha ( c\;{\rm [ ch ] } -\beta).\ ] ] since we do not know the percentage of si and sio@xmath2 in the insensitive layer , we calculated the thickness for the case it consisted of si and the case of sio@xmath2 . the stopping power of sio@xmath2 is a little larger than that of si , therefore , the estimate value of the thickness is a little smaller . in the fitting ,
@xmath20 and @xmath24 were left free and @xmath23 was fixed at 2.24 .
this value was determined by calibration with alpha - particles from @xmath4am in vacuum and decelerated alpha - particles from the same source in the air .
thereby we got @xmath26 with @xmath27 for si and 13% smaller for sio@xmath2 .
at worst , in the case that the insensitive layer consisted of si , the thickness is less than @xmath28 at 95% confidence level .
this is nearly @xmath29 of the value we conservatively assumed before @xcite .
if the insensitive layer were relatively thick as we supposed before , then the detection efficiency for the x - ray would be largely affected by absorption in it .
however , it is not the case because the insensitive layer is thin enough , less than 15% of the absorption length for 5 kev x - rays .
however , one still has to take into account loss of energy caused by escape of x - rays and auger electrons from the detector following photoelectric absorption of the incoming x - ray .
a monte carlo simulation was performed to estimate the peak detection efficiency for x - rays of energy range between 2 kev and 10 kev with the geant4 program @xcite . in the simulation we assumed that the insensitive layer consists of si .
the results are shown in fig .
[ simulation ] .
the errors for the estimation are too small to show on the data points .
the estimated peak detection efficiency is more than 95% in the energy range between 4 and 10 kev .
the decreasing efficiency toward low energy is due to the absorption in the insensitive layer and lower efficiency above 9 kev is caused by the insufficiency of the thickness of the depletion layer .
even if we simulated on the assumption that the insensitive layer consisted of sio@xmath2 , the result did not change significantly .
= 6.8 cm = 6.8 cm
we showed the thickness of an insensitive layer of a pin photodiode , s3590 - 06 produced by hamamatsu , is @xmath0 for si or 13% smaller for sio@xmath2 .
therefore , the upper limit of the thickness is @xmath30 at 95% confidence level in the case of the insensitive layer consisted entirely of si .
we calculated the peak efficiency of a pin photodiode for x - rays in consideration of the thickness of the insensitive layer and escape of x - rays and auger electrons from the detector .
this result showed the efficiency as more than 95% for 4 - 10 kev x - rays , which correspond to the energy range observed in the tokyo axion helioscope experiment @xcite @xcite .
00 s. moriyama , _ et al .
b * 434 * ( 1998 ) 147 .
y. inoue , _ et al .
b * 536 * ( 2002 ) 18 .
y. inoue , _ et al .
methods in phys .
research a * 368 * ( 1996 ) 556 . | we measured the thickness of an insensitive surface layer of a pin photodiode , hamamatsu s3590 - 06 , used in the tokyo axion helioscope .
we made alpha - particles impinge on the pin photodiode in various incidence angles and measured the pulse height to estimate the thickness of the insensitive surface layer .
this measurement showed its thickness was @xmath0 on the assumption that the insensitive layer consisted of si .
we calculated the peak detection efficiency for low energy x - rays in consideration of the insensitive layer and escape of x - rays and auger electrons .
this result showed the efficiency for 4 - 10kev x - rays was more than 95% .
, , pin photodiode , insensitive surface layer , thickness 85.60.dw , 14.80.mz |
the next generation of colliders , and especially the lhc , will boost the search for new physics , and in particular , the supersymmetry ( susy ) .
therefore , it seems important to derive new constraints on the properties of supersymmetric particles . for this purpose ,
a good strategy would be to look at where the standard model ( sm ) can be subdominant in comparison to the supersymmetry , and where qcd corrections are known with high accuracy and branching fractions can be measured at lhc even at initial low luminosities .
rare b decays appear to be _
ideal choices _ for the above goal .
for example , @xmath1 transitions are forbidden at the tree level in the sm . in this work
, we will concentrate on the related exclusive @xmath0 decay , within supersymmetry , considering msugra parameter space and the minimal flavor violation .
the main focus is on isospin asymmetry which , as we will see later , is a valuable observable to get constraints on supersymmetric parameters .
a more detailed analysis is presented in ref .
the isospin asymmetry for the exclusive process @xmath2 can be defined as : @xmath3 and @xmath4 can be obtained by using the charge conjugate modes .
we consider here recent data for exclusive decays from belle @xcite and babar @xcite : @xmath5 the effective hamiltonian for @xmath6 transitions can be written : @xmath7\;\;,\ ] ] where @xmath8 is the fermi coupling constant , @xmath9 are elements of the ckm matrix , @xmath10 are the operators relevant to @xmath2 and @xmath11 are the corresponding wilson coefficients evaluated at the scale @xmath12 .
the most relevant operators @xmath13 in supersymmetry are : @xmath14 where @xmath15 are the projection operators .
+ we consider the wilson coefficients at the next to leading order ( nlo ) in the strong coupling constant @xmath16 .
the contributions from the w boson , the charged higgs and the charginos , as well as the leading @xmath17 corrections to the w boson and the charged higgs are considered .
the contributions from the gluino and neutralino are neglected in our work , as they are known to be negligible in the minimal flavor violating scenario @xcite . for further details of the calculation of the wilson coefficients ,
please refer to @xcite . following the method of ref .
@xcite , one can write the isospin asymmetry as : @xmath18 the expression for @xmath19 , which is related to the wilson coefficients , can be found in @xcite .
+ we generate the susy mass spectrum using isajet-7.74 @xcite and we perform scans in the msugra parameter space ( @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath17 ) . for any msugra parameter space point , we then calculate the isospin asymmetry , and compare it to the combined experimental limits of eq.([babar ] ) and eq.([belle ] ) . for comparison
, we also perform the calculation of the inclusive branching ratio of @xmath24 following ref .
@xcite and compare to the experimental limits of @xcite .
the points which result in too small light higgs masses ( _ i.e. _ such as @xmath25 gev ) or which do not satisfy the constraints of ref .
@xcite are also excluded .
finally , we also test whether the lightest supersymmetric particle ( lsp ) is charged .
indeed , a charged lsp is cosmologically disfavored if the lsp is stable , but might be viable if r - parity is violated .
we perform scans of the msugra parameter space such that @xmath26 $ ] , @xmath27 $ ] , @xmath28 $ ] , @xmath29 $ ] and @xmath30 .
+ + an investigation of the @xmath31 plane for @xmath32 and @xmath33 , and for @xmath34 , is presented in figure [ a0 ] . in this figure ,
the area marked `` isospin '' corresponds to the region excluded by the isospin breaking constraints , whereas the area marked `` @xmath24 '' corresponds to the region excluded by the inclusive branching ratio constraints .
the `` excluded '' area corresponds to the case where at least one of the particle masses does not satisfy the constraints of ref .
@xcite . and finally , `` charged lsp '' is the cosmologically disfavored region .
the various colors represent the intensity of the isospin asymmetry .
+ we can note here that the isospin symmetry breaking provides stringent constraints on the parameter space .
it is more sensitive to the smaller values of @xmath22 , and one can remark that it also decreases with larger @xmath20 and @xmath21 .
furthermore , it appears that the constraints from isospin asymmetry are more stringent than the ones from inclusive branching ratio .
the results of fig .
[ a0 ] concern a fixed value for @xmath17 .
however , the isospin asymmetry increases with @xmath17 and for example , for @xmath35 , we can obtain valuable constraints at @xmath36 .
this behaviour is revealed in fig .
[ tanb ] , where the isospin asymmetry is depicted in function of @xmath17 and @xmath21 for @xmath37 and @xmath33 .
we note that again , isospin asymmetry reveals to be more restricting than the branching ratio . + for @xmath38 , with @xmath37 and @xmath33 .
the definitions of the different regions are in the text.,title="fig:",width=302,height=232 ] for @xmath38 , with @xmath37 and @xmath33 .
the definitions of the different regions are in the text.,title="fig:",width=302,height=232 ] f.m .
acknowledges the support of the mccain fellowship at mount allison university and a partial support from nserc , and would also like to thank the organizers of susy06 .
m.a.s research is partially funded by a discovery grant from nserc . | we study the isospin asymmetry as an observable in exclusive @xmath0 decay that can be used to study the effects of physics beyond the standard model .
we present predictions for isospin asymmetry in supersymmetric extension of the standard model with minimal flavor violation , using parameter values allowed by current experimental constraints .
this observable can provide very tight constraints on the msugra parameter space .
address = department of physics , mount allison university , 67 york street , sackville , + new brunswick , canada e4l 1e6 |
the quark model has been extremely successful in accounting for the observed meson and baryon states and their quantum numbers . ground
state baryons are grouped into an octet * 8 * and a decuplet * 10 * with total angular momentum @xmath9 and @xmath10 , respectively .
the successful prediction of the @xmath11 and of its mass @xcite and its subsequent experimental discovery @xcite was a cornerstone in establishing su(3 ) symmetry .
the pattern of excited baryons was understood in terms of three quarks obeying the pauli principle ( see , e.g. @xcite and refs . therein ) .
the discovery of the j/@xmath12 @xcite and @xmath13 @xcite families gave final support to the quark model .
the development of qcd as renormalizable field theory @xcite and the observation of the partonic structure of nucleons in deep inelastic scattering @xcite seemed to pave the way for an eventual understanding of the structure of hadronic matter .
a completely different approach to understand the nucleon was proposed by skyrme@xcite .
he suggested that the low - energy behaviour of nucleons can be viewed as spherically symmetric soliton solution of the pion field . in this view , the baryonic nature of the nucleon is not due to quarks carrying baryon quantum number @xmath14 .
instead , the baryon number is interpreted as topological quantum number of the pion field @xcite .
a fascinating consequence of this view is the prediction of an anti
decuplet @xmath15 .
it contains a state with exotic quantum numbers , with strangeness @xmath16 and @xmath9 @xcite .
walliser estimated the mass of this exotic state to be at about 1700mev @xcite .
_ assigned the @xmath17 to the non - strange member of the anti decuplet thus fixing the mass of the exotic @xmath0 to 1530mev @xcite , a value consistent with the prediction of weigel one year later @xcite . due to its intrinsic strangeness and low mass ,
the only allowed strong interaction decay modes of the @xmath0 are @xmath2 and @xmath18 .
its width is predicted to be narrow . in quark models
, the @xmath0 has a minimal quark content @xmath19 and is called pentaquark . in the soliton model , it is the lowest mass state beyond the conventional octet and decuplet baryons .
a recent survey can be found in @xcite .
these predictions stimulated an active search for the @xmath0 . at spring-8(leps ) , a @xmath20 @xmath2 peak was reported in the reaction @xmath21 @xcite .
cuts were applied to select quasi
free processes on the neutron .
after correction for the fermi momentum , the mass was determined to @xmath22mev . the observed width was compatible with the instrumental resolution .
an upper limit of @xmath23mev was derived .
the diana collaboration at itep @xcite reported a @xmath24 enhancement produced in low - energy @xmath25xe collisions , with @xmath26mev and a statistical significance of @xmath27 . at jlab ,
a significant @xmath0 peak was observed with the clas detector in the exclusive reaction @xmath28 .
the @xmath0 is photoproduced off the neutron and subsequently decays into @xmath2 . only in case of @xmath29 rescattering ,
the proton received sufficient momentum to be detected ; the neutron was identified by the missing mass @xcite .
the peak in the @xmath2 mass distribution had a statistical significance of @xmath30 ; it was found at a mass of @xmath31mev , its width was compatible with the resolution ( fwhm = 21mev ) .
the authors did not see a peak in the @xmath6 mass distribution .
since the clas acceptance for @xmath6 is not the same as that for @xmath2 , they cautiously suggested the @xmath0 should have isospin zero . a preliminary study of the @xmath32 final state produced by photoproduction off protons by the clas collaboration @xcite suggests a @xmath33 peak in the @xmath2 invariant mass distribution recoiling ( in the centre of mass system ) against a @xmath34 . in this letter we present evidence for the @xmath0 in photoproduction of the @xmath1 final state off protons .
a diagram which may contribute is shown in fig . [ diag1 ] .
the reaction @xmath35 was measured with the saphir detector @xcite at elsa .
saphir is a magnetic spectrometer covering a solid angle of approximately @xmath36 and the full polar range @xmath37 .
the elsa electron beam produced photons via bremsstrahlung in a copper foil radiator .
the energies of the scattered electrons were determined in the tagging system for photon energies from 31% to 94% of the incident electron energy which was 2.8 gev for the data presented in this letter .
the photon beam passed through a liquid hydrogen target ( 3 cm in diameter and 8 cm in length ) .
interacting photons were detected in a photon counter .
the coincidences of tagger and photon counter determined the photon flux .
the target was located in the centre of a central drift chamber with 14 cylindrical layers .
charged particles leaving the target were bent in the field of a c - shaped magnet allowing for a measurement of momentum and sign of charge .
the momentum resolution @xmath38 for 300mev / c particles in the central drift chamber was approximately 2.5% .
in forward direction a planar drift chamber improved the momentum resolution to about @xmath39% .
the central drift chamber was surrounded by a scintillator wall to measure the time of flight ( tof ) .
the minimum momentum to detect tracks of charged pions and protons was 50 and 150 mev / c , respectively , with a track detection threshold depending on the position of the production vertex in the target .
the data presented here were taken in 1997 and 1998 with a trigger requiring at least two charged particles .
about @xmath40 events were recorded .
the data were used to determine cross sections for photoproduction of strange mesons and baryons @xcite,@xcite,@xcite , and of @xmath41 @xcite , @xmath42 @xcite , and @xmath43 @xcite mesons .
we observe the @xmath0 in the reaction @xmath45 the neutron is identified in a kinematical fit using one constraint ; the neutron momentum is reconstructed from energy and momentum conservation .
2a shows the @xmath2 invariant mass distribution after cuts explained below .
a clear signal is observed at @xmath46mev above a smooth background .
the fit ascribes @xmath47 events ( @xmath3 ) to the peak . by counting the four bins around the @xmath0 centre we get 55 events above background and 56 below , corresponding to a statistical significance of @xmath48 . [ cols="^,^ " , ] [ pic : plusplus ]
the number of @xmath49 events we expect depends on the isospin : clebsch gordan coefficients favour @xmath49 over @xmath44 by a factor 3 ( for i=2 ) or 4 ( for i=1 ) while the @xmath49 does not exist in case of i=0 .
experimentally , the final state @xmath50 offers an additional factor 3 since 1/2 of all @xmath51 are produced as @xmath52 and escape undetected , and the @xmath53 is observed only with a fraction @xmath54 . in case
the @xmath44 is an isovector or isotensor we should observe 9 or 12 times more @xmath49 events , respectively . also , the saphir acceptance is considerably larger for the fully constrained @xmath55 events .
we thus expect a peak with more than 5000 @xmath49 in fig . 4 ( right ) , far above the observed level .
we conclude that the @xmath44 is isoscalar .
we have studied photoproduction of the @xmath1 final state . in the @xmath2 invariant mass distribution
we observe a @xmath3 peak at a mass of @xmath4mev .
the peak is strikingly narrow : we quote an upper limit ( at 90% c.l . ) of @xmath56mev .
mass and width are compatible with results obtained elsewhere . here , the @xmath0 is observed for the first time in the reaction @xmath57 .
the cross section for @xmath58 in the photon energy range from 1.7 to 2.6gev is of similar size as the cross section for @xmath59 photoproduction and smaller by a factor 4 than for other final states with open strangeness like @xmath60 , @xmath61 , or @xmath62 .
the @xmath0 has strangeness @xmath16 and can not be a three - quark baryon . in the quark picture , it is a pentaquark @xmath63 . in the related reaction @xmath7 ,
we have searched for a doubly charged isospin partner of the @xmath0 . from the absence of a signal at the expected strength
we deduce that the @xmath0 is an isoscalar .
we would like to thank the technical staff of the elsa machine group for their invaluable contributions to the experiment .
we gratefully acknowledge the support by the deutsche forschungsgemeinschaft in the framework of the schwerpunktprogramm `` investigations of the hadronic structure of nucleons and nuclei with electromagnetic probes '' ( spp 1034 kl 980/2 - 3 ) .
99 m. gell - mann , phys .
* 8 * ( 1964 ) 214 . v. e. barnes _
et al . _ ,
* 12 * ( 1964 ) 204 .
n. isgur and g. karl , phys .
b * 72 * ( 1977 ) 109 .
j. j. aubert _ et al .
_ , phys .
lett . * 33 * ( 1974 ) 1404 .
t. h. skyrme , proc .
lond . a * 260 * ( 1961 ) 127 .
t. h. skyrme , nucl .
* 31 * ( 1962 ) 556 .
e. witten , nucl .
b * 223 * ( 1983 ) 422 , 433 .
m. chemtob , nucl .
b * 256 * ( 1985 ) 600 .
h. walliser , nucl .
a * 548 * ( 1992 ) 649 .
d. diakonov , v. petrov and m. v. polyakov , z. phys .
a * 359 * ( 1997 ) 305 .
h. weigel , eur .
j. a * 2 * ( 1998 ) 391 .
h. walliser and v. b. kopeliovich , arxiv : hep - ph/0304058 .
s. capstick , p.
r. page and w. roberts , arxiv : hep - ph/0307019
. t. nakano _ et al .
_ [ leps collaboration ] , phys .
* 91 * ( 2003 ) 012002 .
v. v. barmin _ et al . _ [ diana collaboration ] , arxiv : hep - ex/0304040 .
s. stepanyan [ clas collaboration ] , arxiv : hep - ex/0307018 . v. kuberovski and s. stepanyan for the clas collaboration , proc . of `` conf . on the intersections of particle and nuclear physics '' ( cipanp 2003 ) , 19 - 24 may , new york , ny w. j. schwille _ et al .
_ , nucl .
instrum .
a * 344 * ( 1994 ) 470 . c. bennhold _ et al . _
[ saphir collaboration ] , nucl .
a * 639 * ( 1998 ) 209 .
m. q. tran _ et al . _
[ saphir collaboration ] , phys .
b * 445 * ( 1998 ) 20 .
s. goers _ et al . _
[ saphir collaboration ] , phys .
b * 464 * ( 1999 ) 331 .
j. barth _ et al .
_ [ saphir collaboration ] , low - energy photoproduction of @xmath41 mesons , eur .
j. a ( 2003 ) accepted for publication .
r. pltzke _ et al . _
[ saphir collaboration ] , phys .
b * 444 * ( 1998 ) 555 .
j. barth _ et al .
_ [ saphir collaboration ] , eur . phys . j. a * 17 * ( 2003 ) 269 . | the positive strangeness baryon resonance @xmath0 is observed in photoproduction of the @xmath1 final state with the saphir detector at the bonn electron stretcher accelerator elsa .
it is seen as a peak in the @xmath2 invariant mass distribution with a @xmath3 confidence level .
we find a mass @xmath4mev and an upper limit of the width @xmath5mev at 90% c.l . from the absence of a signal in the @xmath6 invariant mass distribution in @xmath7 at the expected strength
we conclude that the @xmath0 must be isoscalar .
+ _ pacs : 14.80.-j _ , , , , , , , , , , , , , , , , , , , , , .
other particles , pentaquark with strangeness @xmath8 . |
it has been known for some time that the renormalization of light - front field theories is an extremely challenging task , although systematic procedures have been developed over the years to handle these problems consistently @xcite , including boost - invariant regularization schemes @xcite .
relatively recently , it was observed that a different boost - invariant regularization scheme may be employed in a treatment of light - front matrix field theories with @xmath1 interactions @xcite , following earlier work on the small-@xmath2 behavior of light - cone wave functions @xcite . in this work ,
we follow the approach suggested in @xcite to determine the effective two - loop mass renormalization of the @xmath0 light - front hamiltonian . in section [ themodel ] , we begin by presenting the light - front formulation of scalar @xmath0 field theory in @xmath3 dimensions , and its subsequent quantization via commutation relations . in section [ boundstate ] ,
we employ a manifestly boost - invariant procedure to extract the effective two - loop mass renormalization of the light - front hamiltonian .
this result will be compared to a feynman diagram calculation that can be performed straightforwardly in two dimensions .
a brief discussion of our results will appear in section [ the - end ] .
consider the @xmath4-dimensional field theory described by the action @xmath5 , \label{phi4model}\ ] ] where @xmath6 is a scalar field defined on the @xmath4-dimensional minkowski space @xmath7
. working in the light - cone coordinate frame @xmath8 one may derive from the light - cone energy - momentum tensor expressions for the light - cone hamiltonian @xmath9 and conserved total momenta @xmath10 : @xmath11 , \hspace{1 mm } \mbox{``light - cone hamiltonian '' } \label{pminus } \end{aligned}\ ] ] where we adopt the notation @xmath12 .
light - cone quantization of the @xmath0 theory is performed in the usual way namely , we impose commutation relations at some fixed light - cone time ( @xmath13 , say ) : @xmath14 = \frac{{\rm i}}{2 } \delta ( x^- - y^- ) \delta ( { \bf x}^{\perp } - { \bf y}^{\perp } ) .
\label{commphi3}\ ] ] the light - cone hamiltonian @xmath9 propagates a given field configuration in light - cone time @xmath15 while preserving this quantization condition . at fixed @xmath13 , the fourier representation @xmath16 , \label{phi}\end{aligned}\ ] ]
together with the quantization condition ( [ commphi3 ] ) , imply the relation @xmath17 = \delta ( k^+ - { \tilde k}^+ ) \delta ( { \bf k}^{\perp } - { \tilde { \bf k}}^{\perp } ) .\ ] ] it is now a matter of substituting the fourier representation ( [ phi ] ) for the quantized matrix field @xmath18 into definitions ( [ pminus ] ) , ( [ ptrans ] ) and ( [ pplus ] ) , to obtain the following quantized expressions for the light - cone hamiltonian and conserved total momenta : @xmath19 where @xmath20 .
we are now interested in calculating the effective two - loop contribution to the mass term in the light - cone hamiltonian , which is represented diagrammatically in fig [ twoloop ] .
one advantage of working in light - cone coordinates is the absence of tadpole diagrams .
i.e. the only two - loop contribution to the mass is given by the ` setting sun ' diagram below .
a convenient strategy is to compute this diagram for vanishing ( external ) longitudinal momentum @xmath21 .
the details of this method appear in an earlier article @xcite , so we will mention only the key ideas here . first , note that the two - loop process can be obtained by two applications of the light - cone hamiltonian ; one application corresponds to the creation of three partons from a single parton via the interaction @xmath22 , while the subsequent interaction @xmath23 will absorb these same three partons to leave one parton left over . during the intermediate step , where there are three partons in flight , the sum of the light - cone momenta must be equal to the initial parton momentum , which we write as @xmath24 .
we now consider what happens in fig [ twoloop ] .
when the incoming longitudinal momentum @xmath25 is made vanishingly small by taking the limit @xmath21 .
it is important to stress that in this limit , we never set @xmath25 identically to zero ; we simply investigate the limiting contribution of the diagram as @xmath25 which is always positive is made arbitrarily small . a detailed calculation paralleling the one carried out in @xcite in which one studies the bound state integral equations governing the light - cone wave functions in the limit of vanishing @xmath25 may be used to calculate the limiting contribution of this diagram to the mass renormalization .
the result is @xmath26 this quantity depends on @xmath27 only , since any dependence on @xmath25 was eliminated after taking the limit @xmath21 . in order to find the renormalized mass , we set @xmath28 .
some remarks are in order .
firstly , note that a remnant of the longitudinal integration survives through the variables @xmath29 and @xmath30 .
these quantities represent the _ fractions _ of the external longitudinal momentum @xmath25 , and are thus integrated over the interval @xmath31 .
they are not eliminated in the limit @xmath21 . to further illuminate the significance of these variables
, we consider the special case of @xmath32 dimensions ( i.e. @xmath33 ) . in this case , the mass renormalization is @xmath34 now consider the feynman integral for the ` setting sun ' diagram of scalar @xmath0 theory in @xmath32 dimensions , with zero external momenta @xcite . in this case
, the mass renormalization turns out to be @xmath35}. \label{feynman}\ ] ] if we use feynman parameters to rewrite this last integral @xcite , and perform the momentum integration , we end up with expression ( [ mass2d ] )
. i.e. the integrals ( [ gamma ] ) and ( [ feynman ] ) are equal in @xmath32 dimensions
. it would be interesting to test for equality in @xmath36 dimensions , and we leave this for future work . interestingly , the parameters @xmath37 above which correspond to fractions of longitudinal light - cone momentum in the light - cone approach represent the familiar feynman parameters in the usual covariant feynman diagram approach . we have therefore discovered a physical basis for the feynman parameters .
the actual value of the integral may be computed @xcite , and is given by @xmath38,\ ] ] where @xmath39 is euler s psi function .
the terms in the parenthesis has the numerical value @xmath40 .
we leave an analysis of the self - energy contribution @xmath41 in @xmath42 dimensions for future work .
we have calculated the two - loop mass renormalization for the effective light - cone hamiltonian of scalar @xmath0 field theory .
an explicit evaluation of the corresponding integral in two dimensions was given , and shown to be precisely equal to the corresponding two - loop ` setting sun ' feynman diagram .
moreover , we were able to attribute physical significance to the familiar feynman parameters ; namely , as fractions of vanishingly small external light - cone momenta .
evidently , it would be interesting to study the integral ( [ gamma ] ) in more detail for higher space - time dimensions .
in particular , ultraviolet divergences are expected in dimensions @xmath43 , and so we may invoke dimensional regularization to regulate the integration over transverse momentum space .
this would facilitate a straightforward comparison with existing feynman diagram calculations in @xmath0 theory @xcite .
f.a . is grateful for fruitful discussions with armen ezekelian .
j.o.a was supported in part by a faculty development grant from the physics department at the ohio state university , and by a nato science fellowship from the norwegian research council ( project 124282/410 ) .
9999 s.glazek and k.g.wilson , _ phys.rev _ * d48 * ( 1993 ) 5863 ; _ ibid _ * d49 * ( 1994 ) 4214 .
f.wegner , _ ann .
der physik _ * 3 * , ( 1994 ) 77 .
k.g.wilson , t.s.walhout , a.harindrinath , w.m.zhang , r , j.perry , and s.d.glazek , _ phys.rev _ * d49 * ( 1994 ) 6720 .
s.glazek , a.harindrinath , s.pinsky , j.shigemitsu , and k.g.wilson , _
phys.rev _ * d47 * ( 1993 ) 1599 .
r.perry , `` light - front quantum chromodynamics , '' to be published in proceedings of the apctp - rcnp joint international school on physics of hadrons and qcd , osaka , japan .
( nucl - th/9901080 ) ; brent h. allen , robert j. perry , `` systematic renormalization in hamiltonian light - front field theory , '' _ phys.rev . _ * d58 * ( 1998 ) 125017 ; roger d. kylin , brent h. allen , robert j. perry , `` systematic renormalization in hamiltonian light - front field theory : the massive generalization '' , hep - th/9812080 .
stanislaw d. glazek , `` boost - invariant running couplings in effective hamiltonians , '' hep - th/9904029 .
f.antonuccio , `` light - cone quantization and renormalization of large - n scalar matrix models , '' hep - th/9705045 .
f.antonuccio , s.j.brodsky , and s.dalley , _ phys.lett . _ * b412 * ( 1997 ) 104 - 110 . s.dalley , `` collinear qcd models '' , hep - th/9812080 .
invited talk at ` new non - perturbative methods and quantisation on the light - cone ' , les houches 24 feb - 4 mar 1997 .
( to appear in proceedings ) , ( hep - th/9704211 ) ; s.dalley , _ phys.rev .
_ * d58 * ( 1998 ) 087705 .
jens.o.andersen , `` two - loop calculations for @xmath0 theory in @xmath32 dimensions '' , ( unpublished notes , 1993 ) .
j.c.collins , _ phys.rev .
_ * d10 * , 1213 ( 1974 ) .
gradshteyn , i.s . and ryzhik , i.m , _ table of integrals , series and products _ , academic press , london ( 1980 ) . | we perform a two - loop calculation in light - front @xmath0 theory to determine the effective mass renormalization of the light - front hamiltonian .
the renormalization scheme adopted here is manifestly boost invariant , and yields results that are in perfect agreement with the explicitly covariant feynman diagram approach .
-0.25 in 8.5 in 6.25 in ohstpy - hep - t-99 - 011 + hep - th/9904140 two - loop calculations in @xmath0 light - front field theory * f.antonuccio and j.o.andersen * + .25 in |
electrons and positrons are a relatively rare component of the cosmic radiation , but are distinct from all other cosmic - ray particles because of their low mass and the absence of hadronic interactions with the constituents of the interstellar medium .
previous measurements of electrons @xmath4 up to roughly 1000 gev ( @xcite ) have shown that their intensity amounts to about 1% of the flux of protons around 10 gev , but decreases more rapidly with energy ( @xmath5 ) than the proton spectrum ( @xmath6 ) .
separate measurements of positrons and electrons have only been possible at much lower energies and have indicated a `` positron fraction '' ( @xmath7 ) of a few percent in the region 1 - 10 gev ( @xcite ) .
some measurements have indicated an increase above about 10 gev ( @xcite ) .
these observations have led to a number of conclusions , but have also left some key questions unanswered : \1 ) the predominance of negative electrons can only be explained if electrons are accelerated by primary sources . the alternate production mechanism , as secondary particles from interstellar nuclear interactions of hadronic cosmic rays ( mostly through the @xmath8 decay ) , would yield negative and positive electrons in about equal proportions , and thus can only account for a small portion of the total @xmath9 flux .
it is worth pointing out that electrons are the only cosmic - ray component for which an extragalactic contribution can be excluded with certainty : energy losses through compton interactions with the 2.7k background radiation preclude their propagation through intergalactic distances .
\2 ) the steepness of the observed energy spectrum of electrons is usually explained as a consequence of radiative energy losses during propagation through the interstellar medium : inverse compton scattering with photons and synchrotron radiation in the interstellar magnetic fields . assuming that electrons are accelerated with the same energy spectrum at the source as nuclei ( @xmath10 , @xcite ) the observed spectral slope is roughly consistent with the galactic containment time of @xmath11 years of nuclei at gev energies .
however , there are several problems with this interpretation : first , the energy spectrum of electrons at the source is not known a priori , nor is it known that electrons and nuclei originate at the same acceleration sites .
second , it has been pointed out ( @xcite ) that the energy dependence ( @xmath12 ) of the containment time observed for nuclei may be difficult to reconcile with the observed shape of the electron spectrum .
third , the `` leaky box '' assumption inherent in these explanations should not be applied to electrons , as it requires an unreasonably high density of electron sources in the galactic disk ( @xcite ) .
\3 ) the small flux of positrons appears to be essentially consistent with a secondary origin .
if this is true , the production spectrum of positrons can be calculated from the known flux of primary nuclei ( @xcite ) .
positrons would be produced continuously throughout the galactic disk , and the leaky box model could more easily be taken as an approximation for their propagation through the galaxy . in any case , a direct comparison between the production spectrum and the observed positron spectrum would permit a quantitative study of the propagation mechanism .
however , it must first be ascertained that indeed all positrons are of secondary origin .
the previously observed increase in the positron fraction has led to much speculation about the appearance of primary positrons at high energy ( e.g. @xcite ) .
recent results ( @xcite ) have not confirmed this increase , but have not yet shown conclusively that the positron flux is entirely free of primary contributions .
it is clear from these considerations that many of the open questions can only be answered through measurements of electrons and positrons separately , and over as large an energy range as possible .
this was the motivation for the construction of the high - energy antimatter telescope ( heat ) , an instrument that utilizes a large superconducting magnet spectrometer to separate positive and negative particles , and that incorporates powerful techniques to identify electrons and positrons and to reject hadronic background .
a first balloon flight of heat was conducted in 1994 . in the following
, we shall describe this measurement and the data analysis technique , and we shall present and discuss the results .
the first balloon flight of the heat e@xmath13 instrument took place on 1994 may 3 - 5 , from fort sumner , new mexico , and data were collected at float altitude for about 29 hours .
the payload reached a maximum altitude of 36.5 km and drooped to a minimum of 33 km at night , as illustrated in figure [ fig1 ] .
over the course of the flight , the payload drifted between vertical geomagnetic cutoff rigidities of 4 gv and 4.5 gv , latitudes of @xmath14 n and @xmath15 n and longitudes of @xmath16 w and @xmath17 w. the instrument was recovered undamaged near wellington , texas .
the expected intensity of cosmic - ray positrons is quite low , of the order of @xmath18 of the total cosmic - ray intensity at comparable energies .
a successful measurement then necessitates a detector with large sensitive area to yield statistically significant data , and with very high discrimination power against the overwhelming proton background .
this is accomplished in the heat instrument , shown in figure [ fig2 ] , through the combination of a superconducting magnet spectrometer ( using a drift - tube hodoscope ( dth ) tracking chamber ) with particle identifiers employing a time - of - flight system ( tof ) , a transition - radiation detector ( trd ) , and an electromagnetic calorimeter ( ec ) .
a thorough description of the instrument appears elsewhere ( @xcite ) , and the following just gives a brief summary of the apparatus and its performance .
the tof measures the velocity of the particle , distinguishes downward from upward - going ( albedo ) particles , and measures the magnitude of the particle s charge .
the rejection of upward - going particles is necessary as these would otherwise appear as particles of the opposite charge in the magnet spectrometer . through the measurement of the magnitude of the charge ,
singly - charged particles are identified and discriminated from helium and heavier nuclei .
the tof consists of four scintillator slabs on top of the instrument , each with an active area of 100 cm @xmath19 25 cm , and of the top three scintillation counters of the ec ( described below ) . throughout the instrument , photomultiplier tubes ( pmt )
are used that are resistant to the magnetic fringe fields .
the pmt signals are both charge- ( adc ) and time- ( tdc ) digitized , and the particle s charge is determined from the adc values measured with the top scintillator slabs , each of which has a pmt on either end .
figure [ fig3 ] shows the charge distribution obtained for flight data , with the criteria for selecting singly - charged particles shown as dashed lines ( and summarized in table [ all_cuts ] ) .
the charge resolution obtained is 0.11 charge units , which is sufficient to provide good rejection of he events .
the tdc values , after corrections for the path length through the instrument , give the particle s velocity , with a resolution of 0.13 c ( where c is the speed of light ) .
figure [ fig3 ] shows the velocity distribution obtained for flight data , and indicates the selection criteria used for the analysis .
the transition - radiation detector is used to distinguish electrons and positrons from hadrons .
it is comprised of 6 modules , each consisting of a polyethylene - fiber radiator and a multiwire proportional chamber ( mwpc ) .
the mwpcs contain a gas mixture of xenon : methane ( 70:30 ) and produce signals on both the anode wires and on cathode strips .
the total charge deposited in the chamber is read from the cathode strips and pulse - height analyzed ( `` pha '' analysis ) , while the current signal is read from the anode wires in 25 nsec - wide time slices , which are digitized into three different threshold levels ( `` time - slice '' analysis )
. charged particles of high lorentz factor ( @xmath20 ) produce transition - radiation ( tr ) x - rays ( 530 kev ) in the radiator , which are subsequently detected by the mwpc .
this x - ray signal is superimposed on the signal due to ionization energy loss of the parent particle . in the energy range of 550 gev
, electrons produce a saturated tr signal , while protons and pions produce none , thereby permitting a discrimination between these species .
a maximum - likelihood technique is used to analyze the pha signals ( @xcite ) . to this end ,
first the probability distributions for the pulse heights in the mwpc s must be determined , using clean populations of electrons and protons .
the electron sample is obtained by selecting events with negative rigidity in the magnet spectrometer ( described below ) , and a shower profile in the ec consistent with an electromagnetic shower ( also described below ) .
the proton sample is obtained by selecting events with positive rigidity and which have shower profiles inconsistent with an electromagnetic shower .
the validity of this procedure has been verified with accelerator calibrations at fermilab .
gain differences between the trd chambers are corrected for , as are temporal variations ( 30% ) in the chamber gains caused by temperature and pressure changes inside the instrument gondola , spatial non - uniformities ( 20% ) , and the relativistic rise of the proton ionization signal .
the upper two panels of figure [ fig4 ] show the pha probability distributions p@xmath21 for electrons and p@xmath22 for protons .
for the subsequent data analysis , each event , characterized by the mwpc pulse heights @xmath23 ( @xmath241, ... ,6 ) , is compared with the probability distributions , and a likelihood ratio @xmath25 is formed : @xmath26 figure [ fig5 ] shows as a dashed line labeled `` pha '' the electron and proton efficiencies obtained by applying a selection on the pha likelihood ratio .
the time - slice technique makes use of the fact that an x - ray photon absorbed in the mwpc generates a highly localized `` ionization cluster '' in the chamber , which then drifts to the anode wires and produces a `` spike '' in the time structure of the anode signal . in the analysis ,
again , a likelihood ratio @xmath27 is constructed , using the appropriate single - chamber probability distributions @xmath28 and @xmath29 .
these distributions are based on the number of time slices above each threshold level , and the positions , heights , and number of clusters in the time - slice distribution , after the gain corrections described above are applied .
the lower two panels of figure [ fig4 ] show the time - slice probability distributions obtained for electrons and protons .
figure [ fig5 ] shows as a dotted line labeled `` time - slice '' the electron and proton efficiencies obtained by applying a selection on the time - slice likelihood ratio .
the likelihood ratios @xmath25 and @xmath27 , are combined to form a total ratio : @xmath30 .
figure [ fig6 ] shows the distribution of @xmath31 together with the selection criterion used to identify electrons ( @xmath32 , or dashed line in the figure ) . with this criterion ,
electrons are retained with an efficiency @xmath33 and protons with an efficiency @xmath34 , corresponding to a proton rejection power of 170 .
this is illustrated in figure [ fig5 ] as a solid line labeled `` total . ''
the magnet spectrometer is used to measure the rigidity @xmath35 and charge sign of a traversing charged particle , and consists of a two - coil warm - bore superconducting magnet and a precision tracking detector utilizing drift tubes .
the magnet produces an approximately uniform field of central value 1 t. the fiducial volume within the magnet bore measures 50 cm @xmath19 50 cm @xmath19 61 cm .
the tracking system consists of 479 drift tubes of 2.5 cm diameter , arranged in 26 rows ; 18 rows ( of which 17 were operational during the flight ) contain tubes parallel to the magnet axis , defining the bending view , and 8 contain tubes perpendicular to the magnet axis , defining the non - bending view .
the drift gas mixture used is co@xmath36:hexane ( 96:4 ) .
the dth determines the particle trajectory by measuring the drift times of the ionization tracks in each tube hit , from which , using a `` time - to - space '' function , the impact parameters ( distance of closest approach to wire ) are found .
the trajectory of the particle in three dimensions is reconstructed from the impact parameters using a modified version of the cern program libraries momentm algorithm ( @xcite ) .
figures [ fig7 ] and [ fig8 ] show the residuals from the tracks reconstructed for flight data , as a distribution and as a function of radius , respectively .
a signal is not used in the track fit if it lies more than one tube diameter away from the reconstructed trajectory , or if it has a residual that is worse than 6 times the average residual for the track .
also removed are signals corresponding to an impact parameter smaller than 2 mm , a region where the tracking capabilities of the drift tubes are poor due to the nature of the electron - ion pair statistics and the increased drift speed near the wire .
the single - tube resolution achieved is about 70 @xmath37 m .
the performance of the magnet spectrometer is characterized by the maximum detectable rigidity ( mdr ) .
the mdr is defined as that rigidity where the rms error in the sagitta measurement is equal to the sagitta of the track , and is computed on an event - by - event basis .
the mdr distribution for electron events , given in figure [ fig9 ] , shows that a mean mdr of 170 gv is achieved .
the mdr yields the relative rms error in the rigidity : @xmath38 .
thus , an mdr of 170 gv provides for a rigidity determination with an at least @xmath39 accuracy up to @xmath40 60 gv . in the analysis of the flight data ,
the selections @xmath41 and @xmath42 are imposed to ensure meaningful results .
the electromagnetic calorimeter measures the energy of electrons and discriminates against hadrons .
the ec consists of ten lead plates , each 0.9 radiation length thick and 50 cm @xmath19 50 cm in area , and each followed by a plastic scintillator .
electrons deposit most of their energy in the ec , with well - understood shower profiles that provide a measure of their energy .
protons and other hadrons , on the other hand , rarely interact in the ec , as it represents only about 0.3 proton interaction lengths , and if a nuclear interaction occurs , the longitudinal shower profile is usually quite different from that of an electron .
therefore , the ec also provides powerful discrimination against protons . in order to reconstruct the electron primary energy from the shower profile ,
monte carlo simulations have been carried out using the cern program libraries geant package ( @xcite ) .
the results of these simulations have been verified through accelerator calibrations ( @xcite ) .
a covariance analysis is applied to the monte carlo events , providing the cross - correlations between the signals obtained in each of the layer pairs as a function of energy .
figure [ fig10 ] shows , for 10 gev simulated electrons , the distribution of energies reconstructed by this covariance analysis .
the correlation matrices generated by the covariance analysis of monte carlo events are then applied to flight data to determine the particle energies .
the fractional energy resolution achieved is about 10% for the energy range of interest . during the balloon flight ,
an event trigger is employed ( see below ) that requires a minimum energy deposition in the ec , and thereby removes most non - interacting protons . in order to discriminate against the remaining proton events with the ec , a @xmath43 measure of the agreement between the observed and expected shower profile , generated by the covariance analysis ,
is determined for each event .
in addition , the shower start depth @xmath44 is derived from a separate profile analysis of the shower , and events with @xmath45 radiation length are rejected . finally , agreement between the energy @xmath46 measured by the ec , and the momentum @xmath47 measured by the magnet spectrometer is demanded .
figure [ fig11 ] shows the electron and proton efficiencies obtained for flight data by imposing all of these requirements .
the selection criteria chosen for the analysis are summarized in table [ all_cuts ] , and yield for the combination of shower counter and magnet spectrometer a proton rejection power of 460 at an electron efficiency of 97% .
combined with that of the trd ( see above ) , the total rejection power of the instrument against protons is nearly @xmath48 .
the instrument trigger requires a through - going particle which deposits a signal larger than that of a 0.5 gev electron in the ec .
this excludes non - interacting protons .
in addition , a minimum number of hits in the dth is required .
a `` fast '' trigger is formed by the coincidence of signals in the top and bottom tof scintillators , and of a signal above threshold in the sum of the lower seven ec layers .
a `` slow '' confirming trigger is required within 1.5 @xmath37sec thereafter , based on the occurrence of a three - fold majority of hits in the first , fifth , and last two rows of tubes in the bending plane . in order to obtain a sample of non - interacting protons ,
the ec sum threshold requirement is removed for a small percentage ( 2% ) of the events .
the initial data analysis stage involves examining the dth hits for clean single - particle events .
a rough estimate of the trajectory is made by performing a quadratic fit to the impact parameters in both the bending and non - bending views . signals from tubes well outside this fit are discarded .
events in which tracks can not be identified in the dth through this procedure are rejected .
most events eliminated at this stage result from interactions of particles that penetrate the instrument from the sides . tracks which pass the initial dth filter are then analyzed using the modified momentm algorithm , which determines the particle s rigidity from the measured track points .
a @xmath43 parameter , generated by comparing the measured track points with the fit , is used to reject events with large tracking errors .
in addition , events are rejected based on the average residuals associated with the track , as well as on the number of tubes retained in the final fit in the bending and non - bending views .
finally , the mdr requirements described in section [ dth ] are imposed .
table [ all_cuts ] summarizes all of the track cleanliness requirements .
electron selection criteria are applied to events that satisfy the track cleanliness requirements .
as discussed , these include selections on the magnitude of the particle charge , the trd total likelihood ratio , the ec covariance analysis @xmath43 parameter , the shower start depth , and reconstructed energy . in order to maximize the statistics of the results ,
the electron selection criteria have been chosen relatively loosely , requiring that the number of background protons be less than 25% of the number of accepted positrons .
table [ all_cuts ] summarizes the selections used . for events meeting the electron selection criteria , distributions of the ratio of the measured energy @xmath46 determined by the ec , and the momentum @xmath47 determined by the magnet spectrometer ,
are generated .
figure [ fig12 ] shows these @xmath49 distributions for electrons and positrons in five top - of - the - atmosphere ( toa ) energy intervals .
( the top - of - the - atmosphere energy @xmath50 is obtained from the measured energy @xmath46 by correcting for atmospheric bremsstrahlung losses according to the procedure described in section [ atmoscor ] . ) the differences in @xmath49 between positrons and residual background protons is evident , with the positron distribution peaking at + 1 , and that for protons peaking at about + 0.5 .
the tails at high @xmath51 values in the electron or positron distributions are due to bremsstrahlung losses in the material above the dth , which result in a reduced rigidity in the spectrometer while the entire energy ( of the e@xmath13 plus bremsstrahlung photons ) is usually recorded in the ec .
a fit to the @xmath49 distributions for electrons and positrons is performed to obtain the number of electrons and positrons , and to estimate and subtract the residual proton background .
the @xmath49 distribution for negative electrons , which can be taken as background free , is used to obtain a template of the expected @xmath49 distribution for e@xmath13 .
the @xmath49 distribution for background protons is obtained by inverting the trd selection to select interacting protons ( see table [ all_cuts ] ) .
both template distributions are smoothed , and used to fit the @xmath49 distribution for positrons , shown as the solid curves in figure [ fig12 ] . a bayesian treatment is then used to estimate the electron , positron and proton counts , with prior probability distributions assumed to be flat for the electron , positron and proton background counts , and with conditional distributions taken from the template fits .
the electron and positron counts resulting from this procedure are listed in table [ elec_numb ] .
the uncertainties given in the table are 16 and 84% bayesian limits . for the energy range 50100 gev ,
the ability of the magnetic spectrometer to determine the direction of bending is reduced as the uncertainty on the sagitta of the track is comparable to the sagitta itself .
therefore , in this energy range , electron events are selected without the rigidity - related selection criteria of table [ all_cuts ] , and electrons are not distinguished from positrons . as a test to verify the degree of potential residual contamination of the positron counts by interacting hadrons
, the electron selection criteria were varied from very loose to very restrictive . in all but the loosest sets of selection criteria ,
the electron and positron counts were reduced or augmented in the same proportion , indicative of little , if any , residual contamination .
the absolute differential energy spectra of cosmic - ray primary electrons or positrons are obtained from the raw electron or positron counts @xmath52 of table [ elec_numb ] by calculating : @xmath53 with @xmath54 @xmath55 where an atmospheric secondary ( _ sec _ ) component is subtracted to obtain the primary ( _ pri _ ) component . in equation
( 1 ) , @xmath56 is the instrumental acceptance , @xmath57 is the live time , @xmath58 and @xmath59 are the lower and upper bounds of each toa energy interval , respectively , @xmath60 is the weighted average toa energy in this interval , @xmath61 is the weighted toa energy interval , @xmath62 is a correction factor related to the transformation of the energy scale to the top of the atmosphere , which will be described in section [ atmoscor ] , and @xmath63 ( @xmath64 for primary particles and @xmath65 for atmospheric secondaries ) is the power - law spectral index . the live time @xmath57 of the detector is measured in flight with a scaler which counts clock cycles only while the instrument is available for triggers and not busy processing an event following a trigger .
periods of instability of the instrument s performance are discarded , and a total live time of @xmath66 hours at float altitude is obtained .
this live time includes losses due to transmission errors and data - unpacking errors from glitches in the electronics of individual subsystems .
some of the quantities in equation ( 1 ) are determined with a full monte carlo ( mc ) simulation of the heat instrument , based on the geant software package , and including the actual detector configuration and the experimentally - determined fluctuations in detector response .
for instance , the mc simulation is used to calculate the instrumental acceptance for electrons , i.e. , the product of the absolute efficiency @xmath67 and the geometrical aperture @xmath68 .
the mc - calculated efficiencies are compared to experimental quantities where possible , and in some cases renormalized to ensure agreement .
for example , the trd electron selection efficiency is determined by obtaining a set of electron events based on ec , tof and dth information , and measuring the fraction of such events that satisfy the trd likelihood ratio requirement .
in addition , a visual inspection of several hundred raw experimental and simulated electron events revealed that about 65% of experimental events are accepted by the analysis , whereas 81% of simulated events are accepted . the instrumental acceptance determined by the mc
is therefore corrected by a `` scanning '' efficiency factor of @xmath69 @xmath70 , the uncertainties being an estimate of the spread of values due to independent scannings .
table [ acc_liv ] lists , for each energy interval , the acceptances calculated with the mc , including the @xmath69 correction .
the geometrical aperture , uncorrected for efficiencies , is about 495 @xmath71 .
this indicates an average electron acceptance efficiency of about 37% . in table
[ acc_liv ] , the acceptance uncertainties are obtained by adding in quadrature the uncertainties in the individual efficiency factors .
the increased acceptance of the highest energy bin derives from an increased efficiency due to a different set of selection criteria used in this energy range , as indicated in table [ all_cuts ] .
a comparison of experimental and simulated @xmath49 distributions revealed a systematic energy calibration offset in the flight data .
specifically , the experimentally - determined @xmath51 peak for electrons is shifted upwards by about 14% , whereas the monte carlo simulation predicts a shift of only 4% due to the bremsstrahlung energy losses .
the 10% discrepancy appears to be due to a systematic bias in the conversion of the adc values of the ec into numbers of minimum - ionizing particles .
this energy - scale shift is corrected for in figure [ fig12 ] .
the measured energy spectrum is shifted without a change in spectral index due to the finite , but roughly constant , energy resolution of the ec .
the monte carlo simulation shows that of these events which have a true initial energy in a given energy bin , a fraction of 2 to 6% are reconstructed by the analysis into the next higher - energy bin , while 22 to 32% are reconstructed into the adjacent lower - energy bin . this `` spillover '' effect reduces the reconstructed energy spectrum by about 10% in overall intensity as compared to the true primary spectrum .
the final results have been corrected for this effect .
only events with toa energies above 5 gev , well above the geomagnetic rigidity cutoff at @xmath72 gv , are retained in the analysis . the electron spectrum measured at balloon altitudes
is comprised of primary cosmic - ray electrons and atmospheric secondary electrons .
the secondary component arises as a result of interactions of hadrons or primary electrons within the atmosphere , as well as reentrant albedo electrons .
reentrant albedo particles are significant only below the geomagnetic cutoff energy . to determine the secondary flux generated in the atmosphere , a monte carlo simulation utilizing the cern libraries fluka hadronic interaction algorithm ( @xcite ) is used . for this calculation , a primary proton spectrum with a power - law index of 2.74
is assumed ( @xcite ) .
secondaries produced by primary electrons do not contribute significantly . to account for the contribution of heavy primary nuclei ,
the simulated intensity of atmospheric secondaries is multiplied by 1.2 ( @xcite ) .
the total intensity of 10 gev secondary electrons and positrons at 6 g/@xmath73 is found to be 0.006 ( gev s sr m@xmath74)@xmath75 .
this corresponds to overall corrections to the positron intensity of 2030% , and to the electron intensity of 12% , at the energies of interest .
the absolute atmospheric intensities at an average atmospheric depth of 5.7 g/@xmath73 for electrons and positrons are also presented in table [ atm_sec ] , for each energy interval . as a test of the reliability of the mc calculation ,
atmospheric growth curves for electrons and positrons are measured during the flight .
the positron fraction as a function of depth in the atmosphere is shown in figure [ fig13 ] , for two separate energy intervals .
the flight data are divided into four altitude intervals of 4.0 , 4.5 , 6.8 , and 7.2 g / cm@xmath74 average depth , respectively .
the error bars represent statistical fluctuations , dominated by the number of positrons in each energy interval .
the positron fraction at the top of the atmosphere is obtained by linear extrapolation of the experimental growth curves . from the data on figure [ fig13 ]
, one can determine the secondary - to - primary ratio @xmath77 as a function of atmospheric depth . in table
[ sec_prim_ratio ] , this ratio at a depth of 6 g/@xmath73 is given and compared to the corresponding results of the mc calculations .
the uncertainties indicated are statistical .
the mc simulation is consistent with the measured secondary - to - primary ratios within errors .
the energy of a primary electron detected at a residual atmospheric depth of @xmath44 radiation lengths is reduced from the energy at the top of the atmosphere due to bremsstrahlung energy losses .
if the primary spectrum has the form of a power law @xmath78 , the detected energy spectrum will , for a given depth @xmath44 , retain the same power law but will be reduced in intensity by a factor @xmath79 .
this factor is derived by taking the statistical fluctuations in the radiation loss mechanism into account ( @xcite , @xcite ) . for the heat flight described here ,
the atmospheric depth varies considerably throughout the flight .
therefore , we correct the energy for each detected electron or positron by a factor @xmath80 , where @xmath44 is the current atmospheric depth , and @xmath81 is taken for the spectral index .
typically , this amounts to a @xmath40 5 - 10% shift in the energy scale . because equation ( 1 ) is used to calculate a _
intensity , the energy scale shift necessitates a further correction factor @xmath82 , where @xmath83 is averaged over the flight .
the absolute differential intensities of electrons and positrons are obtained from equation ( 1 ) using table [ elec_numb ] , which lists the number of particles counted in each energy interval @xmath52 , table [ acc_liv ] which tabulates the acceptances @xmath67 @xmath84 , live time @xmath57 and average toa energy correction factor @xmath62 , as well as @xmath60 and @xmath61 ( computed with @xmath81 ) , and table [ atm_sec ] which gives the atmospheric contribution .
all of the statistical uncertainties associated with these quantities are also contained in these tables .
the electron , positron , and all - electron differential intensities are listed in table [ abs_flux ] . the uncertainties are statistical , computed by adding in quadrature all of the contributing uncertainties . the results are plotted in figures [ fig14 ] and [ fig15 ] , scaled by @xmath85 for clarity , together with other measurements of electron and positron energy spectra ( @xcite ) . besides the uncertainty in determining the instrumental efficiencies discussed in section [ monte ] , there are two additional sources of systematic errors .
first , the necessity of applying an energy - shift correction to ensure agreement with the momentum indicates a systematic uncertainty in the absolute energy calibration of a few percent , which translates into a @xmath4010% uncertainty in the overall intensity normalization due to the @xmath81 power law of the spectrum .
second , uncertainties in the normalization of the proton spectrum translate into a @xmath86 systematic uncertainty in simulated atmospheric secondary intensities ; this has a 0.5% effect on the electron and all - electron intensities , but results in 10% shifts in the positron intensity normalization .
thus , the estimated systematic uncertainty from these sources is 10% for the electron and all - electron intensities , and 14% for the positron intensity . in figure [ fig14 ]
, we compare our results with previous measurements for which absolute intensities have been given for positrons and electrons separately .
we find that the energy spectrum of positrons reported by @xcite and @xcite is generally consistent with our data , although our results show much improved statistical error limits and permit a better estimate of the slope of the energy spectrum .
the intensity of positrons of interstellar secondary origin can be calculated , provided the spectrum of parent protons is known , and the parameters of galactic propagation are specified .
such a calculation has been done by @xcite , and , more recently , by @xcite .
we show in figure [ fig14 ] the results of strong et al .
1997 as a dotted line , and notice the good agreement with our data . in figure [ fig15 ]
, we compare the all - electron energy spectrum @xmath87 derived from our data with previous results , also including measurements which did not employ a magnet spectrometer for charge separation .
we arbitrarily chose to include only results published since 1980 .
the dotted line represents a parametrization of the all - electron spectrum used by strong et al .
we find that all results exhibit similar spectral slopes over the energy range of concern , but that our overall intensity is lower than the average of the previous investigations by about 35% .
there are two possible contributors to this systematic discrepancy .
first , the absolute energy calibration for the experiments is likely to be uncertain by about 10% , leading to a possible systematic uncertainty in the intensity of the order of 20% .
second , the assessment of the absolute detection efficiency of the instrument is notoriously difficult for all such detectors .
its determination does involve some intuitive judgment , although the more recent investigations , such as ours , benefit in this respect from more sophisticated monte carlo simulations than were previously available .
the individual spectral indices for positrons and electrons are @xmath88 and @xmath89 , respectively , over the energy range 5.0 to 50 gev , as indicated in table [ pow_fit ] and represented by solid lines in figure [ fig14 ] .
thus , the positron spectrum appears to be slightly steeper than that of electrons .
if all positrons are of interstellar secondary origin , one should expect that radiative energy losses eventually lead to a power - law index that is larger by unity than that of the production spectrum , which follows the ambient spectrum of the parent nuclei , i.e. , @xmath90 for nuclear spectra of the form @xmath91 .
the spectrum of electrons , on the other hand , should steepen less strongly , to a final slope with @xmath92 if electrons are mostly produced at the same primary sources as nuclei , and with the same source spectrum characterized by @xmath93 .
however , these final spectral slopes may not be reached in the energy region below @xmath1 gev .
thus , it would seem proper to fit our data not to a single power law but rather to a spectral form that reflects a transition from the source spectrum ( modified by solar modulation ) to a spectrum that is fully steepened due to radiative energy losses .
such an analysis had previously been performed on the all - electron spectrum by @xcite and @xcite . here
we will not entertain this analysis for our data , as we soon should be able to improve the statistical quality of the results through the inclusion of data from an additional balloon flight .
qualitatively , we may just conclude that the slightly steeper spectrum of positrons , as compared to that of electrons , is to be expected if positrons are predominantly secondary particles .
this conclusion must be confirmed by additional data and by an extension of the measurement over a wider energy range . as mentioned above , a comparison with the calculations of @xcite and strong et al .
1997 indicates also that the absolute intensity of positrons is close to what may be expected if indeed all positrons are generated subsequent to nuclear interactions in the interstellar medium .
more detail on the relative origin of electrons and positrons , and limits on the possible primary contribution to the positron flux , can be obtained if the relative intensity of positrons , i.e. , the positron fraction @xmath94 ( which does not depend on the absolute efficiency of the detector ) , is investigated as a function of energy .
we have presented the results of such an investigation in previous papers ( @xcite ) .
the results described in this paper have been obtained with the first balloon flight of the heat spectrometer .
we believe that the high level of hadronic background rejection achieved with this instrument has led to data of high quality , limited only by the statistical uncertainties of a single balloon flight .
a second balloon flight which has been carried out successfully from lynn lake , manitoba in 1995 august , covers the energy range from @xmath95 gev upwards .
the analysis of the combined data set is presently in progress , and results will soon be reported .
we gratefully acknowledge assistance from d. bonasera , e. drag , w. johnson , p. koehn , d. kouba , r. northrop , and j. robbins .
we are grateful to m. israel for helpful discussions on atmospheric secondary production .
we also thank the nsbf crew that has supported the balloon flight .
this work was supported by nasa grants nagw-1035 , nagw-1995
, nagw-2000 , nagw-4737 , nag5 - 5059 , l + + dth filter algorithm passed + dth rigidity reconstruction algorithm passed + @xmath98 ( number of tubes in dth bending view used in fit ) + @xmath99 ( number of tubes in dth non - bending view used in fit ) + @xmath100 ( dth rigidity goodness of fit ) + @xmath101 cm ( average residual in @xmath102 ( non - bending ) projection ) + @xmath103 cm ( average residual in @xmath104 ( bending ) projection ) + @xmath105 cm ( average residual in @xmath106 ( vertical ) projection ) + @xmath107 kg m ( integrated b - field over the track length ) + @xmath108 gv + @xmath109 + @xmath110 + @xmath111 ( velocity range , downward - going ) + @xmath112 cm ( agreement between dth track and tof timing ) + propagated dth track traverses both tof and ec + propagated dth track traverses @xmath113 trd chambers + + @xmath114 ( charge ) + ec starting depth @xmath115 radiation lengths + @xmath116 ( ec goodness of fit ) + @xmath117 > 3 $ ] ( trd likelihood ratio ) + + same as electron selection criteria except : + @xmath118 < -1 $ ] ( trd likelihood ratio ) + cccc 5.0 6.0 & @xmath119 & @xmath120 & @xmath121 + 6.0 8.9 & @xmath122 & @xmath123 & @xmath124 + 8.9 14.8 & @xmath125 & @xmath126 & @xmath127 + 14.8 26.5 & @xmath128 & @xmath129 & @xmath130 + 26.5 50.0 & @xmath126 & @xmath131 & @xmath132 + 50.0 100.0 & & & @xmath133 + ccccc 5.0 6.0 & 5.45 & 0.991 & @xmath134 + 6.0 8.9 & 7.16 & 2.78 & @xmath135 + 8.9 14.8 & 11.1 & 5.50 & @xmath136 + 14.8 26.5 & 18.9 & 10.7 & @xmath137 + 26.5 50.0 & 34.5 & 21.1 & @xmath138 + 50.0 100.0 & 66.4 & 44.0 & @xmath139 + ccc 5.0 6.0 & @xmath140 & @xmath141 + 6.0 8.9 & @xmath142 & @xmath143 + 8.9 14.8 & @xmath144 & @xmath145 + 14.8 26.5 & @xmath146 & @xmath147 + 26.5 50.0 & @xmath148 & @xmath149 + 50.0 100.0 & @xmath150 & @xmath151 + ccc 4.5 6.0 & 0.04 @xmath152 0.01 & 0.04 @xmath152 0.03 + 6.0 8.9 & 0.04 @xmath152 0.01 & 0.04 @xmath152 0.03 + 8.9 14.8 & 0.04 @xmath152 0.01 & 0.01 @xmath152 0.03 + 14.8 25.6 & 0.05 @xmath152 0.01 & 0.01 @xmath152 0.03 + 25.6 50.0 & 0.06 @xmath152 0.02 & + clll 5.45 & @xmath153 & @xmath154 & @xmath155 + 7.16 & @xmath156 & @xmath157 & @xmath158 + 11.1 & @xmath159 & @xmath160 & @xmath161 + 18.9 & @xmath162 & @xmath163 & @xmath164 + 34.5 & @xmath165 & @xmath166 & @xmath167 + 66.4 & & & @xmath168 + | observations of cosmic - ray electrons and positrons have been made with a new balloon - borne detector , heat ( the `` high - energy antimatter telescope '' ) , first flown in 1994 may from fort sumner , nm .
we describe the instrumental approach and the data analysis procedures , and we present results from this flight .
the measurement has provided a new determination of the individual energy spectra of electrons and positrons from 5 gev to about 50 gev , and of the combined `` all - electron '' intensity @xmath0 up to @xmath1 gev .
the single power - law spectral indices for electrons and positrons are @xmath2 and @xmath3 , respectively .
we find that a contribution from primary sources to the positron intensity in this energy region , if it exists , must be quite small . epsf psu - pal-97 - 1 0.4 cm |
black holes and singularities are certainly some of the most peculiar and interesting features of general relativity@xcite . from the perspective of fundamental theory their properties are far from completely understood , especially their relation to quantum theory@xcite ; from the perspective of astrophysics and observational astronomy they seem more and more to play a central role in the physics of objects ranging from stellar to cosmological size@xcite .
most general relativity textbooks devote considerable space to discussing the simplest example of a black hole containing a singularity , the schwarzschild geometry@xcite , and some also discuss kerr s generalization to a rotating system , although only a few derive it@xcite .
most also explain , at least qualitatively , why black holes should exist in the real world , based on the instability of neutron stars of more than about 4 solar masses ; even neutron degeneracy pressure is not sufficient to provide stability .
a few texts also discuss schwarzschild s interior solution for a constant density star@xcite , which clearly illustrates some of the same instability features as more realistic neutron star models .
but the actual dynamical process of collapse , whereby a massive body becomes a black hole , is a more complex dynamical problem , and is often either neglected or treated heuristically and qualitatively . for a brief history of gravitational collapse see reference@xcite .
the study of gravitational collapse dynamics began with the classic seminal work of oppenheimer and snyder@xcite , who set up the equations and gave a semi - quantitative discussion of the collapse of a spherical non - radiating star .
they also found a complete analytic solution for the case of a uniform perfect fluid of zero pressure - now widely referred to as `` dust . ''
this example is less artificial than it might appear since the effect of pressure in a more realistic treatment turns out to be not very critical ; but the treatment by oppenheimer and snyder involves some awkward algebra in its treatment of the boundary between the interior and exterior of the dust star . some textbooks discuss collapse much like oppenheimer and snyder , notably landau and lifshitz@xcite and misner thorne and wheeler@xcite . in particular these texts use a novikov type coordinate system , in which the dust is co - moving in a bound orbit@xcite .
this coordinate system is synchronous and co - moving , and is wonderfully simple conceptually , but almost equally awful algebraically . due to the algebraic difficulty , we believe , this is not the simplest way to treat the problem .
most other texts give a heuristic but convincing analysis by comparing the motion of a dust particle just outside a collapsing dust sphere with the motion of the surface ; matching this motion is equivalent to matching the exterior and interior geometries , which is a rather obvious but important theorem@xcite .
what we have tried to do in the present paper is to analyze dynamical collapse in as simple a way as possible by our choice of coordinates ; we use eddington - finkelstein@xcite type coordinates for infalling incoherent light or `` null dust , '' and painleve - gullstrand@xcite type coordinates for perfect fluids , including dust .
these coordinates seem almost magically well suited to this purpose , as we hope will become apparent in sections 2 and 5 .
we use zero energy fluid systems in contrast to negative energy bound systems as used in references [ 3 ] and [ 5 ] , largely because the algebra is much simpler@xcite .
this choice seems to us well justified because the zero and nonzero energy systems have the same qualitative behavior as they approach the black hole state as seen from outside in terms of standard schwarzschild time , and have exactly the same behavior as they approach the singularity as seen in terms of the proper time of a falling observer .
moreover the zero energy case does not have the awkward conceptual problem of the motion before the maximum size is reached , since the maximum size occurs at an infinite time in the past .
we therefore believe the zero energy case is more illustrative and economical of effort . almost needless to say we have treated only spherically symmetric collapsing systems .
the birkhoff theorem guarantees that a spherically symmetric system does not emit gravitational radiation ; non - spherically symmetric collapse generally entails gravitational radiation , making the problem vastly more complex , though also more interesting . to achieve our goal of simple textbook type examples of collapse , we must pay a price , which is some degree of artificiality .
the main artificiality is the spherical symmetry ; also the in - falling shells of light or other massless material of sections 2 and 3 are not likely to be found in nature ; similarly we do not expect to find the perfect fluid sphere surrounded by an elastic shell of section 5 , or the zero pressure perfect fluid of sections 6 and 7 . since our purpose in the present paper is overtly pedagogical , much of what we obtain here
is known and available somewhere in the vast research literature , although obtained with different techniques and different coordinates .
however we believe our treatment is the simplest , both conceptually and algebraically , mainly due to our coordinate choice .
we do not know of any use of painleve - gullstrand coordinates in the present context , although they are becoming more widely used@xcite ; as will be seen in section 6 these coordinates combine a schwarzschild type radial coordinate and a friedmann - robertson - walker ( frw ) type cosmological time coordinate to give a natural interpolating system for describing both the interior and exterior of a fluid sphere .
while the eddington - finkelstein and painleve - gullstrand coordinates serve their mathematical purpose quite well , they have a number of drawbacks such as a time `` marker that is not well - behaved everywhere in spacetime , and they are not synchronous ; we have therefore also treated the collapse of a thin shell of light by transforming to kruskal - szekeres coordinates , which are conformally flat and thus display the causal structure of the collapse process quite clearly@xcite . in kruskal - szekeres
coordinates one has a nice picture for distinguishing real world black holes from ' ' eternal " black holes , with their associated wormholes and white hole segments , none of which may be expected to actually exist in our universe .
we use throughout this paper a rather novel technique for handling the boundary conditions between the fluid and the exterior , which we believe is physically clear .
since the pioneering work of oppenheimer and snyder there has been an enormous amount of work on gravitational collapse and related topics , most directed to research rather than pedagogy .
the present pedagogical paper originated from several research problems , one involving a formation mechanism for singularity - free black holes filled with heavy vacuum , and another involving a finite version of the standard or concordance " cosmological model .
for these we needed analytic collapse models that were algebraically simpler and more explicit than we found in the literature .
our hope is that our methods and models might also be of use to students and teachers in illustrating more standard gravitational collapse .
this paper is organized as follows : the first part ( sections 2 to 4 ) deals with collapsing shells of incoherent light or null dust , which is entirely characterized by its null 4-velocity and energy density ; the second part ( sections 6 and 7 ) deals with spheres of perfect fluid , including some with nonzero pressure and some with non - uniform density . in section 2
we use eddington - finkelstein coordinates to obtain the metric for the collapse of a thin light shell onto a black hole to form a larger black hole , and the collapse of a thin light shell to form a black hole ab initio .
the ab initio collapse is amusing in that it is almost certainly the simplest complete scenario for the formation of a black hole , albeit it a rather artificial one . in section 3
we use the results of section 2 to obtain the metric for the collapse of a thick shell of light , layer by layer .
the density profile of the shell is largely arbitrary , and we give one specific example . in section 4
we transform the metric for the thin light shell collapse to conformably flat kruskal - szekeres coordinates in order to display the causal structure of the geometry in a clear way - nearly equivalent to a penrose diagram ; as already noted we use these coordinates to emphasize the difference between black holes formed by collapse and eternal schwarzschild black holes . in section 5
we present our way of doing the classic problem of collapse of a uniform fluid sphere , using painleve - gullstrand coordinates for maximum simplicity .
we also obtain an explicit solution for a nonzero pressure fluid with a linear equation of state : @xmath0 ; to balance the pressure gradient force at the surface we use the artifice of a thin elastic shell with tangential pressure instead of the slowly decreasing radial pressure of a real stellar system ; our model is essentially a balloon with uniform internal density and pressure .
the stress energy tensor of the surface is described by the radius of the sphere , its energy density , and the parameter @xmath1 ; for @xmath2 we recover the standard results for the dust ball . in section 6
we deal with zero pressure non - uniform dust spheres , with the density characterized by a largely arbitrary function ; again using painleve - gullstrand coordinates we build up the system layer by layer , to obtain the standard results of oppenheimer and snyder and landau and lifshitz etc . we do not allow the dust layers to cross , which could give rise to infinite densities and superficial singularities@xcite .
we give one specific example of the metric of a non - uniform sphere ; the special case of uniform density agrees with the analysis of section 5 .
since this paper is mainly pedagogical we have throughout sacrificed brevity and included considerable algebraic detail ; only elementary mathematical methods are used .
it is worth mentioning that some simple and fairly obvious extensions of the present work can be made .
first , by a reversal of time the collapse of light shells can be viewed as outgoing radiation , so we may easily obtain results like those of vaidya for the geometry of a radiating body@xcite .
similarly the collapse of the perfect fluid can be time reversed to yield the metric for a black hole emitting matter , that is a type of white hole .
also it is straightforward to include a cosmological constant term in the equations to obtain the metric for a collapsing system in an exterior schwarzschild de sitter geometry ; due to its somewhat more cumbersome algebra we have not included this in the present paper , and leave it as an exercise .
( see reference@xcite . )
we will say more about further extensions and applications in section 8 .
the motion of light in schwarzschild geometry is algebraically awkward at and inside the schwarzschild radius when schwarzschild coordinates are used , but it becomes quite simple in eddington - finkelstein ( ef ) coordinates . in this section we use ef coordinates to study the motion of light , the growth of a black hole by absorption of light , and the ab initio formation of a black hole by a thin shell of light .
the last process is most likely the simplest model for the formation of a black hole by gravitational collapse .
the schwarzschild metric in schwarzschild coordinates is , with @xmath3 , @xmath4 where @xmath5 and @xmath6 .
the black hole surface is a sphere at the schwarzschild radius ( twice the geometric mass ) @xmath7 .
it is both an infinite redshift surface where @xmath8 and a null surface or one - way - membrane . in terms of the schwarzschild time
coordinate both light and particles take an infinite time to reach the black hole surface from the exterior . specifically , for light falling radially from @xmath9 @xmath10 thus light never reaches the surface but approaches asymptotically with a characteristic time @xmath11 .
it is for this reason that we introduce the ef time coordinate , in terms of which the surface is reached in a finite time .
the ef time coordinate is obtained from the schwarzschild time by a radially dependent shift @xmath12 this leads to the ef form of metric , @xmath13 provided that we choose the transformation function @xmath14 to obey @xmath15 the solution to this is @xmath16 which has an infinite stretch at @xmath17 .
henceforth we refer to the coordinates and metric form in eq.([ef ] ) as ef for any function @xmath18 .
the ef form is extremely convenient because radially infalling light behaves quite simply if the plus sign in eq.([ef ] ) is chosen .
for such light we set the line element equal to zero , to obtain @xmath19 so that @xmath20 thus the path of infalling light is independent of the metric function @xmath21 , and is the same as in flat space .
see fig .
[ thinlightshellef ] .
conversely , outgoing light `` stalls '' where @xmath22 and @xmath8 , that is at the infinite redshift surface
. we will soon need the einstein tensor for the ef metric form eq.([ef ] ) .
its nonzero components are easily calculated to be @xmath23 here a dot denotes a time derivative and a prime denotes a radial derivative .
we define the stress energy tensor in terms of the einstein tensor by using the field equations @xmath24 for pure schwarzschild geometry the einstein tensor is everywhere zero , as is easily verified .
we now study the formation of a black hole by a single thin shell of light , and in the next section we will consider a thick shell of light . for both purposes we begin with a thin shell of light falling into a schwarzschild black hole with radius @xmath11 to form a black hole with radius @xmath25 as shown in fig .
[ thinlightshellef]a ; the light shell obeys @xmath26 everywhere .
we may write the metric in all of spacetime using a step function @xmath27 and its complement @xmath28 , as @xmath29 which is of course time dependent . substituting this into eq.([pgg ] ) and eq.([einsteinequation ] )
we obtain the stress - energy tensor for the thin light shell , which is singular due to the step function , @xmath30 where @xmath31 and @xmath32 .
the null vector @xmath33 corresponds to infalling light . from the energy density
@xmath34 we may calculate the energy or effective mass of the light shell by going to a time in the distant past when the shell was in asymptotically flat space , @xmath35 this verifies that the mass of the initial black hole plus the energy or effective mass of the light shell equals the mass of the final black hole .
the above results hold for the special case when the initial black hole is replaced by flat space , or @xmath36 .
note that everything above is consistent with the interior of a spherical thin light shell being flat minkowski space .
this represents the formation of a black hole by the gravitational collapse of a single thin light shell ( or other massless material ) , as shown in figure 1b ; this would seem to be the simplest complete example of gravitational collapse .
our results from the preceding section may be used to construct a model for the collapse of a thick light shell , layer by layer .
the initial state is a sequence of concentric thin shells with a flat minkowski interior , as shown in fig .
[ thicklightshellef]a .
each region of spacetime has a schwarzschild geometry as discussed in the previous section .
the innermost shell with energy @xmath37 collapses to form a black hole of mass @xmath38 , followed by others to form intermediate black holes of mass @xmath39 , ending with a final black hole with mass @xmath40 and schwarzschild radius @xmath41 . between the shells the geometry is given by eq.([ef ] ) , with a metric function @xmath42 the evolving infinite redshift surface , defined by @xmath22 , is the zigzag line in fig .
[ thicklightshellef]a .
the continuous analog of the sequence is a family of concentric shells with a continuous label @xmath43 , which we choose to run from 0 to 1 .
the parameter @xmath43 plays the same role as the discrete label @xmath44 : that is the energy or effective mass inside the @xmath43 shell is denoted by @xmath45 . as in eq.([discreteu ] ) the metric function in the light shell is @xmath46 the evolving infinite redshift surface is the smooth line in fig . [ thicklightshellef]b . a labeling scheme , which is convenient for both the present light shell and the fluid systems to be discussed later , is to take the energy inside the @xmath43 shell to be proportional to @xmath43 , @xmath47 the time at which the @xmath43 shell reaches the center may be chosen almost arbitrarily as a function of @xmath43 , subject to the constraints that it be 0 for the innermost shell @xmath48 and increase monotonically to the final time @xmath49 for the outermost shell @xmath50 , as shown in fig .
[ thicklightshellef ] .
we denote this function as @xmath51 , for reasons that will become apparent .
the equation of motion for the shell @xmath43 is thus @xmath52 from its definition @xmath53 must have an inverse , so we may invert this to obtain @xmath54 with @xmath55 and @xmath56 .
thus @xmath57 serves as a density profile function , and the metric in the interior of the light shell may be written as @xmath58 the infinite redshift surface , defined by @xmath22 , is thus determined by @xmath59 inverting this we obtain for the infinite redshift surface @xmath60 this implies that if the total duration of collapse @xmath49 is sufficiently large then @xmath61 must be a positive monotonic function of @xmath62 . using the metric function eq.([continuousu ] )
we may calculate the einstein tensor and the stress - energy tensor from eq.([efg ] ) and eq.([einsteinequation ] ) .
this gives for the light shell , @xmath63 where @xmath64 denotes the derivative of @xmath57 with respect to its argument , and @xmath33 is the null vector defined in eq.([thinlightshelleft ] ) . from this
we obtain the energy density and the total energy of the light shell as @xmath65 where the integral is again done in the asymptotically flat spacetime of the distant past , verifying that the energy of the light shell is equal to the final black hole mass . our approach to boundaries and boundary conditions
is somewhat unorthodox ; we do not impose boundary conditions per se on the metric function between regions of spacetime .
instead we use the above solutions ( in vacuum and within the light shell ) to calculate the stress - energy tensor , which is defined by the field equations and eq.([efg ] ) .
the resulting stress energy tensor must be zero in vacuum , correctly represent the stress energy within the light shell , and describe the stress on the boundaries .
if it does , the solution makes physical sense . for the region near the inner and outer boundaries of the light shell we may write the metric function as @xmath66 calculating the stress energy tensor from eq.([efg ] ) and eq.([einsteinequation ] ) we find that there is no singular shell at these boundaries since the relevant singular derivatives cancel ; thus the stress - energy tensor associated with the boundaries is zero and the solution is thus physically reasonable . as a specific example
let us take the density profile function to be linear , @xmath67 this corresponds to a constant rate of energy impacting the center . from eq.([continuousu ] ) the metric function in the various regions is , @xmath68 the infinite redshift surface , from eq.([thicklightshellinfiniteredshiftsurface ] ) , is the linear function @xmath69 the slope of this is positive for @xmath70 , that is when the light shell thickness is greater than its schwarzschild radius . within the light shell the energy density is , from eq.([efg ] ) and eq.([einsteinequation ] ) , @xmath71 which is independent of time . in fig .
[ thicklightshellefhorizon ] we show a qualitative sketch of some outgoing light rays for a fairly general light shell .
the sketch is made by noting that , from eq.([eflight ] ) : the slope of outgoing rays is 1 in the minkowski region and at large distances it approaches 1 ; the slope is 0 along the infinite redshift surface ; finally , the slope is -1 at the center of the black hole .
there is a last - ray - out emitted from the origin to infinity , after which all outgoing rays ( as well as particles ) are trapped within the surface at @xmath11 and eventually fall into the singularity .
the surface defined by the last - ray - out is thus a global horizon .
this illustrates that the horizon and the infinite redshift surface are quite different inside the time dependent light shell , unlike the situation for the time independent schwarzschild geometry . as a further application of our methods we note that our results can be reversed in time to describe radiation being emitted by a spherically symmetric system such as a star or black hole @xcite .
to do this we use the minus sign for the metric in eq.(4 ) and turn the diagram in fig.2b upside down .
it is then easy to modify our algebraic results to describe any reasonable light shell density profile .
this may be useful in studying the final stages of black hole evaporation , when the gravitational field of the radiation becomes comparable to that of the black hole and can not be neglected @xcite ; it might help in determining if a black hole radiates entirely away to vacuum or leaves behind a remnant @xcite .
the thin light shell discussed in section 2 is probably the simplest model of gravitational collapse . however in ef coordinates @xmath72 is negative for the interior of the black hole .
this means that if one is at coordinate rest @xmath73 then the square of the proper time interval @xmath74 has the opposite sign of the square of the coordinate time interval @xmath75 , so that @xmath61 may not be interpreted as a good time marker in that region of spacetime .
the same is true of schwarzschild coordinates .
kruskal - szekeres ( ks ) coordinates , which are discussed in many texts , were developed to solve this problem@xcite . here
we will obtain ks type coordinates for the thin light shell collapse of section 2 . by ks coordinates
we mean a system in which the @xmath76 part of the metric is conformal to flat minkowski space , with no singularities or zeroes .
we first show how a metric in ef form , with @xmath77 , may be put into conformal form . the metric eq.([ef ] )
may be factored as follows , with angular dependence suppressed , @xmath78 the quantities @xmath79 and @xmath80 are termed conformal null coordinates since the line element is zero along lines of constant @xmath79 or @xmath80 ; these thus represent radially moving light rays .
we transform to other null conformal coordinates by choosing any functions @xmath81 and @xmath82 , so that @xmath83 so the metric in terms of @xmath84 is @xmath85 the metric function @xmath86 may be considered to be a function of @xmath62 and @xmath61 , or @xmath87 and @xmath88 .
we may also relate the null coordinates to lorentz - like coordinates , @xmath89 thus @xmath90 and @xmath91 may be interpreted as time and radial coordinates provided that @xmath86 is positive and has no singularities or zeros . outside the schwarzschild radius and the light shell
the function @xmath92 and a convenient choice for the functions @xmath87 and @xmath88 are the following @xmath93 where @xmath94 this transformation is chosen to make @xmath95 independent of time . in order that @xmath95 also be nonsingular and have no zeros we choose @xmath96 , and to map the exterior region to the @xmath97 quadrant we choose @xmath98 and @xmath99 .
then the transformation and metric functions are explicitly @xmath100 note that @xmath101 and @xmath102 are dimensionless and @xmath95 is the square of a distance . for the interior of the black hole , @xmath103 , and outside the light shell ( see fig.1b )
we choose the opposite sign for the logarithm and @xmath104 in eq.([schwarzschildkstransformation ] ) , to obtain the same expression as in eq.([schwarzschildkstransformationout ] ) , which is thus valid throughout the schwarzschild geometry .
the expressions eq.([schwarzschildkstransformationout ] ) are similar to the standard ones used to transform between schwarzschild coordinates and ks coordinates , but differ in important ways .
lines of @xmath105const .
map to hyperbolae with @xmath106 ; in particular @xmath107 corresponds to @xmath108 and @xmath17 corresponds to @xmath109 .
the lines @xmath110 and @xmath17 both map to @xmath111 , while @xmath112 maps to @xmath113 .
( unlike the case with schwarzschild coordinates lines of constant @xmath61 do not map to rays of @xmath114 ! ) see fig .
[ thinlightshellks ] .
there are many ways to transform from ef to ks coordinates for the minkowski geometry inside the light shell ; of course minkowski geometry in ef coordinates is already in ks form , but not one that is useful to us . we must choose a transformation that joins continuously with the transformation eq.([schwarzschildkstransformationout ] ) along the boundary line @xmath26 , and we also demand that the function @xmath86 be continuous along that line . for that space the metric function @xmath115 , so that from eq.([efks ] ) @xmath116 . thus from eq.([ks ] ) the metric function @xmath86 is @xmath117 equating this with @xmath95 from eq.([schwarzschildkstransformationout ] ) along @xmath26 we obtain a differential equation for @xmath118 @xmath119 that is , in terms of the argument , denoted @xmath120 , @xmath121 for continuity we choose @xmath122 to be the same as in the schwarzschild region eq.([schwarzschildkstransformationout ] ) , so with appropriate constants we arrive at @xmath123 these functions are obviously equal to those in eq.([schwarzschildkstransformationout ] ) along the line @xmath26 , as desired .
note that the price we pay for having @xmath95 in the schwarzschild region independent of time is that @xmath124 in the minkowski region is dependent on time .
the nature of the transformation eq.([minkowskiks ] ) is best seen in terms of the mapping of some lines and points : @xmath110 maps to @xmath125 ; @xmath112 maps to @xmath126 ; lines of constant @xmath127 map to constant @xmath87 lines ; lines of constant @xmath128 map to constant @xmath88 lines ; in particular @xmath26 maps to @xmath129 and @xmath130 maps to @xmath131
. finally the origin @xmath107 maps to @xmath132 so @xmath133 henceforth we drop the subscripts on @xmath87 and @xmath88 .
figure [ thinlightshellks ] show the complete collapse process , with the schwarzschild and minkowski geometries stitched together along the line @xmath26 .
only the region to the right of the line corresponding to @xmath107 has physical meaning .
it is evident from the figure that the line @xmath125 defines a horizon , and it is also apparent that the singularity at @xmath107 does not behave like a time independent spatial position .
note also that @xmath91 serves as a radial marker , even though regions of the spacetime have negative @xmath91 .
figure [ thinlightshellks ] may be compared with the `` pure '' or eternal schwarzschild geometry discussed in many textbooks @xcite and shown in ks coordinates in fig .
[ schwarzschildks ] ; this shows the `` maximum analytic extension '' of the schwarzschild solution .
the region @xmath134 is interpreted as a white hole , and is absent in figure 4 ; also absent is the region @xmath135 , interpreted as the `` other side of the wormhole , '' and part of the interior region @xmath136 .
this illustrates that if the formation of the black hole by gravitational collapse is taken into account then the much - discussed white hole and wormhole regions are not present ; there is no reason to expect that such regions occur in nature , as emphasized by wheeler and many others @xcite .
the spherical shells of light used in the preceding sections provide very simple models of gravitational collapse but are rather artificial and do not approximate anything we expect to find in nature .
we now turn to a more realistic system , a sphere of uniform perfect fluid with a linear equation of state .
this includes our approach to the special case of a dust ball , the original @xcite and still a favored system for collapse studies @xcite .
we believe our approach is the simplest available , because the painleve - gullstrand ( pg ) coordinate system is remarkably well suited to the task @xcite .
some of the mathematical techniques we used for light shells , such as use of concentric layers and the handling of the boundary conditions , will also prove useful for fluid spheres .
the fluid collapse involves only two spacetime regions , the fluid interior , and the schwarzschild exterior .
a crucial step is to find a metric form ( the pg form ) which describes both regions simply , and in which the motion of the fluid is simple .
the spacetime geometry corresponding to a uniform fluid is well known from cosmology ; it is described by the friedmann - robertson - walker metric in co - moving coordinates , which covers all of spacetime from the big bang onwards [ 3,4 ] . for simplicity
we consider the spatially flat case , that is @xmath137 , for which the metric is @xmath138 ( this corresponds to zero energy collapsing system . ) the co - moving radial coordinate @xmath139 is dimensionless , while the scale function @xmath140 is a solution of the cosmological equations , with the dimension of a length . to describe a finite sphere
we truncate the radial coordinate at some value .
if the fluid has a linear equation of state , with pressure and density obeying @xmath0 , then the scale function is a power of @xmath61 , specifically @xmath141 in particular dust ( or cold matter " ) has negligible pressure so @xmath2 and @xmath142 , while radiation ( or hot matter " ) has @xmath143 and @xmath144 .
this is the range of normal matter . in the cosmological scenario time
runs from the big bang at @xmath145 to infinity , but in the collapse scenario it will run from negative infinity to @xmath145 .
that is , a very large fluid sphere in the far distant past collapses toward zero size at @xmath145 . to obtain the desired form for the metric inside the fluid we introduce a new radial coordinate , @xmath146 which is not co - moving , to obtain the metric @xmath147dt^2 + 2(r\dot{a}/a)drdt - dr^2 - r^2 d\omega^2 \nonumber \\ & = & [ 1-(nr / t)^2]dt^2 + 2(nr / t)drdt - dr^2 - r^2 d\omega^2 .
\nonumber \\ \label{uniformfluidpg}\end{aligned}\ ] ] this contains the single metric function @xmath148 and is distinctive in having a cross term and @xmath149 .
it has an infinite redshift surface where @xmath8 , or @xmath150 .
the minus sign is appropriate since we will deal with negative times .
the empty region exterior to the fluid is described by the schwarzschild metric in eq.([schwarzschild ] ) , which we now write in the form @xmath151 to make this compatible with the metric in eq.([uniformfluidpg ] ) in the fluid we choose a new time coordinate that makes @xmath149 .
taking @xmath152 we find that @xmath153 provided that @xmath14 obeys @xmath154 as expected this transformation involves an infinite time stretch at the schwarzschild radius , where @xmath155 . for the schwarzschild region the solution to eq.([pgtequation ] ) is @xmath156 the metric eq.([pg ] ) and the transformation eq.([pgtsolution ] ) are those obtained by painleve and gullstrand @xcite .
both regions of spacetime are now described by the metric form eq.([pg ] ) , with @xmath157 allowed to be a function of both @xmath62 and @xmath61 ; specifically @xmath158 we choose the positive signs in eq.([pgtequation ] ) and negative sign in eq.([pgtsolution ] ) to correspond to collapse during negative time , and refer to the metric form and coordinates as generalized painleve - gullstrand or simply pg .
the pg metric has a remarkable property that is crucial to our analysis .
the geodesic equations for radial motion of a particle in the metric eq([pg ] ) lead , after some algebra , to @xmath159 , \quad \quad \psi^{\prime } \equiv \frac{\partial \psi}{\partial r } , \nonumber \\ 1 & = & ( 1-\psi^2)(\frac{dt}{ds})^2 + 2\psi\frac{dt}{ds}\frac{dr}{ds } - ( \frac{dr}{ds})^2 . \label{pggeodesicequation}\end{aligned}\ ] ] one obvious solution to the first is @xmath160 thus coordinate time and proper time intervals are equal for such a freely falling particle , and this holds for any metric function @xmath157 . the second equation in eq.([pggeodesicequation
] ) now becomes quite simple @xmath161 in the schwarzschild region this is the same as the classical newtonian equation for a radially falling test particle of zero energy .
we will soon need the einstein tensor and the stress - energy tensor . from the pg metric form it is straightforward and only slightly tedious to calculate these .
the nonzero components of the einstein tensor are @xmath162 as before we define the stress - energy tensor via the field equations @xmath163 ; in particular the energy density is @xmath164 thus the density is uniform , as expected .
we now join the spacetime regions inside and outside the fluid by demanding that the metric function @xmath157 in eq.([uniformfluidpsi ] ) be continuous across the fluid surface boundary .
this gives @xmath165 the geodesic equation for a zero energy falling particle in the schwarzschild exterior region is @xmath166 this agrees with the motion of the surface eq.([uniformfluidboundary ] ) only for the case of @xmath142 , that is dust .
thus a freely falling particle may hover at the falling surface of a dust ball , as it should .
however , if the pressure is nonzero and @xmath167 then the surface will fall more rapidly .
this is due the stress in the surface layer , as we will see .
figure [ uniformfluidspherepg ] shows the collapse scenario in pg coordinates .
it is straightforward to calculate the stress - energy tensor for the fluid surface with the same technique we used in section 3 . in the vicinity of the surface
the metric function @xmath157 may be written @xmath168 this leads to a singular stress - energy tensor @xmath169 for a zero pressure fluid , @xmath2 and @xmath142 , this vanishes as expected . if the pressure is not zero it represents a surface tension , as we will now show .
the surface tension of a fluid sphere is related to its radius and pressure by @xmath170 ; in the present case this implies from eq.([frwsolution ] ) and eq.([uniformfluidt00 ] ) , @xmath171 which agrees with eq.([uniformfluidt22 ] ) .
thus the stress - energy tensor indeed represents a surface tension that balances the internal pressure of the fluid to keep it stable .
this is why the surface falls faster than a freely falling particle . in summary the collapse of a uniform fluid sphere
is described by the metric function in eq.([pg ] ) and eq.([uniformfluidpsi ] ) , the density in eq.([uniformfluidt00 ] ) , and the surface tension in eq.([uniformfluidt22 ] ) .
its collapse is qualitatively similar to that of the light shells , but rather less artificial .
our use of surface tension to stabilize the surface is a conceptually simple substitute for the gradual pressure gradient of a more realistic model ; such tangential pressures are also mentioned by singh in ref.[26 ] and in the references contained therein .
we next consider the collapse of a sphere filled with zero pressure perfect fluid , that is a dust ball .
this problem is discussed by landau and lifshitz@xcite using lemaitre - like coordinates@xcite . for simplicity and physical clarity
we use the same technique as in section 3 for constructing thick light shells ; that is we build the dust ball layer by layer from a sequence of thin dust shells .
the layers are prohibited from crossing to prevent superficial singularities due to the consequent infinite density@xcite .
pg coordinates will again be used , with the metric form eq.([pg ] ) containing a single metric function @xmath172 .
the einstein tensor for this is given in eq.([pgg ] ) , and we recall that in pg coordinates we may take proper time and coordinate time intervals to be equal along zero energy particle geodesics . in analogy with section 3
we begin with a thin shell of dust falling onto a black hole , shown in fig .
[ dustshellpg ] .
the shell is assumed to be very light , so that @xmath173 and @xmath174 differ by a small @xmath175 , and the schwarzschild radii differ by @xmath176 . as in section 5 the metric function in the two schwarzschild regions
is @xmath177 the equation of the thin dust shell is that of a geodesic for a zero energy particle in the schwarzschild geometry , given in eq.([pggeodesicsolution ] ) .
it is important that coordinate and proper time intervals are equal along the geodesic , so that the equation for the boundary is a relation between the coordinates @xmath62 and @xmath61 . to obtain the stress - energy tensor for the dust shell we write the metric in the vicinity of the boundary as @xmath178 where @xmath179 as before
from the einstein tensor in eq(refpgg this leads , as in section 3 , to a singular energy density for the dust shell , @xmath180 from this we may calculate the mass of the shell by going to large negative times when the shell is in nearly flat space .
@xmath181 we thus verify that the mass of the initial black hole plus the shell mass is equal to the mass of the final black hole .
this elementary result allows us to construct a rather general dust ball with non - uniform density from layers of thin shells . a discrete sequence of thin shells is shown in fig . [ dustspherepg]a .
the initial system is a black hole of vanishingly small mass @xmath182 with a thin shell of mass @xmath37 collapsing onto it at @xmath145 to give a black hole of mass @xmath183 , followed by more shells , and ending with a final shell impacting the origin at @xmath184 to give a final black hole of mass @xmath40 and schwarzschild radius @xmath185 . in terms of the intermediate masses
@xmath186 the metric function in the regions between shells is @xmath187 the infinite redshift surface where @xmath188 and @xmath8 is the zigzag line in the figure .
a continuous version of the dust shell sequence is shown in figure [ dustspherepg]b .
the shells are labelled by a continuous variable @xmath43 ranging from 0 to 1 , with the total geometric mass inside a shell denoted by @xmath45 ; @xmath45 is the continuum analog of @xmath186 , and the metric function in the fluid region is thus @xmath189 the infinite redshift surface is where @xmath190 , or @xmath191 as in section 3 we label the shells by the total mass inside the shell , @xmath192 where @xmath40 is the final mass .
the time at which the @xmath43 shell impacts the center may be chosen rather arbitrarily , and we denote it by @xmath193 . for simplicity and to avoid density singularities
it is important that the shells do not cross each other .
this will be true if @xmath193 increases monotonically ; that is , outer shells impact at times later than inner shells .
thus we choose @xmath193 to increase monotonically from 0 to @xmath49 as @xmath43 runs from 0 to 1 .
the equation for the @xmath43 shell is then the free fall equation for a particle of zero energy , with the particle reaching the origin at @xmath193 , or @xmath194 = 0 .
\label{dustgeodesicequation}\end{aligned}\ ] ] as we will see , the function @xmath193 determines the dust ball density .
if the function @xmath193 is specified we can use eq.([dustinfiniteredshiftsurface ] ) and eq.([mlambda ] ) and eq.([dustgeodesicequation ] ) to get a parametric expression for the infinite redshift surface ; the radius and time are given in terms of @xmath43 by @xmath195 in particular for the central shell with @xmath48 , which impacts the center at @xmath145 , and the outer shell with @xmath50 , which impacts at @xmath184 , we have for the infinite redshift surface @xmath196 thus the surface will move outward with time if @xmath197 that is if the mass impacts the origin sufficiently slowly .
curiously the same expression occurred for the light shell collapse in eq.([thicklightshellinfiniteredshiftsurfaceexample ] ) .
for the special case of zero total impact time , @xmath198 , all of the shells impact the center at @xmath145 .
this is corresponds to the uniform density fluid sphere in the previous section , so @xmath199 clearly implies uniform density .
we can explicitly see the relation between @xmath193 and the density by solving the shell equation eq.([dustgeodesicequation ] ) for the mass @xmath45 inside the shell @xmath43 at a large negative time @xmath200 , when @xmath201 , to get @xmath202 r^3 \approx \frac{2}{9t^2 } r^3.\end{aligned}\ ] ] thus in the distant past the mass is proportional to @xmath203 , meaning that the dust is asymptotically uniform .
as time progresses to smaller negative values the density may deviate more and more from uniform , and may be quite non - uniform at @xmath145 . to obtain the metric and other properties of a collapsing dust ball
we first specify a function @xmath204 .
then the shell equation eq.([dustgeodesicequation ] ) gives the shell parameter as a function of position , that is @xmath205 .
this is consistent because we allow only one shell to pass through a given point . with the parameter known as a function of position the metric in the fluid
is given by eq.([dustspherepsi ] ) .
with this expression for the shell parameter the shell relation eq.([dustgeodesicequation ] ) is an identity in @xmath206 .
a number of interesting properties of the dust region may be expressed in terms of the function @xmath193 .
( later we will consider a specific @xmath193 to illustrate further . ) from the expression eq.([dustspherepsi ] ) for the metric function @xmath157 and @xmath207 we may calculate the einstein tensor from eq.([pgg ] ) , obtaining @xmath208 for the other components we need to relate the derivatives @xmath209 and @xmath210 to each other , which may be done using the shell relation eq.([dustgeodesicequation ] ) .
differentiating eq.([dustgeodesicequation ] ) we find @xmath211 thus the ratio of derivatives is independent of the function @xmath204 . with the use of eq.([lambdadotprime ] ) it is only slightly tedious to show that , for any @xmath204 , @xmath212 thus the pressure terms of the stress - energy tensor are zero , as we expect for dust .
moreover the energy density of the dust is @xmath213 from this we may verify that the total mass of the dust ball is the mass of the final black hole , which we do by again going to large negative times when the dust is in nearly flat space , @xmath214 finally the stress energy tensor on the dust ball surface is zero according to eq.([surfacetension ] ) with @xmath142 .
there are two specific examples of the function @xmath193 for which eq.([pgg ] ) is easily solved .
the first example is @xmath215 we leave it as an exercise to the reader to show that the metric function in the dust is then @xmath216.\end{aligned}\ ] ] in the limit of @xmath217 this gives the same result as eq.([uniformfluidpsi ] for a uniform dust ball .
the energy density is @xmath218\nonumber\\ & \approx & 1/(2\pi g t^2 ) , \quad \quad \textrm{for large $ t$.}\end{aligned}\ ] ] this is a well - behaved positive function with no singularities or zeros for negative @xmath61 . the second example is @xmath219 where @xmath220 is a constant .
this is singular and thus quite different from the first example , and is not represented by fig.8b : it is singular for @xmath221 , meaning that the central dust layer collapsed to the center in the infinite past and the outer layer reached the center at time @xmath222 .
we again leave it as an exercise to show that the metric function is @xmath223 and that the density is @xmath224 the singularity at @xmath107 is mild in the sense that the mass inside a sphere goes like @xmath225 . as @xmath226 the density approaches infinity as expected .
thus this example represents a dust ball with a mildly singular central density in the distant past , which grows stronger with time until complete collapse , and represents a rather realistic and amusing system @xcite .
finally , we briefly note some properties of light rays in the collapsing dust ball . from the pg metric eq.([pg ] ) we may write the coordinate velocity of light as @xmath227 we generally expect that @xmath157 will be zero at @xmath107 , except near @xmath145 , so @xmath228 there ; note that this is not true for the second example considered above . also , @xmath228 for asymptotically large distances . on the infinite redshift surface
@xmath190 so that @xmath229 for outgoing light and @xmath230 for infalling light . for the center in the schwarzschild region
this allows us to make a rough qualitative sketch of some light rays in fig.9 .
( for @xmath232 eq.([pglight ] ) may be solved exactly . )
note that , as in the case of the thick light shell in section 3 , there is a last light ray out , whose path defines a global horizon inside of which neither particles nor photons may escape to infinity .
[ dustspherepghorizon ]
in this paper we have tried to present a simple introduction to the dynamics of gravitational collapse , which we hope can provide a bridge between basic textbook general relativity and numerous research topics of current interest .
the light shells in ef coordinates and the fluid spheres in pg coordinates provide the simplest way we know of to study collapse .
some simple but amusing extensions of these models can be made , as already mentioned : one may reverse the time and study the emission of light and matter from white holes or other spherically symmetric objects , and it is easy to add a cosmological constant in the fluid collapse .
the present research literature is dense with studies involving collapse , ranging from fundamental theory to astrophysical applications .
a sampling of recent topics from the los alamos archives xxx.lanl.gov includes the following , which we paraphrase : @xmath233 entropy in collapse to a black hole - where does the information go ?
@xmath233 unitarity - is collapse consistent with unitary quantum evolution ?
@xmath233 cosmic censorship - are all singularities surrounded by a horizon ?
@xmath233 collapse in context of string theory , anti de sitter space @xmath233 quantization and entropy of the surface of a black hole .
@xmath233 collapse in diverse dimensions .
@xmath233 scalar tensor ( or other ) gravity theories and collapse @xmath233 collapse with a cosmological constant included .
@xmath233 particle production in collapse .
@xmath233 role of pressure ( radial or tangential ) in collapse . @xmath233 stability of stars against collapse due to rotation .
@xmath233 gravitational radiation from ( non - spherical ) collapse .
this work was supported by nasa grant 8 - 39225 to gravity probe b , and by the us department of energy under contract no.de-ac02-76sf00515 .
we thank the members of the gravity probe b theory group for many critical and stimulating discussions , in particular francis everitt , robert wagoner , and alex silbergleit .
b. k. harrison , k. s. thorne , m. wakano , j. a. wheeler , _ gravitational theory and gravitational collapse _ ,
( u. of chicago press , 1965 ) .
this rather early book is mainly devoted to equilibrium configurations and their stability , with dynamical collapse finally discussed in chapter 11 . | most general relativity textbooks devote considerable space to the simplest example of a black hole containing a singularity , the schwarzschild geometry .
however only a few discuss the dynamical process of gravitational collapse , by which black holes and singularities form .
we present here two types of analytic models for this process , which we believe are the simplest available ; the first involves collapsing spherical shells of light , analyzed mainly in eddington - finkelstein coordinates ; the second involves collapsing spheres filled with a perfect fluid , analyzed mainly in painleve - gullstrand coordinates .
our main goal is pedagogical simplicity and algebraic completeness , but we also present some results that we believe are new , such as the collapse of a light shell in kruskal - szekeres coordinates . |
a dozen years ago , it was shown that introducing the 1918 hemagglutinin ( ha ) confers enhanced pathogenicity in mice to recent human viruses that are otherwise non - pathogenic in this host . moreover , like the 1918 one , these recombinant viruses infect the entire lung and induce high levels of macrophage - derived chemokines and cytokines , which results in infiltration of inflammatory cells and severe haemorrhage @xcite . in macaques ,
the whole 1918 virus causes a highly pathogenic respiratory infection that culminates in acute respiratory distress and a fatal outcome @xcite . although the 1918 polymerase genes were found essential for optimal virulence , replacing the 1918 ha by a contemporary human one proved enough for abolishing the lethal outcome of the 1918 virus infection in mices @xcite , further underlining the key role of the 1918 ha in the deadly process .
the first 1918 ha sequences were obtained in 1999 from formalin - fixed , paraffin - embedded lung tissue samples prepared during the autopsy of victims of the influenza pandemic , as well as from a frozen sample obtained by _ in situ _ biopsy of the lung of a victim buried in permafrost since 1918 @xcite . since then , the number of ha sequences determined each year has grown dramatically , jumping from @xmath0100 in the nineties to @xmath03,000 per year nowadays .
the goal of the present work is to take advantage of this wealth of data for identifying features that are unique to the 1918 sequence , the underlying hypothesis being that they may prove responsible for the unique behaviour of viruses displaying the 1918 ha . -0.3
52006 ha different protein sequences were retrieved , 2016 . ] from the ncbi influenza virus resource @xcite , sequences coming from laboratory viral strains being disregarded .
multiple pairwise sequence alignment was performed using blast @xcite , version 2.2.19 , taking as a reference the long h1 sequence from virus a / thailand / cu - mv10/2010 .
mview @xcite , version 1.60.1 , was used for converting the blast output into a standard multiple sequence alignement ( msa ) . in this msa , 3365 different amino - acid residues
are observed at least 20 times , that is , 6.2 per position along the sequence , on average . since 1999 , a dozen 1918 ha sequences have been determined , at least partially .
they differ , at most , by a couple of mutations @xcite .
one of them was chosen as a representative , namely , the sequence found in pdb structure 4eef ( with h3 subtype residue numbering ) @xcite .
as shown in figure [ fig : idoft ] , known ha sequences of post-1919 viruses are _ all _ less than 95% identical to the 1918 one , that is , complete ones differ from the 1918 sequence by at least 25 amino - acid substitutions .
strinkingly , after 1919 , ha sequences of human viruses are less than 91.4% identical to the 1918 ha , a @xmath090% level of sequence identity being observed in the thirties as well as more recently , though in a few instances only . on the other hand , ha sequences of avian viruses more than 93% identical to the
1918 ha have been observed each year since 2005 , suggesting that a selection pressure favoring 1918 ha - like sequences is at work in avian species , in line with the hypothesis of an avian origin for the 1918 - 1919 pandemic @xcite .
moreover , among the 41 post-1950 sequences that are more than 93% identical to the 1918 ha , 35 ( 85% ) come from duck species , further pinpointing aquatic birds as a possible reservoir @xcite .
._number of post-1919 hemagglutinin sequences with same residue as the 1918 sequence .
_ top : residues found in less than 50 human h1 sequences .
bottom : key residues involved in receptor binding @xcite .
bold : residue index of an highly conserved residue , that is , a residue found in more than 95% of the h1 sequences . @xmath1h3 subtype residue numbering ; @xmath2ha1 subunit ; @xmath3ha2 subunit .
[ cols="^,^,^,^,^,^ " , ] however , the 1918 ha sequence is a singular one .
for instance , 17 amino - acid residues in this sequence are found in less than 1% of other human h1 sequences .
the ten less frequent ones are shown on top of table [ tbl : mutations ] .
most of them are frequent in avian h1 sequences , or often found in sequences of other ha subtypes ( last column ) . as a striking exception , after 1919 , gly 188 has _ not _ been observed again in human h1 sequences .
it has also _ not _ been observed in avian ones . as a matter of fact
, it has only been observed in 47 h1 sequences , all of them from swine , a single time in 2003 , once each year between 2009 and 2012 , and several times each year since then . in human ha of other subtypes , it has been observed 11 times , in sequences from h3n2 or h5n1 viruses . since 1919 , gly 188 has _ only _ been observed 82 times , the first time in 2000 , in the sequence of an avian h9 ha .
residue 188 is located at the n - terminus of the `` 190-helix '' , which is involved in the ha receptor binding site , but it does not interact directly with the receptor @xcite .
three residues are usually observed at this position , namely , serine ( 41% ) , isoleucine ( 33% ) and threonine ( 21% ) .
interestingly , proline which , like glycine , can have a direct impact on the secondary structure @xcite , the folding @xcite or the stability @xcite of a protein , is also rarely observed .
this suggests that , taken alone , gly 188 is likely to be deleterious .
however , since the 1918 virus proved efficient , one or several compensatory mutations have to be present in its ha sequence .
* why has this mutation been overlooked ? * after 1919 , gly 188 has _ not _ been found again in human h1 sequences .
overall , it has been found in _ only _ 0.2% of all known ha sequences .
the reason why this singular mutation seems to have been overlooked is probably the following one : in the eleven 1918 - 1919 ha sequences known so far , gly 188 has been found ten times . in other words , there is an exception , namely , the ha sequence of virus a / london/1/1918 @xcite .
* how may it be involved in the deadly process ? * gly 188 is located between his 183 and asp 190 , two key residues of the receptor binding site ( figure [ fig : hook ] ) .
his 183 is highly conserved , being found in 99.9% of h1 sequences .
it is thus likely to be involved in the specific recognition of the sialic acid moiety of the ha receptor . on the other hand
, asp 190 is well conserved in h1 ha ( 88% ) but much less in all ha ( 56% ) , being frequently replaced by glu ( 32% ) .
this suggests that it is involved in more subtle aspects of the recognition .
indeed , a d190e mutation in the 1918 ha results in a preference for the @xmath4 sialic acid ( avian ) receptor @xcite .
thus , given the inherent flexibility of gly residues , introducing gly at position 188 may allow for alternative orientations of asp 190 and , as a consequence , for the recognition of other conformations of the sialic acid moiety , or to sialic acid moities linked to di - saccharides other than gal - glcnac . though a glycan microarray analysis confirmed a specific recognition of @xmath5 sialic acid receptors by 1918 ha , with no binding when glcnac is absent @xcite , a change in its repertoire of ha receptors could indeed explain why the 1918 virus proved so deadly .
gly 188 has rarely been observed in ha sequences ( table [ tbl : mutations ] ) .
since it is located within a key motif of the ha receptor binding site ( figure [ fig : hook ] ) , this mutation may have an impact on receptor recognition and specificity .
this , in turn , could explain why the 1918 ha confers enhanced pathogenicity .
it is thus important to check the effects of this mutation on receptor function .
meanwhile , it may prove important to monitor this mutation , noteworthy in swine viruses since it has been observed in swine h1 sequences each year since 2009 .
reid , ah , fanning , tg , hultin , jv , taubenberger , jk ( 1999 ) origin and evolution of the 1918 `` spanish '' influenza virus hemagglutinin gene .
_ proceedings of the national academy of sciences _ 96:16511656 .
gunasekaran , k , nagarajaram , h , ramakrishnan , c , balaram , p ( 1998 ) stereochemical punctuation marks in protein structures : glycine and proline containing helix stop signals .
_ journal of molecular biology _ 275:917932 .
krieger , f , mglich , a , kiefhaber , t ( 2005 ) effect of proline and glycine residues on dynamics and barriers of loop formation in polypeptide chains .
_ journal of the american chemical society _ 127:33463352 .
nicholson , h , tronrud , d , becktel , w , matthews , b ( 1992 ) analysis of the effectiveness of proline substitutions and glycine replacements in increasing the stability of phage t4 lysozyme .
_ biopolymers _ 32:14311441 .
zhang , w et al . (
2013 ) molecular basis of the receptor binding specificity switch of the hemagglutinins from both the 1918 and 2009 pandemic influenza a viruses by a d225 g substitution .
_ journal of virology _ 87:59495958 .
stevens , j et al .
( 2006 ) glycan microarray analysis of the hemagglutinins from modern and pandemic influenza viruses reveals different receptor specificities . _
journal of molecular biology _ 355:11431155 . | the influenza pandemic of 1918 - 1919 killed at least 50 million people .
the reasons why this pandemic was so deadly remain largely unknown @xcite .
however , it has been shown that the 1918 viral hemagglutinin allows to reproduce the hallmarks of the illness observed during the original pandemic @xcite .
thanks to the wealth of hemagglutinin sequences accumulated over the last decades , amino - acid substitutions that are found in the 1918 - 1919 sequences but rare otherwise can be identified with high confidence .
such an analysis reveals that gly188 , which is located within a key motif of the receptor binding site , is so rarely found in hemagglutinin sequences that , taken alone , it is likely to be deleterious . monitoring this singular mutation in viral sequences
may help prevent another dramatic pandemic . |
recent theoretical proposals , predicting a significant improvements in the laser gyro sensitivity with the use of a `` white - light '' cavity @xcite , have motivated a number of theoretical and experimental studies of negative dispersion in various systems @xcite .
a promising approach relies on manipulation of optical dispersion via coherent interactions of light with resonant atomic media .
processes like electromagnetically induced transparency ( eit ) , stimulated raman scattering ( srs ) , four - wave mixing ( fwm ) , etc . , are known to enable group index @xmath0 variation from subluminal ( `` slow light '' , @xmath1 ) to superluminal ( `` fast light '' , @xmath2 ) @xcite .
demonstration of a white - light laser gyroscope requires a gain medium with group index tunable around @xmath3 .
several experimental groups have pursued various approaches for the realization of such conditions @xcite .
recently our groups demonstrated that the fast light regime with amplification can be achieved in a four - level _
n _ interaction scheme , shown in fig .
[ fig : levels ] @xcite . in such a configuration the two optical fields the strong pump field
@xmath4 and weak signal field @xmath5form a regular @xmath6 system exhibiting eit and slow light @xcite .
the four - level _
n_-scheme is then completed with the introduction of the second strong pump field @xmath7 . a strong amplification of the probe field is possible in the case that the selection rules allow the four - wave mixing process @xcite . at the same time , the additional rabi splitting introduced by the second strong optical field provides an extra control mechanism for the probe field dispersion@xcite .
the possibility of probe field propagation with no optical losses or with gain , with a smooth transition between slow- and fast - light regimes by varying the strength of one of the pump fields ( @xmath7 ) was theoretically demonstrated @xcite , making it an ideal test bed for explorations of the dispersion effects in a ring cavity laser for gyroscope applications . [
cols="^ " , ]
in conclusion , we have analyzed the propagation of a weak resonant probe through a medium of four - level atoms in an _
n_-scheme with allowed four - wave mixing generation , and found it to be a promising candidate for the realization of tunable `` slow - to - fast '' light with no absorption .
this is particularly interesting for the experimental investigation of potential techniques for the enhancement of optical - gyroscope performance .
the authors thank frank a. narducci and jon davis for useful discussions , and gleb romanov for helping with the experiment .
this research was supported by the naval air warfare center sttr program , contract n68335 - 13-c-0227 .
m. s. shahriar , g. s. pati , r. tripathi , v. gopal , m. messall , and k. salit , `` ultrahigh enhancement in absolute and relative rotation sensing using fast and slow light , '' _ phys . rev .
* 75*(5 ) , 053807 ( 2007 ) .
g. s. pati , m. salit , k. salit , and m. s. shahriar , `` demonstration of displacement - measurement - sensitivity proportional to inverse group index of intra - cavity medium in a ring resonator , '' _ opt .
commun . _ * 281*(19 ) , 49314935 ( 2008 ) . c. ciminelli , c. e. campanella , f. dellolio , and m. n. armenise , `` fast light generation through velocity manipulation in two vertically - stacked ring resonators , '' _ opt
. express _ * 18 * , 29732986 ( 2010 ) . n. b. phillips , i. novikova , e. e. mikhailov , d. budker , and s. m. rochester , `` controllable steep dispersion with gain in a four - level _
n_-scheme with four - wave mixing , '' _
opt . _ * 60 * , 6472 ( 2013 ) .
i. novikova , e. e. mikhailov , l. stagg , d. budker , and s. m. rochester , `` tunable lossless slow and fast light in a four - level _ n_-system , '' _ proc .
spie : advances in slow and fast light _ * 8636 * , 86360c ( 2013 ) .
n. b. phillips , a. v. gorshkov , and i. novikova , `` slow light propagation and amplification via electromagnetically induced transparency and four - wave mixing in an optically dense atomic vapor , '' _
opt . _ * 56*(18 ) , 19161925 ( 2009 ) .
t. hong , a. v. gorshkov , d. patterson , a. s. zibrov , j. m. doyle , m. d. lukin , and m. g. prentiss , `` realization of coherent optically dense media via buffer - gas cooling , '' _ phys .
a _ * 79 * , 013806 ( 2009 ) . n. b. phillips , a. v. gorshkov , and i. novikova , `` light storage in an optically thick atomic ensemble under conditions of electromagnetically induced transparency and four - wave mixing , '' _ phys .
a _ * 83 * , 063823 ( 2011 ) . e. e. mikhailov , v. a. sautenkov , i. novikova , and g. r. welch , `` large negative and positive delay of optical pulses in coherently prepared dense rb vapor with buffer gas , '' _ phys .
* 69*(6 ) , 063808 ( 2004 ) . | we investigate a four - wave mixing process in an _
n _ interaction scheme in rb vapor placed inside a low - finesse ring cavity .
we observe strong amplification and generation of a probe signal , circulating in the cavity , in the presence of two strong optical pump fields .
we study the variations in probe field gain and dispersion as functions of experimental parameters with an eye on potential application of such a system for enhanced rotation measurements .
a density - matrix calculation is performed to model the system , and the theoretical results are compared to those of the experiment .
: eugeniy e. mikhailov , the college of william and mary , department of physic , williamsburg , p.o .
box 8795 , va 23187 , usa ; tel : + 1 757 - 221 - 3571 ; fax : + 1 757 - 221 - 3540 ; e - mail : 2 |
from zimin s explicit classification of unavoidable words , a natural question arises in the ramsey theory paradigm : for a fixed unavoidable word @xmath1 , how long can a word be that avoids @xmath1 ?
our approach to this question is to start with avoiding the zimin words , which gives upper bounds for all unavoidable words .
define @xmath58 to be the smallest integer @xmath59 such that every @xmath13-ary word of length @xmath59 encounters @xmath4 . for
the inductive hypothesis , assume the claim is true for some positive @xmath57 and set @xmath68 .
that is , every @xmath13-ary word of length @xmath69 encounters @xmath4 .
concatenate any @xmath70 strings @xmath71 of length @xmath69 with an arbitrary letter @xmath72 between @xmath73 and @xmath74 for each positive @xmath75 : @xmath76 by the pigeonhole principle , @xmath77 for some @xmath78 .
that string , being length @xmath69 , encounters @xmath4 .
therefore , we have some word @xmath79 that is an instance of @xmath4 and shows up twice , disjointly , in @xmath18 .
the extra letter @xmath80 guarantee that the two occurrences of @xmath0 are not consecutive .
this proves that an arbitrary word of length @xmath81 witnesses @xmath82 , so @xmath83 the technique used to prove theorem [ upper ] is first found in lothaire s proof of unavoidability of @xmath4 ( @xcite , 3.1.3 ) .
the technique in zimin s original proof @xcite implicitly gives that for @xmath88 , @xmath89 this is an ackermann - type function for an upper bound , which is much larger than the primitive recursive bound from theorem [ upper ] . + the following binary word of length 10482 avoids @xmath93 .
this implies that @xmath94 .
the word is presented here as an image with each row , consisting of 90 squares , read left to right .
each square , black or white , represents a bit . for example
, the longest string of black in the first row is 14 bits long .
we can not have the same bit repeated @xmath95 times consecutively , as that would be a @xmath92-instance .
a string of 14 white bits is found in the 46th row . + + the code to generate a @xmath92-avoiding word of length 10482 is messy .
the following , easy - to - validate , inefficient , brute - force , sage @xcite code was used for verification of the word above .
it took about half a day , running on an intelcorei5 - 2450 m cpu @xmath96 2.50ghz @xmath97 4 .
+ .... # recursive function to test if v is an instance of z_n def inst(v , n ) : if n==1 : if len(v)>0 : return 1 return 0 else : top = ceil(len(v)/2 ) for i in range(2^(n-1)-1,top ) : if v[:i]==v[-i : ] : if inst(v[:i],n-1 ) : return 1 return 0 | how long can a word be that avoids the unavoidable ?
word @xmath0 encounters word @xmath1 provided there is a homomorphism @xmath2 defined by mapping letters to nonempty words such that @xmath3 is a subword of @xmath0 .
otherwise , @xmath0 is said to avoid @xmath1 .
if , on any arbitrary finite alphabet , there are finitely many words that avoid @xmath1 , then we say @xmath1 is unavoidable .
zimin ( 1982 ) proved that every unavoidable word is encountered by some word @xmath4 , defined by : @xmath5 and @xmath6 . here
we explore bounds on how long words can be and still avoid the unavoidable zimin words . in 1929 ,
frank ramsey proved that , for any fixed @xmath7 , every sufficiently large set @xmath8 with its @xmath9-subsets partitioned into @xmath10 classes is guaranteed to have a subset @xmath11 such that all the @xmath9-subsets of @xmath12 are in the same class @xcite .
this was the advent of a major branch of combinatorics that became known as ramsey theory . often applied to graph theoretic structures , ramsey theory looks at how large a random structure must be to guarantee that a given substructure exists or a given property is satisfied . here
we apply this paradigm to an existence result from the combinatorics of words .
a _
@xmath13-ary word _ is a string of characters , at most @xmath13 of them distinct . over a fixed @xmath13-letter alphabet ,
the set of all finite words forms a semigroup with concatenation as the binary operation ( written multiplicatively ) and the empty word @xmath14 as the identity element .
we also have a binary subword relation @xmath15 where @xmath16 when @xmath17 for some words @xmath18 , @xmath1 , and @xmath19 . that is , @xmath1 appears contiguously in @xmath0 .
we call word @xmath0 an _ instance _ of @xmath1 provided * @xmath20 where each @xmath21 is a letter ; * @xmath22 with each @xmath23 and @xmath24 whenever @xmath25 .
equivalently , @xmath0 is a _
@xmath1-instance _ provided there exists some semigroup homomorphism @xmath2 such that @xmath26 for each @xmath27 .
@xmath28 is an instance of @xmath29 , with @xmath2 defined by @xmath30 , @xmath31 , and @xmath32 .
a word @xmath18 _ encounters _ word @xmath1 provided some subword @xmath33 is an instance of @xmath1 .
if @xmath18 fails to encounter @xmath1 , then @xmath18 _ avoids _ @xmath1 .
( 0,0 ) node@xmath14 ; ( -1,.7 ) node[above]@xmath34(0,.3)(1,.7 ) node[above]@xmath35 ; ( -1.5,1.7 ) node[above]@xmath36(-1,1.3)(-.6,1.7 ) node[above]@xmath37 ; ( .6,1.7 ) node[above]@xmath38(1,1.3)(1.5,1.7 ) node[above]@xmath39 ; ( -1.2,1.7)(-1.8,2.2 ) ; ( -1.2,2.2)(-1.8,1.7 ) ; ( 1.2,1.7)(1.8,2.2 ) ; ( 1.2,2.2)(1.8,1.7 ) ; ( -1.2,2.7 ) node[above]@xmath40(-.6,2.3)(-.4,2.7 ) node[above]@xmath41 ; ( 1.2,2.7 ) node[above]@xmath42(.6,2.3)(.4,2.7 ) node[above]@xmath43 ; ( -.1,2.7)(-.7,3.2 ) ; ( -.1,3.2)(-.7,2.7 ) ; ( .1,2.7)(.7,3.2 ) ; ( .1,3.2)(.7,2.7 ) ; ( -1.7,3.7 ) node[above]@xmath44(-1.2,3.2)(-.6,3.7 ) node[above]@xmath45 ; ( 1.7,3.7 ) node[above]@xmath46(1.2,3.2)(.6,3.7 ) node[above]@xmath47 ; ( -2,3.7)(-1.4,4.2 ) ; ( -1.4,3.7)(-2,4.2 ) ; ( 2,3.7)(1.4,4.2 ) ; ( 1.4,3.7)(2,4.2 ) ; ( -.9,3.7)(-.3,4.2 ) ; ( -.3,3.7)(-.9,4.2 ) ; ( .9,3.7)(.3,4.2 ) ; ( .3,3.7)(.9,4.2 ) ; we see in figure 1 that @xmath48 is avoided by only finitely many words over a two - letter alphabet .
however , it has been known for over a century @xcite that @xmath48 can be avoided by arbitrarily long ( even infinite ) ternary words .
a word @xmath1 is _ unavoidable _ provided for any finite alphabet , there are only finitely many words that avoid @xmath1 .
a. i. zimin proved an elegant classification of all unavoidable words @xcite .
define the _ @xmath49 zimin word _
recursively by @xmath50 and , for @xmath51 , @xmath52 . using the alphabet rather than indexed variables :
@xmath53 equivalently , @xmath4 can be defined over the natural numbers as the word of length @xmath54 such that the @xmath55 letter is the 2-adic order of @xmath27 for @xmath56 .
a word @xmath1 with @xmath57 distinct letters is unavoidable if and only if @xmath4 encounters @xmath1 . |
immersions and embeddings form open subspaces of the space of @xmath4-maps @xmath5 , @xmath6 .
smale @xcite showed that the path components of the space of immersions ( or , which is the same , the regular homotopy classes of immersions ) @xmath5 are in one to one correspondence with the elements of @xmath7 , the @xmath8 homotopy group of the stiefel manifold of @xmath9-frames in @xmath10-space .
this far - reaching result translates problems in geometry to homotopy theory . indicating the way back to geometry ,
smale suggested the following problems ( @xcite , p.329 ) : _ `` find explicit representatives of regular homotopy classes what regular homotopy classes have an embedding for representative?''_. explicit representatives of regular homotopy classes of immersions @xmath0 are given in section [ secgen ] .
an answer to the second problem gives information about how the inclusion of the space of embeddings into the space of immersions is organized . in the case
@xmath0 , the group @xmath11 , enumerating regular homotopy classes is infinite cyclic and the answer to the second problem was found by hughes and melvin @xcite .
they proved that exactly every @xmath12 regular homotopy class have an embedding for representative . in theorem [ thmimem ] below we describe the obstruction for a generic immersion @xmath0 to be regularly homotopic to an embedding in terms of geometric invariants of its self intersection . in dimensions where there are embeddings in different regular homotopy classes , it is impossible to express the regular homotopy class of a generic immersion in terms of its self intersection
this is in contrast to many other cases where this is possible : for example , in the cases @xmath13 , @xmath14 the regular homotopy class is determined by the algebraic number of self intersection points ( modulo 2 if @xmath9 is odd ) , see @xcite , and in the cases @xmath15 , @xmath16 self intersection formulas for regular homotopy are given in @xcite and @xcite .
generic immersions have simple self intersections .
for example , in the case @xmath17 a generic immersion has empty self intersection and thus , generic immersions are embeddings .
form this point of view , the analogue of classical knot theory in other dimensions is the study of path components of the space of generic immersions and we may think of generic immersions as knots with self intersections .
the space of generic immersions is dense in the corresponding space of immersions and its complement is a stratified hypersurface . using the stratification of this complement ,
vassiliev @xcite introduced the notion of finite order invariants in classical knot theory .
there are natural analogies of this notion for invariants of generic immersions in other dimensions ( see section [ 1stord ] ) .
arnold @xcite found first order invariants of generic plane curves ( @xmath18 ) .
theorem [ thmjst ] below shows that , up to first order , the space of generic immersions @xmath0 is similar to the space of generic plane curves .
the composition of an immersions @xmath3 and the inclusion @xmath19 is an immersion into 5-space .
proposition [ 4ss5 ] shows that two immersions of @xmath20 into 4-space are regularly homotopic in 5-space , after composing them with the inclusion , if and only if one of them is regularly homotopic in 4-space to the connected sum of the other one and a finite number of immersions regularly homotopic to the composition of the standard embedding and a reflection in a hyperplane in @xmath21
. theorems [ gi41 ] , [ gi42 ] , and [ gi43 ] give information about the self intersections of a generic immersions @xmath22 and its relation to the self intersection of a generic immersion @xmath23 regularly homotopic to @xmath24 .
in this section , the main theorems of the paper are formulated .
regular homotopy classes of immersions @xmath0 are known to form an infinite cyclic group @xmath25 under connected sum .
the classes that contain embeddings form a subgroup @xmath26 .
we now state theorem [ thmimem ] describing the extension @xmath27 algebraically and devote the rest of this section to identify the homomorphisms involved there in topological terms .
[ thmimem ] the following diagram of abelian groups has exact rows , commutes and the vertical arrows are isomorphisms @xmath28 theorem [ thmimem ] is proved in section [ pfimem ] . _ the homomorphism @xmath29
_ : recall that if @xmath30 is an embedding then there exists a compact orientable 4-dimensional manifold @xmath31 with @xmath32 .
we call such a manifold a _ seifert - surface _ of @xmath33 . its signature @xmath34 is divisible by 16 and is known to depend only on @xmath33 . _ for @xmath35 , define @xmath36 as @xmath37 , where @xmath38 is a seifert - surface of an embedding representing @xmath39 .
_ it is proved in @xcite that @xmath29 induces an isomorphism @xmath40 and that @xmath26 is a subgroup of index @xmath41 . _
the homomorphism @xmath42 _ : any immersion @xmath43 is determined up to regular homotopy by its smale invariant @xmath44 ( see @xcite or definition [ dfnsmin ] ) . _ for @xmath45 , define @xmath46 as @xmath47 , where @xmath48 is an immersion representing @xmath39 .
_ _ the homomorphism @xmath49 _ :
if @xmath43 is a generic immersion then its self intersection @xmath50 is a closed 1-dimensional manifold .
orientations of @xmath20 and @xmath51 induce an orientation of @xmath52 .
we can push @xmath52 off the image of @xmath33 ( see section [ seclk ] ) .
let @xmath53 denote the linking number of the perturbed @xmath52 and @xmath54 in @xmath51 .
_ for @xmath45 , define @xmath55 as @xmath56 modulo 3 , where @xmath48 is a generic immersion representing @xmath39 .
_ _ the homomorphism @xmath57 _ : the normal bundle of an immersion @xmath58 is 2-dimensional orientable and therefore trivial .
hence , @xmath59 admits a normal vector field .
this implies that @xmath59 is regularly homotopic to a ( generic ) immersion @xmath60 , composed with the inclusion @xmath61 ( see @xcite ) . resolving the self intersection of a generic immersion @xmath62
, we obtain a smooth surface @xmath63 ( lemma [ fgcg ] ) and the immersion @xmath33 induces a pin ( i.e. @xmath64 ) structure on @xmath63 ( section [ secbeta ] ) .
there is a one to one correspondence between pin structures on a surface @xmath65 and @xmath66-quadratic functions @xmath67 on its first homology @xmath68 .
pin structures on a surface are classified up to cobordism by the brown invariant @xmath69 of the corresponding function .
let @xmath70 denote the brown invariant of the quadratic function corresponding to the pin structure induced by @xmath33 on @xmath63 .
_ for @xmath45 , define @xmath71 as @xmath72 , where @xmath73 is a generic immersion representing @xmath39 .
_ the author does not know how to calculate @xmath74 for a generic immersion @xmath43 in terms of the geometry of its self intersection , without pushing it down to @xmath21 .
however , @xmath74 modulo 4 , can be calculated in that way ( see section [ secrelinv ] ) . in generic one - parameter families of immersions
there are isolated instances of non - generic immersions . in generic one - parameter families of immersions @xmath0 such instances
are immersions with one self tangency or one triple point .
the same is true for generic one - parameter families of plane curves where one can distinguish two local types of self tangencies : direct ( the tangent vectors point in the same direction ) and reverse ( the tangent vectors point in opposite directions ) .
an _ invariant of generic immersions _ is a function which is constant on the path components of the space of generic immersions .
such a function may change when we pass through instances of non - generic immersions .
arnold @xcite found three independent first order invariants of generic plane curves : the invariant @xmath75 which changes at direct self tangency instances and does not change under other moves , the invariant @xmath76 which changes under reverse self tangency moves and does not change under other moves , and the invariant strangeness @xmath2 which changes under triple point moves and does not change under other moves . for a generic immersion @xmath43 ,
define @xmath77 to be the number of self intersection components of @xmath33 and define @xmath78 , @xmath79 is a lifting of @xmath80 ( see section [ emimsp ] ) .
[ thmjst ] the invariant @xmath1 changes by @xmath81 under self tangency moves and does not change under triple point moves .
the invariant @xmath82 changes by @xmath83 under triple point moves and does not change under self tangency moves .
the invariants @xmath1 and @xmath82 are first order invariants . moreover , if @xmath84 is any first order invariant then the restriction of @xmath85 , where @xmath86 is a path component of the space of immersions , is a linear combination of @xmath87 and @xmath88 . theorem [ thmjst ] is proved in section [ pfjst ] .
although there are two local types of selftangencies in the case @xmath0 ( see proposition [ verdef51 ] ) , in contrast to the case of plane curves , @xmath1 can not be splitted into more refined first order invariants .
arnold defined the invariant @xmath2 in such a way that it is additive under connected summation of plane curves and showed that this property together with the changes of @xmath2 under local moves and its is orientation independence completely characterizes the invariant ( up to a constant ) .
the invariant @xmath82 is neither additive under connected summation nor symmetric with respect to orientation . to get the analogue of arnold s strangeness of plane curves for immersions @xmath43 we define @xmath89 .
then @xmath2 is additive under connected sum , changes exactly as @xmath82 under local moves , changes sign under composing immersions with an orientation reversing diffeomorphism of @xmath20 , and is completely characterized by these properties up to a constant ( see proposition [ strange ] ) .
the reason for stating theorem [ thmjst ] in terms of @xmath82 instead of @xmath2 ( which works equally well ) is that @xmath90 in contrast to @xmath91 can be calculated in terms of the self intersection of @xmath33 .
the self intersection of a generic immersion@xmath62 consists of 2-dimensional sheets of double points , 1-dimensional curves of triple points and isolated quadruple points ( see definition [ dfngi4 ] ) . resolving the quadruple and triple points turns the self intersection into a smooth surface @xmath63 ( lemma [ fgcg ] ) .
the following three theorems are proved in section [ secrelinv ] .
[ gi41 ] a generic immersion @xmath22 has an odd number of quadruple points if and only if its ( resolved ) self intersection surface @xmath92 has odd euler characteristic . as mentioned ( section [ emimsp ] ) , any immersion @xmath0 is regularly homotopic to a composition of an immersion @xmath3 and the inclusion @xmath61 .
[ gi42 ] let @xmath23 be a generic immersion and let @xmath93 be a generic immersion regularly homotopic to @xmath33
. then @xmath48 has an odd number of quadruple points if and only if @xmath33 has an odd number of self intersection components with connected preimage .
we define a @xmath66-valued invariant @xmath94 of generic immersions @xmath0 ( see section [ tau ] ) . if @xmath33 is a generic immersion then @xmath95 is divisible by 2 if and only if @xmath33 has an odd number of self intersection components with connected preimage .
[ gi43 ] let @xmath23 be a generic immersion and let @xmath93 be a generic immersion regularly homotopic to @xmath33 . if @xmath96 then the ( resolved ) self intersection surface @xmath92 of @xmath48 is nonorientable .
in this section , some properties of the smale invariant and the groups where it takes values are collected for later reference . via the smale invariant the set of regular homotopy classes of immersions is endowed with the structure of an abelian group .
we describe the group operations in topological terms .
most of the results presented here are known .
proofs are provided where there were hard to find references .
the stiefel manifold @xmath97 is homotopy equivalent to @xmath98 .
indeed , any orthonormal 3-frame in 4-space can be uniquely completed to a positively oriented @xmath99-frame .
let @xmath100 be the fibration with fiber @xmath101 .
consider @xmath20 as the set of unit quaternions in @xmath102 , where we identify the vectors @xmath103 in the standard base of @xmath21 with @xmath104 .
for @xmath105 let @xmath106 denote quaternionic product of @xmath107 and @xmath108 .
the map @xmath109 , @xmath110 , is a section of @xmath111 .
thus , @xmath98 is diffeomorphic to @xmath112 .
let @xmath113 be the map @xmath114 , where @xmath115 is a pure quaternion and @xmath116 is identified with the set of pure quaternions .
( that is , the span of the vectors @xmath117 . )
let @xmath118 be @xmath119 composed with the inclusion of the fiber over @xmath120 .
the following is immediate .
[ pi3v43 ] @xmath121\oplus{{\mathbb{z}}}[\rho].\ ] ] let @xmath122 be the fibration with fiber @xmath123 that maps an orthonormal @xmath124-matrix to the 3-frame consisting of its first three column vectors . similarly , let @xmath125 be the fibration with fiber @xmath98 . [ pi3v53 ] the homomorphism @xmath126 is an isomorphism .
the homomorphism @xmath127 is an epimorphism with kernel @xmath128 , where @xmath128 is the subgroup generated by @xmath129 - 2[\rho]$ ] .
the first statement follows by inspecting the homotopy sequence of the fibration . for the second , see @xcite , chapter 8 , proposition 12.11 .
using lemma [ pi3v53 ] , we make the identifications @xmath130 we define the smale invariant of an immersion @xmath131 : consider @xmath20 as the unit sphere in @xmath21 and let @xmath132 , denote the standard embedding .
fix a disk @xmath133 containing the south pole and a framing @xmath134 of @xmath135 . using regular homotopy ,
deform @xmath33 so that @xmath136 . choose a diffeomorphism @xmath137 of degree @xmath138 , where @xmath139 is the hemisphere @xmath140 in the unit sphere @xmath141 in @xmath21
let @xmath142 be the map @xmath143 of the unit sphere in @xmath21 .
define @xmath144 , @xmath145 where @xmath146 is the hemisphere @xmath147 .
[ dfnsmin ] the smale invariant @xmath148 of @xmath33 is @xmath149\in\pi_3(v_{n,3}).\ ] ] smale , @xcite showed that @xmath42 gives a bijection between the regular homotopy classes of immersions @xmath150 and the elements of @xmath151 . in computations we will not use definition [ dfnsmin ] literally .
we use a slightly different approach : consider @xmath152 where @xmath153 denotes the quaternions .
the tangent space of @xmath20 at @xmath154 is the span of the vectors @xmath155 .
we trivialize @xmath156 using the _ quaternion framing : _
@xmath157 let @xmath158 be an immersion .
then there is an induced map @xmath159 , @xmath160 .
assume that @xmath161 .
then @xmath162 .
thus , we get a map @xmath163 .
[ calcsmin4 ] let @xmath62 be an immersion then @xmath164-[\phi_4^s]\in\pi_3(so(4)),\ ] ] and @xmath165=[\sigma]\in\pi_3(so(4))$ ] .
assume that @xmath166 .
consider the fibration @xmath167 as in lemma [ pi3v53 ] .
since the normal bundle of @xmath33 is orientable 2-dimensional , it is trivial and we can lift @xmath168 to @xmath169 and @xmath170\in\pi_3(so(5))$ ] is independent of this lifting .
[ calcsmin5 ] let @xmath43 be an immersion then @xmath171-[\theta^s]\in\pi_3(so(5))=\pi_3(v_{5,3}),\ ] ] and @xmath172=[\sigma]+n\in\pi_3(so(5))$ ] .
let @xmath19 denote the inclusion .
[ smin4in5 ] if @xmath62 is an immersion then @xmath173 complete the framing of @xmath33 with a vector in the fifth direction and combine lemma [ calcsmin4 ] and lemma [ calcsmin5 ] .
let @xmath25 denote the infinite cyclic group of regular homotopy classes of immersions @xmath0 .
the smale invariant gives an isomorphism @xmath174 .
first , we consider addition in @xmath25 and other groups of regular homotopy classes of immersions @xmath150 : given two immersions @xmath175 we define an immersion @xmath176 as follows : consider @xmath177 with coordinates @xmath178 as the subset characterized by @xmath179 .
let @xmath180 and @xmath181 , respectively .
choose frames @xmath182 at @xmath183 and @xmath184 at @xmath185 that agree with the orientation of @xmath186 and such that @xmath187 ( @xmath188 ) are tangent to @xmath54 ( to @xmath189 ) at @xmath183 ( at @xmath185 ) .
we can assume , possibly after moving @xmath189 , that @xmath190 , @xmath191 and that @xmath192 .
moreover , we can deform the maps so that @xmath193 the immersion @xmath176 is now obtained by running a tube from @xmath183 to @xmath185 with axis @xmath194 .
details can be found in kervaire @xcite , section 2 , where the following is proved .
@xmath195 we call the immersion @xmath176 _ the connected sum of @xmath33 and @xmath48_. secondly , we consider inversion in @xmath25 : given an immersion @xmath43 we define the immersion @xmath196 as follows : let @xmath197 , where @xmath198 is the restriction of a reflection through a hyperplane in @xmath21 .
[ invrs ] @xmath199 let the disk @xmath200 in the definition of the smale invariant be the hemisphere @xmath140 and let @xmath201 be the map @xmath202
let @xmath203 be the map @xmath204 then @xmath205 and we can use the map @xmath206 to compute @xmath207 .
it is straightforward to check that this map is homotopic to @xmath208 .
hence , @xmath209= ( \phi_s^f\circ r)_\ast[s^3]= ( \phi_s^f)_\ast\left(-[s^3]\right)=-\omega(f).\ ] ]
in this section we present the classification of embeddings @xmath0 up to regular homotopy .
let @xmath210 be embeddings .
then @xmath176 is regularly homotopic to an embedding and @xmath211 is an embedding .
thus , the regular homotopy classes that contain embeddings from a subgroup of @xmath25 .
we denote this subgroup @xmath212 .
we get a classification of embeddings up to regular homotopy as follows : given an embedding @xmath43 , we can find a compact connected orientable manifold @xmath213 embedded into @xmath51 and such that @xmath214 .
we call such a manifold a _ seifert - surface of @xmath33_. the orientation of @xmath54 induces an orientation of @xmath213 . filling the @xmath20 on the boundary of @xmath213 with a 4-disk we get a closed connected oriented 4-manifold @xmath215
. the cohomology sequence of the pair @xmath216 shows that the inclusion induces an isomorphism @xmath217 .
the normal bundle of @xmath213 is a trivial 1-dimensional bundle .
hence , @xmath218 and therefore @xmath219 , where @xmath220 is the second stiefel - whitney class .
thus , @xmath221 is a spin manifold . by rokhlin s theorem ( see milnor and kervaire , @xcite ) , _ the signature @xmath222 of @xmath215 _ is divisible by 16 . define @xmath223 .
( this definition agrees with that given in section [ emimsp ] . )
[ sgntr ] for @xmath35 , let @xmath224 , where @xmath33 is an embedding representing @xmath39 .
then @xmath225 is an isomorphism .
see @xcite . [
24 ] the subgroup @xmath26 has index @xmath41 .
see @xcite .
in this section , generic immersions @xmath150 , @xmath226 and their self intersections are studied .
the space of immersions @xmath150 as described in the introduction , will be denoted @xmath227 .
it is an infinite dimensional manifold .
the set of non - generic immersions in @xmath227 is a stratified hypersurface @xmath228 .
we describe its strata of codimension one , for @xmath226 , and of codimension two , for @xmath166 .
[ dfngi4 ] an immersion @xmath229 is _ generic _ if it satisfies the following conditions : * for any @xmath230 , @xmath231 contains at most four points . * if @xmath232 , @xmath233 for @xmath234 if @xmath235 , then @xmath236 .
an immersion @xmath237 is _ generic _ if it satisfies the following conditions : * for any @xmath230 , @xmath231 contains at most two points . * if @xmath238 , for @xmath239 , then @xmath240 .
if @xmath241 is an immersion of manifolds then the _ self intersection _ of @xmath33 is the subset of points @xmath242 such that @xmath243 contains more than one point .
we denote it @xmath52 .
we denote its preimage @xmath244 .
that is , @xmath245 .
let @xmath43 be a generic immersion .
then @xmath52 and @xmath244 are closed 1-manifolds and @xmath246 is a double cover .
moreover , there is an induced orientation on @xmath52 .
the first statement follows from * g2*. if we order the oriented sheets coming together along @xmath52 then there is a standard way to assign an orientation to @xmath52 .
since the codimension of @xmath244 is even this orientation is independent of the ordering .
let @xmath22 be a generic immersion .
then @xmath247 and @xmath248 are 2-dimensional stratified spaces , @xmath249 where @xmath250 and @xmath251 are smooth manifolds of dimension @xmath252 , @xmath253 . the strata @xmath254 is the set of @xmath252-tuple points , @xmath255 , and @xmath256 is a @xmath252-fold covering .
immediate from * g1 * and * g2*. the next lemma shows how to resolve the self intersection of a generic immersion @xmath3 .
[ fgcg ] let @xmath22 be a generic immersion .
there exist closed surfaces @xmath257 and @xmath92 ( and closed 1-manifolds @xmath258 and @xmath259 ) , unique up to diffeomorphisms , and immersions @xmath260 and @xmath261 ( @xmath262 and @xmath263 ) such that @xmath264 commutes .
the maps @xmath265 and @xmath266 are surjective , have multiple points only along @xmath267 and @xmath268 , respectively and @xmath269 is the orientation double cover .
the maps @xmath29 and @xmath94 are surjective , have multiple points only along @xmath270 and @xmath271 , respectively and @xmath272 is a @xmath273-fold cover .
this is immediate from the local pictures : close to a @xmath252-tuple point @xmath247 is the intersection of @xmath252 3-planes in general position in 4-space .
we call @xmath92 the _ resolved self intersection surface _ of @xmath48 .
we use the notation @xmath274 and @xmath275 , @xmath276 .
the jet transversality theorem implies that the set of generic immersions is an open dense subset of @xmath227 , @xmath226 .
its complement @xmath277 will be called _ the discriminant hypersurface_. the discriminant hypersurface is stratified , @xmath278 , where each stratum @xmath279 , @xmath280 is a smooth submanifold of codimension @xmath9 of @xmath227 and @xmath281 is contained in the closure of @xmath282 for every @xmath252 .
the following two propositions follow by applying the jet transversality theorem to 1-parameter families of immersions .
[ codim14 ] the codimension one stratum @xmath283 is the set of all immersions @xmath62 such that * * g1 * and * g2 * holds except at one double point @xmath284 , @xmath285 where @xmath286 or , * * g1 * and * g2 * holds except at one @xmath252-tuple point , @xmath287 @xmath288 , @xmath289 , if @xmath290 where @xmath291 for @xmath292 but @xmath293 [ codim15 ] the codimension one stratum @xmath294 is the set of all immersions @xmath43 such that either * * g1 * holds , @xmath33 has double points and * g2 * holds , except at one double point @xmath295 , @xmath285 , where @xmath296 or , * * g2 * holds and * g1 * holds except at one triple point @xmath230 with @xmath297 , where @xmath298 , @xmath299 are three distinct points in @xmath20 and @xmath300 if ( a ) above holds for an immersion @xmath33 , we say that the exceptional double point @xmath301 is a _ self tangency point _ of @xmath33 .
recall that a _ deformation of a map @xmath59 _ is a 1-parameter family of maps @xmath302 , parameterized by @xmath303 , where @xmath86 is a neighborhood of @xmath304 .
a deformation @xmath302 of a map @xmath59 is called _ versal _ if every deformation of @xmath59 is equivalent ( up to left - right actions of diffeomorphisms ) to one induced from @xmath302 .
below , @xmath305 , @xmath226 will be an immersion in @xmath306 with its exceptional multiple point at @xmath307 , @xmath308 will denote coordinates in small 3-balls centered at the preimages of @xmath309 and @xmath310 , @xmath311 will be a versal deformation of @xmath59 .
such a deformation can be assumed to be constant in @xmath266 outside of the coordinate balls .
the proofs of the statements in this section and the next one are discussed in section [ discpf ] .
[ verdef41 ] let @xmath312 be an immersion in @xmath313 with exceptional multiple point @xmath309 and let @xmath310 be a versal deformation .
locally , at @xmath309 , up to choice of coordinates around the preimages of @xmath309 and in @xmath21 , * if @xmath309 is double point then @xmath310 is of the form @xmath314 where @xmath315 , or * if @xmath309 is a triple point then @xmath310 is of the form @xmath316 where @xmath315 , or * if @xmath309 is a quadruple point then @xmath310 is of the form @xmath317 or * if @xmath309 is a quintuple point then @xmath310 is of the form @xmath318 [ verdef51 ] let @xmath58 be an immersion in @xmath319 with exceptional multiple point @xmath309 and let @xmath310 be a versal deformation .
locally , at @xmath309 , up to choice of coordinates around the preimages of @xmath309 and in @xmath51 , * if @xmath309 is double point then @xmath310 is of the form @xmath320 where @xmath315 , or * if @xmath309 is a triple point then @xmath310 is of the form @xmath321 if @xmath322 ( @xmath323 ) in cases ( a ) in the above propositions we say that @xmath309 is an _ elliptic self tangency point _ ( a _ hyperbolic self tangency point _ ) of @xmath59 . to study first order invariants of generic immersions
( see section [ 1stord ] ) we need a description of the codimension two stratum of the discriminant hypersurface .
we restrict attention to immersions into 5-space .
[ verdef52 ] the codimension two stratum @xmath324 is the set of all immersions @xmath58 such that exactly one of the following holds * @xmath59 has two distinct self tangency points , * @xmath59 has two distinct triple points , * @xmath59 has one self tangency point and one triple point , * @xmath59 has a degenerate self tangency point at @xmath325 , in which case its versal deformation @xmath326 is of the form @xmath327 up to choice of coordinates @xmath107 , @xmath108 around @xmath269 , @xmath67 , respectively , and coordinates in @xmath51 .
the versal deformations of ( a)-(d ) are evident .
they are just products of 1-dimensional versal deformations .
let @xmath328 , @xmath329 and @xmath330 denote the codimension one parts of @xmath331 consisting of all immersions with one elliptic self tangency point , one hyperbolic self tangency point , and one triple point , respectively .
figure [ figcod2 ] , which is a consequence of proposition [ verdef52 ] , shows the possible intersections of the discriminant hypersurface @xmath331 and a small 2-disk in @xmath332 which meets @xmath333 and is transversal to @xmath331 .
the proofs of the propositions in section [ vdf1 ] and section [ vdf2 ] are all similar : first we need to find coordinates close to the preimages of the exceptional point such that the map is given by the expression stated .
then we show that the deformation given is infinitesimally versal .
a standard theorem in singularity theory then implies that the deformation is versal . as an example
we write out the proof of proposition [ verdef41 ] ( a ) : since @xmath59 has a tangency at @xmath309 the tangent planes of the two sheets @xmath134 and @xmath329 meeting there must agree .
we may assume that this tangent plane is the plane of the first three coordinates . by the implicit function theorem we can choose
coordinates so that the map of the first sheet is @xmath334 we may now look upon the second sheet as the graph of a function @xmath335 .
the requirement that @xmath329 is tangent to @xmath134 at @xmath309 implies that @xmath336 and @xmath337 , @xmath191 .
moreover , the second partials must be nondegenerate since the immersion is in the codimension 1 part of the discriminant hypersurface . changing coordinates in @xmath21 , by adding a function of the first three coordinates that vanishes to order 3 to the fourth
, we may assume that @xmath338 is identical to the second order terms of its taylor polynomial .
choosing the coordinates in @xmath329 appropriately then gives @xmath59 the desired form .
we must check infinitesimal versality .
if @xmath339 are coordinates on @xmath21 then this amounts to showing that any smooth variations @xmath340 can be written as @xmath341 where @xmath9 is a vector field on @xmath21 and @xmath342 is a constant .
writing these equations out we get the system @xmath343 let @xmath344 .
then the fourth equation holds .
let @xmath345 .
let @xmath346 and choose @xmath342 so that @xmath336 .
if @xmath347 let @xmath348 and if @xmath349 let @xmath350 then the last equation holds .
choosing first @xmath351 , @xmath191 so that the three remaining @xmath57-equations hold and then choosing @xmath352 , @xmath191 so that the remaining @xmath353-equations hold , infinitesimal versality is proved .
in this section , we study how the self intersections of generic immersions and their preimages transform as the immersions cross the discriminant hypersurface .
these transformations give rise to invariants of immersions : a function on @xmath227 , which is constant on path components will be called an _ invariant of regular homotopy_. a function on @xmath354 which is constant on path components of @xmath354 is an _ invariant of generic immersions_. in our study of invariants of generic immersion we are interested in how they change when we pass @xmath228 .
this is described in terms of jumps : let @xmath355 be a path in @xmath227 , @xmath226 intersecting @xmath306 transversally at @xmath59 .
let @xmath356 be small .
let @xmath84 be an invariant of generic immersions .
then @xmath357 is a locally constant function on @xmath306 , defined up to sign .
we call it _ the jump of _ @xmath84 . in section [ 1stord ]
, we get rid of the sign ambiguity in the definition of @xmath358 .
we prove theorem [ thmjst ] and end the section with an axiomatic characterization of strangeness @xmath2 for immersions @xmath0 . throughout this section ,
let @xmath310 be a path in @xmath227 intersecting @xmath359 transversally at @xmath59 and let @xmath360 be small enough so that @xmath310 is generic for @xmath361 .
let @xmath166 and let @xmath59 have a self tangency point . by proposition [ verdef51 ] , the self intersection of @xmath362 is obtained from the self intersection of @xmath363 by a single morse modification .
define @xmath364 for generic immersions @xmath33 .
[ 1stj ] @xmath1 is an invariant of generic immersions @xmath0 .
it jumps by @xmath81 when crossing the self tangency part of @xmath319 and remains constant when crossing the triple point part .
immediate from proposition [ verdef51 ] .
let @xmath161 and let @xmath59 have an exceptional quadruple point
. by proposition [ verdef41 ] the number of quadruple points of @xmath362 differs from the number of quadruple points of @xmath363 by @xmath365 .
define @xmath366 for generic immersions @xmath33 .
[ mor4 ] @xmath367 is an invariant of generic immersions @xmath3 .
it jumps by @xmath365 when crossing the part of @xmath313 that consists of immersions with an exceptional quadruple point and does not jump when crossing any other part of @xmath313 .
furthermore , @xmath368 is an invariant of regular homotopy .
the first part is immediate from proposition [ verdef41 ] .
let @xmath33 and @xmath48 be regularly homotopic then @xmath33 and @xmath48 can be joined by a path in @xmath369 which intersects @xmath313 transversally .
since @xmath367 jumps by @xmath365 or @xmath309 on such intersections , the lemma follows .
let @xmath161 and let @xmath59 have an exceptional triple point . by proposition [ verdef41 ] , @xmath370 is obtained from @xmath371 by a single morse modification ( for notation , see section [ genim ] ) .
define @xmath372 for generic immersions @xmath33 .
@xmath330 is an invariant of generic immersions @xmath3 .
it jumps by @xmath83 when crossing the part of @xmath313 that consists of immersions with an exceptional triple point and does not jump when crossing any other part of @xmath313 .
immediate from proposition [ verdef41 ] .
let @xmath161 and let @xmath59 have an exceptional double point ( a self tangency point ) . by proposition [ verdef41 ] , @xmath373 is obtained from @xmath374 by a single morse modification ( for notation , see section [ genim ] ) .
define @xmath375 for generic immersions @xmath33 , where @xmath376 denotes the euler characteristic .
@xmath377 is an invariant of generic immersions @xmath3 .
it jumps by @xmath365 when crossing the part of @xmath313 that consists of immersions with an exceptional double point and does not jump when crossing any other part of @xmath313 .
furthermore , @xmath378 is an invariant of regular homotopy .
similar to the proof of proposition [ mor4 ] .
we construct an invariant of generic immersions @xmath0 which jumps under triple point moves and stays constant under self tangency moves : let @xmath43 be a generic immersion with self intersection @xmath52 . consider the preimage @xmath379 . choose a normal vector field @xmath301 along @xmath244 satisfying the following condition :
if @xmath380 denotes the result of pushing @xmath244 slightly along @xmath301 we require that @xmath381=0\in h_1(s^3-{{\widetilde}m}_f ) .
\tag{$\dagger$}\ ] ] any two vector fields satisfying this condition are homotopic . for existence , note that we can take @xmath301 as the normal vector field of the boundary in a seifert - surface of the link @xmath382 .
we define a vector field @xmath84 along @xmath52 : for @xmath383 , let @xmath384 let @xmath385 denote the result of pushing @xmath52 slightly along @xmath84 .
then @xmath386 .
by alexander duality @xmath387 and @xmath388 since there is a triangulation of @xmath20 giving a triangulation of @xmath54 after identifications in the @xmath309 and @xmath389 skeletons only .
an orientation of @xmath20 gives a canonical generator of @xmath390 : the boundary of a small 2-disk intersecting @xmath54 transversally in one point and oriented in such a way that the intersection number is positive .
note that this intersection number is independent of the ordering of the 2-disk and @xmath54 .
recall that there is an induced orientation on @xmath52 ( section [ genim ] ) and hence on @xmath385 .
define @xmath391\in h_1({{\mathbb{r}}}^5-f(s^3))={{\mathbb{z}}}\quad\text { and } \quad \lambda(f)={\operatorname{lk}}(f){\operatorname{mod}}{3}\in{{\mathbb{z}}}_3.\ ] ] note that @xmath53 is well defined since an homotopy between two vector fields @xmath301 and @xmath392 satisfying ( @xmath393 ) induces a homotopy between the shifted self intersections in @xmath394 .
[ lk ] @xmath395 is an invariant of generic immersions .
it jumps by @xmath396 when crossing the triple point part of @xmath319 and remains constant when crossing the self tangency part .
furthermore , @xmath49 is an invariant of regular homotopy .
the second part follows from the first exactly as in proposition [ mor4 ] .
we prove the first part : suppose that @xmath310 is a path in @xmath332 intersecting @xmath319 transversally at @xmath59 and let @xmath356 be small .
we restrict attention to a small neighborhood of the exceptional multiple point of @xmath59 since all the immersions @xmath310 , @xmath397 can be assumed to agree outside of this neighborhood .
assume that @xmath59 has an elliptic self tangency point @xmath398 .
using proposition [ verdef51 ] we can write @xmath399 where @xmath107 are coordinates around @xmath269 and @xmath108 are coordinates around @xmath67 . the preimages of the newborn self intersection circle @xmath342 of @xmath362 are @xmath400 and @xmath401 .
we may assume that the field @xmath301 is given by @xmath402 and @xmath403 .
now shift @xmath342 a small distance @xmath404 along @xmath405 to get @xmath406 .
then @xmath406 is the boundary of the 2-disk @xmath407 which does not intersect @xmath408 .
hence , @xmath409 .
assume that @xmath59 has a hyperbolic self tangency point .
as above we can write @xmath410 the preimages of the self intersection are @xmath411 and @xmath412 . we can choose the field @xmath301 so that @xmath402 and @xmath403 close to @xmath269 and @xmath67 , respectively .
we note that with vector fields as above and equal outside neighborhoods of @xmath269 and @xmath67 the condition ( @xmath393 ) holds for @xmath363 if and only if it holds for @xmath362 .
the rest of the argument is similar to the elliptic case .
assume that @xmath59 has a triple point . according to proposition [ verdef51 ] we can find coordinates @xmath107 , @xmath108 and @xmath413 centered at @xmath414 , @xmath415 and @xmath416 , respectively such that close to the triple point @xmath417 we have @xmath321 denote the neighborhoods of @xmath414 , @xmath415 and @xmath416 by @xmath134 , @xmath329 and @xmath418 , respectively . if we orient these by declaring the frames @xmath419 to have the positive orientation then the oriented self intersections of @xmath59 are the lines @xmath420 the inverse image of the self intersection of @xmath310 , @xmath421 is shown in figure [ figinvtrip ] .
we calculate @xmath422 in terms of @xmath423 and have to take two things into consideration : first the motion of @xmath329 itself .
second , the changes in the field @xmath301 that this motion causes . assume that the shifting distance is very small in comparison to @xmath424 .
let @xmath425 denote the intersection @xmath426 shifted along @xmath427 .
let @xmath428 denote a part of a disc in @xmath51 bounded by @xmath425 .
we may assume that @xmath428 is a shift along an extension of @xmath84 of a disk in @xmath429 .
first , the intersection @xmath430 does not depend on the field @xmath301 since the shift is assumed to be very small in comparison to @xmath266 .
we calculate the change in @xmath430 from the preimage in @xmath134 . from figure [ figbiglk ]
it follows that the algebraic number of intersection points in @xmath431 differs from that corresponding to @xmath432 by @xmath138 .
we now take the field @xmath301 into consideration . in figure [ figseif ]
we see a seifert - surface in @xmath20 of @xmath244 close to a crossing point . in figure [ figsmalllk ]
we see @xmath301 chosen as the inward normal in a seifert - surface of @xmath433 . as we move through the triple point the direction of rotation of the vector field @xmath301
is changed .
this change in rotation gives rise to a new positive intersection point in @xmath434 for @xmath435 .
as seen in figure [ figtwlk ] .
similarly , we get a new positive intersection point in @xmath436 .
thus , in total the algebraic number of intersection points in @xmath437 increases by @xmath389 when @xmath266 is changed from @xmath438 to @xmath424 . similarly , the intersection numbers corresponding to @xmath439 and @xmath440 increase by @xmath389 when @xmath266 is changed from @xmath438 to @xmath424 .
it follows that @xmath441 .
this proves the lemma .
[ homlambda ] for @xmath39 in @xmath25 , let @xmath442 , where @xmath33 is a generic immersion representing @xmath39 .
then @xmath443 is a homomorphism . as a function , @xmath49 is well defined by lemma [ lk ] .
clearly , it is additive under connected sum and thus a homomorphism .
[ 3signs ] in knot theory one assigns a sign to crossings as in figure [ figinvtrip ] by comparing the orientation given by @xmath444 to that of the ambient space , where @xmath445 is the tangent vector to the first branch , @xmath446 the tangent vector of the second , and @xmath447 a vector from the second to the first branch .
we note that all crossings appearing in figure [ figinvtrip ] are positive .
changing the orientation of one of the sheets changes the sign of all crossings . hence , _ all three crossings appearing close to a triple point always have the same sign .
_ in this section we summarize the properties of finite order invariants of generic immersions that are needed to prove theorem [ thmjst ] .
recall that we have defined the jump @xmath358 of an invariant @xmath84 of generic immersions .
it was defined only up to sign .
to get rid of this sign we need a coorientation of @xmath306 : if @xmath448 then there is a neighborhood @xmath86 of @xmath59 in @xmath227 which is cut into two parts by @xmath306 .
we make a choice of a positive part and a negative part of @xmath86 .
a coherent choice like that for all @xmath448 is a _ coorientation of @xmath228_. a coorientation enables us to make @xmath358 well defined : in the definition @xmath449 we require that @xmath362 is on the positive and @xmath363 on the negative side of @xmath306 at @xmath59 .
as mentioned , @xmath358 is then locally constant on @xmath306 and we may consider @xmath358 as an element of @xmath450 .
let @xmath310 be a path in @xmath332 intersecting @xmath319 transversally at @xmath59 .
we coorient @xmath331 as follows : if @xmath451 then @xmath362 is on the positive side of @xmath452 at @xmath59 if @xmath453 has more components than @xmath454 .
if @xmath455 then @xmath362 is on the positive side of @xmath330 if @xmath456 .
it is easy to check that this coorientation is _ continuous _ ( see @xcite , section 7.6 ) .
this enables us to iterate the above construction and define inductively @xmath457 here @xmath458 is an immersion with @xmath459 distinct degenerate points ( self tangencies or triple points ) and @xmath460 is a path in @xmath461 , where @xmath462 is the space of immersions with @xmath252 distinct degenerate points , intersecting @xmath463 transversally . then @xmath464 is an element in @xmath465 ( see @xcite , remark 9.1.2 ) .
we say that _ an invariant @xmath84 is of order @xmath9 _ if @xmath466 .
finally , we remark that the space of invariants of generic immersions , i.e. @xmath467 , splits as a direct sum over the path components of @xmath227 .
that is @xmath468 , where @xmath86 runs over the path components of @xmath227 . by lemmas [ 1stj ] and [ lk ] , @xmath1 and @xmath82
are invariants which changes as claimed
. clearly , @xmath82 and @xmath1 are independent .
since @xmath469 ( @xmath470 ) is @xmath389 ( @xmath309 ) on all self tangency parts of @xmath319 and is @xmath309 ( @xmath389 ) on all triple point parts of @xmath319 , it follows from proposition [ verdef52 ] ( see figure [ figcod2 ] ) that @xmath471 and @xmath472 .
thus , they both have order one .
let @xmath86 be a path component of @xmath332 and @xmath84 an invariant , the jump of which is constant on @xmath473 and @xmath474 ( for notation see section [ vdf1 ] ) .
then the jump of @xmath85 is a linear combination of the jumps of @xmath87 and @xmath88 and it follows that @xmath85 is ( up to a constant term ) a linear combination of @xmath87 and @xmath88 .
thus , the theorem follows once we show that any first order invariant has this property .
let @xmath33 and @xmath48 be regularly homotopic immersions with one self tangency point of the same kind each or with one triple point each . using a diffeotopy of @xmath51 we can move the exceptional point of @xmath48 to that of @xmath33 and using a diffeotopy of @xmath20 we can make the maps agree in small disks around the preimages of this point .
choose unknotted paths in @xmath20 connecting these disks .
since there are no 1-knots in @xmath51 , we can after diffeotopy of @xmath51 assume that the maps agree also in a neighborhood of these arcs .
after this is done we have two immersions that agree on a 3-disk in @xmath20 and are regularly homotopic .
we must show that they are regularly homotopic keeping this 3-disk fixed .
however , by the smale - hirsch h - principle the obstruction to this regular homotopy is the difference of the smale invariants which is zero . choosing a generic regular homotopy we see that we can join @xmath33 to @xmath48 with a path in @xmath475 intersecting @xmath333 transversally only at immersions with two distinct exceptional points .
let @xmath84 be a first order invariant
. then @xmath476 .
this and the fact that @xmath358 is locally constant on @xmath319 implies that @xmath358 remains constant along a path as above . hence , if @xmath33 and @xmath48 both belong to either one of @xmath477 , @xmath478 , or @xmath473 then @xmath479 .
it remains to show that @xmath480 equals @xmath481 : consider the last part of figure [ figcod2 ] in section [ vdf2 ] .
let @xmath33 be an immersion in the @xmath328-branch and @xmath48 an immersion in the @xmath329-branch .
pick a small loop @xmath482 , @xmath483 , @xmath484 around the cusp such that it intersects the @xmath328-branch in @xmath33 and the @xmath329-branch in @xmath48 . then @xmath485 . hence , @xmath486 the theorem follows .
arnold s invariant @xmath2 for plane curves ( see @xcite ) can be characterized axiomatically as an invariant which jumps by @xmath389 on triple points , does not jump under self tangencies , and is additive under connected summation .
the invariant @xmath487 jumps by one on triple points but is not additive under connected sum .
the reason for this is that the invariant @xmath488 of order zero is not additive under connected sum .
the invariant @xmath395 is additive under connected sum .
it jumps by @xmath396 on @xmath330 and not at all on @xmath489 .
thus , @xmath395 and @xmath1 generates the space of @xmath490-valued invariants .
we want an invariant which is additive under connected sum and together with @xmath1 generates the space of @xmath491-valued invariants . to accomplish this
we note that , either @xmath492 or @xmath493 is divisible by @xmath273 , where @xmath42 is the smale invariant , which we may consider integer valued after the choice of generator of @xmath494 made in lemma [ pi3v53 ] .
in lemma [ lkgen ] we calculate @xmath53 for an immersion with @xmath495 and the result is @xmath496 .
since @xmath497 is an invariant of regular homotopy it follows that @xmath498 is an integer valued invariant which jumps by @xmath81 on triple points , does not jump on self tangencies , and is additive under connected sum ( since both @xmath395 and @xmath42 are ) .
we therefore define @xmath499 to complete the analogy with arnold s @xmath2 , we need to define the coorientation of the triple point stratum @xmath330 of @xmath319 in terms of the local picture near a triple point . recall that all crossings in the preimage close to a triple point have the same sign ( remark [ 3signs ] ) . by lemma [ lk ]
our coorientation of @xmath330 agrees with the following one defined in local terms : _ we say that we pass @xmath330 in the positive direction if the crossing signs in the preimage close to the triple point changes from positive to negative . _
[ strange ] there is a unique invariant @xmath2 such that it jumps by @xmath138 ( @xmath500 ) under positive ( negative ) triple point moves , it does not jump under self tangency moves , it is additive under connected sum , its value on an immersion changes sign if the immersion is composed with an orientation reversing diffeomorphism of @xmath20 , and takes the values @xmath389 on @xmath501 ( see proposition [ theg ] ) . a generic immersion @xmath48 with @xmath502 is regularly homotopic to a connected sum of @xmath503 copies of @xmath501 or @xmath504 , where @xmath505 is an orientation reversing diffeomorphism .
a generic immersion @xmath48 with @xmath506 is regularly homotopic to @xmath507 .
we know the values of our invariants on connected sums of @xmath508 .
thus , we know the value on any other generic immersion by adding the jumps along a path connecting it to such a connected sum .
this proves uniqueness . for existence
, we need to show that @xmath2 changes sign if the orientation is reversed . for @xmath42 we already know this ( lemma [ invrs ] ) .
to see that the same is true for @xmath395 note that the induced orientation of @xmath52 does not change if we reverse the orientation on @xmath20 .
the orientation of @xmath54 does change and hence @xmath509\in h_1({{\mathbb{r}}}^5-f(s^3))$ ] does change sign .
in this section we show that there is an induced pin structure on the self intersection surface of a generic immersion @xmath3 .
the brown invariant of this pin structure is unchanged under regular homotopy .
actually an even stronger result holds , if two generic immersions into @xmath21 are regularly homotopic in @xmath51 after composing them with the inclusion @xmath61 then the corresponding brown invariants are equal . using the geometry of self intersections of generic immersions @xmath0
, we define a @xmath66-valued invariant of regular homotopy .
let @xmath22 be an immersion .
its smale invariant @xmath47 is an element of @xmath511\oplus{{\mathbb{z}}}[\rho]$ ] ( see lemma [ pi3v43 ] ) . the normal bundle of @xmath48 is 1-dimensional and orientable .
hence , it is trivial and @xmath48 admits a normal field @xmath512 , such that the frame @xmath513 ( see section [ calc ] ) gives the positive orientation of @xmath21 . if @xmath514+n[\rho]$ ] then , from lemma [ pi3v43 ] , it follows that @xmath515 , the normal degree of @xmath48 .
let @xmath516 be the standard embedding and @xmath517 , where @xmath505 is a reflection in a hyperplane as in lemma [ invrs ] .
@xmath518+[\rho]\in\pi_3(v_{4,3})=\pi_3(so(4)).\ ] ] let @xmath19 be the inclusion
. then @xmath519 and @xmath520 are regularly homotopic in @xmath51 . hence , by lemma [ smin4in5 ] , @xmath521 .
the normal degree of @xmath522 equals the normal degree of @xmath265 with opposite sign .
thus , @xmath523+n[\rho]$ ] .
but then , @xmath524 implies that @xmath525 .
let @xmath526 be immersions .
[ 4ss5 ] if @xmath24 and @xmath527 are regularly homotopic in @xmath51 then either @xmath48 and @xmath528 or @xmath529 and @xmath530 are regularly homotopic in @xmath21 . if @xmath24 and @xmath527 are regularly homotopic in @xmath51 then @xmath531 .
now , @xmath532 is a generator of @xmath128 and the smale invariant is additive under connected sum .
the invariants we are about to define originates from functionals on curves in self intersections .
before we can construct these functionals we need some preliminaries : let @xmath533 denote the identity matrix .
define @xmath534 and @xmath535 by @xmath536 the group @xmath537 acts on @xmath538 from the right .
denote the corresponding quotient space @xmath539 and let @xmath269 be the projection .
[ pi_1(so / a ) ] for @xmath161 and @xmath166 , @xmath540 see @xcite , proposition 5.2 .
let @xmath541 denote the nontrivial element .
the functional on curves in a self intersection surface of a generic immersion into 4-space is constructed as follows ( for notation , see proposition [ fgcg ] ) : let @xmath542 be a generic immersion .
let @xmath342 be a closed curve in @xmath92 meeting @xmath543 transversally and let @xmath544 .
note that @xmath545 is either a union of two disjoint circles or one circle . choose a parameterization @xmath546 of @xmath342 and a normal vector @xmath547 of @xmath342 in @xmath548 .
then @xmath505 lifts to two parameterizations @xmath549 and @xmath550 .
if @xmath545 is connected then the path product @xmath551 is a parameterization of @xmath545 .
choose normal fields @xmath552 and @xmath553 of @xmath554 along @xmath555 and @xmath556 , respectively .
let @xmath557 and @xmath558 be such that @xmath559 .
assume that @xmath560 is chosen so that @xmath561 is a positively oriented frame , @xmath562 .
then @xmath563 is a framing or twist framing @xmath134 of @xmath564 along @xmath565 .
let @xmath566 denote the loop ( or loops ) in @xmath567 ( the principal @xmath101-bundle of the tangent bundle of @xmath20 ) represented by @xmath568 with the framing @xmath569 along @xmath570 , @xmath562 and let @xmath571 $ ] denote the corresponding element in @xmath572 .
let @xmath573\in\pi_1(so(4)/{{\mathbf a}}(4))$ ] denote the homotopy class of the loop induced by @xmath565 with ( twist ) framing @xmath134 as above ( which we may have to orthonormalize ) . then define [ om_g ] @xmath574\rangle\cdot p_\ast(u)+[c , r , x]+p_\ast(u)\in \pi_1(so(4)/{{\mathbf a}}(4)),\ ] ] where @xmath575 denotes the unique spin structure on @xmath20 .
this definition is independent of all choices ( see @xcite , lemma 7.6 ) .
we extend the functional @xmath576 to collections @xmath577 of oriented closed curves in @xmath92 transversal to @xmath578 : @xmath579 let @xmath82 be a collection of oriented closed curves in @xmath92 with @xmath580 transverse intersections and @xmath581 be the collection obtained from smoothing each intersection , respecting orientation . then @xmath582 and if @xmath583 and @xmath82 are isotopic collections then @xmath584 see @xcite , lemma 7.15 .
the functional on self intersection circles of generic immersions into @xmath585-space is constructed as follows : let @xmath43 be a generic immersion .
let @xmath586 be a component and @xmath587 .
note that @xmath545 is either a union of two disjoint circles or one circle .
choose a parameterization @xmath505 of @xmath342 , agreeing with the orientation induced on @xmath342 if @xmath9 is odd .
then @xmath505 lifts to two parameterizations @xmath549 and @xmath550 .
if @xmath545 is connected then the path product @xmath551 is a parameterization of @xmath545 .
choose orthonormal framings @xmath588 and @xmath589 of @xmath590 ( the normal bundle of @xmath545 in @xmath20 ) along @xmath549 and @xmath550 respectively .
then @xmath591 is a framing or twist framing @xmath134 of @xmath592 along @xmath342 .
let @xmath566 denote the loop ( or loops ) in @xmath593 ( the principal @xmath594-bundle of the tangent bundle of @xmath595 ) represented by @xmath545 with the framing @xmath596 and @xmath571 $ ] the corresponding element in @xmath597 .
let @xmath573\in\pi_1(so(5)/{{\mathbf a}}(5))$ ] denote the homotopy class of the loop induced by @xmath342 with the ( twist ) framing @xmath134 ( which we may have to orthonormalize ) . then define @xmath598\rangle\cdot p_\ast(u)+[c , r ,
x]+p_\ast(u)\in \pi_1(so(5)/{{\mathbf a}}(5)),\ ] ] where @xmath599 denotes the unique spin structure on @xmath595 .
this definition is independent of all choices ( see @xcite , lemma 6.5 ) .
let @xmath38 be a vector space over @xmath600 with a nonsingular symmetric bilinear form @xmath601 .
a @xmath66-quadratic function on @xmath38 is a function @xmath602 such that @xmath603 , for all @xmath604 .
there are four indecomposable @xmath66-quadratic spaces : @xmath605 a @xmath66-quadratic space @xmath38 is called _ split _ if it contains a subspace @xmath606 such that @xmath607 , @xmath608 and @xmath609 .
two @xmath66-quadratic spaces @xmath38 and @xmath221 belong to the same _ witt class _ if there exists spilt spaces @xmath610 and @xmath611 such that @xmath612 .
the witt classes forms the _ witt group _ which is generated by @xmath613 $ ] ( the witt class of @xmath614 ) with relations @xmath615=0 $ ] , @xmath616=[{{\mathcal t}}_4]$ ] , and @xmath613+[{{\mathcal p}}_-]=0 $ ] . given a @xmath66-quadratic space @xmath38 with quadratic function @xmath67 we define @xmath617 then @xmath618 since @xmath619 is an @xmath620 root of unity , @xmath580 modulo 8 is well defined .
brown s invariant_. it is denoted @xmath621 and gives an isomorphism between the witt group and @xmath622 .
we shall sometimes write @xmath623 , dropping @xmath38 .
more details about @xmath66-quadratic spaces can be found in @xcite .
let @xmath65 be a surface . by a pin structure on @xmath65 we shall mean a @xmath64-structure on the tangent bundle @xmath624 of @xmath65 .
there is a 1 - 1 correspondence between pin structures on @xmath65 and @xmath66-quadratic functions on @xmath68 , see @xcite , theorem 3.2 .
let @xmath22 be a generic immersion .
choose , once and for all an isomorphism @xmath626 ( this choice is discussed in section [ secrelinv ] . ) by equations ( @xmath393 ) and ( @xmath627 ) in section [ secfun ] and lemma 3.4 in @xcite , @xmath628 induces a @xmath66-quadratic function @xmath629 ( and hence a pin structure on @xmath548 ) .
we define @xmath630 the brown invariant of the @xmath66-quadratic function @xmath631 .
[ breg4 ] let @xmath632 be regularly homotopic generic immersions .
then @xmath633 connect @xmath634 to @xmath635 by a path @xmath636 in @xmath369 which is transversal to @xmath637 .
this means that @xmath636 intersects @xmath637 only transversally at finitely many points in @xmath313 .
clearly , @xmath57 remains constant as long as we do not cross @xmath637 and we must show that @xmath57 remains unchanged when we cross @xmath638 : passing @xmath313 at @xmath639 , where @xmath639 has a @xmath252-tuple point @xmath640 does not change @xmath641 .
it changes the preimage in @xmath92 of @xmath642 . since
the curves representing a basis of @xmath643 , used to calculate @xmath57 can be chosen so that they do not meet the discs where these changes occur , @xmath57 does not change at such crossings .
passing @xmath313 at @xmath639 , where @xmath639 has a degenerate double point does change @xmath641 by a morse modification .
if it has index @xmath644 or @xmath309 then @xmath641 is changed by addition or subtraction of an @xmath645-component .
this does not affect @xmath57 .
if it has index @xmath389 then @xmath641 is changed by addition or subtraction of a handle . in this handle
there is a newborn circle @xmath342 , @xmath646 and @xmath57 remains unchanged , see @xcite , lemmas 5.2.1 - 8 .
let @xmath43 be an immersion .
the normal bundle of @xmath33 is trivial ( section [ calc ] ) which implies ( @xcite , theorem 6.4 ) that @xmath33 is regularly homotopic to an immersion @xmath60 composed with the inclusion @xmath61 .
moreover , we may assume that @xmath647 is generic .
define @xmath648 [ hombeta ] for @xmath45 , let @xmath649 , where @xmath33 is an immersion representing @xmath39 .
then @xmath650 is a homomorphism .
we show first that @xmath57 is well defined : if @xmath647 and @xmath635 are immersions into @xmath21 which are regularly homotopic to @xmath33 in @xmath51 then @xmath647 is regularly homotopic to @xmath651 ( proposition [ 4ss5 ] ) .
we can perform the connected sum of two generic immersions into @xmath21 in such a way that the self intersection of the immersion obtained is the union of the self intersections of the summands and at most two new @xmath645 self intersection components . hence , performing connected sum with the embedding @xmath652 does not change @xmath57 .
thus , @xmath653 is well defined by lemma [ breg4 ] . by the argument above
, @xmath57 is additive under connected summation .
hence , it is a homomorphism .
let @xmath43 be a generic immersion .
choose , once and for all , an isomorphism @xmath655 ( this choice is discussed in section [ secrelinv ] . ) a component @xmath342 of the self intersection @xmath52 is _ right twist framed _ if @xmath656 .
it is _ left twist framed _ if @xmath657 .
if @xmath545 is disconnected then there is an induced spin structure on @xmath342 , see @xcite , proposition 2.10 .
this spin structure on @xmath342 is trivial ( spin - cobordant to zero ) if @xmath658 and nontrivial if @xmath659 ( see @xcite , proof of theorem 7.30 ) .
we define @xmath660 , @xmath661 and @xmath662 as the number of self intersection components of a generic immersion @xmath33 that are right twist framed , left twist framed and has the nontrivial spin structure , respectively .
define @xmath663 [ homtau ] for @xmath45 , let @xmath664 , where @xmath33 is a generic immersion representing @xmath39 .
then @xmath665 is a homomorphism . that @xmath94 is well defined follows from @xcite , propositions 5.1.1 - 4 .
clearly , @xmath94 is additive under connected sum .
hence , it is a homomorphism .
in this section , we construct immersions @xmath0 with arbitrary smale invariant by rotating a 2-disk with a kink i 5-space and use an immersion @xmath666 to show that the homomorphism @xmath57 ( see proposition [ hombeta ] ) is nontrivial . as in whitney @xcite we consider the map @xmath667 given by the equations @xmath668 where @xmath669 .
this is an immersion with one transversal double point : @xmath670 moreover , if @xmath671 or @xmath672 is large then @xmath48 is close to the standard embedding @xmath673 .
thus , we can change @xmath48 slightly so that it agrees with the standard embedding outside some large disk .
the whitney kink enjoys the following symmetry property : if @xmath674 and @xmath675 are the linear maps @xmath676 then @xmath677 .
the differential of @xmath48 is @xmath678 let @xmath679 and @xmath680 be unit vector fields on @xmath681 in the @xmath107 and @xmath108 directions , respectively .
applying @xmath682 to these we get a map of @xmath681 into @xmath683 , which outside some large disk is constantly equal to @xmath684 , where @xmath685 is the unit vector in the @xmath686-direction in @xmath21 . clearly , this map is homotopic in @xmath683 to a map of the form @xmath687 , where @xmath688 is orthonormal to @xmath689
. let @xmath377 be a large disk in @xmath681 .
a straightforward calculation shows that the degree of @xmath690 is one .
let @xmath691 be a 2-dimensional hemisphere in @xmath21 which is flat close to its north pole . to make this precise ,
fix a small @xmath692 and let @xmath691 be a 2-disk such that @xmath693 let @xmath694 , @xmath695 , as above be coordinates on @xmath691 .
then we can define an immersion @xmath696 with the following properties : the map @xmath583 equals the inclusion for @xmath697 . on the remaining part @xmath698 of @xmath691 the map @xmath583
is a suitably scaled whitney kink and @xmath699 .
let @xmath700 , @xmath701 and @xmath702 be polar coordinates on the disk @xmath698 , then the map @xmath703 induced by the differential of @xmath583 is homotopic to @xmath704 since the first of these equations defines a map @xmath705 of degree one .
let @xmath706 be coordinates on @xmath51 .
let @xmath685 be the unit vector in the @xmath707-direction .
let @xmath708 and let @xmath709 be the subset @xmath710 let @xmath711 be the disk @xmath712 as @xmath713 varies from @xmath309 to @xmath714 , @xmath715 sweeps the standard @xmath20 in @xmath51 .
we note that if @xmath716 , @xmath717 then @xmath718 and @xmath719 .
a straightforward calculation shows that the column vectors @xmath720 of the matrix @xmath721 ( see lemma [ calcsmin5 ] for notation ) are @xmath722 we now modify the standard embedding : replace the disk @xmath715 by @xmath723 , where @xmath723 is obtained by rotating @xmath724 to @xmath725 .
the embedding of @xmath20 so obtained is clearly diffeotopic to the standard embedding .
we may assume that the framing it induces on @xmath726 depends only on @xmath713 and equals @xmath727 .
let the linear map @xmath728 be given by the matrix @xmath729 let @xmath694 be coordinates on @xmath723 , as in section [ whki ] .
let @xmath730 be the map that is given by the matrix @xmath731 in the coordinates @xmath694 .
consider @xmath732 .
for every half integer @xmath733 we define @xmath734 , @xmath735 where @xmath583 is as in section [ whki ] .
then @xmath736 is well defined since it agrees with the inclusion on @xmath737 for each @xmath713 and since @xmath583 is symmetric , i.e. @xmath738 .
the self intersection of @xmath736 is one circle .
it is traced out by the double point of @xmath583 under rotation .
the preimage of this circle is a torus knot of type @xmath739 if @xmath580 is not an integer and a two component link with unknotted components that are linked with linking number @xmath580 if @xmath580 is an integer .
[ theg ] the smale invariant @xmath740 satisfies , @xmath741+n\in\pi_3(v_{5,3}).\ ] ] ( see lemma [ pi3v53 ] for notation . ) thus , as @xmath580 runs through @xmath742 , @xmath736 runs through the regular homotopy classes of immersions @xmath0 .
the proof of proposition [ theg ] constitutes section [ pftheg ] .
the homomorphism ( see proposition [ homtau ] ) @xmath665 is surjective .
the immersion @xmath743 represents a generator of @xmath25 .
it has one self intersection circle with connected preimage .
thus , @xmath744 is a generator of @xmath66 .
[ lkgen ] the invariant @xmath395 ( see lemma [ lk ] ) takes the values @xmath745 we prove lemma [ lkgen ] in section [ pflkgen ] .
the homomorphism ( see corollary [ homlambda ] ) @xmath443 is surjective . by lemma [ lkgen ]
, @xmath746 is a generator of @xmath747 . as in lashof and smale @xcite , theorem 3.4
, we consider an immersion @xmath748 with one transversal double point .
the normal bundle of this immersion has euler number 2 and the boundary of a small tubular neighborhood of it is therefore an immersion @xmath749 , where @xmath750 denotes the unit tangent bundle of @xmath645 .
let @xmath751 be the universal double covering .
let @xmath752 .
then @xmath33 is an immersion .
let @xmath22 be a generic immersion regularly homotopic to @xmath33 .
[ rp3 ] the invariant @xmath72 ( see proposition [ breg4 ] ) is a generator of @xmath622 , the immersion @xmath48 has an odd number of quadruple points and the euler characteristic of @xmath92 is odd .
proposition [ rp3 ] is proved in section [ pfrp3 ] the @xmath600-valued regular homotopy invariants @xmath368 and @xmath378 defined in section [ secmorinv ] are non - trivial .
the homomorphism ( see proposition [ hombeta ] ) @xmath753 is an epimorphism .
in this section we prove theorems [ thmimem ] , [ gi41 ] , [ gi42 ] , and [ gi43 ] , proposition [ theg ] , lemma [ lkgen ] , and proposition [ rp3 ] proposition [ sgntr ] shows that @xmath29 is an isomorphism .
proposition [ 24 ] shows that @xmath212 is a subgroup of the infinite cyclic group @xmath25 of index 24 .
the smale invariant @xmath42 takes values in @xmath11 .
choosing a generator , we identify this group with @xmath491 . for one of the two possible choices ,
the first part of the diagram is correct .
( making the other choice , the second homomorphism in the second row would be @xmath754 instead of @xmath755 ) . to prove the theorem it is enough to show that @xmath756 is surjective .
lemma [ lkgen ] shows that @xmath744 is a generator of @xmath747 and if @xmath73 is as in proposition [ rp3 ] then @xmath72 is a generator of @xmath622 .
hence , if either @xmath757 is a generator of @xmath622 or @xmath758 is a generator of @xmath747 then @xmath756 is surjective .
assume that this is not the case .
then @xmath757 is even and @xmath759 and hence @xmath760 is a generator of @xmath761 .
thus , @xmath756 is surjective .
( since @xmath743 represents a generator of @xmath25 , it actually follows that @xmath757 is odd . )
consider the invariants @xmath57 , @xmath94 , @xmath378 and @xmath368 , defined in section [ secmorinv ] .
since @xmath378 and @xmath368 are both additive under connected summation ( see the proof of lemma [ hombeta ] ) , it follows exactly as for @xmath57 that they are invariant under regular homotopy in @xmath51 .
they both define homomorphisms @xmath762 and lemma [ rp3 ] implies that they are both nontrivial and therefore equal .
_ proof of theorem [ gi41 ] : _ let @xmath62 be a generic immersion . consider @xmath33 as an immersion into @xmath51 and evaluate the @xmath600-valued invariants induced by @xmath378 and @xmath368 . by the above they are equal .
the theorem follows .
let @xmath763 denote the homomorphism obtained from @xmath368 and @xmath378 .
we have the following sequence : @xmath764 where @xmath550 and @xmath765 denote reduction modulo 2 and reduction modulo 4 , respectively .
the brown invariant of a pin structure on a surface reduced modulo 2 equals the euler characteristic of the surface reduced modulo 2 .
therefore , @xmath766 .
moreover , if @xmath43 represents a generator of @xmath25 then @xmath767 is a generator of @xmath66 and so is @xmath95 . as a consequence , @xmath768 for any immersion @xmath769
the sign in this formula depends on the choices of the isomorphisms @xmath338 ( section [ secbeta ] ) and @xmath770 ( section [ tau ] ) . for any one of the two possible choices of @xmath338
there is a unique choice ( out of two possible ) of @xmath770 such that the sign in the above formula is @xmath771 .
_ proof of theorem [ gi42 ] : _ let @xmath43 be a generic immersion and @xmath22 a generic immersion regularly homotopic to @xmath33 .
then @xmath95 is odd if and only if @xmath772 is odd .
but @xmath772 is odd if and only if @xmath72 is odd which is equivalent to @xmath92 having odd euler characteristic .
apply theorem [ gi41 ] .
_ proof of theorem [ gi43 ] : _ let @xmath33 and @xmath48 be as above .
if @xmath92 is orientable then @xmath72 is divisible by @xmath99 .
hence , @xmath773 and therefore @xmath774 . for simplicity of notation
we let @xmath775 denote the _ modified _ standard embedding and @xmath776 denote the map @xmath777 associated to it .
fix @xmath580 and let @xmath778 denote the map @xmath779 associated to the immersion @xmath780 .
let @xmath781 , where @xmath782 is the inverse of the matrix @xmath783 and @xmath784 is the matrix product . to compute the smale invariant of @xmath736
we must compute the homotopy class ( see lemma [ calcsmin5 ] ) @xmath785-[\theta^s]=[f]-[s]=[\lambda_m]\in\pi_3(so(5))=\pi_3(v_{5,3}).\ ] ] since @xmath786 outside the solid torus @xmath787 , we have @xmath788 where @xmath789 .
let @xmath790 , @xmath791 $ ] , @xmath792 $ ] be coordinates on @xmath330 . here
@xmath713 indicates which @xmath726 we are in and on @xmath726 we have @xmath793 using equations ( @xmath393 ) in section [ whki ] we can compute the first columns of @xmath794 .
the result is the following : @xmath795 where @xmath796\\ & + \sin(r ) \bigl[\cos(\varphi - m\theta ) \left(\cos ( \theta)\,\partial_3+\sin ( \theta)\,\partial_4\right ) -\sin(\varphi - m\theta)\,\partial_5 \bigr],\\ v_2(\theta , r,\varphi)= & -\sin(m\theta)\,\partial_1+\cos(m\theta)\,\partial_2.\end{aligned}\ ] ] we note that @xmath797 , for all @xmath798 .
hence , @xmath799 where @xmath800 .
thus , to compute @xmath801\in\pi_3(so(5))$ ] it is enough to compute the homotopy class of @xmath802 which , as we shall see , can be calculated if we know the first two columns of @xmath803 . if @xmath804 then @xmath805 , @xmath806 , where @xmath807 is the usual inner product on @xmath51 .
hence , we have all the information we need to determine @xmath801 $ ] .
in what follows , we first study how to calculate the homotopy class of @xmath808 and then carry out the necessary calculations . in terms of our standard generators of @xmath809 we have , for @xmath802 , @xmath810=u[\sigma]+v[\rho]\in\pi_3(so(4))$ ] , @xmath811 .
the integer @xmath115 is simply the degree of the map @xmath812 , where @xmath100 is the fibration described in section [ sechomth ] and @xmath813 is the first column vector of @xmath803 . to compute @xmath814 $ ] , recall from section [ sechomth ] that @xmath815 and thus , we must compare the homotopy class of the map @xmath816 , where @xmath817 is the projection onto the second factor in the product space @xmath818 , to that of @xmath819 .
the product structure on @xmath98 is obtained by using the section @xmath820 .
the projection @xmath817 is thus determined by the equation @xmath821 we note that @xmath822 .
let @xmath823 .
to compare the homotopy classes of these maps we can use the fibration @xmath824 : since @xmath825 is an isomorphism we might as well compare the homotopy classes of the maps @xmath826 and @xmath827 both mapping @xmath20 to @xmath645 . by the pontryagin construction the homotopy class of any map @xmath828 is determined by its hopf invariant ( the linking number of two regular fibers ) .
the map @xmath113 , has hopf invariant one if endow @xmath20 with the orientation that is coherent with the quaternion framing .
thus , to determine @xmath829\in\pi_3(so(3))$ ] we need only calculate the hopf invariant of @xmath830 , where @xmath831 is the first column vector in the matrix @xmath832 .
this , in turn , means that the integer @xmath84 can be computed as the hopf invariant of the map @xmath833 , where @xmath20 has the same orientation as above .
we now turn to the actual calculations .
we have @xmath834 , @xmath835 where @xmath685 , @xmath836 is the standard basis in @xmath21 and @xmath837 is the standard inner product on @xmath51 .
computing the necessary inner products we find that if @xmath838 then @xmath839 @xmath840 and if @xmath841 then @xmath842 to find the first column of @xmath843 we must calculate @xmath844 . for @xmath845 , @xmath846 , ( for notation see section [ sechomth ] ) . for @xmath847 , let @xmath848 denote its conjugate then @xmath849 .
multiplying gives @xmath850 hence , the the first column of @xmath843 is given by @xmath851 to determine the degree of @xmath852 we consider preimages of the regular value @xmath853 .
for @xmath841 , @xmath854 and for @xmath855 we get the equations @xmath856 where @xmath857 . thus , there are @xmath858 points in the preimage of @xmath859 .
the orientation of @xmath330 induced from the quaternion framing on @xmath20 is the same as that given by the basis @xmath860 .
thus , the orientation of @xmath20 at @xmath859 induced by @xmath852 is @xmath861 .
the orientation of @xmath20 at @xmath859 induced from the standard orientation on @xmath21 is @xmath862 .
so the sign of each point in the preimage is the same .
it is positive if @xmath863 and negative if @xmath864 .
thus , @xmath865 . to compute the hopf invariant of @xmath833 we consider the preimage of the regular value @xmath866 and
see that it is the curve @xmath342 defined by the equations @xmath867 moreover , the vector field @xmath868 along @xmath342 , is mapped by @xmath869 to the vector @xmath870 .
shifting the curve @xmath342 along this vector field we get a curve @xmath406 and @xmath871 using the above orientation on @xmath330 .
so , the hopf invariant of @xmath872 is @xmath500 and thus @xmath873 .
collecting these results , we finally get @xmath874 = [ { { \mathbf m}}]+n = ( ( -2m+2)[\sigma]-[\rho])+n\\ & = -2m[\sigma]+n\in\pi_3(so(5))=\pi_3(v_{5,3}),\end{aligned}\ ] ] as claimed .
let @xmath875 be coordinates on the solid torus where the whitney kinks are placed .
the preimage of the self intersection @xmath876 of @xmath736 is @xmath877 , @xmath878 or @xmath879 , where @xmath404 is some small positive number .
consider the vector field @xmath880 along this preimage .
if we shift @xmath876 along @xmath881 where @xmath882
then we obtain a curve @xmath883 , which bounds a 2-disk in the @xmath884-plane without intersections with @xmath885 .
thus , @xmath886 .
assume that @xmath580 is not an integer
. then @xmath887 is connected .
shifting @xmath545 along @xmath880 gives a curve @xmath888 with @xmath889 .
thus , the framing of @xmath545 which satisfies condition @xmath890 in section [ seclk ] is the one that differs from @xmath880 by @xmath891 rotations .
assume that @xmath580 is an integer
. then @xmath892 have two components and @xmath893 .
shifting @xmath894 along @xmath880 gives a curve @xmath895 with @xmath896 .
thus , the framings of the @xmath894 that satisfy condition @xmath890 in section [ seclk ] differ from @xmath880 by @xmath897 rotations each .
the rotations that we need to add to @xmath880 can be made locally supported . using a local model
it is easy to see how they affect the linking number : let @xmath134 and @xmath329 be the 3-planes @xmath898 and @xmath899 . then @xmath900 is the line @xmath901 . shifting @xmath902 along @xmath903
we get @xmath904 . shifting @xmath902 along @xmath905 , @xmath906
we get the curve @xmath907 .
let @xmath908 , @xmath909 .
then @xmath377 is a 2-chain with boundary @xmath910 , @xmath911 and @xmath912 with sign @xmath500 .
thus , for each rotation in the negative direction we decrease the linking number by 1 . adding the appropriate number of rotations to the framing @xmath880 and shifting the curve @xmath876 along the corresponding vector field we obtain a curve @xmath913 with @xmath914 thus , @xmath915 .
the self intersection of @xmath529 is located close to the transverse double point @xmath916 of @xmath645 .
let @xmath107 and @xmath108 be coordinates on small disks @xmath917 and @xmath378 around @xmath269 and @xmath67 respectively .
we may assume that @xmath918 close to @xmath269 and @xmath67 , @xmath750 is of the form @xmath919 , @xmath562 and the immersion @xmath529 is : @xmath920 where @xmath921 and @xmath922 are coordinates on @xmath923 in the fiber close to @xmath269 and @xmath67 , respectively .
thus , @xmath924 and the preimage @xmath925 is @xmath926 , where @xmath894 is a circle of radius @xmath404 around @xmath927 .
we can find @xmath928 as follows : fix a point @xmath265 in @xmath645 and a unit tangent vector @xmath84 at that point .
the space of 2-planes @xmath929 through @xmath309 in @xmath116 is @xmath930 .
let @xmath931 be the reflection in @xmath929 .
the map @xmath932 is an embedding of @xmath930 into @xmath750 .
consider the double cover @xmath933 .
observe that @xmath934 subdivides @xmath20 into two hemispheres @xmath935 and @xmath936 .
the intersection @xmath937 consists of two circles , isotopic to @xmath938 , @xmath562 . altering the embedding of @xmath930
slightly we may assume that @xmath939 .
the covering @xmath933 is nontrivial on fibers so @xmath940 consists of two tori @xmath941 , @xmath562 and @xmath942 is a double cover .
thus , @xmath943 consists of two meridians in @xmath944 , @xmath562 .
choose a normal vector field @xmath945 to @xmath529 .
let @xmath356 be small and let @xmath946 be positive on @xmath947 , negative on @xmath948 , @xmath309 on @xmath583 , and have nonvanishing derivative in the normal direction of @xmath583 .
( map @xmath949 to @xmath950 and use a suitably scaled height function . )
we use @xmath338 and @xmath945 to perturb @xmath33 . for @xmath798 , let @xmath951 then @xmath48 is an immersion .
clearly , @xmath48 is an embedding outside a small neighborhood of @xmath952 in @xmath20 .
moreover , inside this neighborhood it has only transverse double points if we stay away from @xmath953 . to determine the self intersection of @xmath48 we can therefore restrict attention to neighborhoods of @xmath954 and @xmath955 .
let @xmath956 .
if @xmath957 , @xmath958 are coordinates on @xmath959 and @xmath960 , respectively then @xmath961 is given by @xmath962 normal fields along @xmath963 and @xmath964 are @xmath965 we may assume that the function @xmath966 is given by @xmath967 then , in @xmath968 , @xmath969 is given by @xmath970 thus , @xmath48 has one quadruple point at @xmath971 , with preimages @xmath972 and @xmath48 has four triple lines : @xmath973 where @xmath974 or @xmath975 .
constructing @xmath92 by gluing the nine double point sheets of @xmath48 together gives @xmath976 as shown in figure [ figfg ] , where the six intersection points of the solid lines represents @xmath977 and the remaining twelve open arcs of the solid lines represents @xmath978 .
it follows that @xmath72 as well as @xmath979 are odd .
since @xmath57 and @xmath378 are invariants of regular homotopy in @xmath51 , the proposition follows .
i want to thank oleg viro for valuable ideas and ryszard rubinsztein for fruitful discussions .
i also want to thank the referee for pointing out that the problem on representing regular homotopy classes by embeddings was solved in @xcite .
this shortened the paper considerably .
99 v. i. arnold , _ the vassiliev theory of discriminants of knots _ , in first european congress of mathematics : paris july 6 - 10 , 1992 , volume i , birkhuser ( 1994 ) , 3 - 30 . v. i. arnold , _ plane curves , their invariants , perestroikas and classifications _ , in singularities and bifurcations , edited by arnold , adv
. sov . math .
* 21 * ( 1994 ) 39 - 91 .
t. ekholm , _ immersions in the metastable range and spin structures on surfaces _ , math . scand .
* 83 * ( 1998 ) 5 - 41 .
t. ekholm , _ vassiliev invariants and regular homotopy of generic immersions @xmath980 , @xmath981 _ , j. knot theor . ramif . * 7 * ( 1998 )
1041 - 1064 . t. ekholm , _ geometry of self intersection surfaces and vassiliev invariants of generic immersions @xmath982 _ , preprint , uppsala university ( 1998 ) m. w. hirsch , _ immersions of manifolds _ , trans .
* 93 * ( 1959 ) , 242 - 276 .
j. f. hughes and p. m. melvin , _ the smale invariant of a knot _ , comment .
* 60 * ( 1985 ) , 615 - 627 .
d. husemoller , _ fibre bundles , 3rd ed .
_ , springer verlag ( 1994 ) m. a. kervaire , _ sur le fibr normal une sphre immerge dans une espace euclidien _ ,
33 * , ( 1959 ) 121 - 131 r. c. kirby and l. r. taylor , _ _ pin _ structures on low - dimensional manifolds _ , in `` geometry of low - dimensional manifolds : 2 '' , london math .
notes ser .
* 151 * , cambridge university press ( 1990 ) , 177 - 242 .
r. lashof and s.smale , _ on the immersion of manifolds in euclidean space _ , ann . of math .
* 68 * ( 1958 ) 562 - 583 y. matsumoto , _ an elementary proof of rohlin s signature theorem and its extension by guillou and marin _ , in `` a la recherche de la topologie perdue '' , edited by guillou and marin , birkhuser ( 1986 ) , 119 - 139 . j. w. milnor and m. a. kervaire , _ bernoulli numbers , homotopy groups and a theorem of rohlin _ , proc .
edinburgh ( 1958 ) , 454 - 458 j. w. milnor and j. a. stasheff , _
characteristic classes _ , ann . of math .
study * 76 * , princeton university press ( 1974 ) s. smale , _ classification of immersions of spheres in euclidean space _ , ann . of math .
* 69 * ( 1959 ) , 327 - 344 .
h. whitney , _ the self intersections of a smooth @xmath983-manifold in @xmath984-space _ , ann . of math .
* 45 * ( 1944 ) , 220 - 246 . | regular homotopy classes of immersions @xmath0 constitute an infinite cyclic group . the classes containing embeddings form a subgroup of index 24 . the obstruction for a generic immersion to be regularly homotopic to an embedding
is described in terms of geometric invariants of its self intersection .
geometric properties of self intersections are used to construct two invariants @xmath1 and @xmath2 of generic immersions which are analogous to arnold s invariants of plane curves @xcite .
we prove that @xmath1 and @xmath2 are independent first order invariants and that any first order invariant is a linear combination of these . as by - products ,
some invariants of immersions @xmath3 are obtained . using them
, we find restrictions on the topology of self intersections . |
dyson - schwinger equations ( dses ) provide a viable method for studying the non - perturbative behaviour of field theories . in gauge field theories in particular , a common technique is to analyse the fermion propagator by truncating the infinite tower of dses at the level of the propagator dses .
one then implements an intelligent ansatz for the gauge boson propagator and boson - fermion vertex function@xcite . for quantum electrodynamics in either three ( qed3 ) or four ( qed4 ) dimensions , there has for some years now been an ongoing programme of improving the ansatz for the fermion - photon vertex@xcite .
a principal goal of this programme is to invent an ansatz which respects the gauge covariance of green s functions in accordance with the transformation properties discovered by landau and khalatnikov ( lk)@xcite .
it has traditionally been common practice in dse studies of qed3 or qed4 to assume , either implicitly or explicitly , that in the quenched ( i.e. @xmath0 ) , massless limit , the dses admit the trivial solution of bare fermion propagator and bare vertex in landau gauge@xcite . in the case of qed4 , this assumption has recently been questioned by by bashir et al.@xcite , who use the one - loop perturbative correction to the vertex@xcite to model an unknown additional transverse part of the vertex . in this paper
we explore the problem of constructing a viable vertex ansatz for the case of qed3 , without recourse to the above simplifying assumption . we choose to work with qed3 rather than its four dimensional counterpart because its benign ultraviolet properties render the integrals we encounter finite , obviating the need for awkward numerical regularisations .
the four component version of massless qed3 which we consider here has an interesting chiral - like @xmath1 symmetry which , if dynamically broken , leads to dynamical mass generation .
the evidence from both dse and lattice calculations , suggest that this is almost certainly the case , at least for small numbers of fermion flavours@xcite . here , however , we shall be considering the chirally symmetric solutions , which must also respect the lk transformations , and can therefore be used to place restrictions on the allowed form of the vertex ansatz .
we shall demonstrate the existence of a gauge in which massless qed3 admits a bare fermion propagator solution , though this may not be landau gauge , even in the quenched case .
the vertex function decomposes into two parts , an in principle known part which reduces to the bare vertex in the gauge mentioned , and an extra , unknown transverse part . by studying the gauge parameter dependence of the photon polarisation scalar , we construct a computationally viable vertex ansatz for the first of these two parts .
the massless fermion dse for qed3 in euclidean momentum space is @xmath2 where the euclidean @xmath3 matrices satisfy @xmath4 .
@xmath5 is the photon propagator which , for the class of covariant , nonlocal gauges and for @xmath6 fermion flavours takes the form @xmath7 where @xmath8 the gauge choice @xmath9 with @xmath10 constant , defines the usual covariant gauge .
the regulated photon polarisation scalar@xcite is given by @xmath11 .
\label{vacuum1}\ ] ] a common starting point for constructing vertex anstze is the ball and chiu vertex@xcite . writing the dressed fermion propagator in the chirally symmetric phase as @xmath12
the ball - chiu vertex is given by @xmath13 where @xmath14 \gamma_\mu + \frac{(p + q)_\mu}{p^2 - q^2 } \left[a(p^2)-a(q^2)\right ] \frac{\gamma \cdot p + \gamma \cdot q}{2 } \label{bcvertex},\ ] ] and the transverse part of the vertex @xmath15 satisfies @xmath16 without loss of generality , the transverse part can be written in terms of eight scalar functions @xmath17 as @xmath18 charge conjugation invariance restricts all @xmath17 to be symmetric under interchange of @xmath19 and @xmath20 , except @xmath21 , which is antisymmetric . given the form of the fermion dse , it is reasonable to assume that in the chirally symmetric sector , only those @xmath22 consisting of terms with odd numbers of dirac matrices contribute to the vertex function .
any terms containing even numbers of dirac matrices would need to conspire to produce contributions to the final term in eq .
( [ masslessdse ] ) which integrate to zero . throughout this paper
we shall therefore assume that only the four @xmath22 listed in the appendix contribute to @xmath15 in the chirally symmetric sector .
we also adopt the position that a useful ansatz should be such that the @xmath17 have no dependence on the gauge fixing function @xmath23 other than that entering implicitly via a functional dependence on other green s functions ( such as the @xmath24 dependence in @xmath25 ) .
this allows one to exploit gauge covariance of the solutions of the dse as a constraint on the functions @xmath17 .
then one can always impose gauge covariance on solutions of the dses by fiat as follows : first arbitrarily specify a vertex ansatz @xmath26 of the form of eq .
( [ general ] ) to be the landau gauge vertex , and then _ define _ the ansatz @xmath27 in other gauges to be the lk transform of @xmath26 .
form invariance of the dses under lk transformations then ensures that any propagator solution will automatically have the desired lk transformation properties .
since our choice of landau gauge vertex was arbitrary , a constraint demanding the gauge covariance of propagator solutions then becomes vacuous . ]
the ball - chiu vertex is arrived at by requiring that the following criteria be satisfied : + ( a ) it should satisfy the ward - takahashi identity ( wti ) : @xmath28 , + ( b ) it should satisfy the ward identity : @xmath29 , or equivalently , be free of kinematic singularities as @xmath30 and + ( c ) it must have the same transformation properties as the bare vertex @xmath31 under lorentz transformations and charge conjugation . in many papers dealing with qed dses @xcite the vertex function is effectively assumed to be restricted by a further condition , namely : + ( d ) it must reduce to the bare vertex ( i.e. @xmath15 reduces to zero ) when dressed propagators are replaced by bare propagators .
a consequence of condition ( d ) is that , in landau gauge ( @xmath32 ) , the chirally symmetric solution to the massless , quenched fermion dse eq .
( [ masslessdse ] ) is the free propagator @xmath33 where @xmath34 is the propagator in coordinate space and @xmath35 .
( [ freeprop ] ) follows from the identity @xmath36 which ensures that the perturbative one - loop fermion self energy in this case is zero . in reference
@xcite pennington et al . argue that restriction ( d ) is not justified in qed4 , and indeed not consistent with perturbation theory .
likewise , we have no reason to assume in qed3 that condition ( d ) is valid or that the bare fermion propagator is a solution to the quenched theory in landau gauge . in fact , the vertex function is in principle determined by the vertex dse @xmath37_{dc } k_{cd , ba}(q + \ell , p + \ell,\ell ) , \label{vdse}\ ] ] where the kernel @xmath38 is a sum of skeleton graphs containing the dressed fermion and photon propagators and dressed vertex function .
the photon propagator , and hence the kernel , is determined via eq .
( [ vacuum1 ] ) once the fermion propagator and vertex function are determined .
( [ vdse ] ) is then an integral equation relating @xmath39 and @xmath40 .
we therefore expect the full vertex to be functionally dependent on just the fermion propagator , and , more specifically , for both @xmath39 and @xmath41 to be determined if the fermion propagator is specified to be the bare propagator .
one is tempted to ask whether there may nonetheless be some ( possibly nonlocal ) gauge in which the bare propagator is the solution to the fermion dse .
such a scenario has been suggested by a number of authors in the context of the nonquenched theory@xcite . in these studies , however , the vertex ansatz is taken either to be the bare vertex or a scalar function multiplied by the bare vertex . in either case condition ( d )
is satisfied and in the quenched limit the desired gauge leading to a trivial solution is landau gauge . here
we demonstrate the existence of a nonlocal gauge in which the fermion dse is solved by the bare propagator without assuming condition ( d ) .
let us take the solution to the vertex dse ( [ vdse ] ) to be @xmath42 when the fermion propagator is the bare propagator . with @xmath43 transverse ,
this form is consistent with the ball - chiu form eq .
( [ general ] ) .
substituting eqs .
( [ photon ] ) , ( [ freeprop ] ) and ( [ barver ] ) into ( [ masslessdse ] ) we obtain @xmath44 \frac{\gamma_\mu \gamma\cdot q \gamma_\nu}{q^2 } } \nonumber \\ & = & \int\frac{d^3q}{(2\pi)^3 } \frac{1}{\left\{1 + n_f\pi\left[(p - q)^2\right]\right\}(p - q)^2 } \frac{\gamma_\mu \gamma\cdot q \bar{\gamma}_\mu^{\rm t}(p , q)}{q^2 } \nonumber \\ & & + \int\frac{d^3q}{(2\pi)^3 } \frac{1}{1 + n_f\pi\left[(p - q)^2\right ] } d_{\mu\nu}^{\rm t}(p - q ) \frac{\gamma_\mu \gamma\cdot q \gamma_\nu}{q^2 } .
\label{inteq}\end{aligned}\ ] ] a solution of these equations for the scalar function @xmath23 will then provide us with a gauge in which the fermion dse is solved by the bare propagator . if @xmath43 includes just those @xmath22 whose terms contain an odd number of dirac matrices , lorentz covariance entails that the right hand side of eq .
( [ inteq ] ) takes a form @xmath45 for some scalar function @xmath46 .
fourier transforming both sides then gives @xmath47 which simplifies to @xmath48 = 2\pi\frac{d\phi(u)}{du},\ ] ] with solution @xmath49 so one can in principle extract from eq .
( [ soln ] ) a nonlocal gauge in which the bare propagator solves massless qed3 . in practice , of course , determining the bare propagator gauge will not be a simple matter . now consider the quenched theory , where the only free parameter is the coupling @xmath50 which for qed3 has dimensions of momentum . on dimensional grounds
the chirally symmetric , landau gauge solution to the fermion dse must be of the form @xmath51 for some function @xmath52 . in any other covariant gauge
, the propagator is given by the euclidean lk transformation @xmath53}.\ ] ] specifically , if @xmath54 is the gauge in which the dse admits a bare propagator solution , then @xmath55 } = 1.\ ] ] this can only be achieved if @xmath56 for some dimensionless constant @xmath57 .
it follows@xcite that the gauge in which the propagator reduces to the the bare propagator is one of the usual covariant gauges @xmath58 and that in the general covariant gauge of eq.([covgauge ] ) , the fermion propagator is @xmath59 in coordinate space , or @xmath60
\right\ } , \label{fermion3d}\ ] ] in momentum space .
an analogous result is obtained for qed4 in ref.@xcite , in which the perturbative correction to the vertex is used to model the unknown transverse contribution to the vertex function . in our case
we are faced with a similar problem : without knowing the transverse contribution @xmath43 , we are unable to determine the constant @xmath57 .
the task of determining @xmath43 is a formidable one , and we have nothing more to say about it in this paper . instead ,
in the next section we concentrate on constructing a practical ansatz for the remaining part of the vertex in an arbitrary covariant gauge .
in an arbitrary covariant gauge @xmath10 , the vertex function can be decomposed according to @xmath61 where @xmath62 is obtained by lk transforming the bare vertex @xmath31 from the special gauge @xmath57 obtained in the last section .
likewise , @xmath63 is obtained by lk transforming its value defined in eq .
( [ barver ] ) . by making use of the lk transform of the vertex , namely @xmath64 } , \label{lkver}\ ] ] where @xmath65 , one readily checks that @xmath43 remains transverse and that @xmath66 satisfies the wti .
the function @xmath66 satisfies conditions ( a ) to ( d ) listed in the previous section . in ref.@xcite these properties were exploited to place one further restriction on the form of the vertex ansatz , which we shall henceforth refer to as the _ transverse condition _ : + ( e ) in the limit of zero fermion flavours and zero bare fermion mass , the part @xmath66 of the fermion photon vertex ansatz must be functionally dependent on the chirally symmetric fermion propagator @xmath67 in such a way that @xmath68 where @xmath69 is defined by eq .
( [ dtdef ] ) .
although refs.@xcite and@xcite only claim the transverse condition for the case @xmath70 , we see by the following argument that it also applies to the current case . in the gauge @xmath57
, the transverse condition reduces to the identity eq .
( [ angid ] ) . by fourier transforming eq .
( [ tversecond ] ) to the form @xmath71 it is clear from eq .
( [ lkver ] ) that the transverse condition is invariant with respect to lk transformations , and so should be true in any gauge .
following eq .
( [ general ] ) , we can write @xmath72 by substituting eq .
( [ hatgeneral ] ) into eq .
( [ tversecond ] ) and noting that @xmath73 only has terms containing single @xmath3 matrices we obtain the transverse condition in the form @xmath74 + \frac{1}{(k - \ell)^2 } f(k,\ell ) \right\}= 0 , \label{tversecondition3}\ ] ] where @xmath75 .
\label{fdef}\ ] ] we now turn our attention to the photon polarisation scalar defined in eq ( [ vacuum1 ] ) .
@xmath76 is a gauge invariant quantity@xcite and must remain so in the limit @xmath77 , that is , if calculated using a fermion propagator and vertex function obtained from the quenched version of the theory .
furthermore , if we restrict our attention to that part @xmath78 of the photon polarisation scalar arising from @xmath66 , we see from condition ( d ) that in the gauge @xmath57 it is given by the one - loop result@xcite @xmath79 we are thus led to one more constraint on the form of the ansatz for @xmath66 , namely that : + ( f ) the part @xmath78 of the photon polarisation scalar eq .
( [ vacuum1 ] ) , calculated using @xmath80 given by eq .
( [ fermion3d ] ) and @xmath39 replaced by @xmath66 , must be equal to the one - loop result eq .
( [ vacuumlandau ] ) for _ all _ choices of the gauge fixing parameter @xmath10 .
inserting eq .
( [ hatgeneral ] ) into the polarisation scalar eq .
( [ vacuum1 ] ) and using the wti we obtain @xmath81 \right .
\nonumber\\ & & \left .
\frac{3i}{p^2 } { \rm tr }
\left [ \gamma\cdot p \left ( s(q - \mbox{$\frac{1}{2}$}p ) - s(q + \mbox{$\frac{1}{2}$}p)\right ) \right ] + f(q + \mbox{$\frac{1}{2}$}p , q - \mbox{$\frac{1}{2}$}p ) \right\}. \label{vacuumscalar2}\end{aligned}\ ] ] it is clear that @xmath82 enters only through the function @xmath83 defined by eq .
( [ fdef ] ) . furthermore , given any two vertex anstze satisfying conditions ( a ) to ( e ) , and
corresponding functions @xmath84 and @xmath85 , we must have @xmath86 = 0.\ ] ] this follows from eq .
( [ tversecondition3 ] ) , the first term in the integrand of which is a uniquely determined functional of @xmath87 . integrating over @xmath88 , making the change of variables @xmath89 , @xmath90 , and using eq .
( [ vacuumscalar2 ] ) then formally gives @xmath91 = 0.\ ] ] we now take ansatz number 1 to be any ansatz satisfying only conditions ( a ) to ( e ) , and ansatz number 2 to be the ` correct ' ansatz satisfying conditions ( a ) to ( f ) . we know in principle that ansatz 2 exists .
it is simply the lk transform of the bare vertex in the gauge @xmath57 to any other gauge @xmath10 .
is given in the appendix to ref.@xcite in a form which is unfortunately of no immediate practical use .
] since the transverse condition and the loop integral for the polarisation scalar are form invariant under the lk transformations , ansatz 2 will then have properties ( a ) to ( f ) .
ansatz number 1 must then entail a polarisation scalar satisfying @xmath92 = 0 , \label{psqpiint}\ ] ] provided this integral converges .
we note that the derivation of eq .
( [ psqpiint ] ) requires a change of integration variable which , as we shall see , may or may not be valid , depending on the convergence properties of the integrals involved . in constructing an ansatz for the vertex satisfying conditions ( a ) to ( e ) , care must be taken to remember that not every function @xmath93 satisfying the transverse condition ( e ) corresponds to a viable vertex ansatz satisfying criteria ( a ) to ( d ) .
for instance , while one may be tempted simply to set the integrand of eq .
( [ tversecondition3 ] ) equal to zero , a straightforward substitution shows that this will not do , as the constraint @xmath94 following from eq .
( [ tconstraint ] ) will then not be satisfied .
one is led to take into account the general form of @xmath95 given by eq .
( [ transverse ] ) .
this gives @xmath96 \tilde{g}(k^2,\ell^2,k\cdot\ell ) \right .
\nonumber \\ & & -2 \left[k^2\ell^2 + ( k\cdot\ell)^2 - ( k^2 + \ell^2)k\cdot\ell\right ] g_3(k^2,\ell^2,k\cdot\ell ) \nonumber \\ & & \left.+ 2 ( k^2 - \ell^2)k\cdot\ell g_6(k^2,\ell^2,k\cdot\ell ) \right\ } , \end{aligned}\ ] ] where @xmath97 dong et al.@xcite proceed to satisfy the transverse condition by making the simplifying assumption that the @xmath17 be independent of @xmath98 .
the angular integral in eq .
( [ tversecondition3 ] ) can then be done analytically , giving the transverse condition as @xmath99 , \label{tversecondition4}\end{aligned}\ ] ] where @xmath100 and @xmath88 now mean @xmath101 and @xmath102 respectively , @xmath103 - \frac{1}{4}(k^2 + \ell^2),\ ] ] @xmath104 - \frac{1}{4}(k^2 + \ell^2),\ ] ] and we have set @xmath105 the @xmath106 are well defined if condition ( d ) holds .
an ansatz which satisfies conditions ( a ) to ( e ) can then be found by choosing functions @xmath107 , @xmath108 and @xmath109 satisfying eq .
( [ tversecondition4 ] ) in such a way that they have no @xmath10 dependence other than a possible implicit dependence through @xmath24 .
the simplest way to achieve this , and the way chosen in ref.@xcite , is to set the terms in square brackets in eq .
( [ tversecondition4 ] ) to zero . here
we shall consider the one parameter family of anstze , @xmath110 , namely the chiral limit of the curtis - pennington ( cp ) vertex@xcite was proposed in ref.@xcite .
the chiral limit of the cp vertex is unacceptable as it violates condition ( b ) ( the ward identity)@xcite and leads to a divergent integral for the polarisation scalar eq .
( [ vacuumscalar2 ] ) . ]
@xmath111 the choice @xmath112 gives the ansatz proposed by dong et al.@xcite .
one possible defect of this ansatz for any value @xmath113 is that , for @xmath114 , but @xmath115 , @xmath116 has a logarithmic singularity .
( for @xmath117 , the structure of the transverse tensors @xmath118 is such that @xmath119 , but for @xmath120 and @xmath121 not parallel , no such cancellation occurs . )
however , it is extremely difficult to construct functions @xmath106 satisfying eq .
( [ tversecondition4 ] ) without this defect , so we shall persevere with this family of anstze .
we have numerically calculated @xmath123 using the vertex ansatz containing the transverse piece specified by eq .
( [ fansatz ] ) .
this amounts to evaluating the integral @xmath124 \right .
\nonumber \\ & & + xu \left [ \frac{1}{a(p^2\kappa^2 ) } - \frac{1}{a(p^2\lambda^2 ) } \right ] - \frac{1}{2 } \left [ \frac{1}{a(p^2\kappa^2 ) } + \frac{1}{a(p^2\lambda^2 ) } \right ] \label{vacuumscalar3 } \\ & & \left .
+ \frac{1}{\lambda^2 - \kappa^2 } \left [ \frac{1}{a(p^2\kappa^2 ) } - \frac{1}{a(p^2\lambda^2 ) } \right ] \left[x^2(1 - u^2)\tilde{f}(\kappa^2,\lambda^2 ) + 2\left(x^2u^2 - \mbox{$\frac{1}{4}$}\right ) f_3(\kappa^2,\lambda^2 ) \right]\right\ } , \nonumber\end{aligned}\ ] ] where @xmath125 the results are plotted in fig .
[ fig1 ] for a range of values of @xmath113 , together with the one - loop result eq .
( [ vacuumlandau ] ) . if the fermion propagator of eq .
( [ fermion3d ] ) , which scales as @xmath126 is used , the polarisation scalar will scale as @xmath127 in our numerical results we therefore only consider the case @xmath128 , @xmath129 . by expanding the integrand in eq .
( [ vacuumscalar3 ] ) before the factors of @xmath19 have been scaled out we obtain analytically the first couple of terms of the expansion of @xmath130 about @xmath131 : @xmath132,\ ] ] the first term being the contribution from the minimal ball - chiu part of the vertex .
this infrared behaviour is confirmed by the numerical results plotted in fig.[fig1 ] . on the other hand , inserting the expansion @xmath133 into eq .
( [ vacuumscalar3 ] ) we obtain an expansion for large @xmath19 : @xmath134 the first term in each coefficient of this expansion arises from @xmath25 and has been calculated analytically , while the remaining terms arise from @xmath95 and have been calculated numerically .
for the integral in eq .
( [ psqpiint ] ) to be convergent , the coefficients of @xmath135 and @xmath136 should be zero .
the @xmath135 coefficient can be set to zero by choosing @xmath137 the coefficient of @xmath136 is zero to within the accuracy of our numerical calculations . in fig.[fig2 ] we plot the integrand @xmath138 $ ] for three values of @xmath113 including that given in eq .
( [ beta0 ] ) .
it is clear that the value of @xmath113 which renders the integral convergent leads to a nonzero value for the integral in eq .
( [ psqpiint ] ) .
we suspect that the cause of this discrepancy lies with the logarithmic singularity introduced into our ansatz for @xmath139 , which may have rendered the change of integration variables leading to eq .
( [ psqpiint ] ) invalid .
nevertheless , we see from fig.[fig1 ] that this choice of @xmath113 yields an excellent agreement with the exact one loop result over the range @xmath140 .
the problem of constructing a practicable ansatz for the fermion - photon vertex function in qed3 has been considered .
we observe that the vertex contains a transverse part which can not be determined by considering only the wti and gauge covariance of single particle propagator dses . in order to progress further with determining this transverse part it will be necessary either to consider the vertex dse or , following ref.@xcite , to incorporate information from the perturbative loop expansion of the vertex .
these problems are not considered in this paper . in the process of carrying out our investigations ,
we have demonstrated the existence of a nonlocal gauge for which the chirally symmetric solution to the fermion dse is the bare propagator . in the case of the quenched theory @xmath77
this gauge is in fact one of the usual covariant gauges .
unfortunately the actual gauge which renders the fermion propagator equal to the bare propagator gauge will remain unknown while the full transverse part of the vertex is undetermined .
nevertheless , we have explored the extent to which a vertex ansatz can be constructed working with only single particle propagator dses in the quenched limit .
we write the vertex function as a sum of two pieces , one of which reduces to the bare vertex in the bare propagator gauge , and the other of which is a transverse piece containing information which can only be gleaned from higher @xmath141-point dses .
we give a viable ansatz for the first of these based on the work of dong et al.@xcite .
our ansatz is determined by the constraint that the known contribution to the photon polarisation scalar from this part of the vertex should be reproduced accurately .
our ansatz achieves this goal over practically the entire momentum range , failing only in the extreme infrared .
the authors gratefully acknowledge helpful discussions with p. maris , m. r. pennington , m. reenders and a. h. hams .
we are also grateful to the special research centre for the subatomic structure of matter , adelaide , for hosting the workshop on nonperturbative methods in quantum field theory where part of this work was completed .
of the eight tensors spanning the the transverse part of the fermion - photon vertex function , only those consisting of terms containing an odd number of dirac matrices are used in this paper .
they are given by @xmath142 ; \nonumber \\
t_\mu^3(p , q ) & = & \gamma_\mu(p - q)^2 - ( p - q)_\mu\gamma\cdot(p - q ) ; \nonumber \\ t_\mu^6(p , q ) & = & \gamma_\mu(p^2 - q^2 ) - ( p + q)_\mu\gamma\cdot(p - q ) ; \nonumber \\ t_\mu^8(p , q ) & = & \mbox{$\frac{1}{2}$}[\gamma\cdot p \gamma\cdot q \gamma_\mu - \gamma_\mu \gamma\cdot q \gamma\cdot p ] \end{aligned}\ ] ] our definition of @xmath143 differs from that given in ref.@xcite by a minus sign .
we find this is necessary to reproduce eqs .
( 29 ) and ( 30 ) of ref.@xcite . c. d. roberts and a. g. williams , prog .
nucl . phys . * 33 * , 475 ( 1994 ) .
d. c. curtis and m. r. pennington , phys .
d * 44 * , 536 ( 1991 ) . c. j. burden and c. d. roberts , phys .
d * 47 * , 5581 ( 1993 ) . z. dong , h. munczek , c. d. roberts , phys .
lett b * 333 * , 536 ( 1994 ) .
a. bashir and m. r. pennington , phys . rev .
d * 53 * , 4694 ( 1996 ) .
a. bashir , a. kizilers and m. r. pennington , phys .
d * 57 * , 1242 ( 1998 ) .
l. d. landau , i. m. khalatnikov , zh .
* 29 * , 89 ( 1956 ) [ sov .
jetp * 2 * , 69 ( 1956 ) ] ; b. zumino , j. math . phys .
* 1 * , 1 ( 1960 ) .
kondo , phys .
d * 55 * , 7826 ( 1997 ) .
i. j. r. aitchison , n. e. mavromatos and d. mc neill , phys .
b * 402 * , 154 ( 1997 ) .
a. kizilers , m. reenders and m. r. pennington , phys .
d * 52 * , 1242 ( 1995 ) .
p. maris , phys .
d * 54 * , 4049 ( 1996 ) , and references therein .
m. r. pennington , private communication .
the story behind factor @xmath144 is explained in detail ref.@xcite .
j. s. ball , t - w .
chiu , phys .
d * 22 * , 2542 ( 1980 ) .
e. h. simmons , phys .
d * 42 * , 2933 ( 1990 ) ; t. ebihara et al .
b * 434 * , 85 ( 1995 ) ; k .-
kondo and p. maris , phys .
d * 52 * , 1212 ( 1995 ) , phys .
* 74 * , 18 ( 1995 ) .
t. w. appelquist , m. bowick , d. karabali and l. c. r. wijewardhana , phys .
d * 33 * , 3704 ( 1986 ) .
c .. j. burden , j. praschifka , c. d. roberts , phys . rev .
d * 46 * , 2695 ( 1992 ) . | we consider the problem of designing an ansatz for the fermion - photon vertex function , using three - dimensional quantum electrodynamics as a test case . in many existing studies ,
restrictions have been placed on the form of the vertex ansatz by making the unsubstantiated assumption that in the quenched , massless limit the landau gauge dyson - schwinger equations admit a trivial solution .
we demonstrate , without recourse to this assumption , the existence of a non - local gauge in which the fermion propagator is the bare propagator .
this result is used to provide a viable ansatz for part of the vertex function . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.