article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
streamers of both polarities appear in many phenomena in nature @xcite .
industrial applications , on the other hand , have largely focussed on positive ( cathode directed ) streamers @xcite as they are easier to create around sharp tips than negative ( anode directed ) ones @xcite .
this even has lead to a tendency in the applied physics and electrical engineering literature to use the term `` streamer '' as equivalent to `` positive streamer '' .
positive streamer coronas are used in gas cleaning and ozone production since they were long thought to have a higher energy efficiency than negative streamer coronas @xcite .
however , recent investigations with a new generation of pulsed power supplies have shown that negative streamers at voltages of 50 to 80 kv can convert pulsed electric energy into ozone with an unequalled efficiency of 100 g / kwh in ambient air @xcite .
furthermore , negative dc - corona is used in dust precipitators to charge small particles that then can be drawn out of a gas stream by an electric field .
the properties of these negative streamers at voltages above 50 kv largely resemble those of positive ones @xcite . a qualitative similarity between positive and negative streamers in nitrogen with varying oxygen concentration
is also found in experiments in a 13 cm gap in protrusion - plane electrode geometry at voltages of 82 to 125 kv @xcite while other authors emphasize their difference @xcite . up to now ,
studies have explored limited parameter regimes , seeing either positive streamers only , or both positive and negative streamers . here
we present a systematic study over a wide parameter regime where actually two regimes are seen , one with positive streamers only and another one with positive and negative streamers .
we investigate streamers in ambient laboratory air between a needle and a planar electrode at a distance of 4 cm .
we investigate a voltage range from 5 to 96 kv with three different voltage - supplies that were previously used either in the studies of positive streamers at the physics department @xcite , or for both positive and negative streamers at the electrical engineering department @xcite of eindhoven university of technology .
we attribute similarity or major differences between positive and negative streamers to different voltage ranges . for low voltages ,
thin positive streamers ignite and propagate easily , while negative discharges require much higher ignition voltages , they form thick and short streamers if at all . for voltages above 60 kv ,
positive and negative streamers become more and more alike .
theoretical investigations of the difference between positive and negative streamers in three spatial dimensions ( using the cylindrical symmetry of a streamer to calculate effectively in the two dimensions @xmath6 and @xmath7 ) and including the photo - ionization effect in air are quite rare @xcite , a thorough discussion and new results that closely correspond to the experimental results of the present paper are presented in @xcite . for readers from the geophysical community
, a short reminder is in place why we discuss the dependence on applied voltage rather than on applied field . to create streamers efficiently , and to photograph them with high spatial and temporal resolution ,
streamers in experiments and applications are typically emitted from needle or wire electrodes that create strong electric fields in their neighborhood . as a short consideration
e.g. , of the example for a charged sphere shows , the electric field close to the strongly curved electrode is mostly determined by voltage and electrode geometry and rather independent of the distance to some distant grounded electrode .
it is therefore physically evident and has been confirmed by experiments @xcite , that streamer inception and initial propagation is determined by applied voltage and electrode geometry , and not by some hypothetical average field within the complete discharge gap .
the streamers start in a high field region and consecutively expand into a region with decreasing field .
all streamers presented in this paper are created in a 40 mm gap in a needle to plane electrode geometry in ambient air at atmospheric pressure .
we have chosen to investigate the streamers in this gap length because in longer gaps the streamers can branch and form thinner ones @xcite , and at high voltages in wire - plane electrode geometries they also have been seen to become thicker while propagating @xcite ; in smaller gaps the streamers have not enough space to develop completely @xcite , and it is difficult to measure their velocities . in @xcite we concentrated on the investigation of the thinnest positive streamers at varying pressure and gas composition and used voltage supplies with `` slow '' rise times . in contrast , we here use three voltage supplies with fast rise time to create the thickest streamers possible at the particular peak voltage .
the three supplies together cover the voltage range from 5 to 96 kv , their ranges overlap and we find the resulting streamer properties to depend continuously on the voltage , independently of the used supply .
we also perform control experiments with one slower voltage supply to further confirm the statement in @xcite that power supplies will create similar streamer patterns only if their voltage rise time , peak voltage and internal resistance are similar .
the paper is organized as follows .
the experimental setups are described in section [ setup ] .
the evaluation procedure of photographs and the results are described in section [ res ] ; the section is ordered into a general overview , dependence on the voltage supply , stability field , streamer diameters , streamer velocities , an empirical relation between diameter and velocity , and current and energy .
section [ concl ] contains summary and conclusions .
\a ) + + + b ) + + + c ) a pulsed power supply consists of three parts : ( i ) charge storage , ( ii ) switch , and ( iii ) load .
( i ) the charge is conventionally stored in a capacitor , these are readily available for voltages up to 100 kv .
( ii ) the switch is the crucial part when rise times in the nanosecond range are required .
the prevailing choice still is the spark gap , modern versions combine robustness , high efficiency and long life time @xcite .
alternatives such as semiconductors , whether or not in combination with magnetic compression , offer in principle longer life time , but are vulnerable , have lower efficiency and are very expensive .
( iii ) the load is basically the corona discharge , but impedance matching is a major issue for large systems @xcite .
the three supplies used in the current study are sketched in fig .
[ figsupplies ] ; their properties are listed in table [ tabsupplies ] . in general , the voltage pulse duration is kept short since long pulses at high voltages may allow the initial streamer to develop into a much brighter arc or spark discharge which forms a risk for overexposure of the intensifier of the ccd camera .
the simplest supply is called the c - supply ; it was thoroughly described in @xcite and it is sketched in fig . [ figsupplies](a ) . in a c - supply , a capacitor first is charged and subsequently discharged over the needle plate gap when the spark gap is closed .
it gives the pulse exponential rise and decay times that can be adapted with resistors @xmath8 and @xmath9 .
therefore this supply is very versatile and easy to build .
two versions of the c - supply were used as listed in table [ tabsupplies ] .
the c@xmath10-supply has the advantage of a short rise time @xmath11 ns and quite short duration @xmath12s , but this limits the pulse amplitude .
the c@xmath13-supply has a longer rise time @xmath14 ns and a longer pulse duration @xmath15 ms .
it is used to confirm that the streamer structure changes when the rise time becomes too long @xcite ; the long pulse duration might allow the transition to arc or spark .
the c - supplies deliver up to 60 kv .
._characterization of the electric power supplies in the present experiments .
the resistor @xmath8 influences the rise time of the pulse .
the combination of @xmath9 and c determines the decay time of the voltage pulse .
@xmath16 is a series resistance to determine the discharge current . _ [ cols="^,^,^,^,^,^,^,^,^,^ " , ] the pulse amplitude can be increased through a transmission line transformer ( tlt ) as shown in fig .
[ figsupplies](b ) . in a tlt
several coaxial cables are connected in series or in parallel at the input or the output side in such a way that either voltage or current is multiplied .
a tlt can also be used as impedance matching network between the power source and the corona discharge reactor @xcite .
the second advantage is that the pulse width and the decay time are very short .
this system is further referred to as tlt - supply , it was thoroughly described in @xcite as well .
its main characteristics are also listed in table [ tabsupplies ] ; the voltage rise time is 25 ns , and the pulse duration is 50 ns .
the voltage range is 30 to 60 kv . the third supply charges and discharges using coaxial cables .
this supply is called the power modulator ( pm ) supply and will be abbreviated as pm - supply , it is shown in fig .
[ figsupplies](c ) .
its output pulses have the same characteristics as the tlt - supply . for charging it uses a pulse forming line ( pfl ) which leads to a very high wall - plug efficiency ( 95% ) .
the supply is discussed in more detail in @xcite .
the rise time of the pm - supply is shorter than that of the tlt - supply , namely 15 ns , probably due to the lower inductance of the load ; the pulse duration is 100 ns .
this supply has a range from 40 to 96 kv . with the c - supplies ,
the voltage is measured by a resistive - capacitive divider ( tektronix p6015 for c@xmath10 or northstar pvm4 for c@xmath13 ) at the positions indicated in fig . 2 of @xcite .
the current through the gap is measured with a pearson current monitor ( 6585 ) for c@xmath10 or via the voltage across a small series resistor @xmath17 between cathode and ground for c@xmath13 .
the outer ring across the cathode ensures a well defined , low stray capacity and therefore a fast rise time of the current measurement @xcite . with the tlt - supply
the voltage and current are measured with the tektronix high voltage probe ( p6015 ) and a pearson current monitor ( 6585 ) , respectively , at the locations indicated in fig .
[ figsupplies](b ) .
for all three supplies the signals are digitized on an oscilloscope using 0.2 ns sampling time ( lecroy waverunner 6100a ) .
the pm - supply at the electrical engineering department uses a differentiating - integrating system to measure the fast high amplitude voltage and current waveforms @xcite . in this case a lecroy waverunner 2 is used which has the same sampling time .
the locations of the measurements are indicated in fig .
[ figsupplies](c ) .
most measurements actually those with the c - supplies and the tlt - supply at the physics department were carried out with the electrodes mounted in a large stainless steel vessel as drawn in fig . 2 of @xcite .
this vessel can be evacuated and filled with different gases with pressures in the range of 0.013 to 1 bar , as thoroughly described in @xcite . in the present paper , only measurements at standard temperature and pressure in ambient air are presented , and a large window is taken off the sidewall to ensure the refreshment of the air in the vessel . with the pm - supply at the electrical engineering departement , that generates voltages up to 96 kv , the streamers propagate in the open air without surrounding vacuum vessel .
no significant differences were observed between streamers in open air and in the stainless steel vessel when similar voltage pulses were applied .
all measurements presented in this paper are performed in a 40 mm gap with point - to - plane electrode geometry .
we have selected this gap length because in longer gaps , the streamers typically branch and consecutively form thinner and slower streamers @xcite , while in a 40 mm gap , their diameter and velocity is rather constant .
in shorter gaps the streamers do nt have enough space to develop fully .
the electrodes are as follows .
the needle electrode is made from thoriated tungsten in c@xmath10- and tlt - supply or from pure tungsten in c@xmath13- and pm - supply .
no differences in streamer pattern are observed between these two needles .
the tip is spherical with radius @xmath1815 @xmath19 m , a cone of height 2 mm connects the tip to a cylinder with radius 0.5 mm ; these shapes are not perfect , the surface has microscopic scratches from the grinding process ; even a dropped and misformed asymmetric needle does not create visibly different discharge structures . furthermore , for the c - supplies and the tlt - supply ,
the plate electrode is made from stainless steel , while for the pm - supply , it is from aluminum . at the physics department ,
the discharges generated with the c - supplies and the tlt - supply are photographed with a 4quike intensified ccd camera from stanford computer optics as described in @xcite .
the camera has 1360@xmath201024 pixels and a minimal exposure time of 2 ns . in the wave length range of 200 - 800 nm
, it has a quantum efficiency of @xmath180.3 on average .
each pixel has 12 bit , and different gains can be chosen .
the discharges generated with the pm - supply in the electrical engineering department are photographed with a princeton instruments 576g / rb intensified ccd - camera as described in @xcite .
it should be noted that if the initial streamer channels later evolve into arcs , the intensifier of the cameras could be damaged even after exposure .
therefore a transition into an arc is carefully avoided by keeping the voltage pulse sufficiently short .
a number of pictures are shown in our articles .
however , it must be noted that the actual pictures depend on the chosen representation of the raw data ; this is demonstrated in panels a and b of fig .
a2 of @xcite .
an example of the actual data together with a picture is shown in fig .
5 of @xcite ; profile bars of the real data along two sections through the image are shown on the margin .
streamer diameters are evaluated as the full width at half maximum ( fwhm ) from this original data .
the further evaluation procedure is described in section 3.4 .
kv in the upper row , @xmath21 kv in the middle row , and @xmath22 kv in the lower row .
the discharges in panels a and b were made with the c@xmath13-supply at the physics department , and those in panels c - f with the pm - supply at the electrical engineering department ; therefore different cameras etc . were used .
the white dotted lines indicate the position of the plate electrode below and the height of the needle electrode tip above.,width=604 ] , c@xmath13 or pm are used as indicated in the figure . for positive streamers created by the c@xmath13-supply ,
the data for two different measurement days are shown , they indicate a typical variation between days . discharges below an extension of 10 mm are mostly cloud shaped and will not be called streamers.,width=453 ] as a first overview , figure [ posnegimpression ] shows representative pictures of the discharges as a function of voltage and polarity .
the panels show the complete streamer evolution in a 40 mm point - plane gap in air at 1 bar , i.e. , with a sufficiently long exposure time of the photograph . in panels
a and b , the slow c@xmath13-supply with the long pulse duration is used together with the equipment at the physics department .
no secondary streamers or return strokes appear after the primary streamers at voltages as low as 20 kv , and the panels just show the primary discharge . in panels
c to f , the pm - supply with a voltage pulse duration of 100 ns is used together with the equipment at the electrical engineering department ; the images show the streamers during the complete pulse duration .
streamers similar to those in panels c and d have also been generated with the c@xmath10- or the c@xmath13-supply in the physics department ; in particular , the temporal evolution of positive discharges similar to panel c are described in detail in @xcite .
the initial evolution of negative streamers similar to panel d is shown in fig .
[ figdiag ] below .
the white dotted lines indicate the approximate position of the plate electrode below and the height of the needle electrode tip above . it can be seen that at high voltages , streamers are emitted from the complete needle electrode , not only from its tip . below the electrode spots in panels
c , e and f , mirror images can be seen ; they are reflected from the planar electrode .
the white lines do not precisely intersect with the foot points of all streamers as the camera slightly looks down onto the electrode and the discharge is a three - dimensional structure .
this 3d structure can be resolved with stereographic imaging , for first results , we refer to @xcite .
figure [ lengths ] complements figure [ posnegimpression ] , it shows the maximal length of the streamer channels as a function of voltage and polarity .
the figures illustrate our statement that there are two regimes with a smooth transition zone , a regime of voltages below 40 kv where positive and negative discharges can be clearly distinguished , and a regime above 60 kv where they are quite similar .
this will now be elaborated in more detail . below 40 kv ,
positive and negative discharges are remarkably different .
positive discharges ignite above 5 kv , while negative discharges are only seen above 15 kv .
when the voltage on the positive discharges increases from 5 to 20 kv , thin streamers of increasing lengths form that branch frequently ( for a further characterization of branching , we refer to @xcite ) ; their lengths as a function of voltage are shown in figure [ lengths ] .
they bridge the gap above 20 kv as shown in figure [ posnegimpression]a . at this same voltage ,
the negative discharge only forms a small cloud around the tip and no streamer , as can be seen in figure [ posnegimpression]b .
negative streamers form above 30 kv , but they are then still too short to determine their diameters . whenever their diameter can be determined , they are much thicker than the 0.2 mm diameter observed for the thinnest positive streamers .
negative streamers bridge the complete 40 mm gap only at voltages above 50 kv as shown in figure [ lengths ] .
panels c and d in figure [ posnegimpression ] show positive and negative streamers at 46 kv where both are 1 to 2 mm thick but the negative ones do not reach the plate electrode and hardly branch while the positive steamers cross the gap and form many branches . in both cases ,
the discharge emerges not only from the needle tip , but also above it , most likely at sharp edges .
above 60 kv , positive and negative streamers become increasingly more alike .
the negative streamers cross the electrode gap at 56 kv ( figure [ figdiag ] ) . at voltages above 75 kv ,
streamers of both polarities are 2 to 3 mm thick , bridge the gap and hardly branch .
the positive streamers still branch a little more than the negative ones ( see panels e and f in figure [ posnegimpression ] ) .
we remark that figure [ posnegimpression]f shows an overexposed streamer region in the middle of the gap while the regions near the tip and near the plate are darker .
the length of this region increases with increasing voltage from 16 mm through 18 mm to 23 mm at 64 kv , 75 kv and 82 kv , respectively .
it arises _ after _ the primary streamers have bridged the electrode gap ( where `` primary streamer '' denotes the first group of streamers after the voltage has been applied @xcite ) .
it is not known why this large region forms in the middle of the gap ; usually such a region stretches out from the tip and is interpreted as a secondary streamer or as a glow developing along the trails of the primary streamers ( as in figure [ posnegimpression]e ) .
striated or subdivided secondary streamers are also reported in @xcite , however , without photographs .
the spots on the plate electrodes also appear after the streamer has crossed the gap ; the cathode spots appear more or less spherical , whereas anode spots are contracted at the plate but become broader and diffuser above the surface ; they continue into the streamer more gradually .
it should be noted that the streamers in figure [ posnegimpression ] are created with the c@xmath13-supply at 20 kv and with the pm - supply at higher voltages . for positive streamers at voltages of at least 30 kv
, we have shown in @xcite that the slow voltage rise time of the c@xmath13-supply creates thinner channels than the other supplies as the streamers start to propagate before the voltage has reached its maximal value ; this observation is also confirmed in the present measurements as demonstrated in panels b and c in figure [ posnegfotos ] ; these panels show positive streamer discharges at the same peak voltage of @xmath1840 kv , but for voltage rise times of 30 or 150 ns , respectively .
however , at 20 kv positive streamers do not show this supply dependence as the streamer inception seems to be sufficiently slow at such low voltages .
one might wonder whether a similar supply dependence holds for negative streamers , but we have no evidence for it .
furthermore , one can speculate whether the negative streamers in figure [ posnegimpression](d ) do not cross the gap because the voltage pulse duration in the pm - supply is only 100 ns ; if the streamer inception time would be several tens of nanoseconds , the remaining pulse duration could be insufficient to let the streamers cross the gap . to test these hypotheses ,
we have plotted the measured lengths of negative streamers as a function voltage in figure [ lengths ] .
the streamers at 15 and 20 kv are created with the c@xmath13-supply , and for voltages of 30 kv and larger with the pm - supply .
the figure shows that the data follow a smooth curve across the change of supplies ; this suggests that the negative streamers are more robust against changes of the power supplies than the positive ones , and that the negative streamers in figure [ posnegimpression ] have reached their maximal length during the 100 ns duration of the voltage pulse of the pm - supply .
a common hypothesis is that there is a stability field , i.e. , a minimal field that the streamer needs to propagate ; this field would also be the characteristic field inside the streamer @xcite .
this hypothesis is guided by the empirical observation , that the ratio of applied voltage over maximal streamer length is approximately constant for a given gas and polarity ; this ratio has the dimension of an electric field .
( the role of some hypothetical average background field is discussed in the introduction . ) for positive streamers in point - plane and plane - plane gaps in air , a value of 4.4 to 5 kv / cm is fitted to experiments in @xcite and references therein .
the data for positive streamers in figure [ lengths ] is fitted by @xmath23 kv / cm , which is in agreement with the above values within the error bar . for negative streamers ,
typical field values in the literature are a factor 2 or 3 higher @xcite .
such values can also be fitted to our data for negative streamer in figure [ lengths ] ; however , it should be noted that the length is certainly not a linear function of the voltage in this curve , also because there are no negative streamers below 40 kv , and that the approximation of a constant stability field for negative streamers is questionable in view of this data . .
the light at the bottom of figure ( e ) is a reflection of the streamer light on the anode plate . how velocities are determined from such data ,
is described in section [ vel].,width=604 ] the streamer head diameters are determined from iccd - images such as in figure [ figdiag]d ; for the evaluation of the raw data imaged in such figures , we refer to section [ eval ] .
the full width @xmath24 at half maximum ( fwhm ) is measured and averaged over the @xmath25-direction for as long as the streamer channel is straight to suppress the stochastic single photon fluctuations within the pixels of the camera .
the same definition of the diameter was also used in @xcite . only single in - focus streamers at a place without return stroke or electrode effects are evaluated , typically in the middle of the electrode gap .
depending on voltage and polarity , typically three to ten streamers per photograph were evaluated , and the diameters were furthermore averaged over three to five photographs per voltage . in the measurements with the pm - supply , secondary streamers can appear after the primary streamers have crossed the gap , and occasionally secondary streamers are evaluated instead of primary streamers when there are not sufficiently many suitable primary streamer channels .
however , primary or secondary streamer diameter show no significant difference with the pm - supply with its high voltages , as also reported in @xcite . for a , b and c , and in figure [ posnegimpression]e for d.,width=377 ] .
panels a and b are made with the slow c@xmath13-supply , panel c with the fast c@xmath10-supply .
the maximal voltage is a ) + 10 kv , b ) + 41 kv , and c ) + 40 kv .
the white dotted lines indicate the electrode positions as in figs .
[ posnegimpression ] and [ figdiag ] .
the streamers in panel a at 10 kv have reached their maximal length ; the streamers in panels b and c at @xmath1840 kv are still propagating .
panels b and c demonstrate how strongly positive streamers at a voltage saturation value of 40 kv depend on the voltage rise time of the supply , as discussed in @xcite.,height=147 ] the diameters of positive and negative streamers as a function of applied voltage are shown in figure [ posnegdiam ] .
the different power supplies and polarities are indicated with symbols .
the measurements of positive streamers with supplies c@xmath10 , tlt and pm together span a continuous curve within the error bars , consistently with what we found in @xcite .
the positive streamers created with the c@xmath13-supply remain thin when the peak voltage increases , as is illustrated in panel b of figure [ posnegfotos ] ; this is also consistent with @xcite . in this case
, the streamers already start to propagate while the voltage is still below its saturation value ; furthermore their current is limited by the large series resistance @xmath16 . for negative streamers ,
diameters are only measured for experiments performed with the pm - supply .
positive streamers ignite above 5 kv .
as shown in figure [ posnegdiam ] , their diameter then is 0.2 mm , i.e. , the minimal diameter @xcite . for 20 kv ,
their diameter is 0.4 mm , and the streamers bridge the 40 mm gap for the first time
. then the diameter increases by a factor of 10 ( i.e. , from 0.2 to 2 mm ) when the applied voltage increases only by a factor of @xmath182 ( from 25 to 55 kv ) . above 55 kv
, the diameter continues to increase , but less rapidly ; it reaches 3 mm at 96 kv . whether for higher voltages , the diameter can become even higher , or whether it saturates eventually , is an open question
we observed that at very high voltages , e.g. at + 82 kv in figure [ posnegvi]a , the voltage rise time of the pm - supply again becomes comparable to the streamer inception time as the inception time decreases with increasing voltage ; this entails that at high voltages the streamers start before the voltage has saturated .
this is also observed in @xcite with the same pm - supply at + 74 kv , while it is not observed at -72 kv , in agreement with the discussion in section [ depsupply ] .
the negative discharge becomes visible at 15 kv , but does not yet form streamers .
the diameter of the negative streamers can be measured at voltages of 40 kv or higher . for 40 kv
the diameter is 1.2 mm .
the negative streamers bridge the gap for the first time at @xmath18 56 kv ( figure [ figdiag ] ) with a diameter of 1.5 mm .
their diameter increases with applied voltage to @xmath182.3 mm for voltages between 76 and 82 kv . on average , the positive streamers are about 10% thicker than the negative ones , but this difference is within the error margins .
( mm ns).,title="fig:",height=226 ] a ) ( mm ns).,title="fig:",height=226 ] b ) in the experiments at the physics department with the c - supplies and the tlt - supply , the light trace on the photographs can be interpreted as the path that the streamer head has followed during the exposure time of the camera @xcite ; therefore the streamer velocities are measured as the full width at half maximum ( fwhm ) of the length @xmath25 in the propagation direction , divided by the exposure time @xmath26 of the camera ; the procedure is illustrated in figure [ figdiag ] . to minimize the error ,
sufficiently long exposure times are used such that the streamer head crosses 1/4th to 1/8th of the electrode gap ; these times are 10 to 100 ns ; the method is further described in @xcite . the velocities are evaluated at some distance from the electrodes ; typically three to ten streamers per photograph and three to five photographs per voltage are evaluated . in
the presently investigated 40 mm gap , the streamer velocity and diameter are approximately constant .
in contrast , in the case of a 160 mm electrode gap the streamer velocity does depend on the position and diminishes after branching @xcite . in the experiments at the electrical engineering department with the pm - supply ,
the velocities are measured as the propagation distance between photographs ( @xmath27 in figure [ figdiag]a , b ) divided by the difference in delay time @xmath28 of the photographs , where each photograph is taken during a different discharge pulse .
the streamers that have propagated furthest are evaluated in each picture , they predominantly move straight downward .
this method can be used since the jitter in streamer inception at high voltages is as small as about 1 ns @xcite .
five to ten photographs with different delays ( e.g. , the five pictures in figure [ figdiag ] ) are taken per voltage , depending on the gap transit time of the streamer .
position is plotted as a function of time for each photograph and an average velocity is determined graphically .
one or two propagation sequences per voltage are evaluated .
only the fastest streamers are used since they propagate most parallel to the camera s focal plane .
again the velocity is found to be constant throughout the gap @xcite .
we remark that the ambiguity of the two - dimensional images of the three - dimensional discharge can lead to an underestimation of the velocities that is counteracted by evaluating only the fastest streamers .
velocity measurements with the stereographic imaging method introduced in @xcite are currently under way .
figure [ posnegvel]a shows the resulting velocities of positive and negative streamers as a function of voltage .
the velocities of streamers of both polarities increase almost linearly with voltage .
velocities of negative streamers at voltages lower than 56 kv could not be measured because the streamers did not propagate far enough to measure them reliably .
note that the velocity of positive streamers increases rapidly in this range , from 0.2 mm / ns at 20 kv to @xmath18 1.5 mm / ns at 56 kv .
figure [ posnegvel]a shows that the negative streamers are about 25% slower than the positive ones when their velocities can be measured . in figure
[ posnegvel]b the velocity is plotted as a function of the diameter ; the data is extracted from figures [ posnegdiam ] and [ posnegvel]a .
the figure shows that the velocity @xmath3 increases with increasing diameter @xmath4 ; it can be fitted quite well by the empirical equation @xmath29(mm ns ) where @xmath3 and @xmath4 are measured in their natural units .
the approximation also works well for streamers of intermediate radius as measured in @xcite .
however , streamers with a minimal diameter of about 0.2 mm @xcite are a factor 5 too slow in this approximation .
the velocities of the thin streamers with diameters less than 0.5 mm in @xcite are also underestimated by a factor of about 4 .
82 kv for a ) positive and b ) negative streamers generated with the pm - supply .
the generated discharges are shown in panels e and f of fig .
[ posnegimpression ] .
the current and voltage lines are smoothed by averaging over 5 adjacent points . in panel a , characteristic times and capacitive versus discharge currents are indicated , as discussed in the text.,title="fig:",width=313 ] a ) 82 kv for a ) positive and b ) negative streamers generated with the pm - supply .
the generated discharges are shown in panels e and f of fig .
[ posnegimpression ] .
the current and voltage lines are smoothed by averaging over 5 adjacent points . in panel
a , characteristic times and capacitive versus discharge currents are indicated , as discussed in the text.,title="fig:",width=313 ] b ) a typical evolution of current and voltage for positive and negative streamers with the pm - supply is shown in fig .
[ posnegvi ] , the generated discharges are shown in panels e and f of fig .
[ posnegimpression ] .
the voltage pulse reaches approximately @xmath30 kv and lasts for about 100 ns .
while the voltage @xmath31 rises , it creates a capacitive current @xmath32 in the circuit ; @xmath33 is the capacitance of the discharge vessel in the absence of a discharge .
the capacitive currents are indicated in panel a of fig .
[ posnegvi ] . after approximately 11 ns ( as indicated in the figure ) ,
the current is mainly the discharge current ; its maximum is approximately 45 a for both polarities .
the positive discharge current remains at this level for the duration of the voltage pulse ( a plateau of duration @xmath1880 ns ) ; the oscillations are due to imperfect matching with the external circuit .
the negative discharge current does not have this plateau but slowly decreases to 10 a before the voltage pulse drops back to zero .
estimates for the dissipated energies within the complete discharge are shown in figure [ posnegen ] ; they are calculated by integrating the product of voltage and current over time , after the capacitive part of the total current is subtracted @xcite .
the total energy of the discharge is calculated by integrating to 200 ns when the voltage pulse certainly has finished .
[ figdiag ] illustrates that the first group of streamers , the so - called primary streamers @xcite , cross the gap within 27 ns for @xmath34 kv .
this gap crossing time varies from 80 ns for 30 kv to 10 ns for 96 kv for positive streamers , and from 27 ns for 56 kv to 15 ns for 83 kv for negative ones . at a later time , secondary processes occur as described elsewhere , see e.g. @xcite .
the energy of the primary streamers is calculated by integrating from 11 ns ( when the total current becomes larger than the capacitive current , i.e. , when the particle current begins to flow ) to the time when the primary streamers have bridged the gap .
this arrival time is estimated from the streamer velocity according to section [ vel ] . for 82 kv
, it is 12 ns , therefore duration of the primary streamers @xmath35 is assumed to last from @xmath36 ns to @xmath37 ns , as indicated in panel a of fig .
[ posnegvi ] .
figure [ posnegen ] shows that the total energy of positive discharges is higher than for negative discharges .
this is because the current persists longer in the positive discharge while it decreases for the negative ones ( figure [ posnegvi ] ) .
the energy of the primary streamers is also shown in figure [ posnegen ] . though these values must be regarded as an estimate , the data are similar for positive and negative streamers .
these energies are never higher than 50 mj and thus are considerably lower than the total energy which is maximally 400 mj for a positive discharge and 220 mj for a negative one at 90 kv . our observation that positive and negative streamers have similar energies , is consistent with @xcite .
our measurements indicate that the inception and propagation processes of positive and negative streamers in ambient air are quite different .
positive streamers emerge when a voltage of at least 5 kv is applied to a needle electrode at 4 cm distance from a planar grounded electrode , while negative streamers only propagate above 40 kv ( which is around the dc - breakdown voltage of the gap ) .
two regimes have to be distinguished . * @xmath38 : positive streamers propagate .
their diameters increase from 0.2 mm at 5 kv to 1 mm at 40 kv , they bridge the complete electrode gap above 20 kv and they branch .
their velocity ranges from @xmath39 m / s at 5 kv to @xmath40 m / s at 40 kv .
a negative discharge is only visible above 20 kv as a glowing cloud near the electrode tip ; no negative streamers are formed . *
@xmath41 : negative streamers appear , but they do not cross the gap for voltage below 56 kv . positive and negative streamers have a similar diameter of 1 mm at 40 kv that increases up to 3 mm for 96 kv .
the positive streamer velocity increases from @xmath40 m / s at 40 kv to @xmath2 m / s at 96 kv . the velocity of negative streamers can only be measured for voltages higher than 56 kv which is then about 25% lower than for positive ones .
the energy of the primary positive and negative streamers are similar ; it ranges from 20 to 50 mj for voltages from 74 to 90 kv .
we also find the completely empirical fit @xmath42(mm ns ) for the relation between velocity @xmath3 and diameter @xmath4 of positive and negative streamers . as a counterpart to the present experimental investigation of streamers of both polarities ,
a theoretical investigation is presented in @xcite . in that paper , the low voltage regime
is investigated , and a strong asymmetry between short positive and negative streamers is found , in agreement with the present experiments . in particular , positive simulation streamers
next to needle electrodes easily evolve out of some initial ionization seed ; they have small diameters comparable to experiments and a strong field enhancement at the streamer tip ; this field enhancement allows them to propagate into the regions with lower background fields further away from the needle electrode . in these simulations
, we find approximately the same empirical relation between velocity and diameter for short positive streamers as in panel b of fig .
[ posnegvel ] .
on the other hand , negative simulation streamers evolve as well out of some initial ionization seed in the high field region next to the electrode needle ; but then their diameter increases when penetrating the regions with lower fields ; therefore the field is less enhanced and they easily extinguish .
we conclude that the simulations presented in @xcite do allow to understand essential physical features of the low voltage regime of the experiments presented here .
this work is financially supported by stw under contract number 06501 and by nwo under contract number 047.016.017 .
10 raizer yu p 1991 , _ gas discharge physics _ , berlin : springer bazelyan e m , raizer yu p 1998 , _ spark discharge _ ( crc press , boca raton , florida ) williams e r 2006 , s91 wagner k h 1966 , z. physik * 189 * 465 phelps c t , and r f griffiths 1976 , j. appl . phys . * 47 * , 2929 - 2934 marode e 1975 , j. appl
. phys . * 46 * 2005 - 2015 bastien f , and e marode 1979 , 249 - 263 spyrou n , and c manassis 1989 , 120 - 128 tajalli h , d w lamb , and g a woolsey 1989 , 1497 - 1503 creyghton y l m 1994 , _ pulsed positive corona discharges : fundamental study and application to flue gas treatment _ ,
phd thesis , eindhoven university of technology , the netherlands .
available on http://alexandria.tue.nl/extra3/proefschrift/prf10a/9402886.pdf allen
n l , and a ghaffar 1995 , 331 alexandrov n l , and e m bazelyan 1996 , 2873 pancheshnyi s , s sobakin , s starikovskaia , and a starikovskii 2000 , plasma physics reports * 26 * 1054 - 1065 pancheshnyi s , s m starikovskaia , and a y starikovskii 2001 , 105 - 115 pancheshnyi s and a starikovskii 2003 , 2683 - 2691 pancheshnyi s and a starikovskii 2004 , plasma sources sci .
* b1-b5 pancheshnyi s , m nudnova , and a starikovskii 2005 , phys . rev .
e * 71 * 016407 pancheshnyi s 2005 , plasma sources sci .
* 14 * 645 - 653 ono r , and oda t 2003 , 1952 - 1958 ono r , and oda t 2005 , 730 - 735 ono r , and oda t 2008 , 035204 van veldhuizen e m ( ed . ) 2000 , _
electrical discharges for environmental purposes : fundamentals and applications _
, huntington : nova science publishers winands g j j 2007 , _ efficient streamer plasma generation _ , phd thesis , eindhoven university of technology , the netherlands .
available on http://alexandria.tue.nl/extra2/200710708.pdf van heesch e j m , g j j winands , and a j m pemen , _ evaluation of pulsed streamer corona experiments to determine the o@xmath43 radical yield _ , submitted to , cluster issue on `` streamers , sprites and lightning '' blom p p m , smit c , lemmens r h p and van heesch e j m 1994 , _ combined optical and electrical measurements on pulsed corona discharges gaseous dielectrics _ ,
vol vii , ed : christophorou l g and james d r ( new york : plenum ) blom p p m 1994 , _ high - power pulsed corona _ , phd thesis , eindhoven university of technology , the netherlands .
available on http://alexandria.tue.nl/extra1/prf14a/9702338.pdf winands g j j , liu z , van heesch e j m , pemen a j m and yan k 2008 , _ ads and cds streamer generation as function of pulse parameters _ , ieee trans .
plasma sci .
[ to appear ] winands g j j , liu z , pemen a j m , van heesch e j m , yan k , _ analysis of streamer properties in air as function of pulse and reactor parameters by iccd photography _ , yi w j and williams p f 2002 , 205 van veldhuizen e m and rutgers w r 2002 , 2169 briels t m p , van veldhuizen e m and ebert u 2004 , _ experiments on propagating and branching streamers in air _
, proceedings xv int . conf . on gas discharges and applications http://gd2004.ups.tlse.fr , available through
http://homepages.cwi.nl/@xmath44ebert/gd2004-corona.pdf briels t m p , kos j , van veldhuizen e m and ebert u 2006 , 5201 briels t m p , van veldhuizen e m and ebert u 2008 , _
time resolved measurements of streamer inception in air _ ,
_ ieee trans .
plasma sci .
_ [ in print ] briels t m p , van veldhuizen e m , ebert u 2008 , _ positive sreamers in ambient air and a n@xmath13:o@xmath13-mixture ( 99.8:0.2 ) _ , _ ieee trans . plasma sci . _
[ in print ] briels t m p , van veldhuizen e m and ebert u 2008 , _ positive streamers in air and nitrogen of varying density : experiments on similarity laws _ , cluster issue on `` streamers , sprites and lightning '' in 0805.1364 nijdam s , moerman j s , briels t m p , van veldhuizen e m , ebert u 2008 , appl .
92 * 101502 babaeva n y and naidis g v 1997 , _ ieee trans .
plasma sci .
_ * 25 * 375 liu n , pasko v p 2004 , _ j. geophys .
* 109 * a04301 luque a , ebert u , montijn c , hundsdorfer w 2007 , _ appl .
lett . _ * 90 * , 1501 luque a , ebert u , hundsdorfer w 2007 , _
e - prints : _ 0712.2774 .
luque a , ratushnaya v , ebert u 2008 , _ positive and negative streamers in ambient air : modeling evolution and velocities _ , revised for the cluster issue on `` streamers , sprites and lightning '' in , _
e - prints : _
0804.3539 winands g j j , liu z , pemen a j m , van heesch e j m , and yan k 2005 , _ rev .
instrum . _ * 76 * 085107 liu z , yan k , winands g j j , pemen a j m , van heesch e j m , and pawelek d b 2006 , _ rev .
_ * 77 * 073501 yan k 2001 _ corona plasma generation _ phd thesis technische universiteit eindhoven , available through http://alexandria.tue.nl/extra2/200142096.pdf winands g j j , liu z , pemen a j m , van heesch e j m , yan k and van veldhuizen e
m 2006 , 3010 - 3017 gravendeel b 1987 , _ negative corona discharges : a fundamental study _
, phd thesis , eindhoven university of technology , the netherlands .
available through http://alexandria.tue.nl/extra3/proefschrift/prf5a/8612916.pdf .
van houten m a , _ electromagnetic compatibility in high voltage engineering _
thesis , eindhoven university of technology , the netherlands ( 1990 ) .
available on www.tue.nl/bib .
sigmond r s 1984 , j. appl .
phys . * 56 * 1355 allen n
l and ghaffar a 1995 , 331 ebert u , montijn c , briels t m p , hundsdorfer w , meulenbroek b , rocco a and van veldhuizen e m 2006 , s118 van veldhuizen e m and rutgers w r , j. phys .
* 36 * ( 2003 ) 2692 . | positive and negative streamers are studied in ambient air at 1 bar ; they emerge from a needle electrode placed 40 mm above a planar electrode .
the amplitudes of the applied voltage pulses range from 5 to 96 kv ; most pulses have rise times of 30 ns or shorter
. diameters , velocities and energies of the streamers are measured .
two regimes are identified ; a low voltage regime where only positive streamers appear and a high voltage regime where both positive and negative streamers exist .
below 5 kv , no streamers emerge . in the range from 5 to 40 kv , positive streamers form , while the negative discharges only form a glowing cloud at the electrode tip , but no streamers . for 5 to 20 kv ,
diameters and velocities of the positive streamers have the minimal values of @xmath0 mm and @xmath1 m / s . for 20 to 40 kv ,
their diameters increase by a factor 6 while the voltage increases only by a factor 2 . above the transition value of 40 kv , streamers of both polarities form ; they strongly resemble each other , though the positive ones propagate further ; their diameters continue to increase with applied voltage . for 96 kv ,
positive streamers attain diameters of 3 mm and velocities of @xmath2 m / s , negative streamers are about 20% slower and thinner .
an empirical fit formula for the relation between velocity @xmath3 and diameter
@xmath4 is @xmath5(mm ns ) for both polarities . streamers of both polarities dissipate energies of the order of several mj per streamer while crossing the gap . |
magnesium diboride owes its high critical temperatures of 40 k @xcite to the interplay between two distinct electronic bands crossing the fermi level .
about half of the quasiparticles belongs to the quasi two - dimensional hole @xmath3 band and the cooper pairs there are strongly coupled via the boron vibrational mode @xmath4 .
the rest of the quasiparticles resides in the three - dimensional @xmath2 band with a rather moderate electron - phonon interaction . without the effect from the @xmath3-band
the transition temperature @xmath5 would be just a fraction of those found in the real mgb@xmath0 @xcite .
the two band superconductivity is the most spectacularly revealed by the presence of two very distinguished energy gaps , large in the @xmath3 and small in the @xmath2 bands @xcite .
superconducting properties of such a system are sensitive to the scattering . due to coexistence of charge carriers from two almost completely separated bands two _ intraband _ and one _ interband _ scattering channels have to be distinguished .
the _ interband _ scattering by non magnetic scatterers is supposed to have a particularly strong effect in a two - band superconductor : it will blend the strongly and weakly coupled quasiparticles , merge two gaps and consequently decrease @xmath5 . in the case of mgb@xmath0 @xmath5 can drop down to about 20 k @xcite .
fortunately , a different symmetry of the bands ensures that the interband scattering remains small also in very dirty mgb@xmath0 samples which show about the same @xmath5 as the purest material @xcite .
a systematic decrease of @xmath5 is achieved in substituted mgb@xmath0 samples .
the only on site substitutions by non magnetic elements which are known so far are carbon for boron and aluminium for magnesium .
the carbon and aluminium atoms in mgb@xmath0 take indeed a role of scatterers but they also dope the system with one extra electron which inevitably leads to the filling effect in the @xmath3 band , with strongly coupled holes . recently ,
@xcite have introduced a model incorporating both effects : interband scattering and band filling in mgb@xmath0 .
the former effect leads to an increase of the small gap @xmath6 and decrease of the large @xmath7 while the latter suppresses both @xmath6 and @xmath7 .
there is a rather controversial situation as far as the strength of both effects in the al and c - doped mgb@xmath0 is concerned @xcite .
in contrast to the interband scattering an increase of the scattering within the bands does not have any effect on the two gaps .
but , the selective tuning of the _ intraband _ scattering can lead to the expressive variation of the values of the upper critical magnetic field @xmath8 and its anisotropy @xcite . upon increased c - doping @xmath8 is increased significantly in both principal crystallographic directions ( parallel to the @xmath9 hexagonal boron layers and the @xmath10-axis direction ) . on the other hand al substitution
suppresses @xmath11(0 ) in the @xmath9-plane direction and in the @xmath10-axis direction @xmath11 is hardly changed .
this difference behavior is due to a different influence of c- and al - doping on intraband scatterings .
while the strong increase of @xmath8 with the c - doping is due to graded dirty limit conditions , al - doped samples stays still in the clean limit and the decrease of @xmath11 is just a consequence of the lower @xmath5 @xcite .
a weight of scatterings in separated bands is still not clear . in this paper
we present the systematic study of the superconducting energy gaps in both al- and c - doped magnesium diborid systems at stoichiometries mg@xmath12al@xmath13b@xmath0 with @xmath14 and 0.2 and mg(b@xmath15c@xmath16)@xmath0 with @xmath17 and 0.1 . for
the carbon doping this work is a supplement to our previous studies @xcite on mg(b@xmath15c@xmath16)@xmath0 with @xmath18 and 0.1 .
the influence of both substitutions for the scattering processes is discussed .
polycrystalline samples of al - doped mgb@xmath0 have been prepared by a two step synthesis at high temperatures as mg@xmath19al@xmath20b@xmath0 and mg@xmath21al@xmath22b@xmath0 with @xmath5 = 30.5 k and 23.5 k , resp . @xcite .
the carbon substituted samples mg(b@xmath23c@xmath16)@xmath0 with @xmath24 and with the respective transitions at @xmath5 = 39 k , 33 k , 28.2 k , and 22 k , were synthesized in the form of pellets following the procedure described in ref.@xcite from magnesium lumps and b@xmath25c powder .
for @xmath26 we have worked with the wire segments with @xmath5 = 39 k , 37.5 k , and 36.2 k @xcite .
the undoped mgb@xmath0 sample with @xmath27 k in a dirty limit with the high upper critial field has been prepared from boron and magnesium powder as described elsewhere @xcite the point - contact ( pc ) measurements have been realized via the standard lock - in technique in a special point - contact approaching system with lateral and vertical movements of the pc tip by a differential screw mechanism .
the microconstrictions were prepared in situ by pressing different metallic tips ( copper and platinum formed either mechanically or by electrochemical etching ) on different parts of the freshly polished surface of the superconductor .
point - contact measurements on the wire segments have been performed in a `` reversed '' configuration - with the mg(b@xmath12c@xmath13)@xmath0 wire as a tip touching softly a bulk piece of electrochemically cleaned copper .
s in both series of substitutions have been determined from the resistive transitions and also from the temperature dependences of the point - contact spectra .
the transition temperatures found by both methods were essentially the same , indicating that the information obtained from a more surface sensitive point - contact technique is relevant also for the bulk . the point - contact spectrum measured on the ballistic microconstriction between a normal metal ( n ) and a superconductor ( s ) consist of the andreev reflection ( ar ) contribution and the tunneling contribution @xcite .
the charge transfer through a barrierless metallic point - contact is realized via the andreev reflection of carriers .
consequently at @xmath280 the pc current as well as the pc conductance inside the gap voltage ( @xmath29 ) is twice higher than the respective values at higher energies ( @xmath30 ) .
the presence of the tunneling barrier reduces the conductance at the zero bias and two symmetrically located peaks rise at the gap energy . the evolution of the point - contact spectra between the pure andreev reflection and the giaver - like tunnelling case has been theoretically described by the blonder - tinkham - klapwijk ( btk ) theory @xcite .
the point - contact conductance data can be compared with this theory using as input parameters the energy gap @xmath31 , the parameter @xmath32 ( measure for the strength of the interface barrier ) and a parameter @xmath33 for the spectral broadening @xcite . in any case
the voltage dependence of the n / s point - contact conductance gives direct spectroscopic information on the superconducting order parameter @xmath31 .
for the two - gap mgb@xmath0 superconductor the point - contact conductance @xmath34 can be expressed as a weighted sum of two partial btk conductances : those from the quasi two - dimensional @xmath3-band ( with a large gap @xmath35 ) and those from the 3d @xmath2-band ( with a small gap @xmath6 ) @xmath36 the weight factor @xmath37 for the @xmath38 - band contribution can vary from 0.6 for the point - contact current strictly in the mgb@xmath0 @xmath9-plane to 0.99 of @xmath10-axis direction @xcite .
a large number of the point - contact measurements have been performed on the tested samples of pure mgb@xmath0 and both substituted series .
the point contacts revealed different barrier transparencies changing from more metallic interfaces to intermediate barriers with @xmath39 for the aluminium doped serie and @xmath40 for the c - doped samples . by trial and error
we looked for the spectra showing both gaps of a polycrystalline specimen . for more detailed studies we chose those junctions revealing the spectra with a low spectral broadening @xmath33 ,
i.e. with an intensive signal in the normalized pc conductance . as a result we succeeded to obtain many point contact spectra with well resolved two superconducting gaps directly in the row data for all of the studied substitutions except for the highest dopings , mg@xmath19al@xmath20b@xmath0 and mg(b@xmath21c@xmath22)@xmath0 .
figure 1 displays the representative spectra of the aluminium doped serie and fig .
2 resumes the results for the carbon doped mgb@xmath0 .
the experimental data are presented by the full lines , while the fits to the two - gap formula are indicated by the open circles .
all shown point - contact spectra have been normalized to the spectra measured in the normal state at a temperature above @xmath5 .
the upper curves are shifted for the clarity .
the retention of two gaps for all dopings apart from the highest substitution is evident .
for the highest dopings ( always with a larger spectral broadening , @xmath41 ) , the spectra in figs . 1 and 2 reveal only one pair of peaks without an apparent shoulder at the expected position of the second gap . for those data measured at 1.6 k ( the rest of tha data was measured at 4.2 k ) also the one - gap fit is presented .
the one - gap fit can reproduce either the height of the peaks or the width of them , while the two - gap fit can reproduce both .
the size of the apparent gap is well indicated by the peak position , which is about 2 mev for mg@xmath19al@xmath20b@xmath0 and 1.6 mev for mg(b@xmath21c@xmath22)@xmath0 .
but , this size is too small to explain the superconductivity with the respective @xmath5 s equal to 23.5 k and 22 k within the single - gap bcs scenario suggesting , that we are dealing with a smeared two - gap structure . in the case of the 10% c - doped mgb@xmath0 an existence of the large gap was clearly shown by the specific heat mesurements @xcite .
moreover , an existence of the both gaps was strongly evidenced in a single experiment by our previous point - contact spectroscopy measurements in applied magnetic field @xcite . there , the spectrum at @xmath42 = 0.7 t showed that the contribution from the @xmath2 band was partially suppressed and the large - gap shoulder clearly appeared near 3 mv besides the gap peak at 1.6 mv .
a similar effect has the magnetic field in the case of 20 % al doped mgb@xmath0 : as shown in the top panel of fig .
5 , the peak position being at about 2 mv in zero field , is shifted towards 3 mv at 1 tesla . in the case of a single - gap superconductor
the application of magnetic field can only lead to a shrinkage of the distance between the peak positions in the point - contact spectrum .
it is a simple consequence of an introduction of the vortices and magnetic pair - breaking @xcite .
the shift of the peak of the pc spectra to higher voltages demonstrated in fig .
5 is then due to an interplay between the two gaps , because the dominance of the @xmath6 peak at zero field is suppressed in increasing field and @xmath7 contributes more .
thus , the retention of two gaps is proved in the all presented doped mgb@xmath0 samples with @xmath5 s from 39 down to 22 k. al@xmath13b@xmath0 with @xmath43 0 , 0.1 and 0.2 - solid lines .
symbols show the fitting curves : open circles for the two - gap btk model and solid triangles for the one - gap btk model .
the spectra are vertically shifted for the clarity .
the dashed vertical arrows emphasizes the tendency of the evolution of the gaps.,width=7 ] the statistics of the energy gaps obtained from fitting to the two - band btk formula is shown in fig .
3 , for each c and al concentration .
the left coordinates indicate the transition temperature of the junctions , while the right one counts the number of junctions with the same size of the gaps .
the energy width of a particular count indicates the fitting uncertainty .
all the samples , the undoped mgb@xmath0 as well as the aluminium and carbon doped material reveal a certain distribution of the small and large gaps but the two gaps are well distinguishable in the histogram and no overlap of @xmath7 and @xmath6 is observed .
already from this figure a general tendency is evident , namely that the reduction of the large gap is proportional to the decreasing transition temperature . on the other hand
the changes in the small gap are proportionally smaller .
c@xmath16)@xmath0 with @xmath44 0 , 0.021 , 0.038 , 0.055 , 0.065 and 0.1 - solid lines .
symbols show the fitting curves : open circles for the two - gap btk model and solid triangles for the one - gap btk model .
the spectra are vertically shifted for the clarity .
the dashed vertical arrows emphasizes the tendency of the evolution of the gaps.,width=8 ] figure 4 displays the energy gaps as a function of @xmath5 s .
the points ( open squares for al dopings and solid circles for c ones ) are positioned at the averaged energies of the gap distributions from fig . 3 and the error bars represent the standard deviations .
the large gap @xmath7 is essentially decreased linearly with the respective @xmath5 s .
the behavior of @xmath6 s is more complicated . for both kind of substitutions
the gap is almost unchanged at smaller dopings . in the case of c
- doped mgb@xmath0 it holds clearly down to @xmath45 33 k. in ( mg , al)b@xmath0 @xmath6 of the 10 % al doped sample seems to even slightly increase .
but for the highest dopings @xmath6 decreases in both doping cases . the dashed lines in fig .
4 are calculations of kortus _
@xcite for the case of a pure band filling effect and no interband scattering .
although , the constant @xmath6 is not reproduced , the @xmath5 dependence of @xmath31 s for carbon doping are broadly well described .
the solid lines show the kortus s calculations including the interband scattering with the rate @xmath46@xmath47 ( or @xmath48@xmath47 ) . as can be seen the gaps in the al doped samples
can not be accounted without an interband scattering but it is smaller than in the presented calculations .
recently , similar results on the superconducting energy gaps have been obtained on the al - doped single crystals by some of us and other authors @xcite exploring the specific heat as well as the point contact spectroscopy measurements .
those data also show a tendency of approaching of the two gaps with their possible merging in the al - doped samples with @xmath5 below 10 - 15 k. it is worth noticing that those and here presented results do not show any decrease of the large gap below the canonical bcs value for the al - doped mgb@xmath0 with @xmath5 s below 30 k found earlier @xcite . for the carbon doped samples we have not found any stronger tendency to blend both gaps down to @xmath49k , being in contrast with the data obtained on single crystals by gonnelli _
_ @xcite . on the other hand
our data are very compatible with the recent arpes measurements of tsuda _ et al . _ where both gaps have been directly seen in the row data down to @xmath5 of 23 k. a presence of two gaps is also evidenced by the point contact measurements of schmidt _ et al . _
@xcite on the heavily carbon doped mgb@xmath0 .
we remark that the data showing no merging of the gaps on the heavily doped mgb@xmath0 with @xmath5 down to 22 k have been collected from the measurements on the samples of different forms ( wires , sintered pellets , polycrystals ) prepared by different methods .
moreover it is very important that two gaps have been directly experimentally evidenced in the row data without any dependece on the model or fit .
the theoretical calculations of erwin and mazin @xcite on merging of the gaps due to a substitution are also supporting the picture presented here . by them the carbon substitution on the boron site should have zero effect on the merging of the gaps .
it is due to the fact that a replacing boron by carbon does not change the local point symmetry in the @xmath2 and @xmath3 orbitals which are both centered at the boron sites .
then , the interband scattering is of little probability without incorporating extra defects producing a change of the symmetry of orbitals . on the other hand much bigger effect
is expected for the out of plane substitutions ( al instead of mg ) or defects which would indeed change the local point symmetry .
the substitution of al instead of mg leads also to a significant decrease of the @xmath10-lattice parameter @xcite ( it is basically nullite in case of carbon doping ) , which helps the interlayer hopping from a @xmath50 orbital ( @xmath2-band ) in one atomic layer to a @xmath3 bond orbital in the next one .
of coarse , a particular strength of the interband scattering can scatter among samples due to deffect and thus the merging of two gaps could be sample dependent . in the following we present the effect of magnetic field on the point contact spectra of the aluminium as well as carbon doped mgb@xmath0 samples .
the field effect has been measured on a number of samples of each substitution .
the field was always oriented perpendicularly to the sample surface and parallel with the tip having the point contact area in the vortex state .
figure 5 displays the typical magnetic field dependences of the pc spectra ( lines ) obtained on mg@xmath51al@xmath13b@xmath0 samples with x = 0 , 0.1 and 0.2 .
the zero - field spectrum of the undoped mgb@xmath0 ( bottom panel ) reveals an expressive two - gap structure . at small fields
the peak corresponding to the small gap @xmath6 is very rapidly suppressed and only the peak responsible for the large gap @xmath7 is apparently visible above 1 tesla . at 10 % of al doping
the zero field spectra also show both gaps but the large one is now revealed more as a pronounced shoulder .
again at a small magnetic field the @xmath2-band - gap peak is rapidly suppressed but its signature is still distinguishable at 0.5 tesla . at higher fields the signatures of both gaps
are not resolvable due to an interference of the related peaks placed at closed positions .
the resulting maximum is located in the intermedium position between them . at the highest doping ( 20 @xmath52 al )
such an interference of even closer gaps is graduated , so already at zero magnetic field one can see only one pair of peaks at the position of the small gap ( it has always much higher weight at zero field ) and a minor shoulder at the voltage of the large one . in the increasing field
the peak is broadened and its position shifted to higher voltages ( indicated by arrows ) . as mentioned above the shift is an indication of two gaps present in the spectrum with an increasing weight of the large - gap peak with respect to the small one .
the in - field measurements have also been performed on the carbon doped samples . in our previous paper @xcite
they have been demonstrated on the heavily doped , 10 % c mgb@xmath0 sample . in two sets of measurements and subsequent analysis on the aluminum and carbon doped samples
two different types of undoped mgb@xmath0 samples have been used .
the carbon doping increases the upper critical fields @xcite despite the decreased @xmath5 and both superconducting gaps .
thus , these samples are driven to the dirty limit by doping .
that is why we compare the 6.5 % and 10 % c - doped samples with the undoped mgb@xmath0 crystal being already in a dirty limit ( @xmath53 k , @xmath54 t ) @xcite . on the other hand
the aluminum doping leads only to a decrease of @xmath11 for the field in @xmath9 planes or a constant value in perpendicular orientation with @xmath55 t @xcite , so we have chosen the pure mgb@xmath0 sample with @xmath56 k and @xmath57 t as a reference . from the in - field spectra
the excess current @xmath58 has been determined .
it was calculated by integrating the pc spectra after subtracting an area below unity .
when a single gap superconductor is in the vortex state , each vortex core represents a normal state region and this region /
vortex density increases linearly up to the upper critical magnetic field .
thus , the pc excess current ( @xmath59 ) is reduced linearly in increasing magnetic field by the factor of @xmath60 , where @xmath61 represents a normal state region - the pc junction area covered by the vortex cores .
the field dependences of the excess currents @xmath58 obtained on the carbon and aluminum doped samples are shown in fig .
6a and 6b , respectively .
the magnetic field coordinates have been normalized to an appropriate @xmath62 referred to as @xmath63 @xcite . in a none case the linear decrease of @xmath58 was observed . on the contrary ,
a strong non - linearity observed in all curves approves a presence of two gaps in the quasiparticle spectrum of the material .
there , a more rapid fall of @xmath64 at low magnetic fields is ascribed to a strong filling of the @xmath2-band gap states up to a virtual / crossover @xmath2-band upper critical field @xmath65 .
then , at higher fields , the low temperature superconductivity is maintained mainly by the @xmath3-band @xcite , with the @xmath3 gap getting filled smoothly up to the real upper critical field of the material @xmath66 .
due to a small , but finite coupling of the two bands , the superconductivity is maintained also in the @xmath2-band above @xmath65 , but , as shown e.g. in the tunneling data of eskildsen _ et al . _
@xcite , the @xmath2-gap states are filled much more slowly than in the low field region . a magnitude of the upper critical fields of two bands can be estimated in the clean limit @xcite as @xmath67 and @xmath68 , with @xmath69 or @xmath70 for the fermi velocity of the respective band ( since only @xmath62 is considered the anisotropy due to an effective mass tensor is neglected ) .
a similar estimates can be done in the dirty limit when taking into account the respecive diffussion coefficients @xmath71 .
6a shows that the carbon doping leads to changes in the field dependence of the excess current .
the undoped mgb@xmath0 reveals a very steep fall of @xmath58 already at small fields indicating also proportionally small crossover field @xmath65 . by doping the @xmath2-band crossover field
apparently shifts to higher values . from the previous subsection
we know that the carbon doping does not influence significantly the interband scattering in mgb@xmath0 .
the ratio between the two gaps , @xmath72 remains almost unchanged .
then , following from the dirty limit theory the shift of the normalized crossover field @xmath73 is due to an increase of the ratio of the diffussion coeficients @xmath74 .
this is an indication that the carbon doping leads to a significant enhancement of the @xmath2-band scattering . in fig .
6b the excess currents on the aluminum doped samples show different picture : there is no change in the field dependence of @xmath58 upon the aluminum doping up to 20 % indicating constant value of the crossover field @xmath65 .
qualitatively , it can be understood in the framework of the clean limit persisting in the al - doped mgb@xmath0 up to 20 % al doping region .
indeed , the constant value of the upper critical field @xmath62 , which is equal to approximately 3 tesla for all three dopings , indicates the constant ratio between @xmath7 and @xmath70 .
it means that the electron filling of the hole @xmath3-band decreases both quantities proportionally .
then , @xmath75 must be only weakly doping dependent . as @xmath76 and @xmath6 remains almost constant for x
= 0.1 and 0.2 this would suggest that @xmath77 remains also unchanged .
quantitatively , @xmath58 of the two - gap superconductor in the mixed state can be described by the sum @xcite @xmath78 for the point - contact conductance a similar expression can be written @xmath79g_{\pi } + ( 1-\alpha)[n_{\sigma}+(1-n_{\sigma})]g_{\sigma}.\ ] ] considering that @xmath80 and @xmath81 represent independent fillings of the respective gaps in increasing applied field , their values can be identified with the values of the zero - energy density of states ( dos ) @xmath82 and @xmath83 averaged over the vortex lattice .
this model proposed by bugoslavsky _
_ @xcite enables a direct experimental determination of the field dependent energy gaps and zero - energy dos @xmath84 from the point - contact spectra in mgb@xmath0 .
the calculations of koshelev and golubov @xcite show that the low field slope of @xmath82 is controlled by an important parameter @xmath74 which can thus be established .
unfortunately the fitting of the pc conductance of mgb@xmath0 in the mixed state is not trivial because of too many ( 9 ) parameters . moreover
two pairs of the parameters @xmath85 vs. @xmath86 and @xmath87 vs. @xmath88 manifest very similarly in the fit . hence , before fitting the spectra by the formula ( 3 ) we estimate @xmath82 values from the excess currents . here
some simplifications were made : if the real upper critical field @xmath11 is large compared to @xmath65 , in the region @xmath89 the suppression of the @xmath3 band contribution by the field can be neglected ( @xmath90 .
if we also neglect a small reduction of the energy gaps at these fields @xcite , then @xmath91 can be represented by a single parameter formula expressed as @xmath92 since all other parameters ( @xmath93 , @xmath94 and @xmath37 ) have already been determined from the btk fit at @xmath95 .
the resulting @xmath82 is later used as a first approximation in the fit of the pc conductance spectrum . there , also
the values of @xmath96 and @xmath97 determined from btk fits at @xmath98 are kept unchanged . in the second step
all parameters except for @xmath99 and @xmath37 are adjusted for the best fit .
the resulting values of @xmath82 are decreased by 5 - 20% in comparison with the first estimate from @xmath91 @xcite .
also the values of the smearing parameters @xmath100 and @xmath97 reveal about @xmath101 of increase in the interval @xmath102 t accounting for the magnetic pair - breaking of the superconducing dos . in the same field interval a small @xmath103 ) suppression of the values of the energy gaps has been obtained .
the resulting zero - energy dos @xmath82 and @xmath83 are plotted in fig.6c for the carbon - doped samples and in fig.6d for the al - substituted mgb@xmath0 .
again , the carbon and aluminium dopings reveal very differently . the steep increase of @xmath104 in the undoped mgb@xmath0 can be ascribed within the framework of the dirty superconductivity model @xcite to a small value of the ratio of in - plane diffusivities @xmath105 ( see dotted line , calculated for @xmath106 ) .
it means a dirtier @xmath3-band than the @xmath2-one .
the subsequent decrease of the low field slope of @xmath104 upon c - doping can be explained by an increase of the @xmath107 ratio , i.e. by a more rapid enhancement of the @xmath2-band intraband scattering as compared with the scattering in the @xmath3-band .
the dashed line plots the theoretical @xmath104 dependence at @xmath108 the situation is different for the al - doping . here
the field dependences of @xmath104 reveal no change upon doping .
the same explanation as used above for the excess current versus field can be applied , underlying the conservation of the superconducting clean limit for al doping .
the influence of carbon and aluminium doping on the two - band / two - gap superconductivity in mg@xmath109 has been studied by the point - contact spectroscopy .
it has been shown , that the decreased transition temperatures and the evolution of the gaps upon both substitutions is mainly a consequence of the band filling effect . however , in the case of an increased al - doping also an increase of the interband scattering has to be taken into account . by the analysis of the field dependences of the point - contact spectra we have shown , that the c - doping increases the scattering in the @xmath2-band more rapidly then in the @xmath3 one . on the other hand , the al - doping does not introduce significant changes in the relative weight of the scatterings within two bands .
this work has been supported by the slovak science and technology assistance agency under contract no.apvt-51-016604 .
centre of low temperature physics is operated as the centre of excellence of the slovak academy of sciences under contract no .
ames laboratory is operated for the u.s .
department of energy by iowa state university under contract no .
this work was supported by the director for energy research , office of basic energy sciences .
the liquid nitrogen for the experiment has been sponsored by the u.s .
steel koice , s.r.o .
z. hoanov , p. szab , p. samuely , p. c. canfield , and s. l. budko , phys . rev .
b * 70 * , 064520 ( 2004 ) ; p.
samuely , z. hoanov , p.szab , j. kamark , p. c. canfield and s. l. budko , phys .
b * 68 * , 020505 ( r ) ( 2003 ) .
due to measurements on polycrystals the field orientation is not fully controlled .
but in the present study we are interested mainly in the behavior of the isotropic @xmath2 band , thus the choice of the crystallographic orientation of the `` reference '' upper critical field is arbitrary but once chosen it must be kept for all compared doped samples . | the two - band / two - gap superconductivity in aluminium and carbon doped mgb@xmath0 has been addressed by the point - contact spectroscopy .
two gaps are preserved in all samples with @xmath1 down to 22 k. the evolution of two gaps as a function of the critical temperature in the doped systems suggest the dominance of the band - filling effects but for the increased al - doping the enhanced interband scattering approaching two gaps must be considered .
the magnetic field dependences of the andreev reflection excess currents as well as zero - energy density of states determined from the experimental data are used to analyze the intraband scattering .
it is shown , that while the c - doping increases the intraband scattering in the @xmath2-band more rapidly then in the @xmath3 band , the al - doping does not change their relative weight . |
large - scale studies of nearby molecular clouds provide the ability to study the environment of the earliest stages of star formation .
the `` from molecular cores to planet forming disks '' ( c2d ) _ spitzer _ legacy project @xcite used the _
spitzer space telescope _ @xcite to map five large molecular clouds with the infrared array camera ( irac ; @xcite ) and the multi - band imaging photometer for _ spitzer _ ( mips ; @xcite ) including serpens , chamaeleon ii , lupus , ophiuchus , and perseus .
the c2d team s data and results for all five large clouds were compared and summarized in @xcite , hereafter e09 .
we present a study of the complete irac and mips c2d maps of the perseus molecular cloud .
our work synthesizes the data sets first presented by @xcite and @xcite putting the full _ spitzer _ observations in context of more recent work on perseus .
this paper gives an overview of low - mass star formation across the varied environments in perseus from 3.6 to 70 .
we provide an updated list of young stellar objects ( ysos ) with newly corrected photometry , spectral indices , and classifications .
perseus is a well - known and a much - studied cloud ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) because it presents a range of star formation environments .
perseus contains two young clusters , ic 348 and ngc 1333 , as well as several small dense clumps of the type that often produce one or a few stars ( b1 , b5 , l1448 , and l1455 ) .
perseus is near the middle of the spectrum of star formation in large clouds that spans from a low density ( e.g. , chamaeleon ii ) to a high density ( e.g. , serpens ) of young stellar objects ( @xcite ) .
perseus is not currently forming very high - mass ( mid - b or earlier spectral type ) stars ; although , ic 348 is known to contain a b5 v star ( bd+31634 ) . therefore , perseus has been called a `` typical '' cloud in the study of the early phases of low - mass star formation @xcite .
perseus also contains protostellar cores , very low luminosity objects ( vellos ) , and potentially very young ysos that are too deeply embedded to have been detected by _
spitzer_. six candidate first hydrostatic cores have been identified in perseus ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite identify numerous low - luminosity embedded protostars and as well as vellos in the cloud .
enoch et al .
@xcite combined the _
map with a 1.1 mm bolocam dust continuum map of perseus from the _ caltech submillimeter observatory _ ( cso ) .
they discuss the properties of prestellar @xcite and the class 0 and i cores @xcite identified with the _ spitzer _ and bolocam data .
perseus was also mapped in the submillimeter using the submillimeter common user bolometer array ( scuba ) at the _ james clerk maxwell telescope _
( jcmt ; * ? ? ?
* ; * ? ? ?
discussion of the _ spitzer _ sources in relation to the scuba cores can be found in @xcite and @xcite .
more recently , @xcite identified 28 candidate class 0 sources in perseus using the _ herschel space observatory_. four of these sources had not been previously identified with _
the irac and mips data sets for perseus were analyzed individually by @xcite and @xcite respectively .
this is the 12th paper in a series presenting original _
data of the five large molecular clouds mapped by the c2d team .
we adopt the same distance of 250 @xmath2 50 pc for perseus @xcite for consistency with previous work by the c2d team . however , very long baseline interferometry ( vlbi ) studies of h@xmath3o masers in nearby star - forming regions give a slightly closer distance to the western end of perseus : 235 @xmath2 18 pc for ngc 1333 @xcite and 232 @xmath2 18 pc to l1448 @xcite .
we follow the same data reduction and analysis procedures laid out in previous papers in the series and briefly described in the following sections .
we have updated the c2d list of ysos in perseus , providing new extinction corrected photometry , and discuss the classification and distribution of ysos both within and outside the clusters in the context of the extended emission and visual extinction in perseus .
the perseus molecular cloud was mapped with two of the _ spitzer space telescope s _ instruments , irac and mips .
figure [ fig : color ] is a three - color composite map of perseus from the irac 4.5 and 8 @xmath4 m and mips 24 @xmath4 m data .
the _ spitzer _
maps cover 3.86 square degrees in the irac observation wavelengths of 3.6 , 4.5 , 5.8 , and 8.0 @xmath4 m and approximately 10.5 square degrees with mips at 24 , 70 , and 160 @xmath4 m .
however , due to the nature of the detector , the 160 maps are not complete and provide only limited photometry for a portion of the ysos .
therefore , our discussion will be limited to data from 3.6 to 70 @xmath4 m .
rebull et al .
( 2007 ) discussed the 160 mosaic . in figure
[ fig : color ] , the edges of the image are the mips scan boundaries and the irac observation area is denoted by the dashed lines .
we will focus on the area where the two maps overlap , namely the nearly 4 square degrees observed by irac .
the outline of the perseus mapped region was selected using the a@xmath0 = 2 mag contour from ^13^co extinction maps of @xcite .
the c2d team chose this criteria in order to observe the regions of densest gas and dust in the cloud complex and , by association , observe all of the young and forming stars in perseus .
a correlation also exists between ^13^co and near - infrared dust extinction @xcite .
@xcite showed that the ^13^co column density is correlated with visual extinction ( a@xmath0 ) in perseus for a@xmath0 @xmath5 10 mag .
however , @xcite concluded that , while ^13^co may be a good tracer of gas surface density , it is not linearly correlated with visual extinction in perseus when considering a@xmath0 up to 25 mag and , using standard conversion factors , may underestimate mass .
the irac maps fit tightly to the a@xmath0 = 2 mag outline , whereas mips was able to scan over a much larger area in less observing time due to the design of the instrument .
ngc 1333 and ic 348 were mapped separately from the c2d map in conjunction with _ spitzer _ guaranteed time observers program ( gtos ; * ? ? ?
* ; * ? ? ?
the gto maps have been combined with the larger c2d maps and are included in this analysis .
the irac and mips observations of perseus were discussed in detail in @xcite and @xcite respectively .
we only discuss pertinent details of the observations here .
each instrument mapped perseus twice in separate time epochs anywhere from several hours to 7 months apart , depending on the region , to identify and eliminate asteroids and artifacts .
the raw data were processed through the c2d data reduction pipeline to create mosaic maps , reduce instrumental effects , and produce a band - merged source catalog for perseus that includes both two micron all sky survey ( 2mass ) and _ spitzer _ photometry .
the fluxes given in the c2d catalog at each irac and mips band are the average of the single - epoch fluxes @xcite .
the c2d catalogs for perseus as well as the other molecular clouds and cores included in the survey are available online through the nasa / ipac infrared science archive @xcite . a complete description of the data processing and source extraction for all of the c2d data can be found in the first papers in this series ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and the final c2d data delivery document @xcite .
figure [ fig : fluxratio ] plots the ratio of observed fluxes at 3.6 @xmath4 m ( irac band 1 ) between the two observation epochs versus the 3.6 @xmath4 m flux derived from the combined epochs for all sources in the cloud with a single epoch flux to uncertainty ratio greater than 10 .
the flux measurements are very consistent , differing by less than 15% , between the epochs , with an average flux ratio of @xmath6 @xmath7 = -0.0032 $ ] .
figure [ fig : fluxratio ] shows a slightly larger spread in the flux ratios ( @xmath8 @xmath6 @xmath7 < 0.2 $ ] ) for the faintest sources ( @xmath9 mjy ) .
the few sources with more extreme flux ratios between epochs were mostly flagged in the data processing as extended , double , or with a signal - to - noise ratio that was too small to determine shape ( imtype = 2 , 3 , or 7 ; * ? ? ?
these imtypes are indicated as x s in figure [ fig : fluxratio ] , where one can see that the outliers fall mostly in these categories .
three objects classified by c2d as young stellar objects ( ysos ; see section 3 ) have a large ratio of 3.6 fluxes between the first and second observations ( triangle outliers in figure[fig : fluxratio ] ) .
the variation in flux between the two observation epochs for these objects is likely not physical .
one yso in the ic 348 region was flagged with an uncertainty ratio too low to accurately determine shape ( index number 348 in table [ tab:348 ] ) .
another yso in ic 348 ( 424 in table [ tab:348 ] ) has a relatively low flux at 3.6 , @xmath6 @xmath10 , placing it at the edge of the spread in figure [ fig : fluxratio ] .
this object is also situated in extended nebulosity near a source with a large 3.6 flux .
the third yso outlier ( 209 in table [ tab:1333 ] ) has a combined epoch flux of 39 mjy at 3.6 and a @xmath6 @xmath7 = 0.213 $ ] for observations 7 months apart .
yso 209 is a class 0+i object ( see section 4 ) located in the ngc 1333 region , but again is potentially confused with bright nebulosity .
a major goal of the c2d program was to identify young stellar objects ( ysos ) in the large molecular clouds .
we follow the method of previous papers in this series for selecting yso candidates in perseus , in which ysos are distinguished from reddened background galaxies using color - color and color - magnitude diagrams in the 2mass ( h and k ) , irac ( 4.5 and 8 ) , and mips ( 24 ) bands .
this method is described here briefly , however a detailed description can be found in the c2d data delivery document @xcite and by @xcite .
the c2d data were compared with a galaxy sample in the elais n1 field from the _ spitzer _ wide - area infrared extragalactic ( swire ) catalog @xcite .
elais n1 is well above the galactic plane and not expected to contain any ysos or molecular clouds .
the swire data was used to eliminate confusion of reddened background galaxies with ysos , however , it does not control for reddened background stars in our own galaxy .
irac and mips maps of the field observed by the swire team were resampled to match the c2d sensitivity and reprocessed in the c2d pipeline so accurate color comparisons could be made .
in addition , the swire data were also reddened using the extinction profile of perseus generated from _ spitzer
_ data ( figure [ fig : avhist ] ) .
therefore , galaxies in the swire field would appear reddened as if behind a perseus - like molecular cloud .
the reprocessed swire data were plotted in infrared color - color and color - magnitude spaces to determine the positions of galaxies in the diagrams .
this process resulted in cutoff limits in each color - color and color - magnitude space to divide ysos from galaxies .
the color - color and color - magnitude diagrams used for yso selection in perseus are in figure [ fig : colormag ] .
the lines in each color - color or color - magnitude space are the cutoffs that define the yso candidate criterion in the various color - magnitude spaces created by comparison with the swire data and are the same as those presented and discussed by harvey et al . @xcite .
the c2d team constructed the cutoffs in the color - magnitude diagrams to be smooth , exponentially decaying probabilities around the lines in figure [ fig : colormag ] so that sources far below the cutoffs are assigned the highest probability of being extragalactic .
the probability of extragalactic contamination is highest in the center of the ellipses in the @xmath11 $ ] versus @xmath12 - [ 24]$ ] diagrams and decreases radially .
perseus sources with infrared colors that were consistent with dust emission rather than extincted background stars were assigned an unnormalized total `` probability '' that the source is extragalactic .
this proxy for probability is formed from the product of the individual probabilities from the color - magnitude diagrams in figure [ fig : colormag ] and factors based on source properties such as @xmath13 $ ] color , mips flux density , and whether it was found to be extended at 3.6 and 4.5 @xcite . the cutoffs for @xmath13 $ ] color
are also shown in figure [ fig : colormag ] .
bluer colors reflect a higher probability that the source is extragalactic . in figures [ fig : avhist ] , [ fig : colormag ] , and [ fig : galhist ] and the following text , we divide perseus into three distinct regions for comparison of the yso properties .
the sources associated with the ic 348 and ngc 1333 clusters are shown separately from the remaining cloud " ( hereafter rc ) , which includes all the other sources , such as b1 and l1455 , associated with the perseus cloud complex .
figure [ fig : color ] displays the boundaries of these two cluster regions in red . the ngc 1333 region was defined from 03^h^28^m^00^s^ to 03^h^30^m^00^s^ in right ascension and + 310600 to + 313000 in declination .
the ic 348 region runs from 03^h^43^m^12^s^ to 03^h^46^m^00^s^ in right ascension and + 314800 to + 322400 in declination .
the rc includes all of the irac observation area ( dashed outline in fig .
1 ) outside of the cluster boundaries .
these three regions were previously defined in @xcite and @xcite based on the boundaries of the gto observations of the clusters @xcite .
the regions do not represent physical boundaries of the clusters , but reflect _
spitzer s _ mapping footprint .
therefore , the ic 348 and ngc 1333 regions may include some ysos that are not physically associated with the clusters .
however , for consistency within this series of papers , we maintain these somewhat arbitrary boundaries .
figure [ fig : galhist ] is a histogram of the number of identified sources with infrared flux versus the `` probability '' of being a background galaxy .
sources with low probabilities were considered yso candidates .
the cutoff to separate ysos and galaxies in terms of probability is @xmath6 p @xmath14 @xmath151.47 ( as in @xcite ) .
there is a small tail of sources found in perseus between @xmath6 p = @xmath151.5 and @xmath152.0 .
harvey et al .
@xcite predicted no more than zero to one contaminant galaxies in this tail based on comparisons with the swire catalog .
most of the extragalactic sources appear in the rc histogram rather than being associated with the young clusters in perseus because of their uniform spacing over the cloud area .
previous papers in this series have reported the number of yso candidates ( ysoc ) in a given c2d cloud . however , in recent years , the c2d ysoc lists have been used and refined in a number of studies ( e.g. , e09 , @xcite ) . in the course of these studies ,
all of the ysoc were visually inspected to eliminate resolved galaxies and artifacts ( e09 ) .
therefore , following the terminology in e09 , the `` candidate '' designation has been dropped for this study and we will discuss the ysos found in perseus . using the described selection method
, we identified 385 ysos in perseus .
irac and mips fluxes for these sources , as well as the ysos in the other c2d clouds , were published in online tables as part of a paper summarizing the c2d _ spitzer _ legacy project results ( e09 ) .
the tables also provide additional sed data , classifications , and derived properties , such as bolometric temperature and luminosity , for all of the perseus ysos .
dunham et al .
@xcite updated the seds and other properties of the protostellar population in perseus .
recently , @xcite developed a new method for separating ysos and galaxies in _ spitzer _ data sets using multi - dimensional ( multi - d ) magnitude space .
they analyzed all of the c2d clouds with this new method , adding 100 yso candidates and removing 16 ysos in perseus compared to the e09 c2d list for a total of 469 ysoc in perseus .
figure [ fig : hlmap ] shows the location of the @xcite new yso candidates on the 8 map of the cloud .
sixteen c2d - identified ysos were not included by @xcite for various reasons ( see also their table 5 ) : 6 are jet knots , 6 appear to be confused with nearby sources or cloud structures , 3 were not identified as high reliability sources by c2d , and one was identified by the multi - d process as a giant star .
we agree with hsieh and lai s characterization of these 16 sources and have removed them from our list of ysos in perseus .
the multi - d method developed for identifying yso candidates by @xcite is relatively new and completely different from the established c2d method presented in previous papers in this series and in other works by the c2d team . because our method ,
as described above , does not identify the 100 additional sources in the @xcite candidate list as ysos and for consistency with the previous work from the c2d team , we do not include the new @xcite yso candidates .
the @xcite candidates likely include previously unidentified ysos in perseus .
however , we suggest further observations , such as optical spectra or maps of the molecular cores , are needed to confirm the classification of these faint sources and rule out possibility that the candidates are reddened agb stars or background galaxies
. therefore , we present 369 ysos identified by the c2d method , 385 from e09 minus 16 excluded by @xcite , in perseus .
e09 calculated the star formation rate ( sfr ) for all of the c2d clouds based on the number of ysos ( n@xmath16 ) identified in each cloud . assuming a mean mass per yso of 0.5 m@xmath17 and a period of star formation of 2 myr , e09 report a sfr in perseus of 96 m@xmath17 myr@xmath18 . with 369 instead of 385 ysos and a slightly lower sfr of 92 m@xmath17 myr@xmath18 , perseus maintains the highest sfr among the c2d team s molecular clouds .
if all of yso candidates from @xcite are included , the sfr in perseus increases to 117 m@xmath17 myr@xmath18 .
figure [ fig : hlmap ] shows the distribution of the 369 ysos identified in perseus .
ysos are found in groups and clusters associated with known regions of star formation .
we list and identify the ysos in each cluster region and the rc .
tables [ tab:348 ] and [ tab:1333 ] list the 143 ysos in the ic 348 region and 104 ysos in the ngc 1333 region respectively .
table [ tab : rc ] identifies 122 ysos in the rc .
all of the ysos identified in the c2d molecular cloud survey with sufficient data were placed in one of four classes according to their spectral index , @xmath19 : class 0+i , flat ( i.e. , flat - spectrum ) , class ii , and class iii .
we have reclassified the ysos in perseus according to an updated spectral index .
therefore , some sources are classified differently in this work than in e09 .
the spectral index is defined by a least squares fit to the photometry of the source between 2 and 24 @xmath4 m according to equation 1 in e09 .
the near - infrared flux data are from 2mass .
we follow the class definitions of @xcite , where @xmath20 0.3 is a class i,@xmath21 is a flat source , @xmath15 1.6 @xmath22 is a class ii , and @xmath23 is a class iii .
because @xcite does not distinguish between class 0 and class i sources and the classes can not be separated based on spectra index ( @xcite ; m. m. dunham et al .
2015 , in preparation ) , we classify all sources with @xmath20 0.3 as class 0+i .
in addition , the @xcite classification scheme does not put a lower limit on @xmath19 for class iii sources and therefore includes ysos that with spectral energy distributions that resemble bare stellar photospheres . as noted by @xcite , _ spitzer _ data alone can not identify diskless class iii sources with no infrared excess . therefore , our class iii census represents a lower limit for perseus .
evans et al .
( 2009 ) corrected the c2d large cloud photometry for extinction . using a mean extinction of a@xmath0 = 5.92 in perseus , e09 classified their 385 yso candidates : 76 class 0+i , 35 flat , 244 class ii , and 30 class iii .
following e09 , we assign extinction values to all ysos as follows : 1 .
we adopt extinction values from the literature for class ii and iii ysos ( classified via infrared spectral index ; see e09 for details ) included in published optical studies .
2 . we de - redden the remaining class ii and iii ysos to the intrinsic near - infrared colors of an assumed spectral type of k7 , found to be fairly representative of the majority of class ii and iii ysos in the c2d clouds ( @xcite ; see also e09 for details ) .
3 . we de - redden all of the class 0+i and flat spectrum ysos ( again classified via infrared spectral index ) using the mean extinction toward all class ii ysos in perseus ( a@xmath0 = 5.9 mag ) .
the mean extinction is used since we can not directly determine the line - of - sight extinction from the cloud for these sources .
once the extinction values are assigned , we use these values combined with the @xcite extinction law for @xmath24 to correct the photometry for extinction .
the choice of the @xmath24 law rather than the @xmath25 law is motivated by several studies showing that the former is more appropriate for the dense regions in which stars form ( e.g. , @xcite ) .
while the original extinction corrections in e09 only used the absorption cross section , here we adopt the sum of the absorption and scattering cross sections in order to correct for the total line - of - sight extinction .
tables [ tab:348 ] - [ tab : rc ] list the absorption and scattering corrected spectral indices and photometry for each yso .
the 369 ysos identified in this work were reclassified according to their newly corrected spectral index .
we find 70 class 0+i , 32 flat , 231 class ii , and 36 class iii sources in perseus .
although the number of sources in each class differs from e09 , the fractions of each class compared to the total number of ysos is not significantly different from those reported by e09 in comparing the five c2d clouds .
perseus has a larger fraction , 19% , of class 0+i sources compared to the other c2d star forming clouds .
@xcite classified their 469 ysoc in perseus using the same class definitions , reporting 99 class 0+i , 49 flat , 272 class ii , and 49 class iii sources ( their table 8) .
they find more ysoc in each category compared to the c2d list , however the fractions of each class remain very similar to this work .
namely , @xcite also find a large fraction , 21% , of class 0+i sources .
table [ tab : class ] gives the fraction of the total number of ysos for each classification .
these observational yso classifications are suggestive of physical stages of star formation .
@xcite used hco^+^ observations to classify class 0+i and flat sources identified in the c2d and gould belt ( m. m. dunham et al .
2015 , in preparation ) _ spitzer _ surveys of molecular clouds as physically embedded stage 0+i @xcite objects that retain envelope material .
they find 72% of the _ spitzer _ class i and 48% of flat sources are also stage 0+i protostars .
yso classification has also been connected to the physical nature of disks around young stars . @xcite and @xcite used @xmath19 derived from irac data to place ysos with optically thick disks and anemic disks into distinct categories .
the thick and thin disk classifications developed by @xcite are very similar to the class ii and class iii definitions of @xcite . using these definitions
, we find 228 ysos with thick disks ( -0.5 @xmath26 -1.8 ) and 29 with anemic disks ( -1.8 @xmath26 -2.56 ) in perseus .
the ic 348 cluster region contains nearly half of these sources ( 126 ) and the majority of the thick disks ( 59% ) in the cloud .
more ysos were identified in perseus than in any of the other clouds studied in the c2d survey .
however , it also covers the most area .
perseus , as previously reported in e09 , has a density of 5.0 ysos per square parsec , placing it behind serpens and ophiuchus in terms of number of ysos per area .
table [ tab : class ] lists the number of ysos per square parsec , assuming a distance of 250 pc , for each region of perseus .
if we adopt the distance found by @xcite of 235 pc for ngc 1333 , the density increases to 5.7 ysos pc@xmath27 .
as noted in section 3 , the ysos in perseus are not evenly distributed across the cloud ( figure [ fig : hlmap ] ) .
instead there are two main sections with star formation activity . to the east ,
lie b5 , ic 348 , and l1468 .
ngc 1333 , b1 , l1448 , and l1455 are on the opposite end of the mapped region ( figure [ fig : regions ] ) .
the two clusters , ic 348 and ngc 1333 , contain 67% of the ysos in perseus .
ngc 1333 has the highest density of ysos , 32 pc@xmath27 .
many of the 122 ysos in the rc are associated with the other known star - forming associations such as b1 ( 22 ) , l1448 ( 5 ) , and l1455 ( 11 ) .
l1455 has the lowest density of all the regions at 1.4 ysos pc@xmath27 .
only two ysos are found in the middle of perseus far from the clusters and known star - forming associations ( see figure [ fig : regions ] ) : index numbers 304 and 305 in table [ tab : rc ] and e09 .
yso 304 is within a few arcmin of the dark cloud l1468 .
however , there is no evidence in the literature of star - formation activity in l1468 .
both ysos are found toward regions where the visual extinction is greater than 4 magnitudes and both are class ii sources .
@xcite identified three additional yso candidates ( their numbers 227 , 228 , & 231 ) in the mid - cloud region .
one of their candidates ( 231 ) is within 2 arcmin of our yso number 305 .
section 6 discusses the mid - cloud region in terms of visual extinction and star formation .
@xcite divided perseus into eastern and western regions and concluded that western perseus ( including ngc 1333 ) has the greatest fraction of embedded sources ( class 0+i ) of any of the c2d molecular clouds .
the _ herschel space observatory _ observed more infrared emission in its shorter wavelength bands ( 70 and 160 ) in eastern perseus ( ic 348 ) than the western end of the cloud indicating that the eastern end is warmer @xcite .
further analysis of the _ herschel _ data showed regions in the western end of the cloud ( ngc 1333 and l1448 ) have the greatest fractions of class 0+i ysos @xcite .
figure [ fig : regions ] shows the locations of ysos according to their classifications overlaid on the 8 irac map . named star - forming associations are labeled and the clusters and larger associations ( b1 and l1455 ) used for analysis of the yso populations in this work are outlined in the figure .
table [ tab : class ] lists the regional number and fraction of ysos for each class in the clusters , rc , b1 , l1455 , and the eastern and western sections of the clouds as well as the ratio of class ii to class 0+i ysos for the region .
the charts in figure [ fig : pie ] are a visual representation of the distribution of ysos by class for regions listed in table [ tab : class ] .
our results agree with those of @xcite and the _ herschel _ studies @xcite . by dividing the perseus cloud into eastern and western sections along a line of right ascension at 3^h^38^m^00^s^ ( fig .
[ fig : regions ] ) , we find three - quarters of the class 0+i and flat ysos in the western section .
the eastern section , an area of approximately 1.6 square degrees , includes ic 348 and contains the majority of the more evolved sources , 61% of class ii and 81% of class iii . assuming a constant birthrate of stars , the ratio of the number of class ii to class 0+i ysos is used as a relative age estimate between star - forming regions @xcite .
class ii sources are more abundant than class 0+i ysos in eastern perseus by a ratio of 9.3 to 1 .
when we analyze the yso population in the clusters , the class ii/(0+i ) ratio is 9.5 in ic 348 but only 2.1 in ngc 1333 , implying ic 348 is older .
@xcite found ratios of 9.0 and 2.7 for ic 348 and ngc 1333 respectively and both the separate irac @xcite and mips @xcite studies of perseus agree that ic 348 contains a larger ratio of class ii to class i sources .
the class ii/(0+i ) ratios are consistent with previous age estimates for the clusters ( ngc 1333 @xmath5 1 myr @xcite and ic 348 approx .
2 myr @xcite ) .
we find more than twice the number of class 0+i and flat sources are associated with ncg 1333 as are associated with ic 348 , as first suggested by the analysis of irac data alone @xcite .
jrgensen et al .
@xcite found that the majority of these more embedded sources were in the smaller star - forming regions in the rc , suggesting that these associations are the sites of the most recent star formation activity in the cloud .
however , when rebull et al .
( 2007 ) analyzed the mips data , they showed that the abundance of bright class i and flat sources is not higher in the rc , suggesting that the jrgensen et al .
@xcite result applies only to faint ysos . with the combined irac and mips data , as well as
the extinction corrected source classifications , we find 40 class 0+i and flat sources in the defined ngc 1333 region and 46 in the rc .
further , we find the rc has a greater number of class 0+i and flat ysos ( 28 ) with mips 24 fluxes greater than 100 mjy compared to ngc 1333 ( 15 ) .
the class ii/(0+i ) ratio in ngc 1333 ( 2.1 ) is nearly equal to that of the rc ( 2.2 ) . based solely on source counts in these regions ,
there is not an obvious implied age difference between ngc 1333 and the rc .
however , individual star - forming associations in the rc do contain higher fractions of class 0+i sources than ngc 1333 , suggesting a younger age for these regions and in agreement with @xcite .
for example , l1448 , at the far western edge of the _ spitzer _ map , contains only embedded ysos , 5 class 0+i sources .
we defined large regions ( 0.4 square degrees , fig .
[ fig : regions ] ) for the largest star - forming associations outside of the clusters , b1 and l1455 .
the b1 area is the same as the scuba 850 map of @xcite .
nearly half ( 45% ) of the ysos in the b1 and l1455 areas are class 0+i sources in contrast with 27% in ngc 1333 .
the ratio of class ii to class 0+i sources is only 0.8 in b1 and l1455 .
however , we caution that even with the broadly defined b1 and l1455 regions the total number of ysos in these regions is relatively small ( 33 ) .
we find class iii sources are most common in ic 348 in agreement with @xcite .
64% are found in the cluster .
the rc contains only a few more class iii sources ( 8) than ngc 1333 ( 5 ) .
as noted previously , this is a lower limit for class iii sources in these regions since our data can not identify young objects with star - like spectral energy distributions .
extensive x - ray observations or spectroscopic surveys of the perseus cloud are needed to find the class iii sources missed by this survey .
@xcite cross - referenced a large number of ngc 1333 catalogs , including the _ spitzer _ catalog , and classified ysos with the combined data sets .
rebull identified more than twice as many ysos ( 277 ) associated with the cluster than this work ( 104 ) .
the rebull catalog encompasses a larger region than we define for ngc 1333 , so some ysos in @xcite are in our rc region .
the ysos in rebull s catalog that we do not identify in this work were generally not detected in most or all of the _ spitzer _ bands and were identified as ysos with submillimeter , spectroscopic , or x - ray data .
@xcite lists a large number of class iii sources in ngc 1333 that were identified as young by the young stellar object variability ( ysovar ) project @xcite with x - ray detections .
all of the ngc 1333 ysos presented here are contained in the @xcite catalog and the classifications are generally consistent .
@xcite classifies ysos using the spectral index definitions from @xcite as in section 4 , however , @xmath19 is defined from the observed sed without any reddening correction .
rebull places 15 ysos in a different class than this work due to slightly greater values of @xmath19 .
for example , 7 of our class ii sources are placed in the flat category by @xcite .
the @xcite fractions of class i and flat sources in the cluster are consistent with this work .
however , rebull finds a much larger fraction of class iii sources in ngc 1333 , 29% , because of the greater number of class iii sources identified with x - ray data .
several studies have placed the large scale nebulous emission in the irac 8 and mips 24 maps ( figs .
[ fig : color ] , [ fig : hlmap ] , and [ fig : avmap ] ) of perseus in the context of star formation and related phenomena .
the c2d team compared the irac and mips extended emission with areas of high extinction @xcite .
they found no clear correlation at shorter wavelengths ; however , the 160 mosaic revealed emission that matched extinction maps of the cloud @xcite .
more recently , the complete ( coordinated molecular probe line extinction and thermal emission ) team used the _
maps of perseus to help identify shells and bubbles in the cloud as part of their co survey @xcite .
they found that some of the arc - like nebulosities seen in the irac and mips maps are related to the edges of shells created by stellar winds .
figure [ fig : avmap ] shows a@xmath0 contours overlaid on the mips 24 map of perseus .
the contours are derived from an extinction map produced by the c2d team using 2mass and _ spitzer _ data .
the visual extinctions toward sources classified as stars by c2d ( see figure [ fig : avhist ] ) were convolved with a gaussian beam to create extinction maps of the entire cloud @xcite .
while there are bright regions of 24 emission at high extinction , many areas of high extinction ( a@xmath0 @xmath28 8 mag ) , such as b1-e , show little to no extended emission .
furthermore , the irac and mips extended emission is not necessarily correlated with star formation in perseus .
most notably , the middle region of perseus , between 3^h^37^m^00^s^ to 3^h^41^m^00^s^ , contains only two class ii ysos ( see section 5 ) despite bright extended emission at 8 and 24 ( figures [ fig : hlmap ] and [ fig : avmap ] ) .
co and infrared maps show extended emission and some areas of high extinction in this middle region ( e.g. , * ? ? ?
the region also contains dark clouds such as l1468 and b1-e ( figure [ fig : regions ] ) .
however , there has been little evidence found of star formation , suggesting that the structure in the middle region is perhaps not filamentary enough to form stars @xcite .
the complete survey highlights the lack of ^13^co in the mid - region of perseus as corresponding to a shell , cps 5 , created by a nearby o or b star @xcite .
the size and location of cps 5 is indicated on figure [ fig : avmap ] against the 24 image of perseus .
@xcite mapped perseus at 1.1 mm and found no dense clumps in the middle of perseus .
sadavoy et al .
( 2012 ) studied the b1-e association , located to the east of ngc 1333 and b1 , toward the midsection of the cloud as shown in figure [ fig : color ] , as part of the _ herschel space observatory _
gould belt survey and found it to contain dense cores but no young stars .
figure [ fig : avmap ] shows high extinction ( mean a@xmath0 @xmath29 10 mag ) toward b1-e , however , we do not find any ysos in the region .
@xcite , did report one yso candidate in the region ( their 227 ) using the c2d data .
it has been further postulated that perseus may consist of two separate clouds that are not physically associated ( e.g. , * ? ? ?
* ; * ? ? ?
* ) or possibly a chain of clouds at a range of distances ( 250 to 350 pc ) along the line of sight @xcite .
if this is the case , the central region of perseus as seen from earth may be a physical gap between clouds .
these scenarios would naturally explain the lack of ysos in the mid - regions of perseus .
however , following the lead of earlier papers in this series and other recent surveys of perseus mentioned here , we continue to assume the entire mapped region to be associated and at the same distance , while acknowledging the complexities of such a large cloud .
histograms of visual extinction ( a@xmath0 ) towards ic 348 , ngc 1333 , the rc , and all of perseus using stars observed with irac and mips are presented in figure [ fig : avhist ] .
the a@xmath0 was measured along lines of sight though the cloud using the 2mass and _ spitzer _ spectral energy distribution of sources classified as stars and adopting the @xcite extinction law , r@xmath0 = 5.5 mag @xcite .
the histograms are presented as a fractional number of stars versus a@xmath0 . in this work , we limited our data collection by artificially setting the edge of the cloud at the a@xmath0 = 2 mag contour .
the histograms peak near a@xmath0 = 5 mag and have a tail to higher extinction .
many authors @xcite have presented extinction histograms for perseus using 2mass , iras , and _ herschel _ infrared , as well as co , data in comparison to log - normal probability distribution functions .
log - normal distributions at low column densities suggest the importance of turbulence in the cloud structure ( e.g. , @xcite ) .
several of the studies ( e.g. , @xcite ) also show tails above log - normal distributions at higher a@xmath0 ( @xmath1 2 - 5 mag ) , implying that perhaps self - gravity becomes more important as clouds begin to form stars @xcite .
because of the a@xmath0 = 2 mag cutoff in our data , the histograms in figure [ fig : avhist ] represent mostly the higher a@xmath0 tail of the distribution .
we find mean a@xmath0 values of 10.3 , 11.7 , 15.5 , and 10.1 mag for all of perseus , ic 348 , ngc 1333 , and the rc respectively .
the histogram of extinction for ngc 1333 ( green in fig .
[ fig : avhist ] ) peaks at a higher a@xmath0 of 7.5 mag , has a larger mean a@xmath0 , and a larger tail to high extinction than the other regions .
these features indicate different physical conditions in ngc 1333 and ic 348 @xcite and point to the recent star formation history of ngc 1333 .
recent studies of molecular clouds , including perseus , have suggested an extinction threshold for low - mass star formation ( e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ; see also section 8 of the recent review by @xcite ) .
this threshold ranges from a@xmath0 = 4 - 7 mag , below which very few or no dense cores are found .
@xcite found submillimeter clumps in perseus only where a@xmath0 @xmath1 5 - 7 mag and suggested a threshold of a@xmath0 = 5 mag linked to magnetic support of the clump , where there is only enough time for ambipolar diffusion if extinction is near the threshold extinction .
figure [ fig : avmap ] shows our classified ysos as well as a@xmath0 = 4 , 6 , and 8 mag contours from the c2d extinction map ( see section 6 ) on the 24 mosaic of perseus
. using the background star extinctions averaged over 270 as in @xcite , we determined the a@xmath0 at the source position for each class 0+i and flat yso , since these extinctions can not be determined directly .
extinction values for class ii and iii sources were previously collected from the literature in @xcite .
56% of all the ysos in perseus are found toward an a@xmath0 @xmath1 5 mag .
the distribution of the class 0+i and flat ysos , those sources that are least likely to have dispersed from their place of origin , as a function of a@xmath0 is shown in figure [ fig : ysohist ] .
91% of the class 0+i and flat sources are found toward a@xmath0 @xmath1 5 mag .
only 3 class 0+i and 6 flat ysos are found at a@xmath0 @xmath5 5 mag .
@xcite report 75% of 535 class 0+i and flat sources in the c2d plus gould belt sample ( m. m. dunham et al .
2015 , in preparation ) , including the perseus ysos in this work , are found toward a@xmath0 @xmath28 8 mag .
we find 85% of the perseus class 0+i and flat sources are found toward a@xmath0 @xmath28 8 mag .
these results agree with studies identifying dense cores only in regions of high extinction in perseus @xcite and support the idea of a threshold extinction near a@xmath0 = 5 - 8 mag .
we present irac and mips data from 3.6 to 70 of the perseus molecular cloud and give a broad overview of the current star formation in the region .
perseus contains 369 ysos and has a density of 5 ysos per square parsec .
tables [ tab:348 ] - [ tab : rc ] list updated extinction corrected photometry for the ysos including both absorption and scattering cross sections .
while perseus has a large fraction of embedded class 0+i sources ( 19% ) , 63% of the objects in the cloud are classified as class ii ysos .
perseus contains a variety of star forming environments .
ic 348 and the eastern end of the cloud are shrouded in extended nebulosity at the longer _
spitzer _ wavelengths .
however , the ic 348 region contains fewer embedded sources and the majority of the class ii ( 73% ) and iii ( 64% ) objects in perseus , consistent with previous age estimates of 2 myr @xcite . the ngc 1333 cluster , suggested to be @xmath5 1 myr @xcite ,
contains more than twice the number of class 0+i and flat sources than ic 348 . the ratios of class ii to class 0+i ysos ( 9.5 in ic 348 and 2.1 in ngc 1333 ) are also consistent with the relative age estimates of the clusters .
we find that 56% of all the ysos and 91% of the class 0+i and flat spectrum sources in perseus are in regions where the a@xmath0 @xmath1 5 mag , consistent with an extinction threshold for early stage star formation suggested by other authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , the mid - cloud region is nearly devoid of star formation activity despite regions where the visual extinction ranges from a@xmath0 = 5 - 20 mag .
support of this work , part of the _ spitzer _ legacy science program , was provided by nasa through contracts 1224608 , 1230782 , and 1230779 issued by the jet propulsion laboratory , california institute of technology , under nasa contract 1407 .
k. e. y. and c. h. y. acknowledge support of a louisiana space consortium research enhancement award through nasa epscor grant number nnx10ai40h .
m. m. d. acknowledges support from the submillimeter array through an sma postdoctoral fellowship .
n. j. e. acknowledges support from a grant from the national science foundation , ast-1109116 .
we would like to thank the anonymous referee for comments that improved the focus and clarity of this work .
arce , h. g. , borkin , m. a. , goodman , a. a. , pineda , j. e. & beaumont , c. n. 2011 , , 742 , 105 baraffe , i. , chabrier , g. , allard , f. & ha , p. h. 2002 , , 382 , 563 chapman , n. l. , mundy , l. g. , lai , s .-
, & evans , n. j. , ii 2009 , , 690 , 496 chen , x. , arce , h. g. , zhang , q. , et al .
2010 , , 715 , 1344 di francesco , j. 2013 , star formation newsl .
252 , http://www.ifa.hawaii.edu/ reipurth / newsletter.htm dunham , m. m. , crapsi , a. , evans , n. j. ii , et al .
2008 , , 179 , 249 dunham , m. m. , arce , h. g. , allen , l. e. , et al .
2013 , , 145 , 94 dunham , m. m. , stutz , a. m. , allen , l. e. , et al .
2014 , in protostars and planets vi , eds . h. beuther et al .
( tucson , az : uap ) enoch , m. l. , young , k. e. , glenn , j. , et al .
2006 , , 638 , 293 enoch , m. l. , glenn , j. , evans ii , n. j. , et al .
2007 , , 666 , 982 enoch , m. l. , evans ii , n. j , sargent , a. i. , et al .
2008 , , 684 , 1240 enoch , m. l. , evans ii , n. j. , sargent , a. i. , & glenn , j. 2009 , , 692 , 973 enoch , m. l. , lee , j .- e . , harvey , p. , dunham , m. m. , & schnee , s. 2010 , , 722 , l33 evans , n. j. ii , 1999 , ara&a , 37 , 311 evans , n. j. ii , allen , l. e. , blake , g. a. , et al .
2003 , , 115 , 965 evans , n. j. ii , harvey , p. m. , dunham , m. m. , et al . 2007 ,
final delivery of data from the c2d legacy project : irac and mips ( pasadena , ca : ssc ) evans , n. j. ii , dunham , m. m. , jrgensen , j. k. , et al .
2009 , , 181 , 321 evans , n. j. ii , heiderman , a. , & vutisalchavakul , n. 2014 , , 782 , 114 fazio , g. g. , et al .
2004 , , 154 , 10 federrath , c. , glover , s. c. o. , klessen , r. s. & schmidt , w. 2008 , phst , 132 , 014025 froebrich , d. & rowles , j. 2010 , , 406 , 1350 goodman , a. a. , pineda , j. e. , & schnee , s. l. 2009 , , 692 , 91 greene , t. p. , wilking , b. a. , andre , p. , young , e. t , lada , c. j. 1994 , , 434 , 614 gutermuth , r. , myers , p. c. , megeath , s. t. , et al .
2008 , , 674 , 336 gutermuth , r. , megeath , s. t. , myers , p. c. , et sl .
2009 , , 184 , 18 harvey , p. m. , chapman , n. , lai , s .-
2006 , , 644 , 307 harvey , p. m. , mern , b. , huard , t. l. , et al .
2007 , , 663 , 1149 hatchell , j. , richer , j. s. , fuller , g. a. , et al .
2005 , , 440 , 151 hatchell , j. , fuller , g. a. , richer , j. s. , harries , t. j. & ladd , e. f. 2007 , , 468 , 1009 heiderman , a. , evans , n. j. ii , allen , l. e. , huard , t. , & heyer , m. 2010 , , 723 , 1019 heiderman , a. & evans , n. j. ii , apj , in press ( arxiv:1503.06810 ) .
hirota , t. , bushimata , t. , choi , y .- k . ,
et al . , pasj , 60 , 37 , 2008 hirota , t. , honma , m. , imai , h. , et al . ,
pasj , 63 , 1 , 2011 hsieh , t - h . &
lai , s - p . 2013 , , 205 , 5 jrgensen , j. k. , harvey , p. m. , evans ii , n. j. , et al .
2006 , , 645 , 1246 jrgensen , j. k. , johnstone , d. , kirk , h. & myers , p. c. 2007 , , 656 , 293 kainulainen , j. , beuther , h. , henning , t. & plume , r. 2009 , , 508 , l35 kainulainen , j. , alves , j. , beuther , h. , henning , t. & schuller , f. 2011 , , 536 , a48 kirk , h. , johnstone , d. & di francesco , j. 2006 , , 646 , 1009 klessen , r. s. 2000 , , 535 , 869 kryukova , e. , megeath , s. t. , gutermuth , r. a. , et al .
2012 , , 144 , 31 lada , c. j. , muench , a. a. , luhman , k. l. , et al .
2006 , , 131 , 1574 lada , c. j. , lombardi , m. & alves , j. f. 2010 , , 724 , 687 lada , c. j. , lombardi , m. , roman - zuniga , c. , et al .
2013 , 778 , 133 ladd , e. f. , lada , e. a. & myers , p. c. , 1993 , , 410 , 168 ladd , e. f. , myers , p. c. & goodman , a. a. 1994 , , 433 , 117 lonsdale , c. j. , smith , h. e. , rowan - robinson , m. , et al .
2003 , , 115 , 897 infrared science archive , 2007 , c2d spitzer and ancillary data catalog ( pasadena , ca : ipac ) , http://irsa.ipac.caltech.edu/data/spitzer/c2d/ offner , s. s. r. & mckee c. f. 2011 , , 736 , 53 oliveira , i. , mern , b. , pontoppidan , k. m. , et al . 2009 , , 691 , 672 oliveira , i. , pontoppidan , k. m. , mern , b. , et al .
2010 , , 714 , 778 ostriker , e. c. , stone , j. m. & gammie , c. f. 2001 , , 546 , 980 padoan , p. , bally , j. , billawala , y. , juvela , m. & nordlund , a. 1999 , , 525 , 318 pineda , j. e. , caselli , p. , & goodman , a. a. 2008 , , 679 , 481 pineda , j. e. , arce , h. g. , schnee , s. , et al .
2011 , , 743 , 201 rebull , l. m. , stapelfeldt , k. r. , evans ii , n. j. , et al .
2007 , , 171 , 447 rebull , l. m. , cody , a. m. , covey , k. r. , et al . 2014 , , 148 , 92 rebull , l. m. 2015 , submitted robitaille , t. p. , whitney , b. a. , indebetouw , r. , wood , k. & denzmore , p. 2006
, , 167 , 256 rieke , g. , et al .
2004 , , 154 , 25 ridge , n. a. , di francesco , j. , kirk , h. , et al .
2006a , , 131 , 2921 ridge , n. a. , schnee , s. l. , goodman , a. a. & foster , j. b. 2006b , , 643 , 932 sadavoy , s. , di francesco , j. , andre , p. , et al .
2012 , 540 , a10 sadavoy , s. , di francesco , j. , andre , p. & pezzuto , s. 2013 , protostars and planets vi , heidelberg , poster 1s03 , http://www.mpia-hd.mpg.de/homes/ppvi/posters/ sadavoy , s. , di francesco , j. , andre , p. , et al .
2014 , , 787 , l18 sargent , a. i. , 1979 , , 233 , 163 schnee , s. l. , ridge , n. a , goodman , a. a. & li , j. g. 2005 , , 634 , 442 teixeira , p. s. , lada , c. j. , marengo , m. & lada , e. a. 2012 , , 540 , a83 weingartner , j. c. , & draine , b. t. 2001 , , 548 , 296 werner , m. w. , et al .
2004 , , 154 , 1 wilking , b. a. , meyer , m. r. , green , t. p. , mikhail , a. , & carlson , g. 2004 , , 127 , 1131 young , k. e. , harvey , p. m. , brooke , t. y. , et al .
2005 , , 628 , 283 llcccccccc 337 & j034313.70 + 320045.2 & -1.15 & ii & 16 @xmath2 1.1 & 19 @xmath2 1.2 & 16 @xmath2 1.0 & 21 @xmath2 1.3 & 26 @xmath2 2.7 & @xmath30 + 340 & j034323.57 + 321225.9 & -1.52 & ii & 4.9 @xmath2 0.36 & 4.0 @xmath2 0.31 & 3.2 @xmath2 0.28 & 3.0 @xmath2 0.28 & 2.8 @xmath2 0.37 & @xmath30 + 341 & j034325.48 + 315516.5 & -1.33 & ii & 15 @xmath2 1.1 & 11 @xmath2 0.86 & 7.8 @xmath2 0.70 & 5.9 @xmath2 0.69 & 17 @xmath2 1.9 & @xmath30 + 342 & j034328.21 + 320159.1 & -1.39 & ii & 63 @xmath2 4.3 & 65 @xmath2 4.4 & 46 @xmath2 3.6 & 46 @xmath2 3.7 & 45 @xmath2 4.7 & @xmath30 + 343 & j034328.46 + 320505.8 & -1.19 & ii & 8.2 @xmath2 0.60 & 7.6 @xmath2 0.54 & 6.3 @xmath2 0.48 & 7.6 @xmath2 0.54 & 11 @xmath2 1.2 & @xmath30 + 344 & j034329.43 + 315219.5 & 0.41 & 0+i & 7.7 @xmath2 0.53 & 13 @xmath2 0.71 & 20 @xmath2 1.0 & 43 @xmath2 2.1 & 180 @xmath2 20 & 260 @xmath2 33 + 345 & j034336.02 + 315009.0 & -0.18 & flat & 3.7 @xmath2 0.26 & 4.4 @xmath2 0.28 & 5.2 @xmath2 0.31 & 9.5 @xmath2 0.47 & 30 @xmath2 3.1 & @xmath30 + 347 & j034344.63 + 320817.8 & -0.93 & ii & 22 @xmath2 1.6 & 17 @xmath2 1.4 & 12 @xmath2 1.2 & 12 @xmath2 1.2 & 110 @xmath2 12 & 170 @xmath2 44 + 348 & j034345.17 + 320358.6 & 0.11 & flat & 300 @xmath2 22 & 190 @xmath2 19 & 250 @xmath2 18 & 140 @xmath2 17 & 850 @xmath2 90 & 940 @xmath2 100 + 349 & j034348.83 + 321551.5 & -1.07 & ii & 10 @xmath2 0.79 & 9.9 @xmath2 0.71 & 10 @xmath2 0.69 & 13 @xmath2 0.8 & 15 @xmath2 1.6 & @xmath30 + 350 & j034350.96 + 320324.7 & 1.39 & 0+i & 0.49 @xmath2 0.04 & 1.5 @xmath2 0.12 & 1.9 @xmath2 0.18 & 3.4 @xmath2 0.2 & 62 @xmath2 6.6 & @xmath30 + 351 & j034351.02 + 320308.1 & -0.29 & flat & 9.8 @xmath2 0.85 & 5.4 @xmath2 0.85 & 17 @xmath2 1.2 & 2.9 @xmath2 0.24 & 52 @xmath2 5.5 & 2000 @xmath2 290 + 353 & j034355.24 + 315532.1 & -0.99 & ii & 52 @xmath2 3.7 & 43 @xmath2 3.1 & 31 @xmath2 2.6 & 25 @xmath2 2.5 & 100 @xmath2 10 & @xmath30 + 354 & j034355.28 + 320753.3 & -1.09 & ii & 5.0 @xmath2 0.38 & 4.8 @xmath2 0.34 & 4.6 @xmath2 0.34 & 5.6 @xmath2 0.37 & 5.9 @xmath2 0.66 & @xmath30 + 355 & j034356.03 + 320213.3 & -1.20 & ii & 210 @xmath2 14 & 200 @xmath2 14 & 170 @xmath2 11 & 190 @xmath2 13 & 210 @xmath2 22 & @xmath30 + 357 & j034356.84 + 320304.7 & 1.83 & 0+i & 0.07 @xmath2 0 & 0.27 @xmath2 0.08 & 0.66 @xmath2 0.08 & 0.79 @xmath2 0.16 & 13 @xmath2 1.4 & 3400 @xmath2 460 + 358 & j034357.23 + 320133.7 & -1.28 & ii & 7.8 @xmath2 0.53 & 7.4 @xmath2 0.51 & 5.9 @xmath2 0.43 & 6.2 @xmath2 0.46 & 6.5 @xmath2 0.75 & @xmath30 + 362 & j034358.57 + 321727.5 & -1.10 & ii & 33 @xmath2 2.6 & 31 @xmath2 2.2 & 25 @xmath2 1.9 & 31 @xmath2 2.2 & 52 @xmath2 5.4 & @xmath30 + 363 & j034358.92 + 321127.2 & -0.84 & ii & 46 @xmath2 3.5 & 45 @xmath2 3.2 & 38 @xmath2 2.8 & 50 @xmath2 3.3 & 170 @xmath2 17 & @xmath30 + 364 & j034359.10 + 321421.2 & -1.28 & ii & 20 @xmath2 1.5 & 21 @xmath2 1.4 & 16 @xmath2 1.2 & 19 @xmath2 1.3 & 27 @xmath2 3.0 & @xmath30 + 366 & j034359.65 + 320154.1 & -0.90 & ii & 1200 @xmath2 84 & 1300 @xmath2 97 & 1300 @xmath2 74 & 1800 @xmath2 110 & 1500 @xmath2 160 & 1700 @xmath2 190 + 367 & j034359.88 + 320441.4 & -1.10 & ii & 3.9 @xmath2 0.28 & 3.2 @xmath2 0.25 & 2.8 @xmath2 0.23 & 3.2 @xmath2 0.26 & 7 @xmath2 0.79 & @xmath30 + 368 & j034400.48 + 320432.7 & -0.82 & ii & 4.3 @xmath2 0.29 & 4.0 @xmath2 0.27 & 3.7 @xmath2 0.25 & 4.4 @xmath2 0.29 & 9.2 @xmath2 1.1 & @xmath30 + 369 & j034401.57 + 322358.9 & -1.21 & ii & 6.3 @xmath2 0.47 & 5.6 @xmath2 0.41 & 4.5 @xmath2 0.39 & 5.4 @xmath2 0.40 & 8.1 @xmath2 0.88 & @xmath30 + 370 & j034402.40 + 320204.9 & 1.47 & 0+i & 1.1 @xmath2 0.07 & 2.6 @xmath2 0.2 & 4.8 @xmath2 0.34 & 8.4 @xmath2 0.59 & 130 @xmath2 14 & 870 @xmath2 120 + 371 & j034402.63 + 320159.5 & 0.92 & 0+i & 1.2 @xmath2 0.08 & 3.8 @xmath2 0.22 & 5.1 @xmath2 0.25 & 8.0 @xmath2 0.39 & 63 @xmath2 7.5 & @xmath30 + 372 & j034402.92 + 315227.7 & -1.31 & ii & 5.5 @xmath2 0.39 & 4.7 @xmath2 0.35 & 4.3 @xmath2 0.32 & 4.7 @xmath2 0.35 & 4.4 @xmath2 0.51 & @xmath30 + 373 & j034405.78 + 320001.1 & -0.92 & ii & 5.7 @xmath2 0.40 & 5.2 @xmath2 0.34 & 4.7 @xmath2 0.32 & 5.7 @xmath2 0.36 & 9.9 @xmath2 1.1 & @xmath30 + 374 & j034405.78 + 320028.5 & -0.64 & ii & 55 @xmath2 3.8 & 65 @xmath2 4.0 & 68 @xmath2 3.8 & 68 @xmath2 3.7 & 31 @xmath2 3.3 & @xmath30 + 375 & j034406.01 + 321532.1 & -1.16 & ii & 3.0 @xmath2 0.21 & 2.7 @xmath2 0.21 & 2.7 @xmath2 0.20 & 2.8 @xmath2 0.22 & 2.9 @xmath2 0.38 & @xmath30 + 377 & j034406.80 + 320754.0 & -1.15 & ii & 13 @xmath2 0.95 & 11 @xmath2 0.82 & 9.5 @xmath2 0.77 & 11 @xmath2 0.8 & 15 @xmath2 1.6 & @xmath30 + 378 & j034409.16 + 320709.3 & -2.36 & iii & 98 @xmath2 7.2 & 66 @xmath2 5.9 & 46 @xmath2 5.4 & 37 @xmath2 5.3 & 10 @xmath2 1.1 & @xmath30 + 379 & j034409.20 + 320237.8 & 0.61 & 0+i & 17 @xmath2 1.2 & 34 @xmath2 1.8 & 36 @xmath2 1.9 & 63 @xmath2 3.2 & 320 @xmath2 35 & 280 @xmath2 59 + 380 & j034410.13 + 320404.5 & -1.11 & ii & 10 @xmath2 0.72 & 9.3 @xmath2 0.69 & 8.3 @xmath2 0.64 & 9.0 @xmath2 0.73 & 16 @xmath2 1.9 & @xmath30 + 381 & j034411.63 + 320313.1 & -1.39 & ii & 92 @xmath2 6.5 & 68 @xmath2 5.6 & 68 @xmath2 5.0 & 70 @xmath2 5.3 & 50 @xmath2 5.4 & @xmath30 + 382 & j034412.98 + 320135.5 & 0.29 & flat & 220 @xmath2 14 & 590 @xmath2 40 & 620 @xmath2 32 & 790 @xmath2 47 & 1800 @xmath2 200 & 4700 @xmath2 510 + 383 & j034415.84 + 315936.7 & -1.03 & ii & 4.0 @xmath2 0.27 & 3.4 @xmath2 0.26 & 3.1 @xmath2 0.24 & 4.1 @xmath2 0.28 & 7.6 @xmath2 0.84 & @xmath30 + 384 & j034418.10 + 321053.5 & -1.03 & ii & 50 @xmath2 3.6 & 41 @xmath2 2.9 & 32 @xmath2 2.3 & 35 @xmath2 2.5 & 39 @xmath2 4.2 & @xmath30 + 385 & j034418.17 + 320457.0 & -0.92 & ii & 130 @xmath2 8.9 & 83 @xmath2 7.4 & 93 @xmath2 6.9 & 79 @xmath2 6.8 & 580 @xmath2 61 & @xmath30 + 386 & j034418.21 + 320959.3 & -1.45 & ii & 8.2 @xmath2 0.63 & 6.9 @xmath2 0.53 & 5.6 @xmath2 0.48 & 5.2 @xmath2 0.48 & 9.0 @xmath2 1.0 & @xmath30 + 387 & j034418.27 + 320732.5 & -1.49 & ii & 3.8 @xmath2 0.27 & 2.8 @xmath2 0.24 & 2.1 @xmath2 0.23 & 2.2 @xmath2 0.24 & 1.9 @xmath2 0.42 & @xmath30 + 388 & j034418.59 + 321253.1 & -1.01 & ii & 37 @xmath2 2.8 & 37 @xmath2 2.5 & 35 @xmath2 2.4 & 36 @xmath2 2.5 & 36 @xmath2 3.9 & @xmath30 + 389 & j034419.14 + 320931.3 & -2.38 & iii & 32 @xmath2 2.5 & 23 @xmath2 2.0 & 16 @xmath2 1.8 & 9.8 @xmath2 1.7 & 3.8 @xmath2 0.51 & @xmath30 + 391 & j034419.25 + 320734.7 & -1.30 & ii & 16 @xmath2 1.1 & 14 @xmath2 1.1 & 13 @xmath2 1.0 & 14 @xmath2 1.1 & 8.3 @xmath2 0.94 & @xmath30 + 392 & j034420.18 + 320856.5 & -1.66 & iii & 12 @xmath2 0.90 & 11 @xmath2 0.78 & 9.3 @xmath2 0.73 & 9.8 @xmath2 0.79 & 6.0 @xmath2 0.89 & @xmath30 + 393 & j034421.23 + 320114.5 & -1.39 & ii & 7.3 @xmath2 0.51 & 5.3 @xmath2 0.42 & 3.9 @xmath2 0.34 & 4.0 @xmath2 0.36 & 8.9 @xmath2 0.97 & @xmath30 + 394 & j034421.29 + 321237.2 & -1.48 & ii & 5.0 @xmath2 0.39 & 4.8 @xmath2 0.34 & 4.0 @xmath2 0.32 & 3.5 @xmath2 0.32 & 3.2 @xmath2 0.42 & @xmath30 + 395 & j034421.31 + 321156.3 & -1.35 & ii & 14 @xmath2 1.1 & 12 @xmath2 0.88 & 11 @xmath2 0.86 & 14 @xmath2 0.97 & 15 @xmath2 1.6 & @xmath30 + 396 & j034421.35 + 315932.6 & 0.52 & 0+i & 44 @xmath2 3.0 & 67 @xmath2 4.7 & 79 @xmath2 5.3 & 93 @xmath2 6.5 & 260 @xmath2 28 & 610 @xmath2 88 + 397 & j034421.57 + 321509.7 & -1.73 & iii & 6.3 @xmath2 0.47 & 4.7 @xmath2 0.38 & 3.2 @xmath2 0.35 & 2.2 @xmath2 0.33 & 5.0 @xmath2 0.66 & @xmath30 + 398 & j034421.62 + 321037.7 & -1.09 & ii & 75 @xmath2 5.4 & 67 @xmath2 4.8 & 53 @xmath2 4.2 & 50 @xmath2 4.2 & 68 @xmath2 7.2 & @xmath30 + 399 & j034422.29 + 320542.8 & -0.89 & ii & 41 @xmath2 2.8 & 30 @xmath2 2.5 & 28 @xmath2 2.3 & 40 @xmath2 2.7 & 130 @xmath2 13 & @xmath30 + 400 & j034422.34 + 321200.7 & -1.13 & ii & 20 @xmath2 1.5 & 20 @xmath2 1.4 & 21 @xmath2 1.4 & 26 @xmath2 1.7 & 26 @xmath2 2.7 & @xmath30 + 401 & j034422.58 + 320153.6 & -1.05 & ii & 20 @xmath2 1.4 & 13 @xmath2 1.2 & 10 @xmath2 1.1 & 8.7 @xmath2 1.1 & 73 @xmath2 7.6 & @xmath30 + 402 & j034422.70 + 320142.3 & -1.68 & iii & 4.6 @xmath2 0.32 & 3.4 @xmath2 0.27 & 2.7 @xmath2 0.24 & 2.6 @xmath2 0.26 & 7.8 @xmath2 1.8 & @xmath30 + 403 & j034423.65 + 320152.7 & -1.36 & ii & 13 @xmath2 0.91 & 11 @xmath2 0.8 & 9.7 @xmath2 0.71 & 11 @xmath2 0.82 & 10 @xmath2 1.2 & @xmath30 + 404 & j034424.46 + 320143.7 & -1.25 & ii & 2.6 @xmath2 0.19 & 2.7 @xmath2 0.18 & 2.2 @xmath2 0.17 & 2.4 @xmath2 0.19 & 4.2 @xmath2 1.0 & @xmath30 + 405 & j034424.84 + 321348.4 & 1.65 & 0+i & 0.27 @xmath2 0.02 & 0.32 @xmath2 0.02 & 0.33 @xmath2 0.07 & 0.99 @xmath2 0.21 & 36 @xmath2 3.7 & @xmath30 + 406 & j034425.32 + 321012.7 & -1.42 & ii & 14 @xmath2 1.1 & 12 @xmath2 0.93 & 11 @xmath2 0.88 & 14 @xmath2 1.0 & 12 @xmath2 3.9 & @xmath30 + 407 & j034425.52 + 321131.2 & -0.79 & ii & 63 @xmath2 4.8 & 54 @xmath2 4.2 & 46 @xmath2 3.6 & 67 @xmath2 4.4 & 150 @xmath2 16 & @xmath30 + 408 & j034425.55 + 320617.1 & -1.37 & ii & 22 @xmath2 1.5 & 16 @xmath2 0.13 & 1.3 @xmath2 1.2 & 14 @xmath2 1.4 & 30 @xmath2 3.4 & @xmath30 + 409 & j034425.71 + 321549.2 & -1.06 & ii & 3.1 @xmath2 0.23 & 2.7 @xmath2 0.20 & 2.4 @xmath2 0.2 & 2.7 @xmath2 0.23 & 5.9 @xmath2 0.70 & @xmath30 + 410 & j034426.04 + 320430.4 & -1.15 & ii & 570 @xmath2 39 & 520 @xmath2 39 & 450 @xmath2 33 & 490 @xmath2 40 & 540 @xmath2 58 & @xmath30 + 411 & j034426.70 + 320820.3 & -1.21 & ii & 47 @xmath2 3.6 & 39 @xmath2 2.9 & 28 @xmath2 2.4 & 31 @xmath2 2.7 & 59 @xmath2 6.4 & @xmath30 + 412 & j034427.22 + 322028.8 & -1.39 & ii & 4.5 @xmath2 0.34 & 3.9 @xmath2 0.29 & 3.2 @xmath2 0.27 & 3.4 @xmath2 0.27 & 4.3 @xmath2 0.51 & @xmath30 + 413 & j034427.26 + 321420.9 & -1.17 & ii & 14 @xmath2 1.1 & 13 @xmath2 0.92 & 11 @xmath2 0.83 & 13 @xmath2 0.92 & 23 @xmath2 2.4 & @xmath30 + 414 & j034427.26 + 321037.3 & -1.60 & ii & 9.7 @xmath2 0.72 & 8.3 @xmath2 0.62 & 6.1 @xmath2 0.6 & 6.4 @xmath2 0.61 & 14 @xmath2 3.5 & @xmath30 + 416 & j034428.14 + 321600.2 & -2.35 & iii & 14 @xmath2 1.1 & 10 @xmath2 0.85 & 7.3 @xmath2 0.75 & 4.7 @xmath2 0.72 & 1.8 @xmath2 0.32 & @xmath30 + 417 & j034428.51 + 315954.0 & -1.51 & ii & 23 @xmath2 1.6 & 19 @xmath2 1.5 & 15 @xmath2 1.3 & 16 @xmath2 1.3 & 16 @xmath2 1.7 & @xmath30 + 418 & j034428.95 + 320137.9 & -1.00 & ii & 18 @xmath2 1.3 & 17 @xmath2 1.2 & 14 @xmath2 0.99 & 17 @xmath2 1.2 & 41 @xmath2 4.4 & @xmath30 + 419 & j034429.23 + 320115.7 & -1.33 & ii & 11 @xmath2 0.76 & 8.0 @xmath2 0.64 & 5.9 @xmath2 0.55 & 5.8 @xmath2 0.57 & 17 @xmath2 1.8 & @xmath30 + 420 & j034429.74 + 321039.8 & -1.08 & ii & 53 @xmath2 4.0 & 47 @xmath2 3.4 & 37 @xmath2 3.0 & 45 @xmath2 3.3 & 160 @xmath2 17 & @xmath30 + 421 & j034429.80 + 320054.6 & -0.93 & ii & 7.8 @xmath2 0.54 & 6.9 @xmath2 0.50 & 6.8 @xmath2 0.48 & 8.2 @xmath2 0.54 & 9.9 @xmath2 0.11 & @xmath30 + 422 & j034429.98 + 321922.6 & -2.19 & iii & 14 @xmath2 1.1 & 11 @xmath2 0.87 & 7.0 @xmath2 0.78 & 4.2 @xmath2 0.71 & 2.2 @xmath2 0.32 & @xmath30 + 423 & j034430.14 + 320118.2 & -1.83 & iii & 10 @xmath2 0.71 & 7.2 @xmath2 0.59 & 5.7 @xmath2 0.51 & 5.3 @xmath2 0.54 & 3.9 @xmath2 0.53 & @xmath30 + 424 & j034430.30 + 321135.2 & 0.05 & flat & 3.3 @xmath2 0.24 & 4.0 @xmath2 0.27 & 6.2 @xmath2 0.41 & 8.4 @xmath2 0.66 & 20 @xmath2 2.5 & @xmath30 + 425 & j034430.84 + 320955.7 & -1.46 & ii & 51 @xmath2 3.7 & 35 @xmath2 3.1 & 41 @xmath2 3.3 & 74 @xmath2 4.8 & 26 @xmath2 6.8 & @xmath30 + 426 & j034431.13 + 321848.5 & -0.92 & ii & 8.6 @xmath2 0.64 & 8.5 @xmath2 0.59 & 7.8 @xmath2 0.53 & 9.8 @xmath2 0.61 & 18 @xmath2 1.9 & @xmath30 + 427 & j034431.19 + 320558.9 & -0.59 & ii & 13 @xmath2 0.88 & 13 @xmath2 0.94 & 13 @xmath2 0.9 & 13 @xmath2 0.91 & 31 @xmath2 3.3 & @xmath30 + 428 & j034431.37 + 320014.2 & -1.28 & ii & 69 @xmath2 5.3 & 63 @xmath2 4.7 & 53 @xmath2 3.9 & 67 @xmath2 4.5 & 77 @xmath2 8.1 & @xmath30 + 429 & j034431.55 + 320844.9 & -2.36 & iii & 42 @xmath2 3.2 & 29 @xmath2 2.5 & 20 @xmath2 2.3 & 13 @xmath2 2.2 & 12 @xmath2 3.8 & @xmath30 + 430 & j034432.04 + 321143.7 & -0.62 & ii & 200 @xmath2 16 & 150 @xmath2 13 & 330 @xmath2 22 & 1000 @xmath2 66 & 430 @xmath2 45 & @xmath30 + 431 & j034432.59 + 320855.7 & -1.93 & iii & 23 @xmath2 1.7 & 16 @xmath2 1.4 & 14 @xmath2 1.2 & 13 @xmath2 1.3 & 15 @xmath2 4.0 & @xmath30 + 432 & j034433.22 + 321257.4 & -1.35 & ii & 6.4 @xmath2 0.47 & 4.9 @xmath2 0.38 & 4.1 @xmath2 0.35 & 4.6 @xmath2 0.44 & 7.0 @xmath2 0.89 & @xmath30 + 433 & j034433.79 + 315830.2 & -1.25 & ii & 14 @xmath2 1.0 & 12 @xmath2 0.88 & 9.2 @xmath2 0.77 & 9.7 @xmath2 0.80 & 24 @xmath2 2.5 & @xmath30 + 434 & j034434.05 + 320657.0 & -0.81 & ii & 4.6 @xmath2 0.32 & 4.3 @xmath2 0.32 & 4.5 @xmath2 0.32 & 5.5 @xmath2 0.40 & 5.6 @xmath2 1.3 & @xmath30 + 435 & j034434.14 + 321635.7 & -1.20 & ii & 6.0 @xmath2 0.45 & 5.3 @xmath2 0.39 & 4.5 @xmath2 0.37 & 5.0 @xmath2 0.39 & 7.9 @xmath2 0.94 & @xmath30 + 436 & j034434.21 + 320946.3 & -2.02 & iii & 620 @xmath2 54 & 380 @xmath2 39 & 300 @xmath2 35 & 170 @xmath2 32 & 350 @xmath2 53 & @xmath30 + 437 & j034434.29 + 321240.7 & -1.34 & ii & 29 @xmath2 2.2 & 29 @xmath2 2.0 & 23 @xmath2 1.6 & 27 @xmath2 1.8 & 30 @xmath2 3.2 & @xmath30 + 438 & j034434.69 + 321600.1 & -0.43 & ii & 17 @xmath2 1.3 & 19 @xmath2 1.2 & 18 @xmath2 1.1 & 21 @xmath2 1.3 & 150 @xmath2 15 & @xmath30 + 439 & j034434.71 + 321554.4 & -1.98 & iii & 12 @xmath2 0.96 & 9.3 @xmath2 0.74 & 6.3 @xmath2 0.64 & 3.6 @xmath2 0.62 & 7.2 @xmath2 1.4 & @xmath30 + 440 & j034434.81 + 315655.2 & -1.19 & ii & 13 @xmath2 0.94 & 10 @xmath2 0.82 & 7.4 @xmath2 0.73 & 5.9 @xmath2 0.71 & 23 @xmath2 2.5 & @xmath30 + 441 & j034434.99 + 321531.1 & -1.29 & ii & 10 @xmath2 0.78 & 9.8 @xmath2 0.69 & 9.1 @xmath2 0.67 & 12 @xmath2 0.77 & 8.6 @xmath2 0.94 & @xmath30 + 443 & j034435.38 + 320736.2 & -1.20 & ii & 37 @xmath2 2.7 & 38 @xmath2 2.7 & 30 @xmath2 2.2 & 46 @xmath2 2.9 & 55 @xmath2 5.8 & @xmath30 + 444 & j034435.38 + 321004.6 & -1.57 & ii & 430 @xmath2 37 & 350 @xmath2 32 & 290 @xmath2 26 & 300 @xmath2 27 & 270 @xmath2 30 & @xmath30 + 445 & j034435.47 + 320856.3 & -1.13 & ii & 12 @xmath2 0.90 & 10 @xmath2 0.78 & 9.3 @xmath2 0.79 & 11 @xmath2 1.2 & 50 @xmath2 11 & @xmath30 + 446 & j034435.69 + 320303.5 & -1.06 & ii & 21 @xmath2 1.5 & 19 @xmath2 1.4 & 18 @xmath2 1.3 & 22 @xmath2 0.14 & 27 @xmath2 3.1 & @xmath30 + 448 & j034436.96 + 320645.2 & -1.93 & iii & 230 @xmath2 15 & 160 @xmath2 14 & 130 @xmath2 13 & 98 @xmath2 12 & 47 @xmath2 5.3 & @xmath30 + 449 & j034437.40 + 321224.2 & -0.93 & ii & 21 @xmath2 1.5 & 16 @xmath2 1.3 & 12 @xmath2 1.1 & 15 @xmath2 1.3 & 77 @xmath2 8.2 & @xmath30 + 450 & j034437.88 + 320804.1 & -1.22 & ii & 97 @xmath2 6.8 & 93 @xmath2 6.5 & 68 @xmath2 5.4 & 87 @xmath2 6.2 & 100 @xmath2 11 & @xmath30 + 451 & j034437.98 + 320329.7 & -1.24 & ii & 70 @xmath2 4.8 & 60 @xmath2 4.6 & 51 @xmath2 4.0 & 72 @xmath2 4.7 & 75 @xmath2 8.1 & @xmath30 + 452 & j034438.01 + 321137.0 & -1.49 & ii & 8.2 @xmath2 0.6 & 6.5 @xmath2 0.51 & 6.0 @xmath2 0.5 & 7.0 @xmath2 0.61 & 10 @xmath2 2.9 & @xmath30 + 453 & j034438.46 + 320735.7 & -1.13 & ii & 55 @xmath2 3.8 & 44 @xmath2 3.5 & 44 @xmath2 3.3 & 73 @xmath2 4.6 & 88 @xmath2 9.4 & @xmath30 + 454 & j034438.54 + 320800.6 & -0.97 & ii & 36 @xmath2 2.5 & 29 @xmath2 2.3 & 23 @xmath2 2.0 & 35 @xmath2 2.6 & 120 @xmath2 13 & @xmath30 + 455 & j034438.97 + 320319.7 & -1.25 & ii & 4.1 @xmath2 0.29 & 3.4 @xmath2 0.26 & 2.9 @xmath2 0.25 & 3.8 @xmath2 0.37 & 5.9 @xmath2 1.6 & @xmath30 + 456 & j034439.19 + 322008.9 & -1.80 & iii & 13 @xmath2 0.95 & 10 @xmath2 0.80 & 8.6 @xmath2 0.75 & 8.3 @xmath2 0.81 & 6.2 @xmath2 1.8 & @xmath30 + 457 & j034439.21 + 320944.8 & -1.44 & ii & 22 @xmath2 1.6 & 19 @xmath2 1.4 & 16 @xmath2 1.4 & 18 @xmath2 1.6 & 52 @xmath2 15 & @xmath30 + 458 & j034439.80 + 321804.0 & -1.40 & ii & 27 @xmath2 2.0 & 21 @xmath2 1.7 & 17 @xmath2 1.5 & 18 @xmath2 1.5 & 24 @xmath2 2.5 & @xmath30 + 459 & j034441.74 + 321202.2 & -1.07 & ii & 15 @xmath2 1.1 & 11 @xmath2 0.88 & 8.6 @xmath2 0.81 & 8.9 @xmath2 0.85 & 56 @xmath2 5.9 & @xmath30 + 460 & j034442.04 + 320859.9 & -2.01 & iii & 41 @xmath2 3.2 & 34 @xmath2 2.6 & 25 @xmath2 2.4 & 26 @xmath2 2.6 & 17 @xmath2 4.0 & @xmath30 + 461 & j034442.15 + 320902.1 & -0.52 & ii & 26 @xmath2 2.2 & 35 @xmath2 2.2 & 36 @xmath2 2.2 & 39 @xmath2 2.4 & 74 @xmath2 7.8 & @xmath30 + 462 & j034442.58 + 321002.5 & -1.29 & ii & 14 @xmath2 1.1 & 14 @xmath2 0.99 & 13 @xmath2 0.91 & 14 @xmath2 1.0 & 16 @xmath2 2.5 & @xmath30 + 463 & j034442.76 + 320833.7 & -1.35 & ii & 14 @xmath2 1.0 & 12 @xmath2 0.91 & 12 @xmath2 0.89 & 13 @xmath2 1.1 & 15 @xmath2 3.4 & @xmath30 + 464 & j034443.03 + 321559.6 & -0.65 & ii & 5.4 @xmath2 0.39 & 5.1 @xmath2 0.36 & 5.0 @xmath2 0.37 & 7.9 @xmath2 0.49 & 20 @xmath2 2.1 & @xmath30 + 466 & j034443.32 + 320131.5 & 0.61 & 0+i & 260 @xmath2 19 & 400 @xmath2 27 & 480 @xmath2 25 & 740 @xmath2 38 & 2300 @xmath2 250 & @xmath30 + 467 & j034443.53 + 320742.7 & -1.97 & iii & 36 @xmath2 2.6 & 23 @xmath2 2.0 & 16 @xmath2 1.8 & 11 @xmath2 1.9 & 11 @xmath2 1.5 & @xmath30 + 468 & j034443.78 + 321030.4 & -0.87 & ii & 24 @xmath2 1.7 & 23 @xmath2 1.6 & 24 @xmath2 1.6 & 43 @xmath2 2.4 & 74 @xmath2 7.9 & @xmath30 + 469 & j034443.96 + 320136.2 & 0.74 & 0+i & 3.9 @xmath2 0.30 & 8.5 @xmath2 0.89 & 7.0 @xmath2 0.41 & @xmath30 & 170 @xmath2 23 & 11000 @xmath2 1200 + 470 & j034444.59 + 320812.5 & -0.91 & ii & 33 @xmath2 2.3 & 32 @xmath2 2.2 & 27 @xmath2 2.0 & 39 @xmath2 2.4 & 30 @xmath2 4.0 & @xmath30 + 471 & j034444.72 + 320402.5 & -1.18 & ii & 130 @xmath2 9.1 & 120 @xmath2 8.8 & 100 @xmath2 7.7 & 120 @xmath2 8.4 & 170 @xmath2 18 & @xmath30 + 472 & j034445.20 + 320119.6 & -0.58 & ii & 18 @xmath2 1.2 & 17 @xmath2 1.2 & 19 @xmath2 1.1 & 23 @xmath2 1.3 & 35 @xmath2 3.8 & @xmath30 + 473 & j034450.35 + 315236.0 & -1.12 & ii & 4.0 @xmath2 0.28 & 3.7 @xmath2 0.26 & 3.0 @xmath2 0.24 & 3.0 @xmath2 0.24 & 5.2 @xmath2 0.57 & @xmath30 + 475 & j034452.05 + 315825.2 & -1.27 & ii & 16 @xmath2 1.1 & 1.4 @xmath2 0.99 & 11 @xmath2 0.89 & 9.9 @xmath2 0.87 & 15 @xmath2 1.5 & @xmath30 + 476 & j034455.62 + 320919.8 & -2.23 & iii & 33 @xmath2 2.3 & 22 @xmath2 1.9 & 14 @xmath2 1.6 & 9.2 @xmath2 1.5 & 10 @xmath2 2.3 & @xmath30 + 477 & j034456.14 + 320915.2 & -1.12 & ii & 110 @xmath2 7.8 & 79 @xmath2 6.7 & 65 @xmath2 6.0 & 53 @xmath2 5.8 & 230 @xmath2 25 & @xmath30 + 478 & j034456.84 + 315411.4 & -1.35 & ii & 3.0 @xmath2 0.2 & 2.2 @xmath2 0.18 & 1.7 @xmath2 0.16 & 1.2 @xmath2 0.17 & 3.3 @xmath2 0.39 & @xmath30 + 479 & j034456.90 + 322035.6 & -0.72 & ii & 2.9 @xmath2 0.21 & 3.1 @xmath2 0.19 & 2.7 @xmath2 0.18 & 2.5 @xmath2 0.21 & 6.3 @xmath2 0.80 & @xmath30 + 480 & j034457.72 + 320741.7 & -1.10 & ii & 3.1 @xmath2 0.21 & 2.3 @xmath2 0.18 & 1.7 @xmath2 0.17 & 1.4 @xmath2 0.2 & 6.9 @xmath2 0.8 & @xmath30 + 481 & j034457.85 + 320401.5 & -1.82 & iii & 4.9 @xmath2 0.33 & 3.8 @xmath2 0.30 & 3.0 @xmath2 0.28 & 3.2 @xmath2 0.29 & 1.6 @xmath2 0.35 & @xmath30 + 482 & j034458.55 + 315827.1 & -1.19 & ii & 2.3 @xmath2 0.16 & 2.0 @xmath2 0.15 & 1.8 @xmath2 0.13 & 2.2 @xmath2 0.15 & 1.7 @xmath2 0.30 & @xmath30 + 483 & j034459.85 + 321331.9 & -1.68 & iii & 6.6 @xmath2 0.46 & 5.2 @xmath2 0.41 & 4.4 @xmath2 0.39 & 4.4 @xmath2 0.41 & 2.3 @xmath2 0.37 & @xmath30 + 484 & j034500.45 + 320320.1 & -1.78 & iii & 5.3 @xmath2 0.37 & 4.1 @xmath2 0.32 & 3.2 @xmath2 0.28 & 2.5 @xmath2 0.28 & 2.1 @xmath2 0.38 & @xmath30 + 485 & j034501.43 + 320501.7 & -2.31 & iii & 44 @xmath2 3.0 & 30 @xmath2 2.6 & 20 @xmath2 2.4 & 14 @xmath2 2.3 & 4.4 @xmath2 0.51 & @xmath30 + 486 & j034504.66 + 321501.1 & -2.10 & iii & 10 @xmath2 0.75 & 7.3 @xmath2 0.60 & 5.3 @xmath2 0.53 & 4.4 @xmath2 0.52 & 2.5 @xmath2 0.82 & @xmath30 + 487 & j034507.64 + 321027.9 & -2.48 & iii & 40 @xmath2 3.5 & 34 @xmath2 2.6 & 24 @xmath2 2.3 & 15 @xmath2 2.1 & 3.9 @xmath2 0.49 & @xmath30 + 488 & j034513.08 + 322005.3 & -1.48 & ii & 2.6 @xmath2 0.2 & 2.4 @xmath2 0.17 & 1.9 @xmath2 0.15 & 2.1 @xmath2 0.17 & 1.7 @xmath2 0.27 & @xmath30 + 490 & j034513.82 + 321210.0 & 0.43 & 0+i & 2.9 @xmath2 0.2 & 4.7 @xmath2 0.27 & 8.1 @xmath2 0.42 & 16 @xmath2 0.76 & 69 @xmath2 7.4 & 290 @xmath2 38 + 491 & j034516.34 + 320620.1 & -0.71 & ii & 240 @xmath2 17 & 280 @xmath2 18 & 260 @xmath2 16 & 370 @xmath2 24 & 450 @xmath2 48 & 470 @xmath2 59 + 492 & j034517.82 + 321205.8 & -1.52 & ii & 6.3 @xmath2 0.47 & 5.5 @xmath2 0.41 & 4.2 @xmath2 0.38 & 4.0 @xmath2 0.36 & 4.3 @xmath2 0.53 & @xmath30 + 493 & j034520.46 + 320634.5 & -1.13 & ii & 130 @xmath2 8.9 & 120 @xmath2 8.4 & 91 @xmath2 7.2 & 110 @xmath2 7.9 & 230 @xmath2 24 & 370 @xmath2 42 + 494 & j034525.15 + 320930.3 & -0.83 & ii & 41 @xmath2 2.7 & 35 @xmath2 2.6 & 31 @xmath2 2.4 & 47 @xmath2 2.9 & 120 @xmath2 12 & 91 @xmath2 16 + 495 & j034529.72 + 315919.7 & -1.29 & ii & 4.6 @xmath2 0.31 & 3.6 @xmath2 0.27 & 3.2 @xmath2 0.26 & 3.3 @xmath2 0.27 & 5.1 @xmath2 0.58 & @xmath30 + 497 & j034535.64 + 315954.4 & -1.32 & ii & 12 @xmath2 0.81 & 9.4 @xmath2 0.71 & 8.1 @xmath2 0.65 & 9.9 @xmath2 0.71 & 14 @xmath2 1.5 & @xmath30 llcccccccc 136 & j032832.56 + 311105.1 & 0.51 & 0+i & 1.4 @xmath2 0.13 & 3.1 @xmath2 0.23 & 3.3 @xmath2 0.23 & 5.3 @xmath2 0.37 & 63 @xmath2 6.7 & 63 @xmath2 6.7 + 138 & j032834.49 + 310051.1 & 0.31 & 0+i & 6.6 @xmath2 0.46 & 16 @xmath2 0.86 & 31 @xmath2 1.5 & 45 @xmath2 2.2 & 69 @xmath2 7.4 & 160 @xmath2 19 + 141 & j032837.09 + 311330.8 & 1.93 & 0+i & 43 @xmath2 3.3 & 120 @xmath2 7.7 & 340 @xmath2 17 & 950 @xmath2 54 & 7900 @xmath2 1400 & 55000 @xmath2 5800 + 142 & j032838.78 + 311806.6 & -0.25 & flat & 6.5 @xmath2 0.46 & 9.4 @xmath2 0.58 & 10 @xmath2 0.59 & 13 @xmath2 0.70 & 18 @xmath2 2.0 & @xmath30 + 143 & j032839.10 + 310601.7 & 1.55 & 0+i & 0.19 @xmath2 0 & 0.36 @xmath2 0.03 & 0.55 @xmath2 0.05 & 1.2 @xmath2 0.09 & 28 @xmath2 3.0 & 96 @xmath2 15 + 144 & j032839.71 + 311731.9 & 0.61 & 0+i & 19 @xmath2 1.4 & 31 @xmath2 1.8 & 41 @xmath2 2.2 & 55 @xmath2 2.8 & 220 @xmath2 22 & 370 @xmath2 41 + 145 & j032840.63 + 311756.5 & 0.96 & 0+i & 0.04 @xmath2 0.01 & 0.26 @xmath2 0.02 & 0.62 @xmath2 0.05 & 1.2 @xmath2 0.08 & 6.0 @xmath2 0.69 & @xmath30 + 147 & j032843.24 + 311042.7 & -1.86 & iii & 23 @xmath2 1.6 & 16 @xmath2 1.3 & 13 @xmath2 1.1 & 11 @xmath2 1.1 & 5.8 @xmath2 0.66 & @xmath30 + 148 & j032843.28 + 311732.9 & 0.24 & flat & 260 @xmath2 20 & 470 @xmath2 28 & 620 @xmath2 38 & 1000 @xmath2 61 & 2200 @xmath2 240 & 2100 @xmath2 240 + 149 & j032844.09 + 312052.7 & -1.34 & ii & 6.4 @xmath2 0.47 & 5.2 @xmath2 0.41 & 4.7 @xmath2 0.37 & 5.2 @xmath2 0.39 & 4.8 @xmath2 0.53 & @xmath30 + 151 & j032846.21 + 311638.4 & -1.84 & iii & 51 @xmath2 4.0 & 43 @xmath2 3.3 & 38 @xmath2 3.1 & 36 @xmath2 3.0 & 15 @xmath2 1.6 & @xmath30 + 152 & j032847.65 + 312406.0 & -1.32 & ii & 35 @xmath2 2.7 & 29 @xmath2 2.1 & 25 @xmath2 1.9 & 20 @xmath2 1.9 & 21 @xmath2 2.1 & @xmath30 + 153 & j032847.84 + 311655.1 & -1.10 & ii & 39 @xmath2 3.0 & 46 @xmath2 3.1 & 38 @xmath2 2.5 & 43 @xmath2 2.7 & 53 @xmath2 5.6 & @xmath30 + 154 & j032848.77 + 311608.8 & 0.20 & flat & 2.3 @xmath2 0.17 & 3.1 @xmath2 0.19 & 4.3 @xmath2 0.24 & 7.2 @xmath2 0.37 & 22 @xmath2 2.3 & @xmath30 + 156 & j032850.62 + 304244.7 & -0.95 & ii & 220 @xmath2 19 & 200 @xmath2 16 & 180 @xmath2 15 & 270 @xmath2 22 & 390 @xmath2 40 & @xmath30 + 157 & j032851.08 + 311632.4 & -1.48 & ii & 6.3 @xmath2 0.49 & 5.6 @xmath2 0.42 & 5.1 @xmath2 0.4 & 5.8 @xmath2 0.42 & 4.0 @xmath2 0.49 & @xmath30 + 158 & j032851.20 + 311954.8 & -1.00 & ii & 78 @xmath2 5.8 & 72 @xmath2 5.2 & 69 @xmath2 4.9 & 110 @xmath2 6.4 & 150 @xmath2 16 & @xmath30 + 159 & j032851.26 + 311739.3 & 0.60 & 0+i & 0.96 @xmath2 0.08 & 1.6 @xmath2 0.10 & 2.2 @xmath2 0.01 & 6.5 @xmath2 0.42 & 29 @xmath2 3.0 & @xmath30 + 160 & j032852.15 + 311547.1 & -1.34 & ii & 8.8 @xmath2 0.65 & 7.3 @xmath2 0.56 & 6.6 @xmath2 0.53 & 8.1 @xmath2 0.58 & 5.3 @xmath2 0.62 & @xmath30 + 162 & j032852.17 + 312245.3 & -1.45 & ii & 27 @xmath2 2.1 & 22 @xmath2 1.8 & 19 @xmath2 1.6 & 22 @xmath2 1.6 & 25 @xmath2 2.6 & @xmath30 + 163 & j032852.92 + 311626.4 & -1.49 & ii & 4.8 @xmath2 0.37 & 4.2 @xmath2 0.32 & 3.8 @xmath2 0.3 & 3.9 @xmath2 0.31 & 3.2 @xmath2 0.46 & @xmath30 + 164 & j032853.96 + 311809.3 & -1.29 & ii & 69 @xmath2 5.3 & 60 @xmath2 4.2 & 51 @xmath2 3.6 & 70 @xmath2 4.4 & 130 @xmath2 13 & 350 @xmath2 47 + 165 & j032854.09 + 311654.2 & -1.07 & ii & 15 @xmath2 1.1 & 12 @xmath2 0.91 & 11 @xmath2 0.9 & 16 @xmath2 1.0 & 23 @xmath2 2.7 & @xmath30 + 166 & j032854.63 + 311651.1 & -1.08 & ii & 94 @xmath2 7.1 & 85 @xmath2 6.2 & 80 @xmath2 5.5 & 110 @xmath2 6.6 & 140 @xmath2 14 & @xmath30 + 167 & j032855.08 + 311628.7 & -0.76 & ii & 140 @xmath2 9.9 & 120 @xmath2 8.8 & 110 @xmath2 7.7 & 130 @xmath2 8.9 & 250 @xmath2 26 & @xmath30 + 168 & j032855.55 + 311436.7 & 2.39 & 0+i & 1.4 @xmath2 0.12 & 17 @xmath2 1.2 & 29 @xmath2 1.8 & 41 @xmath2 2.8 & @xmath30 & 140000 @xmath2 15000 + 169 & j032856.12 + 311908.4 & -0.51 & ii & 8.5 @xmath2 0.62 & 25 @xmath2 1.6 & 34 @xmath2 1.7 & 34 @xmath2 1.7 & 24 @xmath2 2.6 & @xmath30 + 170 & j032856.32 + 312227.9 & -0.40 & ii & 23 @xmath2 1.7 & 24 @xmath2 1.6 & 29 @xmath2 1.8 & 41 @xmath2 2.2 & 100 @xmath2 11 & 560 @xmath2 72 + 171 & j032856.60 + 310737.0 & 0.01 & flat & 1.1 @xmath2 0.09 & 3.4 @xmath2 0.22 & 2.5 @xmath2 0.17 & 0.87 @xmath2 0.11 & 14 @xmath2 1.4 & 110 @xmath2 14 + 172 & j032856.65 + 311835.5 & -1.14 & ii & 150 @xmath2 11 & 130 @xmath2 10 & 90 @xmath2 7.5 & 100 @xmath2 8.1 & 280 @xmath2 29 & 650 @xmath2 76 + 173 & j032856.97 + 311622.3 & -0.74 & ii & 24 @xmath2 1.9 & 20 @xmath2 1.6 & 17 @xmath2 1.4 & 26 @xmath2 1.7 & 210 @xmath2 22 & @xmath30 + 174 & j032857.18 + 311534.6 & -0.71 & ii & 5.1 @xmath2 0.40 & 4.8 @xmath2 0.35 & 5.3 @xmath2 0.37 & 8.2 @xmath2 0.45 & 16 @xmath2 1.7 & @xmath30 + 175 & j032857.21 + 311419.1 & 1.40 & 0+i & 46 @xmath2 3.2 & 130 @xmath2 8.1 & 330 @xmath2 17 & 500 @xmath2 30 & 5000 @xmath2 530 & 31000 @xmath2 3300 + 177 & j032857.70 + 311948.1 & -0.96 & ii & 19 @xmath2 1.5 & 19 @xmath2 1.3 & 18 @xmath2 1.2 & 26 @xmath2 1.6 & 41 @xmath2 4.4 & @xmath30 + 178 & j032858.11 + 311803.7 & -2.37 & iii & 12 @xmath2 0.92 & 8.7 @xmath2 0.74 & 6.3 @xmath2 0.66 & 4.7 @xmath2 0.64 & 1.3 @xmath2 0.28 & @xmath30 + 179 & j032858.26 + 312209.2 & -0.99 & ii & 5.3 @xmath2 0.38 & 4.9 @xmath2 0.38 & 4.5 @xmath2 0.34 & 5.7 @xmath2 1.1 & 44 @xmath2 12 & @xmath30 + 180 & j032858.27 + 312202.0 & -1.07 & ii & 13 @xmath2 1.0 & 14 @xmath2 0.9 & 13 @xmath2 0.87 & 17 @xmath2 0.99 & 19 @xmath2 4.1 & @xmath30 + 181 & j032858.43 + 312217.5 & 0.82 & 0+i & 57 @xmath2 4.3 & 110 @xmath2 11 & 230 @xmath2 12 & 340 @xmath2 21 & 990 @xmath2 110 & 950 @xmath2 110 + 182 & j032859.23 + 312032.5 & -0.02 & flat & 0.90 @xmath2 0.10 & 2.0 @xmath2 0.19 & 1.7 @xmath2 0.17 & 1.7 @xmath2 0.24 & 13 @xmath2 2.3 & @xmath30 + 183 & j032859.32 + 311548.7 & -0.08 & flat & 160 @xmath2 11 & 160 @xmath2 12 & 150 @xmath2 11 & 250 @xmath2 14 & 970 @xmath2 100 & @xmath30 + 184 & j032859.56 + 312146.7 & -1.11 & ii & 120 @xmath2 9.0 & 120 @xmath2 8.1 & 96 @xmath2 6.8 & 110 @xmath2 7.4 & 12 @xmath2 13 & @xmath30 + 185 & j032900.55 + 311200.8 & 1.96 & 0+i & @xmath30 & 0.12 @xmath2 0 & 0.14 @xmath2 0.04 & @xmath30 & 17 @xmath2 1.8 & 1300 @xmath2 180 + 186 & j032901.56 + 312020.6 & 2.30 & 0+i & 770 @xmath2 66 & 1500 @xmath2 110 & 2200 @xmath2 270 & 4100 @xmath2 250 & @xmath30 & 24000 @xmath2 2600 + 188 & j032902.81 + 312217.2 & -1.32 & ii & 9.5 @xmath2 0.72 & 9.1 @xmath2 0.61 & 8.5 @xmath2 0.57 & 9.6 @xmath2 0.62 & 7.9 @xmath2 2.1 & @xmath30 + 189 & j032903.15 + 312238.0 & -0.76 & ii & 35 @xmath2 2.5 & 38 @xmath2 2.5 & 35 @xmath2 2.3 & 52 @xmath2 3.0 & 100 @xmath2 11 & @xmath30 + 190 & j032903.22 + 312545.1 & -0.81 & ii & 2.6 @xmath2 0.19 & 2.3 @xmath2 0.18 & 2.1 @xmath2 0.17 & 2.4 @xmath2 0.19 & 9.2 @xmath2 1.0 & @xmath30 + 191 & j032903.33 + 312314.6 & 1.00 & 0+i & 6.5 @xmath2 0.47 & 10 @xmath2 0.59 & 24 @xmath2 1.2 & 61 @xmath2 2.9 & 160 @xmath2 17 & @xmath30 + 192 & j032903.78 + 311603.8 & 0.85 & 0+i & 1300 @xmath2 100 & 480 @xmath2 110 & 3600 @xmath2 420 & 7500 @xmath2 430 & @xmath30 & 130000 @xmath2 13000 + 194 & j032903.87 + 312148.6 & -1.06 & ii & 260 @xmath2 20 & 230 @xmath2 18 & 210 @xmath2 15 & 280 @xmath2 19 & 320 @xmath2 34 & @xmath30 + 195 & j032904.06 + 311446.5 & 1.32 & 0+i & 0.63 @xmath2 0.05 & 1.5 @xmath2 0.12 & 1.7 @xmath2 0.10 & 1.8 @xmath2 0.12 & 83 @xmath2 8.8 & @xmath30 + 198 & j032904.68 + 311659.0 & -0.46 & ii & 11 @xmath2 0.83 & 12 @xmath2 0.78 & 12 @xmath2 0.79 & 20 @xmath2 1.1 & 41 @xmath2 4.5 & @xmath30 + 199 & j032904.73 + 311134.9 & -0.62 & ii & 14 @xmath2 0.94 & 17 @xmath2 1.1 & 16 @xmath2 0.89 & 19 @xmath2 1.1 & 17 @xmath2 1.8 & @xmath30 + 200 & j032904.95 + 312038.4 & 0.06 & flat & 13 @xmath2 0.95 & 16 @xmath2 1.0 & 19 @xmath2 1.1 & 28 @xmath2 1.5 & 110 @xmath2 12 & @xmath30 + 201 & j032905.18 + 312036.9 & -0.98 & ii & 2.9 @xmath2 0.27 & 4.3 @xmath2 0.27 & 5.3 @xmath2 0.34 & 7.8 @xmath2 0.42 & 38 @xmath2 6.0 & @xmath30 + 202 & j032905.78 + 311639.6 & -1.08 & ii & 470 @xmath2 37 & 460 @xmath2 33 & 410 @xmath2 27 & 500 @xmath2 30 & 450 @xmath2 48 & @xmath30 + 204 & j032906.33 + 311346.4 & -1.14 & ii & 48 @xmath2 3.4 & 47 @xmath2 3.1 & 43 @xmath2 2.5 & 52 @xmath2 3.1 & 55 @xmath2 5.9 & @xmath30 + 205 & j032907.78 + 312157.3 & 2.30 & 0+i & 1000 @xmath2 110 & 2400 @xmath2 230 & 3900 @xmath2 360 & 3400 @xmath2 330 & @xmath30 & 50000 @xmath2 5300 + 206 & j032907.96 + 312251.4 & -1.16 & ii & 130 @xmath2 10 & 120 @xmath2 8.6 & 120 @xmath2 8.0 & 160 @xmath2 13 & 110 @xmath2 31 & @xmath30 + 207 & j032908.97 + 312256.1 & -0.21 & flat & 43 @xmath2 3.2 & 54 @xmath2 3.5 & 56 @xmath2 4.6 & 75 @xmath2 16 & 140 @xmath2 31 & @xmath30 + 208 & j032909.10 + 312305.5 & -0.33 & ii & 22 @xmath2 1.6 & 23 @xmath2 1.5 & 23 @xmath2 2.9 & 70 @xmath2 16 & 240 @xmath2 46 & @xmath30 + 209 & j032909.10 + 312128.7 & 1.05 & 0+i & 39 @xmath2 3.3 & 110 @xmath2 6.9 & 170 @xmath2 9.8 & 170 @xmath2 13 & 430 @xmath2 63 & @xmath30 + 210 & j032909.34 + 312104.1 & -0.51 & ii & 5.9 @xmath2 0.44 & 4.7 @xmath2 0.38 & 5.1 @xmath2 0.50 & 8.6 @xmath2 0.76 & 41 @xmath2 5.6 & @xmath30 + 211 & j032909.40 + 311413.8 & -0.87 & ii & 2.9 @xmath2 0.20 & 7.3 @xmath2 0.51 & 8.9 @xmath2 0.45 & 6.0 @xmath2 0.35 & 3.6 @xmath2 0.50 & @xmath30 + 212 & j032909.49 + 312720.9 & -1.10 & ii & 5.2 @xmath2 0.39 & 4.7 @xmath2 0.35 & 4.2 @xmath2 0.32 & 6.0 @xmath2 0.39 & 7.4 @xmath2 0.85 & @xmath30 + 213 & j032909.65 + 312256.3 & -1.60 & ii & 96 @xmath2 6.9 & 68 @xmath2 5.7 & 56 @xmath2 6.0 & 87 @xmath2 16 & 120 @xmath2 29 & @xmath30 + 214 & j032910.47 + 312334.7 & -1.41 & ii & 10 @xmath2 0.72 & 8.9 @xmath2 0.62 & 8.3 @xmath2 0.60 & 9.5 @xmath2 0.87 & 13 @xmath2 3.3 & @xmath30 + 215 & j032910.49 + 311331.0 & 2.49 & 0+i & @xmath30 & 0.11 @xmath2 0.01 & 0.17 @xmath2 0.03 & @xmath30 & 37 @xmath2 3.9 & 34000 @xmath2 3600 + 216 & j032910.65 + 311340.0 & 0.42 & 0+i & 0.07 @xmath2 0.01 & 1.3 @xmath2 0.08 & 0.74 @xmath2 0.05 & 0.37 @xmath2 0.06 & 5.5 @xmath2 0.92 & @xmath30 + 217 & j032910.68 + 311820.6 & 1.84 & 0+i & 2.6 @xmath2 0.26 & 9.0 @xmath2 0.89 & 13 @xmath2 1.1 & 14 @xmath2 1.0 & 890 @xmath2 94 & 21000 @xmath2 2300 + 218 & j032910.84 + 311642.6 & -0.32 & ii & 8.0 @xmath2 0.65 & 10 @xmath2 0.69 & 13 @xmath2 0.76 & 22 @xmath2 1.2 & 28 @xmath2 2.9 & @xmath30 + 220 & j032911.26 + 311831.4 & 1.79 & 0+i & 1.0 @xmath2 0.11 & 5.2 @xmath2 0.35 & 6.4 @xmath2 0.42 & 5.4 @xmath2 0.36 & 600 @xmath2 63 & @xmath30 + 221 & j032911.89 + 312127.0 & 0.32 & 0+i & 23 @xmath2 1.7 & 34 @xmath2 2.0 & 41 @xmath2 2.5 & 51 @xmath2 4.0 & 60 @xmath2 20 & @xmath30 + 222 & j032912.06 + 311305.4 & 0.85 & 0+i & 0.04 @xmath2 0.01 & 1.3 @xmath2 0.09 & 0.80 @xmath2 0.05 & 0.39 @xmath2 0.11 & 21 @xmath2 2.9 & 12000 @xmath2 1300 + 223 & j032912.06 + 311301.7 & 0.07 & flat & 0.98 @xmath2 0.07 & 17 @xmath2 0.91 & 6.2 @xmath2 0.32 & 4.5 @xmath2 0.28 & 44 @xmath2 4.6 & @xmath30 + 224 & j032912.97 + 311814.3 & 1.56 & 0+i & 66 @xmath2 5.2 & 240 @xmath2 13 & 340 @xmath2 18 & 420 @xmath2 24 & 850 @xmath2 89 & 4400 @xmath2 470 + 225 & j032913.14 + 312252.8 & -1.33 & ii & 140 @xmath2 11 & 140 @xmath2 9.3 & 120 @xmath2 8.1 & 120 @xmath2 8.4 & 77 @xmath2 8.3 & @xmath30 + 226 & j032913.54 + 311358.2 & 2.50 & 0+i & 0.04 @xmath2 0 & @xmath30 & 0.32 @xmath2 0.05 & @xmath30 & 32 @xmath2 3.5 & 4000 @xmath2 550 + 227 & j032914.40 + 311444.1 & -0.70 & ii & 4.4 @xmath2 0.35 & 16 @xmath2 1.5 & 8.1 @xmath2 0.65 & 6.7 @xmath2 0.50 & 12 @xmath2 1.3 & @xmath30 + 228 & j032916.61 + 312349.4 & -1.18 & ii & 32 @xmath2 2.3 & 31 @xmath2 2.1 & 32 @xmath2 2.1 & 43 @xmath2 2.5 & 21 @xmath2 3.1 & @xmath30 + 229 & j032916.69 + 311618.2 & -2.43 & iii & 24 @xmath2 1.7 & 17 @xmath2 1.5 & 12 @xmath2 1.3 & 7.2 @xmath2 1.3 & 1.9 @xmath2 0.31 & @xmath30 + 230 & j032916.83 + 312325.1 & -0.57 & ii & 3.0 @xmath2 0.22 & 2.7 @xmath2 0.20 & 2.9 @xmath2 0.22 & 4.2 @xmath2 0.25 & 23 @xmath2 2.7 & @xmath30 + 231 & j032917.17 + 312746.5 & 1.69 & 0+i & 0.6 @xmath2 0.06 & 1.6 @xmath2 0.16 & 1.5 @xmath2 0.13 & 1.1 @xmath2 0.14 & 130 @xmath2 14 & 2000 @xmath2 300 + 233 & j032917.68 + 312245.0 & -0.94 & ii & 450 @xmath2 34 & 430 @xmath2 31 & 490 @xmath2 33 & 520 @xmath2 36 & 580 @xmath2 61 & @xmath30 + 234 & j032917.78 + 311948.0 & -1.23 & ii & 5.2 @xmath2 0.38 & 4.5 @xmath2 0.33 & 3.9 @xmath2 0.31 & 4.4 @xmath2 0.33 & 5.5 @xmath2 0.61 & @xmath30 + 235 & j032918.26 + 312319.9 & 1.09 & 0+i & 1.3 @xmath2 0.10 & 3.8 @xmath2 0.2 & 8.0 @xmath2 0.38 & 14 @xmath2 0.72 & 68 @xmath2 7.7 & @xmath30 + 236 & j032918.67 + 312017.7 & 0.16 & flat & 1.6 @xmath2 0.12 & 1.8 @xmath2 0.11 & 1.8 @xmath2 0.13 & 3.6 @xmath2 0.21 & 25 @xmath2 2.7 & @xmath30 + 237 & j032918.74 + 312325.4 & -0.66 & ii & 52 @xmath2 4.0 & 56 @xmath2 3.9 & 67 @xmath2 3.9 & 120 @xmath2 7.0 & 210 @xmath2 23 & @xmath30 + 238 & j032920.06 + 312407.5 & 0.34 & 0+i & 43 @xmath2 3.4 & 59 @xmath2 3.6 & 76 @xmath2 5.0 & 120 @xmath2 9.9 & 430 @xmath2 45 & @xmath30 + 239 & j032920.44 + 311834.2 & -0.25 & flat & 130 @xmath2 11 & 160 @xmath2 10 & 150 @xmath2 9.4 & 210 @xmath2 12 & 510 @xmath2 54 & 210 @xmath2 30 + 240 & j032921.57 + 312110.3 & -1.24 & ii & 12 @xmath2 0.88 & 1.1 @xmath2 0.80 & 9.2 @xmath2 0.75 & 11 @xmath2 0.80 & 16 @xmath2 1.8 & @xmath30 + 241 & j032921.87 + 311536.2 & -1.34 & ii & 120 @xmath2 8.4 & 120 @xmath2 8.3 & 87 @xmath2 6.8 & 93 @xmath2 7.2 & 95 @xmath2 9.9 & @xmath30 + 242 & j032923.17 + 312030.2 & -1.03 & ii & 19 @xmath2 1.5 & 18 @xmath2 1.3 & 15 @xmath2 1.2 & 20 @xmath2 1.3 & 31 @xmath2 3.3 & @xmath30 + 243 & j032923.25 + 312653.1 & -1.17 & ii & 8.5 @xmath2 0.65 & 8.5 @xmath2 0.59 & 7.7 @xmath2 0.55 & 8.9 @xmath2 0.59 & 7.6 @xmath2 0.84 & @xmath30 + 245 & j032924.09 + 311957.6 & -0.13 & flat & 3.0 @xmath2 0.22 & 2.7 @xmath2 0.19 & 2.7 @xmath2 0.20 & 4.6 @xmath2 0.26 & 77 @xmath2 8.2 & @xmath30 + 246 & j032925.93 + 312640.1 & -0.93 & ii & 100 @xmath2 8.7 & 120 @xmath2 8.2 & 140 @xmath2 8.4 & 210 @xmath2 13 & 150 @xmath2 16 & @xmath30 + 247 & j032926.81 + 312647.6 & -1.81 & iii & 53 @xmath2 4.0 & 35 @xmath2 3.3 & 25 @xmath2 2.9 & 19 @xmath2 2.8 & 20 @xmath2 2.1 & @xmath30 + 249 & j032929.27 + 311834.7 & -0.98 & ii & 18 @xmath2 1.4 & 12 @xmath2 1.1 & 9.1 @xmath2 0.93 & 8.5 @xmath2 0.93 & 110 @xmath2 11 & 290 @xmath2 34 + 250 & j032929.80 + 312102.6 & -1.15 & ii & 18 @xmath2 1.4 & 14 @xmath2 1.1 & 15 @xmath2 1.1 & 21 @xmath2 1.3 & 31 @xmath2 3.3 & @xmath30 + 251 & j032930.40 + 311903.3 & -1.11 & ii & 18 @xmath2 1.3 & 17 @xmath2 1.3 & 17 @xmath2 1.2 & 20 @xmath2 1.3 & 29 @xmath2 3.1 & 120 @xmath2 29 + 252 & j032932.57 + 312436.9 & -1.22 & ii & 27 @xmath2 2.0 & 24 @xmath2 1.7 & 18 @xmath2 1.5 & 19 @xmath2 1.5 & 21 @xmath2 2.2 & @xmath30 + 253 & j032932.88 + 312712.6 & -1.58 & ii & 6.3 @xmath2 0.48 & 5.2 @xmath2 0.41 & 4.4 @xmath2 0.37 & 4.2 @xmath2 0.38 & 2.6 @xmath2 0.31 & @xmath30 + 254 & j032937.73 + 312202.5 & -0.97 & ii & 3.7 @xmath2 0.26 & 3.0 @xmath2 0.23 & 2.6 @xmath2 0.22 & 3.2 @xmath2 0.24 & 7.9 @xmath2 0.87 & @xmath30 + 255 & j032944.16 + 311947.3 & -0.76 & ii & 1.8 @xmath2 0.13 & 2.4 @xmath2 0.16 & 2.4 @xmath2 0.15 & 2.6 @xmath2 0.15 & 2.1 @xmath2 0.28 & @xmath30 + 258 & j032954.04 + 312052.9 & -0.87 & ii & 34 @xmath2 2.6 & 34 @xmath2 2.3 & 31 @xmath2 2.2 & 39 @xmath2 2.4 & 120 @xmath2 13 & 200 @xmath2 22 llcccccccc 121 & j032519.52 + 303424.2 & -0.14 & flat & 3.6 @xmath2 0.26 & 4.4 @xmath2 0.27 & 5.3 @xmath2 0.31 & 7.0 @xmath2 0.38 & 13 @xmath2 1.4 & @xmath30 + 122 & j032522.32 + 304513.9 & 2.21 & 0+i & 0.47 @xmath2 0.06 & 4.4 @xmath2 0.39 & 6.6 @xmath2 0.46 & 14 @xmath2 0.75 & 550 @xmath2 59 & 13000 @xmath2 1400 + 123 & j032536.22 + 304515.7 & 1.51 & 0+i & 0.69 @xmath2 0.26 & 3.8 @xmath2 0.5 & 5.1 @xmath2 0.5 & 6.6 @xmath2 0.67 & 220 @xmath2 33 & @xmath30 + 124 & j032536.49 + 304522.2 & 2.56 & 0+i & 1.0 @xmath2 0.22 & 15 @xmath2 0.99 & 57 @xmath2 2.8 & 180 @xmath2 9.3 & 4500 @xmath2 480 & 25000 @xmath2 2600 + 125 & j032538.83 + 304406.2 & 2.05 & 0+i & 4.3 @xmath2 0.39 & 16 @xmath2 1.6 & 20 @xmath2 1.5 & 26 @xmath2 2.1 & 2100 @xmath2 220 & 32000 @xmath2 3400 + 126 & j032539.12 + 304358.2 & 2.00 & 0+i & 3.4 @xmath2 0.27 & 31 @xmath2 1.9 & 94 @xmath2 4.7 & 170 @xmath2 8.7 & 1400 @xmath2 160 & @xmath30 + 127 & j032637.47 + 301528.1 & 1.03 & 0+i & 5.2 @xmath2 0.42 & 13 @xmath2 0.87 & 14 @xmath2 0.98 & 16 @xmath2 1.1 & 460 @xmath2 48 & 4400 @xmath2 600 + 128 & j032738.25 + 301358.6 & -0.40 & ii & 90 @xmath2 6.3 & 97 @xmath2 6.9 & 100 @xmath2 7.0 & 120 @xmath2 8.7 & 390 @xmath2 41 & 1100 @xmath2 160 + 130 & j032739.08 + 301303.1 & 2.46 & 0+i & 0.83 @xmath2 0.07 & 12 @xmath2 0.93 & 25 @xmath2 1.4 & 32 @xmath2 1.6 & 2000 @xmath2 210 & 21000 @xmath2 2400 + 131 & j032741.47 + 302016.8 & -1.06 & ii & 66 @xmath2 4.6 & 55 @xmath2 4.2 & 49 @xmath2 3.8 & 59 @xmath2 4.1 & 120 @xmath2 12 & 160 @xmath2 19 + 132 & j032743.23 + 301228.9 & 2.23 & 0+i & 1.1 @xmath2 0.10 & 7.3 @xmath2 0.50 & 15 @xmath2 0.83 & 37 @xmath2 1.7 & 880 @xmath2 92 & 7600 @xmath2 800 + 133 & j032747.67 + 301204.5 & -0.25 & flat & 590 @xmath2 46 & 690 @xmath2 50 & 1000 @xmath2 56 & 1700 @xmath2 89 & 2000 @xmath2 210 & 1800 @xmath2 200 + 134 & j032800.09 + 300847.0 & -1.13 & ii & 54 @xmath2 3.8 & 46 @xmath2 3.4 & 35 @xmath2 3.0 & 33 @xmath2 3.0 & 92 @xmath2 9.7 & 160 @xmath2 20 + 135 & j032800.39 + 300801.3 & 0.98 & 0+i & 19 @xmath2 1.3 & 36 @xmath2 2.0 & 57 @xmath2 2.8 & 92 @xmath2 4.5 & 430 @xmath2 45 & 700 @xmath2 76 + 137 & j032834.49 + 310051.1 & 0.88 & 0+i & 79 @xmath2 5.5 & 160 @xmath2 10 & 280 @xmath2 14 & 320 @xmath2 17 & 1400 @xmath2 140 & 2800 @xmath2 300 + 139 & j032834.94 + 305454.5 & -0.06 & flat & 2.7 @xmath2 0.19 & 3.5 @xmath2 0.22 & 4.7 @xmath2 0.25 & 6.3 @xmath2 0.33 & 14 @xmath2 1.4 & @xmath30 + 140 & j032835.03 + 302009.9 & -0.20 & flat & 2.2 @xmath2 0.14 & 1.6 @xmath2 0.13 & 1.3 @xmath2 0.12 & 1.8 @xmath2 0.14 & 48 @xmath2 5.1 & 140 @xmath2 18 + 146 & j032842.41 + 302953.2 & -1.05 & ii & 33 @xmath2 2.3 & 30 @xmath2 2.3 & 29 @xmath2 2.1 & 36 @xmath2 2.4 & 43 @xmath2 4.5 & @xmath30 + 150 & j032845.30 + 310541.9 & 1.09 & 0+i & 1.7 @xmath2 0.13 & 3.6 @xmath2 0.26 & 3.4 @xmath2 0.24 & 3.6 @xmath2 0.25 & 240 @xmath2 26 & 1500 @xmath2 210 + 155 & j032850.62 + 304244.7 & -0.36 & ii & 13 @xmath2 0.9 & 13 @xmath2 0.89 & 12 @xmath2 0.8 & 16 @xmath2 0.95 & 71 @xmath2 7.5 & 76 @xmath2 1.0 + 161 & j032852.17 + 304505.5 & -0.78 & ii & 300 @xmath2 22 & 350 @xmath2 30 & 410 @xmath2 23 & 520 @xmath2 30 & 340 @xmath2 36 & @xmath30 + 193 & j032903.87 + 305629.8 & -0.44 & ii & 1.4 @xmath2 0.10 & 1.3 @xmath2 0.09 & 1.1 @xmath2 0.09 & 1.3 @xmath2 0.09 & 13 @xmath2 1.4 & @xmath30 + 196 & j032904.12 + 305612.8 & -0.97 & ii & 4.2 @xmath2 0.28 & 3.6 @xmath2 0.27 & 3.2 @xmath2 0.26 & 3.9 @xmath2 0.28 & 9.2 @xmath2 1.0 & @xmath30 + 203 & j032906.05 + 303039.2 & 0.64 & 0+i & 0.89 @xmath2 0.06 & 1.3 @xmath2 0.08 & 2.3 @xmath2 0.14 & 5.3 @xmath2 0.26 & 24 @xmath2 2.5 & 110 @xmath2 19 + 244 & j032923.48 + 313329.5 & 1.53 & 0+i & 1.1 @xmath2 0.11 & 3.1 @xmath2 0.23 & 3.2 @xmath2 0.31 & 4.0 @xmath2 0.28 & 220 @xmath2 23 & 1200 @xmath2 160 + 248 & j032928.88 + 305841.9 & -1.05 & ii & 5.0 @xmath2 0.35 & 4.6 @xmath2 0.33 & 4.7 @xmath2 0.33 & 7.5 @xmath2 0.43 & 14 @xmath2 1.5 & @xmath30 + 256 & j032946.00 + 310439.0 & -0.89 & ii & 33 @xmath2 2.3 & 29 @xmath2 2.1 & 25 @xmath2 2.0 & 25 @xmath2 2.0 & 42 @xmath2 4.4 & 96 @xmath2 14 + 257 & j032951.82 + 313906.0 & 3.40 & 0+i & @xmath30 & 0.03 @xmath2 0.01 & 0.13 @xmath2 0.04 & 0.49 @xmath2 0.05 & 56 @xmath2 6.0 & 2000 @xmath2 290 + 259 & j033015.14 + 302349.4 & 1.58 & 0+i & 10 @xmath2 0.76 & 26 @xmath2 1.3 & 28 @xmath2 1.4 & 59 @xmath2 2.9 & 2100 @xmath2 220 & 5700 @xmath2 610 + 260 & j033022.45 + 313240.5 & 0.28 & flat & 1.7 @xmath2 0.12 & 2.2 @xmath2 0.13 & 2.8 @xmath2 0.17 & 5.7 @xmath2 0.28 & 21 @xmath2 2.2 & @xmath30 + 261 & j033024.08 + 311404.4 & -1.20 & ii & 5.3 @xmath2 0.37 & 4.2 @xmath2 0.35 & 3.9 @xmath2 0.32 & 5.2 @xmath2 0.36 & 5.5 @xmath2 0.61 & @xmath30 + 262 & j033025.97 + 310217.9 & -2.44 & iii & 75 @xmath2 5.3 & 48 @xmath2 4.4 & 32 @xmath2 3.9 & 20 @xmath2 3.8 & 4.8 @xmath2 0.53 & @xmath30 + 263 & j033027.14 + 302829.8 & -0.16 & flat & 22 @xmath2 1.4 & 20 @xmath2 1.3 & 14 @xmath2 1.1 & 12 @xmath2 1.1 & 500 @xmath2 52 & 950 @xmath2 100 + 264 & j033032.68 + 302626.6 & 2.66 & 0+i & 0.09 @xmath2 & 0.27 @xmath2 0 & 0.24 @xmath2 0.05 & 0.21 @xmath2 0.04 & 36 @xmath2 3.7 & 320 @xmath2 44 + 265 & j033035.47 + 311558.6 & -0.60 & ii & 2.0 @xmath2 0.14 & 1.6 @xmath2 0.12 & 1.5 @xmath2 0.13 & 2.0 @xmath2 0.14 & 20 @xmath2 2.1 & 130 @xmath2 20 + 266 & j033035.92 + 303024.4 & -1.21 & ii & 720 @xmath2 56 & 560 @xmath2 47 & 520 @xmath2 42 & 550 @xmath2 44 & 600 @xmath2 64 & 180 @xmath2 20 + 267 & j033036.97 + 303127.7 & -0.97 & ii & 260 @xmath2 18 & 280 @xmath2 18 & 310 @xmath2 18 & 430 @xmath2 24 & 410 @xmath2 42 & 120 @xmath2 16 + 268 & j033038.20 + 303211.9 & -0.02 & flat & 1.6 @xmath2 0.11 & 1.3 @xmath2 0.10 & 1.1 @xmath2 0.10 & 0.92 @xmath2 0.09 & 64 @xmath2 6.8 & 250 @xmath2 30 + 269 & j033043.98 + 303247.0 & -0.89 & ii & 410 @xmath2 33 & 390 @xmath2 28 & 380 @xmath2 28 & 430 @xmath2 28 & 780 @xmath2 83 & 810 @xmath2 86 + 270 & j033052.51 + 305417.8 & -0.86 & ii & 49 @xmath2 3.5 & 47 @xmath2 3.2 & 44 @xmath2 3.0 & 71 @xmath2 4.0 & 120 @xmath2 13 & 160 @xmath2 21 + 271 & j033110.68 + 304940.6 & -0.92 & ii & 31 @xmath2 2.2 & 29 @xmath2 2.1 & 30 @xmath2 2.1 & 39 @xmath2 2.5 & 64 @xmath2 6.8 & 82 @xmath2 13 + 272 & j033114.70 + 304955.4 & -0.07 & flat & 55 @xmath2 3.9 & 67 @xmath2 4.2 & 81 @xmath2 4.6 & 110 @xmath2 5.5 & 260 @xmath2 28 & 230 @xmath2 26 + 273 & j033118.30 + 304939.5 & -1.01 & ii & 230 @xmath2 17 & 210 @xmath2 16 & 200 @xmath2 15 & 280 @xmath2 18 & 340 @xmath2 35 & 200 @xmath2 25 + 274 & j033120.11 + 304917.7 & -0.89 & ii & 9.8 @xmath2 0.68 & 8.8 @xmath2 0.64 & 8.3 @xmath2 0.61 & 9.9 @xmath2 0.66 & 16 @xmath2 1.7 & @xmath30 + 275 & j033120.98 + 304530.1 & 1.48 & 0+i & 0.19 @xmath2 & 1.1 @xmath2 0.09 & 1.1 @xmath2 0.09 & 0.82 @xmath2 0.08 & 17 @xmath2 1.7 & 4000 @xmath2 540 + 276 & j033128.87 + 303053.3 & -1.36 & ii & 220 @xmath2 16 & 190 @xmath2 14 & 160 @xmath2 13 & 140 @xmath2 13 & 140 @xmath2 15 & 210 @xmath2 26 + 277 & j033131.17 + 304411.1 & -1.46 & ii & 1.6 @xmath2 0.22 & 5.3 @xmath2 0.48 & 2.3 @xmath2 0.23 & 2.4 @xmath2 0.24 & 1.9 @xmath2 0.30 & @xmath30 + 278 & j033142.40 + 310624.9 & -1.48 & ii & 45 @xmath2 3.2 & 35 @xmath2 2.7 & 26 @xmath2 2.3 & 24 @xmath2 2.4 & 29 @xmath2 3.1 & @xmath30 + 279 & j033217.96 + 304947.5 & 0.86 & 0+i & 0.56 @xmath2 0.11 & 0.56 @xmath2 0.19 & @xmath30 & 1.7 @xmath2 0.14 & 15 @xmath2 1.6 & 1700 @xmath2 180 + 280 & j033229.17 + 310240.8 & 0.16 & flat & 16 @xmath2 1.2 & 23 @xmath2 1.8 & 29 @xmath2 2.0 & 43 @xmath2 3.0 & 210 @xmath2 22 & 1100 @xmath2 160 + 281 & j033232.99 + 310221.7 & -1.12 & ii & 83 @xmath2 6.0 & 79 @xmath2 5.7 & 66 @xmath2 4.8 & 78 @xmath2 5.4 & 90 @xmath2 9.5 & @xmath30 + 282 & j033234.05 + 310055.8 & -1.02 & ii & 140 @xmath2 9.7 & 120 @xmath2 9.0 & 100 @xmath2 7.7 & 120 @xmath2 8.8 & 340 @xmath2 36 & 300 @xmath2 36 + 284 & j033241.70 + 311046.3 & -0.78 & ii & 37 @xmath2 2.7 & 33 @xmath2 2.3 & 29 @xmath2 2.2 & 55 @xmath2 3.8 & 130 @xmath2 14 & 97 @xmath2 13 + 285 & j033247.20 + 305916.3 & -1.27 & ii & 530 @xmath2 35 & 480 @xmath2 34 & 440 @xmath2 28 & 500 @xmath2 31 & 360 @xmath2 39 & 210 @xmath2 28 + 286 & j033257.84 + 310608.3 & 0.33 & 0+i & 1.2 @xmath2 0.08 & 1.3 @xmath2 0.08 & 1.4 @xmath2 0.10 & 1.6 @xmath2 0.1 & 17 @xmath2 1.7 & 83 @xmath2 14 + 287 & j033306.41 + 310804.6 & 0.17 & flat & 10 @xmath2 0.73 & 13 @xmath2 0.79 & 14 @xmath2 0.83 & 28 @xmath2 1.3 & 110 @xmath2 11 & 45 @xmath2 9.9 + 288 & j033309.56 + 310531.2 & 0.70 & 0+i & 0.76 @xmath2 0.05 & 0.67 @xmath2 0.05 & 0.71 @xmath2 0.07 & 2.0 @xmath2 0.12 & 89 @xmath2 9.3 & 100 @xmath2 14 + 289 & j033312.84 + 312124.2 & 0.16 & flat & 430 @xmath2 32 & 590 @xmath2 42 & 810 @xmath2 45 & 1400 @xmath2 78 & 4600 @xmath2 500 & 3400 @xmath2 370 + 290 & j033313.80 + 312005.3 & 1.31 & 0+i & 0.86 @xmath2 0.06 & 1.8 @xmath2 0.10 & 3.2 @xmath2 0.17 & 7.1 @xmath2 0.37 & 68 @xmath2 7.1 & 57 @xmath2 18 + 291 & j033314.38 + 310710.9 & 2.72 & 0+i & 0.23 @xmath2 0 & 1.3 @xmath2 0.09 & 2.5 @xmath2 0.19 & 5.5 @xmath2 0.39 & 130 @xmath2 14 & 830 @xmath2 120 + 292 & j033316.44 + 310652.5 & 1.56 & 0+i & @xmath30 & 0.35 @xmath2 0.05 & 0.56 @xmath2 0.06 & @xmath30 & 23 @xmath2 2.4 & 1800 @xmath2 270 + 293 & j033316.65 + 310755.2 & 1.69 & 0+i & 23 @xmath2 1.7 & 66 @xmath2 3.5 & 110 @xmath2 5.2 & 160 @xmath2 7.2 & 2000 @xmath2 210 & 6600 @xmath2 710 + 294 & j033317.85 + 310931.9 & 2.90 & 0+i & 0.24 @xmath2 0.02 & 8.6 @xmath2 0.44 & 46 @xmath2 2.3 & 130 @xmath2 6.7 & 770 @xmath2 82 & 12000 @xmath2 1300 + 295 & j033320.32 + 310721.5 & 0.61 & 0+i & 6.0 @xmath2 0.42 & 23 @xmath2 1.1 & 33 @xmath2 1.5 & 42 @xmath2 2.1 & 200 @xmath2 21 & 360 @xmath2 45 + 296 & j033327.29 + 310710.2 & 1.82 & 0+i & 3.4 @xmath2 0.24 & 9.0 @xmath2 0.46 & 10 @xmath2 0.51 & 16 @xmath2 0.79 & 1800 @xmath2 200 & 5500 @xmath2 590 + 297 & j033330.41 + 311050.6 & -1.04 & ii & 1300 @xmath2 91 & 1300 @xmath2 92 & 1200 @xmath2 80 & 1200 @xmath2 81 & 980 @xmath2 100 & 520 @xmath2 59 + 298 & j033341.29 + 311341.0 & -0.93 & ii & 100 @xmath2 7.2 & 100 @xmath2 6.7 & 90 @xmath2 5.8 & 95 @xmath2 6.2 & 170 @xmath2 19 & 180 @xmath2 25 + 299 & j033346.92 + 305350.1 & -1.17 & ii & 4.9 @xmath2 0.34 & 4.1 @xmath2 0.31 & 3.7 @xmath2 0.31 & 3.9 @xmath2 0.31 & 6.2 @xmath2 0.67 & @xmath30 + 300 & j033351.07 + 311227.8 & -2.16 & iii & 42 @xmath2 3.0 & 29 @xmath2 2.4 & 19 @xmath2 2.1 & 13 @xmath2 2.1 & 5.5 @xmath2 0.60 & @xmath30 + 301 & j033401.66 + 311439.8 & -1.16 & ii & 100 @xmath2 7.1 & 95 @xmath2 6.8 & 100 @xmath2 6.7 & 120 @xmath2 7.5 & 110 @xmath2 11 & 91 @xmath2 16 + 302 & j033430.78 + 311324.4 & -1.25 & ii & 15 @xmath2 1.1 & 13 @xmath2 0.96 & 13 @xmath2 0.93 & 15 @xmath2 1.0 & 13 @xmath2 1.4 & @xmath30 + 303 & j033449.84 + 311550.3 & -1.48 & ii & 76 @xmath2 5.8 & 51 @xmath2 4.7 & 37 @xmath2 4.1 & 37 @xmath2 4.2 & 52 @xmath2 5.4 & 40 @xmath2 11 + 304 & j033915.81 + 312430.7 & -1.07 & ii & 22 @xmath2 1.5 & 19 @xmath2 1.3 & 14 @xmath2 1.1 & 19 @xmath2 1.3 & 34 @xmath2 3.6 & @xmath30 + 305 & j034001.49 + 311017.3 & -1.54 & ii & 15 @xmath2 1.1 & 12 @xmath2 0.9 & 9.0 @xmath2 0.79 & 8.4 @xmath2 0.8 & 7.5 @xmath2 0.82 & @xmath30 + 306 & j034109.13 + 314437.9 & -0.89 & ii & 670 @xmath2 49 & 570 @xmath2 48 & 540 @xmath2 39 & 520 @xmath2 39 & 930 @xmath2 98 & 2200 @xmath2 230 + 307 & j034114.11 + 315946.3 & -2.09 & iii & 50 @xmath2 3.5 & 32 @xmath2 3.0 & 23 @xmath2 2.7 & 19 @xmath2 2.7 & 9.6 @xmath2 1.0 & @xmath30 + 308 & j034119.19 + 320203.8 & -1.45 & ii & 23 @xmath2 1.7 & 21 @xmath2 1.5 & 17 @xmath2 1.3 & 16 @xmath2 1.3 & 10 @xmath2 1.1 & @xmath30 + 309 & j034124.42 + 315327.9 & -1.06 & ii & 4.0 @xmath2 0.28 & 3.9 @xmath2 0.26 & 3.5 @xmath2 0.25 & 3.7 @xmath2 0.27 & 3.0 @xmath2 0.42 & @xmath30 + 310 & j034139.17 + 313610.7 & -1.15 & ii & 51 @xmath2 3.5 & 36 @xmath2 3.1 & 30 @xmath2 2.9 & 53 @xmath2 3.6 & 150 @xmath2 16 & @xmath30 + 311 & j034141.09 + 314804.6 & -0.16 & flat & 4.7 @xmath2 0.33 & 6.6 @xmath2 0.39 & 8.0 @xmath2 0.45 & 11 @xmath2 0.55 & 13 @xmath2 1.4 & @xmath30 + 312 & j034153.26 + 315019.2 & -0.38 & ii & 4.0 @xmath2 0.29 & 4.3 @xmath2 0.28 & 4.7 @xmath2 0.29 & 6.8 @xmath2 0.38 & 12 @xmath2 1.3 & @xmath30 + 313 & j034155.71 + 314811.4 & 0.28 & flat & 24 @xmath2 1.7 & 32 @xmath2 1.9 & 42 @xmath2 2.3 & 62 @xmath2 3.2 & 140 @xmath2 15 & 250 @xmath2 43 + 314 & j034157.44 + 314836.7 & -1.20 & ii & 580 @xmath2 48 & 530 @xmath2 39 & 480 @xmath2 34 & 480 @xmath2 34 & 330 @xmath2 34 & 180 @xmath2 29 + 315 & j034157.75 + 314800.8 & -1.11 & ii & 92 @xmath2 6.7 & 85 @xmath2 5.9 & 77 @xmath2 5.3 & 98 @xmath2 6.2 & 140 @xmath2 15 & @xmath30 + 316 & j034158.67 + 314821.4 & 0.47 & 0+i & 2.9 @xmath2 0.2 & 4.4 @xmath2 0.26 & 6.2 @xmath2 0.34 & 10 @xmath2 0.5 & 43 @xmath2 4.4 & @xmath30 + 317 & j034201.01 + 314913.4 & -1.42 & ii & 7.8 @xmath2 0.55 & 6.4 @xmath2 0.48 & 5.4 @xmath2 0.45 & 5.4 @xmath2 0.45 & 5.3 @xmath2 0.61 & @xmath30 + 318 & j034202.17 + 314802.1 & 1.23 & 0+i & 1.3 @xmath2 0.09 & 1.6 @xmath2 0.10 & 2.9 @xmath2 0.17 & 9.1 @xmath2 0.45 & 140 @xmath2 14 & 240 @xmath2 36 + 319 & j034204.34 + 314711.6 & -1.02 & ii & 19 @xmath2 1.4 & 18 @xmath2 1.3 & 17 @xmath2 1.2 & 22 @xmath2 1.4 & 25 @xmath2 2.6 & @xmath30 + 320 & j034210.69 + 314705.6 & -1.10 & ii & 6.6 @xmath2 0.46 & 6.2 @xmath2 0.4 & 6.1 @xmath2 0.38 & 7.4 @xmath2 0.45 & 5.0 @xmath2 0.62 & @xmath30 + 321 & j034219.27 + 314327.0 & -0.88 & ii & 84 @xmath2 5.8 & 74 @xmath2 5.2 & 64 @xmath2 4.6 & 79 @xmath2 5.2 & 140 @xmath2 15 & 110 @xmath2 25 + 322 & j034220.32 + 320531.0 & -1.26 & ii & 15 @xmath2 1.1 & 12 @xmath2 0.95 & 9.2 @xmath2 0.85 & 12 @xmath2 0.94 & 20 @xmath2 2.1 & @xmath30 + 323 & j034223.33 + 315742.7 & -2.08 & iii & 52 @xmath2 3.7 & 34 @xmath2 3.1 & 25 @xmath2 2.9 & 22 @xmath2 2.8 & 9.4 @xmath2 1.0 & @xmath30 + 324 & j034227.12 + 314432.9 & -1.22 & ii & 21 @xmath2 1.5 & 15 @xmath2 1.2 & 11 @xmath2 1.0 & 8.7 @xmath2 1.0 & 56 @xmath2 6.0 & @xmath30 + 325 & j034232.10 + 315249.5 & -1.43 & ii & 1.8 @xmath2 0.12 & 1.4 @xmath2 0.11 & 1.2 @xmath2 0.12 & 1.5 @xmath2 0.12 & 1.1 @xmath2 0.28 & @xmath30 + 326 & j034232.91 + 314220.6 & -1.19 & ii & 39 @xmath2 2.8 & 33 @xmath2 2.4 & 26 @xmath2 2.1 & 27 @xmath2 2.2 & 51 @xmath2 5.4 & @xmath30 + 327 & j034233.13 + 315214.7 & -1.07 & ii & 3.9 @xmath2 0.27 & 3.1 @xmath2 0.23 & 2.4 @xmath2 0.22 & 2.9 @xmath2 0.24 & 7.7 @xmath2 0.85 & @xmath30 + 328 & j034234.19 + 315101.0 & -1.19 & ii & 3.4 @xmath2 0.24 & 2.5 @xmath2 0.20 & 1.7 @xmath2 0.19 & 1.7 @xmath2 0.19 & 7.1 @xmath2 0.79 & @xmath30 + 329 & j034236.47 + 315517.6 & -1.21 & ii & 3.1 @xmath2 0.22 & 2.4 @xmath2 0.19 & 2.3 @xmath2 0.19 & 2.5 @xmath2 0.24 & 3.6 @xmath2 0.45 & @xmath30 + 330 & j034244.50 + 315958.7 & -1.49 & ii & 2.7 @xmath2 0.18 & 2.3 @xmath2 0.17 & 2.0 @xmath2 0.15 & 1.8 @xmath2 0.16 & 1.1 @xmath2 0.28 & @xmath30 + 331 & j034249.18 + 315011.2 & -1.54 & ii & 26 @xmath2 1.9 & 20 @xmath2 1.6 & 17 @xmath2 1.4 & 17 @xmath2 1.5 & 15 @xmath2 1.6 & @xmath30 + 332 & j034254.67 + 314345.3 & -2.22 & iii & 26 @xmath2 1.8 & 17 @xmath2 1.5 & 12 @xmath2 1.4 & 7.5 @xmath2 1.4 & 3.6 @xmath2 0.55 & @xmath30 + 333 & j034255.95 + 315842.0 & -0.84 & ii & 510 @xmath2 38 & 430 @xmath2 35 & 450 @xmath2 32 & 650 @xmath2 40 & 190 @xmath2 20 & 2400 @xmath2 270 + 334 & j034256.05 + 315644.8 & 0.06 & flat & 4.0 @xmath2 0.29 & 6.2 @xmath2 0.36 & 9.2 @xmath2 0.48 & 18 @xmath2 0.87 & 13 @xmath2 1.3 & @xmath30 + 335 & j034301.94 + 314435.6 & -1.15 & ii & 3.9 @xmath2 0.27 & 3.6 @xmath2 0.25 & 3.5 @xmath2 0.25 & 4.0 @xmath2 0.27 & 3.4 @xmath2 0.49 & @xmath30 + 336 & j034306.77 + 314820.5 & -2.40 & iii & 25 @xmath2 1.8 & 17 @xmath2 1.4 & 11 @xmath2 1.3 & 7.1 @xmath2 1.2 & 2.0 @xmath2 0.32 & @xmath30 + 338 & j034321.47 + 314246.3 & -1.50 & ii & 6.9 @xmath2 0.48 & 5.5 @xmath2 0.42 & 4.6 @xmath2 0.38 & 5.2 @xmath2 0.41 & 3.3 @xmath2 0.45 & @xmath30 + 339 & j034322.22 + 314613.6 & -1.15 & ii & 36 @xmath2 2.5 & 33 @xmath2 2.3 & 27 @xmath2 2.0 & 31 @xmath2 2.1 & 38 @xmath2 4.0 & @xmath30 + 346 & j034344.48 + 314309.3 & -0.50 & ii & 130 @xmath2 11 & 230 @xmath2 13 & 240 @xmath2 14 & 330 @xmath2 17 & 550 @xmath2 59 & 470 @xmath2 53 + 376 & j034406.47 + 314325.1 & -2.39 & iii & 22 @xmath2 1.7 & 16 @xmath2 1.3 & 10 @xmath2 1.2 & 7.1 @xmath2 1.1 & 2.1 @xmath2 0.30 & @xmath30 + 415 & j034427.90 + 322718.9 & -1.18 & ii & 25 @xmath2 1.8 & 24 @xmath2 1.6 & 18 @xmath2 1.4 & 25 @xmath2 1.6 & 29 @xmath2 3.1 & @xmath30 + 442 & j034435.34 + 322837.2 & -0.16 & flat & 0.42 @xmath2 0.04 & 1.3 @xmath2 0.07 & 0.74 @xmath2 0.08 & 1.1 @xmath2 0.11 & 3.9 @xmath2 0.47 & @xmath30 + 465 & j034443.06 + 313733.7 & -1.29 & ii & 7.4 @xmath2 0.58 & 5.8 @xmath2 0.51 & 5.0 @xmath2 0.43 & 5.4 @xmath2 0.45 & 9.5 @xmath2 1.0 & @xmath30 + 474 & j034452.00 + 322625.4 & -1.62 & iii & 14 @xmath2 1.3 & 12 @xmath2 0.87 & 8.6 @xmath2 0.77 & 7.5 @xmath2 0.73 & 5.7 @xmath2 0.64 & @xmath30 + 489 & j034513.51 + 322434.8 & -1.14 & ii & 5.9 @xmath2 0.41 & 4.7 @xmath2 0.36 & 3.6 @xmath2 0.33 & 3.8 @xmath2 0.33 & 5.5 @xmath2 0.63 & @xmath30 + 496 & j034533.47 + 314555.3 & -1.24 & ii & 4.1 @xmath2 0.31 & 3.7 @xmath2 0.27 & 3.1 @xmath2 0.26 & 3.2 @xmath2 0.25 & 4.9 @xmath2 0.57 & @xmath30 + 498 & j034536.83 + 322557.0 & -1.32 & ii & 140 @xmath2 12 & 120 @xmath2 9.0 & 84 @xmath2 7.9 & 97 @xmath2 7.9 & 200 @xmath2 21 & @xmath30 + 500 & j034548.27 + 322412.0 & -0.97 & ii & 1300 @xmath2 110 & 1200 @xmath2 91 & 930 @xmath2 83 & 990 @xmath2 78 & 3000 @xmath2 320 & 5500 @xmath2 580 + 501 & j034558.25 + 322647.5 & -1.39 & ii & 8.9 @xmath2 0.73 & 7.8 @xmath2 0.66 & 7.0 @xmath2 0.58 & 6.4 @xmath2 0.55 & 4.9 @xmath2 0.58 & @xmath30 + 502 & j034657.38 + 324917.4 & -1.36 & ii & 7.1 @xmath2 0.50 & 5.9 @xmath2 0.45 & 4.7 @xmath2 0.41 & 4.3 @xmath2 0.4 & 5.7 @xmath2 0.64 & @xmath30 + 503 & j034658.51 + 324658.9 & -1.54 & ii & 8.3 @xmath2 0.58 & 6.1 @xmath2 0.50 & 5.2 @xmath2 0.47 & 5.5 @xmath2 0.48 & 4.3 @xmath2 0.51 & @xmath30 + 504 & j034705.43 + 324308.5 & 0.51 & 0+i & 89 @xmath2 6.3 & 110 @xmath2 7.1 & 130 @xmath2 7.4 & 170 @xmath2 9.7 & 690 @xmath2 73 & 930 @xmath2 100 + 505 & j034741.58 + 325144.1 & 1.30 & 0+i & 560 @xmath2 49 & 1100 @xmath2 65 & 1800 @xmath2 92 & 2200 @xmath2 130 & 3900 @xmath2 410 & 11000 @xmath2 1200 all of perseus & 73.5 & 369 & 1.00 & 5.0 & 70 & 0.19 & 32 & 0.09 & 231 & 0.63 & 36 & 0.10 & 3.3 + ic 348 & 6.85 & 143 & 0.39 & 21 & 11 & 0.08 & 5 & 0.03 & 104 & 0.73 & 23 & 0.16 & 9.5 + ngc 1333 & 3.24 & 104 & 0.28 & 32 & 28 & 0.27 & 12 & 0.12 & 59 & 0.57 & 5 & 0.05 & 2.1 + rc & 63.4 & 122 & 0.33 & 1.9 & 31 & 0.25 & 15 & 0.12 & 68 & 0.56 & 8 & 0.07 & 2.2 + b1 & 7.62 & 22 & 0.06 & 2.9 & 10 & 0.45 & 3 & 0.14 & 8 & 0.36 & 1 & 0.05 & 0.8 + l1455 & 7.62 & 11 & 0.03 & 1.4 & 5 & 0.45 & 2 & 0.18 & 4 & 0.36 & 0 & 0.00 & 0.8 + eastern perseus & 31.0 & 193 & 0.52 & 6.2 & 15 & 0.08 & 9 & 0.05 & 140 & 0.73 & 29 & 0.15 & 9.3 + western perseus & 42.5 & 176 & 0.48 & 4.1 & 55 & 0.31 & 23 & 0.13 & 91 & 0.52 & 7 & 0.04 & 1.7 | the _ spitzer space telescope _ mapped the perseus molecular cloud complex with irac and mips as part of the c2d _ spitzer _ legacy project .
this paper combines the observations from both instruments giving an overview of low - mass star formation across perseus from 3.6 to 70 .
we provide an updated list of young stellar objects with new classifications and source fluxes from previous works , identifying 369 ysos in perseus with the _ spitzer _ dataset . by synthesizing the irac and mips maps of perseus and building on the work of previous papers in this series @xcite , we present a current census of star formation across the cloud and within smaller regions .
67% of the ysos are associated with the young clusters ngc 1333 and ic 348 .
the majority of the star formation activity in perseus occurs in the regions around the clusters , to the eastern and western ends of the cloud complex .
the middle of the cloud is nearly empty of ysos despite containing regions of high visual extinction .
the western half of perseus contains three - quarters of the total number of embedded ysos ( class 0+i and flat sed sources ) in the cloud and nearly as many embedded ysos as class ii and iii sources .
class ii and iii greatly outnumber class 0+i objects in eastern perseus and ic 348 .
these results are consistent with previous age estimates for the clusters . across the cloud , 56% of ysos and 91% of the class 0+i and flat sources are in areas where a@xmath0 @xmath1 5 mag , indicating a possible extinction threshold for star formation . |
while the @xmath3-factories have been highly successful , it is useful to ask what further insights could be achieved with up to two orders of magnitude more data .
most straightforward are extensions of measurements and searches that are currently statistically limited . some of the more prominent examples are @xmath11 ( which strongly constrains @xmath7 ) , the dalitz analyses used to obtain @xmath8 , radiative @xmath12 and @xmath13 decays , and lepton - flavor - violating tau decays . more intriguing , though , is the sensitivity of this region to new particles and forces .
the standard model ( sm ) rates for decays involving loop processes are generally suppressed due to ckm cancellation .
furthermore , the @xmath5 asymmetries of many accessible decay modes are expected to be small .
previously unknown processes at higher mass scales and with complex couplings could thus contribute substantially to such decays and be revealed through precision measurements of rates and @xmath5 asymmetries . in the data accumulated to date ,
more than 1.4 ab@xmath14 from belle and babar combined , there are a number of deviations from expectation that hint at the possibility of new physics . by increasing the data two orders of magnitude
, one can study rare decays and improve precision to the point of probing physics well into the tev mass scale .
kek has proposed super - kekb , an upgrade of kekb , with a goal of accumulating 50 ab@xmath14 .
the lhc will commence operations later this year , and many significant measurements on @xmath3 s can be expected to come out of lhc - b as well as the atlas and cms experiments .
the primary advantages of the @xmath3-factory experiments for such measurements are the ability to detect photons and @xmath15 s and their so - called `` hermeticity , '' coverage of the detector over nearly @xmath16 solid angle to contain the well - specified @xmath2 annihilation event and thus enable identification of neutrinos .
the priority of the proposed physics program thus emphasizes measurements for which the @xmath3-factory has a clear advantage : @xmath5 asymmetries in @xmath12 and @xmath13 penguin decays , lepton universality tests in decays such as @xmath17 and @xmath18 , and @xmath5 asymmetries in @xmath19 mixing .
, ( left ) penguin and ( right ) mixing + penguin.,title="fig:",height=60 ] , ( left ) penguin and ( right ) mixing + penguin.,title="fig:",height=60 ] from @xmath12 penguin decays.,width=264 ] @xmath3 decays to a @xmath5-eigenstate exhibit @xmath5 asymmetry due to a non - zero complex phase angle between the amplitudes for the non - mixed and mixed processes . for @xmath20 ,
this is illustrated in figure [ fig : bmixing ] .
for the unmixed `` penguin '' process ( left ) @xmath21 , the dominant amplitude is proportional to the product @xmath22 , or @xmath23 in the wolfenstein parametrization@xcite
. the decay may also proceed through a mixing oscillation , @xmath24 ( figure [ fig : bmixing](right ) ) , for which the amplitude is proportional to @xmath25 .
the penguin parts of the two amplitudes differ by a particle conjugation , which in this case involves no relative complex phase .
the hadronic components of the two are identical .
the amplitude of mixing is well - measured and understood and includes the complex phase of @xmath26 which , expressed in terms of the angles of the unitarity triangle , is @xmath27 ( or @xmath28 ) .
the result is a @xmath5-dependent decay rate @xmath29 , \label{eqn : cpasym}\end{aligned}\ ] ] where @xmath30 for a @xmath31 , @xmath32 if @xmath5 is even ( odd ) , @xmath33 is the mass difference between the two @xmath5-eigenstates , and @xmath34 is the time interval from creation to the @xmath5-eigenstate decay . for
the @xmath20 decays the penguin process involves three amplitudes , where the internal fermion in the loop is @xmath35 , @xmath36 , or @xmath37 , and the unitarity of the ckm matrix would cause them to cancel were it not for the high mass of the @xmath37 .
the decay rate is thus suppressed , so that other forces with weaker but unsuppressed couplings could contribute to observable anomalous decay rates and @xmath5 asymmetries .
the sm @xmath5-asymmetry in particular is theoretically robust , so a precise measurement will have sensitivity to new physics processes . as a function of integrated luminosity.,width=245 ]
the heavy flavor averaging group ( hfag ) maintains a compilation of the world average asymmetry , measured as `` @xmath38 '' ( or `` @xmath39 '' ) , over all @xmath12 hadronic penguin decays ( figure [ fig : hfag ] ) .
a nave averaging of the values gives @xmath40 , which agrees to 2.2@xmath41 or 3% confidence level with the value measured in @xmath42 decays , @xmath43 .
possible reasons for the difference ( @xmath44 ) include statistical fluctuations or systematic errors , uncounted theoretical corrections , and new physics .
the resolution of this possible inconsistency will require substantial additional data .
, ( right ) @xmath45 . ,
title="fig:",width=132 ] , ( right ) @xmath45 . ,
title="fig:",width=105 ] $ ] still allowed ( white , 95.5% cl ) after belle measurement of @xmath46 with 0.41 ab@xmath14 .
( bottom ) allowed region if @xmath46 with 50 ab@xmath14 is consistent with the sm .
note the change in vertical scale .
, title="fig:",width=185 ] + $ ] still allowed ( white , 95.5% cl ) after belle measurement of @xmath46 with 0.41 ab@xmath14 .
( bottom ) allowed region if @xmath46 with 50 ab@xmath14 is consistent with the sm .
note the change in vertical scale .
, title="fig:",width=166 ] within the standard model the charged current coupling to leptons is universal . for decays such as @xmath47 and @xmath48 ( figure [ fig : taunufd ] )
, a deviation of the ratio of the @xmath49 and @xmath50 modes from unity would thus indicate new physics .
for @xmath51 the sm partial width is theoretically robust , so that agreement between the measured and theoretical branching fractions can be used to constrain the mass of the charged higgs in some models .
the measured branching fraction for this channel@xcite is consistent with the sm . in the two - higgs - doublet model
the rate is enhanced by a factor that depends on the higgs mass and on @xmath52 , the ratio of vacuum expectation values of the two higgs doublets@xcite . in figure
[ fig : taunu](top ) the region in @xmath52 and higgs mass that is not yet ruled out is shown in white .
much of this allowed region will be explored with 50 ab@xmath14 of @xmath0(4s ) data ( figure [ fig : taunu](bottom ) ) . the standard model predicts branching fractions of @xmath53 , @xmath54 , and @xmath55 for @xmath56 , @xmath57 , and @xmath58 , respectively .
the muonic decay will likely be observed with @xmath59 ab@xmath14 of data .
the decay @xmath60 was recently observed at belle , with @xmath61 @xcite .
its ratio with the corresponding muonic decay , which has been precisely measured for many years , is sensitive to the charged higgs mass .
the @xmath62 and @xmath51 processes ( figure [ fig : taunufd ] ) have identical couplings to @xmath49 but differ on the quark side , so that a comparison of their behavior can also be used to distinguish between different new physics scenarios , if a deviation is observed . for a @xmath63 decay with a @xmath5-eigenstate in the final state hadronic component ,
the standard model final state is _ not _ a @xmath5-eigenstate because the photon is left- ( right- ) handed for the @xmath12 ( @xmath64 ) process .
the expected @xmath5-asymmetry is thus much smaller than for gluonic penguin @xmath12 modes , of the order of a few percent rather than @xmath65 .
any sizable @xmath5-asymmetry in such modes is thus a signature of right - handed currents in the @xmath12 transition .
the expected precision in the asymmetry @xmath66 as a function of integrated luminosity , extrapolated from belle s result for @xmath67@xcite , is shown in figure [ fig : btosgam_cp ] .
-asymmetry as a function of integrated luminosity for decays @xmath67 ( red ) , @xmath68 ( green ) and other @xmath67 ( blue ) , extrapolated from belle result on @xmath67@xcite , with 465 fb@xmath14.,height=170 ] mixing parameters , @xmath69 , with @xmath70 , and ( bottom ) @xmath5 violation parameters , @xmath71 , with @xmath72 , based on 75 ab@xmath14 of simulated data .
, title="fig:",height=188 ] + mixing parameters , @xmath69 , with @xmath70 , and ( bottom ) @xmath5 violation parameters , @xmath71 , with @xmath72 , based on 75 ab@xmath14 of simulated data .
, title="fig:",height=188 ] the charm sector has long been acknowledged as a window to new physics because of the strong suppression of loop rates and @xmath5 asymmetry that is rooted in the unitarity of the ckm matrix . with the recent observation of mixing in the @xmath9@xcite ,
it is clear that the @xmath3-factories have great sensitivity to rare charm processes and @xmath5 asymmetries .
the parameters pertaining to mixing are @xmath73 and @xmath74 , where @xmath33 and @xmath75 are the differences between the masses and widths of the mass eigenstates , respectively , and @xmath76 is the average width . shown in figure [ fig : dmixing ] are likelihood contours for projected measurements in @xmath77 and in the @xmath5-violation parameters @xmath78 and @xmath79 , based on simulations of 75 ab@xmath14 .
with a goal toward exploring beyond the horizon into the tev mass range as well as leveraging current successes , a second - generation @xmath3-factory has been proposed at kek .
the plan includes major upgrades of the kekb collider and belle detector .
the luminosity goals of `` superkekb '' are @xmath80 instantaneous and @xmath81 integrated luminosities , to be achieved through multiple incremental improvements , each of which is designed to roughly double the luminosity yield .
they include improvements to the positron damping ring , superconducting rf , beam pipe design , crab cavities , and energy exchange c - band .
while it is hard to guarantee performance in a beam collider , the kekb team has an excellent record of meeting and exceeding expectations . in the current plan , which is presented as a baseline , the upgrade
is phased so that it can be achieved with a continuation of the current level of funding . assuming a 3-year shutdown for construction starting in 2009 and 200 days per year of operations
, it will be possible to collect 50ab@xmath14 by 2028 .
a crucial step for kekb is the implementation of `` crab '' cavities , to rotate the beams , which cross at a small angle , so that the beam overlap is maximized and collisions are `` head - on . ''
crab cavities were installed in the winter of 2007 and have been operating since spring 2007 . thus far
the principle of the crab design has been validated through the consistency of achieved specific luminosity with the predictions of simulations .
the maximum luminosity during the recent run has been @xmath82 , approximately equal to the maximum obtained without crab cavities . to achieve the projected factor of two gain in luminosity
, kekb is working to improve the maximum stable beam current .
increased beam current is accompanied by increases in background particles and events and synchrotron radiation , which in turn increase the rates to the detector for general radiation damage , detector occupancy , fake hits , pile - up , and triggers .
figure [ fig : backgroundprojection ] shows the predicted rates for several types of background , based on projected currents and luminosities , in the major sectors of the belle detector .
most will experience a five - fold increase over current levels during early operations of the upgraded ring , increasing to approximately twenty - fold at maximum luminosity .
the upgrades to the belle detector will focus on improving hadron identification and background tolerance .
an upgrade task force has been investigating these questions for several years .
a letter of intent signed by nearly 300 physicsts from 61 institutions was submitted to kek in 2004@xcite .
kek has launched a call for the formation of a new collaboration , temporarily named sbelle , to build and operate the upgraded detector .
the upgraded detector must be able to operate with a twenty - fold increase in background particles and a fifty - fold increase in real event rate over current conditions .
modules targeted for upgrade include data acquisition , silicon tracker , central drift chamber , particle i d device , calorimeter , and @xmath83 detector .
we describe briefly the current concept for each .
the silicon inner tracker is a critical component for vertexing but is at the same time exposed to high radiation doses and experiences high occupancy .
the upgrade design includes two additional outer layers which will have short strips to limit occupancy , fast electronics with reduced readout time ( from 800 ns to 50 ns ) , and a pipelined readout .
as with silicon strip detectors in most current experiments , it is designed to be replaced periodically , and in a later version the inner layers will be closer to the beam line , 10 mm as opposed to the current 15 mm , and consist of pixel detectors , which are less vulnerable to radiation damage .
the central drift chamber will occupy a larger volume , having a larger inner radius ( 77 mm @xmath84 160 mm ) to accommodate the new silicon tracker and an outer radius expanding into space previously occupied by aerogel and time - of - flight detectors ( 880 mm @xmath84 1140 mm ) . to reduce occupancy each drift cell will be smaller , with an attendant increase in the number of sense wires to 15,140 from the current 8,400 .
while the aerogel threshold erenkov and time - of - flight counters have served belle well , there is strong motivation to achieve better hadron separation for sbelle ; for example , @xmath85 separation at momenta in the 2 - 4 gev/@xmath36 range will need significant improvement to bring out a @xmath13 signal above @xmath12 . to accomplish this ,
the current threshold erenkov / tof system in the barrel region will be replaced .
all of the designs under consideration involve the collection of erenkov light from quartz glass radiators with sensitivity to the erenkov angle .
the time - of - propagation ( top ) device measures the arrival time of photons , which depends on the time - of - flight from interaction point to the radiator and on the angle - dependent path length of the photon traveling the length of the bar via total internal reflection .
the `` imaging top '' concept adds a focusing mirror on one end and a photon collection array at the other that corrects for chromaticity and improves the performance .
the `` focusing dirc '' concept measures the angle more directly , through angle - dependent imaging of the light .
in addition to a new barrel detector , a ring - imaging aerogel detector is planned for the endcap region . while the current electromagnetic calorimeter , consisting of tl - doped csi crystals viewed by silicon photodiodes , achieves excellent resolution ,
modifications will be needed to maintain resolution at higher luminosities . in the barrel
the signal waveform will be sampled and fitted to avoid deterioration of resolution .
the relatively slow response ( @xmath86s ) of tl - doped csi would result in degradation of signals in the endcap region due to pileup , the superposition of unrelated pulses over many beam crossings .
the existing endcap will thus be replaced with pure csi crystals , which have a rise time @xmath87ns , viewed by photomultipliers , which are well matched to the spectrum and speed of the csi response .
the belle @xmath83 detector , which consists of 5 cm steel plates interspersed with glass resistive plate chambers ( rpc ) running in streamer mode , has separate barrel and endcap systems .
the expected rates at the outer barrel region are sufficiently low that the existing chambers can be used at super - kekb if run in avalanche mode .
the rates in the endcap regions will be beyond the capacity of the rpc s , so it has been proposed that the endcaps be replaced with plastic scintillator strips , with each layer in a 2-dimensional grid .
the `` kek roadmap '' outlines the long - range plan of the lab on a time scale of 10 + years .
the roadmap report released on january 4 , 2008 , by the director a. suzuki and the kek management has included the kekb luminosity and detector upgrades ; this report was affirmed by the external roadmap review committe in march 2008 .
a key feature in this preliminary version of the plan is that it accomplishes the upgrade with a flat funding profile . according to the plan
, kekb and belle will end datataking in 2009 , and the activity will shift to super - kekb / detector construction for a three - year period , after which super - kekb operations will commence .
the initial luminosity goal is @xmath88 .
additional rf cavities will be installed as funding permits , raising luminosity over several years to @xmath89 . while details are still under discussion between kek and the japanese funding agency ( mext ) , the appearance of the kekb upgrade as an official item in the roadmap is highly significant .
the next step in the project will be to assemble an international collaboration to build and operate sbelle .
any institution is welcome to join and will have excellent opportunities to share in the construction and physics bounty .
interested investigators are invited to contact anyone in the interim steering committee .
the @xmath3-factories will have collected during their lifetime , 1999 - 2009 , over 1.4 billion @xmath3 pairs at the @xmath0(4s ) resonance . based on these data ,
the ckm matrix has been firmly established as the origin of @xmath5 asymmetry in the weak interaction .
it has been possible to make multiple independent measurements of parameters to test the unitarity of the ckm matrix , including the three phase angles @xmath90 , @xmath7 , and @xmath8 , as well as @xmath91 , with a precision that would have been considered optimistic a decade ago .
some internal inconsistencies in the global fit of ckm , as well as mysteries such as unexpected differences in @xmath5 asymmetry among @xmath92 modes , may hint at breaks in the standard model . by increasing the data by two orders of magnitude , one can study rare decays and improve precision to the point of probing physics into the tev mass scale . the kekb upgrade , which will ultimately run at @xmath89 luminosity will focus on @xmath5-asymmetries in @xmath3 penguin decays , tests of lepton universality , lepton flavor violation , and precision measurement of all aspects of the ckm matrix . the project has been included in the kek roadmap in 2008 , a major step in bringing it to reality .
the proposed detector upgrade , currently named sbelle , has been under study for several years .
the technical design will be finalized in the next year , and a new international collaboration is being formed . | the belle experiment at the kekb electron - positron collider is expected to have collected close to one billion @xmath0(4s ) events by the time it comes to an end in 2009 .
an upgrade to kekb has been proposed .
it is designed for an order of magnitude higher luminosity than kekb , following a three - year construction period .
the ultimate goal of @xmath1 luminosity would be reached through further improvements over several years . to exploit the physics accessible through this improved luminosity ,
an upgrade of the belle detector is also planned .
a new international collaboration , temporarily named sbelle , is in the process of being formed .
super - kekb and sbelle were officially placed on the kek 5-year roadmap in early 2008 .
the richness of the physics in the upsilon region in @xmath2 collisions is well - known ; the first upsilon resonance was discovered in the late 70 s and , by the time the @xmath3-factories started up in 1999 , hundreds of associated papers had been published .
the upsilon mine has yet to be depleted ; for example , the belle collaboration has submitted 265 articles for publication in peer - reviewed journals as of june 2008 .
the most influential among these include many investigations of the @xmath4 matrix , including measurements of the @xmath5-violating phase parameter @xmath6 and constraints on @xmath7 and @xmath8 , observation of new charmonia and charmonium - like states , discovery of @xmath9 mixing , and many probes for `` new physics . ''
the events , which are clean , numerous , and triggered on with high efficiency , contain @xmath3-mesons , many types of charm hadrons , tau leptons , 2-photon events , upsilons ( @xmath0(4s ) , @xmath0(10860 ) , @xmath0(3s ) , @xmath0(1s ) ) , and @xmath10 mesons . among the physics topics addressed through them are @xmath5 violation and other weak physics , qcd , heavy quark spectroscopy , lepton flavor violation , and dark matter . in
2009 the belle experiment at kekb will conclude , having collected approximately one billion @xmath0(4s ) events and achieved , even surpassed , its original scientific objectives .
a proposed upgrade , with luminosity two orders of magnitude greater than kekb , is in the process of approval and would extend the heavy quark program by more than a decade , starting in 2012 . |
computing the charateristic numbers of curves in projective spaces is a classical problem in algebraic geometry : how many curves in @xmath0 of given degree and genus that pass through a general set of linear subspaces , and are tangent to a general set of hyperplanes ?
these types of problems have provided a wealth of inspiration for classical algebraic geometers , such as schubert , zenthen , mailard , and have been one of the main drives of the development of intersection theory .
they have achieved remarkable success despite the technology of their time ( in mathematical and literal sense ) .
for example , schubert has correctly computed the number of twisted cubics in @xmath1 that are tangent to @xmath2 quadric surfaces ( @xmath3 ) , and this represents the summit of enumerative geometry at the time ( see @xcite ) .
however , not until recently are characteristic numbers computed in certain generality , as all of the results obtained by classical algebraic geometers are bounded by degree .
moduli space of stable maps provide a powerful tools to attack characteristic numbers problems of curves with low genus .
the case of rational space curve of any degree was solved in @xcite .
the author actually gave algorithms to compute intersections of divisors on @xmath4 and derived characteristic numbers as a corollary .
however , if one is only interested in the characteristic numbers , then a small tweak would give a much faster and easier to implement algorithm .
the case of elliptic plane curves , any degree , was solve in @xcite .
the difficulty to extend that result to @xmath5 was due to two factors .
firstly , there was yet a result calculating incidence - only ( no tangency ) numbers for elliptic space curves in any dimensional projective space . secondly , because the tangency divisor on @xmath6 is related to the divisor of elliptic curves with fixed @xmath7-invariant , hence one also needs to compute the characteristic numbers of elliptic curves with fixed @xmath7-invariant .
the latter was solved in a separate paper by the author ( @xcite ) .
the algorithms for characteristic numbers of genus two plane curves were given in @xcite , although no actual numbers were shown .
in this paper , we solve the full characteristic numbers problem for elliptic curves in projective spaces by using the moduli space of stable maps .
note that , even in the incidence - only case our result is already new .
for example , no incidence - only characteristic numbers for elliptic curves in @xmath8 and @xmath9 have been computed .
the number of elliptic curves in @xmath1 were computed in @xcite and the proof was scheduled to appear in another paper . however , the author of this paper was unable to locate it .
the incidence - only numbers of elliptic curves in @xmath1 were also computed in @xcite . in that paper , only the numbers upto degree @xmath10 elliptic curves were computed , but the method could compute numbers of any degree .
the incidence - only numbers of degree @xmath10 elliptic curves in @xmath1 were first computed in @xcite using a classical argument .
the approach of this paper is as follows .
we first compute incidence - only numbers for elliptic space curves using getzler relation .
this is analogous to the enumeration of rational space curves using wdvv equation , but with a small twist . first we will not be using the entire moduli space of stable maps @xmath11 as intersecting enumerative classes with the virtual fundamental class will have unwanted contributions from other components ( this space is not irreducible ) .
in fact , it was claimed in @xcite that the actual count of elliptic curves is a linear combination of genus one and genus zero gromov - witten invariants .
we will only use the main component , that is the closure of the locus of maps with smooth source curves , denoted by @xmath12 . for our purpose ,
knowledge of the maps on the loci where the main component intersect other components is required , and this is desribed in section @xmath10 . in the case of rational curve ,
the wdvv equation is a rational equivalence of divisors , hence we can pull back via the forgetful morphism and obtain a relation on the stable map space .
this is no longer true for the genus one stable map space .
the reason is that getzler relation is a rational equivalence of codimension @xmath13 strata on @xmath14 , and that the forgetful morphism @xmath15 have fibre dimensions that could jump : for example , the preimage of the stratum @xmath16 , has two components , one of which is of codimension @xmath13 , and the other is a weil divisor .
we can get around this as follows .
first we use enough enumerative constraints to cut down the space @xmath17 into a @xmath13-dimensional family .
then we pushforward via the forgetful morphism , and then intersect with getzler s relation . as a result
we obtain a relation of the enumerative invariants of elliptic curves with that of rational curves , rational cuspidal curves , and elliptic curves with fixed @xmath7 invariants .
the first is well - known , and the latter two were computed in @xcite and @xcite respectively . to go from incidence - only to full characteristic numbers , we study the relation between the tangency divisor and the incident and boundary divisors .
we will not obtain a rational equivalence , due to the present of enumeratively irrelevant divisors , those that are intersections of @xmath17 with other components .
however , we obtain a numerical equivalence whenever we intersect with only curves in @xmath17 that has empty intersection with the irrelevant divisors .
this is true for @xmath18-dimensional families of elliptic curves cut down by enumerative constraints , so this is enough for our purpose .
the structure of this paper is as follows . in section @xmath13
, we introduce basic notions such as various stacks of stable maps and the enumerative constraints . in section
@xmath19 we review getzler s relation and give an example of using the relation to count elliptic curves . in section @xmath10 , we give the recusion computing incidence - only numbers of elliptic space curves based on getzler s relation . in section @xmath20
, we give the recursions computing full characteristic numbers of elliptic space curves .
we end with some tables with numerical examples in section @xmath21 .
the author would like to thank ravi vakil for many helpful suggestions and some of the arguments used in this paper , and for suggisting this problem .
we denote @xmath4 the kontsevich compactification of the moduli space of genus zero curves with @xmath22 marked points of degree @xmath23 in @xmath0 .
let @xmath17 be the main component of the moduli space of stable maps of genus one , that is , the closure of the locus of maps with smooth domains .
we will also use the notation @xmath24 and @xmath25 where the markings are indexed by a set @xmath26 .
we will be concerned with the number of curves passing through a constraint , and each constraint is denoted by a @xmath27-tuple @xmath28 as follows : + * ( i ) * @xmath29 is the number of hyperplanes that the curves need to be tangent to .
+ * ( ii ) * for @xmath30 , @xmath31 is the number of subspaces of codimension @xmath32 that the curves need to pass through . in @xcite and @xcite , the constraints may have @xmath33 coordinates because we want to impose conditions on the node or cusp , but we will not need that here . note that because in general a curve of degree @xmath23 will always intersect a hyperplane at @xmath23 points , introducing an incident condition with a hyperplane has the same effect as that of multiplying the enumerative number by @xmath23 .
for example , if we ask how many genus zero curves of degree @xmath10 in @xmath1 that pass through the constraint @xmath34 , that means we ask how many genus zero curves of degree @xmath10 pass through three lines , four points , are tangent to one hyperplane , and then multiply that answer by @xmath35 .
we will also refer to @xmath28 as a set of linear spaces , hence we can say , pick a space @xmath36 in @xmath28 .
we consider the following ordering on the set of constraints , in order to prove that our algorithm will terminate later on .
let @xmath37\cdot i^2 $ ] , and this will be our rank function .
we compare two constraints @xmath38 using the following criteria , whose priority are in the following order .
we only proceed to using the next criterion if using the current one give us a tie . * if @xmath39 then @xmath40 . * if @xmath41 and @xmath28 has fewer non - hyperplane elements than @xmath42 does , then @xmath43 . *
if @xmath44 then @xmath40 . informally speaking , characteristic numbers where the constraints are more spread out at two ends are computed first in the recursion .
we write @xmath45 if @xmath46 as a parition of the set of linear spaces in @xmath47 we list the following definitions of stacks of stable maps that will occur in our recursions .
+ + * 1 ) * let @xmath49 be the usual moduli space of genus zero stable maps , @xmath50 . + + * 2 ) * let @xmath51 be the closure in @xmath52 of the locus of maps of smooth rational curves @xmath53 such that @xmath54 .
informally , @xmath51 parametrizes degree @xmath23 rational nodal curves in @xmath0 .
+ + * 3 ) * let @xmath55 be the closure in @xmath56 of the locus of maps of smooth rational curves @xmath53 such that the differential @xmath57 is zero .
informally , @xmath58 parametrizes degree @xmath23 rational cuspidal curves in @xmath59 + + * 4 ) * let @xmath60 be the main component of the moduli space of genus one stable maps @xmath61 .
that is @xmath62 .
+ + * 5 ) * for @xmath63 let @xmath64 be @xmath65 where the fibre product is taken over evaluation maps @xmath66 to @xmath59 + + * 6 ) * similarly we can define @xmath67 , @xmath68 ( see figure 1 ) .
+ + * 7 ) * for @xmath69 , let @xmath70 be the closure in @xmath71 ( the projections are evaluation maps @xmath72 ) of the locus of maps @xmath53 such that @xmath54 .
we call maps in this family rational two - nodal reducible curves . + + * 8) * for @xmath73 , let @xmath74 be @xmath75 .
similarly , we can define @xmath76 for @xmath77 .
+ + * 9 ) * we define @xmath78 to be the closure in @xmath60 of the locus of maps whose domains are smooth and have a fixed but generic @xmath7-invariant . the enumerative geometry of this stack is studied in @xcite .
+ + + @xmath79{stackcusp.pdf}\ ] ] + @xmath80 @xmath81 @xmath81 @xmath81 @xmath81 @xmath81 @xmath79{stackell.pdf}\ ] ] + @xmath82 let @xmath83 be a stack of stable maps of curves into @xmath0 . for a constraint @xmath28 , we define @xmath84 be the closure in @xmath85 of the locus of maps that satisfy the constraint @xmath28 . for maps of reducible source curves ,
tangency condition include the case where the image of the node lies on the tangency hyperplane , as the intersection multiplicity is @xmath13 in this case . for a stack @xmath83 that is supported on a finite number of points
then we denote @xmath86 to be the stack - theoretic length of @xmath83 .
if @xmath83 is a closed substack of the stacks @xmath87 then we denote @xmath88 to be the closure in @xmath83 of the locus of maps @xmath53 such that the restriction of @xmath53 on the @xmath89th component satisfies constraint @xmath90 , and that the image of the node lies on a codimension @xmath91 subspace .
we use the notation @xmath92 if we do nt want to distinguish the conditions on each component .
if @xmath91 is @xmath93 we omit it from the notation .
if @xmath83 is a closed substack of the stack @xmath94 then we denote @xmath95 to be the closure in @xmath83 of the locus of maps @xmath53 such that the restriction of @xmath53 on the @xmath89th component satisfies constraint @xmath90 .
we use the notation @xmath96 .
if @xmath83 is a closed substack of the stacks @xmath97 then we denote @xmath98 to be the closure in @xmath83 of the locus of maps @xmath53 such that the restriction of @xmath53 on the @xmath89th component satisfies constraint @xmath90 .
we use the notation @xmath96 if we do nt want to distinguish the conditions on each component .
the enumerative geometry of all the stacks defined above are known , except for the stacks that involve @xmath99 .
but the enumerative geometry of the stacks @xmath100 can be easily deduced from that of @xmath99 ( see @xcite , section @xmath19 ) . note the small difference between the notation here and in @xcite , @xcite , as we have remove some conditions on the nodes as they are not necessary .
we review some of the basic intersection theory on @xmath14 and especially getzler s relation . for a more complete treatment see @xcite .
consider the moduli space @xmath14 with the @xmath101 action by permuting the marked points . in @xcite ,
the following @xmath101-invariant codimension @xmath13 strata are defined : @xmath102{getzler.pdf}\ ] ] @xmath103 getzler computed the intersection matrix of these cycles and found a non - trivial null vector @xmath104 these cycles are understood stack - theoretically .
pandharipande gave a beautiful argument in @xcite showing that the above relation is actually a rational equivalence , by using the auxilliary space of admissible covers .
note that it is easy to get relations on @xmath105 induced by wdvv .
let @xmath106 be the boundary stratum of irreducible nodal curves on @xmath105 .
we have a map @xmath107 by identifying the last two marked points . pushing forward the wdvv relation on @xmath108 yields a relation on @xmath105 .
unfortunately , these type of relations does not help us counting elliptic curves cause the general curves in the strata involed in the relations do not have a smooth elliptic component .
now as an example , we use getzler s relation to compute a trivial example . there is @xmath18 elliptic cubic in @xmath1 that passes through @xmath19 points and @xmath21 lines . _
the @xmath19 point conditions determine the plane of the elliptic cubic .
the other @xmath21 conditions translate to @xmath21 other point conditions for a smooth plane cubic .
there is @xmath18 elliptic plane cubic through @xmath109 points is a fact followed easily from linear algebra .
we give another proof using getzler s relation .
call the number of elliptic cubics satisfying the given constraint @xmath110 .
first we consider a @xmath13-dimensional of @xmath83 maps @xmath53 on @xmath111 satisfying the following conditions : + - @xmath112 each belongs to a general plane .
+ - the image of @xmath53 passes through @xmath19 point and @xmath10 lines . by elementary dimension counting
it is easy to see that @xmath83 is @xmath13 dimension .
let @xmath113 be the forgetful morphism .
now we intersection @xmath114 with the getzler s relation .
+ + - @xmath115 : there are no map of elliptic curves with degree @xmath18
. maps of degree @xmath13 are double covers , but with incidence - only constraints , these will not contribute .
thus the degree on the elliptic components are @xmath19 and the two rational twigs are contracted .
( we will see in section @xmath10 and @xmath20 that the maps that contracted the elliptic component do not contribute either ) . for each possible way of distributing the marked points on the two rational twigs we see that the count is @xmath110 ,
thus we have @xmath116 in total .
multiply this by the coefficient of @xmath16 we get @xmath117 + + - @xmath118 : argue similarly , the two rational twigs must contract .
then the node on the elliptic curve must satisfy incidence conditions of @xmath19 of the points of @xmath119 ( whichever on the contracted components ) .
thus we get the numbers of elliptic curves passing through @xmath10 points and @xmath10 lines , multiplied by @xmath19 due to a marked point on the elliptic component satisfying a hyperplane condition . but
this is @xmath93 cause an elliptic cubic is planar and can not pass through @xmath10 general points . + + - similarly @xmath115 and @xmath120 are @xmath93 .
+ + - @xmath121 : if the rational twig is contracted , then we get the numbers of nodal plane cubics passing through @xmath10 points and @xmath19 lines , which is @xmath93 .
if the genus @xmath18 component is contracted , then by a result in section @xmath10 , the resulting image curve has a cusp ( corresponding to the node on the domain ) .
the cusp must satisfy an hyperplane incidence condition . by dimension
counting , the contribution in this case is also @xmath93 .
+ + - @xmath122 : similarly , if the rational twig get contracted , then there is no contribution .
if the genus @xmath18 component gets contracted , then we the number of rational cuspidal cubic curves passing through @xmath19 points and @xmath10 lines .
translating this to the plane curve counting problem we get the number of rational cuspidal cubic passing through @xmath123 points , which is @xmath124 .
there are @xmath10 marked points each satisfying a hyperplane condition , and each map has automorphism group of order @xmath13 . thus the total contribution is @xmath125 + + - @xmath126 : in general , counting maps with this type of domains ( @xmath94 ) is not trivial , and an algorithm for this is given in @xcite . in this particular example , however , it is easy .
if any of the twig gets contracted , we have rational nodal curve with conditions on the node , but by simple dimension counting we see that there is no contribution . thus the pair has to be line - conic intersecting at @xmath13 points .
the @xmath19 points conditions determine the plane for this pair , so we can translate this into a plane curve counting problem .
how many pair of line - conics passing through @xmath123 points .
the answer is @xmath18 , multiplied by the number of ways to distribute the conditions , which is @xmath127 .
we also have incidence on the marked points , especially on the conics ( each contributes a factor of @xmath13 ) . for each marked points distribution ,
the contribution is @xmath128 .
there are six total ways to distribute the marked points .
thus the total contribution of this stratum is @xmath129 multiplying with the coefficient of the stratum , we get @xmath130 . + +
thus we end up with the equation @xmath131 hence @xmath132 .
first we need a lemma to tell us which maps can occur as boundary of the main component @xmath17 .
the euler characeristic is an important invariant in a family of curves .
it is @xmath18 minus the arithmetic genus .
the euler characteristic can be defined for any scheme @xmath110 as the euler characteristic of its structure sheaf @xmath133 .
the following lemma is useful to compute euler characteristic of reducible curves .
euler characteristic satisfies the inclusion - exclusion property .
that is , if @xmath134 are two subschemes of an ambient scheme , then @xmath135 _ proof .
_ we have the following exact sequence of sheaf on @xmath136 : @xmath137 where all the sheaves on smaller schemes are pushforwarded to @xmath136 .
the identity follows from the fact that euler characteristic is additive along an exact sequence .
let @xmath138 be a map in @xmath139 ( @xmath140 ) from a nodal , @xmath22-pointed curve @xmath141 of arithmetic genus one that contracts a connected union of irreducible components having total arithmetic genus one .
let @xmath142 s ( @xmath143 be the non - contracting rational tails and let @xmath144 be the intersection of @xmath142 with the contracted subcurve
. then the images of the tangent vectors to @xmath142 s at @xmath144 s are not independent . _
_ first note that contracted subcurve has arithmetic genus one , hence each @xmath142 intersects it at exactly one point .
let @xmath145 be a one dimensionsinal family with general fibers smooth of genus @xmath18 and the central fiber being @xmath141 .
if @xmath53 contracts the central fiber then it factors through a contraction @xmath146 .
the central fiber of @xmath147 , denoted by @xmath148 consists @xmath149 rational twigs connected at one node .
however , the tangent vectors to these twigs at the node are not independent for the following reason .
the euler characteristic of the general fiber is @xmath93 ( equals @xmath150 the arithmetic genus ) , and this is a constant of the fibers in the family . if the tangent vectors were independent , then the node ( with the intersection scheme structure ) is reduced , and by a repeat application of lemma @xmath151 , we see that the euler characteristic would be @xmath18 .
it follows that the images of the tangent vectors in @xmath152 are also not independent . using the lemma
, one can see that the following loci are intersections of the main component with other components . for @xmath153
is the closure of the locus of maps @xmath53 having the following property : + - the domain of @xmath53 has a smooth elliptic component joined with @xmath149 rational tails @xmath142 at the node @xmath144 .
+ - @xmath53 has degree @xmath93 on the elliptic component , and the images of the tangent vectors to @xmath142 at @xmath144 are linearly dependent . for @xmath154
, the image curve has a cusp , for @xmath155 , the image curve is a two - component rational curves meeting at a tacnode . in general , for @xmath156 the image curve has a @xmath149-fold elliptical singularity . for @xmath157 , @xmath158 lies entirely on @xmath17 .
we are now ready to derive a recursion counting elliptic space curves .
first we need some notations .
let @xmath28 be a constraint such that @xmath159 , and assume @xmath160 is an index such that @xmath161 .
let @xmath162 be two subspaces in @xmath28 of codimension @xmath32 .
let @xmath163 be two general hyperplanes in @xmath152 and let @xmath164 be subspaces of codimension @xmath165 .
the following constraints are derived from @xmath28 : + - @xmath166 by removing @xmath167 and @xmath168 .
+ - @xmath169 by removing @xmath162 but adding a codimension @xmath165 and a codimension @xmath170 subspaces .
+ - @xmath171 by removing @xmath162 but adding a codimension @xmath13 and a codimension @xmath172 subspaces .
+ - @xmath173 by removing @xmath162 but adding a hyperplane and a codimension @xmath174 subspace .
+ - @xmath175 by removing @xmath162 but adding @xmath176 .
we denote @xmath177 be the indicator function where @xmath178 is a logical statement .
let @xmath101 acts on the set @xmath179 it is easy to see that @xmath180 is of lower rank than that of @xmath28 .
we have the following recursive formula , providing the left - hand side is finite ( it is understood that if a constraint contains the empty subspace then the corresponding enumerative term is automatically @xmath93 ) .
@xmath181\end{aligned}\ ] ] _ proof .
_ we consider the moduli space @xmath182 , where there is a one - to - one correspondence @xmath183 .
let @xmath184 be the subfamily cut out by the constraint @xmath166 and the condition that @xmath185 for each mark point @xmath186 .
@xmath83 is either empty or @xmath13-dimensional .
if @xmath83 is empty then all the summands in the equation are @xmath93 so there is nothing to prove .
assume @xmath83 is @xmath13-dimensional .
let @xmath187 be the pushforward of @xmath83 under the forgetful morphism @xmath188 .
we intersect @xmath187 with getzler s relation .
+ - @xmath115 : if the elliptic component is contracted , then by proposition @xmath189 , then there are two possibilities for the image curves
. it can be a two - component rational curves intersecting at a tacnode . by simple dimension counting ( it lies in a family of one fewer dimension than that of a rational cuspidal curve )
we see that the image curves ( of those in @xmath83 that contract the elliptic component ) move in a family of @xmath19 dimension fewer than @xmath83 . but
@xmath83 is @xmath13-dimensional so , this loci must be empty .
the other possibility is again two - component rational curve , but one has a cusp at the intersection .
dimension counting again shows that this locus on @xmath83 is empty .
( note that if @xmath83 has at least @xmath19 dimensional , then these loci are generally not empty and are in fact a divisor on @xmath83 .
however , a contributor to the dimensions of these loci comes from varying the two nodes on the elliptic components , which is irrelevant for enumerative reason ) .
if both rational components are contracted , then we either get @xmath190 depending on if @xmath191 is @xmath18-dimensional .
if @xmath191 is not @xmath18-dimensional , then we also get @xmath192 . next , if @xmath18 of the rational tails is contracted , then we get @xmath193 the factor @xmath194 comes from the fact that permuting marked points on the same rational tail gives us the same stratum . if non of the rational tails are contracted , then the contribution is @xmath195 the factor @xmath196 comes from the fact that permuting marked points on each tail and permuting the two tails give us the same stratum ( given that we consider all possible degree distributions , contrary to the situation in the previous remark , where one of the tail is contracted while the other is not ) .
@xmath2 is the coefficient of @xmath16 in getzler s relation .
+ - @xmath118 , @xmath197 can be found similarly , and these summands explain all the terms until the first @xmath198 term .
+ - @xmath121 : if the nodal elliptic component is contracted , then by proposition @xmath189 , we must have a cuspidal curve joined with a smooth rational curve by a node . but this locus on @xmath83 must be empty since a smooth rational cuspidal curve is moving in a @xmath13 dimensions fewer family than that of a smooth elliptic curve ( a general member of @xmath83 ) . the rational tail can not be contracted also : if it were contracted , then moving the marked points on the rational tail ( one - dimensional family ) change the map but does not change the image curves , meaning than the image curves can not satisfy all the enumerative constraint .
thus non of the components can be contracted , and we have the contribution from this stratum is @xmath199 the factor @xmath200 comes from the fact that we can permute the marked points on the rational component , and the factor @xmath201 comes from the fact that we can permute the branches over the node of the image for maps in @xmath51 .
+ - @xmath122 : argue similarly , the rational tail can not be contracted . if the nodal elliptic component is not contracted , then the contribution is @xmath202 if the nodal elliptic component is contracted , then the contribution is @xmath203 the factor @xmath201 comes from the fact that any map that contracts the nodal elliptic component has automorphism group of order @xmath13 .
+ - @xmath204 : argue as above , we can show that non of the rational bridges can be contracted .
the contribution is @xmath205 we can permute the marked points on each component and permute the two component themshelves , hence the factor @xmath196 . the factors @xmath201 comes from the fact that we can permute two branches over one of the node of the image for maps in @xmath206 except the terms involving @xmath60 , all other terms can be computed either from algorithms in @xcite and @xcite or recursively . for example @xmath207
terms @xmath208 and @xmath209 can be computed using the splitting the diagonal " method described in section @xmath19 of @xcite .
the term @xmath210 for @xmath211 can be assumed known by induction since @xmath212 is of lower rank than that of @xmath28 .
note that we can choose @xmath162 being any two subspaces in @xmath28 having same codimensions : they could be two points , two lines , two planes etc . for each choice
we have a different recursion .
hence a good check of the formulas is that different choices giving same numbers . we have confirmed this is true for all elliptic space curves in @xmath213 of degree at most @xmath20 , and so far no contradiction has been found
in this section we give a recursive formula counting elliptic space curves with tangency conditions .
first we need the following lemma .
let @xmath28 be a constraint , and @xmath83 a family of maps in @xmath214 such that @xmath215 is one - dimensional .
then the maps in @xmath83 do not contract an elliptic component .
moreover , the number of maps in @xmath83 whose images are tangent to a general hyperplane is given by @xmath216 where @xmath217 is the number of maps in @xmath83 satisfying top incident condition , @xmath218 is the number of maps in @xmath83 whose domain is a smooth elliptic curve having a fixed @xmath7-invariant ( @xmath7 is not @xmath93 or @xmath219 , and @xmath220 is the number of rational tails of domains of maps in @xmath83 that are mapped with degree @xmath32 . _
_ the loci of maps that contract an elliptic component are at most divisors on @xmath61 .
since we can change such maps by moving the nodes on the elliptic component without changing the images , the images must move in a family of at most @xmath13 dimensions fewer than those of maps of smooth elliptic curves . since @xmath83 is
one - dimensional , and enumerative constraints impose conditions on the image curves only , this shows those loci must be empty .
let @xmath221 be the divisor on @xmath83 corresponding to tangency condition and top incidence condition respectively .
assume the @xmath83 is represented by the total family @xmath222 and a map @xmath223 .
let @xmath224 be a general hyperplane in @xmath152 .
then @xmath225 is a smooth curve in @xmath226 and @xmath227 is given by the branch divisor of the covering @xmath228 .
thus @xmath229 where the last equality follows from adjunction .
it is easy to see that @xmath230 let @xmath231 be the divisor of @xmath226 corresponding to rational tails , and let @xmath106 be the divisor of @xmath226 corresponding to nodal elliptics .
then we have @xmath232 this follows from theorem @xmath233 in @xcite , corrected by the rational tails .
since @xmath106 is equivalent to the locus of fibers having fixed @xmath7-invariant , we have @xmath234 using the lemma , we can easily deduce the following result .
let @xmath28 be a constraint with @xmath235 .
let @xmath42 be the constraint derived from @xmath28 by removing a tangency condition and adding a top incidence condition .
let @xmath236 be the constraint derived from @xmath28 by removing a tangency condition .
then we have the following equality , provided the left - hand side is finite .
@xmath237 this gives a recursive formula for all characteristic numbers of elliptic space curves with at least one tangency condition .
the characteristic numbers @xmath238 were computed in @xcite .
in this section we give numerical examples of characteristic numbers of elliptic curves of low degree in @xmath239 for elliptic curves in @xmath240 and @xmath241 of low degrees , we give all characteristic numbers . for curves in higher dimensional projective spaces ,
we give a random sample of these numbers as there are too many possible characteristic numbers ( in addition to long running time ) .
note that degree @xmath13 elliptic curves are understood as degree @xmath13 covers of @xmath242 .
& @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath253 & 1 & @xmath254 & 4 & @xmath255 & 16 + @xmath256 & 64 & @xmath257 & 256 & @xmath258 & 976 + @xmath259 & 3424 & @xmath260 & 9766 & @xmath261 & 21004 + @xmath262 & 33616 & & & & + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath264 & 225 & @xmath265 & 1010 & @xmath266 & 4396 + @xmath267 & 18432 & @xmath268 & 73920 & @xmath269 & 280560 + @xmath270 & 994320 & @xmath271 & 3230956 & @xmath272 & 9409052 + @xmath273 & 23771160 & @xmath274 & 50569520 & @xmath275 & 89120080 + @xmath276 & 129996216 & & & & + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath278 & 87192 & @xmath279 & 411376 & @xmath280 & 1873388 + @xmath281 & 8197344 & @xmath282 & 34294992 & @xmath283 & 136396752 + @xmath284 & 512271756 & @xmath285 & 1802742368 & @xmath286 & 5889847264 + @xmath287 & 17668868832 & @xmath288 & 48034104112 & @xmath289 & 116575540736 + @xmath290 & 248984451648 & @xmath291 & 463227482784 & @xmath292 & 747546215472 + @xmath293 & 1048687299072 & & & & + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath295 & 57435240 & @xmath296 & 278443920 & @xmath297 & 1304259360 + @xmath298 & 5884715280 & @xmath299 & 25491474432 & @xmath300 & 105633321120 + @xmath301 & 417060737040 & @xmath302 & 1561852784760 & @xmath303 & 5519825870880 + @xmath304 & 18304445284032 & @xmath305 & 56582281200000 & @xmath306 & 161827650576960 + @xmath307 & 424685965762560 & @xmath308 & 1013734555246080 & @xmath309 & 2182871531466432 + @xmath310 & 4212351284630880 & @xmath311 & 7256549594647680 & @xmath312 & 11151931379093760 + @xmath313 & 15327503832362880 & & & & + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath315 & 0 & @xmath316 & 0 & @xmath317 & 0 + @xmath318 & 1 & @xmath319 & 14 & @xmath320 & 150 + @xmath321 & 1500 & @xmath322 & 0 & @xmath323 & 0 + @xmath324 & 4 & @xmath325 & 50 & @xmath326 & 498 + @xmath327 & 4740 & @xmath328 & 0 & @xmath329 & 0 + @xmath330 & 16 & @xmath331 & 176 & @xmath332 & 1620 + @xmath333 & 14640 & @xmath334 & 0 & @xmath335 & 64 + @xmath336 & 608 & @xmath337 & 5136 & @xmath338 & 43944 + @xmath339 & 0 & @xmath340 & 256 & @xmath341 & 2048 + @xmath342 & 15744 & @xmath343 & 127104 & @xmath344 & 976 + @xmath345 & 6464 & @xmath346 & 45040 & @xmath347 & 342720 + @xmath348 & 3424 & @xmath349 & 18560 & @xmath350 & 116768 + @xmath351 & 836480 & @xmath352 & 47936 & @xmath353 & 269440 + @xmath354 & 1809040 & @xmath355 & 114248 & @xmath356 & 553176 + @xmath357 & 3439024 & @xmath358 & 1024404 & @xmath359 & 5768584 + @xmath360 & 1774680 & @xmath361 & 8656240 & @xmath362 & 11875120 + @xmath363 & 15480640 & & & & + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath365 & 1 & @xmath366 & 4 & @xmath367 & 32 + @xmath368 & 310 & @xmath369 & 3220 & @xmath370 & 34674 + @xmath371 & 385656 & @xmath372 & 4436268 & @xmath373 & 52832040 + @xmath374 & 12 & @xmath375 & 96 & @xmath376 & 920 + @xmath377 & 9408 & @xmath378 & 99270 & @xmath379 & 1081968 + @xmath380 & 12224484 & @xmath381 & 143419320 & @xmath382 & 36 + @xmath383 & 288 & @xmath384 & 2720 & @xmath385 & 27312 + @xmath386 & 281004 & @xmath387 & 2988144 & @xmath388 & 33049512 + @xmath389 & 381061200 & @xmath390 & 864 & @xmath391 & 8000 + @xmath392 & 78656 & @xmath393 & 783840 & @xmath394 & 8087616 + @xmath395 & 87221808 & @xmath396 & 985671936 & @xmath397 & 2592 + @xmath398 & 23360 & @xmath399 & 224256 & @xmath400 & 2145024 + @xmath401 & 21331136 & @xmath402 & 223311840 & @xmath403 & 2466111936 + @xmath404 & 67440 & @xmath405 & 630720 & @xmath406 & 5721424 + @xmath407 & 54410016 & @xmath408 & 550239168 & @xmath409 & 5919868800 + @xmath410 & 191760 & @xmath411 & 1743488 & @xmath412 & 14766080 + @xmath413 & 133095808 & @xmath414 & 1293435904 & @xmath415 & 13514355840 + @xmath416 & 4724272 & @xmath417 & 36626544 & @xmath418 & 309751664 + @xmath419 & 2876272592 & @xmath420 & 29088348480 & @xmath421 & 12532016 + @xmath422 & 86940920 & @xmath423 & 681603936 & @xmath424 & 6007997008 + @xmath425 & 58587710176 & @xmath426 & 197671204 & @xmath427 & 1413728496 + @xmath428 & 11731399560 & @xmath429 & 109792714096 & @xmath430 & 435015624 + @xmath431 & 2767555376 & @xmath432 & 21376596768 & @xmath433 & 190821802560 + @xmath434 & 5150502848 & @xmath435 & 36418237824 & @xmath436 & 307505812160 + @xmath437 & 9269345984 & @xmath438 & 58355286272 & @xmath439 & 460737967360 + @xmath440 & 88904673408 & @xmath441 & 645526598016 & @xmath442 & 131356680480 + @xmath443 & 853096310656 & @xmath444 & 1076343432320 & @xmath445 & 1321684733280 + & @xmath243 & @xmath244 & @xmath243 & @xmath244 + @xmath447 & 42 & @xmath448 & 354 & @xmath449 & 3492 + @xmath450 & 38049 & @xmath451 & 441654 & @xmath452 & 5378454 + @xmath453 & 68292324 & @xmath454 & 901654884 & @xmath455 & 12358163808 + @xmath456 & 175599635328 & @xmath457 & 2583319387968 & @xmath458 & 1094 + @xmath459 & 10476 & @xmath460 & 111774 & @xmath461 & 1271974 + @xmath462 & 15194034 & @xmath463 & 189441324 & @xmath464 & 2460171444 + @xmath465 & 33232962528 & @xmath466 & 466363099008 & @xmath467 & 6789367904448 + @xmath468 & 3340 & @xmath469 & 31120 & @xmath470 & 324980 + @xmath471 & 3618784 & @xmath472 & 42296604 & @xmath473 & 516526368 + @xmath474 & 6582996888 & @xmath475 & 87478828368 & @xmath476 & 1210578237888 + @xmath477 & 17419712523648 & @xmath478 & 91560 & @xmath479 & 935136 + @xmath480 & 10164264 & @xmath481 & 115886944 & @xmath482 & 1381799016 + @xmath483 & 17235919176 & @xmath484 & 224820084336 & @xmath485 & 3062777447088 + @xmath486 & 43504838611968 & @xmath487 & 267008 & @xmath488 & 2663824 + @xmath489 & 28172256 & @xmath490 & 312141824 & @xmath491 & 3619891072 + @xmath492 & 44047594080 & @xmath493 & 562512674880 & @xmath494 & 7529361687168 + @xmath495 & 105419849015808 & @xmath496 & 7515344 & @xmath497 & 77029344 + @xmath498 & 825583024 & @xmath499 & 9266866944 & @xmath500 & 109570881504 + @xmath501 & 1365937000128 & @xmath502 & 17924770819968 & @xmath503 & 246982965815808 + @xmath504 & 21015744 & @xmath505 & 207770304 & @xmath506 & 2142245344 + @xmath507 & 23132708544 & @xmath508 & 264540003744 & @xmath509 & 3208327374528 + @xmath510 & 41175810566208 & @xmath511 & 557379334146048 & @xmath512 & 553229344 + @xmath513 & 5451732864 & @xmath514 & 56197250608 & @xmath515 & 618092874928 + @xmath516 & 7264503650688 & @xmath517 & 90935869598208 & @xmath518 & 1207034339515008 + @xmath519 & 1456801224 & @xmath520 & 13615688984 & @xmath521 & 132655331216 + @xmath522 & 1393893072176 & @xmath523 & 15805591490496 & @xmath524 & 192394855764288 + @xmath525 & 2498937360190848 & @xmath526 & 33442096324 & @xmath527 & 304042170552 + @xmath528 & 3027600730440 & @xmath529 & 32950822741600 & @xmath530 & 388703233243008 + @xmath531 & 4928984795256768 & @xmath532 & 81107025144 & @xmath533 & 677080161264 + @xmath534 & 6326216895824 & @xmath535 & 65682090217248 & @xmath536 & 747920741035008 + @xmath537 & 9234940120602048 & @xmath538 & 1469692262864 & @xmath539 & 12719562619344 + @xmath540 & 125046496125728 & @xmath541 & 1368071583917408 & @xmath542 & 16398336070362048 + @xmath543 & 3132954358848 & @xmath544 & 24665727975168 & @xmath545 & 227432721556608 + @xmath546 & 2377092153260928 & @xmath547 & 27559370814663168 & @xmath548 & 46380697328576 + @xmath549 & 395961856525056 & @xmath550 & 3925472901986688 & @xmath551 & 43826076405481728 + @xmath552 & 85453000115200 & @xmath553 & 662634098949120 & @xmath554 & 6173065500610048 + @xmath555 & 66008781304150528 & @xmath556 & 1073447856471168 & @xmath557 & 9277390256888448 + @xmath558 & 94369223149298688 & @xmath559 & 1703768515184128 & @xmath560 & 13397726490436608 + @xmath561 & 128504536404712448 & @xmath562 & 18739152863106048 & @xmath563 & 167464728764784128 + @xmath564 & 25701019852524288 & @xmath565 & 210144272908228608 & @xmath566 & 255958547477177088 + @xmath567 & 306095919912649728 & & & & + & @xmath243 & @xmath244 + @xmath569 & 0 & @xmath570 & 0 + @xmath571 & 0 & @xmath572 & 1 + @xmath573 & 14 & @xmath574 & 222 + @xmath575 & 114 & @xmath576 & 4 + @xmath577 & 190 & @xmath578 & 1488 + @xmath579 & 0 & @xmath580 & 0 + @xmath581 & 0 & @xmath582 & 16 + @xmath583 & 640 & @xmath584 & 2280 + @xmath585 & 31044 & @xmath586 & 13680 + @xmath587 & 536304 & @xmath588 & 512 + @xmath589 & 3328 & @xmath590 & 2048 + @xmath591 & 8041776 & @xmath592 & 6464 + @xmath593 & 101408 & @xmath594 & 31624 + @xmath595 & 582888 & @xmath596 & 26848 + @xmath597 & 144400 & @xmath598 & 83312 + @xmath599 & 39311360 & @xmath600 & 497216 + @xmath601 & 203968 & @xmath602 & 76501840 + @xmath603 & 8583182 & @xmath604 & 224882706 + @xmath605 & 1130248810 & @xmath606 & 10539980 + @xmath607 & 72275990 & @xmath608 & 5552993600 + & @xmath243 & @xmath244 + @xmath611 & 29 & @xmath612 & 0 + @xmath613 & 4 & @xmath614 & 1004916 + @xmath615 & 144007483 & @xmath616 & 38 + @xmath617 & 4860 & @xmath618 & 81196 + @xmath619 & 920 & @xmath620 & 9408 + @xmath621 & 11043310 & @xmath622 & 156968 + @xmath623 & 5342984 & @xmath624 & 43094568 + @xmath625 & 420000 & @xmath626 & 13098172 + @xmath627 & 13916950104 & @xmath628 & 6377111884 + @xmath629 & 113231319632 & @xmath630 & 0 + @xmath631 & 864 & @xmath632 & 958400 + @xmath633 & 7666224 & @xmath634 & 139312 + @xmath635 & 47136 & @xmath636 & 9606144 + @xmath637 & 5203072 & @xmath638 & 67440 + @xmath639 & 12588560 & @xmath640 & 49299816 + @xmath641 & 4092688 & @xmath642 & 35732553632 + @xmath643 & 257686679704 & @xmath644 & 214317637545920 + @xmath645 & 965893920 & @xmath646 & 19635766224 + @xmath647 & 194300922530150 & @xmath648 & 3865044261124 + @xmath649 & 23269075459780 & @xmath650 & 709454579680 + @xmath651 & 2593390019349960 & @xmath652 & 680573664677760 + @xmath653 & 1046525920177280 & @xmath654 & 6253213828581120 + & @xmath243 & @xmath244 + @xmath656 & 17812920 & @xmath657 & 8 + @xmath658 & 22141610 & @xmath659 & 83534718240 + @xmath660 & 5574120 & @xmath661 & 21135644112 + @xmath662 & 157704159607499400 & @xmath663 & 317850984 + @xmath664 & 3852663999504 & @xmath665 & 12466665712 + @xmath666 & 324662678747600 & @xmath667 & 6438834486545696 + @xmath668 & 152170558528 & @xmath669 & 4059955629840 + @xmath670 & 17870726781040 & @xmath671 & 495750960 + @xmath672 & 267364637315200 & @xmath673 & 37308946152960 + @xmath674 & 5661536113375616 & @xmath675 & 76189759273400 + @xmath676 & 1261503043660160 & @xmath677 & 202527736950333560 + @xmath678 & 152112968567744 & @xmath679 & 3296407750001057840 + @xmath680 & 75941963996580 & @xmath681 & 13254336373174 + @xmath682 & 4927564446146120 & @xmath683 & 67069699366126280 + @xmath684 & 686078803431960 & @xmath685 & 305992051575799496 + @xmath686 & 36003235532769890160 & @xmath687 & 7621648299681170363680 + @xmath688 & 76323334247130289600 & @xmath689 & 2598949329007010304 + @xmath690 & 2695498962098743296 & & + & @xmath243 & @xmath244 + @xmath692 & 0 & @xmath693 & 112 + @xmath694 & 294 & @xmath695 & 0 + @xmath696 & 8 & @xmath697 & 190 + @xmath698 & 2694 & @xmath699 & 0 + @xmath700 & 80 & @xmath701 & 432 + @xmath702 & 7008 & @xmath703 & 3696 + @xmath704 & 6400 & @xmath705 & 256 + @xmath706 & 976 & @xmath707 & 13536 + @xmath708 & 10000 & @xmath709 & 135504 + @xmath710 & 235888 & @xmath711 & 63034080 + @xmath712 & 307313568 & @xmath713 & 5158008 + @xmath714 & 160210908480 & @xmath715 & 242570957664 + @xmath716 & 617311046976 & @xmath717 & 31799786256 + @xmath718 & 40866566832 & @xmath719 & 219076067112 + & @xmath243 & @xmath244 + @xmath722 & 150696 & @xmath723 & 3220 + @xmath724 & 57960 & @xmath725 & 18866664 + @xmath726 & 139530931995 & @xmath727 & 11150743642205 + @xmath728 & 12 & @xmath729 & 585468 + @xmath730 & 56249988 & @xmath731 & 7544 + @xmath732 & 1444 & @xmath733 & 856860 + @xmath734 & 75690880 & @xmath735 & 36244480608680 + @xmath736 & 203200 & @xmath737 & 10274688 + @xmath738 & 6044208 & @xmath739 & 299599503152 + @xmath740 & 30574272 & @xmath741 & 5402075552 + @xmath742 & 56706560 & @xmath743 & 22175099750880 + @xmath744 & 19987349088 & @xmath745 & 1477032997824 + @xmath746 & 555756395786592 & @xmath747 & 36471958824 + @xmath748 & 803465851328 & @xmath749 & 479197387872 + @xmath750 & 2604953865714080 & @xmath751 & 3956581280704 + @xmath752 & 865530347368728 & @xmath753 & 127384328451776 + & @xmath243 & @xmath244 + @xmath755 & 21370599 & @xmath756 & 13368 + @xmath757 & 6441662050785 & @xmath758 & 120 + @xmath759 & 92992972101933954474544 & @xmath760 & 669439268976 + @xmath761 & 70316304 & @xmath762 & 249642503347264 + @xmath763 & 59596479340997440 & @xmath764 & 4670806268148474812288 + @xmath765 & 19278514559525472 & @xmath766 & 275265022278424 + @xmath767 & 1322240935084886 & @xmath768 & 5731941475630570274830480 + @xmath769 & 93646029368101232 & @xmath770 & 7985056411618359072 + @xmath771 & 55781303615787140368 & @xmath772 & 110267644926473616 + [ 1 ] d. avritzer and i. vainsencher , _ compactifying the space of elliptic quartic curves , _ in complex projective geometry , g. ellingsrud et al .
, cambridge u.p .
: cambridge , 1992 . w. barth , c. peters , a. van de ven , _ compact complex surfaces _ , springer - verlag , new york , 1984 . w.fulton _ intersection theory , _ second edition , springer 1996 .
w.fulton and r. pandharipande , _ notes on stable maps and quantum cohomology _ ,
preprint 1996 , alg - geom/9608011 .
x. hernandez , j. m. miret , _ the characteristic numbers of cuspidal plane cubics in @xmath1 _ , bull .
simon stevin , * 10 * ( 2003 ) no .
1 , 115124 .
x. hernandez , j. m. miret and s. xambo - descamps , _ computing the characteristic numbers of the variety of nodal plane cubics in @xmath1 _ , j. symb .
* 42 * ( 2007 ) 192202 .
e. ionel , _ genus - one enumerative invariants in @xmath774 with fixed j - invariant _ , duke math
. j. * 94 ( 2 ) * ( 1998 ) 279324 .
e. getzler , _ intersection theory on @xmath775 and elliptic gromov - witten invariants _ , j. amer .
* 10 * no .
4 ( 1997 ) 973998 .
t. graber , j. kock and r. pandharipande _ descendant invariants and characteristic numbers , _ american journal of mathematics , vol .
124 , no . *
. s. kleiman , s. a. strmme and s. xamb _ sketch of a verification of schubert s number 5819539783680 of twisted cubics , _ lecture notes in mathematics , 1987 , volume 1266/1987 , 156 - 180 .
d. nguyen , _
characteristic numbers of elliptic curves with fixed @xmath776invariant _ , preprint 2011 , arxiv:1111.6295 .
d. nguyen , _
characteristic numbers of rational cuspidal space curves _ , preprint 2011 , arxiv:1111.6296 .
r. pandharipande , _ intersection of @xmath777-divisors on kontsevich s moduli space @xmath778 and enumerative geometry _ ,
soc , * 351 * ( 1999 ) , 1481 - 1505 .
r. pandharipande , _ a note on elliptic plane curves with fixed @xmath7-invariant _ , proc .
, * 125 * , no . 12 , 34713479 .
r. pandharipande , _ a geometric construction of getzler s elliptic relation , _ math . ann .
* 313 * ( 1999 ) , 715729 r. vakil , _ the enumerative geometry of rational and elliptic plane curves in projective space _
, j. reine angew . math .
( crelle s journal ) , * 529 * ( 2000 ) , 101153 . r. vakil , _
recursions for characteristic numbers of genus one plane curves _ , arkiv for matematik , * 39 * ( 2001 ) , no .
1 , 157180 .
r. vakil , a. zinger , _ a desingularization of the main component of the moduli space of genus - one stable maps to projective space _ ,
* 12 * ( 2008 ) , no . 1 , 1 - 95
. a. zinger , _ enumeration of one - nodal rational curves in projective spaces _
, topology * 43 * ( 2004 ) 793829 . | we solve the problem of characteristic numbers of elliptic curves in any dimensional projective space .
the answers are given in the form of effective recursions .
a c++ program implementing most of the recursions is available upon request . |
the form of the stellar initial mass function ( imf ) is of fundamental importance for many areas of astrophysics and a topic of considerable debate ( see , e.g. , schmidt 1959 ; miller & scalo 1979 ; scalo 1986 ; larson 1998 , 2003 ; kroupa 2002 ; chabrier 2003 , for reviews ) . measurements of the imf are difficult and somewhat model - dependent as they require the conversion of the observed present - day luminosity function of a stellar population to its mass function at birth .
best estimates for the galactic disk suggest that the imf has a powerlaw slope at @xmath15 , and turns over at lower masses ( kroupa 2001 ; chabrier 2003 ) .
this turnover can be modeled by a broken powerlaw ( kroupa 2001 ) or by a log - normal distribution with a characteristic mass @xmath6 ( chabrier 2003 ) .
the value of @xmath6 is @xmath16 in the disk of the milky way , with considerable uncertainty .
the powerlaw slope at high masses is probably close to the salpeter ( 1955 ) value of @xmath17 , with an uncertainty of @xmath18 ( scalo 1986 ; chabrier 2003 ) .
although there is no direct evidence for dramatic variations of the imf within the present - day milky way disk ( e.g. , kroupa 2001 ; chabrier 2003 ) , this does not preclude variations with time , metallicity , and/or environment .
in particular , larson ( 1998 , 2005 ) has argued that the characteristic turnover mass may be largely determined by the thermal jeans mass , which strongly depends on temperature ( @xmath19 at fixed density ) . in the context of this model
one might expect that heating by ambient far - infrared radiation would disfavor the formation of low mass stars in extreme environments , such as in super star clusters and in the center of the milky way .
other models emphasize the role of turbulence as opposed to temperature in determining the distribution of protostellar clumps ( e.g. , padoan & nordlund 2002 ) , and in these models the role of the environment may be less direct ( see mckee & ostriker [ 2007 ] for a recent review of various models to explain the characteristics of the imf ) .
observations may support the notion of a top - heavy ( or `` bottom - light '' ) imf in extreme environments .
some young super star clusters in m82 appear to have a top - heavy mass function ( e.g. , rieke et al .
1993 ; mccrady , gilbert , & graham 2003 ) , as do clusters in the galactic center region ( e.g. , figer et al .
1999 ; stolte et al . 2005
; maness et al . 2007 ) .
the interpretation of observed mass functions is complicated by dynamical effects , which tend to make the mass function more top - heavy over time , in particular in the central regions of star clusters ( see , e.g. , mccrady et al .
2003 ; mccrady , graham , & vacca 2005 ; kim et al .
recently harayama , eisenhauer , & martins ( 2007 ) studied the imf of ngc 3603 , one of the most massive galactic star - forming regions , out to large radii and conclude that its imf is substantially flatter than salpeter for masses @xmath20 .
the imf may also depend on redshift . at earlier times star formation presumably occurred more often in a burst mode than in a relatively gradual `` disk '' mode ( e.g. , steidel et al .
1996 ; blain et al . 1999b
; lacey et al . 2007 ) , which means that the imf could generally be more skewed toward high mass stars at redshifts 13 and beyond .
furthermore , the average metallicity in star forming clouds was lower at higher redshift , which may have led to an extremely top - heavy imf for the first generation of stars ( e.g. , abel , bryan , & norman 2002 ; bromm , coppi , & larson 2002 ) .
finally , the cosmic microwave background ( cmb ) radiation sets a floor to the ambient temperature , and hence the jeans mass , which scales with @xmath21 . beyond @xmath22
the cmb temperature exceeds the typical temperatures of dense prestellar cores in galactic molecular clouds ( e.g. evans et al . 2001 ; tafalla et al .
therefore , at sufficiently high redshift the characteristic mass may be expected to evolve roughly as @xmath23 , leading to imfs which have a reduced fraction of low mass stars ( larson 1998 ) .
the effects of the cmb are even more pronounced when its influence on the pressure in star - forming clouds is taken into account , and larson ( 2005 ) suggests that at @xmath24 the characteristic mass may be higher than today s value by as much as an order of magnitude .
such rapid evolution of the imf would have important consequences for determinations of masses and star formation rates of distant galaxies , and for measurements of evolution in these properties .
it is very difficult to constrain the imf at early times directly , as the light of high redshift galaxies is completely dominated by massive stars .
the extremely blue rest - frame uv colors of galaxies at @xmath25 may imply a top - heavy imf ( stanway , mcmahon , & bunker 2005 ) , although this is just one of several possible explanations . from observations of a lensed lyman break galaxy
there is some evidence that the _ slope _ of the imf at @xmath26 is similar to the salpeter value at the high mass end ( pettini et al .
2000 ) , but there is essentially no information on stars with masses near or below 1 . fortunately , the form of the high redshift imf has implications for the properties of galaxies at much lower redshift , as all stars with masses @xmath27 that formed in the history of the universe are still with us today .
tumlinson ( 2007 ) finds that the properties of carbon - enhanced metal - poor stars in our galaxy are best explained with a relatively high number of stars in the mass range 18 at high redshift .
various other constraints obtained from galaxies at low redshift ( including our own ) are reviewed in chabrier ( 2003 ) .
of particular interest are the stellar populations of massive early - type galaxies , as they are very homogeneous and should reflect conditions in star forming regions at @xmath28 .
recently , cappellari et al .
( 2006 ) used the kinematics of elliptical galaxies to constrain the imf , as the dynamical @xmath29 ratio provides an upper limit to the amount of mass that can be locked up in low mass stars .
current data appear to rule out a salpeter ( 1955 ) ( or steeper ) imf , but are consistent with kroupa ( 2001 ) and chabrier ( 2003 ) imfs ( cappellari et al .
2006 ) . in this paper
, we provide new constraints on the imf at high redshift by comparing the _ evolution _ of the @xmath29 ratios of early - type galaxies to their color evolution .
this method was first suggested by tinsley ( 1980 ) , but data of sufficient accuracy are only now becoming available .
the method is sensitive to the imf in the important mass range around 1 , where the effects of an evolving characteristic mass might be expected to manifest themselves . a plan of the paper follows . in 2 a relation between color evolution , luminosity evolution , and the logarithmic slope of the imf @xmath30
is derived using stellar population synthesis models . in
3 published data and archival hst images of galaxy clusters are used to construct the redshift evolution of the @xmath1 color - mass relation . in
4 the color evolution from 3 is combined with the previously measured evolution of the @xmath2 ratio . the relations from 2
are then used to derive constraints on the imf slope @xmath30 from the combined color and luminosity evolution .
section 5 is devoted to the ( many ) systematic uncertainties in the methodology and in the data , and 6 asks whether our results are consistent with other constraints on the stellar populations of massive early - type galaxies .
although the constraints we derive in this paper are subject to many uncertainties , it is interesting to explore their consequences . in 7 the fitting results of 4 are interpreted in the context of an evolving characteristic mass @xmath6 .
the data on cluster galaxies are combined with previous constraints on @xmath6 for globular clusters and submm - galaxies , and a simple form of imf evolution is proposed .
this evolution is then applied to literature data on the evolution of the volume - averaged star formation rate and stellar mass density .
the key results are summarized in 8 .
we assume @xmath31 , @xmath32 , and @xmath33mpc@xmath34 where needed .
the form of the imf has a strong effect on the evolution of the @xmath29 ratio of galaxies .
the luminosity of a stellar population is expected to evolve even in the absence of star formation or mergers , due to the fact that stars turn off the main sequence and die .
the rate of luminosity evolution in a given passband is determined by a combination of the lifetime of stars near the turn - off and the number of these stars .
the lifetime of turn - off stars is directly determined by the age of the stellar population ( with the lifetime approximately equal to the age ) , and the number of turn - off stars is directly determined by the imf . a flat
, top - heavy imf implies a relatively large number of short - lived massive stars and rapid luminosity evolution , whereas a steep imf implies a relatively large number of long - lived low mass stars and slow evolution .
the color evolution behaves differently , as it is insensitive to the number of stars that evolves off the main sequence over time .
instead it is mainly driven by the location of the turn - off in the herzsprung - russell ( hr ) diagram , which does not depend on the imf but is determined by the age of the stellar population .
the imf does influence the color evolution to some extent , as a top - heavy imf reduces the number of turn - off stars with respect to the more luminous red giants .
the location of the giant branch evolves much less than the location of the turn - off , and a flat imf therefore implies weaker color evolution than a steep imf . as first pointed out by tinsley ( 1980 ) , the imf thus has opposite effects on the luminosity and color evolution : a more top - heavy imf leads to stronger luminosity evolution and weaker color evolution .
= 17.5 cm following tinsley ( 1980 ) and many later studies , we parameterize luminosity and color evolution by power laws .
power law approximations render the predicted evolution independent of the hubble constant and the absolute age calibration of the model , and smooth out artifacts in the evolution of single - age stellar populations caused by numerical effects . in our passbands
( see [ data.sec ] ) luminosity evolution takes the form @xmath35 and color evolution takes the form @xmath36 with @xmath37 the luminosity - weighted mean star formation epoch and @xmath38 and @xmath39 constants .
we use the maraston ( 2005 ) stellar population synthesis models to determine @xmath40 and @xmath41 .
[ mlcol.plot](a ) and ( b ) show the evolution of the @xmath2 ratio and @xmath1 color in this model for three different values of @xmath30 ( 1.3 , 0.7 , 0 ) .
these imfs are power laws with a fixed slope over the mass range @xmath42 .
the predictions are for a metallicity [ z / h]@xmath43 .
this relatively high value may be more appropriate for massive early - type galaxies than the solar value ( e.g. , worthey , faber , & gonzalez 1992 ) , although solar models are much better calibrated . as expected , luminosity evolution has a strong imf dependence but color evolution only has a weak ( and opposite ) imf dependence .
panel ( c ) directly compares the evolution of the @xmath29 ratio to the color evolution .
the imf dependence is relatively strong in this panel , due to the opposite effects of the imf on luminosity and color evolution .
broken lines in panels ( a - c ) are power law fits to the model predictions .
these fits generally provide good descriptions of the behavior of the models .
the largest deviations are for the @xmath1 colors at 9 and 12 gyr , where the fits are off by @xmath44 and @xmath45 mag respectively .
panels ( d - f ) show the relation between the powerlaw coefficients and the imf slope . in each case
there is a well - defined relation , that can be characterized by a simple linear fit .
these fits take the form @xmath46 and @xmath47 the fits are shown by the solid lines in fig .
[ mlcol.plot](d-f ) .
although the fits are excellent for @xmath48 the extrapolations outside that range are obviously uncertain .
note that eq .
[ kappa3.eq ] is a fit , and is not identical to the combination of eqs .
[ kappa1.eq ] and [ kappa2.eq ] . the dotted line in fig .
[ mlcol.plot](f ) shows the relation implied by eqs .
[ kappa1.eq ] and [ kappa2.eq ] ; this relation is consistent with eq .
[ kappa3.eq ] to a few percent .
open symbols and dashed lines are predictions for models with [ fe / h]=0 .
luminosity evolution is insensitive to [ fe / h ] , but color evolution has a strong metallicity - dependence .
therefore , the relation in panel ( f ) also has a strong metallicity dependence , and takes the form @xmath49 for solar metallicity . when determining the imf slope in
[ apply.sec ] results will be given both for super - solar models and solar models ( which are better calibrated ) .
as demonstrated in fig .
[ mlcol.plot ] , the imf mostly influences the luminosity evolution , and the color evolution can be viewed as an imf - insensitive `` clock '' constraining the age . by combining color and luminosity evolution for a well - chosen set of galaxies over a sufficiently large redshift range the slope of the imf @xmath30 and the formation time of the stars @xmath37 can , in principle , be constrained simultaneously ( although there are many caveats
see 5 ) .
the method requires very accurate rest - frame color and @xmath29 measurements for a passively evolving sample of galaxies .
this is not an easy task , as galaxy evolution is in general a complex process involving star bursts , mergers , and morphological changes .
the most massive early - type galaxies in clusters at modest redshifts ( @xmath50 ) come closest to a `` purely passive '' sample .
their colors and @xmath29 ratios have very small scatter ( e.g. , bower , lucey , & ellis 1992a ; jrgensen , franx , & kjrgaard 1996 ) , their luminosity and color evolution are consistent with passive evolution of stellar populations formed at redshifts @xmath51 ( e.g. , van dokkum et al .
1998a ; holden et al .
2004 ) , and they appear to have been largely in place by @xmath52 ( holden et al . 2006 , 2007 ) .
recently a large number of measurements of the @xmath2 evolution of early - type galaxies with masses @xmath53 were compiled from the literature and placed on a consistent system ( van dokkum & van der marel 2007 ) [ hereafter vv07 ]
. color evolution has also been studied by many authors ( e.g. , stanford , eisenhardt , & dickinson 1995 ; ellis et al .
1997 ; holden et al .
2006 ) , but not yet for mass - limited samples and with the required accuracy . with the availability of archival multi - color
_ hubble space telescope _ ( hst ) wfpc2 and acs data of a subset of the clusters in vv07 it is now possible to determine color and luminosity evolution of massive early - type galaxies in clusters simultaneously , and begin to constrain the form of the imf at the mean star formation epoch of massive early - type galaxies .
in this section , data from the literature and the hst archive are used to determine the evolution of rest - frame @xmath1 colors of galaxies in clusters from the vv07 sample .
because mass measurements of individual galaxies are available the color evolution can be determined from the zeropoint of the color - mass ( rather than the color - magnitude ) relation .
also , the same mass limit can be applied as was applied to the @xmath29 data .
we focus on @xmath1 because redder rest - frame colors are not available for clusters beyond @xmath54 .
the color - magnitude relation in the coma cluster was studied by bower et al .
( 1992a , b ) [ hereafter ble92 ] .
an important goal of their study was to determine the relative distance between the virgo and coma cluster , and this required very accurate absolute color measurements .
the data from ble92 are combined with velocity dispersions and effective radii from jrgensen , franx , & kjrgaard ( 1995a , b ) .
masses are calculated using @xmath55 with @xmath56 the velocity dispersion corrected to a @xmath57 diameter aperture and @xmath58 the effective radius in kpc .
radii were converted from arcsec to kpc assuming a hubble flow velocity of @xmath59(see vv07 ) .
figure [ comacm.plot ] shows the color - mass relation of galaxies in common between the ble92 color - magnitude sample and the jrgensen et al .
fundamental plane ( fp ) sample .
there is a very clear relation , of the form @xmath60 the slope of the relation is determined by fitting a linear function to all galaxies with masses @xmath53 , and the offset is determined with the biweight statistic ( beers , flynn , & gebhardt 1990 ) , which gives low weight to outliers .
this relation is indicated by the solid line in fig .
[ comacm.plot ] .
we note that the data in fig .
[ comacm.plot ] are not entirely self - consistent , as the colors were determined in a different aperture ( @xmath61 ) than the velocity dispersions ( @xmath57 ) .
this has implications for the interpretation of eq .
[ colmass.eq ] , but it has no effect on our analysis as the distant galaxies are treated in the same way as the coma galaxies .
the observed scatter in the relation is very small at @xmath62 mag for @xmath53 .
this low scatter is remarkable : from repeat measurements of the same galaxies ble92 estimate that the observational uncertainty for individual galaxies is @xmath63 , which would imply that the intrinsic scatter in @xmath1 is well below 0.03 for the most massive coma galaxies .
however , we note that the biweight estimator gives very little weight to the blue outlier at @xmath64 , and this is somewhat arbitrary given the small sample size .
the rms scatter is 0.05 , similar to values measured for clusters at higher redshift ( see [ colevo.sec ] ) . in this study
we are not concerned with the scatter or slope of the color - mass relation but only with its zeropoint evolution . for each cluster in the sample eq .
[ colmass.eq ] is subtracted from the rest - frame @xmath1 colors of early - type galaxies with @xmath53 , and the offset @xmath65 is determined with the biweight estimator . by construction , @xmath66 for the coma cluster .
the random uncertainty in the offset due to galaxy - galaxy color variations is 0.009 .
ble92 give a systematic error of 0.02 in the @xmath1 colors , reflecting uncertainties in the @xmath67 and @xmath68 zeropoints and small errors due to reddening and aperture corrections .
adding the random and systematic errors in quadrature gives a combined uncertainty of 0.022 .
we note that the ble92 data are still the most accurate available for coma , even though they are now more than 15 years old .
in fact , the comprehensive photometric studies by terlevich , caldwell , & bower ( 2001 ) and eisenhardt et al .
( 2007 ) calibrate their photometry by matching the magnitudes of galaxies to the ble92 values .
no other nearby clusters are currently available with high quality @xmath1 colors , effective radii , and velocity dispersions for their constituent galaxies . as an example , effective radii and velocity dispersions are available for the clusters in the jrgensen et al .
( 1996 ) sample , but accurate @xmath1 colors have not been measured .
conversely , mcintosh et al .
( 2005 ) analyzed the color - magnitude relation in the three nearby clusters abell 85 , abell 496 , and abell 754 using the same apertures and methods as ble92 , but those clusters have not been the subject of fp studies .
lccccccccc[!t ] coma & 0.024 & & @xmath67 & @xmath68 & & & 0.036 & 0.000 & 0.037 + abell 2218 & 0.176 & wfpc2 & f450w & f702w & 1.230 & @xmath69 & 0.021 & @xmath70 & 0.028 + cl1358 + 62 & 0.327 & acs & f475w & f775w & 0.955 & @xmath71 & 0.022 & @xmath72 & 0.033 + cl0024 + 16 & 0.391 & wfpc2 & f450w & f814w & 0.793 & @xmath73 & 0.024 & @xmath74 & 0.031 + cl0016 + 16 & 0.546 & wfpc2 & f555w & f814w & 1.065 & @xmath75 & 0.023 & @xmath76 & 0.035 + ms2053@xmath77 & 0.583&wfpc2 & f606w & f814w & 1.503 & @xmath78 & 0.035 & @xmath79 & 0.039 + ms1054@xmath80 & 0.831 & acs & f606w & f850lp & 0.881 & @xmath81 & 0.033 & @xmath82 & 0.038 + rxj0152@xmath83 & 0.837 & acs & f625w & f850lp & 1.035 & @xmath84 & 0.021 & @xmath85 & 0.024 although the offsets @xmath65 for the three clusters of mcintosh et al .
( 2005 ) can not be determined directly , they can be estimated by making use of the tight relation between rest - frame @xmath68 band luminosity and mass in nearby clusters ( e.g. , jrgensen et al .
fitting directly to the ble92 @xmath68 band magnitudes ( in our cosmology ) and jrgensen et al .
( 1996 ) masses gives @xmath86 with an rms scatter of 0.14 in @xmath87 .
applying this relation to the full ble92 sample ( i.e. , not just to galaxies with measured masses ) produces an offset in the color-pseudomass relation of @xmath88 , almost identical to the offset derived from the actual color - mass relation . absolute @xmath68 band magnitudes and @xmath1 colors were taken from table 6 in mcintosh et al .
the @xmath89 colors are used as they were measured using the same physical aperture size as used by ble92 .
the @xmath68 magnitudes are corrected to our cosmology and converted into pseudo - masses using eq.[pseudo.eq ] .
offsets are determined by subtracting eq .
[ colmass.eq ] from the observed colors and determining the biweight mean residual .
the offsets are @xmath90 , @xmath91 , and @xmath92 for abell 85 , abell 496 , and abell 754 respectively .
the uncertainties are a combination of random and systematic errors .
the systematic errors were calculated from the listed uncertainties in the zeropoints and @xmath93-corrections and a 16% uncertainty in the listed reddening ( schlegel , finkbeiner , & davis 1998 ) .
all three offsets are consistent with zero , in agreement with the analysis of mcintosh et al .
( 2005 ) and with earlier work by andreon ( 2003 ) , who studied a large sample of clusters in the sloan digital sky survey ( sdss ) .
the uncertainties in @xmath65 are considerably larger than for coma due to a combination of larger photometric zeropoint uncertainties and larger galactic reddening toward the mcintosh et al .
( 2005 ) clusters . for each of the 14 distant clusters described in vv07 we determined whether hst data exist in the appropriate filters ( i.e. , roughly corresponding to rest - frame @xmath67 and @xmath68 ) .
hst data are crucial because of their photometric stability and because the combination of seeing and color gradients makes aperture effects difficult to control in ground - based data .
appropriate data are available for the seven clusters listed in table [ sample.tab ] .
clusters from vv07 that are _ not _ included in the analysis are abell 665 ( @xmath94 ) , as the bluest available hst filter corresponds to rest - frame @xmath68 ; abell 2390 ( @xmath95 ) , as the hst filters correspond to rest - frame @xmath96 ; 3c295 ( @xmath97 ) , as the vv07 sample only has one galaxy more massive than @xmath98 , except 3c295 itself ; cl1601 + 42 ( @xmath99 ) , as hst observations were obtained in a single filter only ; and all three clusters with @xmath100 as the reddest available hst colors are much bluer than rest - frame @xmath1 . for two of the seven clusters in table
1 accurate colors have been published : blakeslee et al .
( 2006 ) provide acs measurements of galaxies in ms105403 ( @xmath101 ) and rxj015213 ( @xmath102 ) .
for the five remaining clusters the hst data were obtained from the hst archive , and reduced using standard techniques .
the wfpc2 images were processed using a combination of stsdas tasks and custom scripts ( see , e.g. , van dokkum et al .
for acs the photometrically and astrometrically corrected `` drz '' files were obtained , and further processing was limited to the removal of remaining cosmic rays and some other defects . galaxies in common with the vv07 samples
were deconvolved with the sclean task in the iraf stsdas package , using synthetic psfs generated with tiny tim ( krist 1995 ) .
the effect of deconvolution on the measured colors is typically @xmath103 at @xmath104 and @xmath105 at @xmath54
. colors were measured in apertures scaled to match an @xmath61 diameter aperture at the distance of the coma cluster , to allow a direct comparison with ble92 .
the measurements were converted from instrumental counts to magnitudes on the vega system using the wfpc2 and acs zeropoints . for wfpc2
the zeropoints were taken from 28.1 of the wfpc2 data handbook , which lists values for the pc chip and each of the wide field detectors separately .
for acs the zeropoints were taken from the tables on the stsci website , which include small corrections made after publication of sirianni et al .
for the two @xmath106 clusters the colors were taken from tables 1 and 2 of blakeslee et al .
a small aperture correction was applied to each galaxy , as blakeslee et al .
used apertures containing 50% of the light rather than an aperture of fixed size ( see appendix a ) .
the colors were transformed from the ab to the vega system using the ab offsets listed in sirianni et al .
( 2005 ) .
a key step in the analysis is the transformation of the observed colors in the various hst filters to rest - frame @xmath1 . following van dokkum & franx ( 1996 ) and
many later studies ( e.g. , blakeslee et al . 2006 ) we derive transformations of the form @xmath107 with @xmath108 the rest - frame @xmath67 band , @xmath109 rest - frame @xmath68 , @xmath110 and @xmath111 the observed hst filters ( listed in table 1 ) , and @xmath112 and @xmath113 determined from template spectra .
the rest - frame color is given by @xmath114 unless @xmath110 , @xmath111 correspond exactly to @xmath108 , @xmath109 the values of @xmath112 and @xmath113 have a ( modest ) dependence on the assumed spectral type .
for each cluster the appropriate values were determined in the following way .
we used synthetic templates from yi et al .
( 2003 ) with a large range of ages to establish the relation between the coefficients and the observed color in the hst filters . the yi et al . ( 2003 ) templates were chosen because they may be better calibrated in the rest - frame ultra - violet than the bruzual & charlot ( 2003 ) models ; at rest - frame wavelengths @xmath115 the two sets of models are identical .
typically , @xmath112 and @xmath113 vary by less than a few percent for ages varying from 1 12 gyr ; the amount of variation depends largely on how well the observed hst filters match the rest - frame @xmath67 and @xmath68 bands .
the relations between @xmath112 , @xmath113 and the observed colors @xmath116 were fitted with smooth functions , and the values corresponding to the median observed color of galaxies in the sample were adopted .
values of @xmath117 and @xmath118 are listed in table 1 .
= 13.3 cm the rest - frame @xmath1 colors are subject to several sources of systematic uncertainty .
the zeropoints of the hst filters are accurate to @xmath119 mag .
the dominant source of zeropoint error is thought to be the absolute calibration of vega ; e.g. , updates to the acs zeropoints in 2006 due to a recalibration of vega were @xmath120 . as we are always dividing the flux in two acs or wfpc2 passbands , such errors ( nearly ) cancel .
other effects that plague absolute calibration of ( particularly ) wfpc2 photometry , such as variations in charge transfer efficiency and the long - short anomaly , also cancel when measuring colors .
therefore , we estimate that the systematic uncertainty of the measured colors ( before correcting for reddening ) is 0.010 for wfpc2 and 0.015 for acs ( as it has a shorter calibration history ) .
uncertainties in the galactic reddening corrections are @xmath121% of the applied values ( schlegel et al .
1998 ) , which is typically @xmath122 mag for our clusters .
this error is largest for cl0024 + 16 , at 0.021 . the transformations to rest - frame colors also have some uncertainty , particularly if the observed hst filters are not well matched to rest - frame @xmath1 .
this uncertainty was assessed by varying the templates used for the transformation , comparing results for the yi et al .
( 2003 ) templates , bruzual & charlot ( 2003 ) templates , and empirical templates from coleman , wu , & weedman ( 1980 ) . for most clusters
the error is @xmath123 mag . for ms205304 and ms105403 , which have the worst match of observed and rest - frame filters among the seven clusters , the uncertainties are 0.025 and 0.030 respectively .
we note that , once again , a potentially important systematic effect cancels : the form of the synthetic rest - frame @xmath67 and @xmath68 filters , and their ab zeropoints , is the same for all distant clusters , which means that the distant clusters are all on the exact same synthetic photometric system .
there may be a systematic offset between this system and coma , although we use the same bessell ( 1990 ) passbands as landolt ( 1992 ) .
this possible offset was taken into account by adding a 0.03 mag error in quadrature to the systematic uncertainty of the coma offset .
assuming that the sources of systematic error are independent the combined error is given by @xmath124 the values of @xmath125 are listed in table 1 .
they are generally small , of order 2% 3% .
the reliability of these numbers is empirically assessed below . for each of the distant clusters in the sample
the color - mass relation is constructed by combining the rest - frame @xmath1 colors with masses determined using eq .
[ massdef.eq ] . the required effective radii and velocity dispersions
are taken from the literature and corrected to a consistent system ( see appendix a of vv07 ) .
the color - mass relations are shown in fig .
[ colmass.plot ] .
galaxies at higher redshift have bluer rest - frame @xmath1 colors , as expected from stellar evolution and from many previous studies of color evolution ( e.g. , stanford et al .
1995 , 1998 ; kodama et al .
1998 ; holden et al .
2004 ) . the evolution of the zeropoint is quantified by subtracting eq . [ colmass.eq ] from the datapoints , and determining the center of the distribution of residuals with the biweight statistic . choosing the average or the median rather than the biweight slightly changes the offset of each cluster ( always well within its error ) , but does not have an effect on our conclusions .
we verified that the slopes for the high - redshift clusters are consistent with the coma slope by fitting the color - mass relation of all galaxies in the seven distant clusters simultaneously , after removing the zeropoint evolution derived below .
the difference between this slope and the adopted coma slope is only @xmath126 .
varying the adopted slope within these limits does not change the derived offsets .
the analysis relies on the assumption that the slopes do not vary appreciably from cluster to cluster .
the samples of galaxies with mass measurements are too small to determine the slopes of the color - mass relations in individual clusters , but we note that holden et al .
( 2004 ) find no evidence for either evolution or cluster - to - cluster variation in the slope of the color - magnitude relation .
the offsets are listed in table 1 . for each cluster
the uncertainty in the offset @xmath127 is the quadratic sum of the systematic error and the uncertainty in the biweight mean ( which is caused by the galaxy - galaxy scatter within each cluster ) . for most clusters the random uncertainty is of the same order as the systematic uncertainty .
the evolution of the zeropoint of the color - mass relation is shown in fig .
[ colevo.plot ] .
the evolution is very regular for these clusters , which is perhaps not surprising given previous studies ( e.g. , kodama et al .
1998 ; andreon 2003 ; holden et al . 2006 ) and the small cluster - to - cluster scatter in the evolution of @xmath29 ratios ( see , e.g. , vv07 ) . the solid line in fig .
[ colevo.plot ] is the best fitting linear function to the data ( including the three mcintosh et al .
[ 2005 ] clusters ) : @xmath128 the observed cluster - to - cluster scatter around this line is very small at @xmath129 for all 11 clusters or @xmath130 for the eight clusters with direct mass measurements .
this remarkably small cluster - to - cluster scatter is qualitatively consistent with previous work by andreon ( 2003 ) , although we note that our study uses a much smaller number of clusters .
the small measured scatter provides an external check on the reliability of the systematic errors derived in
[ data.sec ] . focusing on the eight clusters with direct mass measurements ,
the expected scatter from measurement errors alone is 0.032 , higher than the observed scatter of 0.019 . assuming the errors are correct the probability of measuring a scatter @xmath131 is only 7% , which means the systematic errors are much more likely to be overestimated than underestimated .
a formal limit on the systematic error can be derived by requiring that the observed scatter of 0.019 has a one - sided probability @xmath132% , and parameterizing the total error for each cluster by @xmath133 , with @xmath134 the same for all clusters . from monte - carlo simulations we find that @xmath135 , i.e. , the systematic errors are consistent with zero and underestimated by at most @xmath136% .
most studies of the color - magnitude relation at high redshift have focused on the galaxy - galaxy scatter in the relation .
although this is not a focus of the present paper , we can robustly measure the scatter for galaxies with @xmath53 by removing the average trend ( eq . [ colevo.eq ] ) and combining the galaxies in all distant clusters .
the observed biweight scatter in the full sample @xmath137 .
the scatter expected from observational error is less than @xmath105 , and we estimate that the intrinsic scatter is @xmath138 .
averaging the three clusters at @xmath139 , the two clusters at @xmath140 , and the two clusters at @xmath141 , we find no evidence for evolution with redshift : the scatter in each redshift bin is consistent with the value for the full sample .
the value for the scatter that we find is significantly lower than results from previous studies ( e.g. , holden et al . 2004 ; blakeslee et al .
this is most likely due to a dependence of the scatter on galaxy mass .
we also note that the color - mass relation is expected to have a smaller scatter than the color - luminosity relation even within the same sample of galaxies , if the scatter is due to age variations ( see van dokkum et al .
as discussed in [ method.sec ] luminosity and color evolution each depend on the imf and on the age of the stellar population , but the age - dependence drops out when comparing the amount of luminosity evolution to the amount of color evolution .
figure [ direct.plot ] shows the measured evolution in @xmath142 as a function of the evolution in @xmath1 .
this figure is the equivalent of fig .
[ mlcol.plot](c ) .
the values for @xmath143 are taken directly from vv07 , and the values for @xmath65 are listed in table 1 . as expected , there is a clear relation , with galaxies becoming both bluer and more luminous at earlier times .
the solid line is a linear fit to the data ( taking the errors in both parameters into account ) of the form @xmath144 the residuals from this fit are consistent with expectations from the uncertainties in the datapoints .
the solid and dashed red lines in fig .
[ fit.eq ] are predictions for a salpeter - like evolution over the relevant range of ages ; hence the term `` salpeter - like '' or `` standard '' to denote imfs with slope @xmath145 at @xmath15 . ]
imf with @xmath146 and two different metallicities .
remarkably , the observed relation is much steeper than expected from a standard imf , even for solar metallicity . for a given color evolution
the luminosity evolves faster than expected , indicating a top - heavy imf with a relatively large fraction of rapidly evolving massive stars . the logarithmic slope of the imf follows directly from eqs .
[ kappa3.eq ] , [ solkappa3.eq ] , and [ fit.eq ] .
we find @xmath147 for super - solar metallicity and @xmath148 for solar metallicity , where the uncertainties reflect the 68% confidence interval .
both values are well below the canonical salpeter value of @xmath146 .
negative values of @xmath30 have a large systematic uncertainty , as they require significant extrapolation of the maraston ( 2005 ) models .
therefore , the results can best be expressed as upper limits , particularly for the super - solar metallicity model .
the 90% confidence upper limits are @xmath149 for the super - solar model and @xmath150 for solar metallicity . the salpeter value can be ruled out at @xmath151% confidence .
= 17.5 cm directly fitting @xmath143 as a function of @xmath65 , as done in [ direct.sec ] , has several advantages : the same galaxies are used to measure the relevant parameters , limiting selection effects ; redshift - dependent selection effects such as progenitor bias cancel ; and the imf is the only free parameter ( at fixed metallicity ) , as the formation epoch of the stars @xmath152 also cancels .
however , constraining @xmath152 better is of great interest in its own right , and allows us to associate a particular time in the history of the universe with the imf result .
here we fit the redshift evolution of the @xmath2 ratio and the @xmath1 color simultaneously , and derive joint constraints on the slope of the imf @xmath30 and the star formation epoch @xmath152 .
the goals are to determine the luminosity - weighted star formation epoch of massive cluster galaxies in a self - consistent way and to verify the results of [ direct.sec ] .
although the additional parameter implies more freedom in the fits this is compensated by the fact that more data can be used , as we are no longer restricted to the sample of eight clusters which have both @xmath29 and color information .
the data are shown in fig .
[ fitboth.plot ] .
panel ( a ) shows offsets in @xmath143 from vv07 , and panel ( b ) shows the offsets in @xmath1 determined in the present study . the dataset is more extensive than used in the analysis of [ direct.sec ] .
the @xmath29 sample includes data from the sdss and for seven additional distant clusters .
the color sample includes the three clusters from mcintosh et al .
( 2005 ) , which help constrain the low redshift end .
small corrections for progenitor bias have been applied , following vv07 : @xmath153 for @xmath143 and @xmath154 for @xmath65 .
the color correction was chosen to be consistent with the correction for luminosity evolution and with eq .
[ fit.eq ] . as shown in [ prog.sec ] , setting the progenitor bias to zero leads to slightly higher formation redshifts but has virtually no impact on the imf constraints .
the data are fit by creating models over a grid of @xmath30 and @xmath37 , with @xmath30 the logarithmic slope of the imf and @xmath37 the mean luminosity weighted formation time of the stars . for each value of @xmath30 corresponding values of @xmath40 and @xmath41
are determined using eqs .
[ kappa1.eq ] and [ kappa2.eq ] .
next , eqs.[mlt.eq ] and [ colt.eq ] are used to determine the expected evolution with cosmic time for each combination of @xmath30 and @xmath37 .
the @xmath155 values of the fits are then calculated by @xmath156 the fit has four free parameters : @xmath30 , @xmath37 ( or equivalently @xmath152 ) , and the normalizations to the @xmath29 and color data .
figure [ contour.plot ] shows the 68% , 95% , and 99% confidence intervals in the @xmath30 , @xmath37 plane , as determined from the @xmath157 values appropriate for two interesting parameters .
solid contours are for super - solar metallicity ( [ fe / h]=0.35 ) maraston ( 2005 ) models and broken contours are for solar metallicity models .
flat imfs and high formation redshifts are clearly preferred , for both metallicities .
the red , blue , and black dots indicate examples of models with different parameters .
the fits to the luminosity and color evolution for these three models are shown in fig .
[ fitboth.plot ] , and illustrate how the data simultaneously constrain @xmath30 and @xmath37 .
the red model has a standard salpeter imf and a low formation redshift @xmath158 .
this model provides an excellent fit to the evolution of the @xmath2 ratio , as shown in fig .
[ fitboth.plot](a ) and discussed in vv07 .
however , this model fits the data in fig .
[ fitboth.plot](b ) poorly : the observed color evolution is much slower than predicted by this young model .
the blue model is a salpeter model with a high formation redshift @xmath159 .
this model fits the slow color evolution very well , but underpredicts the @xmath29 evolution in fig .
[ fitboth.plot](a ) .
the only models that fit the data in both panels simultaneously have high formation redshifts ( to fit the colors ) and a flat imf ( to fit the @xmath29 ratios ) .
the black model has @xmath159 and @xmath160 .
this model provides an excellent fit to the data in both panels .
the formal best - fitting values are @xmath161 , @xmath162 for super - solar metallicity , and @xmath163 , @xmath164 for solar metallicity .
the best - fit values for the imf slope are in very good agreement with those derived in [ direct.sec ] . as discussed earlier negative values for @xmath30 have large systematic uncertainties , and may not be very meaningful .
the super - solar solutions are therefore somewhat difficult to interpret .
formal 68% confidence intervals for the solar metallicity solution are @xmath165 and @xmath166
. these results are consistent with the analysis of the @xmath29 evolution alone in vv07 , where it was noted that the best - fitting @xmath167 for top - heavy imfs with @xmath168 .
although straightforward in principle , the imf test discussed in this paper has several associated uncertainties , some having to do with the data and some with the models that are used to fit the data .
every cluster for which the required data were available in the literature and/or the hst archive was included in the analysis , and as they come from a wide range of projects by many different research groups this may have introduced systematic errors . in particular , it may be that the sample of clusters with fp measurements is somehow different from the sample of clusters with color measurements .
this potential bias is explicitly addressed in [ direct.sec ] , where the analysis was limited to clusters in common between the two samples . as a further test
, the analysis in [ joint.sec ] was repeated after removing the three clusters with @xmath169 from the sample .
these clusters have a large weight in the @xmath29 analysis as they are at the highest redshifts , and they have no counterparts in the color sample .
the effects are small , but not negligible : for the super - solar model the best - fitting imf slope changes from @xmath170 to @xmath171 and the best fitting formation redshift changes from @xmath172 to @xmath173 when the highest redshift clusters are removed .
redshift - dependent selection effects such as progenitor bias are not a major concern in this particular study . if we systematically miss the youngest progenitors of today s clusters , or the youngest progenitors of today s early - type galaxies within clusters , the formation redshifts that we derive may be too high but the imf results should be robust .
the imf is constrained by the ratio of the amount of luminosity evolution to the amount of color evolution , and progenitor bias would ( to first order ) equally affect the colors and luminosities .
this can be tested explicitly by varying the progenitor bias correction to the @xmath29 ratios and colors , and repeating the analysis .
increasing progenitor bias by a factor of two gives the same imf slope ( @xmath170 ) as the model adopted in
[ joint.sec ] , and a lower formation redshift ( 4.8 rather than 7.6 ) . setting the progenitor bias to zero again gives the same imf slope ( @xmath170 ) , and a formal best - fitting formation redshift @xmath174 . as discussed in
[ syserrors.sec ] the low cluster - to - cluster scatter in the color evolution implies that the errors in the photometry are more likely overestimated than underestimated .
nevertheless , absolute photometry is challenging , and it is difficult to rule out subtle systematics in the various corrections and transformations that have been applied to the data . increasing the errorbars on the colors by 50%
( much more than is plausible see [ syserrors.sec ] ) obviously loosens the constraints on the imf , but not by a large amount . repeating the analysis of [ direct.sec ] changes the upper limits on the imf slope from @xmath149 ( for super - solar metallicity ) and @xmath150 ( for solar metallicity ) to @xmath175 and @xmath176 respectively . aside from general concern about the assumed errors ,
a specific worry is the tie between the low redshift data and the ( homogeneous ) synthetic photometric system that is used for the distant clusters .
absolute photometry in the @xmath67 band is particularly difficult : the bandpass is strongly influenced by the detector and the atmosphere , and the spectral energy distributions of old galaxies have a very steep slope around 3800 .
we dealt with this issue in [ syserrors.sec ] by adding a 3% systematic uncertainty in the coma offset .
a more conservative approach is to simply discard coma altogether .
interestingly , removing coma has the effect of tightening the limits on the imf slope . as an example , for solar metallicity the 90% upper limit changes from 0.9 to 0.3 .
the reason for this is readily apparent from fig .
[ direct.plot ] : coma is the point with the largest deviation from the best - fitting line , and the relation becomes much steeper when it is removed .
we note that the mean @xmath2 ratio of coma galaxies may also be somewhat anomalous , as it deviates from the best - fitting relation in fig .
[ fitboth.plot]a . adopting the @xmath2 ratio from this fit rather than
the actual value brings the cluster in close agreement with the solid line in fig .
[ direct.plot ] . the evolution of the mean @xmath29 ratio , shown in fig .
[ fitboth.plot](a ) , is derived from the evolution of the fundamental plane ( fp ) relation ( djorgovski & davis 1987 ) .
as discussed in detail in several papers ( e.g. , franx 1993 ; van dokkum & franx 1996 ; treu et al . 2001 ) the empirical fp relation can be rewritten as a relation between @xmath29 ratio and mass , if it is assumed that @xmath177 ( with @xmath56 the stellar velocity dispersion and @xmath58 the effective radius ) and early - type galaxies are a homologous family . if these assumptions are valid the observed evolution of the fp should track the evolution of the zeropoint of the underlying @xmath29 mass relation . however , this may not be the case if galaxies undergo significant structural changes over time , due to merging or other processes ( see , e.g. , almeida , baugh , & lacey 2007 ) .
strong gravitational lenses provide an independent check on the fp interpretation .
current results indicate that the evolution of the @xmath29 ratio as derived from strong lenses is consistent with that derived for field early - type galaxies ( rusin & kochanek 2005 ) , although the systematic uncertainties in this comparison are still fairly substantial ( see , e.g. , treu & koopmans 2004 ; van der wel et al .
2005 ; rusin & kochanek 2005 ) .
recently van der marel & van dokkum ( 2007 ) used spatially - resolved photometric and dynamical observations of cluster early - type galaxies at @xmath178 to directly test the assumption that the fp tracks @xmath29 evolution .
@xmath29 ratios derived from modeling the spatially - resolved data were directly compared to those determined from the fp .
interestingly , the validity of the fp for determining @xmath29 ratios appears to depend on the mass : for galaxies with masses @xmath179 or velocity dispersions @xmath180the @xmath29 ratios are consistent with each other , but for low mass galaxies the fp appears to systematically underestimate the @xmath29 ratio .
although these results suggest that the traditional interpretation of fp evolution is substantially correct at the high mass end , further tests would be valuable , particularly at @xmath181 .
another complication is that the measured @xmath29 ratio is not identical to the stellar @xmath29 ratio calculated in stellar population synthesis models . in the models @xmath182 is the sum of the mass of all living stars and the mass of stellar remnants ( black holes , neutron stars , and white dwarfs ) .
by contrast , the measured mass includes all possible contributions , including dark matter and gas .
although the amount of dark matter in elliptical galaxies is still somewhat uncertain ( e.g. , romanowsky et al . 2003 ; mamon & okas 2005 ; cappellari et al .
2006 ) , its contribution within the effective radius is probably small ( mamon & okas 2005 ; koopmans et al .
the contribution of gas to the total mass is also uncertain .
one would expect substantial amounts of gas and dust from stellar mass loss ( see , e.g. , goudfrooij et al .
1994 ) , qualitatively consistent with spitzer observations ( temi , brighenti , & mathews 2007 ) .
however , the data appear to be inconsistent with simple expectations , and temi et al .
( 2007 ) suggest that the gas is transported outward , possibly due to agn activity .
furthermore , in galaxy clusters the galaxies move through very hot , diffuse gas at great speed and it seems likely that any cold gas coming from stellar winds is fairly efficiently stripped or heated .
although dark matter and gas could contribute to the measured @xmath29 ratio , these contributions will to large extent cancel when comparing galaxies over a range of redshifts . in the maraston ( 2005 )
models the amount of mass loss over the relevant age range from 5gyr to 12gyr is only @xmath183% for a salpeter ( 1955 ) imf and @xmath184% for a kroupa ( 2001 ) imf . even if half of this mass was retained rather than lost to the intracluster medium , the @xmath29 evolution would change by only @xmath185% .
uncertainties in the stellar population synthesis models are perhaps the largest source of error , and at the same time they are notoriously difficult to quantify .
the fundamental problem is that the main direct tests and calibrations of the models are offered by galactic open clusters and globular clusters , but that these may not be representative for the integrated stellar populations of massive early - type galaxies .
recently the confidence in these models was somewhat shaken by the large differences that exist between predictions of bruzual & charlot ( 2003 ) and maraston ( 2005 ) . as discussed in detail by maraston ( 2005 ) and others ( e.g. , van der wel et al .
2006b ; maraston et al .
2006 ) , the predicted rest - frame optical to near - infrared colors of simple stellar populations can deviate by @xmath186 magnitudes for identical input metallicities and ages .
the differences are in large part due to a different numerical implementation of the thermally pulsating asymptotic giant branch ( tp - agb ) phase of stellar evolution , and are therefore largest for ages 0.52 gyr and colors which include a near - infrared band ( such as @xmath187 ) .
our analysis avoids the region of parameters space where the differences between the models are most pronounced ( ages of 0.52gyr and rest - frame near - infrared passbands ) .
the models are better calibrated ( and in much better agreement ) for ages @xmath188gyr and rest - frame optical passbands , largely because the contribution of giants to the integrated light is significantly smaller . however , subtle differences between the models also exist at ages @xmath188gyr ; as an example , the @xmath187 colors of the maraston ( 2005 ) models are slightly bluer than those of bruzual & charlot ( 2003 ) ( see van der wel et al
. 2006b ) .
to get some handle on the effects of the choice of model the derivation of eq .
[ kappa3.eq ] was repeated for a solar metallicity bruzual & charlot ( 2003 ) model . by comparing the color evolution to the @xmath29 evolution for two different imf slopes ( @xmath17 and
@xmath189 ) we find @xmath190 .
the relation is less steep than that of the solar metallicity maraston ( 2005 ) model , and is close to maraston s super - solar metallicity model ( the solid line in fig .
[ direct.plot ] ) . for a salpeter imf and solar metallicity the bruzual & charlot ( 2003 ) model
provides an even worse fit to the data than the maraston ( 2005 ) model , and the measured slope of the @xmath29 color relation implies @xmath191 ( compared to @xmath192 for the solar metallicity maraston model ) .
we infer that our conclusions are robust in the context of presently popular models , but note that there are significant variations in the model predictions even in the rest - frame optical . finally , an important check on the entire enterprise is whether the models can match the absolute colors of the galaxies .
so far , we have not used the absolute colors in any way and only considered the rate of color _ evolution _ , leaving the normalization a free parameter .
the absolute colors are difficult to interpret , as they suffer from strong degeneracies between age , metallicity , and dust content
. nevertheless , the models should be able to reproduce the observed colors of massive galaxies in coma for plausible combinations of these parameters .
figure [ colcol.plot ] shows the location of galaxies in the coma cluster in a color - color diagram of @xmath193 vs. @xmath1 , along with model predictions of maraston ( 2005 ) .
all models have an age of 12gyr , corresponding to @xmath167 .
the models have approximately the correct @xmath1 colors at @xmath194 for @xmath195 , and are therefore self - consistent .
the fact that the agreement is particularly good for models with solar metallicity and @xmath168 is probably coincidence , given the uncertainties .
as is well known old stellar populations can include very luminous hot stars .
these stars are rare , but due to their high luminosity they can have a significant effect on the integrated luminosity of a stellar population , particularly in blue passbands .
these hot stars fall in three broad categories : blue horizontal branch stars ( e.g. , zinn , newell , & gibson 1972 ; rich et al .
1997 ) , extreme horizontal branch stars ( e.g. , dorman , rood , & oconnell 1993 ) , and blue stragglers ( bss ) ( e.g. , sandage 1953 ; bailyn 1995 ) .
blue horizontal branch stars are generally included in current stellar population synthesis models .
extreme horizontal branch stars are thought to be responsible for the uv - upturn in elliptical galaxies at @xmath196(burstein et al .
1988 ) , and should not affect the @xmath1 colors ( see , e.g. , yi et al .
2003 , 2005 ) .
blue stragglers are not included in stellar population synthesis models , probably because conditions for their formation are still not well understood .
however , they can have a significant effect on the integrated near - uv and optical colors of a stellar population .
xin & deng ( 2005 ) find that the contribution of bss to the integrated @xmath197 color of old open clusters is typically @xmath198 magnitudes .
their contribution to the @xmath1 color has not been measured directly , but is expected to be @xmath18 magnitudes or so . in practice , bss are not included when fitting isochrones to the stellar populations of open clusters or globular clusters , but we do not have that luxury for unresolved stellar populations .
the effect of a bs population such as that in m67 would be larger than the imf effects discussed in this paper , and there is no a priori reason to suppose that bss do not exist in elliptical galaxies ( see , e.g. , stryker 1993 ) .
nevertheless , they probably do not have a very large effect on our results .
first , the relevant question is not how much bss contribute to the integrated colors , but to what extent they influence their _ evolution _ from @xmath199gyr ( the age at @xmath200 for @xmath201 ) to @xmath202gyr ( the age at @xmath194 ) . to have a significant impact on our results
bss would have to have been absent in elliptical galaxies for five or more gyr and then have `` turned on '' .
this seems rather contrived , particularly since the turn - off mass changes by only @xmath203 over this age range .
if anything , one might expect bss to be slightly more prevalent at younger ages .
one of the preferred models for their formation is through the coalescence of binary stars ( mateo et al .
1990 ) , and if binaries start out with a flat distribution of orbital parameters the rate at which some of them turn into bss will gradually decrease . coupled with the fact that bs lifetimes are short compared to the main sequence stars , this could lead to a gradual decline in the blue straggler population .
more to the point , xin & deng ( 2005 ) have determined the contribution of bss to the integrated light of open clusters with a large range of ages .
they find no correlation between the contribution of bss to the integrated colors and the age of the cluster . as a result
, the integrated colors of clusters with ages varying from 1gyr to 8gyr are reasonably well described by a single stellar population synthesis model with a constant offset .
another reason why the effects of bss are probably fairly mild is that it is difficult to hide them completely in nearby elliptical galaxies .
if bss are as prevalent in elliptical galaxies as they are in open clusters , they would lead to relatively blue near - uv colors and strong balmer absorption lines .
population synthesis models would ( correctly ) infer the presence of a and f main sequence stars , and ( incorrectly ) attribute them to a secondary young stellar population , and/or assign an overall young luminosity - weighted age to the galaxy . as pointed out by schiavon , caldwell , & rose ( 2004 ) , if a galaxy had the same stellar population as m67 it would be classified as `` e+a '' ( dressler & gunn 1983 ) to signify the presence of a post - starburst component in addition to an older component .
xin & deng ( 2005 ) quantify these effects and conclude that the ages of unresolved stellar populations will typically be underestimated by a factor of @xmath7 if blue stragglers are present in similar numbers as in open clusters . turning to many years of population synthesis modeling of massive elliptical galaxies in clusters ,
such dramatic effects can be safely ruled out . as an example , thomas et al .
( 2005 ) infer ages of @xmath202gyr ( i.e. , close to the age of the universe ) for cluster galaxies with @xmath204 , not @xmath205gyr as one might expect from a bs - rich stellar population .
qualitatively , the red colors and weak balmer lines of massive cluster ellipticals place strong limits on the relevance of bss in these galaxies .
finally , we note that trager et al .
( 2000b ) argue that bss do not contribute significantly to the integrated light even in elliptical galaxies that _ do _ have enhanced balmer lines and signs of young populations ( typically relatively low mass field ellipticals ) . rather than discuss their arguments , we note that the finding that the derived age of ellipticals appears to correlate with the presence of morphological fine structure ( schweizer & seitzer 1992 ) provides additional support for the notion that late star formation , rather than a bs population , is generally responsible for the young appearance of some nearby ellipticals .
we are left to wonder _ why _ bss seem to be deficient in elliptical galaxies , as compared to old open clusters .
no speculation will be offered here , as this question is well outside the scope of the present study . as argued in
[ implem.sec ] the population of massive early - type galaxies in clusters is not thought to have experienced significant star formation at redshifts @xmath206 .
young stellar populations have been seen in some field ellipticals ( e.g. , trager et al .
2000a ; thomas et al .
2005 ) , but a large body of observational evidence supports the notion that massive cluster galaxies are remarkably homogeneous ( e.g. , bower et al . 1992a ; stanford et al .
1995 , 1998 ; ellis et al . 1997
; kodama et al . 1998 ; van dokkum et al .
2000 ; holden et al . 2004 ; thomas et al .
2005 ; tran et al . 2007 ) . this apparent homogeneity may be misleading and mask more complex evolution ( van dokkum & franx 2001 ) , but this is not likely for the most massive galaxies ( holden et al .
it may well be that cluster early - types experienced more complex evolution at @xmath207 , in particular during and before virialisation of the clusters in which they now live , but this does not affect the analysis
. the parameter @xmath152 does not refer to a single star formation event , but reflects the luminosity - weighted mean formation epoch of the stars . as shown explicitly in van dokkum et al .
( 1998b ) color and luminosity evolution are nearly identical for stellar populations with different star formation histories but the same luminosity - weighted age , as long as star formation terminated at least @xmath208gyr prior to the epoch of observation .
dust needs to be treated separately , as significant amounts of gas and dust are expected to exist in early - type galaxies as a result of winds emanating from massive stars ( see also [ whatml.sec ] ) .
cluster galaxies at @xmath209 may still have retained a significant fraction of the dust expelled by massive stars , whereas it may have been stripped , blown out , or obliterated at later times ( see , e.g. , goudfrooij et al .
1994 ; temi et al .
2007 ) . quantifying these effects
is very difficult , as they not only depend on the amount of mass loss but also on the detailed processes and timescales for cooling and heating of the dust .
qualitatively , one might expect higher reddening at @xmath209 and therefore a lower apparent color evolution from @xmath209 to @xmath210 .
the true color evolution would then be stronger , and the data might be consistent with a normal imf in combination with a relatively low stellar age .
however , dust would not only affect the colors but also the luminosities .
the relation between reddening and extinction in elliptical galaxies is not well known , but is probably not very different from that in the milky way ( e.g. , goudfrooij et al .
1994 ) . for reference , arrows in fig .
[ direct.plot ] show the effects of milky way - like dust and of the ( relatively grey ) calzetti ( 1997 ) extinction curve .
calzetti - like dust would have essentially no effect on the analysis , as the dust vector is nearly parallel to the observed relation between @xmath29 ratio and color .
milky way - like dust would have an effect , but would lead to unrealistically low ages for the galaxies . to change the observed relation to the one expected for a standard imf the @xmath211 galaxies would have to be reddened by @xmath18 mag in @xmath1 , corresponding to a @xmath212-band extinction of @xmath213 mag . the implied luminosity evolution from @xmath211 to the present
would then be @xmath214 magnitudes ( instead of 1.1 magnitudes ; see vv07 ) .
this in turn would mean that massive cluster galaxies formed their stars at @xmath215 , which is ruled out by many independent observations ( e.g. , kelson et al .
2001 ; thomas et al . 2005
; kodama et al . 2007 ) .
the main result from the fits in [ apply.sec ] is that stellar population synthesis models with a standard , salpeter - like imf in the region around 1 are not able to simultaneously fit the luminosity and color evolution of massive cluster galaxies .
by contrast , models with flat imfs and high stellar formation redshifts provide excellent fits . the slow color evolution measured here
is not easily explained by invoking sample selection effects , photometric errors , a blue straggler population , or differential dust extinction .
the formal limits on the imf slope and formation redshift are obviously very uncertain , as they do not take any of the effects discussed above into account .
furthermore , negative imf slopes are outside of the parameter space covered by the models , which makes the formal super - solar metallicity solutions somewhat suspect . in
what follows the formal solar metallicity solutions ( @xmath3 and @xmath216 ) will be used as a `` default '' model , but it should be kept in mind that the errorbars almost certainly underestimate the true uncertainty .
many previous studies have determined or constrained the properties of the stellar populations of early - type galaxies .
although direct comparisons are difficult because of differences in sample selection and methodology , it is important to verify that our results are broadly consistent with the large body of literature that exists on this topic . as discussed in detail in vv07 ,
the values shown in fig.[fitboth.plot]a are mostly determined from literature data , and therefore `` automatically '' consistent with a large body of previous work .
the only exception is jrgensen et al .
( 2006 ) , who find substantially slower evolution in the first version of their published paper . however , this was due to a conversion error , and the vv07 results are fully consistent with the erratum to jrgensen et al .
( 2006 ) published in 2007 .
the evolution of the rest - frame @xmath1 color can not be directly compared to previous work , as most studies choose to compare the observed colors to redshifted model predictions rather than convert the data to the rest - frame .
furthermore , color evolution has not been studied previously for mass - limited samples of cluster galaxies . nevertheless , it is interesting to compare the constraints on the stellar ages that are reported in the literature , as they should reflect the measured rate of color evolution .
most studies of the color evolution of early - type galaxies in clusters find very old ages , in qualitative agreement with the results found here ( e.g. , stanford et al .
1995 , 1998 ; kodama et al . 1998
; koo et al . 2005 ) . in particular , no studies have reported formation redshifts as low as @xmath7 for the most massive galaxies in clusters , which is the value implied by the evolution of the @xmath29 ratios for a standard imf .
the most comprehensive study to date is by holden et al .
( 2004 ) , who combined data from stanford et al .
( 1998 ) with new measurements to obtain a combined sample of 31 clusters at @xmath217 with multi - band photometry .
they find @xmath218 from solar metallicity bruzual & charlot ( 2003 ) models , in good agreement with our results .
it should be noted that the holden et al .
( 2004 ) study may suffer from some of the same systematic effects as the work described here ; for example , an emerging blue straggler population would influence all studies of color evolution .
kelson et al .
( 2001 ) study the evolution of the balmer absorption line strengths of early - type galaxies in clusters at @xmath219 , and their data offer an independent check on the results derived here .
the balmer lines should behave like the @xmath1 colors , in the sense that they are fairly insensitive to the imf and their rate of evolution is mostly determined by the age of the stellar population .
importantly , the sources of systematic error ( both in the data and in the models ) are very different from those that affect the colors ( see , e.g. , kelson et al .
2006 ) . the kelson et al .
( 2001 ) data are shown in fig .
[ kelson.plot ] .
the data are compared to two solar metallicity bruzual & charlot ( 2003 ) models : a model with a standard imf and a young age , and a model with a top - heavy imf and an old age . as discussed in vv07 and in [ joint.sec ] these are the only two classes of models which provide good fits to the observed evolution of the @xmath29 ratio .
the young model is clearly inconsistent with the data , and can be rejected with high confidence . by contrast , the top - heavy / old model fits the data very well .
this model is the formal best fit to the @xmath29 and color evolution ( as discussed in [ fitsum.sec ] ) , but other models with formation redshifts in the range 36 also fit well .
the evolution of the balmer lines thus provides strong independent support for the analysis in [ apply.sec ] .
we note that the inferred ages are consistent with the analysis in kelson et al .
( 2001 ) : they give 95% confidence lower limits to @xmath152 of 2.4 and 2.9 for two different stellar population synthesis models .
field early - type galaxies probably evolve in more complex ways than their counterparts in clusters .
galaxy - galaxy mergers are expected to be rare in virialized clusters ( e.g. , makino & hut 1997 ) but could be common in groups ( e.g. , van dokkum 2005 ; bell et al .
2006 ) , and the same is probably true for late star formation ( e.g. , trager et al .
field samples may therefore not be suitable for measuring the subtle effects discussed in this paper . nevertheless , it is interesting to note that van der wel et al .
( 2006a ) find that bruzual & charlot ( 2003 ) models with a standard imf do not fit the observed relation between @xmath187 color and @xmath2 ratio of field early - type galaxies at @xmath220 .
models with top - heavy imfs fit better , although maraston ( 2005 ) models with a standard imf also provide good fits .
recent studies of faint @xmath221selected samples have identified a population of apparently `` red and dead '' galaxies at @xmath222 ( franx et al .
2003 ; labb et al . 2005
; daddi et al . 2005 ; kriek et al .
deep rest - frame optical spectra demonstrate strong balmer breaks and no line emission in a substantial fraction of massive galaxies at these redshifts , implying that star formation in these objects mostly terminated at @xmath223 or earlier ( kriek et al .
these objects are found in blind surveys of relatively small fields , and it seems likely that many galaxies in ( proto-)clusters terminated star formation even earlier , and/or that the fraction of ( nearly ) passive galaxies is higher in those environments ( see , e.g. , thomas et al .
2005 ; steidel et al . 2005 ; quadri et al . 2007 ;
kodama et al .
2007 ) . assuming that these galaxies are not `` rejuvenated '' at later times ( which is entirely possible ) , the presence of substantial numbers of apparently old galaxies at this redshift places an interesting constraint on the luminosity - weighted ages of massive early - type galaxies . for any star formation history the time of
_ last _ star formation will be more recent than the luminosity - weighted mean star formation epoch . after a @xmath224-function ,
the simplest star formation history is a top - hat with a constant star formation rate from @xmath225 to @xmath226 .
such models may actually be fairly realistic ( as compared to exponentially declining models , for instance ) , as they superficially resemble the star formation histories of massive galaxies in hydrodynamical simulations ( nagamine et al .
as shown in van dokkum et al .
( 1998b ) , there is a simple relation between the luminosity - weighted age @xmath227 and @xmath228 in these models : @xmath229 . assuming that @xmath230 ( extrapolating from the kriek et al .
results for red field galaxies ) , the luminosity - weighted mean formation epoch @xmath231 for @xmath232gyr .
luminosity - weighted formation redshifts of @xmath233 are entirely consistent with the old / top - heavy model that is implied by the analysis in [ apply.sec ] .
formation redshifts of 22.5 , which are implied by the @xmath29 ratio evolution for salpeter - like imfs ( see vv07 and [ apply.sec ] ) may be difficult to reconcile with the presence of apparently old field galaxies at @xmath234 .
we note , however , that this type of evidence should be treated with caution , as the old high redshift objects could constitute only a small fraction of the progenitors of today s early - type galaxies .
many studies have constrained the properties of the stellar populations of nearby early - type galaxies using absorption line strengths ( e.g. , peletier 1989 ; worthey et al .
1992 ; trager et al .
2000a ; thomas et al .
linestrengths are generally more helpful than colors , as they suffer slightly less from the strong degeneracies that exist between ( particularly ) age and metallicity ( e.g. , worthey 1994 ) .
one of the difficulties in linestrength studies is that the abundance ratios of early - type galaxies do not match solar models ( peletier 1989 ) , which in itself may be indirect evidence for a top - heavy imf at early times ( worthey et al .
thomas et al .
( 2005 ) argue that models which explicitly include non - solar abundance ratios can date the star formation epoch of early - type galaxies .
these models indicate that cluster early - type galaxies formed their stars at @xmath235 , in good agreement with the ages derived here for top - heavy imfs . in [ apply.sec ] we were only concerned with the evolution of the derived @xmath29 ratios , and not with their absolute values .
the absolute @xmath29 ratios of nearby elliptical galaxies provide a powerful additional constraint on the imf , as the stellar @xmath29 ratio implied by the combination of the imf and the observed luminosity should not exceed the dynamical @xmath29 ratio .
as discussed in 1 the observations of cappellari et al .
( 2006 ) are consistent with the stellar mass implied by a chabrier ( 2003 ) imf and rule out imfs which are significantly steeper than the salpeter value , because they have too much mass locked up in low mass stars .
perhaps surprisingly , the dynamical @xmath29 ratios also rule out top - heavy imfs of the form fitted in
[ apply.sec ] .
these imfs have less low mass stars than a standard salpeter ( 1955 ) imf but many more high mass stars . as a result , the stellar mass function at late times is `` remnant heavy '' , dominated as it is by white dwarfs , neutron stars , and black holes .
maraston ( 1998 ) shows that the @xmath29 ratio at old ages reaches a minimum for @xmath236 : for larger values the imf is dwarf - dominated and for smaller values the imf is remnant - dominated .
specifically , solar metallicity maraston ( 2005 ) models predict @xmath237 at 12gyr for @xmath17 and @xmath238 for @xmath168 ! such high @xmath29 ratios are an order of magnitude higher than measured ( see van der marel 1991 ; cappellari et al .
2006 ; gavazzi et al .
2007 , and many other studies ) , which means that imfs with a constant logarithmic slope @xmath168 over the entire mass range @xmath239 can effectively be ruled out . in the next section
we will explore imfs which have a salpeter - like slope at high masses and @xmath240 in the region around 1 .
such imfs are physically more plausible than imfs with a constant slope and do not violate dynamical limits on the stellar @xmath29 ratios of nearby galaxies ( see [ mass.sec ] ) .
in the maraston ( 2005 ) models described in [ starpop.sec ] the imf is parameterized as a power law with constant slope @xmath30 over the entire mass range @xmath42 .
the best - fitting slope of @xmath240 implies an extremely top - heavy mass function , with a very large number of massive stars
. however , other forms of the imf are also consistent with the data , as the analysis of [ apply.sec ] is completely insensitive to the imf at masses exceeding the turn - off mass at @xmath199gyr ( @xmath242 ) .
a physically more plausible imf would have a chabrier ( 2003 ) form , with a fixed high - mass slope of @xmath145 and a varying characteristic mass @xmath6 . as will be quantified later , such imfs are not only physically motivated but are also entirely consistent with the dynamical @xmath29 ratios measured for nearby ellipticals .
so far , no stellar population synthesis modeling has been done with chabrier imfs with varying @xmath6 .
however , we can use the fact that our observations probe the imf only over a limited stellar mass range to derive an approximate relation between @xmath6 and the slope of the imf as parameterized in the maraston ( 2005 ) models .
the chabrier ( 2003 ) imf has the form @xmath243 & ( m\leq 1\,m_{\odot } ) \\ a_h m^{-x } & ( m>1\,m_{\odot } ) , \end{array } \right.\ ] ] with @xmath244 , @xmath245 , @xmath246 , @xmath247 , and @xmath146 , and is shown by the dashed lines in fig.[chabrier.plot ] . as shown in chabrier ( 2003 ) this imf is very similar to the kroupa ( 2001 ) imf , and both imfs are nearly identical to the salpeter ( 1955 ) imf for @xmath248 . the bottom panel of fig .
[ chabrier.plot ] shows the derivative @xmath249 , which can be thought of as the local logarithmic slope of the imf .
the slope gradually steepens until leveling off at the salpeter value for masses @xmath248 .
the derivative of the chabrier ( 2003 ) imf has a discontinuity at 1 , where the log - normal part of the imf and the power law part connect .
this discontinuity is rather minor and has no practical implications , but it becomes more pronounced when considering higher values of @xmath6 .
therefore , we adopt an extension of the chabrier ( 2003 ) disk imf parameterization which does not produce discontinuities in the derivative and explicitly accommodates a varying @xmath6 : @xmath250 & ( m\leq n_c m_c ) \\ a_h m^{-x } & ( m > n_c m_c ) , \end{array } \right.\ ] ] with @xmath251 , @xmath252 , and @xmath56 , @xmath253 , and @xmath30 identical to eq .
[ chabrier.eq ] ( see chabrier ( 2003 ) for an alternative parameterization of imfs with a high characteristic mass ) .
the behavior of this modified chabrier imf for @xmath254 , @xmath255 , and @xmath256is shown in fig .
[ chabrier.plot](a ) and ( b ) . for @xmath254
this imf is nearly identical to the chabrier ( 2003 ) disk imf .
the observations discussed in this paper are sensitive to the imf in a limited stellar mass range near 1 .
the @xmath212-band luminosity - weighted mean stellar mass of a stellar population is typically slightly smaller than the turn - off mass , and ranges from 1.00.8 for ages of 512gyr . as is evident from fig .
[ chabrier.plot ] the imf can be approximated by a powerlaw over this small mass range , and the slope of this powerlaw depends on @xmath6 .
we find @xmath257 with @xmath258 the logarithmic slope of the imf near @xmath259 .
this expression relates salpeter - like imfs with varying slope @xmath30 , as used in the maraston ( 2005 ) models and eqs .
[ kappa1.eq ] and [ kappa2.eq ] , to physically more plausible chabrier - like imfs with fixed high mass slope and varying characteristic mass .
the best - fitting slope of the imf as derived from solar metallicity maraston ( 2005 ) models is @xmath260 ( see [ fitsum.sec ] ) . in the context of a chabrier - like imf
this slope implies a characteristic mass @xmath261 at @xmath262 . for super - solar metallicity
the best - fitting values of @xmath30 are strongly negative and may imply @xmath263 , although this represents a substantial extrapolation of the models . we conclude that the color- and luminosity evolution of early - type galaxies are consistent with imfs which have a salpeter slope at high masses but turn over near @xmath208 . such an imf is perhaps best described as `` bottom - light '' rather than top - heavy : it does not have a larger number of massive stars than a standard chabrier ( 2003 ) imf , but has a deficit of low mass stars .
the form of the imf at masses @xmath264 is not constrained by the data presented here , and it is an open question whether the slope of the imf actually becomes negative at low masses or stays constant at @xmath240 . the evolution of the characteristic mass is shown in fig.[larson.plot ] .
the results obtained in this study are represented by the formal best fit for solar metallicity , as discussed in
[ joint.sec ] .
the point labeled smgs reflects results of blain et al .
( 1999a ) for submm galaxies
blain et al .
( 1999a ) find that a standard salpeter ( 1955 ) imf has too many low mass stars , which would produce too much @xmath265-band light at @xmath194 a similar argument was made by dwek et al .
( 1998 , based on cobe data ) .
blain et al .
find that a top - heavy imf with a simple cutoff at 1 resolves this discrepancy .
no errorbar is given , but we can assume that the uncertainty in the amount of `` missing mass '' is at least 50% .
the value we adopt for @xmath266 , to match the total stellar mass of the truncated salpeter ( 1955 ) imf invoked by blain et al .
( 1999a ) .
the redshift comes from chapman et al .
( 2003 ) .
the figure also includes the imfs of the milky way disk ( from chabrier 2003 ) and of globular clusters .
the age for globular clusters reflects the results of gratton et al .
( 2003 ) , who find that globular clusters in the inner halo have ages ranging from @xmath267gyr to @xmath268gyr with a systematic uncertainty of @xmath269gyr .
the characteristic mass was determined by paresce & de marchi ( 2000 ) , who infer a typical characteristic mass @xmath270 , with formal uncertainty @xmath271 .
the true uncertainty is probably somewhat larger , as dynamical effects may play a role even at low masses ( see , e.g. , de marchi , paresce , & portegies zwart 2005 ) .
the galactic bulge is not included ; no estimates have been made for the characteristic mass of the bulge imf , but we note that there is evidence that the imf in the range 0.15 1 is flatter than a salpeter imf ( zoccali et al . 2000 ) .
the broken lines in fig . [ larson.plot ] are examples of models in which the temperature of the cmb effectively sets the characteristic mass scale of star formation at high redshift ( see larson 1998 , 2005 ; jappsen et al .
the cmb temperature exceeds the typical temperatures in prestellar cores in molecular clouds of @xmath272k ( e.g. , evans et al . 2001 ) beyond @xmath22 .
as discussed in larson ( 2005 ) the exact temperature dependence is very uncertain , and can vary from @xmath273 ( jappsen et al .
2005 , dotted line ) to @xmath274 ( larson 1985 , dashed line ) .
the normalization of these models is also very uncertain .
the value of @xmath18 is the approximate value of the jeans mass in present - day cores ( larson 2005 ) , but other aspects may play a role in setting the characteristic mass at low redshift ( see mckee & ostriker 2007 ) .
the grey solid line is a non - physical `` toy model '' which fits the milky way disk , submm - galaxies , and cluster galaxies and is intermediate between the two physically - motivated models at high redshift .
this model has the form @xmath275 and is used to explore the effects of an evolving characteristic mass in the following sections .
it is interesting that the imf of old globular clusters appears to have a lower characteristic mass than the imf of massive cluster galaxies , even though their stars probably have similar mean ages .
determining the imf in globular clusters is notoriously difficult , as the mass function as observed today is heavily influenced by dynamical effects in combination with stellar evolution ( see , e.g. , chabrier 2003 ) .
in particular , globular clusters with a very top - heavy imf would not be expected to survive in the tidal field of our galaxy for a hubble time ( e.g. , joshi , nave , & rasio 2001 ) , and the clusters that are surviving today might not be representative for the original population .
nevertheless , taking the existing constraints at face value , it seems that other parameters than the cmb temperature ( e.g. , metallicity , star formation rate , or mass of the star forming complex ) may play a role in determining @xmath6 at high redshift . as pointed out by others ( e.g. , larson 2005 ; hopkins & beacom 2006 ; tumlinson 2007 ; fardal et al .
2007 ) , an evolving imf has direct implications for the derived cosmic star formation history .
essentially all indicators of the star formation rate ( such as ultraviolet luminosity , h@xmath112 line luminosity , and infrared luminosity ) measure the effects of luminous o and b stars , which have masses @xmath276 .
inferred total star formation rates therefore rely on an extrapolation of the imf from @xmath277 down to @xmath16(see , e.g. , lilly et al .
1996 ; madau et al . 1996
; madau , pozzetti , & dickinson 1998 ) . as most of the total mass is in low mass stars ,
changes to the imf in the mass range 0.11will have virtually no effect on the observed star formation indicators but potentially large effects on the actual total stellar mass that is produced .
the effects of an evolving imf of the form of eq.[toy.eq ] are shown in fig .
[ madau.plot](a ) .
the broken line shows the integral of eq .
[ mod.eq ] , i.e. , the integrated stellar mass for an increasing value of @xmath6 with redshift .
the solid line shows the integrated mass relative to that in stars with masses @xmath278 , as the energy output of stars in this mass range drives the star formation measurements .
the effects are substantial at high redshift , and at @xmath25 the star formation rate could be overestimated by a factor of 34 relative to @xmath194 . in fig .
[ madau.plot](b ) the correction of panel ( a ) is applied to measurements of the star formation rate at a range of redshifts .
the data are a combination of the extensive compilation of hopkins ( 2004 ) and recent measurements at high redshift by bouwens et al .
extinction - corrected values were used from both studies .
the data as reported in the literature are shown in grey , and the grey curve shows a simple fit to the points .
black points are the same data , corrected to the modified chabrier ( 2003 ) imf proposed in this paper .
this correction is a combination of two effects : a constant offset to account for the difference between a salpeter ( 1955 ) imf ( which is assumed in the quoted literature ) and a chabrier ( 2003 ) imf , and the redshift - dependent effect shown in fig .
[ madau.plot](a ) . as is clear from the black line , the cosmic star formation history has a fairly well - defined peak at @xmath12 if the imf depends on redshift .
the influence of an evolving imf on the @xmath29 ratios of galaxies is complex , as several competing effects play a role .
the number of low mass stars with respect to high mass stars is reduced , which lowers the @xmath29 ratio as these stars contribute little to the integrated light .
however , for @xmath279 the number of turn - off stars is also reduced ( see fig .
[ chabrier.plot ] ) , and these stars dominate the light at rest - frame optical wavelengths . as a result ,
the net effect of an increased @xmath6 on the @xmath29 ratio is generally smaller than on the star formation rate .
a further complication is that the turn - off mass can be similar to the characteristic mass ( e.g. , in elliptical galaxies at low redshift where both values are @xmath208 ) .
this means that the effect on the @xmath29 ratio is not a constant , but depends on the age of the population .
a final complication is the mass in stellar remnants , which is a larger fraction of the total stellar mass for more top - heavy imfs . as discussed by maraston ( 2005 ) and in
[ absml.sec ] imfs that have a powerlaw slope of zero from @xmath239 imply completely remnant - dominated mass functions at old ages , with unrealistically high @xmath29 ratios .
a correct treatment of these issues requires full stellar population synthesis modeling , which is beyond the scope of the present paper .
instead , we used simple stellar evolutionary tracks to estimate what the net outcome is of the various competing mechanisms .
the yale - yonsei isochrones were used ( yi , kim , & demarque 2003 ; demarque et al . 2004 ) .
monte - carlo simulations of 100,000 stars were generated , with solar metallicity , a range of ages , and a salpeter ( 1955 ) imf .
next , the mass function was resampled to match eq . [ mod.eq ] for a grid of values of @xmath6 .
the total @xmath68 band luminosity was determined by linearly adding the light of individual stars .
the mass in living stars was determined by integrating eq .
[ mod.eq ] up to the turn - off mass .
the mass in remnants was calculated by integrating eq .
[ mod.eq ] from the turn - off mass to 100 and multiplying by the fraction of the initial mass that is retained in the form of black holes , neutron stars , and white dwarfs .
this ( age - dependent ) fraction was determined in the same way as described in bruzual & charlot ( 2003 ) .
the results are shown in fig .
[ mlmc.plot ] . at young ages
the light is dominated by relatively massive stars , and the behavior is similar to that of the implied star formation rate ( see fig . [
madau.plot]a ) . at intermediate ages and high values of @xmath6
the turn - off mass is similar to the characteristic mass , and the effect on the luminosity is similar to the effect on the mass .
the net effect is a roughly constant @xmath13 ratio as a function of @xmath6 . at the oldest ages the imf becomes
remnant - dominated , with @xmath280% of the mass in remnants for an age of 10gyr and @xmath281 .
therefore , the @xmath13 ratios are actually _ higher _ than for a standard chabrier ( 2003 ) imf wih @xmath254 , approaching or even exceeding those implied by a salpeter ( 1955 ) imf .
although these results are somewhat uncertain due to the crude nature of our modeling , the qualitative conclusion is that the @xmath13 ratios of galaxies are not necessarily strongly reduced by an evolving imf , in contrast to their star formation rates . in fig .
[ mdens.plot ] we show the effects of an evolving imf on the cosmic stellar mass density , in a similar way as was done in fig.[madau.plot ] for the cosmic star formation history .
panel ( a ) shows the redshift dependence of the @xmath13 ratio , with respect to the @xmath13 ratio implied by a standard chabrier ( 2003 ) imf . to generate this relation
it was assumed that all galaxies start forming stars at @xmath282 , and have a constant star formation rate until the epoch of observation .
the implied mean age of galaxies at @xmath283 is @xmath198gyr , and the mean age of galaxies at @xmath194 is @xmath284gyr .
these values are reasonable , and we note that the results are not very sensitive to the details of the star formation history of the galaxies .
this assumption provides an estimate of the mean mass - weighted formation time of the stars , and hence through eq .
[ toy.eq ] the appropriate value of @xmath6 .
the relations between age , @xmath6 , and @xmath13 ratio shown in fig .
[ mlmc.plot ] are then used to estimate the change in @xmath13 as a function of redshift .
going up in redshift , the @xmath13 ratio first declines due to the increase in @xmath6 from @xmath16 to @xmath203 .
the @xmath13 ratio reaches a minimum at @xmath22 , and then increases as the characteristic mass becomes similar to the turn - off mass .
beyond @xmath8 the @xmath29 ratio once again decreases , as the galaxies become younger and the turn - off mass increases rapidly .
measurements of the evolution of the mass density from cole et al .
( 2001 ) , dickinson et al .
( 2003 ) , drory et al .
( 2005 ) , rudnick et al .
( 2006 ) , and fontana et al .
( 2006 ) are shown in panel ( b ) .
errorbars are taken from the literature sources , and typically do not include systematic uncertainties .
grey points show the literature values , determined assuming a salpeter ( 1955 ) imf , and black points show the corrected values .
the correction is a constant offset ( to account for the difference between a salpeter ( 1955 ) imf and a standard chabrier ( 2003 ) imf ) in addition to the relation shown in panel ( a ) .
the grey and black lines show the evolution of the stellar mass density as implied by the observed evolution of the star formation rate shown in fig .
[ madau.plot ] .
mass loss was taken into account using the same scheme as employed by bruzual & charlot ( 2003 ) .
note that mass loss is a larger effect for chabrier - like imfs than for a salpeter ( 1955 ) imf , as a larger fraction of the total mass is in high mass stars .
the median difference between the data points and the curve is a factor of 2.3 for a non - evolving imf , and a factor of 1.7 for the evolving chabrier - like imf . considering that this type of comparison has many systematic uncertainties quite independent of the imf ( hopkins & beacom 2006 ; fardal et al .
2007 ) , the fact that the two independent measures of the build - up of stellar mass in the universe agree to within a factor of @xmath7 can be considered a success . in any case , the discrepancy is smaller for the evolving imf than for a non - evolving imf , because the star formation rate is reduced by a larger fraction than the mass density .
our evolving imf has a qualitatively similar effect as the `` paunchy '' imf with an increased contribution from stars around 1.54 proposed by fardal et al .
( 2007 ) , which was specifically designed to give better agreement between the star formation history of the universe ( as implied by the extragalactic background radiation ) and the observed evolution of the mass density .
this paper compares the color evolution of massive cluster galaxies to their luminosity evolution , with the aim of constraining the form of the imf at the time when the stars in these galaxies were formed .
it is found that the evolution of the rest - frame @xmath1 color is not consistent with the previously determined evolution of the rest - frame @xmath2 ratio , unless the imf slope is significantly flatter than the salpeter value around 1 . for standard imfs with a slope of 1.3 at @xmath285the luminosity evolution
is too fast for the measured color evolution , and the implied stellar ages derived from @xmath29 evolution and color evolution are not consistent with each other .
the only models that are able to fit the color evolution and the luminosity evolution simultaneously have imf slopes of @xmath286 around 1 and mean luminosity - weighted stellar formation redshifts of @xmath233 ( for solar metallicity ) .
this result is somewhat uncertain , as the currently available sample of cluster galaxies with accurate rest - frame @xmath1 colors and dynamical masses is somewhat limited and there are many systematic effects which may play a role . in particular , it is an open question whether stellar population synthesis models are able to predict color evolution with the required accuracy .
the commonly used bruzual & charlot ( 2003 ) and maraston ( 2005 ) models give broadly similar answers , but that may be because they share many of the same assumptions . as discussed in [ comp.sec ] the higher stellar ages implied by a flat imf are consistent with many other studies , which lends some credibility to the results presented here . of particular importance is the agreement with the data on balmer line strengths of kelson et al .
( 2001 ) , as they do not suffer from the same systematic uncertainties as the color data .
formation redshifts substantially larger than two also fit more comfortably with the direct detection of old galaxies at high redshifts .
a firm independent measurement of the star formation epoch of massive cluster galaxies , combined with a better understanding of selection effects at high redshift , would leave the imf as the only free parameter and greatly simplify the problem .
the implications discussed in [ imply.sec ] are obviously somewhat speculative .
although the interpretation in terms of an evolving characteristic mass is physically plausible according to some models ( e.g. , larson 2005 ) , many other forms of the imf are consistent with the data .
the observations described in this paper are only sensitive to a narrow mass range near 1 , and the imf proposed in eq .
[ mod.eq ] represents a very substantial extrapolation .
this is illustrated in fig .
[ summary.plot ] : both the top - heavy imf ( red line ) and the `` bottom - light '' imf ( green line ) are consistent with the data presented in this paper .
the main reason for preferring the bottom - light imf over a top - heavy form is that the absolute @xmath29 ratios of galaxies are within a reasonable range . as shown in fig .
[ mlmc.plot ] the @xmath13 ratios are similar to those implied by a standard chabrier ( 2003 ) imf , which means that they are consistent with dynamical measurements at @xmath194 ( cappellari et al .
2006 ) .
an `` unintended '' effect of an evolving imf of the form proposed here is that it reduces the discrepancy between the observed stellar mass density and the density implied by the cosmic star formation history .
this result is in excellent ( albeit qualitative ) agreement with several other recent studies ( e.g. , hopkins & beacom 2006 , fardal et al .
2007 , prez - gonzlez et al . 2007 ; wilkins , trentham , & hopkins 2007
; dav 2007 ; see also , e.g. , fields 1999 ) .
the differences between a non - evolving imf and an evolving imf are fairly large at @xmath8 ( as the effect on the star formation rate is strong and the effect on @xmath29 ratios is weak at that redshift ) , and it will be interesting to see where future measurements of the mass density at high redshift will fall in fig .
[ mdens.plot ] .
we note that the effects on the @xmath13 ratios are somewhat uncertain , as they rely on rather rudimentary stellar population synthesis modeling .
the effects on star formation rates are more robust , and suggest that the cosmic star formation rate peaked at @xmath12 .
this paper adds to previous theoretically and observationally motivated suggestions that the imf may evolve with redshift ( e.g. , worthey et al . 1992 ; larson 1998 , 2005 ; fields 1999 ; blain et al .
1999a ; baugh et al .
2005 ; stanway et al . 2005 ; hopkins & beacom 2006 ; tumlinson 2007 ; lacey et al . 2007 ;
fardal et al .
2007 ; prez - gonzlez et al . 2007 ) .
although these studies vary greatly in their parameterization of imf evolution and the range of stellar masses that are considered , they all suggest that the ratio of high - mass stars to low - mass stars was higher in the past .
it should be pointed out that most of these papers invoke a change in the imf as a `` last resort '' possibility , to explain data that are otherwise difficult to interpret . in the present study
a different approach was followed , in that we set out with the specific purpose of constraining the slope of the imf .
an advantage of the applied method is that it is fairly direct , as the rate of luminosity evolution is determined by the number of stars as a function of mass .
disadvantages are that it is only sensitive to a very limited mass range ( see fig .
[ summary.plot ] ) ; that it relies on stellar population synthesis models , which are not well calibrated in the relevant parameter range ; that the progenitors of early - type galaxies may not be representative for the general population of high redshift galaxies ; and that the currently available data are somewhat limited . accepting the possibility of an evolving characteristic mass , it is interesting to speculate what could be the cause , or causes .
the proximate cause may well be a higher temperature in molecular clouds at high redshift , which would raise the jeans mass and could inhibit the formation of low mass stars ( larson 1998 , 2005 ) . the ultimate cause could be the higher temperature of the cosmic microwave background , the fact that star formation tends to proceed in more extreme environments at higher redshift , or a combination .
available information on ir - bright galaxies suggests that dust temperatures in star burst galaxies are of order @xmath287k ( dunne et al . 2000 ; chapman et al .
2005 ) , and hence exceed the cmb temperature for all relevant redshifts . however , it is as yet unclear what fraction of the total star formation has taken place in these extreme environments ( see , e.g. , reddy et al .
2007 ) .
the analysis presented here can be improved in various ways .
the number of clusters with accurate rest - frame @xmath1 is currently smaller than the number of clusters with accurate @xmath2 measurements , and this can be remedied by obtaining accurate ( space - based ) photometry in well - chosen filters of the remaining clusters in the vv07 sample .
it is also important to measure the evolution in a redder rest - frame color , such as @xmath288 .
redder color suffer less from possible contributions of hot stars , and their evolution is probably somewhat better calibrated in stellar population synthesis models .
this requires very accurate photometry in the near - infrared , which should be possible with wfc3 on hst . on the modeling side
, it would be helpful to implement more variations of the imf in stellar population synthesis codes than the standard salpeter , kroupa , and chabrier forms
. ultimately it may be fruitful ( or prove necessary ) to have the characteristic mass , or some other parameter describing the form of the imf , as one of the `` standard '' parameters in these models , on a par with the age and metallicity .
the author acknowledges very helpful discussions with richard larson , marijn franx , charles bailyn , arjen van der wel , and bob zinn .
claudia maraston kindly provided model predictions for different slopes of the imf .
daniel mcintosh shared unpublished data on low redshift clusters .
comments of the anonymous referee significantly improved the presentation .
support from nsf career grant ast 04 - 49678 and nasa ltsa grant
nng04ge12 g is gratefully acknowledged . , t. , bryan , g. l. , & norman , m. l. 2002 , science , 295 , 93 , c. , baugh , c. m. , & lacey , c. g. 2007 , , 376 , 1711 , s. 2003 , , 409 , 37 , c. d. 1995 , , 33 , 133 , c. m. , lacey , c. g. , frenk , c. s. , granato , g. l. , silva , l. , bressan , a. , benson , a. j. , & cole , s. 2005 , , 356 , 1191 , t. c. , flynn , k. , & gebhardt , k. 1990 , , 100 , 32 , e. f. , naab , t. , mcintosh , d. h. , somerville , r. s. , caldwell , j. a. r. , barden , m. , wolf , c. , rix , h .- w . , et al .
2006 , , 640 , 241 , m. s. 1990 , , 102 , 1181 , a. w. , jameson , a. , smail , i. , longair , m. s. , kneib , j .-
p . , & ivison , r. j. 1999a , , 309 , 715 , a. w. , smail , i. , ivison , r. j. , & kneib , j .-
1999b , , 302 , 632 , j. p. , holden , b. p. , franx , m. , rosati , p. , bouwens , r. j. , demarco , r. , ford , h. c. , homeier , n. l. , et al .
2006 , , 644 , 30 , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , , in press ( arxiv:0707.2080 ) , r. g. , lucey , j. r. , & ellis , r. s. 1992a , , 254 , 601 . 1992b , , 254 , 589 , v. , coppi , p. s. , & larson , r. b. 2002 , , 564 , 23 , g. 2007 , in iau symposium no .
241 , stellar populations as building blocks of galaxies , ed .
a. vazdekis , & r. peletier , in press ( astro - ph/0703052 ) , g. & charlot , s. 2003 , , 344 , 1000 , d. , bertola , f. , buson , l. m. , faber , s. m. , & lauer , t. r. 1988 , , 328 , 440 , d. 1997 , , 113 , 162 , m. , bacon , r. , bureau , m. , damen , m. c. , davies , r. l. , de zeeuw , p. t. , emsellem , e. , falcn - barroso , j. , et al .
2006 , , 366 , 1126 , g. 2003 , , 115 , 763 , s. c. , blain , a. w. , ivison , r. j. , & smail , i. r. 2003 , , 422 , 695 , s. c. , blain , a. w. , smail , i. , & ivison , r. j. 2005 , , 622 , 772 , s. , norberg , p. , baugh , c. m. , frenk , c. s. , bland - hawthorn , j. , bridges , t. , cannon , r. , colless , m. , et al .
2001 , , 326 , 255 , g. d. , wu , c .- c . , & weedman , d. w. 1980 , , 43 , 393 , e. , renzini , a. , pirzkal , n. , cimatti , a. , malhotra , s. , stiavelli , m. , xu , c. , pasquali , a. , et al .
2005 , , 626 , 680 dav , r. 2007 , , submitted ( astro - ph/0710.0381 ) , g. , paresce , f. , & portegies zwart , s. 2005 , in astrophysics and space science library , vol .
327 , the initial mass function 50 years later , ed .
e. corbelli , f. palla , & h. zinnecker , 77 , p. , woo , j .- h .
, kim , y .- c . , &
yi , s. k. 2004 , , 155 , 667 , m. , papovich , c. , ferguson , h. c. , & budavri , t. 2003 , , 587 , 25 , s. & davis , m. 1987 , , 313 , 59 , b. , rood , r. t. , & oconnell , r. w. 1993 , , 419 , 596 , a. & gunn , j. e. 1983 , , 270 , 7 , n. , salvato , m. , gabasch , a. , bender , r. , hopp , u. , feulner , g. , & pannella , m. 2005 , , 619 , l131 , l. , eales , s. , edmunds , m. , ivison , r. , alexander , p. , & clements , d. l. 2000 , , 315 , 115 , e. , arendt , r. g. , hauser , m. g. , fixsen , d. , kelsall , t. , leisawitz , d. , pei , y. c. , wright , e. l. , et al .
1998 , , 508 , 106 , p. r. , de propris , r. , gonzalez , a. h. , stanford , s. a. , wang , m. , & dickinson , m. 2007 , , 169 , 225 , r. s. , smail , i. , dressler , a. , couch , w. j. , oemler , a. j. , butcher , h. , & sharples , r. m. 1997 , , 483 , 582 , ii , n. j. , rawlings , j. m. c. , shirley , y. l. , & mundy , l. g. 2001 , , 557 , 193 , m. a. , katz , n. , weinberg , d. h. , & dav , r. 2007 , , 379 , 985 fields , b. d. 1999 , , 515 , 603 , d. f. , kim , s. s. , morris , m. , serabyn , e. , rich , r. m. , & mclean , i. s. 1999 , , 525 , 750 , a. , salimbeni , s. , grazian , a. , giallongo , e. , pentericci , l. , nonino , m. , fontanot , f. , menci , n. , et al . 2006 , , 459 , 745 , m. 1993 , , 105 , 1058 , m. & illingworth , g. 1990 , , 359 , l41 , m. , illingworth , g. , & heckman , t. 1989 , , 98 , 538 , m. , labb ' e , i. , rudnick , g. , van dokkum , p. g. , daddi , e. , f " orster schreiber , n. m. , moorwood , a. , rix , h. , et al .
2003 , , 587 , l79 , r. , treu , t. , rhodes , j. d. , koopmans , l. v. e. , bolton , a. s. , burles , s. , massey , r. , & moustakas , l. a. 2007 , , 667 , 176 , p. , de jong , t. , hansen , l. , & norgaard - nielsen , h. u. 1994 , , 271 , 833 , r. g. , bragaglia , a. , carretta , e. , clementini , g. , desidera , s. , grundahl , f. , & lucatello , s. 2003 , , 408 , 529 harayama , y. , eisenhauer , f. , & martins , f. 2007 , , in press ( astro - ph/0710.2882 ) , b. p. , franx , m. , illingworth , g. d. , postman , m. , blakeslee , j. p. , homeier , n. , demarco , r. , ford , h. c. , et al .
2006 , , 642 , l123 , b. p. , illingworth , g. d. , franx , m. , blakeslee , j. p. , postman , m. , kelson , d. d. , van der wel , a. , demarco , r. , et al .
2007 , , in press ( arxiv:0707.2782 ) , b. p. , stanford , s. a. , eisenhardt , p. , & dickinson , m. 2004 , , 127 , 2484 , a. m. 2004 , , 615 , 209 , a. m. & beacom , j. f. 2006 , , 651 , 142 , a .-
k . , klessen , r. s. , larson , r. b. , li , y. , & mac low , m .-
2005 , , 435 , 611 , i. , chiboucas , k. , flint , k. , bergmann , m. , barr , j. , & davies , r. 2006 , , 639 , l9 jrgensen , i. , franx , m. , & kjrgaard , p. 1995a
, , 273 , 1097 . 1995b , , 276 , 1341 . 1996 , , 280 , 167 , k. j. , nave , c. p. , & rasio , f. a. 2001 , , 550 , 691 , d. d. , illingworth , g. d. , franx , m. , & van dokkum , p. g. 2001 , , 552 , l17 . 2006 , , 653 , 159 , s. s. , figer , d. f. , kudritzki , r. p. , & najarro , f. 2006 , , 653 , l113 , t. , arimoto , n. , barger , a. j. , & aragon - salamanca , a. 1998 , , 334 , 99 , t. , tanaka , i. , kajisawa , m. , kurk , j. , venemans , b. , de breuck , c. , vernet , j. , & lidman , c. 2007 , , 377 , 1717 , d. c. , datta , s. , willmer , c. n. a. , simard , l. , tran , k .- v .
, & i m , m. 2005 , , 634 , l5 , l. v. e. , treu , t. , bolton , a. s. , burles , s. , & moustakas , l. a. 2006 , , 649 , 599 , m. , van dokkum , p. , franx , m. , quadri , r. , gawiser , e. , herrera , d. , illingworth , g. d. , labb , i. , et al .
2006 , , 649 , l71 , j. 1995 , in asp conf .
77 : astronomical data analysis software and systems iv , vol . 4 , 349 , p. 2001
, , 322 , 231 . 2002 , science , 295 , 82 , i. , huang , j. , franx , m. , rudnick , g. , barmby , p. , daddi , e. , van dokkum , p. g. , fazio , g. g. , et al .
2005 , , 624 , l81 , c. g. , baugh , c. m. , frenk , c. s. , silva , l. , granato , g. l. , & bressan , a. 2007 , , submitted ( arxiv:0704.1562 ) , a. u. 1992 , , 104 , 340 , r. b. 1985 , , 214 , 379 . 1998 , , 301 , 569 . 2003 , reports of progress in physics , 66 , 1651 . 2005 , , 359 , 211 , s. j. , le fevre , o. , hammer , f. , & crampton , d. 1996 , , 460 , l1 + , p. , ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , , 283 , 1388 , p. , pozzetti , l. , & dickinson , m. 1998 , , 498 , 106 , j. & hut , p. 1997
, , 481 , 83 , g. a. & okas , e. l. 2005 , , 363 , 705 , h. , martins , f. , trippe , s. , genzel , r. , graham , j. r. , sheehy , c. , salaris , m. , gillessen , s. , et al .
2007 , , in press ( arxiv:0707.2382 ) , c. 1998 , , 300 , 872 . 2005 , , 362 , 799 , c. , daddi , e. , renzini , a. , cimatti , a. , dickinson , m. , papovich , c. , pasquali , a. , & pirzkal , n. 2006 , , 652 , 85 , m. , harris , h. c. , nemec , j. , & olszewski , e.
w. 1990 , , 100 , 469 , n. , gilbert , a. m. , & graham , j. r. 2003 , , 596 , 240 , n. , graham , j. r. , & vacca , w. d. 2005 , , 621 , 278 , d. h. , zabludoff , a. i. , rix , h .- w . , & caldwell , n. 2005 , , 619 , 193 , c. f. & ostriker , e. c. 2007 , ara&a , 45 , 565 , g. e. & scalo , j. m. 1979 , , 41 , 513 , k. , cen , r. , hernquist , l. , ostriker , j. p. , & springel , v. 2005 , , 627 , 608 , p. & nordlund , .
2002 , , 576 , 870 , f. & de marchi , g. 2000 , , 534 , 870 , r. f. 1989 , phd thesis , university of groningen , the netherlands , r. f. , davies , r. l. , illingworth , g. d. , davis , l. e. , & cawson , m. 1990 , , 100 , 1091 prez - gonzlez , p. g. , rieke , g. h. , villar , v. , barro , g. , blaylock , m. , egami , e. , gallego , j. , gil de paz , a. , et al .
2007 , , in press ( astro - ph/0709.1354 ) , m. , steidel , c. c. , adelberger , k. l. , dickinson , m. , & giavalisco , m. 2000 , , 528 , 96 , r. , van dokkum , p. , gawiser , e. , franx , m. , marchesini , d. , lira , p. , rudnick , g. , herrera , d. , et al .
2007 , , 654 , 138 , n. a. , steidel , c. c. , pettini , m. , adelberger , k. l. , shapley , a. e. , erb , d. k. , & dickinson , m. 2007 , , in press ( arxiv:0706.4091 ) , r. m. , sosin , c. , djorgovski , s. g. , piotto , g. , king , i. r. , renzini , a. , phinney , e. s. , dorman , b. , et al .
1997 , , 484 , l25 + , g. h. , loken , k. , rieke , m. j. , & tamblyn , p. 1993
, , 412 , 99 , a. j. , douglas , n. g. , arnaboldi , m. , kuijken , k. , merrifield , m. r. , napolitano , n. r. , capaccioli , m. , & freeman , k. c. 2003 , science , 301 , 1696 , g. , labb , i. , frster schreiber , n. m. , wuyts , s. , franx , m. , finlator , k. , kriek , m. , moorwood , a. , et al .
2006 , , 650 , 624 , d. & kochanek , c. s. 2005 , , 623 , 666 , e. e. 1955 , , 121 , 161 , a. r. 1953 , , 58 , 61 , j. m. 1986 , fundamentals of cosmic physics , 11 , 1 , r. p. , caldwell , n. , & rose , j. a. 2004 , , 127 , 1513 , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 , m. 1959 , , 129 , 243 , f. & seitzer , p. 1992
, , 104 , 1039 , m. , jee , m. j. , bentez , n. , blakeslee , j. p. , martel , a. r. , meurer , g. , clampin , m. , de marchi , g. , et al .
2005 , , 117 , 1049 , s. a. , eisenhardt , p. r. , & dickinson , m. 1998 , , 492 , 461 , s. a. , eisenhardt , p. r. m. , & dickinson , m. 1995 , , 450 , 512 , e. r. , mcmahon , r. g. , & bunker , a. j. 2005 , , 359 , 1184 , c. c. , adelberger , k. l. , shapley , a. e. , erb , d. k. , reddy , n. a. , & pettini , m. 2005 , , 626 , 44 , c. c. , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k. l. 1996 , , 462 , l17 , a. , brandner , w. , grebel , e. k. , lenzen , r. , & lagrange , a .- m .
2005 , , 628 , l113 , l. l. 1993 , , 105 , 1081 , m. , myers , p. c. , caselli , p. , & walmsley , c. m. 2004 , , 416 , 191 , p. , brighenti , f. , & mathews , w. g. 2007 , , 660 , 1215 , a. i. , caldwell , n. , & bower , r. g. 2001 , , 326 , 1547 , d. , maraston , c. , bender , r. , & mendes de oliveira , c. 2005 , , 621 , 673 , b. m. 1980 , fundamentals of cosmic physics , 5 , 287 , s. c. , faber , s. m. , worthey , g. , & gonz ' alez , j. j. 2000a , , 120 , 165 , s. c. , faber , s. m. , worthey , g. , & gonzlez , j. j. 2000b , , 119 , 1645 , k .- v . h. , franx , m. , illingworth , g. d. , van dokkum , p. , kelson , d. d. , blakeslee , j. p. , & postman , m. 2007 , , 661 , 750 , t. & koopmans , l. v. e. 2004 , , 611 , 739 , t. , stiavelli , m. , bertin , g. , casertano , s. , & mller , p. 2001
, , 326 , 237 , j. 2007 , , 664 , l63 , r. p. 1991
, , 253 , 710 , r. p. & van dokkum , p. g. 2007 , , 668 , 756 , a. , franx , m. , van dokkum , p. g. , huang , j. , rix , h .- w . , & illingworth , g. d. 2006a , , 636 , l21 , a. , franx , m. , van dokkum , p. g. , rix , h .- w . , illingworth , g. d. , & rosati , p. 2005
, , 631 , 145 , a. , franx , m. , wuyts , s. , van dokkum , p. g. , huang , j. , rix , h .- w . , & illingworth , g. d. 2006b , , 652 , 97 , p. g. 2005 , , 130 , 2647 , p.
g. & franx , m. 1996 , , 281 , 985 . 2001 , , 553 , 90 , p. g. , franx , m. , fabricant , d. , illingworth , g. d. , & kelson , d. d. 2000 , , 541 , 95 , p. g. , franx , m. , kelson , d. d. , & illingworth , g. d. 1998a , , 504 , l17 , p. g. , franx , m. , kelson , d. d. , illingworth , g. d. , fisher , d. , & fabricant , d. 1998b , , 500 , 714 , p. g. & van der marel , r. p. 2007
, , 655 , 30 wilkins , s. m. , trentham , n. , & hopkins , a. m. 2007 , , submitted , g. 1994 , , 95 , 107 , g. , faber , s. m. , & gonzalez , j. j. 1992 , , 398 , 69 , y. & deng , l. 2005 , , 619 , 824 , s. k. , kim , y .- c . , & demarque , p. 2003
, , 144 , 259 , s. k. , yoon , s .- j . , kaviraj , s. , deharveng , j
.- m . , rich , r. m. , salim , s. , boselli , a. , et al .
2005 , , 619 , l111 , r. j. , newell , e. b. , & gibson , j. b. 1972 , , 18 , 390 zoccali , m. , cassisi , s. , frogel , j. a. , gould , a. , ortolani , s. , renzini , a. , rich , r. m. , & stephens , a. w. 2000 , , 530 , 418
for five of the seven distant clusters discussed in this paper colors were measured directly from hst images in apertures of fixed physical size . for each cluster
this size was chosen to match an @xmath61 diameter aperture at the distance of the coma cluster , allowing a direct comparison to data from ble92 and to mcintosh et al .
for the clusters ms105403 ( @xmath101 ) and rxj015213 ( @xmath102 ) color measurements were taken from blakeslee et al .
( 2006 ) , who used apertures scaled to contain 50% of each galaxy s light . in this section
we give the corrections that were applied to bring the blakeslee et al .
( 2006 ) data onto our system , and test these corrections using data from mcintosh et al .
( 2005 ) .
the reason why aperture corrections need to be applied even when data are corrected for seeing effects is that early - type galaxies have color gradients .
these gradients do not vary much from galaxy to galaxy , when expressed as @xmath289 , with @xmath110 and @xmath111 the relevant filters and @xmath290 the logarithm of the circularized radial distance to the center of the galaxy ( see , e.g. , franx , illingworth , & heckman 1989 ; franx & illingworth 1990 ; peletier et al .
the measured color within an aperture @xmath291 is not simply the integral of the color gradient , as the radial dependence of the luminosity has to be taken into account as well .
we determine the expected dependence of color on aperture size numerically , assuming an @xmath292 law for the radial luminosity dependence . the solid line in fig .
[ colgrad.plot ] shows the expected relation for a color gradient of @xmath293 , appropriate for @xmath1 ( e.g. , peletier et al .
the long - dashed line is a polynomial fit to the numerically - derived relation , of the form @xmath294 with @xmath295 the color measured through an aperture of diameter @xmath296 , @xmath297 the color measured through an aperture containing 50% of the galaxy s luminosity , @xmath58 the half - light radius , and @xmath298 the color gradient in mag per dex .
the long - dashed line is a very good match to the ( exact ) solid line . for reference
, short - dashed lines show the behavior of eq .
[ colgrad.eq ] for color gradients of @xmath299 and @xmath300 .
the data points show binned data from mcintosh et al .
( 2005 ) , who measured @xmath1 colors of three nearby clusters in apertures of fixed size and in apertures of diameter @xmath301 .
the data follow the predicted relation very closely , with an rms of @xmath103 and no apparent systematic deviations .
the datapoints from blakeslee et al .
( 2006 ) were corrected to our fiducial aperture size by calculating @xmath302 for each galaxy ( using the information supplied in blakeslee et al . ) and applying eq.[colgrad.eq ] . as on average @xmath303 for our aperture size the correction is only @xmath123 in the mean for both clusters , with an object - to - object scatter of @xmath304 .
galaxies requiring a correction @xmath305 or @xmath306 were removed from the sample ; this step slightly reduces the scatter in the color - mass relations of the clusters but otherwise has a negligible effect on our results . | theoretical arguments and indirect observational evidence suggest that the stellar initial mass function ( imf ) may evolve with time , such that it is more weighted toward high mass stars at higher redshift . here
we test this idea by comparing the rate of luminosity evolution of massive early - type galaxies in clusters at @xmath0 to the rate of their color evolution
. a combined fit to the rest - frame @xmath1 color evolution and the previously measured evolution of the @xmath2 ratio gives @xmath3 for the logarithmic slope of the imf in the region around 1 , significantly flatter than the present - day value in the milky way disk of @xmath4 .
the best - fitting luminosity - weighted formation redshift of the stars in massive cluster galaxies is @xmath5 , and a possible interpretation is that the characteristic mass @xmath6 had a value of @xmath7 at @xmath8 ( compared to @xmath9today ) , in qualitative agreement with models in which the characteristic mass is a function of the jeans mass in molecular clouds .
such a `` bottom - light '' imf for massive cluster galaxies has significant implications for the interpretation of measurements of galaxy formation and evolution . applying a simple form of imf evolution to literature data
, we find that the volume - averaged star formation rate at high redshift may have been overestimated ( by a factor of @xmath10 at @xmath11 ) , and the cosmic star formation history may have a fairly well - defined peak at @xmath12 . the @xmath13 ratios of galaxies are less affected than their star formation rates , and future data on the stellar mass density at @xmath14 will provide further constraints on imf evolution .
the formal errors likely underestimate the uncertainties , and confirmation of these results requires a larger sample of clusters and the inclusion of redder rest - frame colors in the analysis . |
the infinite - ranged spherical spin glass with @xmath5 body random exchange interactions has attracted significant attention recently for a number of reasons . among these are ( i ) it is exactly soluble in the thermodynamic limit both for its equilibrium properties and for its off - equilibrium macrodynamics ( at least in the sense of coupled equations for macroscopic order functions and _ anstze _ solving them in a long time limit ) , ( ii ) it exhibits replica symmetry breaking ( rsb ) and aging glassy macrostates , ( iii ) the rsb is of the one - step kind ( 1rsb ) even down to the lowest temperatures .
( in its discontinuous variant , 1rsb is believed to be symptomatic of many systems exhibiting glassy behaviour without hamiltonian disorder . )
the original work @xcite and most subsequent advances have concentrated on situations where the exchange distribution is symmetric , albeit possibly with an external field .
more recently @xcite an extension was introduced to allow for an additional ferromagnetic interaction , stimulated by the existence of many physical systems with large coherently coordinated attractors , such as most real spin glass materials with appropriate concentrations @xcite , neural networks @xcite , proteins @xcite , and error - correcting codes@xcite .
this extension was however limited to two - body ferromagnetic exchange , with corresponding second - order ferromagnetic transitions . in this paper
we extend the study further to include @xmath2 body ferromagnetic interactions for which the onset of ferromagnetism is first order .
this is of relevance both for application to real systems ( for example , for error - correcting codes @xmath6 ) and since it brings in new features ( spinodal and thermodynamic transitions , metastability , and suppression of glassy ferromagnetism ) .
the hamiltonian we use is @xmath7 where the @xmath8 are independent quenched random couplings given by a gaussian distribution with zero mean and variance .
the spins @xmath9 are real numbers subject to the spherical constraint @xmath10 we consider so that replica symmetry is broken for low temperatures , spherical spins so that 1rsb is sufficient at all temperatures , and infinite - ranged interactions so that mean field theory is exact . for @xmath11
this reduces to the model of @xcite , for @xmath12 it becomes that of @xcite , but we shall be interested also in @xmath2 . in considering the phase diagram
, we will identify lines of four types . since there are first order transitions , in the statics we must consider the spinodal lines ( where a phase appears ) and the thermodynamic lines ( where it becomes thermodynamically preferred due to its free energy ) . a modified replica analysis employing
the criterion of marginal stability ( ms ) leads to a different set of spinodal lines . finally , there are spinodal lines generated by a study of the langevin dynamics at long times . as in the cases
studied previously , with an external field @xcite or a coupling @xcite , the dynamic lines are identical to the ms lines .
the equilibrium properties of this model are given by the replica method @xcite in which one - step replica symmetry breaking is sufficient .
this generates four order parameters governed by self - consistent equations .
three of these describe the probability distribution @xmath13 of the overlap between pure states : @xmath13 has two @xmath14-function spikes , one corresponding to the self - overlap @xmath15_\mathrm{av } \intertext{and one corresponding to the mutual overlap } q_0 = q^{ss ' } & = \frac{1}{n } \left [ \sum_i \langle \phi_i \rangle^s \langle \phi_i \rangle^{s ' } \right]_\mathrm{av } , \qquad\qquad s \neq s'\,,\end{aligned}\ ] ] where @xmath16 refers to the thermodynamic average over the microstates of the pure state @xmath17 and @xmath18_\mathrm{av}$ ] to the average over the quenched disorder ; the strength of these two are and @xmath19 respectively .
the fourth order parameter is the magnetization @xmath20 . the replica free energy @xmath21 is , in the thermodynamic limit ( @xmath22 ) , given by @xmath23\ , , \label{repfe}\end{gathered}\ ] ] where the replica indices @xmath24 and @xmath25 run from @xmath26 to @xmath27 , and @xmath28 represents the outer product of the vectors . under the 1rsb
@xcite this becomes @xmath29 where we have defined an average over @xmath13 : for any value of @xmath30 , @xmath31 stationarity of @xmath21 leads to the self - consistent equations .
extremising with respect to @xmath20 gives @xmath32 and with respect to @xmath19 gives @xmath33 for the statics , equations ( [ eqnm ] ) , ( [ eqn1q ] ) , ( [ eqn2q ] ) and ( [ eqnx ] ) are solved for @xmath20 , @xmath34 , @xmath35 , and @xmath19 .
we also find it useful to consider the free energy with the magnetization constrained ; in this case ( [ eqnm ] ) does not apply . for the calculation under marginal stability ,
we abandon ( [ eqnx ] ) , and instead insist that the lowest eigenvalue of the hessian matrix of ( [ repfe ] ) in the @xmath36 should vanish , which is to say that the system should be marginally stable against small fluctuations in the overlaps .
the resulting equation is @xmath37
the dynamics used are given by the langevin equation @xmath38 and @xmath39 is introduced to enforce the spherical constraint ( [ spher ] ) .
the standard generating function procedure @xcite yields a self - consistent mean - field equation of motion : @xmath40 this equation can not be solved exactly , but it is possible to obtain self - consistent equations using the standard aging assumption@xcite @xmath41 taking the limit of long times and setting all the time derivatives to zero .
the calculation follows @xcite and , as expected , gives the equations ( [ eqnm ] ) , ( [ eqn1q ] ) , ( [ eqn2q ] ) , and ( [ ms ] ) obtained in the ms version of the replica analysis , identifying @xmath42 and @xmath43 .
the rle played by @xmath19 is that of a factor in a modified fluctuation - dissipation theorem ( qfdt ) in the non - ergodic phase : @xmath44 there is a direct correspondence between the breaking of replica symmetry and the breaking of ergodicity .
phase diagrams for a characteristic set of situations are exhibited in figures [ r1][r5 ] ; the new results are in figures [ r3][r5 ] but figures [ r1 ] and [ r2 ] are included for orientation ( as well as completeness ) . in each case @xmath45 is chosen , but similar results apply for other @xmath46 .
the remaining figures show the free energies of systems with constrained magnetizations : these assist in the interpretation of the phase diagrams which result when the constraint is removed and the free energy minimized with respect to @xmath20 .
figure [ r1 ] is for @xmath11 @xcite , in which case the second term of ( [ ham ] ) corresponds to an applied field @xmath47 .
it is helpful to recall its features .
for @xmath48 the only stable state is paramagnetic .
for @xmath49 between @xmath50 and @xmath51 there is a continuous one - step replica symmetry breaking ( c1rsb ) transition from paramagnet to spin glass , with @xmath52 at the transition ; the transition temperature is the same statically and dynamically , and the transition coincides with the onset of almeida thouless instability . at @xmath53 ,
this temperature reaches a maximum of @xmath54 . for @xmath55
there is a discontinuous one - step replica symmetry breaking ( d1rsb ) transition from paramagnet to spin glass , with @xmath56 at the transition .
the transition temperature is higher for dynamics ( or equivalently marginal stability ) than for statics ; both transition temperatures are higher than that at which small fluctuation almeida thouless instability would onset were it not preempted by the discontinuous transition .
all three temperatures , which we shall label @xmath57 , @xmath58 , and @xmath59 respectively , fall as @xmath49 falls ; for future use we define @xmath60 at @xmath61 . in considering the various systems with ferromagnetic interactions discussed below it will be helpful to make reference to this case . before passing to the generalized models
it is also useful to consider the system with @xmath62 ( for which case the value of @xmath1 is irrelevant ) but with constrained magnetization . for @xmath63 the free energy as a function of @xmath20
has the form shown in figure [ mj0](a ) : the stable state is replica symmetric for all @xmath20 . as @xmath64 is reduced below @xmath54 the character of @xmath65 changes as shown in figure [ mj0](b ) , or in more detail in figure [ mj0](c ) : a gap opens up in which there is no longer an rs solution stable against almeida
thouless fluctuations , and a region in which there is a new rsb solution appears .
the rsb solution spans the gap in the rs solution , with an overlap at its lower end .
that is , the upper end - point of the rsb curve coincides with the lower end - point of the high-@xmath20 section of the rs curve , but the lower end - point of the rsb curve lies on the low-@xmath20 section of the rs curve below its end - point .
the rsb solution has monotonically increasing @xmath66 as @xmath20 is lowered below the upper connection point of rs and rsb .
the coincidence at the upper end of the gap is related to the possibility of a continuous rsb , while the overlap at the lower end is related to a discontinuous rsb . as the temperature
is lowered further the gap in the rs curve and range of rsb both grow , with the latter extending to @xmath67 at @xmath68 , as shown in figure [ mj0](d ) . for @xmath62
the minimum of @xmath65 is always at @xmath67 which is therefore the unconstrained magnetization . for @xmath69 ,
where both rs and rsb solutions exist at @xmath67 , the latter is favoured .
increasing @xmath47 for @xmath11 modifies the curves @xmath65 and moves the minimum to a finite magnetization @xmath70 . for @xmath63
, @xmath65 remains only rs and the unconstrained state remains paramagnetic .
for @xmath71 , where there is an rsb curve which does not extend to @xmath67 , the sequence of events on increasing @xmath49 is as follows : ( i ) the minimum moves out along the lower section of the rs curve , ( ii ) it crosses into the region where the rs and rsb curves overlap and both have minima , the rsb being favoured so that a d1rsb transition to a spin glass takes place ; ( iii ) the rsb minimum continues to move out , while the rs minimum reaches the end of the lower rs branch and disappears , corresponding to crossing the almeida
thouless line , but the rs minimum is irrelevant and no phase change occurs ; ( iv ) the rsb minimum moves up the rsb curve until this gives way smoothly to the upper section of the rs curve and a c1rsb transition back to a paramagnet takes place . for @xmath72 again the rsb curve does extend to @xmath67 and the system already favours the spin glass solution at @xmath61 , so only ( iii ) and ( iv ) occur . for @xmath73 ferromagnetism becomes possible with effective field @xmath74 determined self - consistently . figure [ r2 ] shows the phase diagram obtained recently @xcite for @xmath12 . at low @xmath75 the frustration due to
the disorder in @xmath76 continues to prevent ferromagnetism , as does entropy as the temperature is raised , leading to behaviour similar to that for @xmath62 . at a temperature - dependent @xmath75
a continuous ferromagnetic transition takes place and the system goes over to a solution whose magnetization rises continuously with @xmath75 .
the ferromagnetic region is split into two parts , non - glassy rs ( at higher @xmath75 , @xmath64 ) and glassy 1rsb ( at lower @xmath75 , @xmath64 ) .
the transition between them directly mirrors the behaviour of figure [ r1 ] with @xmath49 replaced by the self - consistently determined @xmath77 and is d1rsb ( c1rsb ) for @xmath75 less ( greater ) than the value @xmath78 for which @xmath79 .
( the transition temperature reaches the maximum of @xmath54 at @xmath80 . )
again there is an almeida thouless curve which lies beneath the d1rsb transition line for @xmath81 but is coincident with the c1rsb transition line for @xmath82 where @xmath83 is the value of @xmath75 at which @xmath84 .
the transitions are apparent in @xmath65 through behaviour directly analogous to that discussed above for @xmath11 . in the case of @xmath12
the ferromagnetic onsets are second order , with the minimum in @xmath65 moving away from @xmath67 continuously as @xmath75 is increased across the transition lines .
figure [ mr2 ] illustrates several aspects of the phase diagram : [ mr2](a ) shows a region of non - glassy ferromagnetism above @xmath54 , where @xmath65 is rs throughout its range ; [ mr2](b ) shows a non - glassy ferromagnet between @xmath54 and @xmath85 for @xmath81 , with a gap in the rs curve and an rsb section but with the minimum in the lower rs region ; [ mr2](c ) shows a glassy ferromagnet at the same temperature , where the minimum is now in the rsb region , the system having undergone a d1rsb transition ; [ mr2](d ) shows a non - glassy ferromagnet at the same temperature for @xmath86 , with the minimum now in the upper rs region , the system having undergone a c1rsb transition .
for @xmath2 the ferromagnetic transitions are first order , with the @xmath67 solution always locally stable . as noted before , there are two kinds of transition as @xmath75 is increased or @xmath64 is decreased : a spinodal transition at which a secondary minimum appears in @xmath65 at a finite @xmath20 whilst the lowest minimum is at @xmath67 , and a thermodynamic transition at which the finite @xmath20 minimum becomes lower than that at @xmath67
. figures [ r3][r5 ] show the full phase diagrams for @xmath45 and @xmath87 respectively , as characteristic illustrations of systems with first order transitions for which @xmath1 is less than , equal to , and greater than @xmath0 . only for the case @xmath3 is a glassy ferromagnet found with glassy / non - glassy transitions ; this transition is analogous to that for the case @xmath12 , with d1rsb for @xmath75 less than @xmath78 and c1rsb for @xmath86 .
figure [ mr3 ] illustrates the underlying character of @xmath65 , which drives the static transitions .
figures [ mr3](a , b ) show the situation for @xmath63 , ( a ) in a region of spinodal ferromagnetism , ( b ) of thermodynamic ferromagnetism . figures [ mr3](c f ) are for @xmath69 , showing the effect of increasing @xmath75 : ( c ) is in the spin glass phase , passing to an rsb ferromagnet ( d ) via a spinodal transition , with the rsb ferromagnet thermodynamically preferable to the spin glass in ( e ) , and eventually crossing via a continuous transition into an rs ferromagnet in ( f ) .
the smallest self - consistent value of @xmath77 ( i.e. that at the spinodal transition ) increases with @xmath1 , and in fact for @xmath6 at any @xmath72 it is exactly that corresponding to the c1rsb line in figure [ r1 ] so that the transition is directly into an rs ferromagnet .
this is shown in figure [ mr45](a ) .
( note that since the c1rsb line coincides with the almeida
thouless line , the rs ferromagnet does become unstable against rsb fluctuations at the transition . ) for @xmath88 the smallest @xmath77 is beyond this critical value and carries the system well into the rs region , so there is no trace of an rsb ferromagnet .
this is shown in figure [ mr45](b ) .
hence for all @xmath4 glassy ferromagnetism is suppressed . in the case of @xmath6
the minimum @xmath77 corresponds to the same value of @xmath75 for any @xmath72 , yielding a vertical transition line between spin - glass and rs ferromagnet . for @xmath88 ( as in figure [ r5 ] )
the ferromagnet becomes marginally stable against small fluctuations in @xmath20 along its spinodal transition line @xmath89 .
the corresponding thermodynamic transition line is in two segments : for @xmath90 it is given by @xmath91 ; whilst for @xmath69 it is where @xmath92 . for @xmath6 ( as in figure [ r4 ] )
the spinodal transition is also in two segments : for both segments it has @xmath89 as before , but in the lower section ( @xmath72 ) it simultaneously becomes unstable against almeida
thouless fluctuations .
the thermodynamic transition is as for @xmath88 . for @xmath3 ( as in figure [ r3 ] ) the spinodal transition line to glassy ferromagnetism
is in three segments ( see figure [ r3spin ] ) : ( a ) the c1rsb line where @xmath93 and @xmath94 ; ( b ) the d1rsb line where @xmath95 and @xmath96 ; ( c ) the line @xmath97 , where it is marginally stable against small fluctuations in @xmath20 again .
non - glassy ferromagnetism onsets at @xmath89 , as shown by curve ( d ) which terminates where it intersects ( b ) .
the static and dynamic results are qualitatively the same , although the lines ( b ) and ( c ) are slightly displaced .
the thermodynamic transition lines are as the spinodal lines for the continuous transitions , but for the first order ferromagnetic transitions are moved to higher @xmath75 .
this curve is in three segments : ( c ) has moved to the line @xmath98 ; the lower section of ( b ) to @xmath99 ; ( d ) has moved to @xmath100 .
in this paper we have solved the infinite range spherical @xmath0-spin glass with an additional @xmath1-spin ferromagnetic interaction , finding the phase diagrams both in statics and dynamics for @xmath5 and general @xmath1 . by further examination of the free energy with constrained magnetization we have clarified the origin of the different phases , both for the previously studied models with @xmath101 and for @xmath2 .
we have related the behaviour of systems with @xmath73 to those with @xmath11 . in all cases , in the replica method ,
the first step of replica symmetry breaking is sufficient . as previously noted @xcite , for @xmath12 there are thermodynamically continuous transitions to ferromagnetism with two types of ferromagnetic region , non - glassy and glassy ( figure [ r2 ] ) .
for @xmath2 the ferromagnet transitions are first order . for @xmath3
there remains a region of parameter space where the system favours a glassy ferromagnet ( figure [ r3 ] ) , whereas for @xmath4 glassy ferromagnetism is suppressed and the ferromagnetic region consists of a single non - glassy phase ( figures [ r4 ] , [ r5 ] ) .
the authors would like to thank andrea cavagna , irene giardina , juan - pedro garrahan and john hertz for helpful discussions .
we would also like to thank eprsc ( uk ) for financial support , pg for research studentship 97304251 and ds for research grant gr / m04426 . | we consider an infinite - range spherical @xmath0-spin glass model with an additional @xmath1-spin ferromagnetic interaction , both statically using a replica analysis and dynamically via a generating functional method . for @xmath2
we find that there are first order transitions to ferromagnetic phases . for @xmath3
there are two ferromagnetic phases , one non - glassy replica symmetric and one exhibiting glassy one - step replica symmetry breaking and aging , whereas for @xmath4 only the replica symmetric phase exists . |
noncommutative differential geometry is a growing subject centering around the exploration of a new kind of geometric spaces which do not belong to the classical geometric world .
the theory is resting on large classes of examples as well as on the elaboration of new general concepts .
the main sources of examples so far have been provided by : * a general principle allowing to understand difficult quotients of classical spaces , typically spaces of leaves of foliations , as noncommutative spaces . *
deformation theory which provides rich sources of examples in particular in the context of quantization " problems .
* spaces of direct relevance in physics such as the brillouin zone in the quantum hall effect , or even space - time as in the context of the standard model with its higgs sector .
we recently came across a whole class of new noncommutative spaces defined as solutions of a basic equation of @xmath8-theoretic origin .
this equation was at first expected to admit only commutative ( or nearly commutative ) solutions . whereas classical spheres provide simple solutions of arbitrary dimension @xmath9 , it turned out that when the dimension @xmath9 is @xmath10 there are very interesting new , and highly noncommutative , solutions
. the first examples were given in @xcite , and in @xcite ( hereafter refered to as part i ) we began the classification of all solutions in the @xmath4-dimensional case , by giving an exhaustive list of noncommutative 3-spheres @xmath11 , and analysing the critical " cases .
we also explained in @xcite a basic relation , for generic values of our modulus " @xmath12 , between the algebra of coordinates on the noncommutative @xmath13-space of which @xmath11 is the unit sphere , and the sklyanin algebras which were introduced in the context of totally integrable systems .
+ in this work we analyse the structure of noncommutative @xmath4-spheres @xmath11 and of their moduli space .
we started from the above relation with the sklyanin algebras and first computed basic cyclic cohomology invariants using @xmath14-functions .
the invariant to be computed was depending on an elliptic curve ( with modulus @xmath15 ) and several points on the curve .
it appeared as a sum of @xmath16 terms , each an integral over a period of a rational fraction of high degree ( @xmath17 ) in @xmath14-functions and their derivatives .
after computing the first terms in the @xmath18-expansion of the sum , ( with the help of a computer ) , and factoring out basic elliptic functions of the above parameters , we were left with a scalar function of @xmath18 , starting as , + @xmath19 in which one recognises the @xmath20 power of the dedekind @xmath21-function .
+ we then gradually simplified the result ( with @xmath22 appearing as an integration factor from the derivative of the weierstrass @xmath23-function ) and elaborated the concepts which directly explain the final form of the result .
+ the main new conceptual tool which we obtained and that we develop in this paper is the notion of central quadratic form for quadratic algebras ( definition [ cent ] ) .
the geometric data @xmath24 of a quadratic algebra @xmath25 is a standard notion ( @xcite , @xcite , @xcite ) defined in such a way that the algebra maps homomorphically to a cross - product algebra obtained from sections of powers of the line bundle @xmath26 on powers of the correspondence @xmath27 . we shall prove a purely algebraic result ( lemma [ alg0 ] ) which considerably refines the above homomorphism and lands in a richer cross - product . its origin can be traced as explained above to the work of sklyanin ( @xcite ) and odesskii - feigin ( @xcite ) . + our construction is then refined to control the @xmath0-norm ( theorem [ c * ] ) and to construct the differential calculus in the desired generality ( section 8) .
it also allows to show the pertinence of the general spectral " framework for noncommutative geometry , in spite of the rather esoteric nature of the above examples .
+ we apply the general theory of central quadratic forms to show that the noncommutative 3-spheres @xmath11 admit a natural ramified covering @xmath3 by a noncommutative @xmath4-manifold @xmath28 which is ( in the even case , cf .
proposition [ siginvar ] ) isomorphic to the mapping torus of an outer automorphism of the noncommutative @xmath29-torus @xmath30 .
it is a noncommutative version of a nilmanifold ( corollary [ h3 ] ) with a natural action of the heisenberg lie algebra @xmath31 and an invariant trace .
this covering yields the differential calculus in its transcendental " form .
another important novel concept which plays a basic role in the present paper is the notion of jacobian for a morphism of noncommutative spaces , developed from basic ideas of noncommutative differential geometry @xcite and expressed in terms of hochschild homology ( section 7 ) .
we compute the jacobian of the above ramified covering @xmath3 by pairing the direct image of the fundamental class of the noncommutative @xmath4-dimensional nilmanifold @xmath28 with the chern character of the defining unitary and obtain the answer as the product of a period ( of an elliptic integral ) by a rational function ( theorem [ vol ] and corollary [ nct0 ] ) . as explained above , we first computed these expressions in terms of elliptic functions and modular forms which led us in order to simplify the results to extend the moduli space from the real to the complex domain , and to formulate everything in geometric terms .
the leaves of the scaling foliation of the real moduli space then appear as the real parts of a net of degree @xmath13 elliptic curves in @xmath32 having @xmath2 points in common .
these elliptic curves turn out to play a fundamental role and to be closely related to the elliptic curves of the geometric data ( cf .
section @xmath13 ) of the quadratic algebras which their elements label .
in fact we first directly compared the @xmath33-invariant of the generic fibers ( of the scaling foliation ) with the @xmath33-invariant of the quadratic algebras , and found them to be equal .
this equality is surprising in that it fails in the degenerate ( non - generic ) cases , where the characteristic variety can be as large as @xmath32 .
moreover , even in the generic case , the two notions of real " points , either in the fiber or in the characteristic variety are not the same , but dual to each other .
we eventually explain in theorem [ iden ] the generic coincidence between the leaves of the scaling foliation in the complexified moduli space and the characteristic varieties of the associated quadratic algebras .
this theorem [ iden ] also exhibits the relation of our theory with iterations of a specific birational automorphism of @xmath32 , defined over @xmath34 , and restricting on each fiber as a translation of this elliptic curve .
the generic irrationality of this translation and the nature of its diophantine approximation play an important role in sections 7 and 8 .
+ another important result is the appearance of the hyperfinite factor of type ii@xmath5 as cross - product of the field @xmath6 of meromorphic functions on an elliptic curve by a subgroup of the galois group aut@xmath7 , ( theorem [ gal ] ) and the description of the differential calculus in general ( lemma [ cyclic ] ) and in rational " form on @xmath11 ( theorem [ fund ] ) .
the detailed proofs together with the analysis of the spectral geometry of @xmath11 and of the @xmath0-algebra @xmath35 will appear in part ii .
let us now be more specific and describe the basic @xmath8-theoretic equation defining our spheres . in the simplest case
it asserts that the algebra @xmath25 of coordinates " on the noncommutative space is generated by a self - adjoint idempotent @xmath36 , ( @xmath37 , @xmath38 ) together with the algebra @xmath39 of two by two scalar matrices .
the only relation is that the trace of @xmath36 vanishes , i.e. that the projection of @xmath36 on the commutant of @xmath39 is zero .
one shows that @xmath25 is then the algebra @xmath40 where @xmath41 is the algebra of coordinates on the standard @xmath29-sphere @xmath42 .
+ the general form of the equation distinguishes two cases according to the parity of the dimension @xmath9 . in the even case the algebra @xmath25 of coordinates " on the noncommutative space is still generated by a projection @xmath36 , ( @xmath37 , @xmath38 ) and an algebra of scalar matrices , but the dimension @xmath43 appears in requiring the vanishing not only of the trace of @xmath36 , but of all components of its chern character of degree @xmath44 .
this of course involves the algebraic ( cyclic homology ) formulation of the chern character . the chern character in cyclic homology @xcite , @xcite
, @xmath45 is the noncommutative geometric analogue of the classical chern character .
we describe it in the odd case which is relevant in our case @xmath46 . given a noncommutative algebra @xmath25 , an invertible element @xmath47 in @xmath48 defines a class in @xmath49 and the components of its chern character are given by , @xmath50 where @xmath51 and summation over repeated indices is understood . by a noncommutative @xmath52-dimensional spherical manifold ( n odd ) , we mean the noncommutative space @xmath53 dual to the @xmath54-algebra @xmath25 generated by the components @xmath55 of a solution @xmath56 , @xmath57 , of the equation @xmath58 which is the noncommutative counterpart of the vanishing of the lower chern classes of the bott generator of the @xmath8-theory of classical odd spheres .
+ the moduli space of 3-dimensional spherical manifolds appears naturally as a quotient of the space of symmetric unitary matrices @xmath59 indeed for any @xmath60 let @xmath61 be the unitary solution of ( [ sph ] ) given by @xmath62 where @xmath63 are the usual pauli matrices and the presentation of the involutive algebra @xmath64 generated by the @xmath65 is given by the relations @xmath66 we let @xmath67 be the associated quadratic algebra , generated by the @xmath65 with presentation , @xmath68 where @xmath69 is the unit of @xmath39 .
the element @xmath70 is in the center of @xmath71 ( cf .
part i ) and the additional inhomogeneous relation defining @xmath64 is @xmath72
. by part i theorem 1 , any unitary solution of ( [ sph ] ) for @xmath73 is a homomorphic image of @xmath61 for some @xmath74 .
the transformations @xmath75
@xmath76 @xmath77 act on the space of unitary solutions of ( [ sph ] ) and preserve the isomorphism class of the algebra @xmath64 and of the associated quadratic algebra @xmath71 .
transformation ( [ eq1.2 ] ) corresponds to @xmath78 transformation ( [ eq1.3 ] ) induces @xmath79 with @xmath80 which in turn induces @xmath81 finally ( [ eq1.4 ] ) reverses the orientation " of @xmath82 and corresponds to @xmath83 we define the _ real moduli space _ @xmath84 as the quotient of @xmath85 by the transformations ( [ scal ] ) and ( [ rot ] ) , and the _ unoriented _ real moduli space @xmath86 as its quotient by ( [ inv ] ) . by construction
the space @xmath87 is the homogeneous space @xmath88 , with @xmath89 acting on @xmath87 by @xmath90 for @xmath91 and @xmath92 . the conceptual description of the moduli spaces @xmath84 and @xmath86 requires taking care of finer details .
we let @xmath14 be the involution of the compact lie group @xmath93 given by complex conjugation , and define the closed subgroup @xmath94 as the normaliser of @xmath95 , i.e. by the condition @xmath96 where @xmath97 is the center of @xmath93 .
the quotient @xmath98 is a riemannian globally symmetric space ( cf .
@xcite theorem 9.1 , chapter vii ) .
one has @xmath99 but the image of @xmath8 in @xmath100 is disconnected .
indeed , besides @xmath101 the subgroup @xmath8 contains the diagonal matrices with @xmath102 as diagonal elements , where @xmath103 is an @xmath2th root of @xmath104 .
[ sym1 ] the real moduli space @xmath84 ( resp .
@xmath86 ) is canonically isomorphic to the space of congruence classes of point pairs under the action of @xmath93 ( resp . of isometries ) in the riemannian globally symmetric space @xmath105 .
this gives two equivalent descriptions of @xmath84 as the orbifold quotient of a @xmath4-torus by the action of the weyl group of the symmetric pair .
in the first @xmath106 we identify the lie algebra @xmath107 with the lie algebra of traceless antihermitian elements of @xmath108 .
we let @xmath109 be the lie subalgebra of diagonal matrices , it is a maximal abelian subspace of @xmath110 .
the weyl group @xmath111 of the symmetric pair @xmath112 is isomorphic to the permutation group @xmath113 acting on @xmath109 by permutation of the matrix elements .
the @xmath4-torus @xmath114 is the quotient , @xmath115 and the isomorphism of @xmath116 with @xmath84 is obtained from , @xmath117 the lattice @xmath118 is best expressed as @xmath119 in terms of the roots @xmath120 of the pair @xmath112 where the root system @xmath121 is the same as for the cartan subalgebra @xmath122 of @xmath123 .
the roots @xmath124 , @xmath125 , @xmath126 are given by @xmath127 where @xmath128 for @xmath129 are the elements of the diagonal matrix @xmath130 . in terms of the primitive roots
@xmath131 the coordinates @xmath132 used in part
i are given by @xmath133 and the lattice @xmath118 corresponds to @xmath134 , @xmath135 .
with @xmath136 we let @xmath137 be the inverse isomorphism .
one has , modulo projective equivalence , @xmath138\ ] ] in these coordinates ( @xmath132 ) the symmetry given by the weyl group @xmath139 of the symmetric pair @xmath112 now reads as follows , @xmath140 where @xmath141 is the transposition of @xmath125 , @xmath142 . in part i we used the parametrization by @xmath143 to label the 3-dimensional spherical manifolds @xmath11 and their quadratic counterparts @xmath144 .
the presentation of the algebra @xmath145 is given as follows , using the selfadjoint generators @xmath146 for @xmath129 related to the @xmath147 by @xmath148 , @xmath149 @xmath150 for @xmath151 @xmath152_-=i \,\sin ( \varphi_\ell-\varphi_m ) [ x^\ell , x^m]_+ \label{eq1.6}\ ] ] @xmath153_-=-i \,\sin ( \varphi_k)[x^0,x^k]_+ \label{eq1.7}\ ] ] for @xmath154 where ( @xmath155 ) is the cyclic permutation of ( @xmath156 ) starting with @xmath157 and where @xmath158_\pm = ab\pm ba$ ] .
the algebra @xmath159 is the quotient of @xmath145 by the two - sided ideal generated by the hermitian central element @xmath160 .
the second equivalent description of @xmath84 relies on the equality @xmath161 , i.e. the identification of @xmath93 with the spin covering of @xmath1 using the ( @xmath13-dimensional ) half spin representation of the latter .
proposition [ sym1 ] holds unchanged replacing the pair ( @xmath162 ) by the pair ( @xmath163 ) , where @xmath164 corresponds under the above isomorphism with the quotient of @xmath8 by the kernel of the covering spin@xmath165 . identifying the lie algebra @xmath166 with the lie algebra of antisymmetric six by six real matrices
, @xmath109 corresponds to the subalgebra @xmath167 of block diagonal matrices with three blocks of the form @xmath168 the @xmath4-torus @xmath169 is the quotient , @xmath170 which using the roots of @xmath171 , @xmath172 becomes , @xmath173 the relation between the parameters ( @xmath174 ) for @xmath175 and the ( @xmath132 ) is given by , @xmath176 and in terms of the @xmath174 the killing metric on @xmath169 reads , @xmath177 the action of the weyl group @xmath111 is given by the subgroup of @xmath178 @xmath179 with @xmath180 and @xmath181 , of elements @xmath182 such that @xmath183 .
the additional symmetry ( [ inv ] ) defining @xmath86 is simply @xmath184 and together with @xmath111 it generates @xmath185 .
another advantage of the variable @xmath174 is that the scaling vector field @xmath97 ( cf .
part i ) whose orbits describe the local equivalence relation on @xmath84 generated by the isomorphism of quadratic algebras , @xmath186 and which was given in the variables ( @xmath132 ) as , @xmath187 is now given by @xmath188 with @xmath189 since @xmath190 is the unit lattice for @xmath191 we can translate everything to the space @xmath192 of conjugacy classes in @xmath191 .
+ [ scaltheo ] the doubling map @xmath193 establishes an isomorphism between @xmath84 and @xmath192 transforming the scaling foliation on @xmath84 into the gradient flow ( for the killing metric ) of the character of the signature representation of @xmath1 , i.e. the super - trace of its action on @xmath194 with @xmath195 .
this flow admits remarkable compatibility properties with the canonical cell decomposition of @xmath192 , they will be analysed in part ii .
the proper understanding of the noncommutative spheres @xmath196 relies on basic computations ( theorem [ vol ] below ) whose result depends on @xmath12 through elliptic integrals .
the conceptual explanation of this dependence requires extending the moduli space from the real to the complex domain .
the leaves of the scaling foliation then appear as the real parts of a net of degree @xmath13 elliptic curves in @xmath32 having @xmath2 points in common .
these elliptic curves will turn out to play a fundamental role and to be closely related to the elliptic curves of the geometric data ( cf . section @xmath13 ) of the quadratic algebras which their elements label . to extend the moduli space to the complex domain we start with the relations defining the involutive algebra @xmath64 and take for @xmath197 the diagonal matrix with @xmath198 where @xmath199 are the coordinates of @xmath200 . using @xmath201
one obtains the homogeneous defining relations in the form , @xmath202 for any cyclic permutation @xmath203 of ( 1,2,3 ) .
the inhomogeneous relation becomes , @xmath204 and the corresponding algebra @xmath205 only depends upon the class of @xmath200 .
we let @xmath206 be the quadratic algebra defined by ( [ homc ] ) . taking @xmath207 , @xmath208 ,
for all @xmath209 and @xmath210 we obtain the defining relations ( [ eq1.6 ] ) and ( [ eq1.7 ] ) ( except for @xmath211 ) .
we showed in part i that for @xmath212 and @xmath213 ( i.e. @xmath214 and @xmath215 ) for any @xmath216 , one can find 4 scalars @xmath217 such that by setting @xmath218 for @xmath129 the system ( [ eq1.6 ] ) , ( [ eq1.7 ] ) reads @xmath219_-=ij_{\ell m } [ s_\ell , s_m]_{+ } \label{eq1.10}\ ] ] @xmath220_-=i[s_0,s_k]_{+ } \label{eq1.11}\ ] ] for @xmath154 , where @xmath203 is the cyclic permutation of ( 1,2,3 ) starting with @xmath157 and where @xmath221 .
one has @xmath222 and the relations ( [ eq1.10 ] ) , ( [ eq1.11 ] ) together with ( [ eq1.12 ] ) for the scalars @xmath223 characterize the sklyanin algebra @xcite , @xcite , a regular algebra of global dimension 4 which has been widely studied ( see e.g. @xcite , @xcite , @xcite ) and which plays an important role in noncommutative algebraic geometry . from ( [ eq1.12 ] )
it follows that for the above ( generic ) values of @xmath12 , @xmath145 only depends on 2 parameters ; with @xmath97 as in ( [ eq1.9 ] ) one has @xmath224 and the leaf of the scaling foliation through a generic @xmath225 is the connected component of @xmath12 in @xmath226 in terms of homogeneous parameters the functions @xmath223 read as @xmath227 for any cyclic permutation @xmath228 of ( 1,2,3 ) , and extend to the complex domain @xmath229 as , @xmath230 it follows easily from the finer theorem [ iden ] that for generic values of @xmath229 the quadratic algebra @xmath206 only depends upon @xmath231 .
we thus define @xmath232 let then , @xmath233 be the lagrange resolvent of the 4th degree equation , @xmath234 viewed as a map @xmath235 where @xmath53 is the following set of 8 points @xmath236 we extend the generic definition ( [ def ] ) to arbitray @xmath237 and define @xmath238 in general as the union of @xmath53 with the fiber of @xmath239 through @xmath12 .
it can be understood geometrically as follows .
let @xmath240 be the net of quadrics in @xmath32 which contain @xmath53 .
given @xmath237 the elements of @xmath240 which contain @xmath12 form a pencil of quadrics with base locus @xmath241 which is an elliptic curve of degree 4 containing @xmath53 and @xmath12 .
one has @xmath242
@xmath243 figure @xmath104 : the elliptic curve @xmath244 we shall now give , for generic values of @xmath245 a parametrization of @xmath238 by @xmath14-functions .
we start with the equations for @xmath238 @xmath246 and we diagonalize the above quadratic forms as follows @xmath247 where @xmath248 where @xmath28 is the involution , @xmath249\ ] ] in these terms the equations for @xmath238 read @xmath250 let now @xmath251 , @xmath252 and @xmath253 be such that one has , modulo projective equivalence , @xmath254 where @xmath255 are the theta functions for the lattice @xmath256 [ theta ] the following define isomorphisms of @xmath257 with @xmath238 , @xmath258 and @xmath259 .
* proof * up to an affine transformation , @xmath260 ( and @xmath261 are ) is the classical projective embedding of @xmath257 in @xmath32 .
thus we only need to check that the biquadratic curve @xmath262 is given by ( [ eq6.11 ] ) .
it is thus enough to check ( [ eq6.11 ] ) on @xmath263 .
this follows from the basic relations @xmath264 and @xmath265 which one uses to check @xmath266 and @xmath267 respectively.@xmath268 + the elements of @xmath53 are obtained from the following values of @xmath269 @xmath270 and @xmath271 ( we used @xmath272 to go back to the coordinates @xmath273 ) . + let @xmath274 be the klein subgroup of the symmetric group @xmath113 acting on @xmath32 by permutation of the coordinates @xmath199 .
+ for @xmath120 in @xmath275 one has @xmath276 , so that @xmath120 defines for each @xmath12 an automorphism of @xmath238 . for @xmath120 in @xmath275
the matrix @xmath277 is diagonal with @xmath278 on the diagonal and the quasiperiodicity of the @xmath14-functions allows to check that these automorphisms are translations on @xmath238 by the following 2-torsion elements of @xmath257 , @xmath279 let @xmath280 be the complement of the @xmath13 hyperplanes @xmath281 with @xmath282 .
then @xmath283 defines an involutive automorphism @xmath284 of @xmath285 and since one has @xmath286 it follows that @xmath287 , so that @xmath284 defines for each @xmath288 an involutive automorphism of @xmath289 which extends canonically to @xmath238 , in fact , [ nice ] the restriction of @xmath284 to @xmath238 is the symmetry @xmath290 around any of the points @xmath291 in the elliptic curve @xmath238 .
this symmetry , as well as the above translations by two torsion elements does not refer to a choice of origin in the curve @xmath238 .
the proof follows from identities on theta functions .
let @xmath292 . by section @xmath29
the torus @xmath293 gives a covering of the real moduli space @xmath84 .
for @xmath294 , the point @xmath295 is real with projective coordinates @xmath296 the corresponding fiber @xmath297 is stable under complex conjugation @xmath298 and the intersection of @xmath297 with the real moduli space is given by , @xmath299 the curve @xmath297 is defined over @xmath300 and ( [ real ] ) determines its purely points .
note that @xmath301 ( [ real ] ) is invariant under the klein group @xmath275 and thus has two connected components , we let @xmath302 be the component containing @xmath303 .
the real points , @xmath304 of @xmath297 do play a complementary role in the characteristic variety ( proposition [ siginvar ] ) .
let us recall the definition of the geometric data @xmath24 for quadratic algebras .
let @xmath306 be a quadratic algebra where @xmath307 is a finite - dimensional complex vector space and where @xmath308 is the two - sided ideal of the tensor algebra @xmath309 of @xmath307 generated by the subspace @xmath310 of @xmath311 .
consider the subset of @xmath312 of pairs @xmath313 such that @xmath314 for any @xmath315 .
since @xmath310 is homogeneous , ( [ eq5.1 ] ) defines a subset @xmath316 where @xmath317 is the complex projective space of one - dimensional complex subspaces of @xmath318 .
let @xmath319 and @xmath320 be the first and the second projection of @xmath118 in @xmath317 .
it is usually assumed that they coincide i.e. that one has @xmath321 and that the correspondence @xmath27 with graph @xmath118 is an automorphism of @xmath322 , @xmath26 being the pull - back on @xmath322 of the dual of the tautological line bundle of @xmath317 .
the algebraic variety @xmath322 is refered to as the characteristic variety . in many cases
@xmath322 is the union of an elliptic curve with a finite number of points which are invariant by @xmath27 .
this is the case for @xmath323 at generic @xmath12 since it then reduces to the sklyanin algebra for which this is known @xcite , @xcite . when @xmath12 is non generic , e.g. when the isomorphism with the sklyanin algebra breaks down , the situation is more involved , and the characteristic variety can be as large as @xmath32 .
this is described in part i , where we gave a complete description of the geometric datas .
our aim now is to show that for @xmath229 generic , there is an astute choice of generators of the quadratic algebra @xmath323 for which the characteristic variety @xmath324 actually coincides with the fiber variety @xmath238 and to identify the corresponding automorphism @xmath27 .
since this coincidence no longer holds for non - generic values it is a quite remarkable fact which we first noticed by comparing the @xmath33-invariants of these two elliptic curves .
+ let @xmath229 be generic , we perform the following change of generators @xmath325 we let @xmath223 be as before , given by ( [ jlm ] ) @xmath326 with @xmath327 given by ( [ eq6.2 ] ) .
finally let @xmath328 be the @xmath13 points of @xmath32 whose homogeneous coordinates ( @xmath329 ) all vanish but one .
+ [ iden ] 1 ) in terms of the @xmath330 , the relations of @xmath331 take the form @xmath332_- & = & [ y_\ell , y_m]_+\label{eq5.6}\\ { [ y_\ell , y_m]}_- & = & - j_{\ell m } [ y_0,y_k]_+\label{eq5.7}\end{aligned}\ ] ] for any @xmath175 , @xmath203 being a cyclic permutation of ( 1,2,3 ) \2 ) the characteristic variety @xmath324 is the union of @xmath238 with the @xmath13 points @xmath328 .
\3 ) the automorphism @xmath27 of the characteristic variety @xmath324 is given by @xmath333 on @xmath238 and @xmath334 on the 4 points @xmath328 .
\4 ) the automorphism @xmath27 is the restriction to @xmath238 of a birational automorphism of @xmath32 independent of @xmath12 and defined over @xmath34 .
the resemblance between the above presentation and the sklyanin one ( [ eq1.10 ] ) , ( [ eq1.11 ] ) is misleading , for the latter all the characteristic varieties are contained in the same quadric ( cf .
@xcite 2.4 ) @xmath335 and ca nt of course form a net of essentially disjoint curves . *
proof * @xmath336 by construction @xmath337 where @xmath338 one checks that it is the union of the fiber @xmath238 ( in the generic case ) with the above @xmath13 points .
the automorphism @xmath27 of the characteristic variety @xmath324 is given by definition by the equation , @xmath339 where @xmath340 is the column vector @xmath341 ( in the variables @xmath342 ) .
one checks that @xmath340 is already determined by the equations in ( [ aut ] ) corresponding to the first three lines in @xmath343 which are independent of @xmath327 ( see below ) .
thus @xmath27 is in fact an automorphism of @xmath32 which is the identity on the above four points and which restricts as automorphism of @xmath238 for each @xmath12 generic .
one checks that @xmath27 is the product of two involutions which both restrict to @xmath324 ( for @xmath12 generic ) @xmath344 where @xmath284 is the involution of the end of section 3 corresponding to @xmath345 and where @xmath346 is given by @xmath347 for @xmath175 and which restricts obviously to @xmath324 in view of ( [ eq6.9 ] ) .
both @xmath284 and @xmath346 are the identity on the above four points and since @xmath346 induces the symmetry @xmath348 around @xmath349 ( proposition [ theta ] ) one gets the result using proposition [ nice ] .
the fact that @xmath27 does not depend on the parameters @xmath350 plays an important role .
explicitly we get from the first 3 equations ( [ aut ] ) @xmath351 for @xmath129 , where @xmath352 and @xmath353 for @xmath354 .
let @xmath306 be a quadratic algebra .
its geometric data @xmath24 is defined in such a way that @xmath25 maps homomorphically to a cross - product algebra obtained from sections of powers of the line bundle @xmath26 on powers of the correspondence @xmath27 ( @xcite ) .
we shall begin by a purely algebraic result which considerably refines the above homomorphism and lands in a richer cross - product .
we use the notations of section 4 for general quadratic algebras . [ cent ]
let @xmath355 be a symmetric bilinear form on @xmath318 and @xmath356 a component of @xmath357 .
we shall say that @xmath358 is on @xmath356 iff for all ( @xmath359 ) in @xmath356 and @xmath315 one has , @xmath360 let @xmath356 be a component of @xmath357 globally invariant under the map @xmath361 given a quadratic form @xmath358 central and not identically zero on the component @xmath356 , we define as follows an algebra @xmath362 as a generalised cross - product of the ring of meromorphic functions on @xmath356 by the transformation @xmath363 .
let @xmath364 , @xmath365 be such that @xmath366 does not vanish identically on @xmath356 .
we adjoin two generators @xmath367 and @xmath368 which besides the usual cross - product rules , @xmath369 fulfill the following relations , @xmath370 where the function @xmath371 is given by the ratio , @xmath372 the a priori dependence on @xmath364 , @xmath373 is eliminated by the rules , @xmath374 which allow to adjoin all @xmath367 and @xmath368 for @xmath364 and @xmath373 not identically zero on the projections of @xmath356 , without changing the algebra and provides an intrinsic definition of @xmath362 . our first result is [ alg0 ] let @xmath358 be central and not identically zero on the component @xmath356 .
\(i ) the following equality defines a homomorphism @xmath120 : @xmath375 @xmath376 ( ii ) if @xmath377 , then @xmath378 where @xmath358 is viewed as an element of @xmath379 . formula ( i )
is independent of @xmath364 , @xmath373 using ( [ compare ] ) and reduces to @xmath380 when @xmath381 is non - trivial on the two projections of @xmath356 .
it is enough to check that the @xmath382 fulfill the quadratic relations @xmath315 .
we view @xmath315 as a bilinear form on @xmath318 .
the vanishing of the terms in @xmath383 and in @xmath384 is automatic by construction of the characteristic variety .
the vanishing of the sum of terms in @xmath385 follows from definition [ cent ] .
let @xmath323 at generic @xmath12 , then the center of @xmath331 is generated by the three linearly dependent quadratic elements @xmath386 with the notations of theorem [ iden ] .
[ central ] let @xmath323 at generic @xmath12 , then each @xmath387 is central on @xmath388 ( @xmath389 ) .
one uses ( [ eq5.11 ] ) to check the algebraic identity .
together with lemma [ alg0 ] this yields non trivial homomorphisms of @xmath331 whose unitarity will be analysed in the next section .
note that for a general quadratic algebra @xmath306 and a quadratic form @xmath355 , such that @xmath390 center(@xmath25 ) , it does not automatically follow that @xmath358 is central on @xmath357 .
for instance proposition [ central ] no longer holds on @xmath391 where @xmath328 is any of the four points of @xmath392 not in @xmath238 .
the algebra @xmath331 , @xmath294 is by construction a _ quadratic @xmath54-algebra _ i.e. a quadratic complex algebra @xmath393 which is also a @xmath54-algebra with involution @xmath394 preserving the subspace @xmath307 of generators .
equivalently one can take the generators of @xmath25 ( spanning @xmath307 ) to be hermitian elements of @xmath25 . in such a case
the complex finite - dimensional vector space @xmath307 has a real structure given by the antilinear involution @xmath395 obtained by restriction of @xmath394 .
since one has @xmath396 for @xmath397 , it follows that the set @xmath310 of relations satisfies @xmath398 in @xmath311 where @xmath399 is the transposition @xmath400 .
this implies [ conj ] the characteristic variety is stable under the involution @xmath401 and one has @xmath402 we let @xmath356 be as above an invariant component of @xmath357 we say that @xmath356 is @xmath33-real when it is globally invariant under the involution @xmath403 by lemma [ conj ] this involution commutes with the automorphism @xmath363 ( [ siginv ] ) and the following turns the cross - product @xmath362 into a @xmath54-algebra , @xmath404 provided that @xmath355 fulfills @xmath405 .
we use the transpose of @xmath33 , so that @xmath406 the homomorphism @xmath120 of lemma [ alg0 ] is a @xmath54-homomorphism . composing @xmath120 with the restriction to the subset @xmath407 one obtains a @xmath54-homomorphism @xmath14 of @xmath393 to a twisted cross - product @xmath0-algebra , @xmath408 . given a compact space @xmath8 ,
an homeomorphism @xmath27 of @xmath8 and a hermitian line bundle @xmath26 on @xmath8 we define the @xmath0-algebra @xmath408 as the twisted cross - product of @xmath409 by the hilbert @xmath410-bimodule associated to @xmath26 and @xmath27 ( @xcite , @xcite ) .
we let for each @xmath411 , @xmath412 be the hermitian line bundle pullback of @xmath26 by @xmath413 and ( cf .
@xcite , @xcite ) @xmath414 we first define a @xmath54-algebra as the linear span of the monomials @xmath415 with product given as in ( @xcite , @xcite ) for @xmath416 so that @xmath417 we use the hermitian structure of @xmath418 to give meaning to the products @xmath419 and @xmath420 for @xmath421 .
the product then extends uniquely to an associative product of @xmath54-algebra fulfilling the following additional rules @xmath422 the @xmath0-norm of @xmath408 is defined as for ordinary cross - products and due to the amenability of the group @xmath423 there is no distinction between the reduced and maximal norms .
the latter is obtained as the supremum of the norms in involutive representations in hilbert space .
the natural positive conditional expectation on the subalgebra @xmath409 shows that the @xmath0-norm restricts to the usual sup norm on @xmath409 . to lighten notations in the next statement
we abreviate @xmath424 as @xmath425 , [ c * ] let @xmath426 be a compact @xmath27-invariant subset and @xmath358 be central and strictly positive on @xmath427 .
let @xmath26 be the restriction to @xmath8 of the dual of the tautological line bundle on @xmath317 endowed with the unique hermitian metric such that @xmath428 \(i ) the equality @xmath429 yields a @xmath54-homomorphism @xmath430 \(ii ) for any @xmath431 the @xmath0-norm of @xmath432 fulfills @xmath433 \(iii ) if @xmath377 , then @xmath434 where @xmath358 is viewed as an element of @xmath379 .
we shall now apply this general result to the algebras @xmath331 , @xmath294 .
we take the quadratic form @xmath435 in the @xmath436-coordinates , so that @xmath358 is the canonical central element defining the sphere @xmath11 by the equation @xmath437 .
proposition [ central ] shows that @xmath358 is central on @xmath388 for generic @xmath12 .
the positivity of @xmath358 is automatic since in the @xmath436-coordinates the involution @xmath438 coming from the involution of the quadratic @xmath54-algebra @xmath331 is simply complex conjugation @xmath439 , so that @xmath440 for @xmath441 .
we thus get , [ ii ] let @xmath442 be a compact @xmath27-invariant subset .
the homomorphism @xmath14 of theorem [ c * ] is a unital @xmath54-homomorphism from @xmath443 to the cross - product @xmath444 .
this applies in particular to @xmath445 .
it follows that one obtains a non - trivial @xmath0-algebra @xmath35 as the completion of @xmath443 for the semi - norm , @xmath446 where @xmath3 varies through all unitary representations of @xmath443 . to analyse the compact @xmath27-invariant subsets of @xmath447 for generic @xmath12 , we distinguish two cases . first note that the real curve @xmath244 is non empty ( it contains @xmath448 ) , and has two connected components since it is invariant under the klein group @xmath275 ( [ eq6.17 ] ) .
@xmath449 figure @xmath29 : the elliptic curve @xmath244 ( odd case ) we say that @xmath294 is when @xmath27 preserves each of the two connected components of the real curve @xmath244 and when it permutes them ( cf . figure @xmath29 ) .
a generic @xmath294 is even ( cf . figure @xmath104 ) iff the @xmath450 of ( [ real0 ] ) ( @xmath451 ) have the same sign . in that case
@xmath27 is the square of a real translation @xmath452 of the elliptic curve @xmath238 preserving @xmath244 .
[ siginvar ] let @xmath294 be generic and even .
\(i ) each connected component of @xmath244 is a minimal compact @xmath27-invariant subset .
\(ii ) let @xmath442 be a compact @xmath27-invariant subset , then @xmath8 is the sum in the elliptic curve @xmath238 with origin @xmath448 of @xmath453 ( cf .
[ real ] ) with the component @xmath454 of @xmath244 containing @xmath448 .
\(iii ) the cross - product @xmath455 is isomorphic to the mapping torus of the automorphism @xmath456 of the noncommutative torus @xmath457 acting on the generators by the matrix @xmath458 $ ] .
more precisely with @xmath459 the generators one has @xmath460 the mapping torus of the automorphism @xmath456 is given by the algebra of continuous maps @xmath461 such that @xmath462 .
[ h3 ] let @xmath294 be generic and even , then @xmath463 is a noncommutative @xmath4-manifold with an elliptic action of the three dimensional heisenberg lie algebra @xmath31 and an invariant trace @xmath464 .
we refer to @xcite and @xcite where these noncommutative manifolds were introduced and analysed in terms of crossed products by hilbert @xmath410-bimodules .
one can construct directly the action of @xmath31 on @xmath465 by choosing a constant ( translation invariant ) curvature connection @xmath466 , compatible with the metric , on the hermitian line bundle @xmath26 on @xmath447 ( viewed in the @xmath467-category not in the holomorphic one ) .
the two covariant differentials @xmath468 corresponding to the two vector fields @xmath469 on @xmath447 generating the translations of the elliptic curve , give a natural extension of @xmath469 as the unique derivations @xmath470 of @xmath465 fulfilling the rules , @xmath471 we let @xmath472 be the unique derivation of @xmath465 corresponding to the grading by powers of @xmath111 .
it vanishes on @xmath473 and fulfills @xmath474 let @xmath475 be the constant curvature of the connection @xmath466 , one gets @xmath476=\,\kappa\ , \delta \,,\quad [ \delta,\ , \delta_j]=\,0\ ] ] which provides the required action of the lie algebra @xmath31 on @xmath465 .
integration on the translation invariant volume form @xmath477 of @xmath238 gives the @xmath31-invariant trace @xmath464 , @xmath478 it follows in particular that the results of @xcite apply to obtain the calculus .
in particular the following gives the fundamental class " as a @xmath4-cyclic cocycle , @xmath479 where the @xmath470 are the above derivations with @xmath480 .
we shall in fact describe the same calculus in greater generality in the last section which will be devoted to the computation of the jacobian of the homomorphism @xmath14 of corollary [ ii ] .
similar results hold in the odd case .
then @xmath244 is a minimal compact @xmath27-invariant subset , any compact @xmath27-invariant subset @xmath442 is the sum in the elliptic curve @xmath238 with origin @xmath448 of @xmath244 with @xmath453 but the latter is automatically invariant under the subgroup @xmath481 of order @xmath29 of the klein group @xmath275 ( [ eq6.17 ] ) @xmath482 the group law in @xmath238 is described geometrically as follows .
it involves the point @xmath303 .
the sum @xmath483 of two points @xmath436 and @xmath484 of @xmath238 is @xmath485 where @xmath486 is the 4th point of intersection of @xmath238 with the plane determined by the three points @xmath487 .
it commutes by construction with complex conjugation so that @xmath488 . by lemma [ conj ]
the translation @xmath27 is imaginary for the canonical involution @xmath438 . in terms of the coordinates
@xmath329 this involution is described as follows , using ( [ change0 ] ) ( multiplied by @xmath489 ) to change variables . among the @xmath4
real numbers @xmath490 two have the same sign @xmath491 and one , @xmath492 , @xmath175 , the opposite sign
. then @xmath493 where @xmath494 is complex conjugation on the real elliptic curve @xmath238 ( section 3 ) and @xmath495 the involution @xmath496 the index @xmath157 and the sign @xmath491 remain constant when @xmath12 varies in each of the four components of the complement of the four points @xmath497 in @xmath498 .
the sign @xmath491 matters for the action of @xmath438 on linear forms as in ( [ inv2 ] ) , but is irrelevant for the action on @xmath238 .
each involution @xmath495 is a symmetry @xmath499 in the elliptic curve @xmath238 and the products @xmath500 form the klein subgroup @xmath275 ( [ eq6.17 ] ) acting by translations of order two on @xmath238 .
the quadratic form @xmath358 of ( [ quad ] ) is given in the new coordinates by , @xmath501 with @xmath502 and @xmath503 defined by ( [ qk ] ) .
let @xmath294 be generic and even and @xmath504 .
let @xmath505 be the minimal compact @xmath27-invariant subset containing @xmath103 ( proposition [ siginvar ] ( ii ) ) .
by corollary [ ii ] we get a homomorphism , @xmath506 whose non - triviality will be proved below in corollary [ nct ] .
we shall first show ( theorem [ gal ] ) that it transits through the cross - product of the field @xmath6 of meromorphic functions on the elliptic curve by the subgroup of its galois group aut@xmath7 generated by the translation @xmath27 . for @xmath507 , @xmath508 , one has using ( [ invole ] ) and ( [ real ] ) , @xmath509 which is a rational function @xmath510 . fixing @xmath511 and substituting @xmath97 and @xmath512 in the formulas ( [ crossed ] ) and ( [ ratio ] ) of lemma [ alg0 ] with @xmath364 real such that @xmath513 ,
@xmath514 and @xmath358 given by ( [ q13 ] ) we obtain rational formulas for a homomorphism @xmath515 of @xmath443 to the generalised cross - product of the field @xmath6 of meromorphic functions @xmath516 on the elliptic curve @xmath238 by @xmath27 .
the generalised cross - product rule ( [ crossed ] ) is given by @xmath517 where @xmath518 is a rational function .
similarly @xmath519 . using integration on the cycle @xmath520 to obtain a trace , together with corollary [ ii ] ,
we get , [ gal ] the homomorphism @xmath521 factorises with a homomorphism @xmath522 to the generalised cross - product of the field @xmath6 of meromorphic functions on the elliptic curve @xmath238 by the subgroup of the galois group aut@xmath7 generated by @xmath27 .
its image generates the hyperfinite factor of type ii@xmath5 after weak closure relative to the trace given by integration on the cycle @xmath520 .
elements of @xmath6 with poles on @xmath520 are unbounded and give elements of the regular ring of affiliated operators , but all elements of @xmath523 are regular on @xmath520 .
the above generalisation of the cross - product rules ( [ crossed ] ) with the rational formula for @xmath517 is similar to the introduction of @xmath29-cocycles in the standard brauer theory of central simple algebras .
in this section we shall analyse the morphism of @xmath54-algebras @xmath525 of corollary [ ii ] , by computing its jacobian in the sense of noncommutative differential geometry ( @xcite ) .
the usual jacobian of a smooth map @xmath526 of manifolds is obtained as the ratio @xmath527 of the pullback of the volume form @xmath528 of the target manifold @xmath529 with the volume form @xmath530 of the source manifold @xmath28 . in noncommutative geometry ,
differential forms @xmath531 of degree @xmath9 become hochschild classes @xmath532 .
moreover since one works with the dual formulation in terms of algebras , the pullback @xmath533 is replaced by the pushforward @xmath534 under the corresponding transposed morphism of algebras @xmath535 .
the noncommutative sphere @xmath11 admits a canonical volume form " given by the hochschild @xmath4-cycle @xmath536 . our goal is to compute the push - forward , @xmath537
let @xmath12 be generic and even .
the noncommutative manifold @xmath538 is , by corollary [ h3 ] , a noncommutative @xmath4-dimensional nilmanifold isomorphic to the mapping torus of an automorphism of the noncommutative @xmath29-torus @xmath30 .
its hohschild homology is easily computed using the corresponding result for the noncommutative torus ( @xcite ) .
it admits in particular a canonical volume form @xmath539 which corresponds to the natural class in @xmath540 ( @xcite ) .
the volume form @xmath307 is obtained in the cross - product @xmath538 from the translation invariant @xmath29-form @xmath477 on @xmath238 . to compare @xmath541 with @xmath307
we shall pair it with the @xmath4-dimensional hochschild cocycle @xmath542 given , for any element @xmath543 of the center of @xmath544 , by @xmath545 where @xmath546 is the fundamental class in cyclic cohomology defined by ( [ 3trace ] ) .
by @xcite , ( 2.13 ) p.549 , the component @xmath536 of the chern character is given by , @xmath547 where @xmath548 . in terms of the @xmath330 one
gets , @xmath549 where @xmath550 , @xmath502 and @xmath551 with @xmath552 and @xmath553 .
the normalization factor is @xmath554 formula ( [ beauty ] ) shows that , up to normalization , @xmath555 only depends on the fiber @xmath238 of @xmath12 .
let @xmath294 be generic , we assume for simplicity that @xmath12 is even , there is a similar formula in the odd case . in our case the involutions @xmath284 and @xmath346 are conjugate by a real translation @xmath452 of the elliptic curve @xmath238 and we let @xmath556 be one of the two connected components of , @xmath557 by proposition [ siginvar ] we can identify the center of @xmath558 with @xmath559 .
we assume for simplicity that @xmath560 $ ] are in cyclic order @xmath561 for some @xmath562 .
[ vol ] let @xmath563 center @xmath564 .
then @xmath565 where @xmath566 is the period given by the elliptic integral of the first kind , @xmath567 and @xmath310 the rational fraction , @xmath568 with @xmath569 .
we first obtained this result by direct computation using the explicit formula ( [ 3trace ] ) and the natural constant curvature connection on @xmath26 given by the parametrisation of proposition [ theta ] in terms of @xmath14-functions . the pairing was first expressed in terms of elliptic functions and modular forms , and the conceptual understanding of its simplicity is at the origin of many of the notions developped in the present paper and in particular of the rational " formulation of the calculus which will be obtained in the last section .
the geometric meaning of theorem [ vol ] is the computation of the jacobian in the sense of noncommutative geometry of the morphism @xmath14 as explained above .
the differential form @xmath570 is independent of @xmath157 and is , up to scale , the only holomorphic form of type @xmath571 on @xmath238 , it is invariant under the translations of the elliptic curve . the integral @xmath566 is ( up to a trivial normalization factor ) a standard elliptic integral , it is given by an hypergeometric function in the variable @xmath572 or a modular form in terms of @xmath18 .
the differential of @xmath310 is given on @xmath238 by @xmath573 where @xmath574 the period @xmath566 does not vanish and @xmath575 , only vanishes on the @xmath13 ramification points " necessarily present due to the symmetries .
[ nct0 ] the jacobian of the map @xmath576 is given by the equality @xmath577 where @xmath578 is the element of the center @xmath559 of @xmath544 given by formula ( [ dr ] ) .
this statement assumes that @xmath12 is generic in the measure theoretic sense so that @xmath21 admits good diophantine approximation ( @xcite ) .
it justifies in particular the terminology of ramified covering " applied to @xmath576 .
the function @xmath578 has only @xmath13 zeros on @xmath556 which correspond to the ramification .
as shown by theorem [ iden ] the algebra @xmath331 is defined over @xmath300 , i.e. admits a natural antilinear automorphism of period two , @xmath518 uniquely defined by @xmath579 theorem [ iden ] also shows that @xmath27 is defined over @xmath300 and hence commutes with complex conjugation @xmath580 .
this gives a natural real structure @xmath518 on the algebra @xmath362 with @xmath581 and @xmath358 as above , @xmath582 one checks that the morphism @xmath120 of lemma [ alg0 ] is real " i.e. that , @xmath583 since @xmath580 reverses the orientation of @xmath238 , while @xmath518 preserves the orientation of @xmath11 it follows that @xmath584 and @xmath578 necessarily vanishes on @xmath585 .
note also that for general @xmath543 one has @xmath586 which shows that both @xmath587 and @xmath588 are non trivial hochschild classes .
these results hold in the smooth algebra @xmath589 containing the closure of @xmath590 under holomorphic functional calculus in the @xmath0 algebra @xmath35 .
we can also use theorem [ vol ] to show the non - triviality of the morphism @xmath521 of ( [ ncu ] ) .
[ nct ] the pullback of the fundamental class @xmath591 $ ] of the noncommutative torus by the homomorphism @xmath521 of ( [ ncu ] ) is non zero , @xmath592 ) \neq 0 \in hh^2 $ ] provided @xmath103 is not a ramification point .
we have shown above the non - triviality of the hochschild homology and cohomology groups @xmath593 and @xmath594 by exhibiting specific elements with non - zero pairing . combining the ramified cover @xmath595 with the natural spectral geometry ( spectral triple ) on the noncommutative @xmath4-dimensional nilmanifold @xmath538 yields a natural spectral triple on @xmath11 in the generic case .
it will be analysed in part ii , together with the @xmath0-algebra @xmath35 , the vanishing of the primary class of @xmath47 in @xmath596 , and the cyclic cohomology of @xmath589 .
theorem [ vol ] suggests the existence of a rational " form of the calculus explaining the appearance of the elliptic period @xmath566 and the rationality of @xmath310 .
we shall show in this last section that this indeed the case .
let us first go back to the general framework of twisted cross products of the form @xmath597 where @xmath27 is a diffeomorphism of the manifold @xmath28 .
we shall follow @xcite to construct cyclic cohomology classes from cocycles in the bicomplex of group cohomology ( with group @xmath598 ) with coefficients in de rham currents on @xmath28 .
the twist by the line bundle @xmath26 introduces a non - trivial interesting nuance .
let @xmath599 be the algebra of smooth differential forms on @xmath28 , endowed with the action of @xmath600 @xmath601 as in @xcite p. 219
we let @xmath602 be the graded algebra obtained as the ( graded ) tensor product of @xmath599 by the exterior algebra @xmath603')$ ] on the augmentation ideal @xmath604'$ ] in the group ring @xmath604 $ ] . with @xmath605 ,
n\in \mathbb z$ ] the canonical basis of @xmath604 $ ] , the augmentation @xmath606\mapsto \mathbb c$ ] fulfills @xmath607)=1 , \forall n\,$ ] , and @xmath608- [ 0]\,,\quad \ ,
n\in \mathbb z\,,\quad n\neq 0\ ] ] is a linear basis of @xmath604'$ ] .
the left regular representation of @xmath600 on @xmath604 $ ] restricts to @xmath604'$ ] and is given on the above basis by @xmath609 it extends to an action @xmath610 of @xmath600 by automorphisms of @xmath611 ' $ ] .
we let @xmath612 be the tensor product action of @xmath600 on @xmath613 ' $ ] .
we now use the hermitian line bundle @xmath26 to form the twisted cross - product @xmath614 we let @xmath615 be as in ( [ gene2 ] ) for @xmath616 and extend its definition for @xmath617 so that @xmath618 is the pullback by @xmath413 of the dual @xmath619 of @xmath615 for all @xmath52 .
the hermitian structure gives an antilinear isomorphism @xmath620 .
the algebra @xmath621 is the linear span of monomials @xmath622 where @xmath623 with the product rules ( [ gene3 ] ) , ( [ gene1 ] ) .
let @xmath466 be a hermitian connection on @xmath26 .
we shall turn @xmath621 into a differential graded algebra . by functoriality @xmath466
gives a hermitian connection on the @xmath624 and hence a graded derivation @xmath625 whose square @xmath626 is multiplication by the curvature @xmath627 of @xmath615 , @xmath628 with @xmath629 the curvature of @xmath26 .
ones has @xmath630 and we extend the differential @xmath9 to a graded derivation on @xmath631 by @xmath632 we can then extend @xmath633 uniquely to @xmath634 so that it fulfills @xmath635 to construct closed graded traces on this differential graded algebra we follow ( @xcite ) and consider the double complex of group cochains ( with group @xmath598 ) with coefficients in de rham currents on @xmath28 .
the cochains @xmath638 are totally antisymmetric maps from @xmath639 to the space @xmath640 of de rham currents of dimension @xmath641 , which fulfill @xmath642 besides the coboundary @xmath643 of group cohomology , given by @xmath644 and the coboundary @xmath645 of de rham homology , @xmath646 the curvatures @xmath647 generate the further coboundary @xmath648 defined on ker @xmath643 by , @xmath649 which maps ker @xmath650 to @xmath651 .
translation invariance follows from ( [ curvcoc ] ) and @xmath652 for @xmath653 , @xmath654 .
we shall now show how the above general framework allows to reformulate the calculus involved in theorem [ vol ] in rational terms .
we let @xmath28 be the elliptic curve @xmath238 where @xmath12 is generic and even .
let then @xmath466 be an arbitrary hermitian connection on @xmath26 and @xmath452 its curvature .
we first display a cocycle @xmath667 which reproduces the cyclic cocycle @xmath668 .
we use the generic hypothesis in the measure theoretic sense to solve the small denominator " problem in ( i ) . in
( ii ) we identify differential forms @xmath678 of degree @xmath9 with the dual currents of dimension @xmath679 .
it is a general principle explained in @xcite that a cyclic cocycle @xmath464 generates a calculus whose differential graded algebra is obtained as the quotient of the universal one by the radical of @xmath464 .
we shall now explicitely describe the reduced calculus obtained from the cocycle of lemma [ dioph ] _
( iii)_. we use as above the hermitian line bundle @xmath26 to form the twisted cross - product @xmath680 of the algebra @xmath599 of differential forms on @xmath28 by the diffeomorphism @xmath27 . instead of having to adjoin the infinite number of odd elements @xmath681 we just adjoin two @xmath682 and @xmath683 as follows .
we let @xmath472 be the derivation of @xmath684 such that @xmath685 we adjoin @xmath682 to @xmath684 by tensoring @xmath684 with the exterior algebra @xmath686 generated by an element @xmath682 of degree 1 , and extend the connection @xmath466 ( [ nabla1 ] ) to the unique graded derivation @xmath687 of @xmath688 such that , @xmath689 with @xmath690 as in lemma [ dioph ] . by construction ,
every element of @xmath691 is of the form @xmath692 one does not yet have a graded differential algebra since @xmath693 .
however , with @xmath669 as in lemma [ dioph ] one has @xmath694 \,,\quad \forall x \in \omega'=\calb \otimes \wedge\{\chi\}\ ] ] and one can apply lemma @xmath695 p.@xmath696 of @xcite to get a differential graded algebra by adjoining the degree @xmath104 element @xmath697 fulfilling the rules @xmath698 and defining the differential @xmath9 by , @xmath699\,,\quad \forall x \in \omega ' \nonumber\\ dx & = & 0\end{aligned}\ ] ] where @xmath700 $ ] is the graded commutator .
it follows from lemma @xmath695 p.@xmath696 of @xcite that we obtain a differential graded algebra @xmath701 , generated by @xmath684 , @xmath702 and @xmath683 . in fact using ( [ x ] ) every element of @xmath701 is of the form @xmath703 and we define the functional @xmath704 on @xmath701 by extending the ordinary integral , @xmath705 first to @xmath706 by @xmath707 then to @xmath691 by @xmath708 and finally to @xmath701 as in lemma @xmath695 p.@xmath696 of @xcite , @xmath709 now in our case the line bundle @xmath26 is holomorphic and we can apply theorem [ fund ] to its canonical hermitian connection @xmath466 .
we take the notations of section 5 , with @xmath581 , and @xmath358 given by ( [ q13 ] ) .
this gives a particular rational " form of the calculus which explains the rationality of the answer in theorem [ vol ] .
we first extend as follows the construction of @xmath362 .
we let @xmath712 be the generalised cross - product of the algebra @xmath713 of meromorphic differential forms ( in @xmath714 and @xmath715 ) on @xmath356 by the transformation @xmath363 .
the generators @xmath367 and @xmath368 fulfill the cross - product rules , @xmath716 while ( [ crossed ] ) is unchanged .
the connection @xmath466 is the restriction to the subspace @xmath717 of the unique graded derivation @xmath466 on @xmath712 which induces the usual differential on @xmath713 and satisfies , @xmath718 where @xmath719 and @xmath720 are the ( partial ) differentials relative to the variables @xmath97 and @xmath721 .
note that one needs to check that the involved differential forms such as @xmath722 are not only invariant under the scaling transformations @xmath723 but are also , i.e. have zero restriction to the fibers of the map @xmath724 , in both variables @xmath97 and @xmath721 . by definition
the derivation @xmath725 of @xmath712 vanishes on @xmath713 and fulfills @xmath726 where @xmath727 is a basic form which when restricted to the subspace @xmath717 is the curvature .
we let as above @xmath472 be the derivation of @xmath712 which vanishes on @xmath713 and is such that @xmath728 and @xmath729 .
we proceed exactly as above and get the graded algebras @xmath730 obtained by adjoining @xmath682 and @xmath701 by adjoining @xmath683 .
we define @xmath687 , @xmath9 as in ( [ deltaext ] ) and ( [ deltaext1 ] ) and the integral @xmath704 by integration ( [ step1 ] ) on the subspace @xmath717 followed as above by steps ( [ step2 ] ) , ( [ step3 ] ) , ( [ step4 ] ) .
+ the computation of @xmath687 only involves rational fractions in the variables @xmath97 , @xmath721 ( [ rational ] ) , and the formula ( [ beauty ] ) for @xmath536 is polynomial in the @xmath367 , @xmath368 .
we thus obtain an a priori reason for the rational form of the result of theorem [ vol ] .
the explicit computation as well as its extension to the odd case and the degenerate cases will be described in part ii .
a. connes .
cyclic cohomology and the transverse fundamental class of a foliation . in _
geometric methods in operator algebras _ , pages 52144 .
pitman res .
notes in math . 123 ( 1986 ) , longman , harlow , 1986 .
m. dubois - violette . lectures on graded differential algebras and noncommutative geometry . in y.
maeda and h. moriyoshi , editors , _ noncommutative differential geometry and its applications to physics _ , pages 245306 .
shonan , japan , 1999 , kluwer academic publishers , 2001 . m. v. pimsner .
a class of @xmath736-algebra generalizing both cuntz - kriger algebras and crossed products by @xmath738 . in _ free probability theory _ , pages 189212 .
fields inst .
12 , amer .
soc . , providence , ri . | we analyse the moduli space and the structure of noncommutative 3-spheres .
we develop the notion of central quadratic form for quadratic algebras , and prove a general algebraic result which considerably refines the classical homomorphism from a quadratic algebra to a cross - product algebra associated to the characteristic variety and lands in a richer cross - product .
it allows to control the @xmath0-norm on involutive quadratic algebras and to construct the differential calculus in the desired generality .
the moduli space of noncommutative 3-spheres is identified with equivalence classes of pairs of points in a symmetric space of unitary unimodular symmetric matrices .
the scaling foliation of the moduli space is identified to the gradient flow of the character of a virtual representation of @xmath1 .
its generic orbits are connected components of real parts of elliptic curves which form a net of biquadratic curves with @xmath2 points in common .
we show that generically these curves are the same as the characteristic variety of the associated quadratic algebra .
we then apply the general theory of central quadratic forms to show that the noncommutative 3-spheres admit a natural ramified covering @xmath3 by a noncommutative @xmath4-dimensional nilmanifold .
this yields the differential calculus .
we then compute the jacobian of the ramified covering @xmath3 by pairing the direct image of the fundamental class of the noncommutative @xmath4dimensional nilmanifold with the chern character of the defining unitary and obtain the answer as the product of a period ( of an elliptic integral ) by a rational function .
finally we show that the hyperfinite factor of type ii@xmath5 appears as cross - product of the field @xmath6 of meromorphic functions on an elliptic curve by a subgroup of its galois group aut@xmath7 .
* moduli space and structure * * of noncommutative 3-spheres * alain connesihes.fr ] and michel dubois-violetteth.u-psud.fr + ] * msc ( 2000 ) * : 58b34 , 53c35 , 14h52 , 33e05,11f11 . +
* keywords * : noncommutative geometry , symmetric spaces , elliptic curves , elliptic functions , modular forms .
lpt - orsay 03 - 34 and ihes / m/03/56 = 0.7 cm |
the first passage of a stochastic process across a certain , pre - set value renders vital information on the underlying dynamics @xcite .
thus , it quantifies how long it takes a share to cross a given price threshold in the stock exchange .
one of the famed historical versions of such a first passage problem is the pascal - huygens gambler s ruin , i.e. , the number of rounds of a game it takes until the first gambler goes broke . for a particle diffusing in space ,
one is normally interested in the time it takes the particle to reach a given position after its initial release at some other position . here
we pursue the question of how we can optimise the first passage of a particle from point o to x , when the values of the potential @xmath0 is different at these two points .
it was previously shown that for a brownian particle the mean first passage time can be significantly reduced in a piecewise linear potential when the particle first has to cross a large potential barrier located close to its starting point and in exchange experiences a drift towards the target for the remaining part of its trajectory @xcite .
can similar effects be observed when instead of a brownian particle we consider a particle in a strongly disordered environment ? to answer this question we study a particle that performs anomalous diffusion @xcite @xmath1 with anomalous diffusion exponent @xmath2 and the generalised diffusion coefficient @xmath3 of physical dimension @xmath4=\mathrm{cm}^2/\mathrm { sec}^{\alpha}$ ] .
microscopically , we assume that the particle follows a scher - montroll continuous time random walk ( ctrw ) , in which successive jumps of the particle are separated by independent , random waiting times @xmath5 with power - law distribution , @xmath6 where @xmath7 is a scaling factor of physical dimension of time , such that no characteristic time scale @xmath8 exists @xcite .
realisations of ctrw subdiffusion were observed in a variety of systems including charge carrier diffusion in amorphous semiconductors @xcite , the motion of submicron particles in living biological cells @xcite , the dynamics of tracer beads in an actin mesh @xcite , or the motion of functionalised colloidal particles along a complementary , functionalised surface @xcite . as we show here based on analytical calculations and numerical analyses
the introduction of a piecewise linear potential indeed renders the first passage behaviour of subdiffusive processes more efficient .
this is demonstrated in terms of the density of first passage times and the associated survival probability , as well as a recently defined efficiency parameter .
we discuss potential applications of our findings to the translocation of polymers through narrow channels .
we assume that the particle starts at point o and diffuses until it reaches the point x located at @xmath9 in normalised units . on its way it passes through a piecewise linear potential with a change of slope at point a at @xmath10 ( see fig .
[ sketch ] ) .
we denote the values of potential at these points as @xmath11 , @xmath12 , and @xmath13 .
thus , by help of thermal fluctuations the particle first crosses the potential maximum at point a before being advected towards the target x for the rest of the way . at the starting point
o a reflecting boundary condition is assumed , while at the target we implement an absorbing boundary to account for the first passage problem @xcite .
the basis for the analytical description of this subdiffusion problem with given distribution ( [ wtd ] ) of waiting times @xmath5 in the long - time limit @xmath14 is given by the fractional fokker - planck equation @xcite which we here write in the integral form @xmath15 where @xmath16 is the initial condition , @xmath17 is the derivative of the external potential , @xmath18 is the particle mass , and @xmath19 the friction experienced by the particle .
the riemann - liouville fractional integral is defined in terms of @xmath20 representing a laplace convolution . in the brownian limit @xmath21 , eq .
( [ fpe ] ) reduces to the regular fokker - planck equation . for segments with a linear potential in our piecewise linear form
@xmath22 the fractional fokker - planck equation reduces to @xmath23 where @xmath24 corresponds to the two different slopes of the piecewise linear potential . in our choice of @xmath25 the @xmath26
correspond to drift velocities , as the dimension of @xmath19 is that of @xmath27=\mathrm{sec}^{\alpha -2}$ ] @xcite .
the solution of this equation with a reflective boundary condition at one end and an absorbing boundary at the other can be found by methods similar to the brownian case , compare ref . @xcite . if both boundary and initial conditions are set as vanishing probability at @xmath9 , @xmath28 , for the absorbing boundary , the flux condition at the origin @xmath29 of the form @xmath30 , and the initial condition @xmath31 @xcite , then @xmath32 applying the laplace transform @xmath33 to eq .
( [ ffpe1 ] ) , we find the ordinary differential equation @xmath34 the solution of this equation has the form @xmath35 with the exponents @xmath36 the coefficients @xmath37 and @xmath38 are determined by the boundary conditions @xmath28 and @xmath39 as well as by the continuity of flux and distribution @xmath40 at @xmath41 .
this produces the following system of linear equations , @xmath42 due to the divergence of the characteristic waiting time @xmath8 for ctrw subdiffusion processes , even in confined geometries no mean first passage time exists @xcite .
below we will therefore analyse the average time for the 50% or 90% probability that the particle has arrived in x. analytically , the relevant quantity for this type of process is the probability density of first arrival , @xmath43 , or the cumulative survival probability , @xmath44 .
both quantities are related through @xmath45 @xcite . in our case of the absorbing boundary condition at x we obtain the first passage density in terms of the flux at @xmath46 ( @xmath47 in our units ) . in laplace space , @xmath48 in terms of the exponents and coefficients defined in eqs .
( [ exponents ] ) and ( [ system ] ) .
we note that for ctrw subdiffusion any process described by the fractional fokker - planck equation ( [ ffpe ] ) can be related to its brownian counterpart simply by the method of subordination , i.e. , a transformation of the number of steps to the process time . for the first passage process this subordination corresponds to the laplace space rescaling @xcite @xmath49 where the factor @xmath50 takes care of the dimensionality : @xmath51=\mathrm{sec}^{\alpha}$ ] . from the above expressions in laplace space we now perform a numerical laplace inversion @xcite and compare the obtained results to simulations of the ctrw process in the external , piecewise linear potential @xmath25 .
the green curves ( top curve on the left , bottom curve on the right ) correspond to the analytical result for normal diffusion ( @xmath21 ) , see ref .
the red ( centre ) curves and triangles represent the numerical laplace inversion of eq .
( [ flux ] ) and simulations results for @xmath52 , respectively . the blue curves and squares
stand for @xmath53 .
lattice size for the ctrw was @xmath54 and @xmath55 .
each symbol represents @xmath56 simulation runs.,width=15 ] the monte carlo simulations of the ctrw process were performed on a lattice with @xmath54 points and the waiting times in between successive jumps were drawn from a waiting time with asymptotic power - law decay , @xmath57 with @xmath2 , for details see ref .
@xcite . in the chosen units
the length of the interval @xmath58 is @xmath59 , and thus the lattice constant is @xmath60 . for comparison with the simulations we make use of the explicit derivation of the ffpe @xcite , such that in the limit @xmath61 and @xmath62 we have @xmath63 is the boltzmann factor and @xmath64 the ( monte - carlo simulations ) temperature .
for simplicity we use @xmath65 .
results are shown in fig .
[ t50t90 ] for the 50% and 90% probability of particle absorption at the target point x. for each case we show results for the cases @xmath66 and @xmath67 , as well as include the analytical result for the brownian case from ref . @xcite . in these simulations
the potential maximum was placed at @xmath55 . moreover , the potential at the starting and end points was chosen as zero : @xmath68 .
the lines for the subdiffusive cases were obtained from numerical laplace inversion of eq .
( [ flux ] ) and subsequent integration such that the plotted times @xmath69 and @xmath70 are implicitly defined through the integral @xmath71 , and analogously @xmath72 for @xmath70 .
the symbols are obtained from the monte carlo simulations .
[ t50t90 ] shows some remarkable properties .
thus , for the case of the 90% probability the brownian case exhibits the shortest absorption times , and the subdiffusive cases with @xmath67 and @xmath66 become increasingly slower . this result would be naively expected .
however , for the 50% probability the behaviour is exactly opposite , i.e. , the 50% first passage is fastest for the most pronounced subdiffusion .
this effect is due to the fact that one - sided stable distributions , to which our waiting time distribution @xmath73 belongs , have long power - law tails , but are also more concentrated around the origin at @xmath74 .
thus , if we cut off extremely long waiting times governed by the long tail of @xmath73 ( particles that do not arrive up to @xmath69 ) , we actually observe that the resulting process becomes faster for decreasing @xmath75 . at 90% probability
this trend is inverted , as the statistics include sufficiently many long(er ) waiting times . for the brownian case and stable exponents @xmath53 and @xmath52 .
while for @xmath21 the decay of @xmath76 is exponential , for the subdiffusive cases a power - law behaviour is observed as @xmath77 . before the crossover at @xmath78 the decay of the subdiffusive particles is faster than the brownian particle .
parameters : @xmath79 and @xmath55.,width=10 ] the second important observation is that the minima of all first passage time curves in fig .
[ t50t90 ] are located at @xmath80 for all @xmath75 as well as for 50% and 90% probability .
thus , if a certain value @xmath12 optimises the first passage behaviour for a brownian particle , it also optimises the corresponding subdiffusive dynamics .
these findings are corroborated by the functional behaviour of the survival probability @xmath81 , as shown in fig .
[ survival ] . at smaller times , corresponding to a smaller percentage of the probability of first passage , indeed the decay is faster for more pronounced subdiffusion and slowest for normal diffusion .
approximately at unit time @xmath82 a crossover is observed , and for longer times we find the naively expected behaviour : brownian motion effects the fastest decay while the subdiffusive motion is slower . at long times
@xmath83 the survival probability for the subdiffusive cases exhibits the power law @xmath84 , compare refs . @xcite .
this follows directly from the subordinated exponential decay of brownian motion also shown in fig . [ survival ] .
on the value @xmath12 of the potential at the maximum point a , for different positions of point a , @xmath55 ( left , asymmetric potential ) and @xmath85 cm ( right , symmetric potential).,width=16 ] yet another way to characterise the first passage behaviour is to evaluate the average rate @xmath86 , which was introduced for superdiffusive search as a non - diverging measure of optimisation @xcite .
this quantity can be computed conveniently numerically via the relation @xmath87 from the laplace space result of the first passage density @xmath43 .
the functional form of this quantity shows another remarkable property : even in the absence of a potential the efficiency @xmath88 increases with decreasing stable exponent @xmath75 .
namely , using the known result for normal brownian diffusion @xcite , from subordination we find that for @xmath89 @xmath90 therefore , @xmath91 where @xmath92 is a constant .
in the limit @xmath93 this expression reduces to @xmath94 which proves the fact that the average rate increases with the decrease of the stable exponent @xmath75 . in the case of the piecewise linear potential this behaviour
is indeed preserved , as shown in fig .
[ effx01x05 ] .
the figure also demonstrates that the maxima of the efficiency are consistent with the minima of the 50% and 90% probability first passage times , @xmath69 and @xmath70 shown in fig .
[ t50t90 ] . note that even in the case of a symmetric potential shown on the right of fig .
[ effx01x05 ] the first passage dynamics profits from the existence of the energy landscape : after crossing the peak the diffusion back towards the starting point o is suppressed . as function of the position @xmath10 of the potential maximum with fixed value @xmath12 .
note the crossover at small values of @xmath10 .
a higher potential peak @xmath12 facilitates the first passage as long as the peak location @xmath10 remains close to the starting point o. the dotted line indicates the efficiency value in absence of an external potential , i.e. , @xmath95.,width=597 ] the dependence of the efficiency @xmath88 on the position @xmath10 of the potential maximum is displayed in fig .
[ effvarxa ] , for different @xmath12 values of the potential ( as indicated in the panels ) .
consistently for all @xmath75 the efficiency increases with growing value @xmath12 when @xmath10 shifts towards the starting point o : the particle requires a larger thermal fluctuation to cross the initial peak @xmath12 , but then experiences a higher , constant drift velocity towards its target x. at growing values for @xmath10 we observe a crossover , and then a higher peak value @xmath12 effects lower efficiency . what would happen if we let the particles age before sending them on their journey from o to x ?
ageing is a characteristic property of subdiffusive ctrw particles : the process is highly non - stationary , and physical observables strongly depend on the ageing time @xmath96 elapsing between system initiation and start of the measurement @xcite . due to
the scale - free form of the waiting time distribution with diverging mean waiting time @xmath97 , longer and longer waiting times occur on average , such that effectively the particle is constantly slowing down .
physically , in the picture of a random energy landscape this effect corresponds to the situation that the particle discovers deeper and deeper traps on its path . after passing the ageing period , the first step of the particle then corresponds to the forward waiting time @xcite @xmath98 for the parameters
@xmath99 and @xmath100 we show the dependence of the efficiency @xmath88 on the ageing time @xmath96 of the process for the subdiffusive cases with @xmath66 and @xmath67 .
indeed , the continued slow - down of the motion due to increase of the typical waiting times leads to a pronounced decrease of @xmath88 of the power - law form @xmath101 corresponding to the product of the first passage density without ageing ( @xmath102 ) and the probability density of the forward waiting time @xcite .
this behaviour is indeed nicely observed in our simulations , as demonstrated in fig .
[ aging ] . on the ageing time @xmath103 elapsing between initiation of the process and start of our observation of the particle at @xmath104 , for @xmath53 ( left ) and @xmath52 ( right ) .
we observe that @xmath105 , as predicted by eq .
( [ e_age ] ) .
parameters : @xmath99 and @xmath100.,width=16 ]
the study of the mean first passage behaviour in finite , flat potential landscapes has experienced remarkable progress during the last few years @xcite . in particular , the distinction of compact versus non - compact exploration strategies led to the concept of `` geometry - controlled kinetics '' @xcite .
moreover , the trajectory - to - trajectory variation of first passage times have been studied on finite domains of various shapes recently @xcite .
much less is known about the first passage in potential landscapes .
previously we found that for normal brownian diffusion the first passage in a finite interval can be sped up significantly by introducing a potential landscape @xcite .
for the case of a piecewise linear potential , the role of the potential is intuitively clear : after crossing an initial potential barrier , the particle continuously slides down towards the target .
when the position of the barrier successively approaches the point of release and the barrier height increases , the mean first passage time decreases @xcite .
here we studied the behaviour in such a piecewise linear potential for the case of a subdiffusing particle whose dynamics is governed by a long - tailed waiting time distribution with a diverging characteristic waiting time . from analysis of the probability percentage of the first passage , the survival probability , and the efficiency parameter we found a number of remarkable properties .
namely , as in the brownian case the introduction of the barrier indeed leads to a more efficient ( i.e. , faster ) first passage behaviour for a given value @xmath75 of the anomalous diffusion exponent . at short times , somewhat surprisingly the more subdiffusive particle performs better than the brownian particle , while at longer times the less subdiffusive particle wins out .
the optimal height of the potential barrier is thereby conserved for varying @xmath75 . moreover , the efficiency increases dramatically when the position of the potential maximum is shifted towards the particle origin o. finally , the ageing of the particle leads to a power - law decay of the efficiency as function of the ageing time . as in the brownian case @xcite a faster first passage of particles through the interval @xmath58 in the presence of the potential barrier does not contradict the corresponding result of the ( fractional ) kramers escape @xcite .
the distribution of escape times for ctrw subdiffusion becomes @xcite @xmath106 where @xmath107 is the mittag - leffler function @xmath108 with @xmath109 at @xmath110 .
the ( fractional ) rate coefficient @xmath111 is given by @xmath112 thus shifting the maximum at @xmath10 towards point o leads to an increase of the curvatures at the maximum and minimum positions in eq .
( [ kramers ] ) , and hence speeds up the rate in the same way as observed in ref .
@xcite , if @xmath12 is kept constant .
let us briefly discuss a technologically relevant , physical application of the above results .
namely , we consider the passage of a polymer chain through a narrow channel , the so - called translocation process @xcite . indeed , in terms of the translocation co - ordinate @xmath18 ( the number of monomers crossing the exit of the channel ) the translocation becomes subdiffusive , see , for instance , refs .
the fractional fokker - planck equation was proposed to model this subdiffusive behaviour , and shown to capture the first passage behaviour of the translocating polymer chain in comparison to simulations @xcite .
more recently , it has become clear that the translocation of a polymer through a narrow channel without interactions between channel wall and polymer chain is stochastically described by fractional langevin equation motion driven by long - ranged gaussian noise @xcite .
however , the long - time translocation dynamics of a polymer may still be governed by the fractional fokker - planck equation if the motion of the chain is successively immobilised with power - law distributed waiting times due to monomer - channel interactions or extra - channel inhibitors .
the translocation time of a polymer through a channel consists of three distinct contributions @xcite : ( i ) the time needed for the free polymer chain on the cis side of the channel to diffuse to the channel entrance , ( ii ) the time for one of the chain ends to thread into the channel entrance , and ( iii ) the passage of the chain through the channel across the entropic potential describing the reduction of the polymer s accessible degrees of freedom due to the imposed constraints plus some external driving potential . with the translocation co - ordinate @xmath18 for a polymer with @xmath113 monomers ,
this gives rise to the free energy landscape @xcite @xmath114 \right ) , \label{entropy}\ ] ] where @xmath115 is the ( non - universal ) lattice connectivity ( e.g. , @xmath116 on a cubic lattice ) and @xmath117 is the topological critical exponent for a self - avoiding chain attached to a wall , @xmath118 for a self - avoiding chain in three dimensions .
for translocation processes only the second term of the free energy function ( [ entropy ] ) depends on the translocation co - ordinate @xmath18 and thus matters to our analysis .
the entropic free energy barrier in eq .
( [ entropy ] ) clearly slows down the translocation of the chain . in the mathematical description in the pseudo - equilibrium approximation @xcite , we are only interested in the very translocation dynamics of the chain , and we therefore impose a reflecting boundary condition at @xmath119 , in order to prevent the chain from escaping the channel .
what happens when instead of a homogeneous polymer chain we consider a gradient or tapered copolymer with a a sequence of monomers of different types with different monomer - channel interactions ?
gradient copolymers show a quite special behaviour with respect to their thermodymamic properties , contrasting other copolymer sequences @xcite .
the effects of interactions between translocating chain and channel were indeed studied previously for the passage of heteropolymers refs .
@xcite . to demonstrate the possibility of modifying the translocation time statistics for such inhomogeneous polymer chains we imagine a polymer sequence , that gives rise to a piecewise linear potential of the shape studied above . combining this with the translocation free energy ( [ entropy ] )
, we obtain the landscape portrayed in fig .
[ landscapetrans ] .
simulations of this modified translocation process produce the 50% and 90% probabilities for translocation across the combined free energy landscape shown in fig .
[ translocation ] .
evidently the piecewise linear potential can lower the translocation time .
as above , the position of the optimum ( at around @xmath120 , different from the value without the entropic potential ) is independent of the exponent @xmath75 and whether we consider the 50% or 90% case .
notably , while the overall translocation times are higher in the 90% case , the effect of the potential is strongest for @xmath66 . for a chain with @xmath54 monomers .
the maximum of the piecewise linear potential component is at @xmath121 .
black triangles show the simulations results for @xmath53 , blue squares for @xmath52 , and red circles for the brownian case .
@xmath56 simulation runs were performed to obtain the data points.,width=16 ] thus the sequence of the copolymer may have a significant influence on the translocation times .
this statement itself was already confirmed by monte - carlo simulations ref .
@xcite , in which the authors found a distinct variation of up to 3 orders of magnitude in translocation times for different sequences but same composition .
however , they did not report the effect of decreased translocation times in comparison with the homopolymer by adding slower translocating segments .
we mention that an essential part of this brief discussion relies on the quasi - equilibrium approximation , necessary to apply a free energy picture @xcite , which was shown to be inapplicable in many cases of translocation @xcite .
however , we assume that these effects on polymer translocation times will remain relevant for non - equilibrium situations , as well .
it is feasible that parts of biopolymer sequences may have gradient or tapered parts which assist translocation .
these specific interaction potentials may also originate from combination of complex pore structure and copolymer sequence as in refs .
further studies of this problem are expected to shed new light on the biophysics of translocation processes .
vvp acknowledges financial support from the deutsche forschungsgemeinschaft ( project pa2042/1 - 1 ) as well as discussions with j. schulz about the simulation of random variables .
rm acknowledges support from the academy of finland within the fidipro scheme .
99 s. m. a. tabei , s. burov , h. y. kim , a. kuznetsov , t. huynh , j. jureller , l. h. philipson , a. r. dinner , and n. f. scherer , proc .
usa * 110 * , 4911 ( 2013 ) .
jeon , v. tejedor , s. burov , e. barkai , c. selhuber - unkel , k. berg - srensen , l. oddershede , and r. metzler , phys .
lett . * 106 * , 048103 ( 2011 ) . for numerical laplace inversion
the additional package by p. p. valk and j. abate was used in mathematica : http://library.wolfram.com/infocenter/demos/4738/ ; p.p .
valk and j. abate , comput .
appl . , * 48 * , 629 ( 2004 ) .
bouchaud , j. phys .
i ( paris ) * 2 * , 1705 ( 1992 ) .
+ c. godrche and j. m. luck , j. stat . phys . * 104 * , 489 ( 2001 ) .
+ e. barkai , phys .
* 90 * 104101 ( 2003 ) .
+ e. barkai , y .- c .
cheng , j. chem .
phys . * 118 * 6167 ( 2003 ) .
+ j. schulz , e. barkai , and r. metzler , phys .
lett . * 110 * , 020602 ( 2013 ) ; e - print arxiv:1310.1058 .
j. l. a. dubbeldam , a. milchev , v. g. rostiashvili , and t. a. vilgis , phys .
e,*76 * 010801(r ) ( 2007 ) .
+ j. l. a. dubbeldam , a. milchev , v. g. rostiashvili , and t. a. vilgis , epl , * 79 * 18002 ( 2007 ) .
+ r. metzler and j. klafter , biophys .
j. * 85 * , 2776 ( 2003 ) .
cohen , a. chandhuri and r. golestanian , phys .
x , * 2 * , 021002 ( 2012 ) .
+ r. h. abdolvahab , r. metzler , and m. r. ejtehadi , j. chem .
phys . * 135 * , 245102 ( 2011 ) .
+ k. luo , t. ala - nissila , s .- c .
ying and a. bhattacharya , j. chem .
phys . * 126 * , 145101 ( 2007 ) . | we show that for a subdiffusive continuous time random walk with scale - free waiting time distribution the first - passage dynamics on a finite interval can be optimised by introduction of a piecewise linear potential barrier .
analytical results for the survival probability and first - passage density based on the fractional fokker - planck equation are shown to agree well with monte carlo simulations results . as an application we discus an improved design for efficient translocation of gradient copolymers compared to homopolymer translocation in a quasi - equilibrium approximation . |
over the past few years , a number of papers have revisited the issue of the non - universality of the stellar initial mass function ( imf ) in the unresolved stellar populations of early - type galaxies ( etgs ) . at present
, there are three independent probes of the imf in the unresolved , nearly quiescent populations of etgs , based on the modelling of the kinematics ( e.g. @xcite ) ; constraining the mass in strong gravitational lenses that probe the projected mass distribution in the central regions of galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or via gravity - sensitive spectroscopic features ( e.g. @xcite ) .
the first two probes are sensitive to the stellar m / l , and their major systematic relates to the contribution of the dark matter halo in the central regions of galaxies .
spectroscopic constraints are mostly sensitive to the ratio between giant and dwarf stars , and its main systematic is the degeneracy with the additional parameters of the underlying stellar populations , most notably the effect of variations in the individual elemental abundances .
we focus here on this issue .
a large effort has been made to understand this systematic in detail .
the bottleneck in the analysis is the lack of robust models of stellar atmospheres , and their response to abundance effects . on this front ,
population synthesis modellers follow either a theoretical approach ( e.g. , * ? ? ? * ) or an empirical methodology ( e.g. , * ? ? ?
the theoretical approach relies on predictions from stellar atmosphere models to construct stellar population models that best match the observed spectra , while the empirical approach targets the deviation of the data from solar - scaled stellar population models to measure the abundance effects . at present ,
both methods are needed to be able to improve our understanding of line strengths in unresolved stellar populations . on the issue of imf variations ,
a number of papers have recently strengthened the idea that etgs feature a non - universal imf ( see @xcite , as well as ( * ? ? ?
* ; * ? ? ?
* hereafter fer13 and lb13 , respectively ) ) .
furthermore , the analysis has gone as far as providing tentative drivers of imf variations , suggesting either [ mg / fe ] @xcite , velocity dispersion ( fer13 , lb13 , @xcite ) , metallicity @xcite , or more complex physical observables @xcite .
studies combining galaxy dynamics and spectroscopic features of the same galaxies have revealed potentially serious systematics @xcite , although such effects could be strongly model - dependent @xcite .
in addition , constraints based on strong gravitational lensing are inconclusive : while @xcite find agreement with the dynamical modelling , several massive lensing etgs feature stellar m / l values consistent with a kroupa - like imf @xcite .
critics have voiced concerns on the interpretation of the line strength variations as an imf trend , proposing instead large departures from the solar - scaled abundance of individual elements such as sodium @xcite .
it is therefore imperative to understand in detail the potential systematics ( most notably , the effect of elemental abundances , age , and metallicity ) . on the spectroscopic analysis side the focus of this paper
it is necessary to put under the microscope several spectral features , ideally with varying levels of dependence on the parameters under scrutiny .
the main target of the present work is a luminous etg , hereafter xsg1 , selected from the spider survey @xcite , more specifically from the subsample of @xmath6 etgs with @xmath7 kms@xmath2 defined by lb13 .
we have gathered new , very deep , `` high '' resolution ( @xmath8 ) , long - slit spectroscopy for xsg1 with the x - shooter spectrograph on the eso - very large telescope ( vlt ) .
the combination of depth , resolution , and wavelength range for xsg1 gives us a unique opportunity to single out the effect of the imf against the other stellar population parameters ( in particular abundance ratios ) , as a function of galacto - centric distance , advancing our previous work ( * ? ? ?
* hereafter mn15a ) . to this effect , we focus on the optical , tio - based , spectral features , combining them with the near - infrared tio feature , @xmath9 , at @xmath108860 , and the wing - ford band , @xmath11 ( @xmath109915 ) .
the layout of the paper is as follows . in sec .
[ sec : obs ] we describe the observations and instrumental setup .
[ sec : red ] deals with the data reduction .
the analysis is detailed in sec .
[ sec : analysis ] , including a description of the radial binning of the data ( sec .
[ sec : kin ] ) , the stellar population models used ( sec . [ sec : models ] ) , the determination of stellar age ( sec . [ sec : age_z ] ) , and abundance - ratio ( sec . [ sec : profs_xfe ] ) profiles , as well as the method to constrain the imf ( sec .
[ sec : method ] ) .
[ sec : res ] presents the main results of the work , i.e. the radial trends of imf- and abundance - sensitive features in xsg1 , for tio features ( sec .
[ sec : tio_profs ] ) and the wing - ford band ( sec .
[ sec : feh ] ) , the inferred radial trend of imf slope ( sec .
[ sec : gamma_trends ] ) , as well as constraints to the radial variation of imf normalization , via a comparison of the stellar population analysis with dynamics ( sec .
[ sec : alpha_trends ] ) .
a discussion is presented in sec .
[ sec : disc ] , followed by a summary in sec .
[ sec : summary ] .
the main target of the present work is xsg1 ( sdss j142940.63 + 002159.0 ) , a massive ( @xmath12 ; @xmath13kms@xmath2 ) etg at redshift @xmath14 , taken from the spider survey @xcite . the galaxy has been selected from the sample of lb13 to have a high abundance ratio ( @xmath0$]@xmath15 ) .
[ fig : image ] shows the sdss r - band image of xsg1 , together with the residual map obtained by fitting the galaxy surface brightness distribution with a two - dimensional , psf - convolved , srsic model , using the software 2dphot @xcite .
xsg1 is a round object ( @xmath16 ) , whose light distribution is well described by a de vaucouleurs profile , with best - fitting srsic @xmath17 , and an effective radius of @xmath18@xmath19 ( see the fig . [
fig : image ] ) , consistent with the sdss estimate of @xmath20 ( based on a de vaucouleurs model ) . we have obtained new deep long - slit spectroscopy for xsg1 with the x - shooter spectrograph at the eso - vlt , on cerro paranal ( proposal ids : 092.b-0378 , 094.b-0747 ; pi : flb ) .
x - shooter is a second - generation eso - vlt instrument a slit echelle spectrograph that covers a wide spectral range ( 300025000 ) , at relatively high resolution @xcite .
the incoming bin is split into three independent arms ( ultraviolet - blue , uvb : 30005900 ; visible , vis : 530010200 ; near - infrared , nir : 980025000 ) , each one equipped with optimized optics , dispersive elements , and detectors , ensuring high sensitivity throughout the entire spectral range .
observations were done through five observing blocks ( obs ) , each including two exposures on target , interspersed by two sky exposures ( through an object - sky - sky - object sequence ) , with the same integration time as for the science target . for each ob
, the on - target exposure time was 20.4 , 22.6 , and 25 min , resulting into a total exposure time of @xmath41.7 , 1.9 , and @xmath21 hr , in the x - shooter uvb , vis , and nir arms , respectively .
the observations were carried out in service mode during two consecutive nights at the end of march 2014 ( 4 out of 5 obs ) , and completed at the end of february 2015 .
all data were taken under photometric conditions , with seeing fwhm varying from @xmath22 to @xmath23arcsec in the optical , with a mean value of @xmath24arcsec .
the x - shooter slit is 11 long , with a spatial scale of 0.16@xmath25 in the uvb and vis , and 0.21@xmath26 in the nir , arms .
we adopted an instrument setup with 0.9- , 0.9- , and 1.0- wide slits , providing a typical resolution power of @xmath27 , @xmath28 , and @xmath29 , in the uvb , vis , and nir arms , respectively .
the pointing was nodded by a few pixels along the slit among exposures , in order to perform an optimal removal of cosmic rays and detector defects ( e.g. hot columns / pixels ) during data reduction . a hot ( white dwarf ) spectro - photometric standard star , together with a few ( 37 ) telluric ( b - type ) standards were observed during each night ( the latter taken before and after each science ob ) in order to perform flux calibration and correct the spectra for telluric lines , respectively . as part of the same observing campaign as xsg1 ,
we have also observed a second target , a massive ( @xmath30 ; @xmath31kms@xmath2 ) etg ( hereafter xsg2 ; sdss j002819.30 - 001446.7 ; z=@xmath32 ; @xmath33 ) also selected from the spider survey , with a more typical @xmath0 $ ] ( @xmath34 dex ) for its velocity dispersion .
xsg2 has been observed with xshooter ( proposal i d : 094.b-0747 ; pi : flb ) through a series of 10 obs , each consisting of an object - sky - object sequence .
this different strategy , resulting into half of the total integration time spent on sky , allowed us to perform shorter individual exposures , in order to test the quality of sky subtraction in the nir , where the fast variability of sky lines may hamper the analysis of the spectra , especially in the outer , low surface - brightness regions .
the data of xsg2 will be the subject of a forthcoming paper . here , we only use the central spectrum of xsg2 to complement the analysis of the wing - ford band for xsg1 , as detailed in sec .
[ sec : feh ] .
the echelle layout of x - shooter produces 1216 curved and highly distorted spectral orders per arm , requiring a complex reduction process .
we refer the reader to @xcite for a detailed description of all issues related to the data reduction of x - shooter uvb- and vis - arm data . for our purposes
, we performed the data reduction using version 2.4.0 of the x - shooter data - reduction pipeline @xcite , optimized with dedicated fortran software developed by the authors .
we ran the x - shooter pipeline in `` physical '' mode , through the eso command - line utility esorex .
for each arm , the data ( i.e. science and sky , as well as standard star frames ) were entirely pre - reduced with the x - shooter pipeline , including de - biasing , correction for dark frames ( nir - arm only ) , rectification of echelle orders , wavelength - calibration , and merging of the orders .
the remaining reduction steps were performed as follows : _ flux calibration _ - : : for the uvb and vis arms , we found significant differences among reference spectra of standard stars to the flux - calibrated ones computed by the pipeline , being as high as 510@xmath35 in regions affected by strong telluric lines ( for the vis arm ) .
for this reason , we applied an iterative procedure to improve the response function of the pipeline .
the spectrum of each flux - calibrated standard star is corrected for tellurics with the same procedure as for the science data ( see below ) .
then , for each echelle order , we derive the ratio of the corrected standard - star spectrum to its tabulated version .
the ratio is fitted , order - by - order , with legendre polynomials of suitable orders ( from one to twelve , depending on the spectral order ) , and the best - fitting polynomials are used to update the response function ( for each observing night ) . the procedure is iterated a few times , until a variation below 1@xmath35 ( the typical accuracy of our flux calibration ) is achieved in the updated response function , at all wavelengths . for the nir arm , the spectrum of the flux - calibrated standard - star , produced by the pipeline , turned out to have large deviations ( up to @xmath36 ) with respect to the tabulated flux values .
these deviations , with typical scales of @xmath37 , are likely due to residual sky - emission and telluric lines affecting the cubic - spline fit performed by the pipeline when deriving the response function . therefore , we applied the flux - calibration of the nir arm with a dedicated procedure . for each observing night , the ( one - dimensional ) non - calibrated standard - star spectrum is corrected for tellurics with the software molecfit ( see below ) .
intrinsic ( mostly hydrogen ) absorption lines , from the stellar atmospheres , are removed , by fitting them with a linear combination of multiple gaussian functions .
we compute the ratio of the resulting spectrum to the tabulated one by carefully masking out regions with strong telluric absorption ( @xmath38 ) and removing pixels with spurious spikes ( e.g. from cosmic rays or residuals of sky lines ) , through a 2-@xmath39 clipping procedure .
the resulting ratio is fitted with a second - order spline to produce an initial response function .
this response is applied to each telluric standard - star observed during the night .
the continuum of the telluric standard is modelled , and removed , with two black - body ( bb ) laws , one for bluer and another one for redder wavelengths with respect to the brackett jump ( @xmath40 ) .
the temperatures of the two bbs are obtained by a best - fitting procedure .
the resulting signal , median - smoothed , and averaged over all telluric - standards observed during the night , was applied as a correction to the initial response , to derive the final response function . comparing the central spectrum of xsg1 among different exposures / nights , we have estimated that our nir flux - calibration is accurate at @xmath41 .
_ sky subtraction _ - : : our targets with an effective radius around @xmath42 fill the whole x - shooter 11 slit .
therefore , sky subtraction can not be performed in the standard way of interpolating the sky from the two sides of the slit .
we performed sky subtraction with the software skycorr @xcite .
given the two - dimensional spectrum of the object , @xmath43 , we consider the nearest sky frame , @xmath44 .
we extract a one - dimensional ( 1d ) spectrum from both @xmath44 and @xmath43 , by median - combining each frame over an 8-pixel region , using four pixels from each edge of the slit ( where the signal from the galaxy is less prominent ) .
we run skycorr to determine the optimal scaling factors ( depending on wavelength ) that match the 1d spectrum of the sky to that of the object frames .
the scaling factors are then applied to rescale @xmath44 , and subtract it from @xmath43 .
_ telluric correction _ - : : telluric lines were removed from each object exposure individually . for the vis arm , we adopted two independent approaches , by ( i ) using telluric standard stars observed during our programme , and ( ii ) using the software molecfit @xcite . in case ( i )
, we obtained , for each telluric standard star , an atmospheric transmission curve , by removing intrinsic absorption lines and the continuum from the stellar spectrum .
the correction itself was performed by deriving the linear combination of ( re - scaled ) transmission curves that minimizes the absolute deviation to the object spectrum , computed over spectral regions with prominent telluric lines . in case ( ii )
, molecfit was run to compute a theoretical transmission model from the object spectrum itself . in general , both approaches were found to provide very consistent results ( with differences below @xmath45 ) , as also shown in the analysis of line strengths throughout the present work . for the nir arm ,
given the excellent quality of molecfit fits , we decided to rely only on this method . _
combination _ - : : to combine different object exposures , we first aligned the photometric centre of the galaxy as a function of wavelength in each two - dimensional spectrum , in order to remove its spatial drift due to the atmospheric dispersion . to this effect , we collapsed the object spectrum along the slit direction in wavelength bins ( with size @xmath46 ) , and derived the photometric centre , for each bin , by fitting the collapsed profile with a linear combination of gaussian functions .
the aligned 2d spectra were then averaged with the iraf tool _
imcombine_. cosmic rays , as well as pixels flagged by the x - shooter pipeline were excluded from the average .
a noise map was propagated accordingly .
black and grey curves correspond to radial binned spectra , along the x - shooter slit , with minimum s / n ratio of 30 and 15 , respectively .
error bars denote @xmath47 uncertainties .
notice that the galaxy has low rotation , with a radial drop in velocity dispersion ( from @xmath48kms@xmath2 in the centre , to @xmath49kms@xmath2 at 1@xmath18 ) .
the inset in the top panel compares the observed @xmath50 ( empty circles ) with best - fitting results from the jeans anisotropic modelling method ( jam , red curve in the inset ; * ? ? ?
, width=309 ] in order to compute the kinematics of xsg1 , we binned the x - shooter two - dimensional spectrum along the spatial direction , starting from the photometric centre of the galaxy .
we chose radial bins with a minimum width of 0.22
( i.e. about 1/4 of the seeing fwhm ) , and increased the bin width adaptively to reach a given , minimum ( median ) signal - to - noise ratio , @xmath51 ( computed per , at @xmath52 ) , in each bin .
this procedure allowed us to reach an outermost radius of @xmath53 ( @xmath54 ) @xmath18 for @xmath55 ( @xmath56 ) .
for each radial bin , we measured the recession velocity and the galaxy velocity dispersion using the software ppxf @xcite , which simultaneously fits the stellar kinematics and an optimal linear combination of spectral templates to the observed spectrum , using a maximum - likelihood approach that works in wavelength space , .
we measured the kinematics independently in different spectral regions of the uvb and vis arms ( @xmath57 ) , combining the corresponding probability distribution functions into final estimates .
from the recession velocity in the innermost bin , we obtain a redshift of z=@xmath58 for xsg1 , fully consistent with the value of z=@xmath59 from the sdss database .
[ fig : kin ] plots the profiles of rotation velocity , @xmath60 , and velocity dispersion , @xmath39 , for xsg1 , showing that the galaxy is likely a pressure - supported system , with @xmath39 decreasing smoothly from @xmath48kms@xmath2 in the centre , to @xmath49kms@xmath2 at 1@xmath18 , along both sides of xsg1 . the galaxy rotation velocity is less than @xmath61 kms@xmath2 at all radii . notice that we get very consistent kinematics when binning the spectra with either @xmath62 or @xmath56 , as shown by the black and grey lines . under the assumption of circular symmetry , the velocity dispersion profile in fig .
[ fig : kin ] implies a luminosity - weighted value @xmath63kms@xmath2 within the sdss fibre aperture radius of 1.5 , consistent with two estimates of @xmath64 and @xmath65kms@xmath2 available from the sdss database .
.relevant properties of radially binned spectra used to constrain the imf . for all bins we fold - up the spectra around the photometric centre of the galaxy . [
cols="^,^,^,^ " , ] [ tab : indices ] we analyze the x - shooter spectra with extended miles stellar population models ( @xcite ; rck et al . , 2015b , a&ar , in preparation ) , the first set of single - burst stellar population models to date that cover both the optical and infrared wavelength range , between 3500 and 50000 , based exclusively on empirical stellar spectra .
these models combine miuscat stellar population models ( @xcite ) , covering the spectral range @xmath66 at a nominal resolution of @xmath67 fwhm @xcite , with the nir model extension based on the irtf stellar library , covering the @xmath68 wavelength range at a spectral resolution of 2000 @xcite .
the miuscat and irtf models are joined in the interval @xmath69 . for @xmath70 ,
the model predictions are identical to miuscat .
the extended miles models rely on solar - scaled isochrones with stellar spectra following the abundance pattern of our galaxy , i.e. approximately solar - scaled at solar metallicity , and are derived from either padova00 or basti isochrones . for comparison with our previous works ( e.g. fer13
; lb13 ; mn15a ) , we base our analysis , throughout the present work , on padova models . because of the irtf implementation , the extended miles ssps cover a restricted parameter range than miles models , i.e. ages from @xmath71 to @xmath72gyr , and four bins of total metallicity , i.e. @xmath73=\{$ ] @xmath74 , @xmath75 , @xmath76 , @xmath77 .
the ssps are provided for several imfs , as for miles models , i.e. unimodal ( single power - law ) and bimodal ( low - mass tapered ) imfs , both characterized by their slope , @xmath78 ( unimodal ) and @xmath3 ( bimodal ) , as a single free parameter ( see , e.g. , @xcite ) .
the bimodal imfs are smoothly tapered off below a characteristic `` turnover '' mass of @xmath79m@xmath80 . for @xmath81
, the bimodal imf gives a good representation of the kroupa imf , while for @xmath82 the unimodal imf coincides with the @xcite distribution .
the lower and upper mass - cutoff of the imfs are set to @xmath83 and @xmath84m@xmath80 , respectively .
notice that younger ages , as well as lower metallicities , than those provided by extended miles models are not relevant to the study of massive etgs , while a mild extrapolation of the models to higher metallicity ( from @xmath85\sim + 0.22 $ ] up to @xmath86 ) is required to match the innermost spectrum of xsg1 ( see sec .
[ sec : age_z ] ) . as in fer13 and lb13
, we consider bimodal imf models with @xmath87 , @xmath88 , @xmath71 , @xmath89 , @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95 , while for unimodal imfs , we restrict the analysis to @xmath96 , where model predictions are safe ( see @xcite ) . to estimate the effect of abundance ratios on line strengths , we also combine the predictions of extended miles models , with those from other two sets of models , i.e. @xmath97miles @xcite and ( * ? ? ?
* hereafter cvd12 ) ssps .
the @xmath98 ssps , having a spectral resolution of 2.51 ( fwhm ) , are based on the miles library and apply corrections from theoretical models of stellar atmospheres to produce spectra of old- and intermediate - age stellar populations with varying total metallicity , imf , and @xmath0 $ ] abundance ratios .
the models are based on basti scaled - solar and @xmath97enhanced isochrones .
the cvd12 models combine optical ( miles ) and near - ir ( irtf ) empirical stellar libraries to produce integrated light spectra , at @xmath99@xmath100 m , with a resolving power of @xmath101 .
the empirical cvd12 models , computed at solar metallicity and in the age range from @xmath102 to @xmath103gyr , are complemented with synthetic ssps modelling the spectral variations due to changes in individual elemental abundances , as well as in generic @xmath97 elements .
we notice that cvd12 models are computed at fixed @xmath104 $ ] , rather than @xmath85 $ ] ( adopted in the extended- and @xmath97-miles models ) , and based on a combination of three different sets of scaled - solar isochrones .
we also emphasize that in the publicly available version of cvd12 models , ssps with varying elemental abundances are only provided for a specific age ( @xmath103gyr ) and imf ( chabrier ) , and @xmath104=0 $ ] .
because of the difference between @xmath98 as well as cvd12 models and our reference extended miles ssps ( i.e. different isochrones , metallicity range , etc ..... ) , we use these other models mostly to perform a qualitative analysis of the data , whereas the quantitative results presented in this work ( e.g. sec .
[ sec : gamma_trends ] ) are entirely based on extended miles models .
notice that the present work relies on the analysis of line - strengths , based on several index - index diagrams ( see sec .
[ sec : res ] ) . in these diagrams mostly used for illustrative purposes
all model line - strengths are computed after smoothing the models to a @xmath39 of 300 kms@xmath2 ( accounting for the intrinsic spectral resolution of the models ) , i.e. the averaged velocity dispersion ( among radial bins ) for xsg1 . in all index - index diagrams ,
observed line - strengths are corrected to a common @xmath39 of 300 kms@xmath2 .
for a given radial bin , the correction is obtained using the corresponding best - fit extended - miles ssp ( see sec . [
sec : gamma_trends ] ) .
we compute the difference of each line - strength when smoothing the model to the actual @xmath39 of the spectrum and the reference value of 300 kms@xmath2 . given the relative narrow range in velocity dispersion for xsg1 , and the fact that we mostly analyze broad features ( e.g. tio and @xmath11 ) , the correction is largely independent of the specific model ( e.g. age and metallicity ) adopted .
however , in our quantitative analysis ( i.e. when fitting model to observed line - strengths ) , we compute , for each bin , line - strengths at the original @xmath39 of the data , and smooth the models accordingly .
we prefer this approach , rather than smoothing all observed spectra to the same @xmath39 , as it extracts the maximum amount of information from the data ( see , e.g. , lb13 ) , and avoids any contamination of the relevant features , when smoothing the spectra , from nearby sky residuals .
, width=309 ] is plotted as a function of the total metallicity index @xmath105'$ ] .
the solid dots represent the observed line - strengths in xsg1 ( @xmath47 error bars ) , corrected to a common velocity dispersion of 300kms@xmath2 ( see sec .
[ sec : models ] ) .
the black and green grids correspond to kroupa - like ( @xmath3@xmath106 ) and bottom - heavy ( bimodal imf ; @xmath3=@xmath102 ) miles model predictions with varying age and metallicity , as labelled .
notice that , for the specific purpose of this plot , the grids have been linearly extrapolated from the maximum value of @xmath85$]@xmath107 in miles models , up to @xmath85$]@xmath108 .
different radial bins are shown ( inside - out ) with red - through - blue colours ( see labels on the left side of the plot ) .
the effect of emission line correction on @xmath109 is shown with magenta arrows .
, width=309 ] an accurate estimate of age and metallicity is needed to constrain the imf , because gravity - sensitive spectral features also depend , to some extent , on such parameters .
we estimate radial profiles of age and @xmath85 $ ] for xsg1 with different methods , as illustrated in fig .
[ fig : agez ] .
in particular , regarding stellar age , we explore several approaches ( see labels in the top panel of the figure ) : _ fit1 _ - : : the age is derived by performing spectral fitting over the wavelength range @xmath110 .
the fit is done individually for extended miles ssps with different choices of @xmath3 ( see sec . [
sec : models ] ) , normalizing both model and observed spectra by the median flux in a reference spectral window ( @xmath111 ) .
the black , solid line , in the figure plots the median age value , for each radial bin , among ssps with different @xmath3 .
the black error bars reflect the statistical uncertainty on the age , summed in quadrature to the standard deviation among estimates for different values of the imf slope , @xmath3 .
we find no significant age gradient throughout the galaxy .
this result is further confirmed when fitting models with extended star - formation histories to the spectra ( see fer13 and lb13 for details ) , in which case we find very small differences ( @xmath112gyr ) between luminosity- ( as well as mass- ) weighted ages with respect to the results with ssp models , meaning that at all radial bins the spectra of xsg1 are very well described by a single ssp .
this is consistent with the high @xmath0 $ ] ( implying a very short star - formation time - scale in the galaxy within the effective radius , see below ) .
the dotted and dashed black lines in the top panel of fig .
[ fig : agez ] show the results for a kroupa - like ( @xmath3@xmath106 ) and a bottom - heavier ( @xmath3@xmath113 ) imf , showing that , for our x - shooter data , the age profile does not depend on @xmath3 . on the other hand
, we found that the absolute value of the age depends on the adopted methodology , e.g. the way one normalizes the continua of observed and model spectra prior to spectral fitting .
the cyan profile in the figure ( labelled @xmath114 ) shows the case when one divides the spectra by their integrated flux in the normalization window ( see above ) , rather than the median flux ( as for the solid black curve ) .
although the age gradient does not change , we find an overall shift of @xmath42gyr in the age values .
in general , changing the setup of the spectral fitting approach ( e.g. the continuum normalization window ) , we found variations in the mean age from @xmath115 to @xmath116gyr . _
fit2 @xmath117+[\alpha / fe]$ ] _ - : : the solid green line in the top panel of fig .
[ fig : agez ] shows the spectral fitting constraints when `` removing '' the effect of @xmath118 $ ] and @xmath0 $ ] abundance ratios from the observed galaxy spectrum at each radial bin . in practice , for each radial bin , we use cvd12 ssp models to estimate the ratio of @xmath118$]- and @xmath0$]- enhanced ssps with respect to the solar - scaled ssp , and divide the observed spectrum by this ratio .
the ratio is estimated using the median values of @xmath118 $ ] and @xmath0 $ ] measured for each bin ( sec .
[ sec : profs_xfe ] ) . notice that we consider here only the effect of @xmath118 $ ] and @xmath0 $ ] as ( i ) both turn out to be enhanced in xsg1 , ( ii ) they are the leading abundances to affect the tio - based optical features analyzed in the present work ( see sec . [ sec : tio_profs ] ) , and ( iii ) cvd12 models suggest that @xmath0 $ ] and @xmath118 $ ] are the main abundance ratios whose enhancement affects the shape of the continuum , possibly leading to a biased determination of age from spectral fitting .
notice that enhancing @xmath0 $ ] tends to make a given spectrum bluer , mimicking the effect of a young age ( see , e.g. , @xcite ) . on the contrary , enhancing @xmath118 $ ] tends to counteract this effect ( i.e. reddening the spectrum ; see @xcite ) . as a result , for xsg1 , accounting for the effect of abundance ratios does not change the age determination with respect to solar - scaled models .
@xmath109 - : : we use predictions from miles ssp models to fit simultaneously the @xmath109 age - sensitive indicator @xcite and the total metallicity indicator @xmath105'$ ] @xcite .
the ews of @xmath109 are corrected for emission contamination as described in lb13 , i.e. estimating the excess of flux in the line with respect to a combination of two ssps that gives the best fit in the @xmath119 spectral region ( @xmath120 ) after excluding the trough of the absorption .
the correction to @xmath109 varies from @xmath121 in the innermost bin , to less than @xmath122 in the outermost regions .
[ fig : hbo ] shows the radial variation of @xmath109 and @xmath105'$ ] line - strengths in xsg1 .
the corrected values of @xmath109 suggest no significant radial gradient in xsg1 , as confirmed by the age trend inferred for a kroupa - like imf from the @xmath109 @xmath105'$ ] diagram ( red solid line in top panel of fig . [
fig : agez ] ) .
notice , however , that miles models predict a decrease of @xmath109 with imf slope , as seen by comparing green vs. black grids in fig .
[ fig : hbo ] , corresponding to models with @xmath3=3 and @xmath3=1.3 , respectively ( see lb13 and @xcite for more details ) . as a result , for a bottom - heavy imf
, one infers younger ages than for a kroupa - like distribution ( see blue solid line in top panel of fig .
[ fig : agez ] ) .
@xmath123 - : : by summing up the three innermost galaxy spectra of xsg1 , the median @xmath124 in the @xmath125 spectral region ( @xmath126 ) is high enough ( @xmath127 ) to constrain the age with the @xmath123 spectral index , proposed by @xcite ( see also @xcite ) .
this index is a pure age indicator , i.e. it is completely independent of metallicity ( provided that the wavelength calibration is accurate enough , as it is the case for our x - shooter data ) .
the magenta symbols in the top panel of fig .
[ fig : agez ] show the age inferred from @xmath123 when using either kroupa - like ( @xmath3=1.3 ) or bottom - heavy ( @xmath3=3 ) ssp models . the age inferred in the galaxy centre from
@xmath123 ( @xmath128gyr ) is independent of the imf , and is consistent with those obtained from spectral fitting ( @xmath129gyr , see black and cyan lines in the figure ) .
$ ] and @xmath0 $ ] abundance ratios .
these are expected to contribute the most to variations in the tio features ( see text , and fig . [
fig : tio_nocorr ] ) . the upper ( lower ) , red - through - blue , curves plot the @xmath0 $ ] ( @xmath118 $ ] ) radial trends , respectively , as labelled .
different line types correspond to different methods to estimate the abundance ratios ( see labels on the upper
right corner of the bottom panel ) .
error bars are quoted at the @xmath47 level .
, width=309 ] in summary , our analysis shows that the spectra of xsg1 feature homogeneously old stellar populations inside the effective radius , with no significant age gradient .
the absolute value of the age depends on the methodology , a well - known issue plaguing most stellar population studies ( see , e.g. , @xcite ) . throughout this work ,
we adopt as a reference the age estimates from spectral fitting ( method @xmath130 above ) , rescaled to match the @xmath123based value ( @xmath128gyr ) at the galaxy centre .
we stress that this assumption serves only for illustrative purposes ( e.g. to estimate the metallicity and abundance ratio gradients of xsg1 ) , while in practice , when constraining the imf , our fitting scheme takes into account the uncertainties on the age zero - point with an ad - hoc procedure ( see sec .
[ sec : method ] ) . the bottom panel of fig .
[ fig : agez ] shows the metallicity profile for our reference age estimates , from the @xmath105'$ ] index .
the @xmath105'$ ] is largely independent of imf slope , implying that , _ at fixed age _ , metallicity estimates are independent of the imf ( see dotted and dashed black lines in the figure ) .
notice that @xmath105'$ ] is also well - known for being independent of @xmath0 $ ] @xcite .
moreover , cvd12 models predict a small variation of @xmath105'$ ] with non - alpha elements ( e.g. @xmath131@xmath105'$]@xmath132 for @xmath131@xmath118$]@xmath133dex = 0 , and assume no effect from varying abundance ratios on the solar - scaled isochrones .
] , at @xmath134kms@xmath2and for an age of @xmath103gyr ) .
hence , after accounting for the uncertainty on the zero - point of the age , the @xmath85 $ ] can be robustly constrained from @xmath105'$ ] .
[ fig : agez ] shows that xsg1 has a negative metallicity gradient of about @xmath135dex per radial decade , a typical value in massive early - type galaxies ( e.g. @xcite ) .
deviations from the solar scale of the abundance ratios involving single elements can partly mimic ( or mask out ) the effect of a varying imf on gravity - sensitive features ( see @xcite ) . since tio features , which are the main focus of the present study ,
are mainly affected by @xmath118 $ ] and @xmath0 $ ] ( see sec .
[ sec : tio_profs ] ) , in the present section we analyze the radial gradients of @xmath118 $ ] and @xmath0 $ ] in xsg1 , pinpointing the uncertainties arising from different model assumptions .
notice that we do not aim at obtaining a precise , quantitative , determination of abundance ratios , but rather to perform a qualitative characterization of the radial behaviour of @xmath0 $ ] and @xmath118 $ ] in xsg1 .
[ fig : abund ] plots different estimates of @xmath0 $ ] and @xmath118 $ ] ( i.e. different methods / models ; see lines with different colours ) as a function of galacto - centric distance .
the solid red - through - blue line , joining filled circles , shows our solar - scaled proxy for @xmath0 $ ] , measured as a difference between the metallicity estimates derived , at fixed age , from @xmath136 and @xmath137 .
the proxy has been calibrated onto @xmath0 $ ] with the aid of ( * ? ? ?
* hereafter tmj11 ) stellar population models , as detailed in lb13 ( see also vaz15 ) , resulting into an accuracy ( rms ) of only 0.025 dex in @xmath0 $ ] . the dotted and dashed lines in fig .
[ fig : abund ] also show @xmath0 $ ] and @xmath118 $ ] obtained by fitting simultaneously @xmath105'$ ] , @xmath136 , @xmath137 , and @xmath138 line - strengths with predictions from miles models with a kroupa - like ( @xmath3=1.3 ) and a bottom - heavy ( @xmath3=3 ) imf , respectively .
the fit is done by minimizing the rms scatter of model and observed line - strengths with respect to @xmath118 $ ] and @xmath0 $ ] ( see eq .
[ eq : method ] in sec .
[ sec : method ] ) , where the sensitivity of line - strengths to @xmath0 $ ] ( @xmath118 $ ] ) is computed with @xmath98 ( cvd12 ) stellar population models , for old ( 13.5gyr ) age and solar metallicity .
the reason for adopting @xmath136 and @xmath137 is that these features are strongly sensitive to @xmath139 $ ] , while @xmath138 is extremely sensitive to carbon abundance ( e.g. @xcite ) .
the pink dotted lines in fig .
[ fig : abund ] are the same as the red - through - blue dotted lines , but estimating the sensitivity of @xmath136 , @xmath137 , and @xmath138 to @xmath0 $ ] in the fitting with @xmath98 ssps having super - solar ( @xmath85$]@xmath140 ) , rather than solar ( see above ) metallicity . in this case
, @xmath138 decreases more with @xmath0 $ ] than for @xmath85$]@xmath141 , implying higher inferred values of @xmath118 $ ] .
notice also that , as mentioned above , cvd12 models are computed at fixed [ fe / h ] , rather than total metallicity . hence combining cvd12 and extended miles model predictions is only meaningful for elements that give a negligible contribution to @xmath85 $ ] .
since this is only approximately true for @xmath118 $ ] , we have also repeated the fits by modelling the sensitivity of @xmath138 to carbon with tmj11 models ( computed at fixed @xmath85 $ ] ) . as shown by the dot - dashed line in fig .
[ fig : abund ] , the tmj11 models , relative to cvd12 , lead to lower @xmath118 $ ] estimates , by @xmath142dex ( compare the dot - dashed versus dashed curves in fig . [
fig : abund ] ) . fig .
[ fig : abund ] shows that there is a good agreement among estimates of @xmath0 $ ] from different methods , with differences below @xmath143dex for each radial bin . in particular
, changing @xmath3 ( dotted vs. dashed red - through - blue curves ) gives differences @xmath144dex , implying that the derivation of @xmath0 $ ] is robust with respect to the assumed imf slope .
we find that the radial profile of @xmath0 $ ] in xsg1 is essentially constant with radius , out to one half of the effective radius , with a very high @xmath0 $ ] ( @xmath145 ) , and decreases to @xmath146 , only in the outermost bin .
regarding @xmath118 $ ] , differences among methods are more significant ( up to @xmath83dex ) than for @xmath0 $ ] . in general , we find that , similar to @xmath0 $ ] , @xmath118 $ ] is approximately constant with radius ( @xmath146dex ) , at least in the radial range from @xmath147 to @xmath148@xmath18 , dropping to @xmath149dex at the largest distance probed ( @xmath150@xmath18 ) .
notice that although the decrease of @xmath118 $ ] in the innermost radial bin might be due to model uncertainties in the high metallicity regime , in practice such trend does not affect at all our imf constraints ( sec .
[ sec : gamma_trends ] ) . also one can notice that , with the exception of the innermost bin , the ratio of @xmath0 $ ] to @xmath118 $ ] throughout the galaxy is approximately constant ( @xmath151 ) .
the radial behaviour of @xmath0 $ ] and @xmath118 $ ] have important implications for the analysis of tio - based gravity - sensitive features , as discussed below , in sec .
[ sec : tio_profs ] .
the approach we adopt to constrain the imf is an improved version of the method developed in our previous works ( e.g. fer13 , lb13 , @xcite ) . for each radial bin of xsg1 , we minimize the following expression , @xmath152 } , [ x / fe]_r ) = \left [ \frac{age - age_m}{\sigma_{age } } \right]^2 + \nonumber
\\ + \rm \sum_i \left [ \frac{e_{o , i}^{ss}- e_{m , i } } { \sigma_{e_{o , i}^{ss } } } \right]^2 , \label{eq : method}\end{aligned}\ ] ] where the summation index @xmath153 runs over a selected set of spectral features ( see sec .
[ sec : features ] ) , @xmath154 and @xmath155 are our reference estimates of the age ( updated through an iterative procedure ; see sec .
[ sec : corrs ] ) and its uncertainty for the given spectrum ( sec .
[ sec : age_z ] ) , @xmath156 are line - strength predictions for an ssp model with age @xmath157 , @xmath158 are observed line - strengths @xmath159 corrected to solar - scale with an approach similar to that of lb13 , i.e. @xmath160 , \label{eq : corr}\ ] ] where @xmath161 is the _ observed _ slope of the correlation between a given index line - strength and @xmath0 $ ] , derived from sdss data ( see lb13 ) .
@xmath162 represents the total metallicity , estimated for a milky - way - like imf ( in practice , we use a bimodal imf with @xmath3=1.3 ) from the ( emission - corrected ) @xmath109 vs. @xmath105'$ ] diagram ( see , e.g. , fig .
[ fig : hbo ] ) .
the uncertainty , @xmath163 , is obtained by adding in quadrature the statistical error on @xmath164 with the uncertainty on the solar - scaled correction ( see sec . 6 of lb13 ) .
notice that our imf constraints for xsg1 rely on the use of single ssp models to fit observed line - strengths ( eq . [ eq : method ] ) . as discussed in sec .
[ sec : age_z ] , this assumption is well motivated from results of spectral fitting , as well as from the high values of @xmath0 $ ] for xsg1 , pointing to homogeneously old ages for this galaxy at all radii .
moreover , as discussed in lb13 , the use of complex star - formation histories ( e.g. 2ssp models ) does not affect significantly the derived imf slopes in etgs with high velocity dispersion , while the effect is measurable , although mild , in etgs with low velocity dispersion .
we note that spectral fitting ( see _ fit1 _ in [ sec : age_z ] ) yields probability distribution functions at all radii with a negligible fraction in young ( @xmath1658gyr ) components . in @xcite
, we explicitly show that the use of multiple ssp models does not change the radial profile of the imf slope in the elliptical galaxy ngc4552 whose velocity dispersion is comparable to that of xsg1 . the definition of the main spectral features used in the present work is summarized in tab .
[ tab : indices ] .
we focus on optical ( mostly tio - based ) gravity - sensitive features , i.e. @xmath166 , @xmath167 , @xmath168 , and @xmath169 , named @xmath170 , has been defined by @xcite to constrain the imf . in practice , @xmath170 and @xmath169 have a very similar sensitivity to imf , and abundance ratios . ] , contrasting the constraints from these features with those from the wing - ford band ( hereafter @xmath11 ) in the near - infrared ( @xmath171@xmath100 m ) .
moreover , we include the total metallicity indicator , @xmath105'$ ] , in our @xmath172 minimization procedure . since @xmath105'$ ] is independent of abundance ratios ( see , e.g. , tmj11 , vaz15 ) , for this index we set @xmath161@xmath141 in eq .
[ eq : method ] .
hence , the input of our fitting procedure consists of empirically - corrected values of @xmath166 , @xmath167 , @xmath168 , @xmath169 , plus @xmath105'$ ] and our reference @xmath173 estimates . given this input , we minimize eq .
[ eq : method ] over a set of extended miles models , deriving the best - fitting values of the relevant model parameters , and in particular the imf slope , i.e. @xmath78 ( @xmath3 ) for unimodal ( bimodal ) models .
notice that for @xmath11 , no empirical correction is available at the moment , and thus the fitting is performed in a different manner , relying on theoretical ( cvd12 ) models alone to estimate the impact of abundance ratios ( see sec .
[ sec : feh ] for details ) .
in addition to @xmath166 , @xmath167 , @xmath168 , @xmath169 , and @xmath11 , we also study the radial behaviour of the nir tio feature , @xmath9 ( @xmath174@xmath100 m ) , defined by cvd12 .
this feature is mostly sensitive to abundance ratios as well as giant ( rather than both dwarf and giant ) stars in the imf ( in contrast to most tio lines ) , and has been recently used , for similar purposes as in our work ( i.e. studying imf and abundance - ratio gradients ) , by @xcite .
these authors combined @xmath9 with gravity - sensitive features arising from different elements ( na and fe ) than tio . to our knowledge
, the present work is the first one where optical and nir tio features sensitive to the _ same _ elemental abundances ( @xmath175 $ ] and @xmath176 $ ] ; see sec .
[ sec : tio_profs ] ) are studied simultaneously to constrain the imf .
combining @xmath9 with features sensitive to temperature and in particular the optical tio - related indices allows us to disentangle the contribution of giants and abundance ratios , with respect to dwarfs , in the line - strengths . notice
that while we discuss the radial behaviour of @xmath9 , we do not include it into our @xmath172 minimization procedure ( eq . [ eq : method ] ) , as we fit imf slope , age , and metallicity , but not abundance ratios , to the ( solar - scale corrected ) line - strengths .
indeed , obtaining a precise determination of abundance ratios is not the main purpose of the present work . the approach adopted in the present work ( eq . [ eq : method ] )
has some relevant difference with respect to the one adopted in our previous works , i.e. 1 - : : we do not consider any residual abundance - ratio term in eq .
[ eq : method ] , i.e. we assume that our solar - scaled correction is able to completely remove ( within uncertainties ) the combined effect of deviations of single elements from the solar pattern .
this choice is motivated by the fact that corrected line - strengths are reasonably well fitted , at all radial bins , with solar - scaled models ( see sec . [
sec : res ] ) ; 2 - : : in order to account for uncertainties in the absolute calibration of the age ( see sec .
[ sec : age_z ] ) , we apply an iterative procedure .
first we minimize eq .
[ eq : method ] using our reference age estimate for each radial bin ( see sec . [
sec : age_z ] ) , then we re - scale _ all _ age estimates ( i.e. for _ all _ bins ) to the median value of the best - fitting @xmath177 values , and repeat the fitting process .
the procedure typically converges after two iterations , with differences between the average @xmath154 and @xmath178 less than a few percent . 3 - : : to apply the solar - scaled correction , @xmath179 $ ] , we assume that the slope @xmath180 is a function of metallicity ( @xmath162 ) , rather than velocity dispersion ( as in lb13 ) .
this choice deserves some clarification . in lb13 , we binned sdss spectra of etgs with respect to central velocity dispersion , @xmath39 , and @xmath0 $ ] $ ] , see above ] , resulting into 18 stacks with @xmath181 kms@xmath2 , and a set of seven @xmath0 $ ] sub - stacks at fixed @xmath39 .
we showed that @xmath0 $ ] provides a good estimate of the deviation of gravity - sensitive features from solar - scaled , in the sense that , after removing the _ observed _ departure of different indices from @xmath0$]@xmath141 by means of the index@xmath0 $ ] correlations at fixed @xmath39 one can fit a variety of features simultaneously with solar - scaled models . in this
regard , @xmath0 $ ] should be seen as a metric that measures the deviation of a given spectrum from solar scale , and not only as an estimate of the actual enhancement of @xmath97-elements .
because of the @xmath85$] @xmath39 relation in etgs , metallicity also increases with @xmath39 in the lb13 stacks , and thus , the lb13 correction can be applied , in general , either as a function of metallicity , or @xmath39 .
based on the radial behaviour of gravity - sensitive features analyzed in this work ( and in particular the @xmath169 indicator , see app .
[ app : atio_mgf ] ) , we argue that a metallicity - driven correction provides the best way to generalize the lb13 approach also to the case of radially binned spectra ( where the effect of @xmath39 and @xmath85 $ ] , are not equivalent , in contrast to sdss data ) . in practice , for a given spectrum
, we derive the metallicity from the @xmath109 vs. @xmath105'$ ] diagram , using miuscat ssp models with a kroupa - like imf ( @xmath3=1.3 ) .
we calculate the corresponding @xmath161 by interpolating the index-@xmath0 $ ] slopes from the lb13 sdss stacks with respect to @xmath162 ( estimated in the same way for the given spectrum and the sdss stacks ) .
we notice that our approach , where the @xmath180 are a function of @xmath182 , is also motivated by the correlation of imf slope and metallicity , recently found by ( * ? ? ?
* hereafter mn15b ) . in a forthcoming paper ( la barbera et al . , in preparation )
, we will publish the coefficients @xmath161 for all gravity - sensitive features covered by sdss spectra , providing a detailed description of how one should apply our empirical corrections to galaxy spectra .
we start by discussing the radial profiles of imf - sensitive features in xsg1 . in sec . [
sec : tio_profs ] , we focus on the @xmath166 , @xmath167 , and @xmath9 spectral features , as they allow us to disentangle the imf trends with other population features , such as abundance ratios . the radial behaviour of other imf - sensitive optical features ( i.e. @xmath168 and @xmath169 ) is discussed in app .
[ app : atio_mgf ] . in sec .
[ sec : feh ] , we perform a qualitative comparison of observations and model predictions for the wing - ford band .
[ sec : gamma_trends ] shows results of our fitting procedure applied to all observed line - strengths , i.e. the relation between imf - slope and galacto - centric distance in xsg1 ; sec .
[ sec : alpha_trends ] deals with implications for the stellar mass - to - light ratio gradient in xsg1 , in comparison with dynamical constraints .
[ fig : tio_nocorr ] shows one of the main results presented here , namely the radial trends in the @xmath9 vs. @xmath167 ( _ left _ ) and @xmath9 vs. @xmath166 diagrams ( _ right _ ) for xsg1 . in order to perform a qualitative analysis of these trends , the figure also shows predictions of different stellar population models . according to cvd12 models ,
all three features are sensitive to @xmath118 $ ] and @xmath0 $ ] ( see pink and green solid arrows in the figure ) , in such a way that the effect of an enhancement in @xmath118 $ ] tends to cancel out an increase in @xmath0 $ ] .
the sensitivity to @xmath0 $ ] mostly stems from @xmath175 $ ] and @xmath176 $ ] , as shown by the dashed pink arrows in the figure .
however , @xmath9 is far more sensitive to abundance ratios than @xmath166 and @xmath167 , while it is insensitive to the imf .
in fact , changes of the imf slope ( either unimodal or bimodal ) correspond to an almost horizontal shift in both diagrams , as shown by the black solid ( miuscat ) and dashed ( cvd12 ) lines in the figure .
in other words , the effect of abundance ratios and imf are orthogonal , and thus can be singled out in the plots of @xmath9 vs. both @xmath167 and @xmath166 . the observed line - strengths for xsg1 ( plotted with red through blue circles , moving from the innermost to the outermost bin ) show a radial gradient _ only for @xmath167 and @xmath166 _ , while @xmath9 is almost constant out to the largest radius probed .
notice that the lack of gradient for @xmath9 is consistent with the competing effect between @xmath0 $ ] and @xmath118 $ ] and the fact that the ratio of @xmath0 $ ] to @xmath118 $ ] is roughly constant with radius in xsg1 ( see sec . [
sec : profs_xfe ] ) .
the grey arrows in fig .
[ fig : tio_nocorr ] also show the effect of varying the effective temperature of dwarfs , can also ( partly ) mimic a varying imf in the @xmath166 and @xmath167 indices ( see mn15a ) .
however , as shown by @xcite , when taking total metallicity into account , the effect of @xmath183 is irrelevant to constrain the imf .
varying @xmath183 produces an `` oblique '' , rather than horizontal ( imf ) , shift in the @xmath9 vs. @xmath167/@xmath166 plots , in contrast to the observed ( horizontal ) trends . ] , by adopting the prescription from @xcite , cooler than the miuscat reference temperature for dwarf stars ( see * ? ? ?
varying @xmath184 has a minor effect on both @xmath166 and @xmath167 , and thus can not explain the observed gradients .
regarding the sensitivity of tio features to other stellar population properties , we notice that , for @xmath85$]@xmath185 , miuscat models predict both @xmath166 and @xmath167 to be independent of ( total ) metallicity ( see , e.g. , @xcite ) . for old ages , @xmath167 ( and to less extent @xmath166 ) tend to decrease with decreasing age , in such a way that a change in age by @xmath186gyr , from an old centre to younger outskirts ( solid arrows , labelled @xmath187gyr ) might account for @xmath188 ( @xmath189 ) of the observed radial gradient of @xmath167 ( @xmath166 ) .
in practice , however , xsg1 has no significant radial gradient in age ( sec .
[ sec : age_z ] ) , and thus a varying imf is left as the only possible explanation for the @xmath9 vs. @xmath167 and @xmath166 trends . as mentioned above ( sec .
[ sec : method ] ) , the use of @xmath9 in combination with a dwarf - sensitive indicator ( e.g. the wing - ford band ) has been originally proposed by cvd12 , to disentangle the effect of giants and dwarfs on the imf . the present work is the first attempt to combine @xmath9 and the optical tio features ( in particular , @xmath167 and @xmath166 ) to disentangle the effect of the imf with respect to giant stars _ and _ abundance ratios .
recently , @xcite have used @xmath9 to investigate the origin of radial trends of imf - sensitive features in two nearby etgs , with @xmath190kms@xmath2 ( ngc@xmath1911023 ) and @xmath192kms@xmath2 ( ngc@xmath1912974 ) .
they found a slightly - decreasing radial trend of @xmath9 for ngc@xmath1911023 ( with a variation of @xmath193 out to 1@xmath18 ) , and a constant trend for ngc@xmath1912974 , the latter being consistent with our findings for xsg1 . while there is some difference between solar - scaled cvd12 and miuscat model predictions for tio features ( see black circle and small black square in fig .
[ fig : tio_nocorr ] ) , the most remarkable offset in the figure is the one between the @xmath9 model predictions and the observed points . the observed @xmath9 is significantly higher ( by @xmath40.03 ) than in cvd12 and miuscat models .
this might be driven by non - solar abundance ratios ( i.e. @xmath0 $ ] and @xmath118 $ ] ) , or some other effects is strongly sensitive to giant stars ( see section 4.3 and figure 16 of cvd12 ) . as mentioned above , the lack of a gradient in @xmath9 for xsg1 points to no significant radial change in the properties of giants , but rather a genuine radial variation of the dwarf - to - giant ratio in the imf . ] .
the effect of abundance ratios is further illustrated in fig .
[ fig : tio_corr ] , where we use cvd12 ( theoretical ) model predictions to remove the expected effect of @xmath0 $ ] and @xmath118 $ ] from all tio lines ( see empty circles in the plot ) . in practice , for each feature , we estimate the difference of line - strengths between the @xmath0$]- ( @xmath118$]- ) enhanced cvd12 ssp , and the solar - scaled one , for old age ( @xmath103gyr ) , a chabrier imf , and the same velocity dispersion as the observed spectrum at a given radial bin . we rescale these differences to the measured values of @xmath0 $ ] ( @xmath118 $ ] ) for xsg1 . the rescaled line - strength differences ( i.e. the _ theoretical abundance corrections _ ) are subtracted off from the observed values .
the error bars account for the statistical errors on line - strengths as well as for uncertainties on @xmath118 $ ] and @xmath0 $ ] from different methods ( sec .
[ sec : profs_xfe ] ) .
remarkably , the theoretically corrected values of @xmath9 are fairly consistent with miuscat ( and cvd12 ) solar - scaled predictions . however
, one should notice that theoretical corrections are estimated at a specific point in parameter space ( i.e. the cvd12 models provide a grid of values for the age , imf , and metallicity ) , and might be not exactly the same for . more importantly ,
theoretical stellar population models rely on ingredients , such as the stellar atmosphere calculations , that carry notoriously difficult uncertainties .
hence , we use theoretical corrections in a qualitative manner throughout the present work , whereas we rely on _ empirical abundance corrections _
( see also sec .
[ sec : method ] ) to perform a quantitative analysis .
[ fig : tio_corr ] compares the theoretical ( circles ) and empirical ( squares ) abundance corrections on @xmath166 and @xmath167 line - strengths .
notice that for @xmath9 only the theoretical approach is currently possible , as this feature is not covered by the sdss spectra analyzed in our previous works .
therefore , the offset between empirically - corrected points ( squares ) and model predictions ( black arrows ) for @xmath9 in fig .
[ fig : tio_corr ] should not be regarded as a discrepancy between models and data , but rather as the effect of abundance ratios that , at present , can not be described with the lb13 `` empirical '' approach .
since @xmath9 is not used in the fitting procedure , the above offset does not affect any of the quantitative estimates regarding imf variations in xsg1 ( sec .
[ sec : gamma_trends ] ) . in lb13
, we found that @xmath166 and @xmath167 do not depend much on the abundance ratio at fixed velocity dispersion ( metallicity ) .
more precisely , the tio line - strengths are found to increase ( decrease ) slightly with abundance ratio at the lowest ( highest ) velocity dispersions ( i.e. metallicity , in our refined approach , see sec .
[ sec : method ] , and section 5.2 of lb13 ) . figs .
[ fig : tio_nocorr ] and [ fig : tio_corr ] shows that this small dependence on abundance ratio likely arises , _ qualitatively _ , from the counteracting effect of @xmath0 $ ] and @xmath118 $ ] . as a result ,
the empirical corrections tend to stretch out the radial variations of @xmath166 and @xmath167 , making the radial gradient of @xmath166 ( and also @xmath167 ) even more pronounced than in the original , uncorrected data .
we add in quadrature the statistical errors of @xmath166 and @xmath167 to the error budget derived from our empirical correction procedure , giving the error bars shown at the open squares in fig .
[ fig : tio_corr ] .
notice that although ideally theoretical and empirical corrections should agree , in practice they do not .
the theoretical approach tends to shift both @xmath166 and @xmath167 towards lower values , in contrast to the empirical one , where the trends are stretched out towards higher values .
this might be due to limitations in the theoretical estimates ( see above ) .
in addition , a combination of departures of other element abundance ratios , such as @xmath139 $ ] , @xmath194 $ ] that individually do not produce a major change in the tio line strengths can compensate , along with @xmath118 $ ] , the increase in @xmath166 and @xmath167 due to @xmath176 $ ] and @xmath175 $ ] . in practice , when inferring the imf , the fact that @xmath166 and @xmath167 are lower when relying on theoretical abundance corrections would be compensated by a younger age in our fitting procedure , implying similar imf - radial trends as those presented in sec.[sec : gamma_trends ] . since we have not been able to find a consistent interpretation of all the observed indices when relying on theoretical models ( see below and app .
[ app : atio_mgf ] ) and because of the current limitations of the publicly available cvd12 models ( see above ) , our quantitative analysis relies fully on the empirical approach .
one might wonder to what extent the trends in figs .
[ fig : tio_nocorr ] and [ fig : tio_corr ] might be driven by an issue related to the data reduction process .
the grey arrows in fig .
[ fig : tio_corr ] show the effect of a change in the telluric correction method in the reduction process ( i.e. telluric - standard stars vs. molecfit , see sec . [
sec : obs ] ) on tio line - strengths .
the effect is negligible for @xmath9 and @xmath167 , while it is mild for @xmath166 .
notice that although the spectral region of @xmath9 is affected by telluric lines , the telluric correction is remarkably accurate ( see also fig . [
fig : spectra ] ) , thanks to the excellent quality especially the high resolution of our x - shooter data .
the telluric correction uncertainty is added in quadrature to the error bars on the empirically corrected line - strengths .
other changes in the reduction process ( e.g. sky subtraction ) have been found not to affect at all the observed tio line - strengths .
, as a function of the wing - ford band , @xmath11 .
circles with different colours ( from red to purple , as labelled ) mark different radial bins for xsg1 .
the empty star corresponds to another massive ( @xmath195kms@xmath2 ; @xmath0$]@xmath34 ) etg , xsg2 ( see the text ) .
error bars mark @xmath47 uncertainties on line - strengths .
we show predictions for extended - miles bimodal ( solid - black lines ) and unimodal ( solid - green line ) models with solar and super - solar @xmath85$](plotted as small and big diamonds , respectively , for an age of @xmath196gyr ) , as well as predictions for cvd12 unimodal models ( dashed line ) , with a chabrier , salpeter and bottom - heavy ( @xmath197 ) imf , at solar metallicity , and an age of @xmath103gyr .
the dotted ( dot - dashed ) line joins predictions for unimodal ( bimodal ) extended - miles models with solar @xmath85 $ ] ( and should be compared with the dashed line for cvd12 models ) .
both @xmath198 and @xmath11 decrease with increasing @xmath0 $ ] , as shown by the pink arrows , for extended - miles models with @xmath3@xmath113 and @xmath199 .
notice that the effect of @xmath0 $ ] for @xmath3@xmath106 ( not shown to make the plot more readable ) is very similar to the case @xmath3@xmath113 . all data and model line - strengths refer to a velocity dispersion of 300kms@xmath2 ( see sec . [
sec : models ] ) .
, width=328 ] the optical gravity - sensitive spectral features constrain the mass fraction of low - mass stars in the imf ( see lb13 ) , and thus do not allow for different functional forms of the imf to be singled out .
this is illustrated in fig .
[ fig : tio_nocorr ] , where one can see that an increase of the imf slope of both bimodal and unimodal models produces the same shift in @xmath166 and @xmath167 , and only a tiny difference , with respect to the observed scatter , in @xmath9 . on the other hand ,
the wing - ford band , @xmath11 , is sensitive to very low - mass stars in the imf ( see figure 17 of cvd12 ) .
hence , a combination of this feature with optical indicators allows to better constrain the functional form of the imf at the low - mass end , discriminating , for instance , between unimodal and bimodal models .
[ fig : feh_rad ] , shows the abundance - sensitive ( [ fe / h ] ) feature @xmath198 @xmath200 , in the optical , as a function of @xmath11 ( see figure 12 of cvd12a ) .
we show extended - miles ssp predictions for bimodal ( solid - black lines ) as well as unimodal ( green line ) models with solar and super - solar metallicities ( small and big diamonds , respectively ) .
the extended - miles predictions are compared to those for cvd12 unimodal models ( black - dashed line ) .
both sets of models agree for a mw - like imf , for which the @xmath11 line - strengths from extended - miles and cvd12 differ by only @xmath40.03 .
however , the change of @xmath11 with @xmath78 ( black - dotted for extended - miles , black - dashed line for cvd12 models ) is remarkably weaker in extended - miles with respect to the cvd12 models .
notice that for many imf - sensitive indicators ( e.g. na- and ca - dependent features ) the response to the imf is also significantly model - dependent ( e.g. lb13 , @xcite ) .
on the other hand , the behaviour of tio features seems quite robust between different models ( see solid vs. dashed black lines in fig . [
fig : tio_nocorr ] ) . fig .
[ fig : feh_rad ] shows that , using the extended - miles models , @xmath11 increases with total metallicity ( cvd12a models do not vary @xmath85 $ ] ) , and increases with the imf slope more strongly for the unimodal case .
the variation of the index is much weaker for a bimodal imf , where the low - mass end is tapered off by construction .
the pink arrows in the figure show the effect of varying @xmath0 $ ] ( by @xmath201dex ) , at fixed @xmath85 $ ] , on both @xmath11 and @xmath198 .
the @xmath11 and @xmath198 variation with @xmath0 $ ] is computed from @xmath97-enhanced and solar - scaled cvd12 models ( chabrier imf , @xmath103gyr , solar metallicity ) , removing the effect of the change in total metallicity due to the fact that cvd12 models are computed at fixed @xmath104 $ ] ( rather than @xmath85 $ ] ) .
we remind the reader that , at the moment , we do not have any empirical method to correct the @xmath11 for abundance variations ( see sec . [ sec : tio_profs ] ) .
however , at least for @xmath198 , we have verified that the predicted effect of @xmath0 $ ] is robust , as the cvd12 model predictions are fairly consistent with those from the @xmath98 models .
in fact , for @xmath202=+0.2 $ ] , we get @xmath131@xmath198@xmath203 from cvd12a models ( see pink arrow in the figure for @xmath3@xmath113 ) , while from @xmath98 we get @xmath131@xmath198 in the range from @xmath204 to @xmath205 , depending on age , metallicity , and imf ( with the largest variations found in models at the highest @xmath85 $ ] ) .
[ fig : feh_rad ] also shows the observed @xmath11 for xsg1 as a function of galacto - centric distance , with red - through - purple filled circles . despite of the high s
/ n ratio of our x - shooter data , the estimates in the outermost radial bins have large uncertainties in @xmath11 to provide useful constraints on the imf .
therefore , we focus here on the three innermost radial bins .
the most remarkable aspect in the figure is that the observed @xmath11 is significantly lower than expected for a bottom - heavy unimodal imf compare the largest black circle ( cvd12 bottom - heavy estimate ) with the red dot with error bars ( observed data ) .
this mismatch also applies to the extended - miles models .
specifically , for an old , metal - rich population ( @xmath412.6gyr ; @xmath85$]@xmath206 dex ) , typical values in the central regions of xsg1 ( see fig .
[ fig : agez ] ) , and @xmath197 ( an imf slope even lower than the one implied by optical imf - sensitive indicators alone , see fig .
[ fig : gamma_rad ] ) , the extended - miles models give @xmath11@xmath207 ( @xmath134kms@xmath2 ) . for @xmath0$]@xmath15
( see fig .
[ fig : abund ] ) , the @xmath11 should be lower by @xmath208 , implying an expected value of @xmath209 in the innermost radial bin , in disagreement with the measured value of @xmath11@xmath210 . instead
, for a bottom - heavy imf with @xmath3@xmath113 ( 12.6gyr , @xmath85$]@xmath206 ) , the extended - miles prediction ( accounting for @xmath0 $ ] ) is @xmath11@xmath211 , fully consistent with the observations .
a similar kind of discrepancy with a unimodal imf is present in the second and third radial bins of xsg1 , although it is less significant because of the larger uncertainties on @xmath11 .
we emphasize that the sensitivity of @xmath11 with respect to @xmath0 $ ] is computed in fig.[fig : feh_rad ] by use of the cvd12 models for a chabrier imf .
if such sensitivity were larger for a unimodal bottom - heavy case , relative to a chabrier imf , the @xmath11 might decrease more strongly with @xmath0 $ ] than shown in fig .
[ fig : feh_rad ] , making this discrepancy less significant . however , it seems not to be the case : the second massive galaxy , xsg2 ( with @xmath212kms@xmath2 ) , observed as part of the same programme ( sec .
[ sec : obs ] ) , is shown as an empty star with error bars in fig .
[ fig : feh_rad ] .
xsg2 has a _ lower _
@xmath0$](@xmath213 ) than xsg1 , and its @xmath11 is far too - low with respect to the bottom - heavy unimodal predictions . in the following section ,
we show that a bottom - heavy bimodal imf matches both optical indicators and the wing - ford band in xsg1 . , in xsg1 . in panels a , b , c , and d , the observed equivalent widths , empirically corrected to solar - scale ( see sec . [
sec : method ] ) , are plotted in black .
the green curves are the uncorrected values , and the bimodal best - fit line - strengths , obtained by simultaneously fitting all the optical indicators ( @xmath166 , @xmath167 , @xmath168 , and @xmath169 ) are shown in red . in panel
e ( wing - ford band ) , the corrected values are obtained by removing the effect of the best - fitting @xmath0 $ ] from the measured line - strengths ( see sec . [
sec : feh ] ) , while the blue line shows extended - miles ( solar - scaled ) best - fitting line - strengths .
error bars denote @xmath47 uncertainties , including errors on the correction procedure , as well as the telluric correction ( see grey arrow in panel b ) and dichroic continuum variations ( see grey arrows in panel c ) .
all line - strengths have been corrected , based on the best - fitting extended - miles model in each bin , to @xmath134kms@xmath2 ( see sec .
[ sec : models ] ) .
, width=321 ] by use of the method described in sec .
[ sec : method ] , we fit imf - sensitive features of xsg1 with either unimodal or bimodal extended - miles ssp models . fig .
[ fig : ind_rad ] shows the fitting results for bimodal models , plotting @xmath166 , @xmath167 , @xmath168 , @xmath169 , and @xmath11 as a function of galacto - centric distance .
the unimodal fits are not shown in the figure , as the comparison of best - fitting and observed line - strengths is very similar to that for bimodal ssps .
we fit separately ( i ) the optical indicators and ( ii ) the wing - ford band : ( i ) : : for the optical indices ( @xmath166 , @xmath167 , @xmath168 , and @xmath169 ) , we have applied the empirical corrections with respect to abundance pattern , and used our iterative approach to account for the uncertainty on the absolute age zeropoint , as described in sec .
[ sec : method ] .
( ii ) : : the fit with @xmath11 uses simultaneously @xmath11 , @xmath136 , @xmath105'$ ] , and @xmath198 without any iterative approach , but assuming the age constraint ( i.e. the term @xmath173 in eq . [ eq : method ] ) from the fit of optical features ( i ) .
indeed , the age constraints do not affect significantly the results for @xmath11 , as this feature is largely independent of age ( see fig .
[ fig : ind_rad ] ) . in the fit for @xmath11
, we do not apply the empirical corrections for abundance pattern , as they are not available for this feature . instead
, we include @xmath0 $ ] as a free - fitting parameter , using @xmath11 , @xmath136 , @xmath105'$ ] , and @xmath198 to constrain , simultaneously , the parameters @xmath85 $ ] , @xmath0 $ ] , and the imf slope . to this effect ,
the sensitivity of @xmath11 to @xmath0 $ ] is modelled , at fixed @xmath85 $ ] , as discussed in sec .
[ sec : feh ] . we have found that our results ( see below ) remain unchanged if ( a ) we assume our proxy - based estimate of @xmath0 $ ] in each radial bin ( fig . [
fig : abund ] ) , and fit simultaneously only @xmath11 and @xmath105'$ ] to infer @xmath85 $ ] and imf slope ( either bimodal or unimodal ) , and ( b ) we also fix @xmath85 $ ] to the output from the step ( i ) , and use @xmath11 only to infer the imf slope . black symbols and lines in fig .
[ fig : ind_rad ] refer to corrected indices .
uncorrected values are plotted as green lines .
all line - strengths in the figure are corrected to a common velocity dispersion of 300kms@xmath2 using the best - fit from the ( bimodal ) extended - miles models ( see sec .
[ sec : models ] ) .
the best - fitting line - strengths are shown as red and blue lines in fig .
[ fig : ind_rad ] for the optical and @xmath11 indices , respectively .
the best - fit values of @xmath78 ( unimodal ) and @xmath3 ( bimodal ) are shown in the lower and upper panels of fig .
[ fig : gamma_rad ] , respectively .
we also display , with different line - types , the results for bimodal models with different combinations of optical indicators .
we also show the result when lifting the constraint on the age ( grey curve ; where the first term in eq .
[ eq : method ] is neglected ) .
[ fig : ind_rad ] and [ fig : gamma_rad ] show the main results of the present work , as follows : 1 - : : all the optical ( tio - based ) imf - sensitive features in xsg1 show a significant radial gradient , implying an imf gradient in this massive galaxy , from bottom - heavy in the centre to mw - like at @xmath214 .
fitting all the optical features simultaneously ( red lines in the figures ) , we measure a radial decrease from @xmath3@xmath215 ( @xmath216 ) in the centre to @xmath3@xmath217 ( @xmath218 ) at @xmath219 .
notice that all tio features analyzed in the present work are insensitive to imf slope variations when the imf is top - heavier than kroupa ( i.e. for @xmath3@xmath220 , or @xmath221 ; see figures 11 and 18 in lb13 ) .
the fact that the lower error bars for the outermost bin in fig .
[ fig : gamma_rad ] are relatively small ( going , e.g. , from @xmath222 to @xmath15 for unimodal models ) is only due to the lowest boundary value of imf slope ( @xmath78 and @xmath223 ) for extended - miles models .
hence , our results in the outermost radial bin of xsg1 can be considered fully consistent with a kroupa - like imf . 2 - : : for unimodal models , the wing - ford band provides inconsistent constraints ( i.e. significantly lower @xmath78 ) with respect to those from the optical indicators ( blue and red dots in lower panel of fig . [ fig : gamma_rad ] , respectively ) .
in the innermost radial bin , the @xmath78 derived with @xmath11 is smaller than the optical - based one by @xmath224 , i.e. the discrepancy is significant at more than @xmath225 . in particular
, this result seems not to be specific to xsg1 , as also for xsg2 the second galaxy we have observed with x - shooter the @xmath11 is far too low to be consistent with a unimodal imf ( see sec .
[ sec : feh ] ) . 3 - : : for a low - mass tapered bimodal imf , we get fully consistent results between optical lines and the @xmath11 . in the innermost bin ,
the difference between @xmath11- and optical - derived @xmath3 is @xmath226 . regarding the quality of the fits ( i.e. fig . [
fig : ind_rad ] ) , one should notice that : i. : : the fit using optical features ( red curves in the figure ) describes remarkably well the @xmath167 and @xmath169 observed trends ( black circles and lines ) .
we emphasize that the uncorrected line - strengths of @xmath169 do not show any clear radial gradient ( green curve in panel d ) .
it is our empirical abundance - pattern correction that reveals a radial gradient in this feature , making it fully consistent with @xmath167 .
notice also that , as discussed in app .
[ app : atio_mgf ] , in the innermost bin , the empirical correction is the same when applying it either as a function of velocity dispersion or metallicity .
however , in the former , the external corrected values of @xmath169 in panel d would be too high with respect to the best - fitting solution .
this is consistent with what we reported in mn15a , where the radial gradient of @xmath169 was found to be too small with respect to that of other features , when applying our empirical approach in terms of velocity dispersion .
our refined method , where abundance - pattern corrections are performed as a function of metallicity , provides consistent results .
ii . : : for @xmath166 , the three innermost data points appear too high ( at the @xmath227 level ) with respect to the best - fits . this discrepancy might be explained by uncertainties in our telluric correction procedure ( see sec . [
sec : tio_profs ] ) . using molecfit rather than telluric standard stars in the reduction process
( see sec .
[ sec : obs ] ) , we get systematically lower @xmath166 , by @xmath228mag ( see grey arrow in panel b of fig . [
fig : ind_rad ] ) , making @xmath166 more consistent with our best - fitting solution at all radii .
: : for @xmath168 , the best - fitting solution shows a shallow radial gradient .
these results from two competing effects : @xmath168 is expected ( from extended - miles models ) to increase with the imf . however , while the index is independent of metallicity for a kroupa - like imf , it _ decreases _ with @xmath85 $ ] for a bottom - heavy imf .
hence , in the inner regions , i.e. towards higher @xmath85 $ ] and imf slope , one would not expect large variations in the index .
these competing trends , as well as the strong sensitivity of @xmath168 to the continuum ( see app . [ app : atio_mgf ] ) make this index , in practice , a less useful imf indicator than @xmath166 , @xmath167 , and also @xmath169 .
the fact that the observed @xmath168 shows a gradient , but the best - fitting solution does not , might be due to a combination of uncertainties in the empirical correction of this index , along with issues with the absolute / relative calibration of @xmath168 because of the instability of the x - shooter dichroic response with time ( see sec .
[ sec : obs ] , app .
[ app : atio_mgf ] , and grey arrows in panel c of fig .
[ fig : ind_rad ] ) .
: : for the wing - ford band , we obtain excellent agreement between the observed and best - fit line - strengths ( blue and black lines in panel e of the figure ) , although such a result is to be expected , considering that , in this case , the only imf - sensitive feature used in the fitting procedure is @xmath11 . as mentioned above , the remarkable result is that one gets very consistent constraints , for a _ bimodal _ imf , between the optical features and the wing - ford band . finally , we want to emphasize that , although there are a few discrepancies between the observed and best - fit optical indicators in some radial bins ( i.e. @xmath166 in the three innermost , and @xmath168 in the two outermost bins ) , fig . [
fig : gamma_rad ] shows that the quality of the fit is good , within error bars .
moreover , _ the inferred radial profile of the bimodal imf slope , @xmath3 , is consistent when excluding any one index from the analysis _
( see different line - types in the figure ) , i.e. if we include in the fitting , for each bin , only those indices for which models give a better matching to the data .
this proves that our results are robust , and not affected by possible mismatches between data and models .
we also emphasize that _ when completely neglecting the age constraints _ in the fitting procedure we get very consistent results to our reference imf profile ( solid red curve ) in fig .
[ fig : gamma_rad ] , implying that our findings are not affected by the uncertainty on the absolute - age zeropoint . ) and unimodal ( bottom ; @xmath78 ) imf slope as a function of galacto - centric distance for xsg1 .
error bars denote @xmath47 uncertainties .
black lines and symbols correspond to different methods to constrain the slope , using different optical features , as labelled in the top panel .
red curves and symbols show the imf slopes derived by combining all the optical imf - sensitive indicators analyzed in the present work ( i.e. @xmath166 , @xmath167 , @xmath168 , and @xmath169 ) . blue symbols and curves
plot the imf slope inferred using only the wing - ford band .
notice that in this latter case only the three innermost radial bins are plotted , as measurement errors on @xmath11 do not allow us to obtain significant constraints for the three outermost bins . for a bimodal imf , we find an excellent agreement among optical indicators and @xmath11 , while unimodal imf slopes , inferred from the @xmath11 , are on average too low ( at more than the @xmath229 level ) with respect to the optical constraints .
notice that the inferred radial profile of @xmath3 , is consistent when excluding any one index from the analysis ( see different line - types , in black ) , as well as when completely neglecting the age constraints in the fitting procedure ( grey solid curve ; see the text).,width=325 ] the radial trend of imf slope found in xsg1 ( fig .
[ fig : gamma_rad ] ) leads to a radial gradient in the stellar mass - to - light ratio , @xmath230 . to quantify it
, we use the mismatch parameter @xmath231 , defined as the ratio between the stellar @xmath232 corresponding to the best - fit age , metallicity , and imf slope , and the one derived for the same age and metallicity , assuming a mw - like imf ( we adopt as mw reference a bimodal @xmath3@xmath106 imf ) .
we derive @xmath97 as a function of galacto - centric distance .
notice that for a mw - like normalization , @xmath233 , whereas either bottom- or top - heavy imf give @xmath234 because of the higher fraction of low - mass stars , or remnants , respectively .
a salpeter normalization implies @xmath235 . at each radius , we compute @xmath230 from bimodal imf models , as unimodal models are inconsistent with constraints from optical features and the wing - ford band . to this effect , we random shift the values of @xmath3 , from the best - fitting estimate at the given radius ( fig . [ fig : gamma_rad ] ) , according to their uncertainties , and compute the average @xmath230 and its error .
the computation excludes top - heavy models with @xmath3@xmath236 , as these models have @xmath230 much larger than the mw reference ( @xmath3@xmath237 ) , but they are indistinguishable , concerning our imf - sensitive indicators , from the case with @xmath3@xmath238 . fig .
[ fig : alpha ] plots @xmath97 as a function of galacto - centric distance ( black circles , with @xmath47 error bars ) . as expected from the radial trend of @xmath3 , the @xmath97 parameter also decreases with @xmath239 , from @xmath151 in the innermost radial bin , to @xmath240 beyond half of the effective radius .
the red curve in the figure shows the integrated constraints on @xmath97 , i.e. the luminosity - weighted values of @xmath97 computed within concentric circular apertures , under the assumption of circular symmetry . as expected , the integrated version of @xmath97 shows a slower decline with radius , with respect to the local one , becoming slightly lower than the salpeter value already within @xmath241 ( second outermost red point , and dashed horizontal line ) .
[ fig : alpha ] also shows as a blue square the mismatch parameter , @xmath242 , obtained by normalizing the total mass - to - light ratio from the jeans anisotropic modelling ( sec .
[ sec : kin ] ) to the stellar m / l for a kroupa - like imf ( see above ) .
notice that we find @xmath243 , implying @xmath244 .
@xmath245 can be interpreted as the ratio between total mass ( including dark matter , which is likely @xmath246 within the central galaxy regions ; see @xcite ) and the @xmath247-band luminosity within a sphere of radius r@xmath42 .
this is actually the region where the data are more reliable and the fit using jam describes them better . for r@xmath42
( @xmath248 ) , our bimodal - imf fits imply an aperture integrated value ( see red circles in the figure ) of @xmath249 , i.e. smaller than @xmath242 .
hence , not only our spectroscopic constraints are consistent between optical and nir indicators , but they are also consistent with the dynamical analysis .
the ratio @xmath250 gives the dark - matter fraction ( i.e. the ratio of non - stellar to total mass ) for xsg1 , within half the effective radius , and amounts to @xmath251 , consistent with the expectation that this galaxy should have a minor dark - matter contribution in the central region @xcite .
, for bimodal imf models ( see text for details ) .
the black line and symbols show the local values of @xmath97 , while the luminosity - weighted value , @xmath252 , measured over circular apertures , is plotted in red .
error bars denote @xmath47 level uncertainties .
the blue square plots the mismatch parameter , @xmath242 , obtained from the jam dynamical analysis ( sec . [ sec : kin ] ) .
notice that unimodal models would produce an overly high @xmath252 ( @xmath215 ) with respect to @xmath242 .
, width=321 ]
the existence of a radial gradient of the imf in xsg1 agrees well with our previous results for ngc4552 and ngc5557 two massive nearby etgs with central velocity dispersion of @xmath253 and @xmath254kms@xmath2 , respectively ( mn15a ) . for ngc4552 ( the best target in mn15a ) , the bimodal imf slope decreases from @xmath255 in the galaxy centre , to @xmath151 at @xmath256 , consistent with the profile in fig .
[ fig : gamma_rad ] for xsg1 .
notice , however , that the velocity dispersion profile of xsg1 ranges from @xmath48kms@xmath2 in the centre to @xmath49kms@xmath2 at @xmath257 .
hence , our results extend the findings of mn15a at significantly higher velocity dispersion , suggesting that imf gradients are common in high velocity dispersion etgs . on the other hand
, we do not expect such gradients to be universal . in (
* hereafter mn15c ) , we derived the imf profile of the massive and compact `` relic '' galaxy ngc1277 ( @xmath258kms@xmath2 ) , finding only a mild imf gradient , with @xmath3 decreasing from @xmath215 in the galaxy centre , to @xmath259 at @xmath260 . in mn15b , analyzing a sample of etgs from the califa survey ( with @xmath261kms@xmath2 ) we found that locally
i.e. at a given position within the galaxy
@xmath3 is best correlated with total metallicity .
remarkably , no radial variation is found in any of the observed properties of xsg1 ( e.g. kinematics profile , stellar population properties , abundance ratios ) , except for imf and metallicity , reinforcing the idea of a tight connection between these two parameters .
similarly to ngc1277 , xsg1 has a high @xmath39 , and very high abundance ratios ( @xmath0$]@xmath145 ; comparable to ngc1277 , with @xmath0$]@xmath262 ) , with no age gradient and no radial variation of @xmath0 $ ] within @xmath263 ( see fig . [
fig : abund ] ) .
nevertheless , xsg1 and ngc1277 have very different radial variations of the imf , suggesting that one may expect , in general , a wide range of imf profiles in massive etgs .
this variety , also expected from the scatter of the imf-@xmath85 $ ] relation ( see mn15b ) , is similar to the observed metallicity gradients in etgs . on average , a correlation of @xmath85 $ ] gradients with velocity dispersion is detected ( see , e.g. , @xcite ) .
however , this correlation is tighter at low - velocity dispersion ( @xmath264kms@xmath2 ) , whereas a wide scatter of metallicity gradients is observed in more massive etgs ( see , e.g. , figure 8 of @xcite ) , ranging from almost null to significantly negative ( @xmath265/\delta \log r \sim -0.6 $ ] ) .
this scatter originates possibly from the different merging histories of massive etgs @xcite .
our results for xsg1 and other massive galaxies suggests that each specific channel of etg formation will shape the radial imf profile of etgs . while an imf gradient will be produced by the different star - formation and chemical enrichment processes at different radii during the initial stages
likely through a time - varying imf scenario @xcite this gradient is expected to be modified at later times through the different merging events .
this scenario naturally yields a range of imf radial gradients in nearby etgs , possibly without a sharp correlation with any specific property of the galaxy . while in our previous works ( mn15a and mn15c ) we presented evidence against abundance ratios and other systematic effects , such as the effective temperature of giants / dwarfs , as the major driver of radial gradients of imf - sensitive features in etgs , we separate more explicitly in xsg1 the effect of the imf from other confounding factors . we find a significant radial gradient of features sensitive to the dwarf - to - giant ratio in the imf ( e.g. @xmath166 and @xmath167 ) , whereas no gradient is detected either in the abundance ratios ( @xmath0 $ ] ; @xmath118 $ ] ) or @xmath9 , sensitive to giants and abundance effects , but insensitive to dwarfs ) . to our knowledge , this is the clearest evidence so far , of a radial imf gradient in a massive galaxy .
the combination of no gradient in @xmath0 $ ] and a radial trend in the imf confirms our previous claim that abundance ratio variations can not be the major driver of the observed trends ( lb13 and @xcite ) .
these results are in good agreement with the relation between dynamically - determined imf mass normalization factors and stellar population parameters @xcite .
notice that a better correlation of the imf with @xmath0 $ ] , rather than velocity dispersion , has been originally suggested by @xcite , but it is not confirmed from studies where the effects of @xmath39 and @xmath0 $ ] are singled out .
indeed , the correlation of imf with @xmath0 $ ] in @xcite might be driven by the fact that cvd12 models are computed at fixed [ fe / h ] , rather than total metallicity . recently , @xcite investigated the origin of radial trends of imf - sensitive features in two nearby etgs , with @xmath190kms@xmath2 ( ngc@xmath1911023 ) and @xmath192kms@xmath2 ( ngc@xmath1912974 ) , respectively .
they detected a radial gradient in the dwarf - sensitive feature nai8190 ( @xmath266 ) , with only a mild variation of @xmath11 , favouring abundance ratio variations ( in particular , @xmath267 $ ] and @xmath268 $ ] ) as the main explanation for the observed trends of imf - sensitive features in etgs .
although the radial behaviour of the imf might be flat in some massive etgs ( see above ) , we emphasize that , in practice , the presence of abundance- and imf - gradients can coexist , hampering a direct interpretation of the indices .
one example is the @xmath169 feature ( btio ; see @xcite ) .
no significant at most a very shallow
gradient is observed for this imf - sensitive feature for xsg1 as well as ngc4552 ( see mn15a ) .
however , note that abundance ratios tend to wash out the underlying imf trend , and it is only by correcting for abundance effects that one recovers consistent results with other features ( see app .
[ app : atio_mgf ] ) . for some features , like @xmath11 ,
the interpretation is even trickier . according to predictions from new nir ( extended miles ) stellar population models with varying total metallicity _ and _ imf , the @xmath11 response to ( total ) metallicity and thus also to @xmath0 $ ] at fixed @xmath85$] is larger at higher metallicity .
hence , even if @xmath0 $ ] were constant ( as in xsg1 ) , the @xmath0 $ ] tends to further decrease the observed line - strength of @xmath11 in the galaxy centre with respect to the outer regions , counteracting the effect of a steeper imf in the centre ( i.e. reducing the radial gradient of @xmath11 ) . notice that in lb13 we also reported that the abundance - ratio response of some gravity - sensitive features might be coupled with the imf slope .
for instance , in lb13 ( see figure 15 ) the response of nai8200 ( cat ) to @xmath267 $ ] ( @xmath269 $ ] ) was shown to be likely stronger ( weaker ) for ssp models with a bottom - heavy imf , relative to kroupa .
this coupling , not included in any publicly - available version of stellar population models with varying abundance ratios ( e.g. cvd12 ) , might be responsible for some of the apparent contradictions between model predictions and observations @xcite .
our analysis of optical and nir features in xsg1 is inconsistent with a unimodal , single power - law imf , favouring a functional form tapered off at the low - mass end , such as the bimodal imf of @xcite .
we emphasize that this result does not imply that the imf functional form has to be bimodal .
as discussed in fer13 and lb13 , unimodal imf models are also disfavoured by the fact that they predict overly high stellar @xmath232 ratios with respect to simple dynamical constraints .
however , a possible way out is to assume a unimodal parametrization , increasing the low mass - end cutoff , @xmath270 , in the imf ( with @xmath271 ; see @xcite ) .
in fact , very - low mass stars ( @xmath272 ) do not contribute significantly to the integrated galaxy light , while they provide a major contribution to the @xmath273 . since the wing - ford band is sensitive to the very low - mass stars in the imf ( see figure 17 of cvd12 ) , increasing @xmath274
might also reconcile optical and nir indicators for xsg1 , in the same way as adopting a bimodal imf .
however , this option is rather contrived , and would beg the question of why such a gap exists between low - mass stars and massive brown dwarves .
an alternative approach , taken by @xcite , is to assume a three - segment set of power laws for the imf , describing individually the behaviour of very - low , and low , mass stars . in this parametrization
, our results for xsg1 would imply that the imf has to be steep at low , but not at very - low mass .
while the information on the detailed shape of the imf is imprinted in the optical and nir features of a galaxy spectrum ( cvd12 ) , it is not clear if this information can be recovered , mostly because of current uncertainties on stellar population models , and on modelling the effect of variations in the abundance ratios on the targeted spectral features . _ regardless of the detailed shape of the imf , one important result from the present work is that spectroscopic constraints are inconsistent with a pure unimodal parametrization of the imf extended down to the h - core burning stellar mass limit .
_ we notice that our dynamical analysis , based on jam modelling with an approach as close as possible to that of the atlas@xmath275 sample , provides @xmath232 constraints fully consistent with our stellar population analysis .
recently , ( * ? ? ?
* hereafter slc15 ) analysed a sample of nearby etgs where @xmath232 can be constrained through strong gravitational lensing in the inner regions of galaxies . for three systems
, slc15 estimated a `` mass excess factor '' , @xmath97 , only marginally consistent with a salpeter - imf normalization ( @xmath235 ) , in tension with stellar population models . among the three lensing galaxies of slc15 ,
the more similar one to xsg1 is perhaps snl-2 , with a velocity dispersion of @xmath276kms@xmath2 , @xmath0$]@xmath277 , and an effective radius of @xmath278kpc ( in the j band ) . within the einstein radius ( @xmath279 ) , that probes a region @xmath280 for snl-2 , slc15 derive @xmath281 ( @xmath282 ) , for the case where no dark matter ( dark matter ) is assumed . at a galacto - centric distance of @xmath283 , xsg1 has an ( aperture-)integrated value of @xmath284 , for a bimodal imf .
this value is only marginally higher ( below the @xmath285 level ) with respect to the no - dark - matter estimate of slc15 .
moreover , as discussed above , one may expect , and we do observe , a variety of imf profiles in etgs .
hence , we can not exclude that the imf normalization is different when comparing different galaxies ( e.g. xsg1 and snl-2 ) .
furthermore , the estimate of the mass excess , @xmath97 , changes when adopting different imf paramaterizations in the stellar population analysis ( see fer13 and lb13 ) . the case of snl-1 in scl15 is also interesting .
this lensing galaxy has @xmath286kms@xmath2 , @xmath0$]@xmath287 , and an effective radius of @xmath151kpc , being more compact than both xsg1 , and snl-2 . within an @xmath279 of @xmath288
, slc15 find @xmath289 ( @xmath290 ) , for the case with no - dark - matter ( dark - matter ) .
in the case of xsg1 , at the maximum galacto - centric distance probed of @xmath291 , we find an ( aperture-)integrated mass excess @xmath292 ( see red curve in fig . [
fig : alpha ] ) , fully consistent with the estimate of slc15 for snl-1 .
these comparisons show that radial imf , and mass - to - light , gradients are an important ingredient when comparing constraints from different techniques .
* hereafter sbk15 ) have recently suggested an anticorrelation between imf slope and surface density , defined as @xmath293 in units of @xmath294 . since etgs with high @xmath0 $ ] are also the more compact ones around the fundamental plane relation @xcite , the result of sbk15 seems consistent with our findings ( see @xcite ) , namely , at fixed velocity dispersion , the etgs with the highest-@xmath0 $ ] might also have the lower imf slopes .
however , our preliminary results for xsg2 , the second galaxy observed in our x - shooter campaign , challenges this scenario . xsg1 and
xsg2 have @xmath295 and @xmath151 , respectively , i.e. xsg2 is significantly less dense than xsg1 .
nevertheless , it has a lower @xmath11 ( see fig . [
fig : feh_rad ] ) .
this result , combined with its higher central metallicity ( @xmath85$]@xmath296 ) and lower @xmath0 $ ] ( @xmath86 ) with respect to the values for xsg1 ( @xmath85$]@xmath86 and @xmath0$]@xmath145 ) , implies a lower imf slope .
moreover , @xcite found a mismatch parameter @xmath297 for stacked spectra of compact etgs in the sdss , at @xmath298kms@xmath2 , implying a higher imf normalization in more compact galaxies with respect to the general population of etgs ( see , e.g. , figure 5 of cvd12b ) .
we point out that the presence of a variety of imf gradients in massive etgs complicates the comparison of imf slopes among galaxies with different degree of compactness .
larger samples are required to address this issue in detail .
we present a detailed analysis of a set of targeted imf - sensitive line strengths in a massive ( @xmath1kms@xmath2 ) early - type galaxy ( xsg1 ) , at @xmath299 , observed with the vlt / x - shooter instrument . the lack of a significant gradient in the age and @xmath0$] two stellar population properties out to one half of the effective radius ( figs . [
fig : agez ] and [ fig : abund ] ) , makes this target especially suited to the analysis of radial variations in the imf . by combining several tio - based indices in the optical and nir
, we exploit the different sensitivities to the underlying stellar population parameters to disentangle the contribution from the stellar imf .
the observed indices for xsg1 on the @xmath9 vs. optical tio diagrams ( fig .
[ fig : tio_nocorr ] ) show a significant gradient in the optical indices , not followed by the nir index , which is insensitive to the imf .
our findings strengthen the case for radial variations within an effective radius of the imf in massive etgs , further supporting the idea ( mn15a ) that the imf is a local property of etgs . in xsg1 , we find a bottom - heavy population ( @xmath3@xmath43 ) in the innermost regions , gradually changing to a standard kroupa - like imf ( @xmath3@xmath106 ) at @xmath300 ( fig . [
fig : gamma_rad ] ) .
the lack of a measurable gradient in abundance ratios , over the spatial scale where the imf is found to vary , points to a possible relationship between imf and total metallicity in regions with high velocity dispersion , in agreement with @xcite .
in addition we include in the analysis the feh - based wing - ford band , sensitive to very - low mass stars .
a remarkable mismatch is found when adopting an imf with a unimodal ( single power law ) functional dependence .
in contrast , a bimodal imf ( which substitutes the power law at the low mass end with a constant value ) , is fully consistent with all indicators ( fig .
[ fig : feh_rad ] ) .
hence , for the first time , we are able to rule out the unimodal imf based on a pure stellar population analysis ( i.e. beyond the fact that the unimodal stellar
m / l is overly high with respect to dynamical constraints ; see lb13 ) . remarkably , the dynamical analysis based on the jeans anisotropic modelling of @xcite gives a very consistent stellar m / l with respect to the ( bimodal ) model predictions of the spectroscopic method via gravity - sensitive spectral features ( fig .
[ fig : alpha ] ) . from a more technical point of view
, the present work generalizes the empirical approach of lb13 to deal with spectra at different galacto - centric distances . remarkably , once _ empirical solar - scale corrections _
are applied as a function of total metallicity ( rather than velocity dispersion , see lb13 , mn15a ) , we are able to describe fairly well a variety of gravity - sensitive features , including those for which a radial gradient is apparently not detected ( mg4780 ) . in a future work ,
we plan to extend the present analysis , by combining tio- and @xmath11- based results with constraints from ca and na features , to discuss how abundance ratios affect the combined use of these features to constrain the stellar imf .
based on observations made with eso telescopes at the paranal observatory under programmes i d 092.b-0378 and 094.b-0747 ( pi : flb ) .
flb acknowledges the instituto de astrofsica de canarias for the kind hospitality while this paper was in progress .
we thank dr .
j. alcal for the insightful discussions and help with the reduction of xshooter spectra .
we also thank the anonymous referee for his / her useful comments that helped us to improve our manuscript .
we have made extensive use of the sdss database ( http://www.sdss.org/collaboration/credits.html ) .
mc acknowledges support from a royal society university research fellowship .
we acknowledge support from grant aya2013 - 48226-c3 - 1-p from the spanish ministry of economy and competitiveness ( mineco ) .
in addition to the @xmath166 and @xmath167 indices , we also consider the @xmath168 and @xmath169 optical features to constrain the imf in xsg1 ( sec .
[ sec : method ] ) . in this appendix
, we describe the radial behaviour of these features . fig .
[ fig : atio_nocorr ] shows @xmath9 ( see sec . [
sec : tio_profs ] ) , as a function of @xmath168 , at different galacto - centric distances ( red through blue , from the centre outwards ) .
@xmath168 increases with imf slope , slightly more in the bimodal case , in contrast to @xmath166 and @xmath167 , for which the line - strengths of models with @xmath301 correspond exactly to those for @xmath3@xmath113 ( see fig .
[ fig : tio_nocorr ] ) .
the index is independent of age ( see @xcite ) , while it is anticorrelated with metallicity . as shown in fig .
[ fig : atio_nocorr ] , the dependence on @xmath85 $ ] is rather complex , as it is coupled with the imf .
@xmath168 decreases slightly with @xmath85 $ ] for a mw - like imf , while it strongly decreases with @xmath85 $ ] for a bottom - heavy distribution ( see the two black arrows in the figure , for @xmath265=+0.2 $ ] ) . in other words ,
@xmath168 is , _ in general _ , not a good imf - sensitive indicator , as at high metallicity it is expected to have essentially the same value for either a mw - like or a bottom - heavy imf . hence , for a galaxy with a bottom - heavier imf in the centre , with respect to the outer regions as it is the case for xsg1 one would expect a shallow radial gradient in @xmath168 . in contrast ,
[ fig : atio_nocorr ] shows a significant radial variation of @xmath168 in xsg1 . using cvd12 models
, we find that in addition to the anticorrelation with @xmath302 $ ] ( consistent to the one with @xmath85 $ ] in the miuscat models ) , @xmath168 also depends on @xmath0 $ ] and @xmath118 $ ] . as for @xmath166 and @xmath167 ,
the index increases with @xmath0 $ ] .
however , this dependence is not compensated by an anticorrelation with @xmath118 $ ] , as the index is also expected to increase with carbon abundance ( see pink and green solid curves in fig .
[ fig : atio_nocorr ] ) .
notice that since the radial trends of @xmath0 $ ] and @xmath118 $ ] are rather flat in xsg1 , based on cvd12 model predictions , one would not expect any gradient of @xmath168 throughout the galaxy .
the strong radial variation of @xmath168 in fig .
[ fig : atio_nocorr ] seems thus unexplained .
this is further shown in fig .
[ fig : atio_corr ] , where empty circles plot line - strengths corrected to the solar scale with cvd12 models .
one can explain the innermost corrected value with miuscat models but there is no model prediction matching the outermost data - points .
however , one should consider that ( i ) cvd12 predictions for varying abundance ratios are only available for ssps with solar metallicity , and a chabrier imf . in practice
, the effect of @xmath0 $ ] and @xmath118 $ ] might be dependent on metallicity and imf ; ( ii ) @xmath168 is a broad feature , and its sensitivity to @xmath118 $ ] may be more uncertain , _ as it is entirely due to a variation of the continuum in the cvd12 models _
( see green dashed line in fig .
[ fig : atio_corr ] ) ; ( iii ) in our data , the absolute value of @xmath168 is affected by the time - varying response of the uvb vis x - shooter dichroic .
we measured this effect by computing the rms of @xmath168 values , in the central spectrum of xsg1 , among different exposures ( see grey double - sided arrow at top
right of fig .
[ fig : atio_corr ] ) . to gain insight into these issues , we also plot in fig .
[ fig : alpha ] the empirically abundance - corrected values of @xmath168 ( see empty squares in the figure ) . indeed , following the approach of lb13
, we find that , independently of the bin in velocity dispersion from the sdss spectra , @xmath168 increases with @xmath0 $ ] , as @xmath303 .
this implies a shift of @xmath304 mag ( for @xmath0$]@xmath15 ) for all , but the outermost , uncorrected values of @xmath168 ( fig .
[ fig : atio_nocorr ] ) .
however , the value of @xmath305 might be underestimated for the innermost data - points of xsg1 . at fixed velocity dispersion in the sdss data , there is also a mild increase ( @xmath149 dex ) of metallicity with @xmath0 $ ] an effect that we have removed following lb13 , by subtracting off the index variation expected from the change in metallicity for kroupa - like miuscat models ( see sec . 5 of lb13 ) .
while this is correct for indices whose sensitivity to @xmath85 $ ] does not depend on imf ( e.g. @xmath166 and @xmath167 ) , it is not for @xmath168 . accounting for this effect ,
would make our empirical correction larger , by a factor of @xmath215 , in the innermost point of xsg1 , implying the corrected value of @xmath168 in the centre to be @xmath193 mag , in very good agreement with results from our best - fitting procedure to applied all optical indices ( sec .
[ sec : gamma_trends ] ; fig .
[ fig : ind_rad ] ) . in summary , we think that the large radial gradient of @xmath168 in our target galaxy is a combination of ( 1 ) an underestimation of our empirical correction at high metallicity ( we are currently working to improve method in this regard ) , and ( 2 ) the uncertainty in the zero - point of the index due to the dichroic response . regarding point ( 1 ) , it does not affect our results ( see sec .
[ sec : gamma_trends ] ) , whereas for point ( 2 ) , we are currently working to improve the treatment of the dichroic issue in the reduction procedure . for the present work ,
we have decided to simply add in quadrature the uncertainty on the dichroic response to the observed error bars on @xmath168 . but plotting @xmath9 as a function of the @xmath168 imf - sensitive feature .
pink and green dashed arrows show the sensitivity of @xmath9 and @xmath168 to @xmath0 $ ] and @xmath118 $ ] , respectively , when removing the continuum from cvd12 models .
the black two - sided arrow in the top - right of the plot shows the uncertainty in the absolute value of @xmath168 because of the time - variable response of the dichroic of the x - shooter spectrograph in the vis arm .
, width=321 ] fig .
[ fig : mgf_nocorr ] plots the @xmath9 as a function of @xmath169 .
interestingly , no measurable radial gradient is found in @xmath169 .
the index tends to increase radially , from the innermost to the second outermost bin , while the outermost bin value is fully consistent with the central one ( see blue and red solid circles in the figure ) . using cvd12 models
, we find that the line - strength of @xmath169 is expected to increase with @xmath0 $ ] ( mostly because of the sensitivity to @xmath139 $ ] , consistent with @xcite ) and decrease with @xmath118 $ ] , similarly to @xmath166 and @xmath167 .
subtracting off the expected variation of the index from our @xmath0 $ ] and @xmath118 $ ] estimates ( fig .
[ fig : abund ] ) with cvd12 models , we obtain the `` theoretically - corrected '' values ( empty circles ) in fig .
[ fig : mgf_nocorr ] , still showing ( as expected ) no significant radial gradient . notice that the theoretically - corrected line - strengths are only consistent with a kroupa - like imf ( especially in the innermost bin ) and an age of @xmath306@xmath307gyr , inconsistent with results for @xmath166 and @xmath167 .
mn15a already noticed that @xmath169 is also dependent ( to a lesser extent than @xmath0 $ ] and @xmath118 $ ] ) on other single - element abundance ratios , i.e. @xmath267 $ ] and [ si / fe ] .
in particular , @xmath169 is expected to be _ anti - correlated _ with @xmath267 $ ] because of its sensitivity to sodium on the blue pseudo - continuum , as illustrated by the cyan arrow in fig .
[ fig : mgf_nocorr ] .
hence , a large @xmath267 $ ] radial - gradient would tend to cancel out the effect of an imf - gradient on @xmath169 .
the empty squares in fig .
[ fig : mgf_nocorr ] illustrate the results of our empirical correction procedure .
as discussed in lb13 , quite unexpectedly , the index tends to increase with @xmath0 $ ] in the sdss stacks at the lowest velocity dispersion ( metallicity ) , while it
_ decreases _ with @xmath0 $ ] at the other end . as a result ,
the range of values for @xmath169 as a function of @xmath39 ( see panel b in fig . 8 of lb13 ) , is significantly larger at high @xmath0 $ ] , than at @xmath0$]@xmath308 .
this is fully consistent with the fact that , at such high @xmath0 $ ] ( @xmath15 ) , as is the case in xsg1 , we do not see any radial gradient in @xmath169 .
furthermore , in the massive etg analyzed in mn15a we saw @xmath169 to feature a shallower radial gradient than other imf indicators ( mostly @xmath167 ) .
remarkably , our empirical correction performed as a function of metallicity
recovers a clear radial gradient in xsg1 , with values fully consistent with those obtained from the other optical indicators ( see sec . [
sec : gamma_trends ] . perhaps , the inconsistency between theoretical and empirical corrections for @xmath169 arise from the fact that some other elements ( e.g. sodium ) also give a prominent contribution to the observed line - strengths , and/or the fact that the effect of @xmath0 $ ] and @xmath118 $ ] is strongly dependent , for this index , on @xmath85$] something that can not be tested with the available cvd12 models . but ( i ) correcting @xmath168 and @xmath9 for their expected sensitivity to @xmath0 $ ] and @xmath118 $ ] ( see circles and dashed line ) , and ( ii ) correcting @xmath168 for its sdss - based sensitivity to abundance ratios , as in lb13 ( see squares and solid line ) . ,
width=321 ] modigliani , a. , goldoni , p. , royer , f. , et al .
2010 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 7737 , observatory operations : strategies , processes , and systems iii , ed .
d. r. silva , a. b. peck , & t. soifer | at present , the main challenge to the interpretation of variations in gravity - sensitive line strengths as driven by a non - universal initial mass function ( imf ) , lies in understanding the effect of the other population parameters .
most notably , @xmath0$]-enhanced populations or even departures in the _ individual _ element abundances with respect to the solar - scaled ratio may lead to similar observational results .
we combine various tio - based , imf - sensitive indicators in the optical and nir spectral windows , along with the feh - based wing - ford band to break this degeneracy .
we obtain a significant radial trend of the imf slope in xsg1 , a massive early - type galaxy ( etg ) , with velocity dispersion @xmath1kms@xmath2 , observed with the vlt / x - shooter instrument .
in addition , we constrain for the first time both the shape and normalization of the imf , using only a stellar population analysis .
we robustly rule out a single power - law to describe the imf , whereas a power law tapered off to a constant value at low masses ( defined as a bimodal imf ) is consistent with all the observational spectroscopic data _ and _ with the stellar m / l constraints based on the jeans anisotropic modelling method .
the imf in xsg1 is bottom - heavy in the central regions ( corresponding to a bimodal imf slope @xmath3@xmath43 , or a mass normalization mismatch parameter @xmath52 ) , changing towards a standard milky - way like imf ( @[email protected] ; @xmath51 ) at around one half of the effective radius .
this result , combined with previous observations of _ local _ imf variations in massive etgs , reflects the varying processes underlying the formation of the central core and the outer regions in this type of galaxies .
[ firstpage ] galaxies : stellar content galaxies : fundamental parameters galaxies : formation galaxies : elliptical and lenticular , cd |
it has recently been proposed that it may be possible to observe neutral atomic hydrogen gas during the epoch of reionization ( eor ) , in absorption against the cosmic microwave background ( cmb ) in its 3-cm fine - structure line ( @xmath4 , * ? ? ? * hereafter s07 ) .
the possible detectability of the 3-cm line derives from the fact that the 2s state of atomic hydrogen is metastable @xcite .
that is , the average lifetime of an undisturbed atom in the 2s - state is @xmath9 s , after which it decays into the 1s state by emitting 2 photons @xcite . for comparison ,
the mean life - time of atoms in the 2p state is @xmath10 s. hence , when atoms are excited into their 2s and 2p states at comparable rates , the 2s level is expected to be highly overpopulated relative to the 2p level .
this may result in a non - negligible optical depth in the @xmath11 fine - structure line ( see e.g. * ? ? ? * ) .
the detectability of this line has been discussed in the context of hii regions in the local universe ( e.g. * ? ? ? * for a more detailed discussion ) . in local hii regions ,
the 2s level is populated mainly following recombination , while both recombination and photoexcitation by ly@xmath12 photons populate the 2p state .
the ly@xmath12 excitation rate of the 2p - state is limited by dust , and the 2s - state is typically heavily overpopulated in hii regions which result in 3-cm optical depths of order @xmath13 ( ershov 1987 ; dennison et al , 2005 , where the absorption is measured relative to the free - free continuum that is produced in the hii region ) .
s07 have argued that during the eor , bright quasars that are luminous in the restframe soft - uv ( 10.2 ev @xmath14 13.6 ev ) may indirectly photoexcite the 2s state of neutral hydrogen gas in the low density intergalactic medium to levels such that @xmath15 for cmb photons passing through these regions levels of atomic hydrogen can produce a detectable opacity .
one can estimate @xmath16 by comparing it to the gunn - peterson optical depth @xmath17 : @xmath18 @xmath19 .
since @xmath20 a few times @xmath21 for a fully neutral igm at @xmath22 , @xmath23 only requires that @xmath24 . it will be shown in [ sec : sim ] that these levels can be reached in regions around high - redshift quasars . ] .
absorption of these cmb photons would result in an absorption feature with a brightness temperature of a few tens of @xmath6k at @xmath25 ghz , which may be detectable with existing dish antennas such as the _ green bank telescope _ ( gbt ) , _ parkes _ and the most compact configuration of the _ australia telescope compact array _ ( atca , e.g. * ? ? ?
this result is important , because it implies that neutral intergalactic atomic hydrogen gas that exists during eor may reveal itself in a line that is different ( and possibly easier to detect ) than the well studied 21-cm hyperfine transition ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . motivated by this exciting result , and by observational efforts to detect this transition ,
we re - analyze the detectability of the fine - structure line during the eor .
we perform more detailed modeling of the transfer of ly@xmath26 radiation than originally discussed by s07
. we will show that ( unfortunately ) proper modeling of ly@xmath26 ( and ly@xmath27 ) radiative transfer radically lowers the value of @xmath16 to levels that can not be probed with existing radio telescopes .
the outline of this paper is as follows .
we review the dominant excitation mechanisms of the 2s and 2p levels in [ sec : mech ] . in [ sec : ex ] , we compute the effective excitation rate of the 2s - state during the eor , by properly accounting for radiative transfer , and discuss the detectability of the 3-cm fine - structure line in [ sec : detection ] . the validity of our model assumptions , and the implications of this work are discussed in [ sec : discuss ] . finally , we present the conclusions in [ sec : conc ] .
the parameters for the background cosmology used throughout are @xmath28 @xcite .
figure [ fig : diagram ] depicts the first three principle energy levels of the hydrogen atom . on the left , fine and hyperfine - structure
splitting of the lines are ignored .
the notation that is used is @xmath29 , where @xmath30 is the principle quantum number , and @xmath31 denotes the orbital angular momentum number ( @xmath32 corresponds to @xmath33 , @xmath34 corresponds to @xmath35 , @xmath36 corresponds to @xmath37 ) . in most astrophysical conditions , the vast majority of hydrogen atoms are in the @xmath38-state .
atoms can be excited out of the ground state through the following processes : ( _ i _ ) collisional excitation , ( _ ii _ ) recombination following photoionization or collisional ionization , and ( _ iii _ ) photoexcitation .
following s07 , processes ( _ i - ii _ ) are ignored ( see [ sec : sim ] - under roman numeral i -for a more detailed motivation ) .
the quantum mechanical selection rules state that the only atomic transitions that are allowed by emitting or absorbing a photon are those in which @xmath39 changes by @xmath40 .
hence , absorption only results in transitions of the form @xmath41 , while subsequent emission only results in @xmath42 .
the allowed transitions among the first 3 levels of the h - atom are indicated in figure [ fig : diagram ] as _ dotted lines_. in a 3-level atom ( the impact of using a multilevel atom is discussed in [ sec : sim ] under roman numeral ii ) the selection rules only allow the 2s state to be populated by absorbing a ly@xmath26 photon , and by subsequently emitting an h@xmath12 photon .
hence , the level population of atoms in the 2s state is determined by the _
effective ly@xmath26 scattering rate _ , denoted by @xmath43 , which is defined as the rate at which ly@xmath26 photons absorbed , and subsequently converted into h@xmath12 + 2 continuum photons .
this rate is lower than to the `` ordinary '' ly@xmath26 scattering rate , @xmath44 , which is the rate at which ly@xmath26 photons are absorbed .
as mentioned above , the effective ly@xmath26 scattering rate , @xmath45 , is crucial in calculating the number of atoms in the 2s - state .
the goal of this section is to evaluate @xmath46 . in [ sec : mc ] we describe the monte - carlo technique that was used to compute the general and effective ly@xmath26 scattering rates , and in [ sec : result ] we present our results .
we compute the general and effective ly@xmath26 scattering rate by performing monte - carlo ( mc ) calculation of the ly@xmath26 radiative transfer in the neutral gas surrounding a spherical hii region of radius @xmath47 mpc , that is based on the physical size of the hii region that possibly exist around @xmath22 quasars ( * ? ? ?
* ; * ? ? ?
* but see lidz et al 2007 , bolton & haehnelt 2007 ) .
this mc - approach to radiative transfer , in which the trajectories of individual photons are followed , is common for ly@xmath12 radiation ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we use the ly@xmath12-mc code originally described in @xcite , which is easily adapted for ly@xmath26 radiative transfer .
the calculation is performed as follows : * 1 . *
a photon is emitted with a frequency that is drawn randomly from the range @xmath48 $ ] .
the transfer calculation therefore focuses on photons in a frequency range @xmath49 centered on @xmath50 , similar to the optically thin calculation that was originally performed by s07 ( and which is discussed in appendix [ sec : lya1 ] ) .
the photon propagates an optical depth @xmath51 , where @xmath52 is a random number between 0 and 1 .
the optical depth translates to a physical distance via @xmath53)$ ] .
each time a ly@xmath26 photon is absorbed , there is a @xmath54 probability for the atom to decay directly back to the ground state by emitting a ly@xmath26 photon @xcite .
this is simulated in the mc - calculation by generating a random number @xmath55 $ ] .
the photon scatters as a ly@xmath26 photon if @xmath56 , in which case a new direction is chosen and we return to step * 2*. on the other hand , a transition of the form @xmath57 occurs when @xmath58 .
after this process , the ly@xmath26 photon is destroyed , and we return to step * 1.*. we compute total ly@xmath26 scattering rate by keeping track of the number of scattering events that occur in shell @xmath59 , @xmath60 ( in the mc - simulation , the gas is sampled by a discrete number of gas shells , see dijkstra et al 2006 ) . the total ly@xmath26 scattering rate transfer in shell @xmath59 follows from @xmath61 , where @xmath62 is the total rate at which the quasar is emitting photons in the frequency range @xmath48 $ ] ( we have assumed @xmath63 s@xmath64 to compare to s07 as closely as possible ) , @xmath65 is the total number of hydrogen atoms in shell @xmath59 , and where @xmath66 is the total number of photons that is used in the mc - simulation . the effective ly@xmath26 scattering rate is obtained similarly from @xmath67 , where @xmath68 is the total number of scattering events of the sequence @xmath69 in shell @xmath59 .
the general and effective ly@xmath26 scattering rates we obtain from the monte - carlo calculation are shown in figure [ fig : lybscatrate ] as the _ black _ and _ red solid histograms _ , respectively .
the _ blue dotted line _ shows the scattering rate that is obtained when the gas is assumed to be optically thin ( which corresponds to the original calculation of s07 , see appendix [ sec : lya1 ] for more details ) .
furthermore , the _ blue dashed line _ shows the scattering rate that one obtains if one ignores that ly@xmath26 photons undergo multiple scattering events , and simply suppresses the incoming ly@xmath26 flux by a factor of e@xmath70 ( this calculation is described in appendix [ sec : noscat ] ) .
figure [ fig : lybscatrate ] shows that the effective ly@xmath26 scattering rate is much lower than the rate that is obtained in the optically thin calculation . at small distances into the hi region ( @xmath71 kpc , which is shown in the inset ) ,
the effective scattering rate is @xmath72 s@xmath64 .
the effective scattering rate decreases rapidly with radius , and levels off at a level that is @xmath73 orders of magnitude smaller than the rate one obtains from the optically thin calculation .
this rapid decrease owes to the large line - center optical depth of the neutral igm to ly@xmath26 photons , which causes resonant scattering only to occur near the edge of the hii region .
this point is illustrated by the _ green dotted line _ in the main plot of figure [ fig : lybscatrate ] , which shows the scattering rate of photons with frequencies such that they would have redshifted into resonance at @xmath74 mpc ( i.e. 3 mpc into the neutral igm ) . when these photons reach @xmath75 mpc , they find themselves @xmath76 km s@xmath64 blueward of the resonance , and their absorption cross - section is suppressed by a factor of @xmath77 relative to the resonance absorption cross - section .
the scattering rate of these photons is therefore also suppressed by a factor of @xmath78 ( relative to the optically thin rates ) .
the scattering rate increases rapidly with radius as the photons redshift closer to resonance , and the maximum scattering rate occurs at @xmath79 mpc , where the photons have redshifted to within @xmath80 doppler widths from resonance .
the optical depth of the igm increases rapidly as these photons redshift closer to resonance , which results in a sharp cut - off in the scattering rate at @xmath81 mpc .
in other words , photons that were emitted somewhat blueward of the ly@xmath26 resonance frequency -which need to travel far into neutral gas from the hii region edge before redshifting into the line center- have a high probability of being absorbed in the damping wing of the line , and do not generally make it to line center .
the absorption cross section in the damping wing is low compared to that at line center , and so these photons yield a correspondingly small scattering rate
. therefore the ly@xmath26 scattering rates are strongly suppressed at @xmath82 kpc , i.e. photons are conserved . on the other hand , @xmath83 in the optically thin limit ( [ sec : lya1 ] ) .
each emitted ly@xmath26 photon can only undergo one effective scattering event , and the total effective scattering rate can never exceed the total rate at which ly@xmath26 photons are emitted .
the optically thin calculation is therefore logically inconsistent . ] .
the sharp cut - off of the effective ly@xmath26 scattering rate is also discussed in appendix [ sec : noscat ] .
furthermore , figure [ fig : lybscatrate ] shows that the effective scattering rate that is extracted from the mc - simulation ( _ red solid histogram _ ) is well reproduced by the model in which the mean ly@xmath26 intensity is suppressed by a factor of e@xmath70 ( the _ blue dashed line _ ) .
the reason for this is that each ly@xmath26 photon can only excite 1 atom into the 2s state : each time a transition of the from @xmath84 occurs , the ly@xmath26 photon is lost permanently .
this almost corresponds to a calculation in which one ignores scattering and simply suppresses the intensity with a factor e@xmath70 .
the only difference is that ly@xmath26 photons typically scatter @xmath85 times before being converted into h@xmath12 + 2 continuum photons .
the random walk associated with the scattering process transports the ly@xmath26 photons on average @xmath86 mean free paths through the neutral igm , before indirectly exciting an atom into its 2s - state .
in practise , this introduces a difference with the scattering rate obtained from the e@xmath70-calculation only at the smallest separations from the edge of the hii region ( @xmath87 kpc , see inset ) . at these smallest radii ,
the gas is optically thin to ly@xmath26 photons , and here the _
ly@xmath26 scattering rate ( _ black solid histogram _ ) as extracted from the mc - simulation agrees best with the analytic solution given by eq ( [ eq : pscat2 ] ) with @xmath88 .
naturally , here the _ effective _ scattering rate is suppressed by a factor of @xmath89 .
the change in brightness temperature for cmb radiation passing through the neutral gas with an enhanced population of atoms in the 2s - state is given by @xmath90 where @xmath91 is the cmb temperature at redshift @xmath92 , and @xmath93 is the excitation temperature of the 2s-2p transition , which is defined as @xmath94 , where @xmath95 is the standard source function ( see * ? ? ?
* eq 1.79 ) , @xmath96 ( @xmath97 ) denotes the number density of atoms in the 2s level ( 2p level ) , and @xmath98 ( @xmath99 ) denotes the statistical weight of this level . throughout
we ignore that hyperfine splitting of the 2s and 2p levels actually results in 3 possible 2s-2p transitions ( see fig [ fig : diagram ] ) .
we discuss the impact of hyperfine splitting on our results in
[ sec : sim ] ( under roman numeral v ) .
furthermore , the optical depth to 2s-2p absorption , @xmath16 is given by ( * ? ? ?
* eq 1.78 ) @xmath100 where @xmath101 , @xmath102 is the einstein b - coefficient for the fine - structure transition , and @xmath103 is the line profile function ( see * ? ? ? * ) .
s07 move the factor @xmath104 outside of the integral and arrive at their eq ( 7 ) .
however this is not allowed , because we found @xmath105 , and therefore @xmath106 , to be a rapidly changing function of radius ( see fig [ fig : lybscatrate ] ) .
this makes the evaluation of @xmath107 somewhat trickier .
fortunately , we may simplify eq [ eq : tb ] and eq [ eq : tau2s ] by assuming that @xmath108 ( which is verified in [ sec : sim ] , under roman numeral iv ) . under this assumption , @xmath109 and @xmath110 , and the optical depth to 2s-2p absorption at frequency
@xmath111 simplifies to @xmath112),\ ] ] where @xmath113 . for simplicity
, we assume that the edge of the hii region is infinitely sharp ( i.e. the hii region contains no neutral hydrogen , and the neutral fraction jumps from @xmath114 to @xmath115 instantly at the edge of the hii region ) and extends out to @xmath116 mpc ( see [ sec : sim ] , roman numeral iii ) . under this assumption
, the optical depth along a path that is pointing radially outward from the quasar is given by @xmath117 ) , \label{eq : tau2sf}\ ] ] where we used that @xmath118 .
the last approximation is allowed because the vast majority of all hydrogen atoms are in the 1s state .
the life - time of the 2p state is set by the einstein coefficient of the @xmath119 transition , @xmath120 s@xmath64 , and the line profile of the 2s-2p transition has a voigt parameter @xmath121 @xcite .
the line profile is therefore completely dominated by the damping wing , and the cross - section displays a much weaker frequency dependence than is common among the lyman - series transitions : e.g. even for @xmath122 , where @xmath123 , the cross - section is only reduced by a factor of @xmath124 .
we compute @xmath107 by using @xmath105 we found in [ sec : result ] ( and we account for the additional contribution of higher lyman series following [ sec : sim ] ) into eq [ eq : tau2sf ] . the result of this calculation is shown in figure [ fig : tau ] , which shows that the maximum optical depth is @xmath125 .
this is 5 orders of magnitude smaller than the value that was found by s07 .
the reason for this large difference is the assumption that the factor @xmath104 could be taken outside the integral in eq [ eq : tau2s ] .
this is allowed only when the dependence of this factor on @xmath126 is weaker than that of @xmath127)$ ] . however , because the 2s-2p absorption cross - section is completely dominated by the damping wing , it has a tremendous fwhm that is @xmath128 . at @xmath129 , hubble expansion over
4 mpc introduces a comparable frequency shift of @xmath130 , and s07 therefore implicitly assumed that their calculated level population in the 2s - state applied to a shell that was at least @xmath131 mpc thick . in sharp contrast , we found that their calculation applied only to a shell of gas that was @xmath132 kpc thick ( see the inset of fig [ fig : lybscatrate ] ) . in other words , the total column of neutral h atoms in the 2s state
is lower by @xmath1335 orders of magnitude .
it is worth emphasizing that our calculation of @xmath134 assumed that the quasar emitted @xmath63 photons s@xmath64 in the frequency range @xmath135\nu_{\beta}$ ] .
this is very likely a factor of @xmath136 too high ( see appendix [ sec : lya1 ] ) .
therefore , in reality the value of @xmath137 is likely even lower by an additional factor of @xmath136 .
( i ) : in [ sec : mech ] we mentioned that the collisional excitation rate of -and the recombination rate into- the 2s state , were negligibly small compared to the effective ly@xmath26 scattering rate . for example
, the collisional excitation rate is given by @xmath138 ( * ? ? ?
* p54 - 55 ) .
if we assume that the gas is half ionized ( @xmath139 ) , then we find that @xmath140 s@xmath64 for @xmath141 k. this rate is orders of magnitude less than the effective ly@xmath26 scattering rates in the regions of interest .
similarly , we find that the recombination rate is @xmath142 s@xmath64 , which is even smaller .
( ii ) : so far we have ignored the fact that the 2s state of atomic hydrogen can also be accessed following the absorption of higher order lyman series photons , e.g. via a scattering event of the form @xmath143 . in principle , it is straightforward to expand the mc - approach to include scattering of these higher - order lyman series photons , but this is not necessary .
we found in
[ sec : result ] that the effective ly@xmath26 scattering rate was calculated reasonably well if one simply suppresses the incoming ly@xmath26 flux with a factor of e@xmath70 . for simplicity , we also use this approach to calculate the effective scattering rates of the higher order lyman series photons , and cast the total effective scattering rate in a form similar to that of eq ( [ eq : pscat2 ] ) : @xmath144 where @xmath145 is the oscillator strength , @xmath146 is the energy , and @xmath147 is the voigt function for the lyman - n transition ( i.e. @xmath41 ) , where @xmath148 .
the first term of the series corresponds to eq ( [ eq : pscat2 ] ) .
the result of summing the first 10 lyman series transitions is shown as the _ green dot - dashed line _ in figure [ fig : lybscatrate ] . including these higher lyman series photons only boosts the effective scattering rates by a factor of @xmath149 in the region that was optically thin to ly@xmath26 photons , which is due to the fact that the oscillator strength of @xmath150 transitions -and hence the absorption cross section- decreases quite rapidly with @xmath30 ( e.g. for @xmath151 , @xmath152 , which is @xmath153 ) . between @xmath154 kpc from the edge of the hii region
, the contribution from higher lyman series photons is the largest and boosts the overall scattering rate by a factor @xmath155 .
this larger boost is due the fact that at these distances the gas is not yet optically thick in these higher lyman series lines .
note that we have accounted for this boost in our calculation of @xmath16 ( which is shown figure [ fig : tau ] ) .
( iii ) : we have so far assumed that the edge of the hii region is infinitely sharp . in reality , the thickness of the hii region is determined by the mean free path of the ionizing photons emitted by the quasar .
especially the x - rays emitted by the quasar can penetrate deeply into the neutral igm and produce a low level of ionization .
however , the presence of a partially neutral region does not affect our results at all : although the distance from the edge of the hii region at which the igm becomes optically thick is increased by a factor of @xmath156 ( where @xmath157 is the neutral fraction ) , @xmath16 is reduced by the same factor .
( iv ) : the approximation @xmath158 is valid only when @xmath109 , or when @xmath159 .
the 2s and 2p populations are given by @xmath160 and @xmath161 , respectively .
that is , @xmath159 when @xmath162 .
our assumption @xmath159 therefore breaks down in the regions of interest only when @xmath163 s@xmath64 , which is orders of magnitude larger than the actual ly@xmath12 scattering rate .
( v ) : our calculation has ignored the hyperfine splitting of the 2s and 2p levels that was illustrated in figure [ fig : diagram ] .
hyperfine splitting results in 3 allowed 2s-2p transitions at @xmath164 and @xmath165 ghz , and the fine - structure absorption cross - section has maxima at these frequencies .
the most prominent of these 3 maxima reaches @xmath166 of the value that one obtains when hyperfine - structure splitting is ignored @xcite . including hyperfine - structure therefore lowers the maximum observable brightness temperature by a factor of @xmath167 .
we have shown that the optical depth in the 2s-2p fine - structure line around quasar hii regions to be negligible because of the low effective ly@xmath7 scattering rates .
early during the eor , the igm was exposed to a background lyman series photons generated by early generations of galaxies before it was significantly ionized ( e.g. * ? ? ?
* ) . in principle
, one might imagine looking for fine - structure absorption in the igm during this early phase of the eor .
unfortunately , the effective ly@xmath26 scattering rate during this phase of the eor will be low for the same reasons it is low around bright quasars , as we considered previously : it is well established that ly@xmath12 scattering rate plays an important role in determining the excitation temperature of the 21-cm hyperfine transition via the wouthuysen - field effect ( see e.g. * ? ? ?
* ; * ? ? ?
. however , each soft - uv photon that redshifts toward ( and through ) the ly@xmath12 resonance frequency scatters on average @xmath168 times . in sharp contrast
, each higher lyman series photon only excites one atom into the @xmath3-state .
the effective ly@xmath26 ( and total ) scattering rate is therefore lower than the ly@xmath12 scattering rate by a factor of @xmath169 . since the ly@xmath12 scattering rate in the igm during the eor is typically @xmath170 s@xmath64 ( e.g. * ? ? ?
* ) , the effective ly@xmath26 scattering rate is of the order @xmath171 s@xmath64 , which barely competes with the recombination rate . from this
we conclude that wherever the wouthuysen - field effect operates in the low density neutral igm , the optical depth in the fine - structure line is expected to be @xmath172 .
it has recently been proposed that neutral intergalactic atomic hydrogen gas may be detected in absorption in its 3-cm fine - structure line ( @xmath4 ) against the cosmic microwave background ( cmb ) out to very high redshifts . in particular
, bright quasars that are luminous in the restframe soft - uv ( 10.2 ev @xmath14 13.6 ev ) may indirectly photoexcite the 2s state of neutral hydrogen gas in the intergalactic medium ( igm ) during the epoch of reionization ( eor ) to levels such that @xmath5 for cmb photons passing through these regions @xcite .
the resulting brightness temperature of @xmath173 tens of @xmath6k could be detected with existing radio telescopes ( e.g. * ? ? ?
* ) . motivated by this proposal , and by observational efforts to detect this transition , we have performed a detailed analysis of the transfer of ly@xmath7 radiation , and have re - analyzed the detectability of the fine - structure line in neutral intergalactic gas surrounding bright quasars during the eor .
we have found that properly considering radiative transfer effects reduces the signal significantly compared to optically thin estimates , which radically complicates the detectability of the 3-cm fine structure transition .
the main reason for this negative result is the large opacity of the ( partially ) neutral igm to lyman series photons .
in such a medium , ly@xmath12 scattering proceeds completely different than ly@xmath7 scattering : ly@xmath12 photons scatter @xmath174 times .
however , higher lyman series photons have a finite probability for being converted into balmer , paschen , .. , etc photons at each scattering event , which subsequently propagate to the observer unobstructed . these higher lyman series photons therefore scatter on average only @xmath175 times ( [ sec : result ] ) before being destroyed @xcite . furthermore
, only one of these @xmath175 scattering events - defined as an _ effective scattering event_- indirectly excites atoms into their 2s - state ( e.g. ly@xmath26 scattering corresponds to a sequence @xmath176 , and does not affect the level population of the @xmath3-state ) . for these reasons
, we found that the effective ly@xmath26 scattering rate was calculated reasonably well if one simply suppresses the incoming ly@xmath26 flux by a factor of e@xmath70 , in which @xmath177 is the ( frequency dependent ) optical depth of the igm to ly@xmath26 radiation ( see fig [ fig : lybscatrate ] ) .
the same applies to ly@xmath27 radiation .
because of the reduced effective scattering rates , we found a substantial lower column of neutral atoms in the 2s - state .
this , in combination with the fact that the profile of the fine - structure absorption cross section is dominated by the damping wings ( [ sec : detection ] ) , results in @xmath178 .
this is 5 - 6 orders of magnitude smaller than what was originally found by @xcite , and we conclude that the 3-cm fine - structure absorption line from neutral intergalactic gas surrounding high - redshift quasars is presently undetectable . * acknowledgments * this research was supported by harvard university funds .
jrp is supported by nasa through hubble fellowship grant hst - hf-01211.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
smo is supported by nsf / ast0457585 .
14 natexlab#1#1 ahn , s .- h . , lee , h .- w . , & lee , h. m. 2000 , journal of korean astronomical society , 33 , 29 bolton , j. s. , & haehnelt , m. g. 2007 , mnras , 381 , l35 breit , g. , & teller , e. 1940 , apj , 91 , 215 carilli , c. l. 2008 , arxiv e - prints , 802 , arxiv:0802.1727 chuzhoy , l. , & zheng , z.2007 , apj , 670 , 912 dennison , b. , turner , b. e. , & minter , a. h. 2005 , apj , 633 , 309 dijkstra , m. , haiman , z. , & spaans , m. 2006 , apj , 649 , 14 dijkstra , m. , & loeb , a.2008 , mnras , 386 , 492 ershov , a. a. 1987 , soviet astronomy letters , 13 , 115 fan , x. , et al .
2006 , aj , 132 , 117 furlanetto , s. r. , oh , s. p. , & briggs , f. h. 2006 , physics reports , 433 , 181 hirata , c. m. 2006 , mnras , 367 , 259 komatsu , e. , et al . 2008 , arxiv e - prints , 803 , arxiv:0803.0547 lidz , a. , mcquinn , m. , zaldarriaga , m. , hernquist , l. , & dutta , s. 2007 , apj , 670 , 39 loeb , a. , & rybicki , g. b.1999 , apj , 524 , 527 loeb , a. , & zaldarriaga , m. 2004 , physical review letters , 92 , 211301 madau , p. , meiksin , a. , & rees , m. j. 1997 , apj , 475 , 429 mesinger , a. , & haiman , z.2004 , apjl , 611 , l69 osterbrock , d. e. 1989 , _ astrophysics of gaseous nebulae and active galactic nuclei _ , university of minnesota , et al .
mill valley , ca , university science books , 1989 , 422 p. , pritchard , j. r. , & furlanetto , s. r. 2006 , mnras , 367 , 1057 rybicki , g. b. , & lightman , a. p. 1979
, new york , wiley - interscience , 1979 .
393 p. , scott , d. , & rees , m. j. 1990 , mnras , 247 , 510 sethi , s. k. , subrahmanyan , r. , & roshi , d. a. 2007 , apj , 664 , 1 spitzer , l. j. , & greenstein , j. l. 1951 , apj , 114 , 407 wild , j. p. 1952 ,
apj , 115 , 206 wyithe , j. s. b. , & loeb , a.2004 , _ nature _ , 427 , 815 zheng , z. , & miralda - escud , j. 2002 , apj , 578 , 33
where @xmath180 is the mean intensity in the uv radiation field near the ly@xmath26 resonance ( in erg s@xmath64 @xmath181 hz@xmath64 sr@xmath64 ) , and @xmath182 is the ly@xmath26 absorption cross - section .
the mean intensity @xmath180 relates to the luminosity of the central source ( in erg s@xmath64 hz@xmath64 ) as @xmath183 , where @xmath126 is the distance to the central source . assuming that @xmath180 is constant across the ly@xmath26 resonance , which is extremely narrow in frequency , eq [ eq : pscat ] simplifies to where the photon energy @xmath185 , that was originally in the integrand , was taken outside of the integral since it barely varies over the limited range of frequencies where the integrand is not negligibly small . using that @xmath186 , where @xmath187 is the ly@xmath26 oscillator strength
, @xmath188 is the speed of light , @xmath189 and @xmath190 are the electron charge and mass , respectively ( see e.g. * ? ? ? * their eq 3.66 ) , the total ly@xmath26 scattering rate simplifies further to the only quantity that remains to be determined is @xmath192 . under the assumption that @xmath193 does not vary with frequency near the ly@xmath26 resonance frequency , @xmath194 , where @xmath31 is the total luminosity ( in erg s@xmath64 ) of the central source between frequency @xmath195 and @xmath196 , and @xmath197 .
s07 assume that erg s@xmath64 hz@xmath64 , which corresponds to an absolute ab magnitude of @xmath198 . for comparison ,
the brightest @xmath22 quasars have @xmath199 @xcite . given that quasars have a relatively flat spectrum between @xmath200 and @xmath201 (rest - frame ) ,
we conclude that the quasar luminosity assumed by s07 is likely too high by about a factor of @xmath136 . ]
@xmath63 photon s@xmath64 and @xmath202 , and obtain @xmath203 where we used that @xmath204 erg s@xmath64 , and @xmath205 mpc ( [ sec : mc ] ) .
this scattering rate was plotted as the _ dotted line _ in figure [ fig : lybscatrate ] . where @xmath207 is the optical depth to photons _ emitted _ at frequency @xmath111 to a hydrogen atom at a distance @xmath126 from the central source .
furthermore , the factor @xmath208 $ ] accounts for the fact that the hubble expansion redshifts the photons by a factor of @xmath209 over a distance @xmath126 .
eq [ eq : pa1 ] does not account for the fact that individual ly@xmath26 photons are scattered multiple times , which modifies the actual value of mean intensity ( see [ sec : result ] ) . here
, @xmath213 denotes a dimensionless frequency variable of the form @xmath214 , where @xmath215 and @xmath216 .
furthermore , @xmath217 is the voigt function ( see eq 10.77 of rybicki & lightman 1979 .
note that this function obeys @xmath218 .
hence , we recover eq [ eq : pscatb ] when we set @xmath88 . )
the function @xmath211 quantifies the inaccuracy in assuming that the absorbing gas is optically thin .
we again assume that the edge of the hii region is infinitely sharp ( as in [ sec : detection ] ) and extends out to @xmath116 mpc ( see [ sec : sim ] , roman numeral iii ) .
the optical depth @xmath219 is then @xmath220 ) .
\label{eq : tau}\ ] ] we plot the suppression factor @xmath211 in figure [ fig : lyarates ] , which shows that @xmath221 for @xmath222 mpc .
the reason that @xmath211 drops so rapidly can be understood as follows : regardless of their frequency , photons traverse by definition an average optical depth @xmath223 into the neutral intergalactic medium before being absorbed . for ly@xmath26 photons
, this implies that the vast majority of photons are absorbed well before they have redshifted into resonance .
that is , @xmath224 when @xmath225 and therefore @xmath226 , in the notation of eq [ eq : sr ] .
only those photons that have redshifted into resonance at the very edge can scatter near resonance ( i.e. @xmath227 while @xmath228 , also see [ sec : result ] ) .
this special region is very thin : the line - center optical - depth of neutral intergalactic gas in the ly@xmath26 line is @xmath229 .
that is , the line center optical depth approaches unity at @xmath230 kpc=@xmath231 mpc .
the scattering rate that is obtained by combining eq [ eq : pscat2]-[eq : tau ] plotted as the _ dashed line _ in figure [ fig : lybscatrate ] . | a soft ultraviolet radiation field , @xmath0 ev@xmath1 ev , that permeates neutral intergalactic gas during the epoch of reionization ( eor ) excites the @xmath2 ( directly ) and @xmath3 ( indirectly ) states of atomic hydrogen . because the @xmath3 state is metastable , the lifetime of atoms in this level is relatively long , which may cause the @xmath3 state to be overpopulated relative to the @xmath2 state .
it has recently been proposed that for this reason , neutral intergalactic atomic hydrogen gas may be detected in absorption in its 3-cm fine - structure line ( @xmath4 ) against the cosmic microwave background out to very high redshifts . in particular , the optical depth in the fine - structure line through neutral intergalactic gas surrounding bright quasars during the eor may reach @xmath5 .
the resulting surface brightness temperature of tens of @xmath6 k ( in absorption ) may be detectable with existing radio telescopes .
motivated by this exciting proposal , we perform a detailed analysis of the transfer of ly@xmath7 radiation , and re - analyze the detectability of the fine - structure line in neutral intergalactic gas surrounding high - redshift quasars .
we find that proper radiative transfer modeling causes the fine - structure absorption signature to be reduced tremendously to @xmath8 .
we therefore conclude that neutral intergalactic gas during the eor can not reveal its presence in the 3-cm fine - structure line to existing radio telescopes .
[ firstpage ] cosmology : theory , diffuse radiation radio lines : general radiative transfer line : profiles |
large libraries of stellar models and isochrones covering wide ranges of age and metallicities are an essential tool to investigate the properties of resolved and unresolved stellar populations .
we have recently published ( pietrinferni et al .
2004 , hereafter paper i ) an extended database of scaled - solar stellar evolution models and isochrones that is very well suited to accomplish this goal .
it is however well known that the scaled - solar metal mixture is not universal . in particular , observations of metal abundance ratios in the galaxy ( i.e. , ryan , norris & bessel 1991 ; carney 1996 ; gratton , sneden & carretta 2004 ; sneden 2004 ; ) have shown that [ @xmath0/fe ] ( where with @xmath0 we denote collectively the so called @xmath0-elements o , ne , mg , si , s , ar , ca and ti ) is larger than zero in population ii field stars , galactic globular clusters ( see , e.g. , carney 1996 ) and in galactic bulge stars ( see , e.g. , mcwilliam & rich 2004 ) that is , the abundance ratio @xmath0/fe is larger than the solar counterpart .
typical values of [ @xmath0/fe ] in the galactic halo are @xmath12 dex . in addition , also stellar populations in elliptical galaxies appear to have formed with a metal mixture characterized by [ @xmath0/fe]@xmath130 ( see , i.e. , worthey , faber & gonzalez 1992 ; tantalo , chiosi & bressan , 1998 ; trager et al
. 2000 ) from a theoretical point of view , the first extended set of isochrones accounting for the enhancement of oxygen was published by bergbusch & vandenberg ( 1992 ) ( and oxygen - enhanced horizontal branch models by dorman 1992 ) ; the enhancement of all @xmath0-element was exhaustively addressed by salaris , chieffi & straniero ( 1993 ) in the regime of low mass , metal poor objects , while weiss , peletier & matteucci ( 1995 ) studied the high metallicity regime .
an important and useful property of @xmath0-enhanced models shown by salaris et al .
( 1993 ) is that , for @xmath0-enhanced mixtures that satisfy the constraint [ ( c+n+o+ne)/(mg+si+s+ca+fe)]=0 ( as for example in case of a constant enhancement of all @xmath0 elements ) low mass stellar models computed with an [ @xmath0/fe]@xmath130 metal distribution are equivalent to scaled - solar ones with the same global metal content @xmath7 ( see also chaboyer , sarajedini & demarque 1992 ) .
this property breaks down at values of z larger than about z=0.002 ( see , e.g. , weiss et al .
1995 , vandenberg & irwin 1997 , salaris & weiss 1998 , salasnich et al .
2000 , vandenberg et al .
2000 , kim et al .
2002 ) .
these theoretical results have been obtained neglecting the effect of an @xmath0-enhancement on the bolometric corrections to be applied to the theoretical isochrones .
very recently , cassisi et al . ( 2004 ) addressed this issue for the first time , showing that broadband colors involving the @xmath14 part of the spectrum are affected by the enhancement of the @xmath0-element , whereas redder wavelength bands are not affected .
both @xmath15 and @xmath16 colors appear bluer than the scaled - solar ones at either the same [ m / h ] or the same [ fe / h ] , the difference increases with increasing metal content and/or decreasing effective temperature .
the best way to mimic @xmath0-enhanced color transformations is to use scaled - solar ones with the same [ fe / h ] , but this equivalence breaks down when the iron content increases above [ fe / h]@xmath171.6 .
these differences are mainly due to the enhancement of mg with lower contributions from the enhancement of si and o. the use of scaled - solar isochrones to mimic their @xmath0-enhanced counterpart is not valid over the full metallicity range where ratios [ @xmath0/fe]@xmath130 have been observed , so that it is important to compute models and isochrones that include appropriate values of [ @xmath0/fe ] . in the last years salaris & weiss ( 1998 ) ,
vandenberg et al . ( 2000 ) , vandenberg ( 2000 ) , salasnich et al . ( 2000 ) , kim et al . (
2002 ) have published sets of @xmath0-enhanced isochrones that cover to different degrees the age and metallicity parameter range .
none of these sets adopt appropriate @xmath0-enhanced bolometric corrections and color transformations . in this paper
we enlarge the parameter space covered by the model and isochrone library presented in paper i , by extending our computations to an @xmath0-enhanced metal distribution , where the [ @xmath0/fe ] value is consistent with observations of the galactic halo population .
this new library covers exactly the same @xmath18 space , and the same mass ranges of our scaled - solar one presented in paper i , and it has been computed with the same stellar evolution code and homogeneous physical inputs .
it includes models computed both with and without overshooting from the convective cores , two different choices for the efficiency of mass loss and , for the first time , appropriate @xmath0-enhanced color transformations and bolometric corrections for both the low- and high @xmath7 regimes .
the paper is organized as follows : 2 briefly summarizes the adopted physical inputs , while the model library is presented in 3 .
comparisons with widely used isochrone databases , and with selected empirical constraints are discussed in 4 , and 5 , respectively .
a summary and final remarks follow in 6 .
we used the same stellar evolution code as in paper i , and the reader is referred to this paper for more information . here
we just describe the new chemical and physical inputs necessary for the computation of the @xmath0-enhanced library .
the adopted @xmath0-enhanced mixture is the same as in the models by salaris , deglinnocenti & weiss ( 1997 ) and salaris & weiss ( 1998 ) and it is reported in table 1 .
the @xmath0-elements have been enhanced with respect to the grevesse & noels ( 1993 ) solar metal distribution ( that is the heavy element distribution used in paper i ) by variable factors , following mainly the results by ryan et al .
( 1991 ) about field population ii stars .
the overall average enhancement is [ @xmath0/fe]@xmath190.4 .
this @xmath0-enhanced distribution has been used in the nuclear network , the radiative ( iglesias & rogers 1996 , alexander & ferguson 1994 ) and conductive opacities ( potekhin et al .
1999 ) and the equation of state .
although our stellar evolution code can account for the atomic diffusion of helium and heavy elements , we decided to neglect this process in our model grid computation .
as largely discussed in paper i , we have accounted for atomic diffusion when computing the standard solar models .
this choice allowed us to properly calibrate the mixing length ( see below ) and to estimate the initial solar chemical composition .
although in the sun atomic diffusion is essentially fully efficient , spectroscopic observations of stars in galactic globular clusters or field halo stars ( see section ii of paper i ) point to a drastically reduced efficiency of this process , due possibly to the counteracting effect of rotationally induced mixing processes .
this evidence raises the possibility that the almost uninhibited efficiency of diffusion found in the sun might not be an occurrence common to all stars .
superadiabatic convection is treated according to the cox & giuli ( 1968 ) formalism of the mixing length theory ( bhm - vitense 1958 ) and the mixing length value is , as in the case of the scaled - solar models , equal to 1.913 , as obtained from a calibration of the solar model ( see paper i for details ) .
all models include mass loss using the reimers formula ( reimers 1975 ) with the free parameter @xmath20 set to two different values : 0.2 and 0.4 , namely @xmath21 . however , later we decided to recompute the model library using also the value @xmath22 .
as for all other models , these additional computations can be found at the url site : http://www.te.astro.it/basti/index.php ] .
we made the same assumptions as in paper i regarding the treatment of convection at the border of convective core during the core h - burning phase of stars with total mass larger than @xmath23 .
two different values are adopted for the overshoot from the schwarzschild convective boundary , namely @xmath24 and 0.2 .
with @xmath25 we denote the length - expressed as a fraction of the local pressure scale height - crossed by the convective cells in the stable region outside the schwarzschild boundary .
the amount of core overshoot is decreased in the same way as in paper i , when the convective core size decreases to zero as a consequence of the decreasing total stellar mass .
the treatment of convection in the core of central he - burning stars is exactly the same of paper i. the possible occurrence of overshooting from the bottom of the convective envelopes has been neglected as in paper i for the reasons discussed in that paper .
finally , the theoretical tracks and isochrones have been transformed to various observational color - magnitude diagrams ( cmds ) using color-@xmath26 transformations and bolometric corrections ( that we will denote collectively as ct ) obtained from an updated version of atlas9 model atmospheres ( castelli & kurucz 2003 ) . the scaled - solar and [ @xmath0/fe]=0.4 ct have been described in paper i and in cassisi et al .
( 2004 ) . in the latter paper
a subset of the @xmath0-enhanced ct has been used to investigate the impact of an @xmath27>0 $ ] metal distribution on broadband colors .
grids of model atmospheres , energy distribution , color indices in the @xmath28 , strmgren @xmath29 and walraven photometric systems have been computed for @xmath30 $ ] : @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 . for a more detailed discussion about the adopted passbands and the model atmosphere grid the reader
is referred to paper i and to cassisi et al .
( 2004 ) . here
we only add that the strmgren colors used in this paper were computed according the prescriptions given by relyea & kurucz ( 1978 ) .
the grid of chemical compositions of our library covers 11 pairs of @xmath18 values , as listed in table 2 , ranging from @xmath7=0.0001 to @xmath7=0.04 .
they are the same @xmath18 values of our scaled - solar library , although the correspondence between @xmath7 and [ fe / h ] is different , due to the effect of the @xmath0-enhancement .
we wish to note that in the scaled - solar library presented in paper i , the @xmath39 metallicity was originally missing ; it has been added later , in order to improve the sampling of the low - metallicity regime .
we have used @xmath40 for the lowest metallicity , together with @xmath41 that allows to match the calibrated initial he abundance in the sun .
we have used the same stellar mass range and sampling as in paper i , to be consistent with the scaled - solar library , and to optimize the use of both sets of models in a stellar population synthesis code . for each chemical composition , we have computed the evolution of up to 41 different stellar masses , with a minimum mass equal to @xmath42 , and a maximum mass always equal to @xmath6 .
we adopted a mass step of @xmath43 ( or lower ) for masses below @xmath44 , @xmath45 for masses in the range @xmath46 , and @xmath47 for more massive models .
models less massive than @xmath48 have been computed starting from the pre - ms phase , whereas the more massive ones have been computed starting from a chemically homogeneous configuration on the ms .
all models , apart from the less massive ones whose central h - burning evolutionary times are longer than the hubble time , have been evolved until the c ignition or after the first few thermal pulses along the asymptotic giant branch ( agb ) phase .
we are already working to extend the evolution of low mass and intermediate mass stars up to the end of the thermal pulses along the agb , using the synthetic agb technique ( see , e.g. , marigo , bressan , & chiosi 1996 and reference therein ) . in case of models undergoing a violent he flash at the red giant branch ( rgb ) tip
, we do not perform a detailed numerical computation of this phase , but we use the evolutionary values of both the he core mass , @xmath49 , and the chemical profile at the rgb tip to compute the corresponding zero age horizontal branch ( zahb ) model ( see vandenberg et al .
2000 for a careful check of the validity of this approach ) .
a subsample of the evolutionary tracks ( models without convective overshoot and @xmath20=0.2 ) is shown in fig .
[ htracks ] for the entire chemical composition range . for each chemical composition
, we have also computed an extended set of he burning models that is , hb models in addition to the one obtained with our chosen mass loss law with various different values of the total stellar mass after the rgb phase , using the fixed core mass and chemical profiles given by an rgb progenitor having an age ( at the rgb tip ) of @xmath50gyr .
these models allow one to compute the cmd of synthetic hb populations , including an arbitrary spread of the amount of mass lost along the rgb phase ( typical of galactic globular clusters ) .
the rgb progenitor of these additional hb models has a mass typically equal to @xmath51 at the lowest metallicities , increasing up to 1.0@xmath52 for the more metal - rich compositions .
table [ zahb ] lists the main characteristics of these hb models , namely , the he core mass at the zahb , the envelope he content , the total mass , luminosity and absolute visual magnitude of the object whose zahb location is at @xmath53 - taken as rapresentative of the mean temperature inside the rr lyrae instability strip . for these additional he - burning computations
we have employed a fine mass spacing , and more than 30 hb models have been computed for each chemical composition . a subset of these models is displayed in fig .
[ hetracks ] .
the same figure shows also the location of the zahb and central he - exhaustion loci .
all the post - zahb computations have been extended either to the onset of thermal pulses for more massive models , or until the luminosity along the white dwarf cooling sequence has decreased down to @xmath54 for less massive objects .
they allow a detailed investigation of the various evolutionary paths that follow the exhaustion of the central he burning .
the individual evolutionary tracks have been reduced to the same number of points , to facilitate the computation of isochrones and their use in a population synthesis code . along each evolutionary track
some characteristic homologous points ( key points kps ) corresponding to well - defined evolutionary phases have been identified .
the choice and number of kps had to fulfill two conditions : 1 ) all main evolutionary phases have to be properly accounted for ; 2 ) the number of kps has to be large enough to allow a suitable sampling of the track morphology , even for fast evolutionary phases .
the second condition requires also that the number of points between two consecutive kps has to be properly chosen . for a careful description of the adopted kps we refer to paper i. in the original version of the database
, we adopted 14 key stages along the evolutionary tracks ( see paper i ) . for our actual stellar models library ,
we have accounted for 16 key stages along the evolutionary track to obtain a more accurate sampling of the track morphology ( we thank a. dolphin for pointing out to us the need of a more accurate sampling of the rgb bump region )
. of course all evolutionary tracks for the scaled - solar metal distribution have been re - reduced by accounting for the same 16 kps .
the whole set of evolutionary computations ( but the additional hb models discussed above ) have been used to compute isochrones for a large age range , namely , from 30 myr up to 16 gyr ( older isochrones can also be computed ) , from the zero age main sequence ( zams ) up to the first thermal pulse along the agb or to the c ignition.100 myr , stars of @xmath55 would be still evolving along the pre - ms phase , whereas in our isochrones they appear along the zams .
to give an idea of the systematics involved , at intermediate metallicities and ages of 30 myr , a 0.5@xmath56 star along our isochrones is fainter by @xmath190.4 mag in @xmath57 and hotter by about 70 k than its real location .
when our evolutionary computations of very low mass stars will be completed , we will consistently include in the isochrones also the pre - ms evolution . ]
the entire library of evolutionary tracks and isochrones is available upon request to one of the authors or can be retrieved at the web site http://www.te.astro.it/basti/index.php . at the same url site
, we have made available a world wide web interface ( more details will be provided in a forthcoming paper , d. cordier et al .
2006 in preparation ) that allows a user to compute isochrones , evolutionary tracks and luminosity functions for any specified stellar mass , chemical composition , age .
detailed tables displaying the relevant evolutionary features for all computed tracks are also available . the files containing the isochrones provide at each point ( 2000 points in total ) the initial value of the evolving mass , the actual mass value ( in principle different , due to the effect of mass loss ) , @xmath58 , @xmath59 , and absolute magnitudes in different photometric systems ( see the discussion in the next section ) .
these isochrones - as well as those for a scaled - solar metal distribution - can be used as input data for our fortran code @xmath60 that we have written in order to compute synthetic cmds , integrated colors and integrated magnitudes of a generic stellar population with an arbitrarily chosen star formation rate and age metallicity relation .
a presentation of this code can be found in paper i and it will not be repeated here . in the near future , we will allow users to run this code through a p.e.r.l based , web interface .
in addition to a chosen star formation history the user will be free to fix various other parameters like photometric and spectroscopic errors , color excess , fraction of unresolved binaries , geometrical depth of the population .
the access to this tool will require registration , the results being sent to the user by e - mail . before closing this section
, we wish to briefly discuss the effect of an @xmath0-enhanced metal distribution on the hb evolution .
although the effect of @xmath1enhanced mixtures on h - burning models is well known and widely discussed in the literature ( see the references in the introduction ) the behavior of @xmath0-enhanced he - burning models compared to their scaled - solar counterpart has been investigated only by salaris et al .
( 1993 ) and vandenberg et al .
( 2000 ) , although the latter studied the effect on the zahb location only . an earlier investigation by bencivenni et al . ( 1989 ) did not take into account the effect of an [ @xmath0/fe]@xmath130 mixture on the hb progenitor evolution .
figure [ hba ] shows the comparison between an @xmath1enhanced hb model in the instability strip and the corresponding scaled - solar one at fixed total mass for various values of @xmath7 . at fixed total mass and global metallicity ,
@xmath1enhanced hb models appear brighter and hotter than the scaled - solar ones , and in particular for values of @xmath7 larger than @xmath190.002 . at @xmath61 ,
the luminosity difference is negligible , @xmath62 , while at @xmath63 , @xmath64 .
these differences increase to @xmath65 and @xmath66 at @xmath67 and @xmath7=0.04 , respectively . these results about the brighter and hotter location of @xmath0-enhanced hb models in the instability strip agree well with the findings by salaris et al .
( 1993 ) and vandenberg et al .
for @xmath680.002 the @xmath1enhanced zahb models have a he core mass lower than the scaled - solar case by only @xmath69 ( regardless of the metallicity ) and an envelope he abundance essentially equal to or lower by at most @xmath700.001 ; this means that the difference in the zahb location and hb evolution is due to the increased efficiency of the cno - cycle in the h - burning shell with respect to the core burning in models with an [ @xmath0/fe]@xmath130 distribution . as for the lifetime along the hb phase for a given @xmath7
, we find that it is only marginally affected by the use of an @xmath1enhanced metal distribution , irrespective of the precise value of @xmath7 .
the central he - burning lifetime decreases compared to the scaled - solar mixture by a negligible @xmath71% at low metallicity , and about @xmath72% at the highest metallicity of our model grid .
the same change holds for the agb lifetime up to the thermal pulse phase .
this has the important consequence that the theoretical calibrations of the @xmath73 parameter ( number ratio of agb stars to hb stars ; see e.g. caputo et al .
1989 ) is not affected by the adopted metal distribution .
this parameter is very important for its sensitivity to the extension of the convective cores during the central he - burning phase , and is used to assess the efficiency of the breathing pulses at the exhaustion of the central he .
a further important evolutionary feature is the brightness of the agb clump , marking the ignition of the he - burning in a shell .
we have verified that the brightness of the agb clump is slightly larger with respect the scaled - solar case , the difference tracking the luminosity difference of the zahb models .
this implies that the brightness difference between the zahb and the agb clump is preserved when passing from a scaled - solar mixture to an @xmath1enhanced one with the same @xmath7 , for any value of @xmath7 .
cassisi et al . (
2004 ) analized the impact of @xmath0-enhanced ct in the @xmath74 filters , for the metallicity regime of galactic globular clusters , and their results have been already summarized in the introduction . here
we extend the analysis to higher metallicities .
figure [ ubvihz ] shows the comparison between 12 gyr old @xmath0-enhanced isochrones and zahb for @xmath7=0.0198 and @xmath7=0.04 in various cmds , using different ct .
more in detail , we used the appropriate @xmath0-enhanced ct , a scaled - solar ct with the same [ fe / h ] of the isochrone , and a scaled - solar ct with the same [ m / h ] ( total metallicity ) of the isochrone . we find ( in agreement with the results at lower @xmath7 ) that the @xmath57 magnitudes are hardly affected , whereas the @xmath16 and @xmath15 colors predicted by the scaled - solar ct with the same [ m / h ] of the isochrone are systematically redder along the whole isochrone .
differences are smaller when considering scaled - solar ct with the same [ fe / h ] of the isochrone .
this occurrence means that it is the abundance of fe and of all other scaled - solar elements that contribute mostly to the observed color , although not completely , given the difference with the appropriate ct
but they still reach up to @xmath19 0.1 mag along a large fraction of the isochrone .
this is due to the effect of the @xmath0-enhancement in the wavelength range of the @xmath75 and @xmath76 filter ( see discussion in cassisi et al .
2004 ) .
a different approach to mimic @xmath0-enhanced ct with scaled solar ones , is to use scaled - solar ct with the same [ @xmath0/h ] ratio of the [ @xmath0/fe]@xmath130 distribution .
so doing , one assumes that it is essentially the abundance of @xmath0-elements that affects the colors .
the data in figure [ ubvihz ] show that this choice is much less accurate , since it corresponds to a scaled - solar ct with a [ m / h ] value 0.4 dex larger that the actual one , that yields even redder colors .
interestingly , the @xmath77 color is also affected by the ct choice at high metallicities and low temperatures .
the two scaled - solar ct produce colors this time systematically bluer than the appropriate @xmath0-enhanced ct , the ct with the same [ fe / h ] of the isochrone giving the worst results . on the basis of our previous discussion
it is clear that in this case a scaled - solar ct with the same [ @xmath0/h ] ratio of the isochrone would provide colors closer to the appropriate ones .
this suggests that the abundance of the @xmath0-elements ( or at least some of them ) starts to dominate the ct in this wavelength range for solar and super - solar metallicities , and low temperatures
. figure [ strom1 ] shows another comparison of ct applied to a 12 gyr old @xmath0-enhanced isochrone at various @xmath7 , considering the appropriate @xmath0-enhanced transformations and the same two choices for the scaled - solar ct as in fig .
[ ubvihz ] , this time involving the strmgren filters .
the strmgren @xmath78 , @xmath79 and @xmath80 filters are located in the wavelength range of the johnson @xmath75 and @xmath76 bands ; therefore one expects a relevant influence of the @xmath0-elements on , for example , the @xmath81 $ ] color - color plane , used to estimate ages of star clusters and field stars , independently of the knowledge of their distance .
the use of different ct affects strongly the isochrone morphology in this plane , as shown clearly by fig .
[ strom1 ] .
the effect is appreciable from @xmath820.001 and is , as expected , larger at larger metallicities .
the sections of the isochrone that are more affected are , as for the case of johnson colors , the rgb and the lower ms .
again , scaled - solar ct at the same [ fe / h ] of the appropriate ones give a better approximation to the real ct , apart from the highest @xmath7 , where differences are huge for both scaled solar ct . in fig .
[ strom2 ] we show isochrones for various ages and @xmath83 , transformed to the @xmath81 $ ] plane using the same three choices of ct .
one can notice that the 12 gyr old isochrone transformed using the scaled - solar ct with the appropriate [ fe / h ] matches well turn off ( to ) and subgiant branch obtained with the appropriate colors . on the contrary , using scaled - solar ct with the same [ m / h ] of the @xmath0-enhanced ones causes not only a change of the morphology of the lower ms and the rgb , but also a shift of both @xmath84 and @xmath85 of the upper ms , to and subgiant region ; this induces systematic errors in both the age and reddening determined from isochrone fitting .
in this section we discuss the comparison in the h - r diagram of selected isochrones with similar predicitions from the databases by salasnich et al .
( 2000 ) , salaris & weiss ( 1998 ) , vandenberg et al . ( 2000 ) and kim et al . ( 2002 ) .
these grids of models are computed employing some different choices of the physical inputs ( the source of the opacities is quite often the same , but equation of state , nuclear reaction rates , boundary conditions and neutrino energy loss rates are often different ) and also the chemical compositions are often different .
we perform the comparison on the theoretical h - r diagram , thus bypassing the additional degree of freedom introduced by the choice of the color transformations . in each comparison
we have selected chemical compositions as close as possible to our choices .
it is important to notice that kim et al ( 2002 ) models do include he diffusion , and therefore their isochrones for old ages ( above a few gyr ) are affected by the efficiency of this process .
figure [ sw1 ] shows the comparison in the h - r diagram of a @xmath86 , 12 gyr old isochrone from salaris & weiss ( 1998 - hereafter sw98 ) with our own isochrone with the same age and metallicity , from the zams to the trgb .
the zahbs are also displayed .
metal mixture and opacities are exactly the same in two models , and also the initial he mass fraction is practically identical ( @xmath87 in sw98 models , compared to our adopted value of @xmath88 ) .
there are some differences in the luminosities of the zahb and to a lesser extent the to of the two isochrones .
the zahb from sw98 is brighter by @xmath890.03 - 0.04 , whereas the to is fainter by @xmath890.02 .
effective temperatures along the ms are in good agreement , but the temperature at the to region and along large part of the rgb of our isochrones is hotter by about 70 k and @xmath1950 k , respectively . a similar comparison but with a 12 gyr old @xmath0-enhanced isochrones by salasnich et al .
( 2000 - hereafter s00 ) is displayed in fig .
[ salas1 ] .
s00 models have also been computed using the same metal mixture and opacities employed in our own calculations ; at @xmath86 the initial he abundance of their models ( @xmath90 ) is close enough to ours , to provide a meaningful comparison . as for our models
, we display the zahb and the full 12 gyr isochrone computed with the reimers parameter @xmath20=0.2 .
the effective temperatures agree well along ms and rgb for @xmath26 larger than about 5000 k. at lower temperatures the s00 isochrone is systematically hotter .
the zahb luminosity of s00 models is fainter by @xmath890.04 .
given that s00 provide also @xmath0-enhanced isochrones for low ages , we show in fig .
[ salas2 ] a comparison of 500 myr old isochrones .
current and s00 isochrone include overshooting from the convective core , albeit with different prescriptions .
the to region of our isochrone is brighter and hotter than s00 counterpart .
this is consistent with the result of a similar comparison between our 500 myr scaled - solar isochrone of solar metallicity and the corresponding one from girardi et al .
( 2000 ) , that should be computed with physical prescriptions very similar to s00 .
as we remarked in paper i , the interplay between the different treatment of the core overshooting and some different input physics should play a major role in causing this difference . as for the scaled solar case our he burning phase is brighter than s00 one , but the post - ms effective temperatures of our isochrones are slightly cooler than s00 , whereas in the scaled solar case they were hotter .
figure [ dv ] shows the comparison with a 12 gyr old isochrone by vandenberg et al .
( 2000 ) for @xmath7=0.001 and @xmath7=0.01 .
although the selected @xmath7 is the same of our isochrones , there exists some difference at the level of 0.01 in the initial he mass fraction , and the average @xmath1enhancement of our models is equal to 0.4 dex , whereas the vandenberg et al .
( 2000 ) isochrones displayed in this comparison have [ @xmath0/fe]=0.3 , all @xmath0-element being enhanced by the same amount .
bearing in mind these differences between the two data sets , one notices that at @xmath91 the ms almost completely overlap , whereas the to of our isochrone is slightly brighter , by @xmath92 and @xmath93 k hotter . along the rgb
our isochrone is hotter by about 140 k , a difference larger than what we found in the case of the scaled - solar models compared in paper i. this occurrence can not be attributed to a combination of different eos and solar mixing - length calibration , since we found a very good agreement for scaled - solar rgbs .
perhaps , it is mainly an effect due to the different @xmath1element enhancement and mixture and , to a lesser extent , the higher initial he content of our models .
the brightness difference at @xmath94 along the zahb - taken as representative of the mean effective temperature of the rr lyrae instability strip - is of the order of @xmath95 . the differences between the two sets of isochrones increase for increasing @xmath7 .
at @xmath96 the to of our isochrone is brighter by @xmath97 and @xmath98 k hotter .
our rgb effective temperature is hotter by @xmath99 k , while the brightness difference at the rr lyrae instability strip is equal to @xmath100 .
finally , we show in fig .
[ yy ] a comparison with the @xmath0-enhanced isochrones by kim et al .
( 2002 ) , with [ @xmath0/fe]=0.3 , @xmath7=0.02 and @xmath9=0.27 . the @xmath0-enhanced metal distribution of kim et al .
( 2002 ) is the same as vandenberg et al .
we selected two ages , 600 myr and 13 gyr , respectively .
our 600 myr old isochrone is computed with convective overshoot , given that kim et al .
( 2002 ) models include overshoot from the convective core .
the outcome of the comparison is very similar to what we found in paper i in case of the yi et al .
( 2001 ) scaled - solar models , which are homogeneous with the kim et al .
( 2002 ) library .
the ms appear in good agreement , apart from the lowest luminosities , and the to brightness of the 600 myr isochrones are very similar , although our to is hotter .
the rgbs of our isochrones are systematically hotter , by about 200 k , similar to what we found for the scaled solar counterparts .
the to region of the 13 gyr isochrone from kim et al .
( 2002 ) is fainter and cooler , mainly due to the effect of he - diffusion .
in this section we present a set of empirical tests , to show the level of agreement between our models and a number of photometric constraints .
our @xmath0-enhanced isochrones have been already compared to globular cluster ( gc ) cmds in the acs filters ( bedin et al .
2005 ) and employed to determine the initial he content ( from the theoretical calibration of the @xmath101-parameter ) , distances ( from the theoretical zahb models ) , reddening and relative ages of a large sample of galactic gcs ( salaris et al . 2004 ,
recio - blanco et al . 2005 and de angeli et al . 2005 ) .
more in detail , the distances obtained from fitting our zahb models to the observed hbs of a sample of galactic gcs ( in the hst flight system ) provides a distance scale in very good agreement with the empirical ms - fitting distances determined by carretta et al .
( 2000 ) , who employed a sample of field subdwarfs with accurate hipparcos parallaxes and metallicity determinations .
the relationship between @xmath102 of the zahb and [ fe / h ] from our models ( see table 3 ) is in agreement with the similar relationship calibrated by carretta et al .
( 2000 ) from their empirical gc distance , within their quoted error bars .
moreover , the initial he content ( @xmath103 ) obtained for the large sample of gcs analyzed by salaris et al .
( 2004 ) , is in line with predictions from primordial nucleosynthesis calculations , when using the value of the cosmological baryon density estimated from wmap data ( spergel et al .
2003 ) . in the following ,
we will make use of the extinction laws by schlegel , finkbeiner & davis ( 1998 ) .
figure [ subdw ] shows the comparison of the cmd of the field subdwarf sample by carretta et al .
( 2000 ) with isochrones of the appropriate metallicity .
these objects have accurate hipparcos parallaxes , spectroscopic metallicities , individual reddening determinations , and have been grouped into four subsamples of different mean [ fe / h ] .
we have compared the cmd of these objects with our 13 gyr old isochrones , whose [ fe / h ] is within at most 0.04 dex from the mean values displayed in the figure .
the precise choice of the isochrone age does not play any role when @xmath104 mag .
theory and observations appear in satisfactory agreement , particularly in the magnitude range unaffected by age . at @xmath102=6 the theoretical isochrones trace very well the change of subdwarf colors with changing metallicity . at [ fe / h]@xmath171.60 and @xmath171.0
the faintest subdwarfs appear to be systematically redder than the models , but at [ fe / h]@xmath171.3 they agree within the quoted 1@xmath105 error bars .
the figures [ m68 ] and [ n362 ] show the comparison of our isochrones with the cmd of two galactic gcs , namely @xmath106 data for m68 ( data from walker 1994 ) and @xmath107 data for ngc 362 ( data from bellazzini et al .
2001 ) . a detailed analysis of the fitting to globular cluster cmds and the study of their age distribution is outside the scope of the paper .
as mentioned before , here we wish only to show the level of agreement between our set of models and selected cmds of galactic gcs of different [ fe / h ] values .
for m68 we display in fig .
[ m68 ] our 11 and 12 gyr old isochrones with [ fe / h][email protected] , corrected for a reddening @xmath1090.07 and distance modulus @xmath110=15.25 .
the chosen metallicity is somewhat intermediate among the values [ fe / h]=@xmath111 on the carretta & gratton ( 1997 cg97 ) scale , [ fe / h]=@xmath112 on the zinn & west ( 1984 zw84 ) scale and [ fe / h][email protected] on the kraft & ivans ( 2003 ki03 ) scale .
the adopted reddening ( necessary to match the location of the observed ms ) is in good agreement with the value @[email protected] estimated by walker ( 1994 ) .
the distance is obtained from the fitting of the theoretical zahb to the observed counterpart , and agrees with the value we determined in the hst system ( recio - blanco et al . 2004 ) . together with the predicted zahb we also display the hb and agb portion of the 12 gyr isochrone , for values of the mass loss parameter @xmath20=0.2 and 0.4 , respectively .
figure [ n362 ] displays the comparison with ngc362 .
ngc 362 has a metallicity [ fe / h][email protected] on both zw84 and ki03 scales , and [ fe / h][email protected] on the cg97 scale ; we compare the observed cmd to our isochrones with [ fe / h][email protected] ( t=10 gyr @xmath1090.02 ) and an isochrone with [ fe / h][email protected] ( t=9 gyr , @xmath1090.0 ) . together with the full theoretical zahb we display the hb and agb portion of the isochrones computed with @xmath22 for [ fe / h][email protected] , and both @xmath20=0.2 and 0.4 for [ fe / h][email protected] .
the distance moduli adopted from the fitting of the observed zahb are @xmath114 for [ fe / h][email protected] and @xmath110=14.86 for [ fe / h][email protected] , consistent with recio - blanco et al .
( 2004 obtained from our models transformed into the hst flight system ) and very close to the ms - fitting distance by carretta et al .
( 2000 @xmath115 ) .
we have also performed some comparisons with near infrared data of galactic globular clusters .
figure [ m92 ] shows our 13 gyr old isochrone with [ fe / h][email protected] , including the full zahb plus the hb and agb evolution computed with @xmath20=0.2 , together with a @xmath116 cmd of m92 ( del principe et al .
we recall that m92 has [ fe / h][email protected] on the cg97 scale and [ fe / h][email protected] on the zw84 scale .
we have shifted the models to account for the well established reddening @xmath1090.02 ( schlegel et al .
1998 ) and a distance modulus @xmath117 , that corresponds to @xmath118 .
this value is within @xmath119 of the empirical ms - fitting distance obtained by carretta et al .
( 2000 ) , e.g. , @xmath120 .
the dispersion of the observational cmd does not allow us to firmly constrain the derived distance and age ; nevertheless the displayed isochrone properly fits the various branches of the observed diagram .
a @xmath121 cmd of the hb and bright rgb of the [ fe / h]@xmath122 cluster m69 ( valenti , origlia & ferraro 2005 ) is shown in figure [ m69 ] , compared to our 10 gyr , [ fe / h]=@xmath123 isochrone ( including zahb plus the hb and agb evolution for both @xmath20=0.2 and 0.4 ) .
the observational data were on the 2mass system and have been transformed to the @xmath121 system consistent with our color transformations using the empirical relationships by carpenter ( 2001 ) .
we have adopted a distance modulus @xmath124 from fitting the theoretical zahb to the observed hb , and a reddening @xmath125=0.15 ; both values are very close to @xmath126 and @xmath125=0.16 listed in the harris ( 1996 ) catalog .
the displayed isochrone traces very well the position of the rgb up to the brightest detected stars .
figures [ m92strom ] and [ m13strom ] show the comparison of our isochrones with the strmgren @xmath127 $ ] diagrams of m92 and m13 ( grundahl et al .
1998 , 2000 ) . in case of m92
we show a 13 gyr old [ fe / h][email protected] isochrone computed with @xmath21 , together with the full zahb .
the only correction applied to the models is for the reddening @xmath125=0.02 , where we have used the relationships @xmath128 , and @xmath129 . notice how the zahb sequence runs through the center of the observed hb distribution .
this is exactly what one expects from theory , given that the evolution off the zahb along the blue part of the hb on this @xmath127 $ ] diagram overlaps with the zahb itself .
when the zahb sequence turns toward the red , theoretical models show a value of @xmath84 slightly larger than observed .
the isochrones overlap to various degrees all other main branches of the observed diagram , including the rgb .
one has also to notice the spread along the observed rgb , due to the large scatter in @xmath84 caused by star - to - star differences in the abundance of nitrogen ( see , e.g. grundahl et al .
1998 ) . in case of m13
( see fig .
[ m13strom ] ) we have displayed two isochrones and zahbs with , respectively , t=13 gyr , [ fe / h][email protected] , and t=11 gyr , [ fe / h][email protected] .
this accounts for the uncertainty in m13 iron content , which is [ fe / h]=@xmath130 according to cg97 , [ fe / h]@xmath171.5 according to ki03 and [ fe / h]=@xmath131 according to zw84 . only for the [ fe / h][email protected] isochrone we have displayed
the full hb evolution obtained using the reimers parameter @xmath20=0.4 along the rgb .
the models have been shifted for a reddening @xmath125=0.03 , very close to the value @xmath125=0.02 given by schlegel et al .
the level of agreement between theory and observations is similar to the case of m92 .
notice again the wide rgb , as for m92 ( see fig .
[ m92strom ] ) .
we show here a few comparisons of integrated properties extracted from our isochrone sets , e.g. surface brightness fluctuations ( sbfs ) in @xmath57 and @xmath132 , plus @xmath15 and @xmath77 integrated colors of a sample of galactic gcs .
they can be considered as a global test of the accuracy of the bolometric corrections , colors and evolutionary timescales along the relevant phases .
we selected gcs whose sbfs in @xmath57 and @xmath132 have been determined by ajhar & tonry ( 1994 ) , and whose distances have been estimated by recio - blanco et al .
( 2005 ) using our @xmath0-enhanced zahb models .
as for the integrated colors , we considered the clusters with both @xmath77 and @xmath15 integrated photometry plus reddenings and [ fe / h ] values available in the latest version of the harris ( 1996 ) catalog , which are an average of various sources .
it is important to remark that the @xmath57 and @xmath132 sbfs , and the integrated @xmath15 and @xmath77 colors are very weakly affected by the full thermal pulse phase and by ms masses below 0.5@xmath56 ( presently not included in our database see , e.g. , brocato et al .
2000 , cantiello et al .
2003 ) .
the theoretical fluctuation magnitudes have been as usual computed from the fluctuation luminosity @xmath133 in the selected photometric band , given by @xmath134 where @xmath135 is the number of stars of type @xmath136 and luminosity @xmath137 .
these sums have been performed along the appropriate isochrone , populated according to a salpeter initial mass function ( the integrated colors have also been computed using the same initial mass function ) .
figure [ sbf ] shows the comparison between the observed gc @xmath138 and @xmath139 fluctuation magnitudes , and the theoretical counterpart obtained from our 10 gyr @xmath22 isochrones . in order to show the effect of the uncertain [ fe / h ] scale we have considered both the zw84 and cg97 values .
the 1@xmath105 error bars of the observational points have been obtained by adding in quadrature the 1@xmath105 errors given by ajhar & tonry ( 1994 ) and the errors on the individual distance moduli given by recio - blanco et al .
the theoretical fluctuation magnitudes trace very well the observations ; the effect of a larger age ( e.g. 13 instead of 10 gyr ) and a bluer hb ( @xmath20=0.4 instead of @xmath20=0.2 ) are negligible on this plot .
as noticed by our referee , two clusters , namely ngc 6652 and ngc 6723 ( with [ fe / h]@xmath171.0 ) are systematically discrepant in both the @xmath139 and @xmath138 fluctuation magnitudes , by about 0.2 - 0.4 mag .
it is difficult to point out a precise reason for this discrepancy which is at the level of at most @xmath140 although we believe it is possible to exclude an underestimate of the distances to these two clusters by such a large amount .
here we just notice that a similar discrepancy in @xmath139 and @xmath138 can be found in the analysis by cantiello et al .
( 2003 their fig .
12 ) , who use their own sbf models , [ fe / h ] and distances from the harris ( 1996 ) catalog . a comparison with the observed @xmath77 and @xmath15 integrated colors is displayed in fig . [ intcol ] .
the observed colors have been dereddened using the @xmath125 values given in the harris catalog .
given the heterogeneous sources for the observational data one can only discuss the general agreement with theory .
not surprisingly , the integrated @xmath77 is very weakly sensitive to age and also to the assumed value of @xmath20 , being mainly affected by [ fe / h ] .
our models trace very well the observed trend of @xmath77 with [ fe / h ] .
the @xmath15 color is more sensitive to both the age and hb morphology ; even in this case the models reproduce well the observations .
this paper presents an extension of the scaled - solar stellar model and isochrone database of paper i , to include an @xmath0-enhanced ( [ @xmath0/fe]=+0.40 ) metal distribution .
the models have been computed by using the same sources for the input physics and the same stellar evolution code of paper i. they are based on the most updated stellar physics available , and are fully consistent with our scaled - solar database .
this allows one to disentangle unambiguously the effect of different heavy element distributions when analizing the properties of stellar populations .
the database covers a wide metallicity range suitable for both metal - poor stars , such as extreme population ii objects , and metal - rich stars such as the ones hosted by the galactic bulge and elliptical galaxies .
the adopted initial he content for the most metal - poor stellar models is consistent with recent predictions based on cmb analysis and the _
r_-parameter investigations on a large sample of galactic gcs , and the adopted he - enrichment ratio accounts for the initial solar he abundance based on our calibration of the solar standard model . for each chemical composition in our grid , we computed an extended set of hb models corresponding to an rgb progenitor whose age at the he flash is of the order of @xmath141 gyr , i.e. , suitable for investigating the properties of hb stars in old stellar systems .
so far , this is the most complete and homogeneous set of hb models for an @xmath1enhanced metal distribution , available in literature . as in paper
i , we devote great care to bolometric corrections and color-@xmath26 transformations .
we have been able to use ct based on a metal distribution consistent with that adopted for the evolutionary computations .
this opportunity allowed us to extend the investigation performed by cassisi et al .
( 2004 ) to the high metallicity regime and more narrow photometric bands such as the strmgren filters .
our analysis makes more evident the need to employ appropriate @xmath0-enhanced ct , especially at high metallicities .
we have discussed some relevant comparisons between our theoretical framework and empirical data for both unevolved field ms stars with accurate distance and metallicity estimates , and galactic gcs with accurate photometry in various photometric systems . as a whole we notice an agreement between theory and observations .
the whole library is made available to the scientific community in an easy and direct way through a purpose built web site , the same that makes public our scaled - solar database .
this web site is continuously updated to include additional theoretical models , fully consistent with those presented here and in paper i. in collaboration with d. cordier , a big effort is being made to prepare some web interactive interfaces that allow users to compute their desired isochrones , evolutionary tracks , luminosity functions and synthetic cmds .
we wish to warmly thank d. cordier and m. castellani for the relevant help provided when preparing the basti web pages .
we thank also m. quintini for his painstaking help in updating the basti database in the web server at the observatory of teramo .
it is a pleasure to thank g. bono for a detailed reading of an early draft of this paper .
this research has made use of nasa s astrophysics data system abstract service and the simbad database operated at cds , strasbourg , france .
ccc c & 0.108211 & 0.076451 n & 0.028462 & 0.023450 o & 0.714945 & 0.672836 ne & 0.071502 & 0.084869 na & 0.000652 & 0.000882 mg & 0.029125 & 0.041639 al & 0.000900 & 0.001428
si & 0.021591 & 0.035669 p & 0.000086 & 0.000157 s & 0.010575 & 0.019942 cl & 0.000096 & 0.000201 ar & 0.001010 & 0.002373 k & 0.000040 & 0.000092 ca & 0.002210 & 0.005209 ti & 0.000137 & 0.000387 cr & 0.000145 & 0.000443 mn & 0.000075 & 0.000242 fe & 0.009642 & 0.031675 ni & 0.000595 & 0.002056 cccr 0.0001 & 0.245 & @xmath1082.62 & @xmath1082.27 + 0.0003 & 0.245 & @xmath1082.14 & @xmath1081.79 + 0.0006 & 0.246 & @xmath1081.84 & @xmath1081.49 + 0.0010 & 0.246 & @xmath1081.62 & @xmath1081.27 + 0.0020 & 0.248 & @xmath1081.31 & @xmath1080.96 + 0.0040 & 0.251 & @xmath1081.01 & @xmath1080.66 + 0.0080 & 0.256 & @xmath1080.70 & @xmath1080.35 + 0.0100 & 0.259 & @xmath1080.60 & @xmath1080.25 + 0.0198 & 0.273 & @xmath1080.29 & 0.06 + 0.0300 & 0.288 & @xmath1080.09 & 0.26 + 0.0400 & 0.303 & 0.05 & 0.40 + [ chemcomp ] cccccccc 0.0001 & @xmath1082.62 & 0.80 & 0.253 & 0.5028 & 0.802 & 1.7573 & 0.375 + 0.0003 & @xmath1082.14 & 0.80 & 0.255 & 0.4969 & 0.710 & 1.7142 & 0.479 + 0.0006 & @xmath1081.84 & 0.80 & 0.258 & 0.4934 & 0.668 & 1.6955 & 0.518 + 0.0010 & @xmath1081.62 & 0.80 & 0.259 & 0.4905 & 0.642 & 1.6764 & 0.559 + 0.0020 & @xmath1081.31 & 0.80 & 0.262 & 0.4877 & 0.613 & 1.6510 & 0.615 + 0.0040 & @xmath1081.01 & 0.80 & 0.266 & 0.4845 & 0.589 & 1.6161 & 0.691 + 0.0080 & @xmath1080.70 & 0.90 & 0.276 & 0.4795 & 0.567 & 1.5705 & 0.812 + 0.0100 & @xmath1080.60 & 0.90 & 0.279 & 0.4781 & 0.561 & 1.5545 & 0.825 + 0.0198 & @xmath1080.29 & 1.00 & 0.296 & 0.4713 & 0.542 & 1.5025 & 0.948 + 0.0300 & @xmath1080.09 & 1.00 & 0.310 & 0.4656 & 0.529 & 1.4753 & 0.998 + 0.0400 & 0.05 & 1.00 & 0.324 & 0.4603 & 0.519 & 1.4657 & 1.015 + [ zahb ] | we present a large , new set of stellar evolution models and isochrones for an @xmath0-enhanced metal distribution typical of galactic halo and bulge stars ; it represents a homogeneous extension of our stellar model library for a scaled - solar metal distribution already presented in pietrinferni et al .
( 2004 ) .
the effect of the @xmath1element enhancement has been properly taken into account in the nuclear network , opacity , equation of state and , for the first time , the bolometric corrections , and color transformations .
this allows us to avoid the inconsistent use - common to all @xmath0-enhanced model libraries currently available - of scaled - solar bolometric corrections and color transformations for @xmath0-enhanced models and isochrones .
we show how bolometric corrections to magnitudes obtained for the @xmath2 portion of stellar spectra ( i.e. not only the johnson - cousins filters but also strmgren @xmath3 magnitudes ) for @xmath4 k , are significantly affected by the metal mixture , especially at the higher metallicities .
our models cover both an extended mass range ( between @xmath5 to @xmath6 , with a fine mass spacing ) and a broad metallicity range , including 11 values of the metal mass fraction @xmath7 , corresponding to the range @xmath8\le0.05 $ ] .
the initial he mass fraction is @xmath9=0.245 for the most metal - poor models and increases with @xmath7 according to @xmath10=1.4 , to reproduce the initial solar he obtained from the calibration of the solar model .
models with and without the inclusion of overshoot from the convective cores during the central h burning phase are provided , as well as models with different mass loss efficiencies .
we also provide complete sets of evolutionary models for low - mass , he - burning stellar structures covering the whole metallicity range , to enable synthetic horizontal branch simulations .
we compare our database with several widely used stellar model libraries from different authors , as well as with various observed color magnitude and color - color diagrams ( johnson - cousins @xmath11 and near infrared magnitudes , strmgren colors ) of galactic field stars and globular clusters .
we also test our isochrones comparing integrated optical colors and surface brightness fluctuation magnitudes with selected globular cluster data .
we find a general satisfactory agreement with the empirical constraints .
this database , used in combination with our scaled - solar model library , is a valuable tool for investigating both galactic and extra - galactic simple and composite stellar populations , using stellar population synthesis techniques . |
in topological data analysis one studies high dimensional data sets by extracting shapes . many of these shapes are @xmath0-dimensional simplicial complexes where it is computationally possible to calculate topological invariants such as the fundamental group or homology groups ( see for example @xcite ) .
@xmath0-complexes that are amenable for more detailed analysis are foams and @xmath0-stratifolds .
foams include special spines of 3-dimensional manifolds @xcite , @xcite .
khovanov @xcite used trivalent foams to construct a bigraded homology theory whose euler characteristic is a quantum @xmath2 link invariant and carter @xcite presented an analogue of the reidemeister - type moves for knotted foams in @xmath3-space . a closed @xmath0-_stratifold _ is a @xmath0-dimensional cell complex @xmath4 that contains a collection of finitely many simple closed curves , the components of the @xmath1-skeleton @xmath5 of @xmath4 , such that @xmath6 is a @xmath0-manifold and a neighborhood of each component @xmath7 of @xmath5 consists of @xmath8 sheets .
it is not not known which @xmath9-manifolds have spines that are @xmath0-stratifolds .
there are significant differences : for any given @xmath0-stratifold there are infinitely many non - homeomorphic @xmath0-stratifolds with the same fundamental group .
@xmath9-manifold groups are residually finite , but every baumslag - solitar group ( some of which are hopfian , others are non - hopfian ) can be realized as the fundamental group of a simple @xmath0-stratifold .
however it can be shown that fundamental groups of @xmath0-stratifolds have solvable word problem ( work in progress ) .
@xmath0-stratifolds arise in the study of categorical invariants of @xmath9-manifolds .
for example if @xmath10 is a non - empty family of groups that is closed under subgroups , one would like to determine which ( closed ) @xmath9-manifolds have @xmath10-category equal to @xmath9 .
in @xcite it is shown that such manifolds have a decomposition into three compact @xmath9-submanifolds @xmath11 , where the intersection of @xmath12 ( for @xmath13 ) is a compact @xmath0-manifold , and each @xmath14 is @xmath10-contractible ( i.e. the image of the fundamental group of each connected component of @xmath14 in the fundamental group of the @xmath9-manifold is in the family @xmath10 ) . the nerve of this decomposition , which is the union of all the intersections @xmath12 ( @xmath13 ) , is a closed @xmath0-stratifold and determines whether the @xmath10-category of the 3-manifold is @xmath0 or @xmath9 .
a @xmath0-stratifold is essentially determined by its associated bipartite labelled graph ( defined in section 2 ) and a presentation for its fundamental group can be read off from the labelled graph .
thus the question arises when a labelled graph determines a simply connected @xmath0-stratifold . in @xcite
it is shown that a necessary condition is that the underlying graph must be a tree ; if the graph is linear then sufficient and necessary conditions on the labelling are given , and if the graph is trivalent ( definition in section 2 ) , an algorithm on the labelled graph was developed for determining whether the graph determines a simply connected @xmath0-stratifold . in @xcite
an algorithm is given that decides whether a given labelled graph ( not necessarily trivalent ) determines a @xmath0-stratifold that is homotopy equivalent to @xmath15 .
the main result of the present paper ( in section 3 ) is a classification of all trivalent labelled graphs that represent simply connected @xmath0-stratifolds .
then in sections 4 and 5 we develop three operations on labelled graphs that will construct from a single vertex all trivalent graphs that represent simply connected @xmath0-stratifolds .
we first review the basic definitions and some results given in @xcite and @xcite .
a @xmath0-_stratifold _ is a compact , hausdorff space @xmath4 that contains a closed ( possibly disconnected ) @xmath1-manifold @xmath5 as a closed subspace with the following property : each point @xmath16 has a neighborhood homeomorphic to @xmath17 , where @xmath18 is the open cone on @xmath19 for some ( finite ) set @xmath19 of cardinality @xmath20 and @xmath21 is a ( possibly disconnected ) @xmath0-manifold . +
a component @xmath22 of @xmath5 has a regular neighborhood @xmath23 that is homeomorphic to @xmath24 ) /(y,1)\sim ( h(y),0)$ ] , where @xmath25 is the closed cone on the discrete space @xmath26 and @xmath27 is a homeomorphism whose restriction to @xmath26 is the permutation @xmath28 .
the space @xmath29 depends only on the conjugacy class of @xmath30 and therefore is determined by a partition of @xmath31 .
a component of @xmath32 corresponds then to a summand of the partition determined by @xmath33 . here
the neighborhoods @xmath34 are chosen sufficiently small so that for disjoint components @xmath7 and @xmath35 of @xmath5 , @xmath34 is disjoint from @xmath36 .
the components of @xmath37 are called the _ sheets _ of @xmath34 .
+ for a given @xmath0- stratifold @xmath38 there is an associated bipartite graph @xmath39 embedded in @xmath4 as follows : + in each component @xmath40 of @xmath5 choose a black vertex @xmath41 . in the interior of each component @xmath42 of @xmath43
choose a white vertex @xmath44 .
in each component @xmath45 of @xmath46 choose a point @xmath47 , an arc @xmath48 in @xmath42 from @xmath44 to @xmath47 and an arc @xmath49 from @xmath47 to @xmath41 in the sheet of @xmath50 containing @xmath47 .
an edge @xmath51 between @xmath44 and @xmath41 consists of the arc @xmath52 . for a fixed @xmath53 , the arcs @xmath54
are chosen to meet only at @xmath44 .
+ we label the graph @xmath55 by assigning to a white vertex @xmath56 its genus @xmath57 of @xmath56 and by labelling an edge @xmath58 by @xmath59 , where @xmath59 is the summand of the partition @xmath33 corresponding to the component @xmath58 of @xmath32 where @xmath60 .
( here we use neumann s @xcite convention of assinging negative genus @xmath57 to nonorientable surfaces ) .
note that the partition @xmath33 of a black vertex is determined by the labels of the adjacent edges .
if @xmath55 is a tree , then the labeled graph determines @xmath4 uniquely . + * notation*. if @xmath55 is a bipartite labelled graph corresponding to the @xmath0-stratifold @xmath4 we let @xmath61 and @xmath62 .
+ the following was shown in @xcite .
[ retraction ] there is a retraction @xmath63 .
[ simplyconnected ] if @xmath4 is simply connected , then @xmath64 is a tree , all white vertices of @xmath64 have genus @xmath65 , and all terminal vertices are white .
the proof uses proposition [ retraction ] and the following pruning construction : + * pruning at a subgraph*. for a subgraph @xmath66 of @xmath55 and the retraction @xmath67 , let @xmath68 .
this is almost a @xmath0-stratifold , except that @xmath25 has possibly boundary curves corresponding to edges of @xmath69 , where @xmath70 is the star of @xmath66 in @xmath55 .
let @xmath71 be the quotient of @xmath4 obtained by collapsing the closure of each component of @xmath72 to a point i.e. @xmath71 is obtained from @xmath25 by attaching disks to its boundary curves .
then @xmath71 is a @xmath0-stratifold , @xmath73 , whose graph @xmath74 is the union of @xmath66 and the labeled edges ( with their vertices ) of @xmath69 which are adjacent to a black vertex of @xmath66 . and @xmath75,width=384 ] @xmath76 is a pruned subgraph of @xmath55 if @xmath77 , for some subgraph @xmath66 of @xmath55 .
[ pruning ] for a pruned subgraph @xmath76 of @xmath55 , the quotient map @xmath78 induces surjections @xmath79 and @xmath80 .
therefore , if @xmath81 is simply connected , @xmath82 is simply connected . in this paper
we consider trivalent graphs .
a @xmath0-stratifold @xmath4 and its labeled bicolored graph @xmath64 are defined to be _ trivalent _ , if each black vertex @xmath83 is incident to either three edges each with label @xmath1 or to two edges , one with label @xmath1 , the other with label @xmath0 , or @xmath83 is a terminal vertex with adjacent edge of label @xmath9 .
this means that a neighborhood of a point of a component @xmath7 of the @xmath1-skeleton @xmath5 has @xmath9 sheets , so the permutation @xmath84 of the regular neighborhood @xmath23 has partition @xmath85 or @xmath86 or @xmath9 .
+ one of the main results in @xcite ( theorem 5 ) is that a trivalent 2-stratifold is @xmath1-connected if and only if @xmath87 .
the second condition is only needed to insure that @xmath64 has no terminal black vertices , therefore : [ pi = h ] let @xmath4 be a trivalent 2-stratifold such that all terminal vertices are white .
then @xmath4 is @xmath1-connected if and only if @xmath88 .
the building blocks for constructing labeled trivalent graphs for 1-connected @xmath0-stratifolds are called @xmath89-graphs and @xmath90-graphs : \(1 ) the @xmath90-tree is the bipartite tree consisting of one black vertex incident to three edges each of label @xmath1 and three terminal white vertices each of genus @xmath65 . + ( 2 ) the @xmath89-tree is the bipartite tree consisting of one black vertex incident to two edges one of label @xmath1 , the other of label @xmath0 , and two terminal white vertices each of genus @xmath65 .
-tree and @xmath90-tree , width=240 ] first we consider special trivalent trees that do not contain @xmath90-subtrees , which we call _ barycentrically subdivided rooted @xmath91-labeled trees _ or short _ @xmath91-collapsible trees _ : + a _ barycentrically subdivided rooted @xmath91-labeled tree _
is constructed from a rooted tree @xmath92 with root @xmath93 ( a vertex of @xmath92 ) as follows : color with white and label @xmath65 the vertices of @xmath92 , take the barycentric subdivision @xmath94 of @xmath92 , color with black the new vertices ( the barycenters of the edges of @xmath92 ) and finally label an edge @xmath95 of @xmath94 with @xmath0 ( resp .
@xmath1 ) if the distance from @xmath95 to the root @xmath93 is even ( resp .
this labeled @xmath94 is the _
@xmath91-collapsible tree _ determined by @xmath96 .
we allow a one - vertex tree ( with white vertex ) as a @xmath91- collapsible tree . +
a typical example is shown in figure 3 , where regions enclosed by the dashed curves are @xmath91-collapsible trees , bold white vertices are roots . [ 12-components ] let @xmath81 be a trivalent @xmath1-connected @xmath0-stratifold . then after removing the open stars of all black vertices of degree @xmath9 , the components @xmath97 are barycentrically subdivided rooted @xmath91-labeled trees .
furthermore , for each black vertex @xmath83 of degree @xmath9 , at least one of its ( three white ) neighbors is the root of some @xmath98 .
note that each @xmath98 is a pruned subgraph of @xmath55 and so @xmath99 .
therefore for the first part of the proposition it suffices to show that if @xmath7 has no black vertices of degree @xmath9 and @xmath100 is simply connected , then @xmath7 is a barycentrically subdivided rooted @xmath91-labeled tree . by proposition
[ simplyconnected ] , @xmath7 is a tree with all white vertices of genus @xmath65 and all terminal vertices white and , if @xmath7 is not a vertex , contains a @xmath89-subgraph @xmath19 with terminal edge of label @xmath1 .
let @xmath101 be the white vertex of @xmath19 which is not a terminal vertex of @xmath7 .
let @xmath102 , let @xmath103 ( @xmath104 ) be the components of @xmath105 and @xmath106 .
then @xmath107 is simply connected for @xmath108 . by induction on the number of vertices ,
each @xmath109 is a barycentrically subdivided rooted @xmath91-labeled tree with a root @xmath110 . if @xmath111 for each @xmath104 then @xmath7 is a barycentrically subdivided rooted @xmath91-labeled tree with root @xmath101 . if @xmath112 the label on the edge @xmath113 of @xmath109 incident to @xmath101 is @xmath1 .
it follows that there is at most one @xmath110 not equal to @xmath101 . for otherwise ,
if @xmath114 the unioin of the edges and vertices of the simple path in @xmath115 from @xmath110 to @xmath116 is a pruned linear subgraph @xmath117 of @xmath7 for which @xmath118 is not simply connected by theorem 3 of @xcite .
this is a contradiction since @xmath119 is a quotient of @xmath120 .
it follows that if @xmath112 then @xmath7 is a barycentrically subdivided rooted @xmath91-labeled tree with root @xmath110 .
for the second part of the proposition , suppose that @xmath55 contains a @xmath90-subgraph with black vertex @xmath83 such that none of its white vertices @xmath121 is a root of any of the @xmath40 and let @xmath98 be the pruned subgraph of @xmath55 containing @xmath44 ( @xmath122 ) and with roots @xmath123 .
then the pruned subgraph @xmath66 of @xmath55 which is the union of the edges and vertices of the three simple paths in @xmath55 from the roots @xmath110 to @xmath83 ( @xmath122 ) has all terminal edges of label @xmath0 . by lemma 4 of @xcite , @xmath124 , a contradiction .
the figure below shows that the converse of lemma [ 12-components ] is false .
there are two @xmath90-vertices , all labels are @xmath1 except as indicated , and @xmath125 . deleting the two @xmath89-trees at the white center vertex ( but not the center vertex )
horned tree_. a horned tree @xmath126 is a finite connected bipartite labeled tree such that + ( 1 ) every black vertex @xmath83 whose distance to a terminal white vertex is @xmath1 has degree @xmath0 ; otherwise @xmath83 has degree 3 ; + ( 2 ) every nonterminal white vertex has degree @xmath0 ; + ( 3 ) every terminal edge has label @xmath0 , every nonterminal edge has label @xmath1 .
+ ( 4 ) there is at least one vertex of degree @xmath9 .
a horned tree @xmath126 may be constructed from a tree @xmath92 all of whose nonterminal vertices have degree @xmath9 , with at least one such vertex , as follows : color a vertex of @xmath92 white ( resp .
black ) if it has degree @xmath1 ( resp .
@xmath9 ) . trisect the terminal edges of @xmath92 and bisect the nonterminal edges , obtaining the graph @xmath126 .
color the additional vertices @xmath127 so that @xmath126 is bipartite , that is , @xmath127 is colored black if @xmath127 is a neighbor of a terminal vertex of @xmath126 and white otherwise .
then label the edges so that ( 3 ) holds .
+ the main property of @xmath126 is that @xmath128 .
+ finally we consider a reduced graph " of @xmath55 which encodes information on how the @xmath91-collapsible trees of lemma [ 12-components ] are attached to the stars of the black vertices of degree @xmath9 .
+ denote by @xmath129 the union of all the black vertices of degree @xmath9 of @xmath55 , let @xmath130 be the ( closed ) star of @xmath129 in @xmath55 and @xmath131 be the the open star of @xmath129 .
note that @xmath132 consists of the components @xmath97 as in lemma [ 12-components ] .
let @xmath55 be a bipartite labeled tree such the the components of @xmath133 are barycentrically subdivided rooted @xmath91-labeled trees .
the reduced subgraph @xmath134 of @xmath55 is the graph obtained from @xmath130 by attaching to each white vertex @xmath101 of @xmath130 that is not a root , a @xmath89-graph such that the terminal edge has label @xmath0 . as an example the reduced graph @xmath134 for the graph in figure 3
is shown below .
[ h(rg ) ] @xmath135 .
@xmath136 is generated by the black vertices .
let @xmath101 be a white vertex of @xmath55 of degree @xmath137 with incident edges @xmath138 .
suppose the label on @xmath113 is @xmath0 for @xmath139 ( @xmath140 ) .
split @xmath101 into @xmath141 ( disjoint ) vertices @xmath142 so that each @xmath143 has degree @xmath1 with adjacent edge @xmath144 and @xmath145 has degree @xmath146 with adjacent edges @xmath147 .
this change of the graph does not change @xmath136 .
now let c be a @xmath91-collapsible component of @xmath132 .
if @xmath101 is a terminal ( white ) vertex of @xmath7 that is also a terminal vertex of @xmath55 , delete the @xmath89-subgraph of @xmath7 that contains @xmath101 , if there is one ( @xmath7 might consist of a single vertex ) .
continue doing this operation until all terminal vertices of @xmath7 belong to @xmath130 .
this does not change @xmath148 . if @xmath101 is a non - terminal white vertex of ( the new @xmath7 ) then ,
if @xmath101 is the root , all edges incident to @xmath101 have label @xmath0 ; if @xmath101 is not the root , all edges but one incident to @xmath101 have label @xmath0
. do the above construction on each such @xmath101 to change @xmath7 to @xmath35 .
then each component of @xmath35 that does not contain a terminal white vertex of @xmath7 is a linear graph of type @xmath149 with successive edge labels @xmath150 ( or consists of a single white vertex ) .
deleting these homologically trivial components from @xmath35 does not change @xmath136 . doing this for all @xmath91-collapsible components @xmath7 of @xmath132 results in @xmath134 .
we now state the classification theorem .
let @xmath81 be a trivalent connected 2-stratifold .
the following are equivalent : + ( 1 ) @xmath81 is 1-connected + ( 2 ) @xmath64 is a tree with all white vertices of genus @xmath65 and all terminal vertices white such that the components of @xmath133 are barycentrically subdivided rooted @xmath91-labeled trees and the reduced graph @xmath134 contains no horned tree .
if @xmath81 is 1-connected then by proposition [ simplyconnected ] and lemma [ 12-components ] the components of @xmath151 are barycentrically subdivided rooted @xmath91-labeled trees .
if the reduced graph @xmath134 contains a horned tree @xmath152 let @xmath7 be the component of @xmath134 containing @xmath152 .
note that @xmath152 is a pruned subgraph of @xmath7 and since @xmath153 , it follows from remark [ pruning ] that @xmath154 .
lemma [ h(rg ) ] then shows that @xmath155 .
hence @xmath134 does not contain a horned tree .
+ conversely , suppose the components of @xmath151 are barycentrically subdivided rooted @xmath91-labeled trees and @xmath134 contains no horned trees .
let @xmath7 be a component of @xmath134 .
+ first we show by induction on @xmath156 number of black vertices of degree @xmath9 in @xmath7 , that @xmath157 .
+ if @xmath158 , then @xmath7 is a @xmath90-tree with at most two @xmath159-trees attached to its terminal vertices , and so @xmath160 .
+ let @xmath161 .
we claim that at least one terminal label of @xmath7 is @xmath1 .
+ if not , then @xmath7 satisfies conditions ( 1 ) and ( 3 ) of the definition of horned tree .
we can find a sequence @xmath162 , where @xmath163 is a horned tree and @xmath164 is obtained from @xmath98 by deleting all but two components of @xmath165 for some nonterminal white vertex @xmath101 of @xmath98 , contradicting the assumption that @xmath134 contains no horned trees .
this proves the claim . + now let @xmath166 where @xmath167 is a terminal edge of @xmath7 with label @xmath1 .
then @xmath168 which by induction is @xmath65 .
therefore @xmath169 and it follows from theorem [ pi = h ] that @xmath81 is @xmath1-connected .
on a labeled graph @xmath170 with all white vertices of genus @xmath65 consider the following operation @xmath171 that changes @xmath170 to @xmath172 : , title="fig:",width=288 ] @xmath173 if @xmath174 is the black vertex incident to the edge labeled @xmath110 then the corresponding relation @xmath175 in @xmath176 is changed to the relations @xmath177 , @xmath178 , @xmath179 in @xmath180 and it follows that @xmath180 is a quotient of @xmath176 . in particular we note : [ remop1 ] if @xmath4 is simpy connected , then operation @xmath171 does not change the fundamental group . [ op1 ]
let @xmath4 be a trivalent @xmath0-stratifold such that @xmath181 is a tree with all white vertices of label @xmath65 , all terminal vertices white , and all edge labels @xmath1 .
then @xmath181 can be reconstructed from any white vertex @xmath101 of @xmath181 by successively performing @xmath171 .
if @xmath170 consists of @xmath101 only , there is nothing to show .
clearly the lemma is true if @xmath181 has only one black vertex .
let @xmath83 be a black vertex incident to @xmath101 .
deleting @xmath83 and its three incident edges from @xmath66 , we obtain three subtrees @xmath182 , @xmath183 , @xmath184 that satisfy the conditions of the lemma and with fewer black vertices than @xmath66 .
denote by @xmath145 , @xmath185 , @xmath186 the three white vertices adjacent to @xmath83 , where @xmath187 , @xmath188 and @xmath189 . by induction on the number of black vertices , @xmath182 is obtained from @xmath101 by repeated applications of operation @xmath171 .
now apply @xmath171 to @xmath190 to put back @xmath83 with its three edges and vertices @xmath185 and @xmath186 , then apply a sequence of @xmath171 s to @xmath185 and to @xmath186 to engulf @xmath183 and @xmath191 .
[ 111model ] let @xmath4 be a trivalent @xmath0-stratifold such that each edge of @xmath181 has label @xmath1 .
then the following are equivalent : + ( 1 ) @xmath192 . + ( 2 ) @xmath181 is a tree with all white vertices of label @xmath65 and all terminal vertices white .
+ ( 3 ) @xmath181 can be constructed from the @xmath90-graph by successively performing operation @xmath171 .
( 1)@xmath193(2 ) by corollary 1 of @xcite . + ( 2)@xmath193(3 )
by induction on the number of black vertice of @xmath181 : if this number is @xmath1 , then @xmath181 is a @xmath90-graph , so suppose @xmath181 has at least two black vertices . by corollary 1 of @xcite
, @xmath181 contains a @xmath90-subgraph @xmath194 with a white vertex @xmath101 which is a terminal vertex of @xmath181 .
let @xmath83 be the black vertex and @xmath145 , @xmath185 the other white vertices of @xmath194 .
deleting the edges of @xmath194 together with @xmath83 and @xmath101 splits @xmath181 into two subgraphs @xmath195 and @xmath196 , each with fewer black vertices than @xmath181 and @xmath197 , @xmath198 .
now @xmath199 and @xmath200 are simply - connected and satisfy the conditions of lemma [ op1 ] . by induction
, @xmath195 is obtained from @xmath194 by successively performing operation @xmath171 .
one further operation @xmath171 ( starting at @xmath145 ) adds @xmath194 to @xmath195 and by lemma [ op1 ] we can add @xmath196 by performing successively operation @xmath171 , starting at @xmath185 .
+ ( 3)@xmath193(1 ) : the @xmath90-graph is simply connected and by remark [ remop1 ] operation @xmath171 does not change the fundamental group .
for two disjoint labeled graphs @xmath201 and @xmath202 with all white vertices of genus @xmath65 , operation @xmath203 described in figure 7 , creates a new graph @xmath204 : + we now describe the collection @xmath10 of all trivalent graphs that can be obtained from a single white vertex by successively applying operations @xmath171 and @xmath209 . + for a collection @xmath210 of bipartite labeled graphs denote by @xmath211 the collection of all compact , connected bipartite labeled graphs obtained by starting with any @xmath212 and successively performing operations @xmath171 or @xmath209 .
we express this as + @xmath213 + let @xmath214 denote ( the collection containing only ) the graph consisting of one white vertex and let @xmath215 .
+ for two connected bipartite labeled graphs @xmath66 and @xmath182 denote by @xmath216{heil_symbol_y_v2.pdf}}\gamma'$ ] a graph obtained by joining any white vertex of @xmath66 to any white vertex of @xmath182 by operation @xmath217 .
note that that there are @xmath218 such @xmath216{heil_symbol_y_v2.pdf}}\gamma'$ ] , where @xmath127 ( resp .
@xmath219 ) is the number of white vertices of @xmath66 ( resp .
@xmath182 ) . let + @xmath220{heil_symbol_y_v2.pdf}}\mathcal{g}_0 = \{\,\gamma { \includegraphics[width=1.3em]{heil_symbol_y_v2.pdf}}\gamma'\,\vert\,\gamma , \gamma ' \in \mathcal{g}_0\,\}$ ] + in particular , @xmath220{heil_symbol_y_v2.pdf}}\emptyset = \mathcal{g}_0 $ ] and @xmath221{heil_symbol_y_v2.pdf}}\emptyset = \emptyset$ ] .
let + @xmath222{heil_symbol_y_v2.pdf}}\mathcal{g}_0}$ ] , and inductively @xmath223{heil_symbol_y_v2.pdf}}\mathcal{g}_n}$ ] + then @xmath224 suppose @xmath192 . if @xmath181 has no black vertices , or exactly one black vertex , then @xmath226 . in any case
@xmath181 is a tree with all white vertices of genus @xmath65 and all terminal edges white .
furthermore by lemma 4 of @xcite , @xmath181 contains a terminal vertex @xmath101 with incident edge @xmath95 of label @xmath1 .
let @xmath83 be the black vertex incident to @xmath95 .
then the star of @xmath83 in @xmath181 is a @xmath89-graph or a @xmath90-graph . in the first case star@xmath227
has two ( open ) edges @xmath95,@xmath228 where @xmath228 has label @xmath0 .
the subgraph @xmath229 obtained from @xmath181 by deleting @xmath230 is simply connected ( by remark [ remop2 ] ) . by induction on the number of black vertices @xmath231 and since @xmath232 is obtained from @xmath233 by operation @xmath209 , it follows that @xmath225 . in the second case , star(@xmath83 ) has three edges @xmath95,@xmath228,@xmath234 , each with label @xmath1 .
the subgraphs @xmath229 and @xmath235 obtained from @xmath181 by deleting @xmath236 are simply connected by remark [ remop1 * ] . by induction , @xmath237 and @xmath235
are in @xmath10 and since @xmath232 is obtained from @xmath233 and @xmath235 by operation @xmath217 , it follows that @xmath225 .
p. lum , g. singh , j. carlsson , a. lehman , t. ishkhanov , m. vejdemo - johansson , m. alagappan , g. carlsson , extracting insights from the shape of complex data using topology .
nature scientific reports 3 , 12 - 36 ( 2013 ) . | trivalent @xmath0-stratifolds are a generalization of @xmath0-manifolds in that there are disjoint simple closed curves where three sheets meet .
we obtain a classification of @xmath1-connected @xmath0-stratifolds in terms of their associated labeled graphs and develop operations that will construct from a single vertex all graphs that represent @xmath1-connected @xmath0-stratifolds .
mathematics subject classification : 57m20 , 57m05 , 57m15 keywords : stratifold , simply connected , trivalent graph . |
the aim of this paper is to investigate some particular quantum processes taking place in an arbitrary space - dependent scalar potential and a time- and space - dependent vector potential .
vector potential is periodic in time and describes a laser field .
its space - dependence results from the interaction of the laser field with electrons in solids .
such conditions are met for example in semiconductor nanostructures @xcite ( like quantum wires or wells ) , photoemission from a metal tip @xcite , carbon nanotubes or graphene @xcite or in surface physics @xcite . to make our presentation as clear as possible we shall restrict ourselves to the one - space - dimensional case , although extension of the presented method to systems of higher dimensionality is possible ( see , e.g. @xcite ) .
we shall apply our method to investigation of the photo - emission process .
this paper is organized as follows . in sec .
ii the most general solution of the schrdinger equation is introduced .
the transfer - matrix method and matching conditions are analyzed in sec .
iii , whereas reflection and transition probabilities are introduced in sec .
iv . these probabilities must sum up to 1 , which puts a very strong check for the accuracy of numerical calculations .
the most important part of this paper , i.e. the concept of the scattering - matrix method , is discussed in the next section , where it is shown why the scattering - matrix algorithm has to be introduced , instead of a much simpler transfer - matrix algorithm .
numerical illustrations of the applicability of this algorithm are presented in sec .
vi , and are followed by short conclusions . in our numerical illustrations we use atomic units , if otherwise stated .
let us start with one - dimensional schrdinger equation of the form @xcite , @xmath0 \psi(x , t ) .
\label{sec_1_1}\end{aligned}\ ] ] space - dependent mass @xmath1 , scalar potential @xmath2 and vector potential @xmath3 are spatially constant in finite intervals .
their values in any interval @xmath4 will be denoted as @xmath5 , @xmath6 and @xmath7 .
we require also that the function @xmath3 is periodic in time , that is @xmath8 where @xmath9 and @xmath10 is the frequency of the oscillating in time electric field . defining in a standard way the probability density @xmath11 , @xmath12 and the probability current @xmath13 , @xmath14^{\mathrm { * } } , \nonumber \label{sec_1_4}\end{aligned}\ ] ] we show using eq .
( [ sec_1_1 ] ) that the conservation of probability condition is satisfied .
indeed , assuming the above definitions , we get the continuity equation , @xmath15 space dependence of mass in eq .
( [ sec_1_1 ] ) forces one to impose non - standard continuity conditions on any solution of this equation .
it is now the wavefunction @xmath16 and the quantity @xmath17 that have to be continuous at points of discontinuity of mass @xmath1 and both potentials @xmath2 and @xmath3 @xcite . before passing to a general solution @xmath16 of eq .
( [ sec_1_1 ] ) in any given interval @xmath4 , which we shall denote as @xmath18 , let us note that due to time periodicity of the hamiltonian , @xmath18 can be chosen such that the floquet condition , @xmath19 is satisfied , where @xmath20 is the so - called quasienergy . a general solution @xmath18 of eq .
( [ sec_1_1 ] ) in any interval @xmath21 takes then the following form @xcite , @xmath22 where @xmath23 are arbitrary complex numbers to be determined and @xmath24 with @xmath25 being the ponderomotive energy , where @xmath26 means the time - average of @xmath27 over the laser - field oscillation .
components for which @xmath28 are purely imaginary are called closed channels .
these channels are not observed for a particle in initial or final states , but they have to be taken into account in order to satisfy the unitary condition of the time evolution . in a general case ,
the @xmath29 functions are components of the following fourier expansion , @xmath30 provided that the vector potential @xmath3 is periodic in time .
functions @xmath31 are defined as follows : @xmath32\mathrm{d } t.\ ] ] it is easily seen from the above equation that the @xmath29 functions depend on the form of the vector potential @xmath3 , that is on the laser field applied .
continuity conditions discussed above and applied to a general solution ( [ sec_1_7 ] ) of the schrdinger equation ( [ sec_1_1 ] ) lead to an infinite chain of equations connecting constants @xmath23 in the neighboring domains .
these matching conditions can be written in the matrix form , @xmath33 where @xmath34_{n}$ ] are the components of the columns @xmath35 .
the matrices @xmath36 and @xmath37 are defined as follows , @xmath38 the elements of @xmath36 can be computed in the following way . for an arbitrary function @xmath3 ,
periodic in time with the period @xmath39 we have @xmath40 where @xmath41 . in the interval
@xmath21 coefficients @xmath42 assume constant values , which we shall denote as @xmath43 .
using the condition of the continuity of the wavefunction @xmath18 at the point @xmath44 , we compute the elements of the matrices @xmath45 and @xmath46 , @xmath47 on the other hand elements of the @xmath48 matrix can be evaluated by substituting a general solution ( [ sec_1_7 ] ) to the expression ( [ sec_1_5 ] ) and applying the continuity condition to it at @xmath44 . after some algebraic manipulations we obtain finally the expression for the @xmath48-matrices , @xmath49 and a set of equations for vectors @xmath37 , @xmath50 where @xmath51^{-1 } b(i-1,x_{i-1 } ) .
\label{sec_2_12}\ ] ] these relations allow to connect a solution in a given domain @xmath52 with an analogous solution in any other domain @xmath53 , @xmath54 where @xmath55 is the so - called transfer matrix @xcite .
it is clear now that on the basis of eq.([sec_2_7a ] ) we can connect solutions in the boundary domains @xmath56 and @xmath57 .
values of mass @xmath1 , scalar potential @xmath2 and vector potential @xmath3 in these domains will be denoted as @xmath58 , @xmath59 , @xmath60 and @xmath61 , @xmath62 , @xmath63 , respectively .
we can then write down solutions of ( [ sec_1_1 ] ) for each of these domains .
these solutions represent incident ( @xmath64 ) , reflected ( @xmath65 ) and transmitted ( @xmath66 ) waves , and take the following form , @xmath67 @xmath68 @xmath69 where @xmath70 constants @xmath71 and @xmath72 will be denoted from now on as @xmath73 and @xmath74 , respectively . using continuity conditions for functions defined above
, we get the probability conservation equation for reflection and transition amplitudes , @xmath73 and @xmath74 , @xmath75 where summations are over such @xmath76 for which @xmath77 and @xmath78 are real , i.e. , over the open channels .
this equation permits us to interpret @xmath79 and @xmath80 as reflection and transition probabilities for a tunneling process in which absorption ( @xmath81 ) or emission ( @xmath82 ) of energy @xmath83 by electrons occurred @xcite . in case of a monochromatic laser field
this process can be interpreted as absorption or emission of @xmath76 photons from the laser field .
the unitary condition ( [ unit_5 ] ) can be also interpreted as the conservation of electric charge . to this end , let us define the quantities proportional to the density of electric currents , @xmath84 then eq . ( [ unit_5 ] ) adopts the form of the first kirchhoff low , @xmath85 using ( [ sec_2_7a ] ) we can calculate constants @xmath86 and @xmath87 appearing in equations ( [ unit_1 ] ) - ( [ unit_3 ] ) . indeed , since @xmath88 where transfer matrix @xmath89 , and because @xmath90 , @xmath91 and @xmath92 adopt the following block forms , @xmath93 we arrive at @xmath94 where @xmath95 and @xmath96 denote the columns of @xmath73 i @xmath74 , and @xmath97_{n}=\delta_{0,n}$ ] .
thus , after some algebraic manipulations , we have , @xmath98 which allows us to determine the quantities @xmath73 and @xmath74 for a given transfer matrix @xmath90 . for open channels , these quantities are the amplitudes of reflection ( @xmath73 ) and transition ( @xmath74 ) probabilities , from which one can compute reflection and transition probabilities using equations ( [ unit_6 ] ) and ( [ unit_7 ] ) .
we note from equations ( [ sec_2_5 ] ) and ( [ sec_2_10 ] ) that each of the @xmath99 matrices that constitute the transfer matrix @xmath55 contain elements @xmath100 that depend on the @xmath101 coordinates at which the discontinuities appear . for closed channels , that is when the @xmath102 momenta are purely imaginary , these numbers are real and may assume arbitrary values , very large or very small , depending again on the @xmath101 coordinates .
number of the @xmath99 matrices is equal to the number of discontinuity points , that is it depends on how we divide the space into short intervals in order to make our potential tractable by our algorithm .
it may therefore turn out that in order to compute the transfer matrix @xmath55 , we have to multiply a large number of the @xmath99 matrices , each containing both very small and very large numbers .
it is clear that such a procedure is numerically unstable .
we have to find a way to modify our method of calculations in order to compute the elements of each @xmath99 matrix at the same point @xmath103 independently of where the ` real ' @xmath101 is .
this would eliminate `` dangerous '' @xmath100 elements ( turning them to @xmath104 ) , however at the cost of appearing somewhere else .
we shall see later that these ` left - overs ' of the shift into @xmath103 appear only as differences @xmath105 and therefore do not cause any harmful side - effects .
we shall see now that such a modification is possible and the price we pay for it is worth the effort .
it follows from eq .
( [ sec_2_7a ] ) that in the neighboring domains , @xmath106 and @xmath4 , we have , @xmath107 although the elements of the transfer matrix @xmath108 have been computed from the continuity conditions at point @xmath44 , one can compute them at any other point , for example @xmath103 . to this end , let us notice what follows from the form of the solution ( [ sec_1_7 ] ) .
translation of the system by a certain distance @xmath109 along the @xmath110-axis causes only multiplication of each member of the sum over @xmath76 in ( [ sec_1_7 ] ) by a constant @xmath111 .
these constants can be included in coefficients @xmath23 . in this way
we get a new set of constants which we shall denote as @xmath112 , @xmath113 we shall interpret these constants as coefficients in solution ( [ sec_1_7 ] ) , given by the continuity conditions at point @xmath114 .
( [ sec_3_2 ] ) written in the matrix form becomes , @xmath115 where @xmath116 and @xmath117 in the equation above @xmath118 is a diagonal matrix , @xmath119_{nn'}=\delta_{nn'}\exp{(\mathrm{i}\sigma p_{in}\delta ) } , \label{sec_3_5}\ ] ] whereas @xmath35 and @xmath120 are the columns of the constants @xmath121 and @xmath122 respectively , that is @xmath123_{n}=c_{in}^{\pm}$ ] and @xmath124_{n}=c_{in}^{\pm}$ ] .
it follows from the form of the matrix @xmath125 that the following relations are satisfied : @xmath126 @xmath127 let us notice also that translation of the system defined above modifies the transfer matrix @xmath108 .
we have @xmath128 thus @xmath129 and we can write it down as @xmath130 where @xmath131 matrix elements denoted with the tilde symbol refer to the translated system . using the method defined above and the relation ( [ sec_2_7a ] ) , we can connect now the solution in the domain @xmath56 with the solution in any other domain @xmath4 . in this way the elements of the transfer matrix , which have been computed until now at the points of discontinuity @xmath132 , are computed now each time at the same point @xmath103 .
let us illustrate this method for a special case of @xmath133 @xmath134 equation ( [ sec_3_12 ] ) connects constants @xmath91 and @xmath135 using the matrices @xmath136 all computed at @xmath103 independently of @xmath137 , and diagonal matrices @xmath138 , given by the relations ( [ sec_3_4b ] ) and ( [ sec_3_5 ] ) , where @xmath139 . edge matrices @xmath140 and @xmath141 in the equation ( [ sec_3_12 ] ) can be omitted while computing the transmission and reflection probability amplitudes since their only role is to multiply the amplitudes by phase quotients which disappear while computing the probabilities .
although these matrices lead to significant modifications of the closed channels in the domains of @xmath142 and @xmath143 in this particular case , these channels do not influence the reflection and transition amplitudes .
transmission and reflection probabilities can thus be computed using a modified transfer matrix , @xmath144 the matrices @xmath145 are equal to the matrices @xmath146 in eq .
( [ sec_2_12 ] ) calculated however for @xmath147 .
this fact speeds up numerical calculations since now matrix @xmath148 in eq .
( [ sec_2_12 ] ) have to be inverted only once .
further on we shall omit the superscript @xmath149 in @xmath90 and the tilde over @xmath150 in order to simplify notation . the method presented above
is still numerically unstable .
the reason for this instability lies in the existence of large numerical values of elements of the @xmath151 matrix for imaginary momenta @xmath28 .
in other words , for @xmath152 the source of numerical instabilities are matrix elements @xmath153 that contain large numbers .
there is however a chance for improving the stability , if only its reverse will be used , @xmath154 .
it appears that it is possible provided that in our numerical algorithm only the so - called scattering matrix will be applied . for this reason we will show below how to compute the scattering matrix , @xmath155 , using only elements of the transfer matrix , @xmath156 . for the transfer matrix @xmath156
we have , @xmath157 thus , @xmath158 on the basis of ( [ sec_3_16 ] ) we now want to compute the elements of the @xmath155 matrix .
this matrix is supposed to connect the coefficients @xmath35 and @xmath159 in the following way , @xmath160 using the set of linear equations ( [ sec_3_16 ] ) , we easily compute the coefficients @xmath161 and @xmath162 on the left - hand side of equation ( [ sec_3_17 ] ) , as functions of the coefficients @xmath163 and @xmath164 .
we get then the following relations , @xmath165 finally we compute the elements of the matrix @xmath155 , @xmath166 as expected , the matrix @xmath155 contains only numerically stable elements @xmath167 .
it follows from eq .
( [ sec_2_7a ] ) that the transfer matrix @xmath156 can be written as the product of two transfer matrices , @xmath168 and @xmath169 ( @xmath170 ) , @xmath171 where matrices @xmath168 and @xmath169 are defined as follows , @xmath172 applying the method presented above , for each of the transfer matrices @xmath168 and @xmath169 we can construct a scattering matrix , @xmath173 and @xmath174 respectively .
elements of the scattering matrix @xmath155 can be computed using only elements of the matrices @xmath173 and @xmath174 . using the notation above ,
we obtain the following expressions for the elements of the @xmath155 matrix , @xmath175 it is clear from the above that the @xmath155 matrix is not merely a product of two matrices @xmath173 and @xmath174 , but rather a complicated nonlinear composition of them .
it is important however to note that despite its evident complexity , such a construction of the scattering matrix is numerically stable , as opposed to the transfer matrix method which fails if a system with a large number of discontinuity points @xmath101 is considered .
stability of such an algorithm has been proven in our numerical investigations by checking that the condition ( [ unit_5 ] ) is satisfied with an error smaller than @xmath176 .
such an accuracy can never be achieved for systems with a large number of discontinuity points if the transfer matrix is applied .
in our model investigations , we concentrate on some essential features of the solid - vacuum interface , as exemplified by the sommerfeld model , in which the band structure is neglected .
this simplification allows us to consider a quite general form of the laser field . to be more specific ,
the solid surface is described by a continuous step potential @xmath177 with @xmath178 the parameter @xmath179 determines the skin depth of a surface . for @xmath180 ,
the surface potential represents the step function , commonly used in the sommerfeld model . in our illustrations
, we put @xmath181 .
we apply our theory to the gold surface and assume that the electron effective mass is close to the free electron mass . the work function and the fermi energy for the gold metal
are equal to 5.1 and 5.53 ev , respectively .
this means that the constant @xmath182 above ( as the sum of the work function and the fermi energy ) equals 10.63 ev .
the surface potential described above can be generalized further to meet conditions suitable for other solids . in particular
, one can take into account the space - dependent effective mass of electrons in semiconductor heterostructures or metals with effective masses different from the free electron mass .
on the other hand , the form of the laser field is assumed to depend on both space and time coordinates . since , for laser pulses of duration @xmath183 fs and the 800 nm wavelength , the monochromatic approximation works well , we therefore adopt the following form for the laser electric field : @xmath184 where @xmath185 and similarly @xmath186 the parameter @xmath187 defines the plasmon - enhanced part of the laser field . for the incident laser beam ,
we choose the ti : sapphire laser beam of frequency @xmath188 ev ( @xmath189 nm ) .
this means that inside the solid the laser field intensity averaged over the time period decays exponentially , @xmath190 on the other hand , in vacuum , it stays constant close to the surface , and then again decays exponentially . in this way
, we can mimic a real physical situation in which the radiation - filled space is finite . in our illustrations , we take @xmath191 , which means that the penetration depth of the laser field intensity equals @xmath192 . the parameter @xmath193 describes the distance in a solid at which the intensity is not reduced substantially . on the other hand , @xmath194 corresponds to the laser focus diameter in vacuum , whereas @xmath195 alone determines the intensity reduction rate outside the focus .
similar parameters with the subscript @xmath196 refer to the plasmon - enhanced part of the laser field .
the remaining parameters have been chosen as follows : @xmath197 , @xmath198 , @xmath199 , @xmath200 , @xmath201 , @xmath202 , @xmath203 , and @xmath204 .
all dimensional parameters are in atomic units . , @xmath205 and @xmath206 , respectively .
, width=264 ] in our discussions presented below , the laser field intensity is characterized by the dimensionless parameter @xmath207 , where @xmath208 is the ponderomotive energy of electrons in the monochromatic electromagnetic plane wave of frequency @xmath10 ; hence @xmath209 . in fig .
[ syst ] we draw the space - dependence of the continuous step potential @xmath2 and the electric field amplitude @xmath210 for @xmath211 and @xmath212 .
and for six values of @xmath187 .
, width=264 ] the total photo - emission probability is equal to @xmath213 we plot it in fig .
[ trans10 ] as a function of the electron kinetic energy for @xmath212 and for six values of @xmath187 .
we clearly see the multi - photon structure in this distribution , i.e. , the total probability jumps sometimes by a few orders of magnitude if the smaller number of laser photons is sufficient for photo - emission .
as expected , the plasmon effect usually increases the photo - emission probability .
moreover , the energy of the multi - photon channel opening increases with increasing @xmath187 , which is due to the increase of the space - dependent ponderomotive energy of the laser field .
the significance of this effect for the tunneling phenomena is going to be discussed in due course .
as mentioned above , our algorithm is convergent provided that a sufficient number of discretization points is introduced . for systems considered here , this number should not be smaller than 100 .
if the laser field is very weak , this does not create significant numerical problems , except that calculations become longer .
however , when the laser field is sufficiently intense , the algorithm based on the transfer matrix is unstable .
this instability is due to the existence of closed channels , which introduce into numerical calculations very small and very large numbers at the same time . augmenting precisions
significantly slows down the calculation and does not diminish the problem .
we have found that it is possible to make this algorithm numerically stable by just applying nonlinear matrix transformations , without introducing higher precisions .
illustrations presented in this paper show that photo - emission of electrons can be changed significantly by applied nonperturbative oscillating in time and space - dependent electric field .
the efficiency of the algorithm presented in this contribution opens up the possibility of investigating surface phenomena in the presence of more realistic laser pulses that gradually decrease within solids and extend on a mesoscopic scale in vacuum .
this work is supported by the polish national science center ( ncn ) under grant no .
2012/05/b / st2/02547 . | we develop the concept of scattering matrix and we use it to perform stable numerical calculations of photo - emission from nano - tips .
electrons move in an external space and time dependent nonperturbative electric field .
we apply our algorithm for different strengths and spatial configurations of the field .
pacs : 03.65.xp,72.20.dp,73.40.gk |
interdependent behaviour and causality in coupled complex systems continue to attract considerable interest in fields as diverse as solid state science , biology , physiology , climatology @xcite . coupling and synchronization effects
have been observed for example in cardiorespiratory interactions , in neural signals , in glacial variability and in milankovitch forcing @xcite . in finance , the _ leverage effect _ quantifies the cause - effect relation between return @xmath9 and volatility @xmath10 and eventually financial risk estimates @xcite . in dna sequences ,
causal connections among structural and compositional properties such as intrinsic curvature , flexibility , stacking energy , nucleotide composition are sought to unravel the mechanisms underlying biological processes in cells @xcite .
many issues still remain unsolved mostly due to problems with the accuracy and resolution of coupling estimates in long - range correlated signals .
such signals do not show the wide - sense - stationarity needed to yield statistically meaningful information when cross - correlations and cross - spectra are estimated . in @xcite ,
a function @xmath11 , based on the detrended fluctuation analysis - a measure of autocorrelation of a series at different scales @xmath4 - has been proposed to estimate the cross - correlation of two series @xmath1 and @xmath2 .
however , the function @xmath11 is independent of the lag @xmath3 , since it is a straightforward generalization of the detrended fluctuation analysis , which is a _ positive - defined _ measure of autocorrelation for long - range correlated series .
therefore , @xmath11 holds only for @xmath12 .
different from autocorrelation , the cross - correlation of two long - range correlated signals is a _ non - positive - defined function of @xmath3 _ , since the coupling could be delayed and have any sign . in this work ,
a method to estimate the cross - correlation function @xmath0 between two long - range correlated signals at different scales @xmath4 and lags @xmath3 is developed .
the asymptotic expression of @xmath0 is worked out for fractional brownian motions @xmath13 , @xmath14 being the hurst exponent , whose interest follows from their widespread use for modeling long - range correlated processes in different areas @xcite .
finally , the method is used to investigate the coupling between ( _ i _ ) returns and volatility of the dax stock index and ( _ ii _ ) structural properties , such as deformability , stacking energy , position preference and propeller twist , of the escherichia coli chromosome .
the proposed method operates : ( i ) on the integrated rather than on the increment series , thus yielding the cross - correlation at varying windows @xmath4 , as opposed to the standard cross - correlation ; ( ii ) as a sliding product of two series , thus yielding the cross - correlation as a function of the lag @xmath3 , as opposed to the method proposed in @xcite .
the features ( i ) and ( ii ) imply higher accuracy , @xmath4-windowed resolution while capturing the cross - correlation at varying lags @xmath3 .
the _ cross - correlation _ @xmath15 of two nonstationary stochastic processes @xmath1 and @xmath2 is defined as : @xmath16[y^\ast(t+\tau)-\eta_y^\ast(t+\tau)]\big\rangle\ ] ] where @xmath17 and @xmath18 indicate time - dependent means of @xmath1 and @xmath19 , the symbol @xmath20 indicates the complex conjugate and the brackets @xmath21 indicate the ensemble average over the joint domain of @xmath1 and @xmath19 .
this relationship holds for space dependent sequences , as for example the chromosomes , by replacing time with space coordinate .
( [ crosscovariance ] ) yields sound information provided the two quantities in square parentheses are jointly stationary and thus @xmath22 is a function only of the lag @xmath3 . in this work
, we propose to estimate the cross - correlation of two nonstationary signals by choosing for @xmath17 and @xmath18 in eq .
( [ crosscovariance ] ) , respectively the functions : @xmath23 and @xmath24 the wide - sense stationarity of eq .
( [ crosscovariance ] ) can be demonstrated for fractional brownian motions . by taking
@xmath25 , @xmath26 , @xmath17 and @xmath18 calculated according to eqs .
( [ xtil],[ytil ] ) , @xmath15 writes : @xmath27\big[b_{h_2}^*(t+\tau)-\widetilde{b}_{h_2}^*(t+\tau)\big]\big\rangle \;\;\;.\end{aligned}\ ] ] when writing @xmath25 and @xmath26 , we assume the same underlying generating noise @xmath28 to produce a sample of @xmath29 and @xmath30 .
( [ dcab0 ] ) is calculated in the limit of large @xmath4 ( calculation details are reported in the appendix ) .
one obtains : @xmath31\;\;\;,\end{aligned}\ ] ] where @xmath32 is the _ scaled lag _ and @xmath33 is defined in the appendix .
( [ theta ] ) is independent of @xmath34 , since the terms in square parentheses depend only on @xmath35 , and thus eq . ( [ crosscovariance ] )
is made wide - sense stationary .
it is worthy of note that , in eq .
( [ theta ] ) , the coupling between @xmath36 and @xmath37 reduces to the sum of the exponents @xmath38 .
( [ theta ] ) , for @xmath12 , reduces to : @xmath39 indicating that the coupling between @xmath36 and @xmath37 scales as the product of @xmath40 and @xmath41 .
the property of the variance of fractional brownian motion @xmath42 to scale as @xmath43 is recovered from the eq .
( [ zero ] ) for @xmath44 and @xmath45 , i.e. : @xmath46 eq .
( [ auto ] ) has been studied in @xcite .
the leverage effect is a _ stylized fact _ of finance .
the level of volatility is related to whether returns are negative or positive .
volatility rises when a stock s price drops and falls when the stock goes up @xcite .
furthermore , the impact of negative returns on volatility seems much stronger than the impact of positive returns ( _ down market effect _ ) @xcite . to illustrate these effects
, we analyze the correlation between returns and volatility of the dax stock index @xmath47 , sampled every minute from 2-jan-1997 to 22-mar-2004 , shown in fig .
[ fig : figure1 ] ( a ) . the returns and volatility
are defined respectively as : @xmath48 and @xmath49 ^ 2}/{(t-1 ) } } \;.$ ] dax stock index : ( a ) prices ; ( b ) returns with @xmath50 ; ( c ) volatility with @xmath51 ; ( d ) volatility with @xmath52.,width=302 ] fig .
[ fig : figure1 ] ( b ) shows the returns for @xmath53 .
the volatility series are shown in figs .
[ fig : figure1 ] ( c , d ) respectively for @xmath51 and @xmath52 .
the hurst exponents , calculated by the slope of the log - log plot of eq .
( [ auto ] ) as a function of @xmath4 , are @xmath54 ( return ) , @xmath55 ( volatility @xmath51 ) and @xmath56 ( volatility @xmath52 ) .
[ fig : figure2 ] shows the log - log plots of @xmath57 for the returns ( squares ) and volatility with @xmath58 ( triangles ) .
the scaling - law exhibited by the dax series guarantees that its behaviour is a fractional brownian motion .
the function @xmath59 with @xmath60 and @xmath61 with @xmath52 is also plotted at varying @xmath4 in fig .
[ fig : figure2 ] ( circles ) . from the slope of the log - log plot of @xmath59 vs @xmath4 , one obtains @xmath62 , i.e. the average between @xmath5 and @xmath6 as expected from eq .
( [ zero ] ) .
+ next , the cross - correlation is considered as a function of @xmath3 . the plots of @xmath0 for @xmath60 and @xmath61 with @xmath51 and @xmath52 are shown respectively in fig . [
fig : figure3 ] ( a , b ) at different windows @xmath4 .
log - log plot of @xmath57 for the dax return ( squares ) and volatility ( triangles ) and of @xmath59 with @xmath60 and @xmath61 ( circles ) .
red lines are linear fits .
the power - law behaviour is consistent with eqs .
( [ zero],[auto]).,width=302 ] cross - correlation @xmath0 with @xmath60 and @xmath61 with ( a ) @xmath51 and ( b ) @xmath52 ; ( c ) with @xmath60 and @xmath63 with @xmath52 .
@xmath4 ranges from 100 to 500 with step 100.,width=302 ] plot of the function @xmath64 with @xmath60 and @xmath61 with @xmath51 .
@xmath65 and @xmath66 @xmath4 ranges from 100 to 500 with step 100 . one can note that the five curves collapse , within the numerical errors of the parameters entering the auto- and cross - corerlation functions .
this is in accord with the invariance of the product @xmath64 with the window @xmath4.,width=302 ] leverage function with volatility windows @xmath67 = @xmath68 , @xmath69 , @xmath70 , @xmath71 .
the value of @xmath4 is @xmath72 equal for all the curves.,width=302 ] the function @xmath0 for @xmath60 and @xmath73 , is shown in fig .
[ fig : figure3](c ) .
the cross - correlation takes negative values at small @xmath3 and reaches the minimum at about 10 - 12 days .
this indicates that the volatility increases with negative returns ( i.e. with price drops ) .
then @xmath0 changes sign relaxing asymptotically to zero from positive values at large @xmath3 .
the positive values of @xmath0 indicate that the volatility decreases when the returns become positive ( i.e. when price rises ) and are related to the restored equilibrium within the market ( _ positive rebound days _ ) .
it is worthy of remark that the ( positive ) maximum of the cross - correlation is always smaller than the ( negative ) minimum .
this is the stylized fact known as _
down market effect_. a relevant feature exhibited by the curves in figs .
[ fig : figure3 ] ( a - c ) is that the zeroes and the extremes of @xmath0 occur at the same values of @xmath3 , which is consistent with wide - sense - stationarity for all the values of @xmath4 .
a further check of wide sense stationarity is provided by the plot of the function @xmath74 . in fig .
[ fig : figure4 ] , @xmath74 is plotted with @xmath60 and @xmath61 with @xmath51 , @xmath65 and @xmath66 , @xmath4 ranges from 100 to 500 with step 100 .
one can note that the five curves collapse in accord with the invariance of the product @xmath64 with @xmath4 . in fig .
[ fig : figure5 ] , the leverage correlation function @xmath75 according to the definition put forward in @xcite , is plotted for different volatility windows @xmath67 .
the function @xmath76 has been calculated by means of eq.([crosscovariance ] ) .
the negative values of cross - correlation ( at smaller @xmath3 ) and the following values ( _ positive rebound days _ ) at larger @xmath3 can be clearly observed for several volatility windows @xmath67 .
the function @xmath77 for the dax stock index , estimated by means of the standard cross - correlation function , is shown in figs .
1,2 of ref . @xcite . by comparing the curves shown in fig .
[ fig : figure5 ] to those of ref .
@xcite , one can note the higher resolution related to the possibility to detect the correlation at smaller lags ( note the @xmath3 unit is hours , while in ref.@xcite is days ) and at varying windows @xmath4 , implying the possibility to estimate the degree of cross - correlation at different frequencies .
as a final remark , we mention that the cross correlation function between a fractional brownian motion and its own width can be computed analytically in the large @xmath4 limit , following the derivation in the appendix for two general fbm s .
the width of a fbm is one possible definition for the volatility , therefore the derivation in the appendix provides a straightforward estimate of the leverage function .
several studies are being addressed to quantify cross - correlations among nucleotide position , intrinsic curvature and flexibility of the dna helix , that may ultimately shed light on biological processes , such as protein targeting and transcriptional regulation @xcite .
one problem to overcome is the comparison of dna fragments with di- and trinucleotide scales , hence the need of using high - precision numerical techniques .
we consider deformability , stacking energy , propeller twist and position preference sequences of the escherichia coli chromosome
. the sequences , with details about the methods used to synthetize / measure the structural properties , are available at the cbs database - center for biological sequence analysis of the technical university of denmark ( ) . in order to apply the proposed method , the average value is subtracted from the data , that are subsequently integrated to obtain the paths shown in fig . [
fig : figure6 ] .
the series are @xmath78 long and have hurst exponents : @xmath79 ( deformability ) , @xmath62 ( position preference ) , @xmath80 ( stacking energy ) , @xmath79 ( propeller twist ) .
the cross - correlation functions @xmath0 between deformability , stacking energy , propeller twist and position preference are shown in fig .
[ fig : figure7 ] ( a - e ) .
there is in general a remarkable cross - correlation along the dna chain indicating the existence of interrelated patches of the structural and compositional parameters .
the high correlation level between dna flexibility measures and protein complexes indicates that the conformation adopted by the dna bound to a protein depends on the inherent structural features of the dna .
it is worthy to remark that the present method provides the dependence of the coupling along the dna chain rather than simply the values of the linear correlation coefficient @xmath81 . in table 4 of ref . @xcite one can find the following values of the correlation obtained by either numerical analysis or experimental measurements ( in parentheses ) over dna fragments : ( a ) @xmath82 ; ( b ) @xmath83 ; ( c ) @xmath84 ; ( d ) @xmath85 ; ( e ) @xmath86
moreover , also for the genomic sequences the function @xmath64 is independent of @xmath4 within the numerical errors of the parameters entering the auto- and cross - correlation functions . in fig .
[ fig : figure8 ] , @xmath64 is shown for @xmath1 the deformability , @xmath2 the stacking energy , @xmath87 and @xmath88 .
@xmath4 ranges from 100 to 500 with step 100 .
structural sequences of the escherichia coli chromosome.,width=302 ] cross - correlation @xmath0 between ( a ) deformability and stacking energy ; ( b ) position preference and deformability ( c ) propeller twist and position preference ; ( d ) propeller twist and stacking energy ; ( e ) propeller twist and deformability .
@xmath4 ranges from 100 to 500 with step 100.,width=302 ] plot of the function @xmath64 with @xmath1 the deformability , @xmath2 the stacking energy , @xmath87 and @xmath88 .
@xmath4 ranges from 100 to 500 with step 100 .
one can note that the five curves collapse , within the numerical errors of the parameters entering the auto- and cross - corerlation functions .
this is in in accord with the invariance of the product @xmath64 with the window @xmath4.,width=302 ]
a high - resolution , lag - dependent non - parametric technique based on eqs .
( [ crosscovariance]-[ytil ] ) to measure cross - correlation in long range - correlated series has been developed .
the technique has been implemented on ( _ i _ ) financial returns and volatilities and ( _ ii _ ) structural properties of genomic sequences @xcite .
the results clearly show the existence of coupling regimes characterized by positive - negative feedback between the systems at different lags @xmath3 and windows @xmath4 .
we point out that - in principle - other methods might be generalized in order to yield estimates of the cross - correlation between long - range correlated series at varying @xmath3 and @xmath4 .
however , techniques operating over the series by means of a box division , such as dfa and r / s method , are _ a - priori _ excluded .
the box division causes discontinuities in the sliding product of the two series at the extremes of each box , and ultimately incorrect estimates of the cross - correlation .
the present method is not affected by this drawback , since eqs .
( [ crosscovariance]-[ytil ] ) do not require a box division .
let us start from eq .
( [ dcab0 ] ) : @xmath89\big[b_{h_2}^*(t+\tau)-\widetilde{b}_{h_2}^*(t+\tau)\big]\big\rangle \;\;\;,\end{aligned}\ ] ] that , after multiplying the terms in parentheses , becomes : @xmath90\big\rangle \;\;\;.\end{aligned}\ ] ] in general , the moving average may be referred to any point of the moving window , a feature expressed by replacing eqs .
( [ xtil],[ytil ] ) with @xmath91 with @xmath92 . in the limit of @xmath93 ,
the sums can be replaced by integrals , so that : @xmath94 where @xmath95 , @xmath96 , @xmath97 .
for the sake of simplicity , the analytical derivation will be done by using the harmonizable representation of the fractional brownian motion @xcite : @xmath98 where @xmath99 is a representation of @xmath28 in the @xmath100 domain . in the following
we will consider the case of @xmath101 and @xmath102 . by using eq .
( [ harmo ] ) , the cross - correlation of two fbms @xmath36 and @xmath103 can be written as : @xmath104 since @xmath105 is gaussian , the following property holds for any @xmath106 : @xmath107 by using eq .
( [ gaussian ] ) , after some algebra eq .
( [ xy ] ) writes : @xmath108 where @xmath33 is a normalization factor which depends on @xmath5 and @xmath6 . in the harmonizable representation of fbm , @xmath33 takes the following form @xcite : @xmath109\gamma[-(h_1+h_2)]\ ] ] normalized such that @xmath110 when @xmath111 .
different representations of the fbm lead to different values of the coefficient @xmath33 @xcite .
( [ teo2 ] ) can be used to calculate each of the four terms in the right hand side of eq .
( [ dca ] ) .
the mean value of each term in eq .
( [ dca ] ) is obtained from the general formula in eq .
( [ teo2 ] ) ; thus , substituting the right hand side of eq .
( [ teo2 ] ) and eq .
( [ txy ] ) into each term in eq .
( [ dca ] ) we obtain : @xmath112\end{aligned}\ ] ] where each term in round parentheses corresponds to each of the four terms in eq .
( [ dca ] ) . summing the terms in eq .
( [ dca1 ] ) , one can notice that time @xmath34 cancels out , thus one finally obtains : @xmath113\ ; , \nonumber \\\hspace{-25 mm } & \end{aligned}\ ] ] consistently with the large @xmath4 limit , we take @xmath114 , namely @xmath115 .
the integral ( [ integral ] ) admits four different solutions , depending on the values taken by the parameters @xmath116 and @xmath117 .
let us consider each case separately .
[ [ case-1-hattautheta - and - hattautheta1 ] ] case 1 : @xmath118 and @xmath119 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath120 \nonumber \\\hspace{-25mm}&\end{aligned}\ ] ] [ [ case-2-hattautheta - and - hattautheta1 ] ] case 2 : @xmath118 and @xmath121 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath122 \nonumber \\\hspace{-25mm}&\end{aligned}\ ] ] [ [ case-3-hattautheta - and - hattautheta1 ] ] case 3 : @xmath123 and @xmath119 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath124 \nonumber \\\hspace{-25mm}&\end{aligned}\ ] ] it is easy to see that this case includes the eq . ( 2.5 ) treated in the paper .
[ [ case-4-hattautheta - and - hattautheta1 ] ] case 4 : @xmath123 and @xmath121 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath125 \nonumber \\\hspace{-25mm}&\end{aligned}\ ] ]
10 m. rosenblum and a. pikovsky , ( 2007 ) phys . rev .
lett . * 98 * , 064101 . t. zhou , l. chen and k. aihara , ( 2005 ) phys
. rev .
lett . * 95 * , 178103 .
s. oberholzer et al .
( 2006 ) phys .
lett . * 96 * , 046804
. m. dhamala , g. rangarajan , ( 2008 ) m. ding , phys .
lett . * 100 * , 018701 .
f. verdes , ( 2005 ) phys .
e * 72 * , 026222
. m. palus and m. vejmelka , ( 2007 ) phys .
e * 75 * , 056211
. t. kreuz et al . ( 2007 ) physica d * 225 * , 29 .
lu - chun du and dong - cheng mei , ( 2008 ) j. stat .
tass et al .
( 1998 ) phys .
81 * , 3291 .
p. huybers , ( 2006 ) w. curry , nature * 441 * , 7091 . y. ashkenazy , ( 2006 ) climate dynamics * 27 * , 421
. f. black , ( 1976 ) j. of fin .
econ . * 3 * , 167 .
w. schwert , j. of finance ( 1989 ) * 44 * , 1115 .
r. haugen , e. talmor , w. torous , ( 1991 ) j. of finance * 44 * , 1115
. l. glosten , j. ravi and d. runkle , ( 1992 ) j. of finance * 48 * , 1779 .
g. bekaert , g. wu , ( 2000 ) the review of financial studies * 13 * , 1 .
s. figlewski , x. wang ( 2000 ) _ is the leverage effect a leverage effect ? _ , working paper , stern school of business , new york .
j. p. bouchaud , a. matacz and m. potters , ( 2001 ) phys .
* 87 * , 228701 . j. perello and j. masoliver , ( 2003 ) phys .
e * 67 * , 037102
. t. qiu , b. zheng , f. ren and s. trimper , ( 2006 ) phys .
e * 73 * , 065103(r ) .
r. donangelo , m. h. jensen , i. simonsen , k. sneppen , ( 2006 ) j. stat .
i varga - haszonits and i kondor , ( 2008 ) j. stat .
m. montero , ( 2007 ) j. stat .
j. moukhtar , e. fontaine , c. faivre - moskalenko and a. arneodo , ( 2007 ) phys .
98 * , 178101 .
allen , n.d .
price , a. joyce and b.o .
palsson , ( 2006 ) plos computational biology , * 2 * , e2 .
pedersen , l.j .
jensen , s. brunk , h.h .
staerfeld and d.w .
ussery , ( 2000 ) j. mol .
biol . * 299 * , 907 .
jun , g. oh and s. kim , ( 2006 ) phys .
e * 73 * , 066128 .
b. podobnik , h.e .
stanley , ( 2008 ) phys .
lett . * 100 * , 084102
. b. b. mandelbrot , j. w. van ness , ( 1968 ) siam rev . * 4 * , 422 .
a. carbone , g. castelli , h. e. stanley , phys .
e * 69 * , 026105 ( 2004 ) .
a. carbone , phys .
e * 76 * , 056703 ( 2007 ) .
s. arianos and a. carbone , physica a * 382 * , 9 ( 2007 ) .
a. carbone and h. e. stanley , physica a * 384 * , 21 ( 2007 ) .
a. carbone and h. e. stanley , physica a * 340 * , 544 ( 2004 ) . the matlab and c++ codes implementing the proposed method , the dax and e - coli sequences used in this work are downloadable at : | a method for estimating the cross - correlation @xmath0 of long - range correlated series @xmath1 and @xmath2 , at varying lags @xmath3 and scales @xmath4 , is proposed . for fractional brownian motions with hurst exponents @xmath5 and @xmath6 , the asymptotic expression of @xmath0 depends only on the lag @xmath3 ( wide - sense stationarity ) and scales as a power of @xmath4 with exponent @xmath7 for @xmath8 .
the method is illustrated on ( _ i _ ) financial series , to show the leverage effect ; ( _ ii _ ) genomic sequences , to estimate the correlations between structural parameters along the chromosomes .
* keywords * : persistence ( experiment ) , sequence analysis ( experiment ) , scaling in socio - economic systems , stochastic processes |
the fokker - planck equation has been the focus of many decades of study due to its relevance in physics , finance , probability and statistics @xcite .
@xcite provides a particularly early example examining analytically tractable solutions to the fokker - planck equation , with more contemporary examples provided by @xcite . essentially , this work focuses on time dependent densities @xmath0 , @xmath1 , of a diffusion process described by range @xmath2 given that it started at position @xmath3 , governed by , @xmath4 eq.([begin ] ) is defined on some interval on @xmath5 with endpoints @xmath6 where @xmath7 .
the continuous functions @xmath8 and @xmath9 are referred to as the diffusion , drift and the sink coefficients respectively .
the applications of eq.([begin ] ) are wide ranging . for @xmath10 , eq.([begin ] ) is commonly referred to as the fokker - planck equation and one can readily show that @xmath0 is conserved for @xmath1 .
the corresponding solutions to the fokker - planck equation are the time dependent probability densities associated with the ( it ) stochastic langevin process , @xmath11 where @xmath12 is a gaussian white noise term with unit variance .
the langevin equation is ubiquitous in a wide range of applications , from its beginnings in brownian motion ( see chap.1 of @xcite ) , to finance @xcite , biological processes @xcite and the synchronisation of networked oscillators @xcite . for @xmath13 ,
the corresponding continuity equation is sinked , thus we expect the quantity being measured in eq.([begin ] ) to seep away with time .
it is conceptually important to develop analytically tractable solutions to eq.([begin ] ) as they are green s functions , which are both inherently mathematically interesting , and highly applicable - for an account of their application in physics see chap.7 of @xcite .
following @xcite , greens / density functions appearing in this work also figure heavily in random matrix theory .
we shall indicate some of these connections throughout this work . additionally , many past results for the conserved fokker - planck equation ( we offer @xcite as a typical example ) rely simply on the steady state density to gain insights .
sinked densities allow no such avenue for inquiry as their solutions decay with time .
we shall highlight this behaviour in the proceeding sections .
the general strategy for solving eq.([begin ] ) is as follows : we first obtain the _ weight function _
@xmath14 , found by solving the corresponding pearson s equation , @xmath15 for constants @xmath16 and @xmath17
. given the form of the diffusion , drift and sink coefficients , we categorise the spectrum of eq.(1 ) as either discrete or mixed discrete / continuous .
this gives us the general form of @xmath0 as , @xmath18 where the sum will be an integral for the continuous parts of the spectrum .
we then find the corresponding eigenvalues and eigenfunctions of the system through standard techniques @xcite .
the final step involves solving for the normalisation constants which satisfy the initial condition .
for the discrete spectrum eigenfunctions we utilise the orthogonal polynomial relation ( chap.3 of @xcite ) , @xmath19 and for the continuous spectrum eigenfunctions we employ a macrobert s inverse integral transform of the form , @xmath20 see @xcite for an instance involving the whittaker functions , and @xcite and chap.14 of @xcite for examples involving bessel / hankel functions .
these inverse integral transforms usually rely on some key results attributable to macrobert @xcite which we shall exploit when deriving the continuous normalisation for the romanovski case . given the pervasive nature of the fokker - planck equation , green s functions and orthogonal polynomials ,
most of the cases presented in this work have been fully solved in the literature _ without _ the sink term .
what is new in this work is we present the full time dependent solutions for the sinked variants ( @xmath13 ) , and apply these results to obtain new solutions to the stochastic bertalanffy - richards ( b - r ) equation @xcite . for an introduction to the application of the b - r equation in population modelling and biological processes see @xcite . in the next section
we detail the forms of the diffusion , drift and sink coefficients that lead to the orthogonal polynomial eigenfunctions considered in this work . in sec.[1dsl ] we give necessary information about sturm - liouville ( s - l ) operators , the hilbert spaces their eigenfunctions span and how the sink terms in the s - l operators form associated variations of the orthogonal polynomials / eigenfunctions . in sec.[classification ] we detail how the form of the s - l operator determines the exact form of the spectra for the eigenfunctions , along with the corresponding solutions to eq.([begin ] ) . in sec.[applici ] we apply the results of this work to the stochastic b - r equation .
finally we offer implications of these results and flag future work .
applying the weight function , we decompose @xmath0 in eq.([begin ] ) as , @xmath21 hence the ensuing equation for @xmath22 is @xmath23 where @xmath24 is the s - l operator @xmath25 we require that the operator @xmath24 be negative , i.e. all relevant eigenfunctions of @xmath24 have negative eigenvalues , @xmath26 the eigenvalues @xmath27 may be discrete or continuous depending on boundary conditions @xcite , to be specified explicitly in sec.[classification ] . for this work @xmath8 and @xmath9 in eq.([s - lspectrum ] )
have the specific forms : @xmath28 focusing on the @xmath29 case , theorem 4 of @xcite states that there are six classes of discrete spectrum eigenfunctions to eq.([s - lspectrum ] ) : hermite , laguerre , jacobi , bessel , fisher - snedecor ( shifted - jacobi ) , and romanovski ( pseudo - jacobi ) polynomials - the so - called _ continuous classical orthogonal polynomials_. for @xmath30 the eigenfunctions generally have the same polynomial form , but are multiplied by the diffusion coefficient @xmath31 raised to some power , @xmath32 , specified in sec.([1dsl ] ) of this work .
we do not formally consider the hermite and jacobi case in this work ( we only refer to them ) as the hermite case offers nothing new , and the jacobi case only has finite support in @xmath2 . in @xcite , hypergeometric polynomial solutions to eq.([s - lspectrum ] ) for quite general forms of the diffusion and drift coefficients were studied .
using the liouville transformation ( eq.([changevars ] ) of this work ) to transform eq.([s - lspectrum ] ) into a corresponding schrdinger equation , we note that the laguerre , bessel and romanovski potentials correspond to the _ coulomb _ ( chap.6 of @xcite ) , _ morse _ @xcite and _ trigonometric scarf _ @xcite potentials , albeit with an additional parameter corresponding to the sink coefficient .
more recently the laguerre polynomials were applied as solutions to the non - linear madelung fluid equation @xcite and a burger s equation with a time - dependent forcing term @xcite .
both @xcite and @xcite highlight the connection between the laguerre polynomials and the algebra @xmath33 , which is then exploited to construct the coherent laguerre function , and explore squeezed states in the calogero - sutherland model .
@xcite also highlights connections between lie algebras and the associated laguerre functions .
additionally , the non - sinked variant of eq.([solll ] ) for the laguerre case features in the financial cox - ingersoll - ross ( cir ) model @xcite , amongst other applications . with regards to the bessel polynomials , in @xcite the ladder operators of the associated bessel functions were explored .
the non - sinked variant of eq.([solll ] ) for the bessel case was derived as the fokker - planck equation of an ergodic diffusion with reciprocal gamma invariant distribution in @xcite , and features in many financial models ( see sec.6.5 of @xcite ) .
following chap.4 of @xcite , the fisher - snedecor polynomials are a variant of the jacobi polynomials under a simple linear transformation such that the corresponding weight function s support is extended to the positive real line . in @xcite the non - sinked variant of eq.([solll ] ) for the fisher - snedecor case was derived as the fokker - planck equation for an ergodic diffusion with fisher - snedecor invariant distribution .
refer to @xcite for the corresponding jacobi expression of eq.([solll ] ) which has finite support in @xmath2 and an entirely discrete spectrum .
the romanovski polynomials have received a fair amount of attention lately due to their application in supersymmetric quantum mechanics @xcite , quantum chromodynamics @xcite , and connections with yang - mills integrals @xcite . the non - sinked variant of eq.([solll ] ) for the romanovski case ( without a closed form expression for the continuous spectrum normalisation )
was first given in @xcite as the fokker - planck equation for an ergodic diffusion with the symmetric scaled student invariant distribution .
multidimensional generalisations of the classical polynomials of course exist ( see @xcite , amongst other works ) and are a current active field of study .
of particular relevance to this work we see in chapters 2 and 3 of @xcite that the probability density functions of the eigenvalues of the _ chiral _ , _ laguerre _ , _ jacobi _ and _ cauchy _ ensembles of random matrices give the multidimensional generalisations of the hermite , laguerre , jacobi and romanovski weights respectively .
additionally , chap.11 of the aforementioned work considers potentials which correspond to various classes of quantum calogero - sutherland models . in particular
we see that propositions 11.3.1 and 11.3.2 give multidimensional generalisations of the corresponding schrdinger hermite , laguerre and jacobi potentials ( amongst other more general cases ) , with the ( restricted ) green s functions of these three cases constructed in chap.11.6 .
due to @xmath24 in eq.([s - lspectrum ] ) being a non - positive , self - adjoint s - l operator , the full set of solutions to eq.([s - lspectrum ] ) - present in eq.([solll ] ) - necessarily form a ( weighted ) square - integrable hilbert space @xmath34 with respect to the weighted inner product @xcite , @xmath35 the emergence of the continuous spectrum in eq.([solll ] ) for certain classes of eigenfunctions is due to the discrete spectrum eigenfunctions possessing so - called _ finite orthogonality _ ( see chapters 3 and 4 of @xcite ) : only a finite subset of the bessel , fisher - snedecor and romanovski polynomials obey eq.([finite1 ] ) .
thus the continuous spectrum eigenfunctions are required to construct the hilbert space .
following chap.22 of @xcite and chapters 7 and 8 of @xcite , if the spectrum of @xmath24 is mixed , the corresponding hilbert space is separable into the following orthogonal subspaces , @xmath36 where @xmath37 denotes the subspace of the hilbert space containing _ pure point _
( discrete ) spectrum , and @xmath38 denotes the subspace of the hilbert space containing _ absolutely continuous _ spectrum . recently in @xcite the associated variants of the laguerre , bessel and romanovski polynomials , respectively , were considered ( the associated fisher - snedecor functions are a simple variation on the romanovski case ) .
the new results in this paper involve applying the aforementioned results and constructing the associated sinked densities .
we list the canonical forms of the diffusion , drift and sink coefficients of the four relevant cases in tab.[tab1 ] , and give the weight functions and the corresponding support of @xmath2 for each case in tab.[tab2 ] .
.canonical forms of @xmath31 , @xmath39 and @xmath9 considered in this work .
[ cols="^,^,^,^ " , ] [ tab7 ] hence , the expression for the density of the bessel case is , @xmath40 and the corresponding expression for the fisher - snedecor case is , @xmath41 if @xmath24 exhibits oscillatory behaviour at both boundaries for @xmath42 , and @xmath43 , then eq.([solll ] ) is given by , @xmath44 where ( following sec.5.3 of @xcite ) the eigenfunctions with the continuous eigenvalues , @xmath45 , @xmath46 , are the linearly independent solutions to eq.([s - lspectrum ] ) which are square - integrable with @xmath14 , and are valid in the neighbourhood of the natural boundaries ( in this case , @xmath47 ) for which @xmath24 exhibits oscillatory behaviour .
similar to spectral category ii , we note that @xcite , @xmath48 the discrete normalisation constants in eq.([genexp3 ] ) are given by eq.([ortho1 ] ) , and in [ romnormapp ] we explicitly derive the continuous normalisations @xmath49 , using the aforementioned macrobert s style proof . from the forms of @xmath50 in tab.[tab4 ] and the corresponding support of @xmath2 , we see that the romanovski case fall under this particular mixed spectral category , and its highest discrete eigenvalue satisfies , @xmath51 concerning the discrete spectrum eigenfunctions , the associated romanovski functions @xmath52 have the following hypergeometric form , @xmath53 following chap.9.9 of @xcite , the eigenvalues and normalisation constants for this case are given by , @xmath54 under the restrictions , @xmath55 rescaling the continuous spectrum eigenfunctions , @xmath56 ( where @xmath57 is the complex conjugate of @xmath58 ) , the eigenvalues are parameterised by , @xmath59 and hypergeometric forms of @xmath58 are ( see @xcite and chap.15 of @xcite ) , @xmath60 where , @xmath61 the continuous orthogonality relations are given by , @xmath62 where @xmath63 and , @xmath64,\nonumber\\ \lambda_{1,2}(\mu)=\lambda_{2,1}(\mu ) = 2^{2\gamma+2 } \pi \left\ { \left| \tilde{\gamma}(\mu ) \right|^2 \cosh\left [ \frac{\pi}{2 } ( \sigma_2 + 2 \mu ) \right ] \right.\nonumber\\ \left
. + \left| \tilde{\gamma}(-\mu ) \right|^2 \cosh\left [ \frac{\pi}{2 } ( \sigma_2 - 2 \mu ) \right]\right\}.\label{newresult}\end{aligned}\ ] ] thus the continuous normalisations are given by , @xmath65 we provide the detailed derivation of eq.([newresult ] ) in [ romnormapp ] using the aforementioned macrobert s method .
hence the complete density function for the romanovski case is , @xmath66 we give an example of eq.([romres ] ) in fig.[fig2 ] .
notice that as time increases the total area of the density ( which begins at unity ) decreases .
for @xmath67 we notice that the density is barely distinguishable visually .
we compare this to the stationary density of the non - sinked case - the left most density - where area is conserved for all @xmath68 . ) at various times with parameter values @xmath69 , @xmath70 and @xmath71.,width=491,height=188 ]
we now present an application of this work - time dependent distributions corresponding to various instances of the b - r langevin equation .
the b - r equation is a deterministic system given by , @xmath72 where @xmath73 .
we note that when @xmath74 , the @xmath75 and @xmath76 terms act as growth and decay terms respectively ; the greater the value of @xmath77 , the more pronounced the decay .
the choice @xmath78 gives the famous logistic equation .
to proceed we consider the two uncorrelated noise terms @xmath79 and @xmath80 with variance @xmath81 , @xmath82 where @xmath83 means an ensemble average over the noise .
we perturb the growth and decay coefficients by @xmath79 and @xmath80 respectively to obtain the following ( it ) stochastic langevin equation , @xmath84 where @xmath85 .
eq.([ito ] ) can be _ solved exactly _ through the transformation @xmath86 leading to the linear langevin equation , @xmath87,\end{aligned}\ ] ] and the formal solution , @xmath88 we refer to the above solution for @xmath89 as formal as it contains integrals of specific instances of the noise terms ( meaning that each solution will be different for different noise instances ) . in order to make general statements about the above system
, we shall construct its probability density function .
following chap.4.5 of @xcite , the stochastic process in eq.([ito ] ) obeys the following fokker - planck equation , @xmath90 as mentioned in earlier sections , since the above equation contains no sink term @xmath0 is conserved , and its most natural interpretation is density of probability , where @xmath2 and @xmath3 are the population of a species .
our goal of this section is to analytically solve for various cases of eq.([qands ] ) using our polynomial solutions for density functions given in sec.[classification ] .
applying the standard decomposition in eq.([decomp ] ) , and the nonlinear transformation in eq.([itotrans ] ) , the resulting expression for @xmath91 is , @xmath92 \xi + b + \frac{\omega \zeta \beta^2}{2 \xi } \right)\frac{\partial}{\partial \xi } \right\}g(\xi , t|\xi').\label{huen}\end{aligned}\ ] ] since the b - r equation is used extensively in population modelling , where the variable @xmath2 represents the number of living members of a species , only eigenfunctions in the range @xmath93 will be considered , hence leaving out the romanovski example .
this leaves three relevant cases , laguerre , bessel and fisher - snedecor .
the s - l operator on the right hand side of eq.([huen ] ) leads to the heun differential equation ( see chap.31 of @xcite ) . due to the heun equation possessing four distinct singular points , there is no equivalent hypergeometric closed form expression for the heun functions @xcite .
nevertheless , we find the following mapping between the heun system and hypergeometric solutions : * @xmath94 leads to the laguerre case * @xmath95 leads to the bessel case * @xmath96 leads to the fisher - snedecor case we shall only detail the laguerre and fisher - snedecor cases in this work as the bessel case was first solved in @xcite and along with @xcite is one of the earlier results involving analytical expressions of densities with mixed spectra .
the case @xmath97 in eq.([huen ] ) leads to the biconfluent heun equation , whose solution suffers the same non - closed properties as the heun equation ( see chap.31 of @xcite ) .
additionally , the case @xmath98 leads to the _ bessel process with constant drift _
@xcite , which is a peculiar hypergeometric case ( beyond the scope of this work ) where the spectrum is mixed but the discrete part contains an _ infinite _ number of eigenvalues .
setting @xmath94 , the weight function @xmath99 for this case is , @xmath100 applying the following change in variables , @xmath101 eq.([huen ] ) becomes , @xmath102 where eq.([brlag ] ) is the standard laguerre fokker - planck equation .
hence , due to the initial condition , the time dependent solution for the density in this section is , @xmath103 to the best of our knowledge , eq.([brlag2 ] ) is a new result of a specific example of a b - r fokker - planck equation . making the connection with the langevin equation this density
is generated from , @xmath104 since for @xmath105 the deterministic system is divergent , but the density is normalisable , this particular case is an example of multiplicative noise stabilising the system @xcite . setting @xmath96 the weight function @xmath99 for this case is , @xmath106 applying the following change in variables , @xmath107 eq.([huen ] ) becomes , @xmath108 where eq.([brfish ] ) is the standard fisher - snedecor fokker - planck equation .
hence the time dependent solution for the density is , @xmath109 as with the laguerre case , to the best of our knowledge , eq.([brfs ] ) is a new result of a specific instance of a b - r fokker - planck equation . in fig.[fig5 ] , we give a specific example of eq.([brfs ] ) at various times .
it is elementary to show that the weight function , which is proportional to the steady state density , peaks at the value @xmath110 . making the connection with the langevin equation this density
is generated from , @xmath111 we see that this example models linear deterministic growth , with linear and quadratic multiplicative stochastic terms .
again , like the laguerre case , since the deterministic system is divergent , but the density is normalisable , this system provides another example of multiplicative noise stabilising a system @xcite . ) at various times with parameter values @xmath112 , @xmath113 , @xmath114 , @xmath115 and @xmath116.,width=491,height=188 ]
in this work we have given closed form expressions of sinked densities associated with ( at most ) quadratic diffusion and linear drift .
the eigenfunctions relating to the discrete part of the spectrum are associated variants of classical orthogonal polynomials . we have given a macrobert s style proof to obtain a new closed form expression for the continuous spectrum normalisation associated with the romanovski density .
this technique is sufficiently generalisable , given one knows enough about the analytic continuation properties of the hypergeometric function under consideration .
we then applied these results to obtain the time dependent fokker - planck solutions associated with various cases of the b - r langevin equation .
given the pervasive nature of langevin equations ( and the densities and green s functions associated with them ) in the physical sciences , we anticipate that these results are a stepping stone to a richer understanding of a variety of processes , both conserved and non - conserved .
specifically , we hope that processes involving mixed spectra eigenfunctions become increasingly commonplace , as more analytic examples of solution appear which increase our mathematical understanding and our ability to apply such results in novel ways . paraphrasing the relevant introduction of @xcite ; in a world of ever increasing computing power , we must never overlook the benefits provided from analytic solutions in terms of special functions .
they provide insight for understanding non - trivial relationships among physical variables with unsurpassed economy of effort , and are an invaluable tool for the validation of more complicated models which require computational treatment .
the author graciously acknowledges alexander kalloniatis for his fruitful discussions and constant encouragement .
we begin by conveniently labelling the double integral of eq.([newresult ] ) by , @xmath117 focusing on the case @xmath118 , we apply eq.([chi12 ] ) to split up @xmath119 for the region @xmath120 and deform the @xmath121 integral onto the contours @xmath122 and @xmath123 as shown in fig.[figcont ] to obtain , @xmath124 contours @xmath122 and @xmath123 used for the macrobert s proof.,width=377,height=188 ] following chap.14 of @xcite and chap.7 of @xcite , we may reverse the order of integration as each term in eq.([int11 ] ) falls off like @xmath125 , @xmath126 , on the respective contours @xmath122 and @xmath123 , as @xmath127 .
hence @xmath128 becomes , @xmath129 to proceed we note that ( chap.15.5 of @xcite ) @xmath130 , and their complex conjugates , obey the same governing s - l equation as @xmath131 and @xmath132 , as they are the corresponding linearly independent solutions in the neighbourhood of the singular point @xmath133 . thus decomposing
either of the aforementioned eigenfunctions as , @xmath134 the resulting governing equation for the @xmath135 s is , @xmath136 we now recast eq.([niceii ] ) in terms of the @xmath135 s . through considering two copies of eq.([definej ] ) , one for @xmath137 and one for @xmath138 , we multiply the equation for @xmath137 by @xmath138 , and vice versa . subtracting the two expressions
awards us with , @xmath139 integrating eq.([unamed ] ) over all of @xmath2 , and applying integration by parts we obtain , @xmath140^{x \rightarrow \infty}_{x \rightarrow- \infty}. \end{aligned}\ ] ] using eq.([chi12 ] ) the asymptotic forms of the desired limits are given by , @xmath141 where , @xmath142 using the above asymptotic forms , eq.([niceii ] ) becomes , @xmath143 where @xmath144 and , @xmath145 in the following we consider the _ dirichlet integral _ expressions from chap.1 of @xcite and chap.3 of @xcite : @xmath146 where @xmath147 and the analytic function @xmath148 obeys _
dirichlet s conditions _ on the interval @xmath149 on the interval @xmath149 entail : ( i ) @xmath148 contains only a finite number of discontinuities on the interval , ( ii ) @xmath148 contains a finite number of turning points on the interval . ] .
thus we deform the contours @xmath122 and @xmath123 back to the real line segment @xmath150 , and let @xmath151 and @xmath152 . assuming that the function @xmath153 obeys dirichlet s conditions
, we immediately obtain the following expression for @xmath128 , @xmath154 which is the required form given in eq.([newresult ] ) .
the expression for @xmath155 is simply the complex conjugate of the case just considered .
the remaining cases can be verified in an equivalent method considered in this appendix .
10 m abramowitz and i stegun ( editors ) , _ handbook of mathematical functions with formulas , graphs , and mathematical tables _ , in : applied mathematical series , vol .
government printing office , washington d.c .
, ( 1972 ) f avram , n leonenko and n uvak , _ spectral representation of transition density of fisher - snedecor diffusion _ , stochastics - an international journal of probability and stochastic processes , 85(2 ) , ( 2013 ) , 346 - 369 l von - bertalanffy , _ problems of organic growth _ , nature , 163 , ( 1949 ) , 156 - 158 a borodin and p salminen , _ handbook of brownian motion _ , birkhauser , boston , ( 1996 ) m brics , j kaupus and r mahnke , _ how to solve the fokker - planck equation treating mixed eigenvalue spectrum ? _ , condensed matter physics , 16(1 ) , 13002 , ( 2013 ) , 1 - 13 p broadbridge , _ the forced burgers equation , plant roots and schrdinger s eigenfunctions _ , journal of engineering mathematics , 36 , ( 1999 ) , 25 - 39 s bykaik and o pashaev , _ exactly solvable madelung fluid and complex burgers equations : a quantum sturm - liouville connection _
, journal of mathematical chemistry , 50 , ( 2012 ) , 2716 - 2745 e celeghini and m del almo , _ coherent orthogonal polynomials _ , annals of physics , 335 , ( 2013 ) , 78 - 85 m compean and m kirchbach , _ trigonometric quark confinement potential of qcd traits _ , european physical journal a , 33 , ( 2007 ) , 1 - 4 j cox , j ingersoll and s ross , _ a theory of the term structure of interest rates _ , econometrica , 53 , ( 1985 ) , 385 - 407 b davies , _ integral transforms and their applications _ , ( 3rd ed . ) , in : texts in applied mathematics , vol .
41 , springer - verlag , new york , ( 2002 ) d davydov and v linetsky , _ pricing options on scalar diffusions : an eigenfunction expansion approach _ , operations research , 51 , ( 2003 ) , 185 - 209 n dunford and j schwartz , _ linear operators .
part ii : spectral theory , self - adjoint operators in hilbert space _ , wiley , new jersy , ( 1988 ) h fakhri and a chenaghlou , _ ladder operators for the associated laguerre functions _
, journal of physics a , 37 , ( 2004 ) , 7499 - 7507 h fakhri and a chenaghlou , _ ladder operators and recursion relations for the associated bessel polynomials _ , physics letters a , 358 , ( 2006 ) , 345 - 353 h fakhri and b mojaveri , _ the remarkable properties of the associated romanovski functions _ , journal of physics a , 44 , ( 2011 ) 195205 p forrester , _ log gases and random matrices _ , in : the london mathematical society monograph series , vol .
34 , princeton university press , princeton nj , ( 2010 ) h fu and r sasaki , _ exponential and laguerre squeezed states for @xmath156 algebra and the calogero - sutherland model _ , physical review a , a53 , ( 1996 ) , 3836 - 3844 o garca , _ a stochastic differential equation model for the height growth of forest stands _ , biometrics , 39(4 ) , ( 1983 ) , 1059 - 1072 u graf , _ introduction to hyperfunctions and their integral transforms : an applied and computational approach _ , springer science and business media , new york , ( 2010 ) e hille and r phillips , _ functional analysis and semigroups _ , in : american mathematical society colloquium publications , vol .
31 , american mathematical society , rhode island , ( 1957 ) m hortasu , _ heun functions and their uses in physics _ , proceedings of the 13th regional conference on mathematical physics , antalya , turkey , october 27 - 31 , 2010 , edited by u camci and i semiz , world scientific , ( 2013 ) , 23 - 39 a inayat - hussain , _ a new generalization of the hankel integral transform _ , journal of mathematical physics , 30(1 ) , ( 1989 ) , 41 - 44 r koekoek , p leskey and r swarttoue , _ hypergeometric orthogonal polynomials and their q - analogues _ , springer - verlag , berlin , ( 2010 ) t koornwinder , _ two variable analogues of the classical orthogonal polynomials _
, in : theory and application of special functions , edited by r askey , academic press , ( 1975 ) , 435495 h langer and w schenki , _ generalised second - order differential equations , corresponding gap diffusions and susuperharmonic transformations _ , mathematische nachrichten , 148 , ( 1990 ) , 7 - 45 n leonenko and n uvak , _ statistical inference for reciporical gamma diffusion process _ , journal of statistical planning and inference , 140 , ( 2010 ) , 30 - 51 n leonenko and n uvak , _ statistical inference for student diffusion process _ , stochastic analysis and applications , 28(6 ) , ( 2010 ) , 972 - 1002 v linetsky , _ the spectral decomposition of the option value _ , international journal of theoretical and applied finance , 7(3 ) , ( 2004 ) , 337 - 384 v linetsky , _ the spectral representation of bessel processes with constant drift : applications in queueing and finance _ , journal of applied probability , 41 , ( 2004 ) , 327 - 344 t macrobert , _ spherical harmonics : an elementary treatise on harmonic functions , with applications _ , ( 2nd ed . ) , dover publications , new york , ( 1947 ) x mao , g marion and e renshaw , _ environmental brownian noise suppresses explosions in population dynamics _ , stochastic processes and their applications , 97(1 ) , ( 2002 ) , 95 - 110 h mckean , _ elementary solutions for certain parabolic partial differential equations _
, transactions of the american mathematical society , 82 , ( 1956 ) , 519 - 548 e merzbacher , _ quantum mechanics _ , ( 3rd ed . ) , wiley , new jersey , ( 1997 ) p morse , _ diatomic molecules according to the wave mechanics .
vibrational levels _ , physical review , 34 , ( 1929 ) , 57 - 64 p morse and h feshach , _ methods of theoretical physics _ , in : international series in pure and applied physics , mcgraw hill , new york , ( 1953 ) f olver , d lozier , r boisvert , and c clark ( editors ) , _ nist handbook of mathematical functions _
, cambridge university press , new york , ( 2010 ) c quesne , _ extending romanovski polynomials in quantum mechanics _ , journal of mathematical physics , 54 , ( 2013 ) , 102103 a raposo , h weber , d alvarez - castillo and m kirchbach , _ romanovski polynomials in selected physics problems _ , central european journal of physics , 5(3 ) , ( 2007 ) , 253 - 284 m reed and b simon , _ functional analysis _ , in : methods of modern mathematical physics , vol .
1 , academic press , san diego , ( 1981 ) f richards , _ a flexible growth function for empirical use _ , journal of experimental botany , 10 , ( 1959 ) , 290 - 300 h risken , _ the fokker - planck equation _ , ( 2nd ed . ) , springer , heidelberg , ( 1989 ) n saad , r hall and h ciftci , _ criterion for polynomial solutions to a class of linear differential equations of second order _ , journal of physics a , 39 , ( 2006 ) , 13445 - 13454 f scarf , _ new soluble energy band problem _ , physical review , 112 , ( 1958 ) , 1137 - 1141 a schenzle and h brand , _ multiplicative stochastic processes in statistical physics _ , physics letters , 69a(5 ) , ( 1979 ) , 313 - 315 z schuss , _ theory and applications of stochastic processes _ , in : series in applied mathematical sciences , vol . 170 , springer , new york , ( 2010 ) i stewart and d tall , _ complex analysis _ , cambridge university press , cambridge , ( 1999 )
s strogatz , _ from kuramoto to crawford : exploring the onset of synchronization in populations of coupled oscillators _ , physica d , 143 , ( 2000 ) , 1 - 20 m tierz , _
@xmath157 matrix model and supersymmetric yang - mills integrals _ , physical review d , 76(10 ) , ( 2007 ) , 107701 j weidmann , _ spectral theory of ordinary linear operators _ , in : lecture notes in mathematics , vol .
1258 , springer , berlin , ( 1987 ) j wimp , _ a class of integral transforms _
edinburgh math .
14 , ( 1964 ) , 33 - 40 e wong , _ the construction of a class of stationary markoff processes _ , in : stochastic processes in mathematical physics and engineering , edited by r bellman , ( proceedings of symposia of applied mathematics , vol .
26 ) , american mathematical society , providence , r.i . , ( 1964 ) , 264 - 276 m yadav , _ solutions of a system of forced burgers equation in terms of generalized laguerre polynomials _ , acta mathematica scientia , 34b(5 ) , ( 2014 ) , 1461 - 1472 m zuparic and a kalloniatis , _ stochastic ( in)stability of synchronisation of oscillators on networks _ , physica d , 255 , ( 2013 ) , 35 - 51 | we analytically solve for the time dependent solutions of various density evolution models . with specific forms of the diffusion , drift and sink coefficients , the eigenfunctions can be expressed in terms of hypergeometric functions .
we obtain the relevant discrete and continuous spectra for the eigenfunctions . with
non - zero sink terms the discrete spectra eigenfunctions are generalisations of well known orthogonal polynomials : the so - called _ associated - laguerre , bessel , fisher - snedecor _ and _ romanovski functions_. we use a macrobert s proof to obtain closed form expressions for the continuous normalisation of the romanovski density function .
finally , we apply our results to obtain the analytical solutions associated with the bertalanffy - richards langevin equation .
_ keywords _ : fokker - planck equation , density function , hypergeometric function , classical orthogonal polynomial , bertalanffy - richards langevin equation |
in the mssm ( `` minimal supersymmetric extension of the standard model '' ) gauge couplings unify at an energy scale of about @xmath1 gev . adding particles arbitrarily to
the mssm easily destroys this attractive feature .
thus , relatively few susy models have been discussed in the literature which have a larger than mssm particle content at experimentally accessible energies .
neutrino oscillation experiments @xcite , however , have shown that at least one neutrino must have a mass @xmath2 ev .
a ( majorana ) neutrino mass of this order indicates the existence of a new energy scale below @xmath3 . for models with renormalizable interactions and perturbative couplings , as for example in the classical seesaw models
@xcite , this new scale should lie below approximately @xmath4 gev .
from the theoretical point of view gut models based on the group @xmath5 @xcite offer a number of advantages compared to the simpler models based on @xmath6 .
for example , several of the chains through which @xmath5 can be broken to the sm gauge group contain the left - right symmetric group @xmath7 as an intermediate step @xcite , thus potentially explaining the observed left - handedness of the weak interactions .
however , probably the most interesting aspect of @xmath5 is that it automatically contains the necessary ingredients to generate a seesaw mechanism @xcite : ( i ) the right - handed neutrino is included in the @xmath8 which forms a fermion family ; and ( ii ) @xmath9 is one of the generators of @xmath5 .
left - right ( lr ) symmetric models usually break the lr symmetry at a rather large energy scale , @xmath10 .
for example , if lr is broken in the susy lr model by the vev of @xmath11 triplets @xcite or by a combination of @xmath11 and @xmath12 triplets @xcite , @xmath13 gev is the typical scale consistent with gauge coupling unification ( gcu ) .
the authors of @xcite find a lower limit of @xmath14 gev from gcu for models where the lr symmetry is broken by triplets , even if one allows additional non - renormalizable operators or sizeable gut - scale thresholds to be present . on the other hand , in models with an extended gauge group it is possible to formulate sets of conditions on the @xmath15-coefficients for the gauge couplings , which allow to enforce gcu independent of the energy scale at which the extended gauge group is broken .
this was called the `` sliding mechanism '' in @xcite .
however , @xcite was not the first to present examples of `` sliding scale '' models in the literature .
in @xcite it was shown that , if the left - right group is broken to @xmath16 by the vacuum expectation value of a scalar field @xmath17 then the resulting @xmath18 can be broken to @xmath19 of the sm in agreement with experimental data at any energy scale . in @xcite
the authors demonstrated that in fact a complete lr group can be lowered to the tev - scale , if certain carefully chosen fields are added and the lr - symmetry is broken by right doublets .
a particularly simple model of this kind was discussed in @xcite .
finally , the authors of @xcite discussed also an alternative way of constructing a sliding lr scale by relating it to an intermediate pati - salam stage .
we note in passing that these papers are not in contradiction with the earlier work @xcite , which all have to have large @xmath10 .
as discussed briefly in the next section it is not possible to construct a sliding scale variant for an lr model including pairs of @xmath20 and @xmath21 .
three different constructions , based on different @xmath5 breaking chains , were considered in @xcite . in chain - i
so(10 ) is broken in exactly one intermediate ( lr symmetric ) step to the standard model group : @xmath22 in chain - ii so(10 ) is broken first to the pati - salam group : @xcite @xmath23 and finally , in chain - iii : @xmath24 in all cases the last symmetry breaking scale before reaching the sm group can be as low as @xmath25 tev maintaining nevertheless gcu .
the papers discussed above @xcite give at most one or two example models for each chain , i.e. they present a `` proof of principle '' that models with the stipulated conditions indeed can be constructed in agreement with experimental constraints .
it is then perhaps natural to ask : how unique are the models discussed in these papers ?
the answer we find for this question is , perhaps unsurprisingly , that a huge number of variants exist in each class . even in the simplest class ( chain - i ) we have found a total of 53 variants ( up to 5324 `` configurations '' , see next section ) which can have perturbative gcu and a lr scale below 10 tev , consistent with experimental data .
for the two other classes , chain - ii and chain - iii , we have found literally thousands of variants .
with such a huge number of variants of essentially `` equivalent '' constructions one immediate concern is , whether there is any way of distinguishing among all of these constructions experimentally .
tests could be either direct or indirect .
direct tests are possible , because of the sliding scale feature of the classes of models we discuss , see section [ sec : models ] .
different variants predict different additional ( s)particles , some of which ( being colored ) could give rise to spectacular resonances at the lhc .
however , even if the new gauge symmetry and all additional fields are outside the reach of the lhc , all variants have different @xmath15 coefficients and thus different running of mssm parameters , both gauge couplings and susy soft masses .
thus , if one assumes the validity of a certain susy breaking scheme , such as for example msugra , indirect traces of the different variants remain in the susy spectrum , potentially measurable at the lhc and a future ilc / clic .
this was discussed earlier in the context of indirect tests for the susy seesaw mechanism in @xcite and for extended gauge models in @xcite .
we generalize the discussion of @xcite and show how the `` invariants '' , i.e. certain combinations of susy soft breaking parameters , can themselves be organized into a few classes , which in principle allow to distinguish class - ii models from class - i or class - iii and , if sufficient precision could be reached experimentally , even select specific variants within a class and give indirect information about the new energy scale(s ) .
the rest of this paper is organized as follows : in the next section we first lay out the general conditions for the construction of the models we are interested in , before discussing variants and example configurations for all of the three classes we consider .
section [ sect : invariants ] then discusses `` invariants '' , i.e. susy soft parameters in the different model classes .
we then close with a short summary and discussion .
several technical aspects of our work are presented in the appendix .
before entering into the details of the different model classes , we will first list some general requirements which we use in all constructions . these requirements are the basic conditions any model has to fulfill to guarantee at least in principle that a phenomenologically realistic model will result .
we use the following conditions : * perturbative so(10 ) unification .
that is , gauge couplings unify ( at least ) as well as in the mssm and the value of @xmath26 is in the perturbative regime . *
the gut scale should lie above ( roughly ) @xmath27 gev .
this bound is motivated by the limit on the proton decay half - live . *
sliding mechanism .
this requirement is a set of conditions ( different conditions for different classes of models ) on the allowed @xmath15 coefficients of the gauge couplings , which ensure the additional gauge group structure can be broken at any energy scale consistent with gcu . * renormalizable symmetry breaking .
this implies that at each intermediate step we assume there are ( at least ) the minimal number of higgs fields , which the corresponding symmetry breaking scheme requires .
* fermion masses and in particular neutrino masses .
this condition implies that the field content of the extended gauge groups is rich enough to fit experimental data , although we will not attempt detailed fits of all data .
in particular , we require the fields to generate majorana neutrino masses through seesaw , either ordinary seesaw or inverse / linear seesaw , to be present .
* anomaly cancellation .
we accept as valid `` models '' only field configurations which are anomaly free .
* @xmath5 completable .
all fields used in a lower energy stage must be parts of a multiplet present at the next higher symmetry stage .
in particular , all fields should come from the decomposition of one of the @xmath5 multiplets we consider ( multiplets up to @xmath28 ) . *
correct mssm limit .
all models must be rich enough in particle content that at low energies the mssm can emerge .
a few more words on our naming convention and notations might be necessary .
we consider the three different @xmath5 breaking chains , eq .
( [ eq : chaini])-([eq : chainiii ] ) , and will call these model `` classes '' . in each class
there are fixed sets of @xmath15-coefficients , which all lead to gcu but with different values of @xmath26 and different values of @xmath29 and @xmath30 at low energies .
these different sets are called `` variants '' in the following . and finally , ( nearly )
all of the variants can be created by more than one possible set of superfields .
we will call such a set of superfields a `` configuration '' .
configurations are what usually is called `` model '' by model builders , although we prefer to think of these as `` proto - models '' , i.e. constructions fulfilling all our basic requirements .
these are only proto - models ( and not full - fledged models ) , since we do not check for each configuration in a detailed calculation that all the fields required in that configuration can remain light .
we believe that for many , but probably not all , of the configurations one can find conditions for the required field combinations being `` light '' , following similar conditions as discussed in the prototype class - i model of @xcite .
all superfields are named as @xmath31 ( in the left - right symmetric stage ) , @xmath32 ( in the pati - salam regime ) and @xmath33 ( in the @xmath18 regime ) , with the indices giving the transformation properties under the group .
a conjugate of a field is denoted by , for example , @xmath34 , however , without putting a corresponding `` bar '' ( or minus sign ) in the index .
we list all fields we use , together with their transformation properties and their origin from @xmath5 multiplets , complete up to the @xmath28 of @xmath5 in the appendix .
we start our discussion with the simplest class of models with only one new intermediate scale ( lr ) : @xmath35 we do not discuss the first symmetry breaking step in detail , since it is not relevant for the following discussion and only mention that @xmath5 can be broken to the lr group either via the interplay of vevs from a @xmath36 and a @xmath37 , as done for example in @xcite , or via a @xmath36 and a @xmath38 , an approach followed in @xcite .
in the left - right symmetric stage we consider all irreducible representations , which can be constructed from @xmath5 multiplets up to dimension @xmath28 .
this allows for a total of 24 different representations ( plus conjugates ) , their transformation properties under the lr group and their @xmath5 origin are summarized in table [ tab : list_of_lr_fields ] ( and table [ tab : list_of_patisalam_fields ] ) of the appendix .
consider gauge coupling unification first .
if we take the mssm particle content as a starting point , the @xmath15-coefficients in the different regimes are given as : and @xmath39 we use the sm particle content plus one additional higgs doublet . ]
@xmath40 where we have used the canonical normalization for @xmath9 related to the physical one by @xmath41 . here
, @xmath42 stands for the contributions from additional superfields , not accounted for in the mssm . as is well known , while the mssm unifies , putting an additional lr scale below the gut scale with @xmath43 destroys unification
. nevertheless gcu can be maintained , if some simple conditions on the @xmath42 are fulfilled .
first , since in the mssm @xmath44 at roughly @xmath45 gev one has that @xmath46 in order to preserve this situation for an arbitrary lr scale ( sliding condition ) .
next , recall the matching condition @xmath47 which , by substitution of the lr scale by an arbitrary one above @xmath48 , allows us to define an artificial continuation of the hypercharge coupling constant @xmath49 into the lr stage .
the @xmath15-coefficient of this dummy coupling constant for @xmath50 is @xmath51 and it should be compared with @xmath52 ( @xmath53 ) ; the difference is @xmath54 and it must be equal to @xmath55 in order for the difference between this @xmath49 coupling and @xmath44 at the gut to be independent of the scale @xmath48 .
these are the two conditions imposed by the sliding requirement of the lr scale on the @xmath15-coefficients [ see eq .
( [ eq : variantslr ] ) ] .
note , however , that we did not require ( approximate ) unification of @xmath29 and @xmath30 with @xmath56 and @xmath57 ; it was sufficient to require that @xmath58 . in any case , we can always achieve the desired unification because the splitting between @xmath29 and @xmath30 at the @xmath48 scale is a free parameter , so it can be used to force @xmath59 at the scale where @xmath56 and @xmath57 unify , which leads to an almost perfect unification of the four couplings .
also , we require that unification is perturbative , i.e. the value of the common coupling constant at the gut scale is @xmath60 . from the experimental value of @xmath61 @xcite one
can easily calculate the maximal allowed value of @xmath55 as a function of the scale , where the lr group is broken to the sm group .
this is shown in fig .
[ fig : maxdb ] for three different values of @xmath62 .
the smallest max(@xmath55 ) is obtained for the smallest value of @xmath10 ( and the largest value of @xmath62 ) . for @xmath62 in the interval
@xmath63 $ ] one obtains max(@xmath64 ) in the range [ @xmath65 , i.e. we will study cases up to a max@xmath66 ( see , however , the discussion below ) . [ cols="^,^ " , ] here , the invariants depend on @xmath67 with a milder or stronger dependence , depending on the value of @xmath68 . for almost all the solutions with @xmath69 ,
the values @xmath70 , @xmath71 , @xmath72 are constants and only in @xmath73 a mild variation with @xmath67 is found .
this fact was already pointed out in @xcite .
however , we have found that class - iii models can be made with @xmath74 and these , in general , lead to invariants which are qualitatively similar to the case of class - i discussed above . in fig .
( [ fig : invbl ] ) we show two examples of invariants for class - iii , one with @xmath69 and one with @xmath75 .
the solutions with @xmath76 fall in two kinds : the minimum value of @xmath48 is very large .
then , the invariants have the same behavior than those in which @xmath69 .
the minimum value of @xmath48 is low .
the invariants are not constants and look similar to the ones in the class - i models .
the generally mild dependence on @xmath67 can be understood , since it enters into the soft masses only through the changes in the abelian gauge couplings .
class - iii models are therefore the hardest to `` test '' using invariants . the classification of variants that we have discussed in section [ sect : invclasses ] only takes into account what happens when the lowest intermediate scale is very low , @xmath77 .
when one varies continuously the lowest intermediate scale ( @xmath48 in the lr and ps - class models or @xmath67 in the bl - class of models ) , each variant draws a line in the 4-dimensional space @xmath78 .
the dimensionality of such a plot can be lowered if we use the ( approximate ) relations between the invariants shown above , namely @xmath79 and @xmath80 .
we can then choose two independent ones , for example @xmath73 and @xmath72 , so that the only non - trivial information between the 4 invariants is encoded in a @xmath81 plot .
in this way , it is possible to simultaneous display the predictions of different variants .
this was done in fig .
( [ fig : leqe_parametric_plot_rgb ] ) , where lr- , ps- and bl - variants are drawn together .
the plot is exhaustive in the sense that it includes all lr - variants , as well as all ps- and bl - variants which can have the highest intermediate scale below @xmath82 gev . in all cases
, we required that @xmath83 at unification is larger than @xmath84 when the lowest intermediate scale is equal to @xmath85 .
parametric @xmath81 plot for the different variants ( see text ) .
the thicker lines labeled with i , ii , iii and iv indicate the result for the four prototype models presented in @xcite . ]
there is a dot in the middle of the figure - the msugra point - which corresponds to the prediction of msugra models , in the approximation used .
it is expected that every model will draw a line with one end close to this point .
this end - point corresponds to the limit where the intermediate scales are close to the gut scale and therefore the running in the lr , ps and bl phases is small so the invariants should be similar to those in msugra models .
so the general picture is that lines tend to start ( when the lowest intermediate scale is of the order of @xmath86 gev ) outside or at the periphery of the plot , away from the msugra point and , as the intermediate scales increase , they converge towards the region of the msugra point , in the middle of the plot . in fact , note that all the blue lines of lr - class models do touch this point , because we can slide the lr scale all the way to @xmath87 .
but in ps- and bl- models there are two intermediate scales and often the lowest one can not be increased all the way up to @xmath87 , either because that would make the highest intermediate scale bigger than @xmath87 or because it would invert the natural ordering of the two intermediate scales .
it is interesting to note that the bl - class with low @xmath48 can produce the same imprint in the sparticle masses as lr - models .
this is to be expected because with @xmath48 close to @xmath67 the running in the u(1)-mixing phase is small , leading to predictions similar to lr - models .
the equivalent limit for ps - class models is reached for very high @xmath88 , close to the gut scale ( see below ) .
on the other hand , from fig .
( [ fig : leqe_parametric_plot_rgb ] ) we can see that a low @xmath88 actually leads to a very different signal on the soft sparticle masses .
for example , a measurement of @xmath89 and @xmath90 , together with compatible values for the other two invariants ( @xmath91 and @xmath92 ) would immediately exclude _ all classes of models except ps - models _ , and in addition it would strongly suggest low ps and lr scales . parametric @xmath81 plots for different ps - variants showing the effect of the ps scale . ] fig .
( [ fig : leqe_parametric_plot_ps_only ] ) illustrates the general behavior of ps - models as we increase the separation between the @xmath93 and @xmath88 scales .
the red region in the @xmath81 plot tends to rotate anti - clockwise until it reaches , for very high @xmath88 , the same region of points which is predicted by lr - models .
curiously , we also see in fig .
( [ fig : leqe_parametric_plot_ps_only ] ) that some of these models actually predict different invariant values from the ones of lr models .
what happens in these cases is that since the ps phase is very short , it is possible to have many active fields in it which decouple at lower energies .
so even though the running is short , the values of the different gauge couplings actually get very large corrections in this regime and these are uncommon in other settings .
for example , it is possible in this special subclass of ps - models for @xmath29 to get bigger than @xmath56/@xmath94 before unifying !
one can see from fig .
( [ fig : leqe_parametric_plot_ps_only ] ) that many , although not all ps - models can lead to large values of @xmath73 .
this can happen for both low and high values of @xmath88 and is a rather particular feature of the class - ii , which can not be found in the other classes .
we have discussed @xmath5 based supersymmetric models with extended gauge group near the electro - weak scale , consistent with gauge coupling unification thanks to a `` sliding scale '' mechanism . we have discussed three different setups , which we call classes of models .
the first and simplest chain we use breaks @xmath5 through a left - right symmetric stage to the sm group , class - ii uses an additional intermediate pati - salam stage , while in class - iii we discuss models which break the lr - symmetric group first into a @xmath18 group before reaching the sm group . we have shown that in each case many different variants and many configurations ( or `` proto - models '' ) for each variant can be constructed .
we have discussed that one can not only construct sliding models in which an inverse or linear seesaw is consistent with gcu , as done in earlier work @xcite , but also all other known types of seesaws can , in principle , be found .
we found example configurations for seesaw type - i , type - ii and type - iii and even inverse type - iii ( for which one example limited to class - ii was previously discussed in @xcite ) . due to the sliding scale property
the different configurations predict potentially rich phenomenology at the lhc , although by the same reasoning the discovery of any of the additional particles the models predict is of course not guaranteed .
however , even if all the new particles - including the gauge bosons of the extended gauge group - lie outside of the reach of the lhc , indirect tests of the models are possible from measurements of susy particle masses .
we have discussed certain combinations of soft parameters , called `` invariants '' , and shown that the invariants themselves can be classified into a few sets .
just determining to which set the experimental data belongs would allow to distinguish , at least in some cases , class - i from class - ii models and also in all but one case our classes of models are different from the ordinary high - scale seesaw ( type - ii and type - iii ) models .
depending on the accuracy with which supersymmetric masses can be measured in the future , the invariants could be used to gain indirect information not only on the class of model and its variant realized in nature , but also give hints on the scale of beyond - mssm physics , i.e. the energy scale at which the extended gauge group is broken .
we add a few words of caution .
first of all , our analysis is done completely at the 1-loop level .
it is known from numerical calculations for seesaw type - ii @xcite and seesaw type - iii @xcite that the invariants receive numerically important shifts at 2-loop level .
in addition , there are also uncertainties in the calculation from gut - scale thresholds and from uncertainties in the input parameters . for the latter
the most important is most likely the error on @xmath95 @xcite . with the huge number of models we have considered , taking into account all of these effects
is impractical and , thus , our numerical results should be taken as approximate .
however , should any signs of supersymmetry be found in the future , improvements in the calculations along these lines could be easily made , should it become necessary .
more important for the calculation of the invariants is , of course , the assumption that susy breaking indeed is msugra - like . tests of the validity of this assumption can be made also only indirectly .
many of the spectra we find , especially in the class - ii models , are actually quite different from standard msugra expectations and thus a pure mssm - msugra would give a bad fit to experimental data , if one of these models is realized in nature .
however , all of our variants still fulfill ( by construction ) a certain sum rule , see the discussion in section [ sect : invclasses ] . of course
, so far no signs of supersymmetry have been seen at the lhc , but with the planned increase of @xmath96 for the next run of the accelerator there is still quite a lot of parameter space to be explored .
we note in this respect that we are not overly concerned about the higgs mass , @xmath97 gev , if the new resonance found by the atlas @xcite and cms @xcite collaborations turns out to be indeed the lightest higgs boson . while for a pure mssm with msugra boundary conditions it is well - known @xcite that such a hefty higgs requires multi - tev scalars , all our models have an extended gauge symmetry .
thus , there are new d - terms contributing to the higgs mass @xcite , alleviating the need for large soft susy breaking terms , as has been explicitly shown in @xcite for one particular realization of a class - iii model @xcite .
finally , many of the configuration ( or proto - models ) which we have discussed contain exotic superfields , which might show up in the lhc .
it might therefore be interesting to do a more detailed study of the phenomenology of at least some particular of the models we have constructed .
this work has been supported in part by eu network grant unilhc pitn - ga-2009 - 237920 .
m.h . also acknowledges support from the spanish micinn grants fpa2011 - 22975 , multidark csd2009 - 00064 and by the generalitat valenciana grant prometeo/2009/091 . the work of r.m.f has been supported by _
fundao para a cincia e a tecnologia _ through the fellowship sfrh / bd/47795/2008 .
and j. c. r. also acknowledge the financial support from grants cftp - fct unit 777 , cern / fp/123580/2011 and ptdc / fis/102120/2008 .
we have considered @xmath5 based models which may contain any irreducible representation up to dimension 126 ( @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , @xmath106 ) .
once the gauge group breaks down to @xmath107 or @xmath108 these @xmath5 fields divide into a multitude of different irreducible representation of these groups .
in addition , if @xmath109 is broken down further to @xmath110 the following branching rules apply : @xmath111 ; @xmath112 ; @xmath113 .
the standard model s hypercharge , in the canonical normalization , is then equal to the combination @xmath114+\sqrt{\frac{2}{5}}\left[u(1)_{b - l}\textrm { hypercharge}\right]$ ] . in tables
[ tab : list_of_lr_fields ] , [ tab : list_of_patisalam_fields ] and [ tab : list_of_lr_fields_u1 ] we present the list of relevant fields respecting the conditions above . in these tables
we used an ordered naming of the fields .
sometimes it is also useful , like in table [ tab : lr_field_configuration_a ] , to indicate explicitly the quantum numbers under the various groups . in order for a group @xmath115 to break into a subgroup @xmath116 , there must be a field transforming non - trivially under @xmath115 which contains a singlet of @xmath117 that acquires vacuum expectation value . from this observation alone
we know that certain fields must be present in a fundamental model if we are to achieve a given breaking sequence : * the breaking @xmath118 is possible only with the ( * 15*,*1*,*1 * ) while @xmath119 requires the combination ( * 15*,*1*,*1 * ) + ( * 1*,*1*,*3 * ) . for the direct breaking @xmath120
there are two choices : ( * 4*,*1*,*2 * ) , ( * 10*,*1*,*3 * ) or their conjugates ; * the breaking @xmath121 requires the ( * 1*,*1*,*3*,0 ) representation while the direct route @xmath122 is possible with the presence of ( * 1*,*1*,*2*,-1 ) , ( * 1*,*1*,*3*,-2 ) or their conjugates ; * the group @xmath123 can be broken down to @xmath124 with the representations @xmath125 , @xmath126 or their conjugates .
super - kamiokande collaboration , y. fukuda _ et al .
_ , phys.rev.lett . * 81 * , 1562 ( 1998 ) , arxiv : hep - ex/9807003 .
sno collaboration , q. ahmad _ et al .
_ , phys.rev.lett . * 89 * , 011301 ( 2002 ) , arxiv : nucl - ex/0204008 .
kamland collaboration , k. eguchi _
et al . _ ,
phys.rev.lett . * 90 * , 021802 ( 2003 ) , arxiv : hep - ex/0212021 .
d. forero , m. tortola , and j.w.f .
valle , phys.rev .
* d86 * , 073012 ( 2012 ) , arxiv:1205.4018 .
p. minkowski , phys.lett .
* b67 * , 421 ( 1977 ) .
t. yanagida , kek lectures , ed .
o. sawada and a. sugamoto , tsukuba , japan ( 1979 ) .
m. gell - mann , p. ramond , and r. slansky , conf.proc .
* c790927 * , 315 ( 1979 ) .
r. n. mohapatra and g. senjanovic , phys.rev.lett . * 44 * , 912 ( 1980 ) .
h. fritzsch and p. minkowski , annals phys .
* 93 * , 193 ( 1975 ) .
r. n. mohapatra , contemporary physics ( springer , berlin , germany ) ( 1986 ) .
m. cvetic and j. c. pati , phys.lett . *
b135 * , 57 ( 1984 ) .
r. kuchimanchi and r. mohapatra , phys.rev . * d48 * , 4352 ( 1993 ) , arxiv : hep - ph/9306290 . c. s. aulakh , k. benakli , and g. senjanovic , phys.rev.lett .
* 79 * , 2188 ( 1997 ) , arxiv : hep - ph/9703434 . c. s. aulakh , a. melfo , a. rasin , and g. senjanovic , phys.rev . * d58 * , 115007 ( 1998 ) , arxiv : hep - ph/9712551
. s. k. majee , m. k. parida , a. raychaudhuri , and u. sarkar , phys.rev . *
d75 * , 075003 ( 2007 ) , arxiv : hep - ph/0701109 . v. de romeri , m. hirsch , and m. malinsky , phys.rev . *
d84 * , 053012 ( 2011 ) , arxiv:1107.3412 .
l. calibbi , l. ferretti , a. romanino , and r. ziegler , phys.lett . *
b672 * , 152 ( 2009 ) , arxiv:0812.0342 .
m. malinsky , j. romao , and j.w.f .
valle , phys.rev.lett .
* 95 * , 161801 ( 2005 ) , arxiv : hep - ph/0506296 .
p. b. dev and r. mohapatra , phys.rev .
* d81 * , 013001 ( 2010 ) , arxiv:0910.3924 .
j. c. pati and a. salam , phys.rev . * d10 * , 275 ( 1974 ) . m. r. buckley and h. murayama , phys.rev.lett . *
97 * , 231801 ( 2006 ) , arxiv : hep - ph/0606088 .
m. hirsch , s. kaneko , and w. porod , phys.rev .
* d78 * , 093004 ( 2008 ) , arxiv:0806.3361 .
j. esteves , j. romao , m. hirsch , f. staub , and w. porod , phys.rev . *
d83 * , 013003 ( 2011 ) , arxiv:1010.6000 .
particle data group , j. beringer _
et al . _ , phys.rev .
* d86 * , 010001 ( 2012 ) .
j. romao _ et al . _ , http://porthos.ist.utl.pt/arxiv/allso10guts .
r. n. mohapatra and j. w. f. valle , phys .
rev . * d34 * , 1642 ( 1986 ) . e. k. akhmedov , m. lindner , e. schnapka , and j.w.f .
valle , phys.lett .
* b368 * , 270 ( 1996 ) , arxiv : hep - ph/9507275 .
e. k. akhmedov , m. lindner , e. schnapka , and j.w.f .
valle , phys.rev . * d53 * , 2752 ( 1996 ) , arxiv : hep - ph/9509255 .
r. foot , h. lew , x. he , and g. c. joshi , z.phys .
* c44 * , 441 ( 1989 ) . k. babu , b. dutta , and r. mohapatra , phys.rev . * d60 * , 095004 ( 1999 ) , arxiv : hep - ph/9812421 .
s. davidson , d. c. bailey , and b. a. campbell , z.phys .
* c61 * , 613 ( 1994 ) , arxiv : hep - ph/9309310 .
a. kuznetsov , n. mikheev , and a. serghienko , ( 2012 ) , arxiv:1210.3697 .
r. m. fonseca , m. malinsky , w. porod , and f. staub , nucl.phys .
* b854 * , 28 ( 2012 ) , arxiv:1107.2670 . h. baer , v. barger , a. lessa , and x. tata , ( 2012 ) , arxiv:1207.4846 .
atlas collaboration , g. aad _
et al . _ , atlas - conf-2012 - 109 ( 2012 ) .
atlas collaboration , g. aad _
et al . _ ,
phys.lett . *
b716 * , 1 ( 2012 ) , arxiv:1207.7214 .
cms collaboration , s. chatrchyan _ et al .
_ , phys.lett . *
b716 * , 30 ( 2012 ) , arxiv:1207.7235 .
a. arbey , m. battaglia , a. djouadi , f. mahmoudi , and j. quevillon , phys.lett .
* b708 * , 162 ( 2012 ) , arxiv:1112.3028 . h. baer , v. barger , and a. mustafayev , phys.rev . *
d85 * , 075010 ( 2012 ) , arxiv:1112.3017 .
o. buchmueller _ et al .
_ , eur.phys.j .
* c72 * , 2020 ( 2012 ) , arxiv:1112.3564 . j. ellis and k. a. olive , eur.phys.j .
* c72 * , 2005 ( 2012 ) , arxiv:1202.3262 . m. hirsch , f. joaquim , and a. vicente , jhep * 1211 * , 105 ( 2012 ) , arxiv:1207.6635
. h. e. haber and m. sher , phys.rev . *
d35 * , 2206 ( 1987 ) .
m. drees , phys.rev . *
d35 * , 2910 ( 1987 ) .
m. hirsch , w. porod , l. reichert , and f. staub , phys.rev .
* d86 * , 093018 ( 2012 ) , arxiv:1206.3516 .
m. hirsch , m. malinsky , w. porod , l. reichert , and f. staub , jhep * 1202 * , 084 ( 2012 ) , arxiv:1110.3037 . | we construct lists of supersymmetric models with extended gauge groups at intermediate steps , all of which are based on so(10 ) unification .
we consider three different kinds of setups : ( i ) the model has exactly one additional intermediate scale with a left - right ( lr ) symmetric group ; ( ii ) so(10 ) is broken to the lr group via an intermediate pati - salam ( ps ) scale ; and ( iii ) the lr group is broken into @xmath0 , before breaking to the sm group .
we use sets of conditions , which we call the `` sliding mechanism '' , which yield unification with the extended gauge group(s ) allowed at arbitrary intermediate energy scales .
all models thus can have new gauge bosons within the reach of the lhc , in principle .
we apply additional conditions , such as perturbative unification , renormalizability and anomaly cancellation and find that , despite these requirements , for the ansatz ( i ) with only one additional scale still around 50 different variants exist that can have an lr symmetry below 10 tev . for the more complicated schemes ( ii ) and ( iii ) literally thousands of possible variants exist , and for scheme ( ii ) we have also found variants with very low ps scales .
we also discuss possible experimental tests of the models from measurements of susy masses .
assuming msugra boundary conditions we calculate certain combinations of soft terms , called `` invariants '' , for the different classes of models .
values for all the invariants can be classified into a small number of sets , which contain information about the class of models and , in principle , the scale of beyond - mssm physics , even in case the extended gauge group is broken at an energy beyond the reach of the lhc . |
hadron physics explores the least understood sector of the standard model ( sm ) , namely quantumchromodynamics ( qcd ) at large gauge coupling . in the region of energies below a few gev ,
one is faced with two different types of challenges : _ precision calculations / measurements : _ many very precise data exist which let one extract fundamental qcd parameters ( condensates , quark masses , etc . ) or explore bounds for physics beyond the sm . in such cases , precision calculations to accuracies of a few percent or better
have to be done , which in certain cases can be achieved utilizing effective field theory techniques .
however , it should be stressed that contrary to popular opinion , more such precise data are needed , e.g. in processes involving strange quarks .
space does not allow to discuss these very interesting developments here . _
the spectrum : _ to really understand the issue of confinement in the light quark sector , we must first bring order into the spectrum .
the nature and the spectrum of hadrons can only be understood if we are able to say with certainty which states are genuine quark model states , which are dynamically generated through channel couplings or which ones are exotic .
for that , precise measurements of many properties of these states are needed and theoretical tools have to be developed to separate the often overlapping resonances . in this talk , i will address two topics related to this type of problems , which are a ) some properties of the exotic @xmath0 ( the so - called pentaquark ) and b ) the way to achieve a unified description of resonances . clearly , this choice is very subjective , so many other interesting developments and/or challenges could not be discussed .
the parity of the recently discovered exotic state @xmath0 with positive strangeness is not yet determined @xcite ( i refer to the various plenary talks given at this conference for a detailed discussion on the various experiments and possible criticism as well as on theoretical approaches ) .
its parity is considered a decisive quantity regarding its substructure .
the most appealing proposal to determine the parity is based on the application of the pauli principle in the process @xmath1 @xcite since it links spin and parity . in @xcite
the spin correlation coefficient @xmath2 was identified as the crucial observable since in the low - energy region its sign is directly related to parity of the @xmath3 , see fig . 1 and also table 1 ( assuming that the @xmath3 has spin 1/2 ) . due to the self - analyzing hyperon decay
, the energy dependence of spin transfer coefficient @xmath4 might also be used to determine @xmath5 ( under certain favorable conditions ) , cf .
again table 1 ( where the threshold values / ranges are given ) . in these arguments
, one assumes naturalness of the contributing partial waves , in particular for the s - waves that dominate the region of small excess energy @xmath6 . in @xcite
are more detailed study of these proposals was performed , in particular , models were constructed were the leading s - waves are suppressed due to cancellations , thus explicitely violating the naturalness assumption .
still , it was shown that the energy dependence of @xmath7 , with @xmath8 the unpolarized cross section , guarantees unambiguous information on the parity of the @xmath3 .
this can be achieved e.g. by measurements at two @xmath6-values , say at 20 and 40 mev , to avoid the coulomb effects at smaller excess energies but still staying sufficiently close to threshold . for a more detailed discussion ( in particular the the usefulness of the quantity @xmath9 ) and more references on this issue
, please consult @xcite .
another interesting property of the @xmath3 is its width .
most experiments which have seen evidence of this state only quote upper limits for the width given by the resolution of the experiment .
for example , the fairly clean signal in proton - proton collisions reported by the cosy - tof collaboration @xcite gives @xmath10 due to the energy resolution .
in various papers , information on kaon
nucleon scattering @xcite was analyzed leading to the conclusion that the width of the @xmath3 has to be less than a few mev . a strongly interacting particle with such a small width
could truly be called `` exotic '' . in @xcite
the impact of the pentaquark on differential and integrated cross sections for the reaction @xmath11 , where experimental information is available at kaon momenta below 640 mev ( the mass range of 1520 to 1555 mev corresponds to kaon momenta in the range @xmath12 mev ) , was investigated .
the calculation utilizes the jlich kaon - nucleon model ( see @xcite and references therein ) and an extension of it that includes the contribution of a @xmath0 with a variable width .
the evaluation of the reaction @xmath11 takes into account effects due to the fermi motion of the nucleons within the deuteron and three body kinematics . as shown in fig . 2 ,
one concludes from that analysis that the data constrain the width of the @xmath0 to be less than 1 mev . in @xcite , the reaction @xmath13 was investigated in a meson - exchange model including rescattering of the secondary protons with the aim to analyze the evidence for the @xmath0 reported by the diana collaboration @xcite .
it was confirmed that the kinematical cuts introduced by the diana collaboration efficiently suppress the background to the @xmath14 reaction which may contribute to the @xmath3 production .
these kinematical cuts do not produce a narrow structure in the @xmath15 effective mass spectrum near 1540 mev , cf .
the left panel of fig .
the effect of a narrow resonance with both positive and negative parity in comparison to the diana data was studied in @xcite .
it is shown that the @xmath13 calculations without @xmath3 contribution as well as the results obtained with a @xmath3 width of 1 mev are in comparably good agreement with the diana results , see fig .
the @xmath16/dof is @xmath17 , @xmath18 and @xmath19 for no @xmath3 , a positive and a negative parity spin-1/2 pentaquark with a width of 1 mev , in order ( with the cuts applied to the @xmath20 invariant mass spectrum ) . in view of these results ,
the three star rating for the @xmath3 in the recent pdg tables seems somewhat optimistic @xcite .
more dedicated experiments are called for to establish ( or rule out ) this exotic baryon resonance . [ cols="^,^ " , ]
this section first presents some results on the baryon spectrum ( for definiteness , a few states are selected to make the point ) using two very different approaches . in the second more speculative part
some ideas are presented how one could unify these approaches and eventually gain a deeper insight into the confinement regime of qcd .
the quark model has been particularly successful in bringing order into the many observed meson and baryon resonances . in the last years , it has become evident that a covariant framework is necessary to deal with _ all _ the ( ir)regularities found in the hadron spectrum .
here , i will focus on the baryon resonances and pick one particular covariant quark model , in which the residual interaction of the constituent quarks are a flavor - dependent instanton induced interaction and a linear confinement potential ( for details , see @xcite ) .
this approach leads to a fairly good description of the spectrum , including states with very high spin and is consistent with regge phenomenology .
the t hooft interaction also explains the near degeneracy of many positive and negative parity states @xcite , which is very different from the speculations about chiral restauration in the hadron spectrum at masses above @xmath21gev which one finds frequently in the current literature .
many electroweak observables @xcite and heavy quark properties have also been successfully calculated .
strong pion decays come out typically a factor two too small , but the overall pattern of pion couplings is well described since @xcite the so - called `` missing resonances '' have couplings that are at least an order of magnitude smaller than for the states that couple with normal strength to the pion - nucleon continuum .
just to be precise , in this approach the @xmath22 , the @xmath23 and the @xmath24 are three
quark states @xcite .
note , however , that the @xmath22 is not well described in this approach , pointing towards the bound - state scenario discussed below .
furthermore , in the framework of this model , one also generates a whole zoo of states with quark content @xmath25 , some of them exotic and many of them with conventional quantum numbers , which also has to be kept in mind in the discussion of the pentaquark @xmath3 - there simply can not just be one state like that , there must be many ( so far one experiment has a hint for another exotic baryon from the anti - decuplet @xcite ) .
[ for a different modern quark model , see e.g. @xcite ] .
next , i discuss how hyperon resonances can be generated as meson - baryon bound states .
for that , consider @xmath26 scattering .
a purely perturbative treatment is not possible due to the strong channel couplings and the appearance of a subthreshold resonance , the @xmath22 .
a non - perturbative resummation scheme is mandatory to generate a bound state or a resonance .
there exist many such approaches , but it is possible and mandatory to link such a scheme tightly to the chiral qcd dynamics .
such an improved approach was developed for pion
nucleon @xcite and later applied to @xmath27n scattering @xcite .
the starting point is the t
matrix for any partial wave , which can be represented in closed form if one neglects for the moment the crossed channel ( left - hand ) cuts , @xmath28\,$ ] , with @xmath29 the cm energy .
@xmath30 collects all local terms and poles and @xmath31 is the meson - baryon loop function ( the fundamental bubble ) that is resummed by e.g. dispersion relations in a way to exactly recover the right - hand ( unitarity ) cut contributions .
the function @xmath31 needs regularization , this can be best done in terms of a subtracted dispersion relation and using dimensional regularization .
it is important to ensure that in the low - energy region , the so constructed t
matrix agrees with the one of chpt ( matching ) .
in addition , one has to recover the contributions from the left - hand cut .
this can be achieved by a hierarchy of matching conditions , which enforce that to a given order in the chiral expansion , the unitarized amplitude agrees with the chpt amplitude ( eventually in some unphysical region ) .
such a procedure tightly constrains the unitarization procedure , for details see @xcite .
it was observed in @xcite that there are indeed two poles close to the nominal @xmath22 resonance .
the physics behind these two poles was recently revealed in @xcite .
starting from an su(3 ) symmetric lagrangian to couple the meson octet to the baryon octet ( in that limit , all octet goldstone boson masses and all octet baryon masses are equal ) , one could in principle generate a variety of resonances according to the su(3 ) decomposition , @xmath32 . as it turns out ,
the leading order transition potential is attractive only in the singlet and the two octet channels , so that one a priori expects a singlet and two octets of bound states .
however , the two octets come out degenerate .
this has no particular dynamical origin but rather is a consequence of the actual values of the su(3 ) structure constants . in the real world , there is of course su(3 ) breaking of various origins .
this was parameterized in @xcite in terms of a symmetry breaking parameter @xmath33 in the expressions for the meson @xmath34 and baryon masses @xmath35 as well as the subtraction constants @xmath36 via @xmath37 and @xmath38 , with @xmath39mev , @xmath40mev and @xmath41 , where @xmath42 .
the motion of the various poles in the complex energy plane as a function of @xmath33 is shown in fig .
we note that the two octets split , in particular , one moves to lower energy ( @xmath43mev ) close to the position of the singlet ( @xmath44mev ) .
these are the two poles which combine to give the @xmath22 as it appears in various reactions .
recent kaon photoproduction data from spring-8 seem to give support to this two - pole scenario .
we note that the @xmath23 and the @xmath24 are also generated ( the latter is the right - most pole in fig . 4 ) , besides other resonances with strangeness @xmath45 , see ref .
@xcite and also the extensive studies in @xcite .
here , i will speculate how one can bring together the quark model and the coupled channel approach to achieve a unified description of resonances .
for that , one can follow a two - step procedure , as discussed below . before going into the details , i remark that step 1 has been done ( to some extent ) whereas the second step so far is science fiction , but i consider it an important topic to be addressed in the coming years .
more precisely , i envision the following : _ step 1 : _ the coupled channel approach can be extended in that one can also include resonance fields by saturating the local contact terms in the effective lagrangian through explicit meson and baryon resonances ( for details , see @xcite ) . in particular , in this framework one can cleanly separate genuine quark resonances from dynamically generated resonance like states .
the former require the inclusion of an explicit field in the underlying lagrangian , whereas in the latter case the fit will arrange itself so that the couplings to such an explicit field will vanish .
this framework was applied in @xcite to the case of pion nucleon scattering below the @xmath46 threshold .
some typical results are : 1 ) the delta pole is located at @xmath47 mev , in agreement with dispersion - theoretical studies of pion nucleon scattering and pion photoproduction . furthermore
, a genuine 3-quark component is needed to build up this state .
2 ) the rho requires a quark - antiquark component , consistent with large n@xmath48 investigations and its tensor coupling comes out large , @xmath49 .
3 ) no light scalar is needed to describe the data , all the interaction in the t - channel with scalar - isoscalar quantum numbers is build up by pion rescattering ( loops ) .
this is consistent with the analysis of other reactions , like e.g. @xmath50 ( with @xmath51 @xcite or the scalar form factor of the pion @xcite as well as the intermediate range attraction in the central part of the two nucleon potential .
_ step 2 : _ a further step would be to merge the coupled channel dynamics with quark model predictions . in fig . 5
the underlying idea is sketched for the example of the strong @xmath52 transition . in a pure quark model description
based e.g. on the bethe - salpeter formalism , one can only describe part of the @xmath52 decay , since there is a sizeable modification of the vertex through final state interactions ( often called `` dressing '' ) and also channel coupling ( some typical graphs are shown on the right - hand - side of that figure ) .
in fact , the idea would be to use the vertices and masses generated in the quark model as input in these loop corrections .
that way , the aforementioned problem of the too small strong width should be solved and one also obtains a better representation of the underlying dynamics .
ultimately , this should be extended to many channels so as to achieve a truly unified description of the spectrum of the strongly interacting particles and their properties .
to come back to the hyperon states discussed before , such a scheme would allow one to really pin down the nature of these states , my guess is that while the @xmath24 and the @xmath53 are dominantly three - quark states , the @xmath22 will not have a three - quark component but simply be a dynamically generated state with different response to external probes and decay patterns . to end this section ,
i should point out that such ideas are not entirely new , but in my opinion only now the theoretical tools have become available to seriously tackle this problem .
i would like to stress again that hadron structure and dynamics is one of the central issues of qcd ( of the standard model ) and require non perturbative methods in the few gev region and below .
apart from the topics discussed before , there are many further challenges related to ( i just name a few ) : * lattice qcd and chiral extrapolations .
most lattice data are obtained for large unphysical values of the quark ( pion ) masses and thus must be interpolated by means of chiral perturbation theory to the physical value of @xmath54 . on the example of the nucleon mass ,
the possible loopholes and limitations of such methods are discussed in @xcite .
it is evident that for pion masses larger than 500 mev , chiral extrapolations cease to be useful .
+ * as noted in the introduction , precision hadronic calculations are needed to e.g. really be able to address the issue whether physics beyond the sm has been seen in the precise bnl measurements of the muon anomalous magnetic moment @xcite ( my guess would be not ) or if one wants to precisely test ckm unitarity in b - decays .
another recent example are the radiative corrections to tritium @xmath55-decay @xcite to reduce the upper limit on the ( electron ) neutrino mass @xcite .
in addition , there are some real puzzles when comparing precision data to precision calculations . as an example , in fig . 6 a calculation of the ground - state energy shift and width of
kaonic hydrogen @xcite compared to earlier measurements @xcite and the recent result from dear @xcite is shown .
as the only input , the s - wave scattering length combination @xmath56 was taken from the dispersive analysis of martin @xcite and from the unitarized chpt analysis of @xcite . in both cases , the theoretical prediction ( filled boxes )
is in stark disagreement with the dear result . at present
, this discrepancy is unexplained .
* there has been impressive progress made in nuclear effective field theory .
in particular , two- and three - nucleon forces can be described in one consistent scheme . by now , the nucleon
nucleon force has been worked out to next - to - next - to - next - to - leading order in the eft expansion @xcite and has reached the accuracy needed to do precise calculations in few - nucleon systems .
other advances in this field concern the quark mass dependence of light nuclei @xcite and the speculation of an infrared renormalization group limit cycle in qcd @xcite . for many other applications , in particular low energy reactions of relevance for astrophysics and element synthesis , see e.g. refs.@xcite .
i would like to thank the organizers to give me this opportunity , my collaborators j. haidenbauer , c. hanhart , d. jido , j.a .
oller , e. oset , u. raha , a. ramos , a. rusetsky and a. sibirtsev as well as b.c .
metsch and h.r .
petry for sharing their insight into the topics discussed here .
i have also profited from collaborations with v. bernard , e. epelbaum , s. gardner , h .- w .
hammer and w. glckle .
special thanks to paulo bedaque and the nuclear theory group at lbnl for their hospitality where part of this work was done .
a. w. thomas , k. hicks and a. hosaka , prog .
* 111 * ( 2004 ) 291 [ arxiv : hep - ph/0312083 ] . c. hanhart
_ et al . _ ,
b * 590 * ( 2004 ) 39 [ arxiv : hep - ph/0312236 ] . c. hanhart , phys .
* 397 * ( 2004 ) 155 [ arxiv : hep - ph/0311341 ] . c. hanhart , j. haidenbauer , k. nakayama and u .- g .
meiner , arxiv : hep - ph/0407107 .
m. abdel - bary _ et al . _
[ cosy - tof collaboration ] , arxiv : hep - ex/0403011 .
s. nussinov , arxiv : hep - ph/0307357 ; r. n. cahn and g. h. trilling , phys .
d * 69 * ( 2004 ) 011501 [ arxiv : hep - ph/0311245 ] ; r. a. arndt , i. i. strakovsky and r. l. workman , phys .
c * 68 * ( 2003 ) 042201 [ erratum - ibid .
c * 69 * ( 2004 ) 019901 ] [ arxiv : nucl - th/0308012 ] ; j. haidenbauer and g. krein , phys .
c * 68 * ( 2003 ) 052201 [ arxiv : hep - ph/0309243 ] ; w. r. gibbs , arxiv : nucl - th/0405024 .
a. sibirtsev , j. haidenbauer , s. krewald and u .- g .
meiner , arxiv : hep - ph/0405099 .
m. hoffmann , j. w. durso , k. holinde , b. c. pearce and j. speth , nucl .
a * 593 * ( 1995 ) 341 [ arxiv : nucl - th/9504029 ] .
a. sibirtsev , j. haidenbauer , s. krewald and u .- g .
meiner , arxiv : nucl - th/0407011 . v. v. barmin _ et al . _ [ diana collaboration ] , phys . atom .
* 66 * ( 2003 )
1715 [ yad . fiz .
* 66 * ( 2003 ) 1763 ] [ arxiv : hep - ex/0304040 ] .
u. lring , k. kretzschmar , b. c. metsch and h. r. petry , eur .
phys . j. a * 10 * ( 2001 ) 309 [ arxiv : hep - ph/0103287 ] .
u. lring and b. metsch , arxiv : hep - ph/0110412 .
d. merten , u. lring , b. metsch and h. petry , eur .
phys . j. a * 18 * ( 2003 ) 193 .
b. metsch , u. lring , d. merten and h. petry , eur .
j. a * 18 * ( 2003 ) 189 .
u. lring , b. c. metsch and h. r. petry , eur .
phys . j. a * 10 * ( 2001 ) 447 [ arxiv : hep - ph/0103290 ] . c. alt _ et al .
_ [ na49 collaboration ] , phys .
* 92 * ( 2004 ) 042003 [ arxiv : hep - ex/0310014 ] . l. y. glozman , w. plessas , k. varga and r. f. wagenbrunn , phys .
d * 58 * ( 1998 ) 094030 [ arxiv : hep - ph/9706507 ] ; r. f. wagenbrunn , s. boffi , w. klink , w. plessas and m. radici , phys .
b * 511 * ( 2001 ) 33 [ arxiv : nucl - th/0010048 ] ; s. boffi , l. y. glozman , w. klink , w. plessas , m. radici and r. f. wagenbrunn , eur .
j. a * 14 * ( 2002 ) 17 [ arxiv : hep - ph/0108271 ] .
meiner and j. a. oller , nucl .
a * 673 * ( 2000 ) 311 [ arxiv : nucl - th/9912026 ] .
j. a. oller and u .- g .
meiner , phys .
b * 500 * ( 2001 ) 263 [ arxiv : hep - ph/0011146 ] .
d. jido , j. a. oller , e. oset , a. ramos and u .- g .
meiner , nucl .
phys . a * 725 * ( 2003 ) 181 [ arxiv : nucl - th/0303062 ] .
e. oset , a. ramos and c. bennhold , phys .
b * 527 * ( 2002 ) 99 [ erratum - ibid .
b * 530 * ( 2002 ) 260 ] [ arxiv : nucl - th/0109006 ] .
s. sarkar , e. oset and m. j. v. vacas , arxiv : nucl - th/0407025 .
e. e. kolomeitsev and m. f. m. lutz , phys .
b * 585 * ( 2004 ) 243 [ arxiv : nucl - th/0305101 ] ; m. f. m. lutz and e. e. kolomeitsev , nucl .
a * 730 * ( 2004 ) 110 [ arxiv : hep - ph/0307233 ] ; arxiv : nucl - th/0402084 .
meiner and j. a. oller , nucl .
phys . a * 679 * ( 2001 ) 671 [ arxiv : hep - ph/0005253 ] .
meiner , comments nucl .
phys . * 20 * ( 1991 ) 119 .
v. bernard , t. r. hemmert and u .- g .
meiner , nucl .
a * 732 * ( 2004 ) 149 [ arxiv : hep - ph/0307115 ] . g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
* 92 * ( 2004 ) 161802 [ arxiv : hep - ex/0401008 ] .
s. gardner , v. bernard and u .- g .
meiner , arxiv : hep - ph/0407077 .
a. osipowicz _ et al .
_ [ katrin collaboration ] , arxiv : hep - ex/0109033 .
meiner , u. raha and a. rusetsky , eur .
j. c * 35 * ( 2004 ) 349 [ arxiv : hep - ph/0402261 ] .
j. d. davies _ et al .
_ , phys .
b * 83 * ( 1979 ) 55 ; m. izycki _ et al .
_ , z. phys . a * 297 * ( 1980 ) 11 ; p. m. bird _
et al . _ ,
nucl . phys .
a * 404 * ( 1983 ) 482 ; m. iwasaki _ et al . _ ,
* 78 * ( 1997 ) 3067 .
g. beer _ et al . _
[ dear collaboration ] , prog .
* 149 * ( 2003 ) 240 .
a. d. martin , nucl .
b * 179 * ( 1981 ) 33 .
d. r. entem and r. machleidt , phys .
c * 68 * ( 2003 ) 041001 [ arxiv : nucl - th/0304018 ] .
e. epelbaum , w. glckle and u .- g .
meiner , arxiv : nucl - th/0405048 . s. r. beane and m. j. savage , nucl . phys .
a * 713 * ( 2003 ) 148 [ arxiv : hep - ph/0206113 ] .
e. epelbaum , u .- g .
meiner and w. glckle , nucl .
a * 714 * ( 2003 ) 535 [ arxiv : nucl - th/0207089 ] .
e. braaten and h. w. hammer , phys .
* 91 * ( 2003 ) 102002 [ arxiv : nucl - th/0303038 ] .
s. r. beane , p. f. bedaque , w. c. haxton , d. r. phillips and m. j. savage , in m. shifman ( ed . ) : _ at the frontier of particle physics _
, vol . 1 .
, pp . 133 - 269 , [ arxiv : nucl - th/0008064 ] . t. s. park , k. kubodera , d. p. min and m. rho , nucl .
a * 684 * ( 2001 ) 101 [ arxiv : nucl - th/0005069 ] . | in this talk , i address some open problems in hadron physics and stress their importance for a better understanding of qcd in the confinement regime .
hiskp - th-04/16 |
given a graph @xmath0 we orient each edge with equal probability for the two possible directions and independent of all other edges .
this model has been studied previously in for instance @xcite .
let @xmath1 , @xmath2 and @xmath3 be three distinct vertices in @xmath4 .
the object of this paper is to study the correlation of the two events @xmath5 , that there exists a directed path from @xmath1 to @xmath2 , and @xmath6 .
one might intuitively guess that they are always negatively correlated , i.e. that @xmath13 .
this is however not true for all graphs .
in fact , the smallest counterexample is the graph on four vertices with all edges except @xmath14 present , see section [ s : ce ] .
in section [ s : kn ] we prove that for the complete graph , @xmath8 , the events are negatively correlated for @xmath9 , independent for @xmath10 and positively correlated for @xmath11 .
the complementary events , @xmath15 , that there does not exist a directed path from @xmath1 to @xmath2 , and @xmath16 have the same covariance , and we show that their relative covariance , @xmath17 converges to @xmath18 as @xmath19 . in section [ s : rec ]
we give exact recursions for the probabilities @xmath20 and @xmath21 , and compute these for @xmath22 . for completeness , in section [
s : ct ] we show that the events are negatively correlated when @xmath7 is a cycle and negatively correlated or independent when @xmath7 is a tree ( or a forest ) .
we end with stating a number of conjectures and open problems . in a coming paper , @xcite
, we will study this problem when @xmath7 is the random graph @xmath23 .
the question studied here was posed in @xcite .
there it was proved that under this model for any vertices @xmath24 the events @xmath25 and @xmath6 are never negatively correlated .
this was shown to be true also if we first conditioned on @xmath26 , i.e. as a sort of converse it was also proved that @xmath27 .
the proofs in @xcite relied heavily on the results in @xcite and @xcite , where similar statements were proved for edge percolation on a given graph and a result from @xcite that relates the random orientation with edge percolation .
this cluster of questions on correlation of paths have been inspired by an interesting conjecture due to kasteleyn , named the bunkbed conjecture by hggstrm @xcite , see also @xcite and remark 5 in @xcite .
* acknowledgment : * we thank svante janson , stanislav volkov and johan wstlund for fruitful discussions .
let @xmath7 be a graph with four vertices and all edges except the one between @xmath1 and @xmath3 present , see figure [ f : pos ] , and let @xmath28 and @xmath29 .
then , @xmath30 and @xmath31 , so that @xmath32 note that if we relabel the vertices in this graph as in figure [ f : neg ] , we still get @xmath33 , but @xmath34 , so that @xmath35 in fact , the labeling in figure [ f : pos ] is the only graph with four vertices that gives a positive correlation .
let @xmath7 be the complete graph , @xmath8 , and orient the edges either way independently with probability @xmath36 .
for three different vertices , @xmath1 , @xmath2 and @xmath3 , of @xmath8 we want to know if the event @xmath37 and the event @xmath6 are positively or negatively correlated .
it turns out to be easier to study the correlation of the complementary events , i.e. @xmath15 , that there does not exist a directed path from @xmath1 to @xmath2 , and the event @xmath16 , which have the same correlation .
think of the vertices of @xmath8 as @xmath38:=\{1,\dots , n\}$ ] . to estimate @xmath20
, the following lemma will be useful .
[ l : a ] for all @xmath39 , @xmath40 as @xmath41 , we will assume that @xmath42 .
we will use that @xmath43 , if @xmath44 and @xmath45 , and split the sum into two parts , @xmath46 and @xmath47 .
@xmath48 the lemma follows by showing that @xmath49 holds for all @xmath42 . as @xmath46 contributes @xmath50 , this gives a lower bound for @xmath51 .
this is in fact the dominating term , so that @xmath51 asymptotically is of order @xmath52 , where @xmath53 is a polynomial in @xmath54 . by splitting the sum into three parts , we can
, for example , show that @xmath55 for all @xmath39 .
[ l : a ] for all @xmath42 , @xmath56 a necessary condition for @xmath57 is that the edge between @xmath1 and @xmath2 is directed from @xmath2 to @xmath1 .
let @xmath58 , with @xmath59 , denote this event .
let @xmath60 and @xmath61 denote the sets of points in @xmath38\setminus\{a , s\}$ ] that can be reached from @xmath1 and @xmath2 respectively in one step .
similarly , let @xmath62 and @xmath63 denote the sets of points in @xmath38\setminus\{a , s\}$ ] that can reach @xmath1 and @xmath2 respectively in one step .
for the lower bound , note that @xmath64 and @xmath65 , so that @xmath66 for the upper bound , note that @xmath67 , where @xmath68 is the event that the points in @xmath60 have no directed edges to points in @xmath63 .
note that , if @xmath69 and @xmath70 , then @xmath71 is necessary , and there are @xmath72 edges in @xmath68 .
@xmath73 where the function @xmath1 was defined in lemma [ l : a ] .
[ t : a ] @xmath74 follows immediately from theorem [ l : a ] . to estimate @xmath21 ,
the following lemma , in combination with lemma [ l : a ] , is useful .
[ l : b ] for all @xmath39 , @xmath75 note first that @xmath76 , so that we may assume that @xmath77 .
note also that for @xmath78 , @xmath79 , so that @xmath80 .
this gives @xmath81 [
l : ab]for all @xmath77 , @xmath82 a necessary condition for @xmath83 is that the edge between @xmath1 and @xmath2 is directed from @xmath2 to @xmath1 , that the edge between @xmath2 and @xmath3 is directed from @xmath3 to @xmath2 and that the edge between @xmath1 and @xmath3 is directed from @xmath3 to @xmath1 .
let @xmath58 , with @xmath84 , denote this event .
let @xmath60 , @xmath61 and @xmath85 denote the sets of points in @xmath38\setminus\{a , s , b\}$ ] that can be reached from @xmath1 , @xmath2 and @xmath3 respectively in one step .
similarly , let @xmath62 , @xmath63 and @xmath86 denote the sets of points in @xmath38\setminus\{a , s , b\}$ ] that can reach @xmath1 , @xmath2 and @xmath3 respectively in one step .
for the lower bound , note that @xmath87 , @xmath88 and @xmath89 , so that @xmath90 as @xmath91 . for the upper bound
, we note that @xmath92 , where @xmath68 denotes the event that the points in @xmath60 have no directed edges to points in @xmath63 and the points in @xmath61 have no directed edges to points in @xmath86 , so that @xmath93 .
let @xmath94 , @xmath95 and @xmath96 .
then @xmath97 , @xmath98 and @xmath99 and the direction of all @xmath100 edges connected to @xmath1 , @xmath2 and @xmath3 are determined .
further , the event @xmath68 determines the direction of @xmath101 edges , so that @xmath102 where the triple sum is split into seven parts : + 1 : @xmath103 , 2 : @xmath104 , 3 : @xmath105 , + 4 : @xmath106 , 5 : @xmath107 , + 6 : @xmath108 , + 7 : @xmath109 .
+ the first three cases correspond to the three cases of the lower bound , @xmath110 the next three can be expressed by the function @xmath1 of lemma [ l : a ] , @xmath111 and the last by the function @xmath3 of lemma [ l : b ] , @xmath112 collecting the estimates gives the lemma .
[ t : ab ] @xmath113 follows immediately from theorem [ l : ab ] .
[ t : relkorr ] @xmath114 follows immediately from theorems [ t : a ] and [ t : ab ] as @xmath115 .
[ r : korr ] note that @xmath116 so that theorem [ t : relkorr ] can be formulated as @xmath117 theorem [ t : relkorr ] shows that the events @xmath118 and @xmath119 are positively correlated for sufficiently large @xmath54 . from this
follows that the complementary events @xmath120 and @xmath121 also are positively correlated for sufficiently large @xmath54 .
it is in fact true for all @xmath122 as the next theorem shows .
[ t : korr ] the events @xmath118 and @xmath119 are negatively correlated for @xmath9 , independent for @xmath10 and positively correlated for @xmath122 . from lemmas [ l :
a ] and [ l : ab ] we get @xmath123 where @xmath124 is a decreasing function of @xmath54 , with @xmath125 , so that the theorem holds for @xmath126 .
the remaining cases , @xmath127 , are proved using the recursion formulas in lemmas [ l : f ] and [ l : g ] in the next section .
in this section we will derive recursions for @xmath20 and @xmath21 . for @xmath128 ,
@xmath129 $ ] and @xmath130\setminus \{s\}$ ] let @xmath131 denote the event @xmath132 for every @xmath133 .
with @xmath134 define @xmath135 where in particular @xmath136 .
also set @xmath137 for convenience . for @xmath138 and @xmath139 , k\subset [ n]\setminus \{s , b\}$ ] , @xmath140 and @xmath134
define : @xmath141 where in particular @xmath142 . also let @xmath143 .
[ l : f ] for @xmath144 we have @xmath145 we first deduce the following recursion : @xmath146\setminus ( k\cup \{s\ } ) } \frac{1}{2^{k(n - k-|l|)}}\big(1-\frac{1}{2^k}\big)^{|l| } p_{n - k}(l\nto s).\ ] ] this can be seen as follows .
we think of the set @xmath147 as the vertices that we can reach from vertices in @xmath148 in one step .
clearly , we must have @xmath149 .
every edge from @xmath38\setminus ( k\cup l)$ ] to @xmath148 must be oriented in that direction , which gives the first power of @xmath150 . to reach a vertex @xmath151 in one step
it must not be the case that all edges @xmath152 are directed away from @xmath153 , which gives the second factor .
the edges within @xmath148 make no difference and we have considered all edges going between @xmath148 and @xmath38\setminus k$ ] .
we are left with a situation where we must make sure that there is no directed path from any vertex in @xmath147 to @xmath2 passing over vertices in @xmath38\setminus k$ ] , which gives the last term .
the recursion in the lemma is easily deduced from this , by summing over the size , @xmath154 , of @xmath147 .
[ l : g ] for @xmath155 we have @xmath156 first note that @xmath131 and @xmath157 implies @xmath158 .
reasoning as in the previous proof , we obtain @xmath159\setminus ( k\cup \{s\}\cup \{b\ } ) } \frac{1}{2^{k(n - k-|l|)}}\big(1-\frac{1}{2^k}\big)^{|l| } p_n(l\nto s , s\nto b).\ ] ] using the lemmas , we can recursively compute the desired probabilities @xmath160 and @xmath161 .
exact and numerical values of @xmath20 and @xmath21 were computed and are given in table [ tab : corr ] together with numerical values of the relative covariance @xmath17 for @xmath162 .
.probabilities and relative covariances [ cols= " > , > , > , > , > , > " , ]
let @xmath163 , the cycle with @xmath54 vertices , and let @xmath153 , @xmath164 and @xmath165 denote the distances ( number of edges ) between @xmath1 and @xmath2 , @xmath2 and @xmath3 , and @xmath3 and @xmath1 , respectively .
we assume the three vertices to be distinct , so that @xmath166 .
further , let @xmath28 and @xmath29 . then , @xmath167 where the last term corresponds to the case when there is a directed path @xmath168 .
this gives @xmath169 with equality if and only if @xmath170 .
we have proved the following theorem .
[ t : cycles ] when @xmath7 is the cyclic graph with @xmath54 nodes , @xmath12 , the covariance between the events @xmath5 and @xmath6 is at most @xmath171 , with equality if and only if the vertices @xmath1 and @xmath3 are adjacent to @xmath2 . for trees
the situation is even simpler , as there are no cycles so that there is a unique path between any two vertices .
two cases can occur .
if the path between @xmath1 and @xmath3 passes @xmath2 , the paths between @xmath1 and @xmath2 and between @xmath2 and @xmath3 have no edges in common , so that the events @xmath5 and @xmath6 are independent . on the other hand , if the path between @xmath1 and @xmath3 does not pass @xmath2 , then the events are disjoint , so that @xmath172 , and the covariance is strictly negative .
for a forest , if not all three vertices are in the same tree , then trivially @xmath172 and at least one of @xmath173 and @xmath174 is zero , so that the events are independent . when @xmath7 is a tree ( or a forest ) , the events @xmath5 and @xmath6 are either independent or mutually exclusive .
from theorem [ t : cycles ] and the observations in section [ s : ce ] , we make the following conjecture . for any connected graph @xmath0 and three distinct vertices @xmath1 , @xmath2 and @xmath3 in @xmath4 ; if @xmath2 has degree at most two , then the events @xmath5 and @xmath6 are independent or negatively correlated . * for any three distinct vertices @xmath176 , the events @xmath5 and @xmath6 are non - positively correlated .
* there exist three distinct vertices @xmath176 , such that the events @xmath5 and @xmath6 are negatively correlated and there exist three distinct vertices @xmath177 , such that the events @xmath178 and @xmath179 are positively correlated . or there exist three distinct vertices @xmath176 , such that the events @xmath5 and @xmath6 are independent . * for any three distinct vertices @xmath176 , the events @xmath5 and @xmath6 are non - negatively correlated . we have shown that trees and cycles belong to class i , @xmath180 belongs to class iii and @xmath181 minus one edge belongs to class ii . note that when we have independent events there may be some overlap between the classes
, in particular @xmath181 belongs to all three classes .
in fact we guess that for @xmath54 large enough , the graphs in class i are joins ( in some vague sense ) of cycles and trees
. it would be interesting if it was possible to characterize the graphs in class i. we formulate the following more specific questions .
recall that outerplanar graphs are the graphs that do not have @xmath181 or @xmath182 as minors .
is it true that if @xmath7 belongs to class i , but not to class ii , then so does any connected subgraph obtained by removing one edge ?
similarly , is it true that if @xmath7 belongs to class iii , but not to class ii , then so does @xmath7 plus any new edge ?
olle hggstrm , probability on bunkbed graphs , _ proceedings of fpsac03 , formal power series and algebraic combinatorics _
linkping , sweden 2003 .
available at + http://www.fpsac.org/fpsac03/articles/42.pdf | consider a randomly oriented graph @xmath0 and let @xmath1 , @xmath2 and @xmath3 be three distinct vertices in @xmath4 .
we study the correlation between the events @xmath5 and @xmath6 .
we show that , when @xmath7 is the complete graph @xmath8 , the correlation is negative for @xmath9 , zero for @xmath10 , and that , counter - intuitively , it is positive for @xmath11 . we also show that the correlation is always negative when @xmath7 is a cycle , @xmath12 , and negative or zero when @xmath7 is a tree ( or a forest ) . |
bose - einstein condensates are ideal artificial systems for quantum manipulation .
the spinor condensates have very rich spin structures , and their swift response to the external field is notable @xcite .
usually , a condensate would contain more than @xmath2 atoms .
due to the progress in techniques , much smaller condensates ( say , particle number @xmath0 ) could be produced .
note that the strength of the effective interaction between the particles @xmath3 is proportional to the average particle density , which is proportional to @xmath4 ( evaluated based on the thomas - fermi approximation ) .
thus the particles are subjected to a stronger interaction in smaller systems .
in fact , in the study of spin - dynamics , the rate of evolution depends on @xmath5 rather than @xmath6 @xcite , where @xmath6 is the time . therefore , the evolution will be swifter in the smaller systems
. thus they might be even more suitable for manipulation . in principle
, these systems could be more precisely prepared ( such as @xmath7 and @xmath8 , the total magnetization , could be more rigorously given ) .
therefore , some properties not contained in large condensates might emerge ( say , if @xmath9 , whether @xmath7 is even or odd might be serious ) , and some properties might depend on the external field very sensitively in some particular cases ( as shown below ) . there are already a number of literatures dedicated to the large condensates of spin-1 atoms .
the commonly used theoretical tool is the mean field theory ( mft ) @xcite .
the ground states with the ferromagnetic phase and polar phase are found .
their modes of excitation and their dynamical behavior have been studied , and very rich physical phenomena have been found ( say , the formation of spin - domains and spin - vortices ) . on the other hand , for small condensates ,
a theory goes beyond the mft might be more appropriate . in this paper ,
the spin - structures of spin-1 small condensates under a magnetic field @xmath1 are studied by using a many - body theory in which the parentage coefficients of spin - states are used as a tool @xcite .
the variation of @xmath1 is considered as adiabatic and the discussion is limited to static behavior .
one - body and two - body spin correlative probabilities are defined and calculated .
they are used to extract information from the spin eigenstates.@xcite furthermore , the correlation coefficients are defined and calculated to measure quantitatively the spin correlation , and the fidelity susceptibilities are also calculated to measure quantitatively the sensitivity of the ground states against the change of @xmath1 .
thereby a detailed and deeper description on spin - spin correlation has been obtained that might lead to a better understanding on spin structures . in particular , the spin probabilities defined in this paper are observables , they might serve as a bridge to relate theories and experiments , therefore can be used to clarify various spin - structures and interactions .
the emphasis is placed on the response of the condensates to @xmath1
. the related knowledge might be useful for quantum manipulation .
both the cases with the temperature @xmath10 zero and nonzero are considered .
let @xmath7 spin-1 atoms be confined by an isotropic and parabolic trap with frequency @xmath11 .
the interaction is @xmath12 , where @xmath13 and @xmath14 are , respectively , the position vector and the spin operator of the @xmath15-th particle . a magnetic field @xmath1 lying along the @xmath16-axis is applied .
we consider the case that the size of the condensate is small so that it is smaller than the spin healing length . in this case
, the single mode approximation ( sma ) is applicable.@xcite under the sma all particles will have the same spatial wave function @xmath17 . after the integration over the spatial degrees of freedom , we arrive at a model hamiltonian @xmath18 where @xmath19 is the total spin operator of the @xmath7 particles , @xmath20-component of @xmath14 , @xmath21 , @xmath22 , @xmath23 the bohr magneton and @xmath24 the hyperfine splitting energy .
the last two terms of @xmath25 are the linear and quadratic zeeman energies , respectively .
the eigenstates of @xmath25 will have @xmath26 and its @xmath27-component @xmath8 ( namely , the magnetization , @xmath28 is assumed ) conserved when the quadratic term is neglected .
therefore they can be denoted as @xmath29 , @xmath30 , @xmath31 , to @xmath32 or @xmath33 .
it has been proved that @xmath34 is unique without further degeneracy @xcite .
they together form a complete set for all the totally symmetric spin - states of @xmath35 systems . when the quadratic term is taken into account
, @xmath8 will remain to be conserved , but @xmath26 will not . in this case
@xmath34 can be used as basis functions for the diagonalization of @xmath25 . in this approach , a powerful tool , the fractional parentage coefficients , that we have developed
previously is used for the calculation of related matrix elements @xcite . using these coefficients ,
a particle ( say , the particle @xmath32 ) can be extracted from @xmath34 as @xmath36 , \label{e02_varthetasm}\end{aligned}\ ] ] where @xmath37 is the spin - state of the particle @xmath32 in component @xmath38 or @xmath39 .
the fractional parentage coefficients have analytical forms as @xmath40(n - s)(s+1 ) } { 2n(2s+1 ) } } c_{1\mu,\ s+1,m-\mu}^{sm } , \\
\label{e04_bsmu } b_{sm\mu}^{\{n\ } } & = & \sqrt{\frac{[1+(-1)^{n - s}]s(n+s+1 ) } { 2n(2s+1 ) } } c_{1\mu,\ s-1,m-\mu } ^{sm},\end{aligned}\ ] ] where the clebsch - gordan coefficients are introduced .
once a particle has been extracted , the calculation of the matrix elements is straight forward , and we have @xmath41 + qq_{s's}^{nm } \ } , \label{e05_vsm}\end{aligned}\ ] ] where @xmath42 + \delta_{s',s-2 } a_{s-2,m\mu}^{\{n\ } } b_{sm\mu}^{\{n\ } } \nonumber \\ & & + \delta_{s',s+2 } a_{sm\mu}^{\{n\ } } b_{s+2,m\mu}^{\{n\}}\}.\end{aligned}\ ] ] for an arbitrary @xmath7 , after a procedure of diagonalization , the set of eigenenergies @xmath43 and the corresponding eigenstates @xmath44 can be obtained , where @xmath43 is in the order of increasing energy .
since the set @xmath34 is complete , the set @xmath45 is exact for @xmath25 .
for the condensates with nonzero spins , it has been shown theoretically that there are various spin - structures . in order to confirm these structures experimentally
, one has to define some measurable physical quantities . for the spatial degrees of freedom
, it is reminded that the one - body density @xmath46 can provide information on the spatial distribution of the particles , and the two - body density @xmath47 can describe the spatial correlation .
similar quantities can be defined in the spin space .
each spin eigenstate can be written as @xmath48 where @xmath49 can be obtained via .
the normality @xmath50 implies that @xmath51-component .
then , we define the one - body probability @xmath52 note that @xmath53 is just the average population of the @xmath54-component , and is an observable which can be directly measured via the stern - gerlach technique . in the case with @xmath55 , @xmath26 is a good quantum number , and the @xmath15-th state has @xmath56 ( if @xmath57 ) , or @xmath58 ( if @xmath59).@xcite in this case , @xmath60 , and we have @xmath61 when one more particle is extracted from the right side of , we have @xmath62 where @xmath63 or @xmath64 otherwise .
then the @xmath15-th state can be rewritten as @xmath65 from the normality as before , we have @xmath66 .
thus , it is straight forward to define the 2-body spin correlative probability as @xmath67 @xmath68 is the probability that the spin of a particle is in @xmath54 while another in @xmath69 when the two are observed , obviously , @xmath70 .
if more particles are extracted , higher order spin correlative probabilities could also be similarly defined .
these probabilities do not have a counterpart in the mft , therefore additional information might be provided by them .
incidentally , the technique for the measurement of the correlative probabilities is mature in particle physics and nuclear physics , but not in condensed matter physics .
the development of related technique is desired . in general
, one can define the correlation coefficient @xmath71 to measure quantitatively how large the correlation is .
if @xmath72 deviates remarkably from @xmath32 , the correlation is strong . whereas if @xmath73 , the correlation is weak , and the system can be well understood simply from the 1-body probabilities .
numerical examples will be given below .
@xmath74 and @xmath75 are used as units for energy and length , respectively , where @xmath76 is the mass of atom . the spatial wave function @xmath17 is obtained under the thomas - fermi approximation , and
thereby we have the strength @xmath77 .
since a slight inaccuracy that may exist in @xmath17 would cause only a slight deviation in the magnitude of @xmath3 , the approximation is acceptable in the qualitative sense .
@xmath78 ( in @xmath79 ) and @xmath80 are in general assumed ( unless particularly specified ) . @xmath81rb and @xmath82na condensates will be used as examples for the @xmath57 and @xmath59 species , respectively . in the units adopted , we have @xmath83 and @xmath84 for @xmath81rb , and @xmath85 and @xmath86 for @xmath82na .
the diagonalization of @xmath25 is straight forward when all the parameters are given .
then , the coefficients @xmath87 can be known . from equations , and , the 1-body and 2-body probabilities can be obtained .
the condensate will fall into its ground state @xmath88 with a specified @xmath8 when @xmath89 .
@xmath8 is determined by how the species is prepared . the case with @xmath90 will be considered later . to evaluate how strong the correlation in @xmath88 would be ,
the correlation coefficients @xmath91 of rb against @xmath1 are shown in ( a ) .
the magnitudes of all the curves are extremely close to one in ( a ) .
this is also true for @xmath92 with @xmath93 .
it implies that the correlation in @xmath88 is very weak .
when @xmath55 , @xmath88 will have @xmath30.@xcite its expression does not depend on the details of interaction but only on the sign of @xmath94 , and can be uniquely written as @xmath95 where @xmath96 implies that the spins of @xmath97 and @xmath98 are coupled to spin @xmath99 , and so on .
the special way of spin coupling in ( i.e. , the combined spin of an arbitrary group of @xmath100 particles is @xmath100 ) assure that the state is normalized and symmetrized . where all the spins are roughly aligned along a common direction , but the azimuthal angle of this direction is arbitrary . from the analytical forms of @xmath101 can be derived as @xmath102 can be approximately rewritten as @xmath103 .
thus , if @xmath104 , half of the particles would have @xmath38 . if @xmath105 , the increase of @xmath8 would lead to a decrease of @xmath106 .
in particular , @xmath107 when @xmath108 as expected .
incidentally , the above analytical forms of @xmath101 are identical to those from the mft when @xmath109.@xcite when @xmath1 varies , the one - body probabilities @xmath101 of @xmath88 against @xmath1 are calculated and shown in .
the left ends of the curves coincide exactly with those from the above analytical expressions .
note that the @xmath110 particles will gain additional energy from the quadratic zeeman term . in order to reduce the energy
the number of @xmath38 particles will increase with @xmath1 .
this is clearly shown in . for @xmath104 as shown in a , the derivative of @xmath111 varies very swift from a positive value to zero at the vicinity of a turning point @xmath112 where @xmath113 , i , e , the number of @xmath38 particles @xmath114 is very close to its limit @xmath7 .
the swift variation at the turning point will become a sudden transition when @xmath7 is large , and the point becomes a critical point @xmath115 . passing through this point , the derivative varies abruptly , but the spin - state @xmath116 itself varies continuously . when @xmath117 , the ground state remains unchanged as @xmath118 , where @xmath119 is a fock - state with @xmath120 , @xmath121 , and @xmath122 particles in @xmath123 , 0 , and -1 , respectively . when @xmath105 , not all the particles can be changed to @xmath38 because of the conservation of @xmath8 .
accordingly , the increase of @xmath106 is hindered and the curves in b are smoother . when @xmath8 is large , the curves are very flat implying that @xmath88 is affected by @xmath1 very weakly .
when @xmath124 , we found that @xmath125 .
accordingly , @xmath126 , @xmath127 , and @xmath128 . in this way , @xmath129 is maximized so that the quadratic zeeman energy is minimized . in conclusion of this section for small condensates of rb ,
the ground state @xmath130 is continuously changed from @xmath131 to @xmath132 without a transition when @xmath1 increases . however , the derivative @xmath133 undergoes a transition at @xmath115 when @xmath104 .
it is recalled that the mft , which is correct when @xmath7 is very large , has predicted the transitions between the ferromagnetic , broken - axisymmetry , and the polar phases @xcite .
the broken - axisymmetry phase is caused by a rapid quenched field , therefore it is not expected to appear in our case with @xmath1 varying adiabatically .. contrary to the previous case , the spin correlation is very strong in the ground states @xmath130 of @xmath59 condensates as shown by @xmath91 in b , where the left ends of all the curves are remarkably larger than @xmath32 .
it implies that , when two particles are observed , the probability of a particle in @xmath38 would be much larger if the other one has @xmath134 .
starting from @xmath55 there is a domain of @xmath1 where @xmath91 remains large and nearly unchanged ( refer to the @xmath135 curve in b ) . outside the domain ,
the correlation vanish rapidly .
the domain is called a strong correlation domain ( scd ) , where the spin - structure at @xmath55 will keep unchanged against the increase of @xmath1 .
the scd does not have a clear border .
for each curve in b , let @xmath136 be the value where @xmath137/2 $ ] .
then , @xmath136 is considered as the outward border of the scd . a larger @xmath8 will lead to a larger @xmath136 ( say , for b , if @xmath138 , @xmath139 , and @xmath140 , respectively , @xmath141 , @xmath142 , and @xmath143 ) .
thus the spin structure at @xmath55 will better keep when @xmath8 is large .
when @xmath55 , the ground state will have @xmath26 as small as possible.@xcite thus , @xmath144 if @xmath145 is even , or @xmath146 if @xmath145 is odd . due to the fact that @xmath147 is unique , we have @xmath148^{(n - m)/2 } \ } , \label{e20_theta1m}\ ] ] where @xmath149 is the symmetrizer , i.e. , a summation over the @xmath150 particle permutations .
so this state is composed of a group of @xmath123 particles together with a group of singlet pairs . and @xmath151^{(n - m-1)/2 } \ } , \label{e20_varthetam1mn}\ ] ] which is composed of a group of @xmath152 polarized particles ( but the direction of polarization deviates a little from the z - axis ) together with a group of singlet pairs . from the one - body probabilities are @xmath153 for the case @xmath104 and @xmath7 being even , @xmath154^{n/2}$ ] .
this state is named the polar state where all the particles are in the singlet pairs . from equations . and
, the polar state has @xmath155 for all @xmath54 .
in other words , the particles are equally populated among the three components .
this is a common property of @xmath156 states .
when @xmath1 increases , @xmath101 against @xmath1 are shown in .
since the ground state will have more and more @xmath38 particles , @xmath111 increases sharply from @xmath157 to @xmath158 as shown in a. correspondingly , the ground state @xmath159 is sharply transformed from @xmath160 to @xmath161 . comparing the curve of @xmath111 in a with
the one in a for rb , the rise of the former is much faster than the latter .
thus the na condensate is highly sensitive to the appearance of @xmath1 if @xmath104 .
a very weak field ( a few @xmath162 ) is sufficient to break nearly all the pairs and turn every spin lying on the @xmath163-@xmath164 plane independently .
the independence is supported by the curve of @xmath165 in b , where @xmath166 except @xmath1 is close to zero .
however , the high sensitivity will be lost when @xmath167 . in b
the curve of @xmath168 remains to be horizontal when @xmath169 ( refer also to the curve with @xmath135 in b ) .
thus the spin - structure inside the scd might remain unchanged .
to clarify , the 2-body probabilities are calculated and shown in , where all the @xmath170 remain unchanged in the scd .
thus the invariance of the structure against @xmath1 is strongly supported .
the invariance can also be seen by observing the overlap @xmath171 against @xmath1 ( @xmath7 and @xmath8 are assumed to be even , a superscript @xmath1 is added to emphasize the dependence on @xmath1 ) . with the parameters of and with @xmath135 ,
this overlap is @xmath172 when @xmath173 . thus the invariance is directly confirmed .
however , it decreases very fast when @xmath1 is close to @xmath136 , and is equal to @xmath174 when @xmath175 .
afterward , it tends to zero rapidly when @xmath1 is further larger .
the existence of the scd demonstrates that the mixture of a group of singlet pairs together with a group of @xmath8 unpaired particles ( each has @xmath176 ) is capable to keep its structure against @xmath1 .
however , the capability will be lost when @xmath1 is close to or @xmath177 . afterward
the mixture will begin to change .
note that the change is characterized by the increasing of @xmath38 particles , which come from the breaking of pairs .
therefore the change would be less probable if the original number of pairs is small .
thus a larger @xmath8 ( implying a fewer original pairs ) will lead to a better stability and therefore a larger @xmath136 .
a larger @xmath94 will also lead to a better stability and therefore a larger @xmath136 ( say , if @xmath94 is one time larger than the experimental value of @xmath178 , then the @xmath136 would be enlarged from @xmath142 to @xmath179 for the curve with @xmath135 in b ) . since @xmath8 affects the stability , @xmath101 is in general sensitive to @xmath8
. if @xmath1 is weak , the sensitivity would be very high when @xmath8 is small .
e.g. , it is shown in b that @xmath106 will decrease dramatically at @xmath180 when @xmath181 is simply changed from @xmath33 to @xmath182 . when @xmath124 , all the pairs will be destroyed and @xmath130 will tend to @xmath132 .
thus , disregarding @xmath183 or @xmath184 , both species tend to the same structure . in order to understand the sensitivity of the ground states @xmath185 against @xmath1 quantitatively , the fidelity susceptibility @xcite .
@xmath186 is calculated and given in .
this figure demonstrates that each species has its own region highly sensitive to @xmath1 . for rb
, the most sensitive region is surrounding the critical point @xmath115 where the increase of @xmath114 stops suddenly ( refer to a ) . when @xmath1 is small , the sensitivity does not depend on @xmath8 and is in general very weak .
whereas for na , the most sensitive region is surrounding the border of scd , @xmath136 , where @xmath114 begin to increase suddenly ( refer to b ) . the fact that @xmath136 will become larger with @xmath8
is clearly shown in b. the sensitivity can be very high when @xmath1 is small if @xmath8 is also small .
since the linear zeeman term does not affect the spin - structures , the ratio @xmath187 involved in @xmath25 is crucial .
this quantity is proportional to @xmath188 .
therefore , a larger @xmath11 would reduce the effect of @xmath1 .
consequently , for @xmath189 , all the curves plotted in to will extend horizontally to the right by a common factor @xmath190 .
in particular , the scd will become larger ( say , @xmath136 of the ground states @xmath191 of na would increase from @xmath142 to @xmath192 if @xmath11 is from @xmath193 to @xmath194 ) .
on the other hand , the change of @xmath7 causes not simply a change of scale , because the numbers of degrees of freedom are thereby changed . for an example , if @xmath80 , @xmath139 , and @xmath195 , @xmath111 of rb would be @xmath196 when @xmath197 , @xmath198 , and @xmath199 , respectively .
thus , the curve of @xmath111 in a would rise up faster if @xmath7 decreases from 1000 . on the contrary ,
if @xmath80 , @xmath139 , and @xmath195 , @xmath111 of na would be @xmath200 when @xmath201 , @xmath202 , and @xmath203 , respectively .
thus , the curve of @xmath111 in a would rise up slower if @xmath7 decreases from 1000 .
thus , the effect of @xmath7 on @xmath57 and @xmath184 species is different .
the effect of @xmath7 on @xmath106 with @xmath105 is shown in .
the curve with @xmath80 in a is identical to the curve with @xmath135 in b for @xmath57 . in a for rb the curves with a smaller @xmath7 will rise up faster against @xmath1 just as the previous case with @xmath104 . for @xmath59 ,
the curve with @xmath80 in b is identical to the curve with @xmath135 in b. similar to the case with @xmath104 , a smaller @xmath7 will cause also a smoother change . in particular , the abrupt change appearing in the vicinity of @xmath136 as shown in b will disappear when @xmath7 is small .
since the level density in a condensate is usually dense , thermo - fluctuation is in general not negligible .
what actually measured is the weighted probabilities @xmath204 and @xmath205 , where @xmath206 is the weight , @xmath43 the energy , @xmath10 the temperature , and the summation in principle runs over all the states . however , when @xmath207 , the contribution arises essentially from the ground band , and all the higher states can be neglected.@xcite the members of the ground band have nearly the same spatial wave functions , and their spin - states together with @xmath43 can be obtained via the diagonalization of @xmath25 .
as an example , @xmath208 of the condensates with @xmath57 at @xmath209 is plotted in a. in order to see the effect of interaction , @xmath94 is given at three values .
when @xmath1 is very small ( @xmath210 ) or sufficiently large ( @xmath211 ) , the three curves associated with different values of @xmath94 overlap nearly .
however , in between , they are rather sensitive to @xmath212 as shown in a .. to see clearer , both @xmath208 and @xmath213 against @xmath94 under @xmath214 are plotted in b. the figure demonstrates that the weighted 2-body probabilities are very sensitive to @xmath94 .
furthermore , the patterns of the curves can be tuned by altering @xmath1 .
therefore , the measurement of the weighted probabilities under various @xmath1 can provide rich information on the parameters of interaction .
one more example for the case @xmath59 is shown in .
similar to the previous case , high sensitivity to the interaction is found . in a ,
the left ends of the curves are flat .
this is caused by the existence of the scd , where the spin - structure remains nearly unchanged .
we have studied the spin structures of small spin-1 condensates ( @xmath0 ) under a magnetic field @xmath1 .
the theory is beyond the mft , and the single mode approximation has been adopted .
the fractional parentage coefficients have been used as a tool for the analytical derivation .
the 1-body , 2-body , and weighted probabilities are defined and calculated to extract information from the spin eigenstates .
the correlation coefficients @xmath215 and the fidelity susceptibility @xmath216 have also been calculated .
the following results are mentioned ( @xmath145 is assumed to be even for simplicity ) .
\(i ) when @xmath55 , the ground state @xmath130 is either the ferromagnetic state @xmath131 if @xmath57 , or is the polar state @xmath217 if @xmath59 . for @xmath131 ,
all the @xmath7 spins are coupled to total spin @xmath30 , @xmath218 is close to 1 and @xmath219 is very small ( and ) , therefore the ferromagnetic state does not have spin - correlation , and is inert to @xmath1 when @xmath1 is small .
the state @xmath217 has @xmath8 particles in @xmath220 together with @xmath221 singlet pairs , its @xmath91 is very large .
thus the polar state contains strong spin - correlation .
furthermore , its @xmath222 is very large when @xmath8 is small .
it implies a high sensitivity against the appearance of @xmath1 .
\(ii ) when @xmath1 increases , the number of @xmath38 particles @xmath114 in @xmath130 will in general increase so as to reduce the quadratic zeeman energy . for @xmath57 , @xmath130
will be changed from @xmath131 to @xmath132 when @xmath1 is from 0 to @xmath223 .
the change goes on continuously , no transition in spin - structure occur . in accord with the change of @xmath130
, @xmath114 is changed from @xmath224 ( refer to ) to @xmath145 .
therefore , @xmath225 , which is the maximal number of particles allowed to be changed from being @xmath110 to @xmath38 .
when @xmath8 is close to @xmath7 , @xmath226 is very small implying that the room left for changing is very small , therefore @xmath130 is inert to @xmath1 ( refer to b where the curve with @xmath227 is very flat ) .
when @xmath8 is smaller , @xmath130 has much room for changing and therefore would be more sensitive to @xmath1 .
in particular , when @xmath104 , there is a critical point @xmath115 .
once @xmath228@xmath229 and the ground state varies with @xmath1 no more .
there is an abrupt change in the derivative @xmath133 at @xmath115 .
thus , although the spin - structures vary continuously with @xmath1 , the related derivatives might not .
@xmath115 is equal to @xmath230 in
. it will be smaller when @xmath7 decreases , and larger when @xmath11 increases .
\(iii ) for @xmath59 , the increase of @xmath1 from 0 to @xmath223 causes a change of @xmath130 from @xmath231 ( or @xmath232 ) to @xmath132 .
@xmath130 of both cases @xmath57 and @xmath184 tend to the same state because it is the most advantageous state for reducing the quadratic zeeman energy .
the change of @xmath130 goes on also continuously without transitions . in this process
@xmath114 is in general increasing via a mechanism , i.e. , a breaking of pairs as @xmath233 ( whereas the process @xmath234 is suppressed under @xmath1 because it causes an increase of quadratic zeeman energy ) .
it is recalled that the particle correlation in @xmath130 is very weak when @xmath57 , but strong when @xmath59 due to the formation of pairs .
when all the particles are paired ( i.e. , @xmath104 ) , the structure is extremely sensitive to @xmath1 . a very weak @xmath1 ( a few @xmath162 )
is sufficient to break all the pairs . for a comparison , @xmath111 in a for
rb will be equal to 0.9 when @xmath235 , but only @xmath236 in a for na .
however , when unpaired particles emerge ( i.e. , @xmath105 ) , the pairs will have an additional ability to keep themselves .
the mechanism underlying this phenomenon deserves to be studied further .
this leads to the appearance of scd ranging from @xmath55 to @xmath136 .
a larger @xmath8 will lead to a larger @xmath136 , while @xmath237 when @xmath104 .
the spin - structure will remain unchanged when @xmath238 , but @xmath114 will begin to increase when @xmath239 .
the derivative of the probabilities varies very swiftly in the vicinity of @xmath136 .
the swift variation will become a sudden jump when @xmath7 is large ( refer to b ) .
thus @xmath136 is also a critical point when @xmath7 is large . as a numerical example ,
when @xmath80 and @xmath135 , @xmath240 in b. similar to @xmath115 , @xmath136 will become smaller when @xmath7 decreases ( with @xmath181 remaining unchanged ) , and will become larger when @xmath11 increases .
\(iv ) the sensitivity of the ground states against @xmath1 is quantitatively shown in .
\(v ) when @xmath1 is appropriately chosen and @xmath10 is sufficiently low , the measurable weighted probabilities may provide rich information on the parameters of interaction .
the suggestion on the calculation of the fidelity and the comment from prof .
d. x. yao is gratefully appreciated .
the support from the nsfc under the grant 10874249 is also appreciated .
99 sadler l e , higbie j m , leslie s r , vengalattore m , and stamper - kurn d m 2006 nature ( london ) * 443 * 312 . | the spin structures of small spin-1 condensates ( @xmath0 ) under a magnetic field @xmath1 has been studied beyond the mean field theory ( mft ) . instead of the spinors ,
the many body spin - eigenstates have been obtained .
we have defined and calculated the spin correlative probabilities to extract information from these eigenstates .
the correlation coefficients and the fidelity susceptibility have also been calculated .
thereby the details of the spin - structures responding to the variation of @xmath1 can be better understood .
in particular , from the correlation coefficients which is the ratio of the 2-body probability to the product of two 1-body probabilities , strong correlation domains ( scd ) of @xmath1 are found .
the emphasis is placed on the sensitivity of the condensates against @xmath1 .
no phase transitions in spin - structures are found .
however , abrupt changes in the derivatives of observables ( correlative probabilities ) are found in some particular domains of @xmath1 . in these domains
the condensates are highly sensitive to @xmath1 .
the effect of temperature is considered .
the probabilities defined in the paper can work as a bridge to relate theories and experiments .
therefore , they can be used to discriminate various spin - structures and refine the interactions . |
we wish to thank the anonymous reviewer for his valuable comments that helped in improving the quality of the article .
our work is supported by conicyt under the anillo project act1122 .
a. h. guth , phys .
d * 23 * ( 1981 ) 347 ; + d. h. lyth and a. riotto , phys .
* 314 * ( 1999 ) 1 [ arxiv : hep - ph/9807278 ] ; + n. straumann , annalen phys .
* 15 * ( 2006 ) 701 [ arxiv : hep - ph/0505249 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys .
j. suppl .
* 192 * ( 2011 ) 18 [ arxiv:1001.4538 [ astro-ph.co ] ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1303.5062 [ astro-ph.co ] ; + p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1303.5082 [ astro-ph.co ] .
f. englert and r. brout , phys . rev
. lett .
* 13 * ( 1964 ) 321 ; + p. w. higgs , phys .
* 13 * ( 1964 ) 508 .
g. aad _ et al .
_ [ atlas collaboration ] , phys .
b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] ; + s. chatrchyan _ et al . _
[ cms collaboration ] , phys . lett .
b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] .
f. l. bezrukov , a. magnin and m. shaposhnikov , phys .
b * 675 * ( 2009 ) 88 [ arxiv:0812.4950 [ hep - ph ] ] ; + f. bezrukov and m. shaposhnikov , jhep * 0907 * ( 2009 ) 089 [ arxiv:0904.1537 [ hep - ph ] ] ; + a. salvio , phys .
b * 727 * ( 2013 ) 234 [ arxiv:1308.2244 [ hep - ph ] ] ; + d. p. george , s. mooij and m. postma , arxiv:1310.2157 [ hep - th ] .
c. p. burgess , h. m. lee and m. trott , jhep * 0909 * ( 2009 ) 103 [ arxiv:0902.4465 [ hep - ph ] ] ; + j. l. f. barbon and j. r. espinosa , phys .
d * 79 * ( 2009 ) 081302 [ arxiv:0903.0355 [ hep - ph ] ] .
s. r. coleman and e. j. weinberg , phys .
d * 7 * ( 1973 ) 1888 .
e. j. chun , s. jung and h. m. lee , phys .
b * 725 * ( 2013 ) 158 [ arxiv:1304.5815 [ hep - ph ] ] .
m. hashimoto , s. iso and y. orikasa , arxiv:1310.4304 [ hep - ph ] .
a. d. linde , phys .
b * 108 * ( 1982 ) 389 ; + a. albrecht and p. j. steinhardt , phys . rev .
* 48 * ( 1982 ) 1220 .
g. barenboim , e. j. chun and h. m. lee , arxiv:1309.1695 [ hep - ph ] . m. u. rehman , q. shafi and j. r. wickman , phys .
d * 78 * ( 2008 ) 123516 [ arxiv:0810.3625 [ hep - ph ] ] . c. destri , h. j. de vega and n. g. sanchez , phys .
d * 77 * ( 2008 ) 043509 [ astro - ph/0703417 ] .
v. faraoni , phys .
d * 53 * ( 1996 ) 6813 [ astro - ph/9602111 ] .
a. d. linde , phys .
b * 114 * ( 1982 ) 431 .
d. z. freedman , i. j. muzinich and e. j. weinberg , annals phys .
* 87 * ( 1974 ) 95 ; + d. z. freedman and e. j. weinberg , annals phys .
* 87 * ( 1974 ) 354 .
t. futamase and k. -i .
maeda , phys .
d * 39 * ( 1989 ) 399 ; + n. makino and m. sasaki , prog .
* 86 * ( 1991 ) 103 ; + d. i. kaiser , [ astro - ph/9507048 ] ; + d. i. kaiser , phys .
d * 52 * ( 1995 ) 4295 [ astro - ph/9408044 ] ; v. faraoni , int .
j. theor .
* 40 * ( 2001 ) 2259 [ hep - th/0009053 ] .
d. boyanovsky , h. j. de vega and n. g. sanchez , phys .
d * 72 * ( 2005 ) 103006 [ astro - ph/0507596 ] ; + d. boyanovsky , c. destri , h. j. de vega and n. g. sanchez , int .
j. mod .
phys . a * 24 * ( 2009 ) 3669 [ arxiv:0901.0549 [ astro-ph.co ] ] ; + c. destri , h. j. de vega and n. g. sanchez , annals phys .
* 326 * ( 2011 ) 578 [ arxiv:0906.4102 [ astro-ph.co ] ] . w. h. kinney , e. w. kolb , a. melchiorri and a. riotto , phys .
d * 74 * ( 2006 ) 023502 [ astro - ph/0605338 ] .
n. okada , m. u. rehman and q. shafi , phys .
d * 82 * ( 2010 ) 043502 [ arxiv:1005.5161 [ hep - ph ] ] . | in the present work we study gut coleman - weinberg inflation with a nonminimal coupling to gravity . in this kind of model one
usually finds that either the nonminimal coupling to gravity is large @xmath0 or the inflaton self - coupling is unnaturally small @xmath1 .
we have shown that the model is in agreement with the recent results from planck for natural values of the couplings .
inflation @xcite has become the standard paradigm for the early universe , because it solves some outstanding problems present in the standard hot big - bang cosmology , such as the flatness and horizon problems , the problem of unwanted relics , such as magnetic monopoles , and produces the cosmological fluctuations for the formation of the structure that we observe today .
the spectacular cmb data , first from the wmap satellite @xcite and recently from planck @xcite , have strengthen the inflationary idea , since the observations indicate an _ almost _ scale - free spectrum of gaussian adiabatic density fluctuations , just as predicted by simple models of inflation . however , inflation is not a theory yet , as we do nt know how to integrate it with ideas from particle physics .
the brout - englert - higgs mechanism @xcite in the framework of the standard model breaks electroweak symmetry , gives masses to the charged fermions and the massive gauge bosons , and predicts the existence of the higgs boson .
the recent lhc discovery of the higgs boson @xcite completed the particle spectrum predicted by the standard model , and indicates for the first time that fundamental scalar particles exist in nature .
therefore , it is a natural thing to assume that inflation is driven by the higgs boson .
although the higgs potential is not suitable for a viable inflationary model , the presence of a higgs non - minimal coupling to gravity can change things to the better .
a large value of the non - minimal coupling @xmath2 is required , and given the uncertainties in the top quark mass and the strong coupling constant the model is still allowed @xcite .
sadly , that large value of @xmath3 questions the validity of the scenario in the sm higgs inflation , as the inflationary scale exceeds the effective ultraviolet cut - off scale @xcite . as it was realized long ago
, radiative corrections can be the origin of electroweak symmetry breaking @xcite .
the coleman - weinberg mechanism can not work in the standard model due to the large value of the top yukawa coupling , but it can work in models beyond the standard model @xcite .
this is an interesting possibility given the higgs naturalness problem , in the following sense .
if heavy particles are coupled to the higgs boson , like in gut models , then the higgs receives large radiative corrections that bring its mass close to the gut mass scale .
supersymmetry at the tev scale can solve the problem , but given the severe experimental constraints on the masses of the superpartners , it was proposed recently the flatland scenario @xcite , according to which electroweak symmetry is broken radiatively a la coleman - weinberg in the infrared region starting from a flat scalar potential in the ultraviolet region .
regarding inflation , the coleman - weinberg type of potential is a simple and well motivated one , since it naturally arises when loop corrections are taken into account , and it is typical for the new inflation scenario @xcite where inflation takes place near the maximum .
recently it has been studied in @xcite in a b - l extension of the standard model , and a few years ago in @xcite in a gut inflationary model . as shown in @xcite in the context of the effective theory of inflation , cosmic microwave background together with large scale structure data
prefer double - well inflaton potentials .
the 2013 planck data confirmed this result , and not surprisingly the coleman - weinberg potential considered here belongs to this class , and succeeds to reproduce the @xmath4 value and the r bound from the 2013 planck release .
however , in @xcite the inflaton was minimally coupled to gravity . setting @xmath5 , although it is a popular choice , is often unacceptable as was pointed out in @xcite .
non - minimal couplings are generated by quantum corrections even if they are absent in the classical action @xcite , and as a matter of fact the coupling is required if the scalar field theory is to be renormalizable in a classical gravitational background @xcite . for early works on non - minimal
inflation see for example @xcite and references therein .
in the present brief report we imagine a scenario where inflation is driven by the scalar sector of some particle physics model beyond the standard model in which electroweak symmetry breaking takes place radiatively a la coleman and weinberg . since the scale of inflation is close to the gut scale
, we consider a gut model rather than a low energy one , and we also allow for a non - vanishing non - minimal coupling , as required by the quantum corrections already needed to give rise to the coleman - weinberg type of the inflaton potential .
we find that the model is viable as it is in agreement with the recent data from planck , and even more importantly for natural values of the parameters of the model .
we start defining the model by the action @xmath6 where @xmath7 is the non - minimal coupling and @xmath8 with a coleman - weinberg type of potential of the standard form @xcite @xmath9 where a is related to the inflaton quartic self - coupling @xcite , and m is the inflaton vacuum expectation value ( vev ) at the minimum , @xmath10 .
the vacuum energy density at the origin is given by the constant term @xmath11 so that @xmath12 , and the shape of the potential can be seen in figure 1 , and the inflaton mass is given by @xmath13 .
at this point we stress the fact that the coleman - weinberg potential considered in the present work is the standard one obtained from one - loop corrections in minkowski spacetime .
the one - loop corrections in de sitter spacetime ( which approximates an inflationary background ) lead to a very different effective potential as shown in @xcite . [ cols="^,^ " , ] in summary , in this brief report we have considered a gut inflationary model where inflation is driven by the scalar sector of the model with a coleman - weinberg type of potential for the inflaton and a non - minimal coupling of the inflaton to gravity .
as is known , at tree level the @xmath14 potential can not trigger electroweak symmetry breaking , and in addition the corresponding chaotic inflationary model @xmath15 is ruled out by the wmap and planck data .
however , quantum loop corrections modify the inflaton potential and at the same time generate the non - minimal coupling term . the model is characterized by two dimensionless parameters , namely the non - minimal coupling @xmath7 and the inflaton quartic self - coupling @xmath16 .
we have shown that the model leads to predictions in agreement with the 95 per cent cl contours in the r-@xmath4 plane for natural values of the parameters of the model , both for @xmath7 and @xmath16 of the order of @xmath17 . |
in the search for physics beyond the standard model , the observation of @xmath0 violation could be a very usefull tool .
many models shows @xmath0 nonconservation effects , as , for example , the superweak model @xcite , the kobayashi - maskawa model @xcite , the penguin model @xcite or the weinberg - higgs model @xcite . today
, there are three systems where @xmath0 violation has been observed .
the first one was the @xmath10 decay , where it was shown @xcite that @xmath11 is not a pure eigenstate of @xmath0 , and then , the parameter @xmath12 is nonzero .
later , direct @xmath0 violation in @xmath13 decays has been observed .
very recently , @xmath0 violation has also been observed in the @xmath14 and other related modes ( for a review , see @xcite ) . in 1957 , okubo @xcite noted that @xmath0 violation could cause differences in the branching ratios of the @xmath15 and @xmath16 decays .
pais @xcite extended this proposal also to @xmath17 and @xmath18 decays .
in more recent works the @xmath0 violation signs were also investigated using nonleptonic hyperon decays @xcite-@xcite where @xmath6 decays were stiudied too . in @xcite
the study has been made in the framework of the standard model , and in @xcite , @xcite , new physics was considered . at the experimental level , there are experiments searching for @xmath0 violation in hyperon decays @xcite .
the nonleptonic @xmath19 decays has only been studied in @xcite where the the strong phases were estimated at leading order in heavy - baryon chiral perturbation theory .
the aim of this paper is to calculate the @xmath20 strong phases using an effective lagrangian model , without the heavy - baryon approximation , and their effects to the asymmetry parameters in the @xmath21 decays .
the size of the @xmath0 violation depends on the final - state strong interaction between the produced particles .
so , in order to perform the calculations , the strong phase shifts are needed .
at the moment , the situation of the @xmath22 and @xmath23 decays is very confortable , the @xmath24 strong phase shift analysis is very well known experimentally @xcite . at the theoretical level , this system is very well described , and , at least at low energies , the chiral perturbation theory is very precise @xcite , @xcite .
however , in the decays that produce hyperons , the situation is not so good , no experimental data is available to the @xmath25 interactions .
in fact , some information can be obtained in the study of hyperonic atoms ( see , for example , @xcite , where the @xmath26 coupling constant is estimated ) , but it is not enough to fully understand the @xmath27 interaction . thus , to investigate the @xmath25 interactions , the only way is to use a model .
as it was said , the chiral perturbation theory is very accurate when applied to the @xmath24 interactions , so , we hope that it works in the @xmath25 system too . in @xcite-@xcite , @xcite
that was done to the @xmath28 decay , and the calculated phases were very small . in this work , the @xmath29 is studied , and the strong phase shifts for @xmath30 interactions are calculated using the model presented in @xcite . in @xcite ,
chiral lagrangians are used , describing processes with @xmath6 , @xmath7 and @xmath8 in the intermediate states .
the @xmath9-term is also included , but not only as a parametrization ( as it was done in @xcite ) , but relating it with the scalar form factor @xmath31 , based in the results of @xcite , @xcite .
this paper will show the following contents : in section ii it will be shown the @xmath21 decay and how to calculate the observables . in section iii
we will calculate the phase shifts in the @xmath5 interactions .
the results and conclusions are in section iv .
in the @xmath21 ( @xmath33 ) decays , the transitions are of the form 3/2 0 + spin 1/2 , and the contributing phases are the @xmath34 @xmath3(parity conserving ) and @xmath4(parity violating ) waves in the @xmath35 rest frame .
the @xmath36 @xmath37 nonleptonic decays are @xmath38 and @xmath39 .
the experimental observables are the total rate @xmath40 , and the asymmetry parameters , that can be written as = 2 re(p^*d)/(|p|^2+|d|^2 ) = 2 im(p^*d)/(|p|^2+|d|^2 ) = ( |p|^2-|d|^2)/(|p|^2+|d|^2 ) and obeys the relation ^2+^2+^2=1
. for antihyperons decays the expressions are = 2 re(p^ * d)/ ( |p|^2+|d|^2 ) = 2 im(p^ * d)/ ( |p|^2+|d|^2 ) .
the @xmath3 and @xmath4 amplitudes can be parametrized as p=_ia_2i|p_2i|e^i(_p^2i+_p^2i ) [ 2.12 ] d=_ia_2i|d_2i|e^i(_d^2i+_d^2i ) , [ 2.13 ] where @xmath41 is the isospin state , @xmath42 are the strong phase shifts and @xmath43 are the weak @xmath0 violating phases .
the respective @xmath0 conjugated amplitudes are = _
ia_2i|p_2i|e^i(_p^2i-_p^2i ) = -_ia_2i|d_2i|e^i(_d^2i-_d^2i ) , using eq . (
[ 2.12 ] ) , ( [ 2.13 ] ) for the @xmath38 decay we have p(_-^-)=-p_1e^i(_p^1+_p^1)+ p_3e^i(_p^3+_p^3 ) d(_-^-)=-d_1e^i(_d^1+_d^1)+ d_3e^i(_d^3+_d^3 ) and for @xmath39 p(_0 ^ -)=p_1e^i(_p^1+_p^1)+ p_3e^i(_p^3+_p^3 ) d(_0 ^ -)=d_1e^i(_d^1+_d^1)+ d_3e^i(_d^3+_d^3 ) . in the limit of @xmath0 conservation , the @xmath0 asymmetry parameters a=+- and b=+- vanish , since @xmath44=-@xmath45 and @xmath46=-@xmath47 . in hyperon decays , the @xmath48=3/2 amplitudes are much smaller then the @xmath48=1/2 , then , in the first order in @xmath48=3/2 amplitudes , a(_-^-)&=&-tan(_p^1-_d^1)tan(_p^1-_d^1 ) \ { 1 + + & & + 1p_3p_1 . + & & . - ( _ p^3-_d^1)(_p^3-_d^1)(_p^1-_d^1)(_p^1-_d^1 ) + & & + 1d_3d_1 .
-(_p^3-_d^1)(_p^3-_d^1)(_p^1-_d^1)(_p^1-_d^1 ) } [ 2.10 ] + b(_-^-)&=&cot(_p^1-_d^1)tan(_p^1-_d^1 ) \ { 1 + + & & + 1p_3p_1 .
-(_p^3-_d^1)(_p^3-_d^1)(_p^1-_d^1)(_p^1-_d^1 ) + & & + 1d_3d_1 . + & & .
-(_p^3-_d^1)(_p^3-_d^1)(_p^1-_d^1)(_p^1-_d^1 ) } [ 2.11 ] and similar expressions for the @xmath49 decays ( replacing the factors @xmath50 for @xmath51 in the expressions ( [ 2.10 ] ) , ( [ 2.11 ] ) ) . at leading order , a(_-^-)=a(_-^0)= -tan(_p^1-_d^1)tan(_p^1-_d^1 ) ,
[ 1.30 ] and b(_-^-)=b(_-^0)= cot(_p^1-_d^1)tan(_p^1-_d^1 ) .
[ 1.31 ] in the next section we will calculate the phase shifts @xmath42 , that are needed to estimate @xmath52 and @xmath53 .
in order to describe the low energy @xmath5 interaction , a reliable way is to use effective lagrangians , as it was done in a previous work @xcite .
a very important feature of this model is to allow the inclusion of spin 3/2 ressonances in the intermediate states . in the low energy @xmath55 interactons , for example , the @xmath56(1232 ) dominates almost completely the total cross section ( when @xmath57 is near the @xmath56 mass ) .
consequently , it is expected that in some reactions of @xmath25 scattering , the same behaviour will occur @xcite .
the lagrangians to be considered are @xmath58 where @xmath6 , @xmath7 , @xmath59 and @xmath60 are the cascade , the resonance @xmath7(1530 ) , the pion and the rho fields .
@xmath61 is the off - shell parameter @xcite , @xmath62 and @xmath63 are the magnetic moments .
the lagrangians are almost the same as the @xmath24 ones @xcite , because the @xmath24 system is very similar to the @xmath30 .
@xmath64 and @xmath6 are particles with isospin 1/2 , the only difference is that @xmath56(1232 ) has isospin 3/2 and @xmath7(1530 ) , 1/2 , so , a @xmath65 matrix is included in ( [ 3.1 ] ) .
the spin 3/2 propagator for a mass @xmath66 particle , is then g^(p ) & = & - g^ - .
+ & & . - + - . [ 2.3 ] = 80.mm the contributing diagrams are shown in the fig . 1 ( we show only the direct diagrams , but in the calculations , the crossed diagrams are also included )
. the scattering matrix will have the general form t_^ba&= & ( ) a^+ + ( k + k)2b^+_ba + & & + a^- + ( k + k)2a^- i_bac^cu(p ) , where @xmath67 and @xmath68 are the initial and final @xmath69 momenta . calculating the amplitudes from the diagrams , the contributions from fig .
1(a ) ( intermediate @xmath6 ) are @xmath70 fig 1(d ) , the @xmath8 exchange , gives @xmath71 the contribution from fig .
1(b ) , the interaction with the intermediate @xmath7 , is @xmath72 with @xmath73 where @xmath74 is the pion mass , and @xmath75 and @xmath76 are defined in the appendix a. one must remark that eq .
( [ 3.2 ] ) is different from the expression of @xcite , where there was a mistake .
the correct expression is presented here . in @xcite ,
@xcite the @xmath9 term ( diagram 1.c ) was simply considered as a parametrization @xmath77 in fact , the @xmath9 term represents the exchange of a scalar isoscalar system in the @xmath78-channel .
this contribution is related to the scalar form factor of the baryon , and at large distances is dominated by triangle diagrams ( figure 2 ) involving the exchange of 2 pions @xcite . in the @xmath30 interaction , this contribution is associated with two triangle diagrams , with @xmath6 and @xmath7 intermediate states , as it was calculated in @xcite .
= 90.mm the scalar form factor for a spin @xmath79 baryon b
is defined as @xmath80 , where @xmath81 is the chiral symmetry breaking lagrangian .
the contribution of an intermediate particle of spin @xmath82 and mass @xmath66 to the scalar form factor is given by & & _
s(t;m ) u = - ^2 ^2 t_a^t_a + & & , [ 2.4 ] where @xmath83 is the average of the internal pion momenta , @xmath84 is a vertex isospin matrix and & & [ _ 1/2u ] = -(m+m ) + .
+ 1 + u , [ 2.6 ] + & & [ _ 3/2 u ] = - ( m+m ) ( ^2-t/2 ) ... + & & ... -^4 .. + & & .. + -1 ( m+m ) ( 2 m - m ) + 2 ^ 2 .. + & & .. -+ ( ^2-t/2 ) + ( m+m ) ... + & & ... - - + -1 .. + & & .. m^2 + 2 m m - m^2 + 2 ^ 2 - u [ 2.7 ] . calculating @xmath31 in the @xmath5 interaction , and using the loop integrals
@xmath85 defined in appendix b , we obtain & & _
1/2(t ) = ^2 ( m_+m_^ * ) _cc^(000)- . + & & .
_c^(000 ) - _c^(001 ) , [ 2.8 ] + & & _
3/2(t ) = ^2 - ( m_+m_^*)^2 .. + & & .. ( 2m_^*-m _ ) + 2 ^2 ( m_+m_^ * ) + ( ^2-t/2 ) m__cc^(000 ) . + & & . -
2 ^2 m _ _ cc^(000 ) + ( m_^2-m_^*^2 ) ( m_+m_^*)^2 ( 2m_^*-m _ ) .. + & & .. + 2 ^ 2 ( m_+m_^*)(m_^2-m_^*^2 ) .. + & & .. + 6 ( ^2- t/2 ) m_^*^2 ( m_+m_^ * ) -^4 ( 2m_^*+m _ ) . + & & .
+ ( m_+m_^*)^2 ( 4m_m_^*-m_^2-m_^*^2 ) + 6 m_^*^2 ( ^2- t/2 ) .. + & & .. - 2 ^2 ( m_+m_^ * ) ( 2m_^*-m _ ) - ^4 .
[ 2.9 ] more details , as , for example regularization and the determination of @xmath86 can be found in @xcite .
the partial wave amplitudes are obtained summing the contributions from the diagrams of fig .
1 and making a straightforward application of the expressions found in appendix a. this can be done , for example , in the @xmath5 center - of - mass frame , where @xmath87 is the momentum and @xmath88 , where @xmath89 is the scattering angle .
one notes that the @xmath90 amplitudes are real , and , so , the corresponding @xmath91 matrix is not unitary . to unitarize the amplitudes , we reinterpret them as elements of the @xmath92 matrix @xcite , and then a_l^u = a_l1-i a_l , where @xmath93 means unitarized . now the phase shifts are _
l = tg^-1 ( f_l ) .
the parameters used are the same that were used in @xcite and are @xmath94=1.318 gev , @xmath95=1.533 gev , @xmath96 , @xmath97 , @xmath98=4 and @xmath99=@xmath100 , with @xmath101=93 mev .
the coupling constant @xmath102 can be calculated comparing the ressonant @xmath103 phase with the breit - wigner expression _
l= where @xmath104 is the center - of - mass momentum at the peak of the resonance .
the obtained value is 4.54 @xmath105 .
the numerical results of the phase shifts at @xmath106 are _
p^1= -10.173^o and _ d^1= 0.208^o [ 3.20 ] _ p^3= 0.106^o and _ d^3= -0.078^o .
we have calculated the strong @xmath3 and @xmath4 phase shifts for the @xmath21 decay at its mass including the contributions from the diagrams of fig .
the numerical values of the phases are shown in the expressions ( [ 3.20 ] ) , ( [ 3.21 ] ) .
the respective values calculated in @xcite are @xmath107 and @xmath108 , that are greater in magnitude than the ones obtained in this work .
one must remark that the calculations presented here have no heavy - baryon approximation and the @xmath8-exchange and the @xmath9-term are included , that are the sources of the differences . as we can see in eq .
( [ 3.20 ] ) , ( [ 3.21 ] ) , the @xmath4 phases are much smaller than the @xmath3 ones , as it is expected at low energies , and can even be neglected in a first approximation .
we expect that the same pattern occurs for the weak phases .
in the other hyperon decays , the weak phases are of the order @xmath109 in the weinberg - higgs model and @xmath110 in the kobayashi - maskawa model @xcite . in @xcite , using the km model , the estimated value was @xmath109 , which is significantly larger when compared with the other hyperons . the asymmetry parameter @xmath52 , eq .
( [ 1.30 ] ) , in the decays @xmath111 depends on the factor @xmath112 - 0.18 . in the @xmath113 decay ,
the factor is @xmath114 - 0.12 , and using the results of @xcite , for the @xmath115 phases , in the @xmath116 decay , @xmath1170.05 .
one notes that the greatest value happens in the @xmath35 decay ( and the weak phases are also larger ) , so , we conclude that the @xmath52 parameter must be the largest in this decay . on the other hand ,
considering the @xmath53 parameter eq .
( [ 1.31 ] ) , the @xmath35 decay shows a smaller term , from @xmath118 , but the term that depends on the weak phases is larger .
so , the @xmath53 parameter is probably of the same order of the one that appears in the other hyperon decays .
the @xmath53 parameter seems to be the one where the @xmath0 violation would be most evident .
i wish to tank professor y.hama and professor m. r. robilotta , for many helpful discussions .
this work was supported by fapesp and cnpq .
in this paper @xmath119 and @xmath120 are the initial and final hyperon 4-momenta , @xmath67 and @xmath121 are the initial and final pion 4-momenta , so the mandelstam variables are @xmath122 with these variables , we can define @xmath123 where @xmath124 , @xmath125 and @xmath126 are , respectively , the hyperon mass , the resonance mass and the pion mass . the scattering amplitude for an isospin @xmath41 state is t_i=(p)a^i + b^iu(p ) , where @xmath127 and @xmath128 are calculated using the feynman diagrams .
so the scattering matrix is m_i^ba = = f_i ( ) + .n g_i ( ) = f_1^i + f_2^i , with @xmath129 where @xmath130 is the hyperon energy , and a^12=a^++2a^- , a^32=a^+-a^- , and similar expressions holds to @xmath131 .
the partial - wave decomposition is done with a_l = _
-1 ^ 1p_l(x)f_1(x ) + p_l1(x)f_2(x ) dx . in our calculation ( tree level )
@xmath90 is real . with the unitarization ,
as explained in section iii , we obtain a_l^u = e^2i_l -1= e^i_lsen(_l)a_l .
the basic loop integrals needed in order to perform the calculations of fig
. 2 are j. f. donogue , e. golowich , w. a. ponce and b. r. holstein , phys .
* d21 * , 186 ( 1980 ) ; j. f. donogue , x. -g .
he and s. pakvasa , phys . rev .
* d34 * , 833 ( 1986 ) ; x. -g .
he , h. steger and g. valencia , phys .
b * 272 * , 411 ( 1991 ) ; o. e. overseth and s. pakvasa , phys . rev . * 184 * , 1163 ( 1969 ) . | in the study of @xmath0 violation signals in @xmath1 nonleptonic decays , the strong @xmath2=3/2 @xmath3 and @xmath4 phase shifts for the @xmath5 final - state interactions are needed .
these phases are calculated using an effective lagrangian model , that considers @xmath6 , @xmath7(1530 ) , @xmath8 and the @xmath9-term , in the intermediate states .
the @xmath9-term is calculated in terms of the scalar form factor of the baryon . |
ever since the seminal study by davies et al . ( 1983 ) it is known that elliptical galaxies can be divided in two distinct classes .
the bright ellipticals ( @xmath3 ) generally have a boxy appearance , are pressure supported , and often show some form of radio and/or x - ray activity .
low luminosity ellipticals , on the other hand , often have disky isophotes , are rotationally supported , and generally lack significant amounts of activity ( bender 1988 ; bender et al .
1989 ; nieto et al . 1988 ) . before the hubble space telescope ( hst ) came online , very little
was known about the nuclear properties of these systems .
based on ground - based observations , with a typical seeing of @xmath4 arcsec , ellipticals were considered to have constant density cores with a size that correlates with global galaxy properties ( e.g. , lauer 1985 ; kormendy 1985 ) .
high - resolution imaging surveys with the hst , however , revealed that the central brightness profiles of early - type galaxies are well approximated by a power - law , @xmath5 with @xmath6 ( e.g. , lauer et al .
1991 , 1992 ; crane et al . 1993 ; ferrarese et al . 1994 ; kormendy et al .
1994 ; lauer et al . 1995 ; carollo et al .
furthermore , the distribution of @xmath2 was found to be bimodal and to follow the dichotomy of global properties inferred from ground - based observations ( jaffe et al .
1994 ; lauer et al . 1995 ; faber et al . 1997 ) and to be correlated with the global x - ray emission ( pellegrini 1999 ) .
bright ellipticals have luminosity profiles that are well fit by a double power - law with a break radius of typically a few hundred parsecs .
the inner power - law slope has values of @xmath7 .
this contrasts strongly with the low - luminosity ellipticals whose luminosity profiles lack a clear break radius and have central power - law profiles with @xmath8 on average ( jaffe et al .
1994 ; ferrarese et al .
1994 ; lauer et al . 1995 ; byun et al . 1996 ; gebhardt et al . 1996
; faber et al . 1997 ; quillen , bower & stritzinger 2000 ) . following the nomenclature of lauer
et al . ( 1995 ) we refer to the former as core galaxies and to the latter as power - law galaxies .
in addition to the central density cusps , the images from the hst have revealed that a large fraction of early - type galaxies harbor significant amounts of dust ( @xmath9 ) , either in the form of a nuclear disk , or of a more complex , filamentary or patchy morphology ( jaffe et al .
1994 ; lauer et al . 1995 ; van dokkum & franx 1995 ) .
in several cases , nuclear _
stellar _ disks with scale - lengths as small as @xmath10 pc , are found ( e.g. , van den bosch et al . 1994 ; van den bosch , jaffe & van der marel 1998 ; scorza & van den bosch 1998 ) .
both the gaseous and stellar disks have been shown to be powerful tools for determining the central densities of their parent galaxies , and for detecting massive black holes ( harms et al .
1994 ; ferrarese , ford & jaffe 1996 ; van den bosch & de zeeuw 1996 ; macchetto et al . 1997 ; bower et al . 1998 ; cretton & van den bosch 1999 ; van der marel & van den bosch 1999 ) . mainly because of these nuclear disk structures
the number of galaxies with solid detections of massive black holes ( bhs ) has increased strongly over the past years , and we have reached the point where we can start to address the demographics of bhs ( see kormendy & richstone 1995 , ford et al . 1998 ,
ho 1999 , van der marel 1999a , and macchetto 1999 for recent reviews ) .
black holes also play an important role in connection with the observed central cusps .
two different scenarios have been proposed that rely on the presence of massive black holes to explain the correlation between cusp slope and luminosity .
faber et al .
( 1997 ) have suggested that luminous galaxies are the result of mergers , whereby bhs in the progenitors sink to the center and excavate a core ( see also begelman , blandford & rees 1980 ; makino & ebisuzaki 1996 ; quinlan 1996 ; quinlan & hernquist 1997 ) .
van der marel ( 1999b ) , on the other hand , starts with the assumption of constant density cores , and shows that the adiabatic growth of central bhs can create cusps whose slopes correlate with luminosity as observed .
thus , whereas the origin of the relation between cusp slope , luminosity and black hole mass is still unknown , it has become apparent that massive black holes are omnipresent in galactic nuclei . despite the enormous wealth of new data ,
we are still far from a proper understanding of the structure and formation of the central regions of early - type galaxies .
not only have the new hst observations of these systems changed our view of their central regions , they have also raised several new problems : * is the bimodality in cusp slopes real ?
data on elliptical galaxies with kinematically distinct cores seems to suggest a much more continuous transition ( carollo et al .
* what is the origin of the correlation between cusp slope and global characteristics of early - type galaxies ? * what fraction of early - type galaxies contain nuclear disks of stars or dust / gas , and how are these structures related to other properties of their parent galaxies ?
* is there a relation between the central structure of galaxies and the presence or absence of nuclear activity ?
* how does the mass of a central bh correlate with other properties as bulge mass and cusp slope , and what role do bhs play in determining the properties of the central regions of galaxies ? in order to address these questions
one needs a large and unbiased sample of early - type galaxies that have been imaged at comparable resolution . on the order of 80 early - type galaxies
have been imaged with the hst , most of them before the refurbishment mission .
we have obtained hst images with wfpc2 of the central regions of 67 early - type galaxies as part of our hst snap shot program # 6357 .
this roughly doubles the total number of early - type galaxies that have been imaged with the hst . together with the data taken so far and with similar samples of spiral bulges ( phillips et al .
1996 ; carollo et al . 1997b ; carollo , stiavelli & mack 1998 ) and radio galaxies ( de koff et al . 1996
; mccarthy et al . 1997 ; martel et al . 1999 ; verdoes kleijn et al .
1999 ) these images provide a data base for the investigation of correlations of properties such as dust , metallicity , colors , activity , and nuclear structure . in this paper , the first in a series , we present the data and describe the luminosity profiles and isophotal structures .
more detailed analyses of the dust , stellar disks , and the central parameter relations are deferred to future papers .
the sample is compiled from the lyon / meudon extragalactic database ( leda ) , by selecting all early - type galaxies with radial velocities less than @xmath11 , an absolute @xmath12-band magnitude less than @xmath13 , and absolute galactic latitude exceeding 20 degrees ( to minimize the effects of galactic extinction ) . in total 130 galaxies
made it into the sample , of which 68 have been observed successfully by hst .
thus , although our sample is by no means complete , it has been homogeneously selected .
one galaxy , ugc 5467 , turned out to be a spiral galaxy and will be discarded in what follows .
note that , although we have attempted to discard obvious duplicates ( wfpc2 ) from our sample , several of our objects have been observed in other hst programs since the time our sample was selected . in most cases , however , a different filter was used than for the data presented here .
global properties of the 67 galaxies in our sample , taken from the leda ( see paturel et al .
1997 ) , are listed in table [ tab : sample ] .
absolute magnitudes ( column 2 ) are based on the distances listed in column 4 , and based on the virgo - centric infall corrected heliocentric velocities assuming @xmath14 ( see paturel et al .
1997 for details ) .
in addition , we list the 1.4 ghz radio fluxes , taken from the nrao vla sky survey ( nvss ; cordon et al .
1998 ) , as well as a far - infrared magnitudes defined as @xmath15 ( cf
. de vaucouleurs et al . 1991 ) . here
@xmath16 and @xmath17 are the q60 @xmath18 m and 100 @xmath18 m iras fluxes measured by one of us , adopting a similar approach to that taken by knapp et al .
we used the wfpc2 to obtain broad band images of the galaxies in our sample , using the f702w filter .
this filter , centered on @xmath19@xmath20 and with a fwhm of @xmath21 , was chosen to compromise between maximizing the number of photons and minimizing the effects of dust on the observed morphologies .
the nuclei of the galaxies were centered on the planetary camera ccd ( pc ) , which consists of @xmath22 pixels of @xmath23 each .
all exposures were taken with the telescope guiding in fine lock , yielding a rms telescope jitter of @xmath24 milli - arcseconds .
more detailed information on the wfpc2 can be found in biretta et al .
( 1996 ) .
previous hst / wfpc2 observations of ellipticals have shown that the power - law galaxies have typically a central surface brightness in the @xmath12-band of @xmath25 , whereas core - galaxies typically have @xmath26 .
our aim is to obtain a signal - to - noise ( s / n ) of 20 or larger in the center of the galaxies .
this prompted us to use a total exposure time of 1000 seconds with the analogue - to - digital gain set to 13.99 electrons / dn ( dn is the number of counts ) . with this setting the read - out noise
is @xmath27 electrons . to facilitate cosmic ray removal , and to guard against saturation in the presence of a bright nuclear point source , each exposure was split in two separate exposures of 300 and 700 seconds respectively .
the images are calibrated by the hst calibration ` pipeline ' maintained by the space telescope science institute .
the standard reduction steps include bias subtraction , dark current subtraction and flat - fielding , and are described in detail by holtzman et al .
( 1995a ) .
subsequent reduction is done using standard iraf tasks . using the data quality files and the wfixup task ,
bad pixels are corrected by means of a linear one - dimensional interpolation .
the alignment between the 300 and 700 exposures is determined by comparing the isophote centers from initial analyses of the separate exposures .
if the misalignment is significant the 300 sec exposure is shifted appropriately , using linear interpolation .
a difference - image of the 700 sec and ( shifted ) 300 sec images is used to check the accuracy of the alignment .
next , the two images are combined with simultaneous removal of cosmic rays , using the iraf task imcombine .
the cosmic ray removal and bad pixel correction is checked by inspection of the residual images between the cleaned , combined image and each of the two original frames .
sky levels are determined from empty areas in the wf images , and accordingly subtracted .
the sky - level is found to be small , typically @xmath28 dn .
the count rates in the broad - band f702w images are calibrated to magnitudes in the landoldt @xmath0-band , as described in holtzman et al .
( 1995b ) , and assuming a @xmath12-@xmath0 color of @xmath29 .
the magnitudes thus derived are checked using the inverse sensitivity and the zero point of the magnitude scale provided by the hst pipeline , whereby it is assumed that ellipticals have the spectrum of a k - iii giant .
the consistency is found to be better than 0.01 dex .
the 300 and 700 sec exposures are reduced separately as well .
cosmic rays are removed using the iraf task cosmicray .
these images are used to check the consistency of structures , for obtaining estimates of the uncertainties due to random noise , and for cases where saturation has affected the 700 sec exposure ( see [ sec : lumprof ] ) . despite the high spatial resolution of the refurbished hst ,
the central few tenths of an arcsec of the galaxies are influenced by the effects of point - spread function ( psf ) smearing .
because of the stability of the hst psf , these effects can be corrected for by means of deconvolution with the richardson - lucy ( rl ) algorithm ( lucy 1974 ) . based on tests with model galaxies
( see [ sec : models ] below ) we deconvolve the reduced 300 sec , 700 sec and combined images of each galaxy with 40 iterations of the rl algorithm . for three galaxies with faint and shallow surface brightness profiles ( ngc 3613 , ngc 4168 and ngc 4365 ) , the rl - algorithm with 40 iterations ends up amplifying the noise too strongly for an accurate analysis of the central surface brightness profile , and only 15 iterations are used in these cases . for each galaxy
we use a separate model psf @xmath30 , created with the tinytim software package ( krist 1992 ) , centered on the location of the galaxy s center in the pc ccd .
contour plots of the deconvolved images are presented in the appendix .
the main goal of this paper is to analyze the isophotal structure and luminosity profiles of the galaxies . for the isophotal analysis we use the isophote fitting task ellipse in iraf . for each isophote , the center , ellipticity ( @xmath31 ) , and position angle ( @xmath32 )
are computed .
in addition , the third and fourth order deviations of the isophote from a pure ellipse are determined .
these are described by the amplitudes ( @xmath33 and @xmath34 with @xmath35 ) of the sin and cos @xmath36 and @xmath37 terms of the following fourier expansion ( see jedrzejewski 1987 ) : @xmath38\ ] ] for a perfectly elliptical isophote these coefficients are all equal to zero .
typically the absolute values of @xmath33 and @xmath34 are small , rarely exceeding @xmath39 .
the @xmath40 coefficient is of special interest , since positive ( negative ) @xmath40 values correspond to disky ( boxy ) isophotes . in order to parameterize the luminosity profiles
they are fitted with a double power law , known as the `` nuker''-law profile ( lauer et al .
1995 ) : @xmath41^{(\gamma-\beta)/\alpha}\ ] ] which has @xmath42 at @xmath43 , and @xmath44 at @xmath45 .
the parameter @xmath46 describes the sharpness of the break from the outer to the inner power law .
the break - radius , @xmath47 , is the radius of maximum curvature in log - log coordinates , and @xmath48 is the surface brightness at @xmath49 . the nuker law has been shown to accurately fit the central luminosity profiles of early - type galaxies ( e.g. , byun et al .
note that equation ( [ nukerlaw ] ) is only intended to fit the luminosity profiles in the inner @xmath50 of early - type galaxies .
for typical fitted values of @xmath51 there must be a further downturn in the profile at larger radii for the total luminosity to be finite . in order to test the accuracy of our reduction and analysis we constructed a large set of model galaxies .
each model galaxy has a luminosity profile of the form of equation ( [ nukerlaw ] ) , has constant ellipticity and position angle , and has perfectly elliptical isophotes ( i.e. , @xmath52 for @xmath53 ) .
a fake image is created by integrating the model surface brightness distribution over each pixel , and by adding photon and read - out noise .
pixel sizes are taken the same as for the true data .
the frame size of @xmath54 pixels is smaller than the actual hst images , but satisfies the requirements for our testing purposes .
the resulting image , refered to as ` unconvolved ' image , is convolved with the hst psf ( the ` convolved ' image ) , and subsequently deconvolved using @xmath55 iterations of the rl algorithm ( the ` deconvolved ' image ) .
the psfs used for the convolution and deconvolution are constructed with tinytim , and differ slightly from each other in order to mimick errors in the exact psf shape .
the unconvolved and deconvolved model images are analyzed in the same way as the data : an isophotal analysis is performed , and we fit the nuker - law profile to the major axis luminosity profile determined from the isophote fitting . the deconvolution is tested with 10 , 15 , 25 , 40 and 80 iterations .
the best results were obtained for @xmath56 , in good agreement with lauer et al .
however , for galaxies with shallow central surface brightness profiles @xmath57 already suffices to accurately recover the unconvolved image , and more iterations only amplify the noise . in total
we have constructed @xmath58 model galaxies .
each of the 15 sets of model galaxies is characterized by a value of @xmath2 , ranging from @xmath59 to @xmath60 in steps of @xmath61 . for each
set 50 model galaxies are created whose luminosity profile parameters are varied at the few percent level .
the ellipticity and orientation of each model galaxy are chosen randomly from the intervals @xmath62 $ ] and @xmath63 $ ] , respectively .
in addition , the center of the galaxy , in fractional pixels , is drawn randomly as well . in figure
[ fig : modeliso ] we plot the isophotal parameters of one of the model galaxies with @xmath64 .
even though this model intrinsically has @xmath65 over its entire radial range , the ellipticities found by the isophote - fitting routine are significantly lower for @xmath66 , even for the unconvolved image ( solid triangles ) .
in addition , the higher - order parameters become noisy and reveal fluctuations inside this radius .
these effects owe mainly to the discrete pixel sampling and the sub - pixel interpolation used by the isophote - fitting routine , and indicate that one can not trust the isophotal parameters for @xmath67 ( @xmath68 pixels ) .
note , however , that the isophotal parameters of the deconvolved model galaxy are remarkably similar to those of the unconvolved model , indicating that the deconvolution routine used does not introduce spurious artefacts . in order to examine the accuracy with which we can recover the nuker parameters of the luminosity profile used as input for the model , we first focus on the unconvolved images . for each model
@xmath69 we compute @xmath70 with @xmath71 the input surface brightness profile ( in magn .
arcsec@xmath72 ) of equation ( [ nukerlaw ] ) , and @xmath73 the major axis lumninosity profile determined from the isophotal analysis of the unconvolved image . for each of the 15 sets of model galaxies ( that have the same value of @xmath2 ) , we then compute the average @xmath74 and the corresponding standard deviation @xmath75 , where the average is over the 50 models in each set . figure [ fig : deltamu ] plots @xmath76 and @xmath75 for 8 of the 15 sets . for @xmath77
the luminosity profiles derived from the isophotal analysis are in excellent agreement with the intrinsic profile . for smaller radii
, however , the error increases strongly , especially for the profiles with steep central cusp slopes .
even more worrysome is that the average residual @xmath78 strongly deviates from zero at small radii , implying that the main contribution to the error is of systematic rather than random nature .
this is a reflection of the problems with the isophotal analysis due to the discrete , pixelized nature of the data and the sub - pixel interpolation .
as jedrzejewski ( 1987 ) already pointed out , the sub - pixel interpolation scheme used can introduce spurious artefacts especially when the gradients in the luminosity profiles are large ( i.e. , when they are steeply cusped ) . clearly , when determining the intrinsic central cusp slopes of galaxies one can not use the surface brightness profiles determined from the isophotal analysis .
we therefore use a different approach .
for @xmath79 we fit the nuker law profiles ( equation [ [ nukerlaw ] ] ) directly to the major axis luminosity profile determined from the isophotal fit to the image . for @xmath80 , however , we extract the actual pixel values of the image from the data , which we compare to the model predictions given by @xmath81 with @xmath82 the nuker law profile and @xmath83 here @xmath84 are the coordinates centered on the galaxy center @xmath85 with the @xmath86-axis rotated by an angle @xmath87 with respect to the pixel coordinate system , and @xmath88 and @xmath87 are the ellipticity and position angle of the isophotes inside @xmath89 , respectively .
we make the assumption that @xmath90 , @xmath91 , @xmath88 , and @xmath87 are all constant inside @xmath89 and we compute their values as the average of the corresponding isophotal parameters in the radial interval from @xmath89 to @xmath92 . in cases where the isophotes in this radial range are not representative of the galaxy , e.g. , distorted by dust , a different more appropriate radial range is chosen .
the advantage of this method is that by using the observed pixel values to determine the best fit nuker profile we do not suffer from any problems with sub - pixel interpolation which hampers a proper isophotal analysis at small radii .
the disadvantage , however , is that if the actual isophotal parameters vary with radius inside @xmath89 , this will introduce systematic errors in the parameters of the best fit nuker law profile . however
, detailed tests , described below , show that these errors are much smaller than those stemming from sub - pixel interpolation .
note that a somewhat similar scheme was used by byun et al .
( 1996 ) , who used the total integrated light inside @xmath93 as an additional constraint for the nuker - law fitting procedure .
figure [ fig : modelsb ] illustrates the advantage of our new method .
the solid lines correspond to the intrinsic major axis luminosity profile of the same model as in figure [ fig : modeliso ] ( i.e. , with @xmath64 ) .
the upper profiles with the filled circles correspond to the luminosity profiles determined from the isophotal analysis .
as is evident , inside @xmath94 the sub - pixel interpolation used by the isophotal analysis introduces systematic errors .
note again that the results for the unconvolved ( left panels ) and deconvolved ( right panels ) model images are virtually identical , indicating that the deconvolution procedure used can accurately correct for the psf smearing .
the lower profiles with open symbols , offset by 1.5 magn , correspond to the luminosity profile determined with the new method .
the open circles ( @xmath95 ) are the luminosities determined from the isophotal analysis .
the open squares correspond to the actual pixel values of the model image inside @xmath89 defined such that @xmath96 , i.e. , the difference between the pixel value and the nuker - law profile at @xmath97 is equal to the difference between the actual pixel value and the prediction @xmath98 of the best - fit model . this way the residual is accurately visualized .
] . as is evident from the lower panels , this new method allows us to recover the intrinsic luminosity profile with high accuracy down to very small radii . in order to determine the parameters of the nuker law profile that best fits the data we use a @xmath99-minimization technique . for a given set of parameters ( @xmath46 , @xmath51 , @xmath2 , @xmath47 , @xmath48 ) , we determine @xmath100 . here
@xmath101 and @xmath102 are the @xmath99-values of the pixel - value fitting inside @xmath89 and the luminosity profile fitting at @xmath103 , respectively . as errors we use poisson statistics for the pixel values and the nominal errors given by the isophotal analysis , respectively . in practice ,
these latter errors are underestimated , as is evident from the fact that the normalized @xmath102 of the best fit model is typically significantly more than unity .
we therefore introduce the parameter @xmath104 , which scales the nominal error given by the isophotal analysis , and which thus sets the relative contributions of @xmath101 and @xmath102 to the total @xmath99 of the fit . in the left panel of figure [ fig : fchi ]
we plot the average difference between the intrinsic cusp steepness @xmath105 and the best - fit nuker - law profile @xmath106 , as function of @xmath104 for the set of models with @xmath107 .
the averages are taken over all 50 model galaxies in this set . as expected , the fits are better for @xmath108 .
a value of @xmath109 yields values for @xmath106 in good agreement with @xmath105 . for @xmath110 ,
too much relative weight is given to the pixel values inside @xmath89 , and @xmath106 is too large .
for too large @xmath104 , @xmath99 is completely dominated by the isophotal luminosities outside @xmath89 and the central cusp slope becomes poorly constrained , as is evident from the large errorbars . in what follows
we therefore adhere to @xmath111 , for which in general we find a normalized @xmath99 of the best fit model close to unity .
we have constructed luminosity profiles for all 750 deconvolved model images using our new method , and determined the best fit nuker law parameters using the @xmath99-minimization technique described above with @xmath111 . in figure
[ fig : fchi ] we plot @xmath106 versus @xmath105 for all 750 models .
as is evident , this method of analyzing luminosity profiles allows an accurate determination of the central cusp slope , and we use the same method to analyze the luminosity profiles of the actual hst images .
for each galaxy there are six reduced images : the non - deconvolved and deconvolved 300 sec , 700 sec , and combined images .
an isophotal analysis on each of these six images is performed .
all images are checked for contamination such as cosmic rays that have not been properly removed , foreground stars , dust filaments , and bright globular clusters . if necessary , these objects are masked out .
ten galaxies are so severely influenced by dust that meaningful estimates of their luminosity profiles and isophotal structures can not be obtained ( all galaxies with filamentary dust of `` type '' iii as well as ngc 4233 and ngc 4494 , which have a large dust disk , see
[ sec : dust ] below ) . in most of the following discussion
we discard these galaxies and focus on the sample of 57 galaxies for which the hst images provide a meaningful picture of the structure of the stellar component . in what follows
we refer to this sub - sample as the _ unperturbed _ sample .
isophotal parameters are determined down to @xmath112 , and presented in the appendix .
note that , although we plot the isophotal parameters down to @xmath113 , only the parameters for @xmath114 are considered for further analysis . for comparison , the isophotal parameters of the deconvolved image ( open circles ) are overplotted with the ones of the non - deconvolved image ( solid triangles ) . at larger radii , unaffected by convolution effects , the latter data , which is less noisy ,
nicely complements the deconvolved data .
previous high - resolution imaging with the hst has revealed that the central regions of early - type galaxies are very complex environments .
this is strikingly confirmed with our sample , and is clearly reflected in the isophotal parameters : the isophotal shapes can vary strongly with radius and are very different from one galaxy to the other .
this makes it very hard to categorize the central regions of early - type galaxies in a small number of classes .
it has become customary to classify ellipticals as either disky or boxy , depending on the sign of the @xmath40-parameter .
this generally works well for the outer parts of ellipticals , where the changes in the isophotal shape are only modest .
however , the central regions reveal a vastly more complex behavior
. a single galaxy can have isophotes that change from disky to boxy and back to disky all within the inner @xmath115 , even when there are no signs of dust .
clearly , a bimodal classification does not suffice to describe the bewildering variety of morphologies apparent in the inner @xmath116 .
another approach is therefore employed : rather than classifying galaxies as either disky or boxy a simple letter coding , referred to as _
isophotal code _ , is used to describe the individual isophotal shapes of each galaxy .
the letters indicate regions with common isophotal shape , starting from the center at @xmath117 out to @xmath118 , where _ d _ , _ b _ , _ 0 _ , _ x _ and _ ? _ refer to disky , boxy , neither disky nor boxy , undetermined due to dust , and undetermined due to small surface brightness gradient , respectively . as an example , the isophotal code _
x0d _ for ngc 2592 indicates that for small radii the isophotal shape could not be determined due to dust ( in this case a small dust disk , see
[ sec : dust ] ) , followed by a region where the isophotes are elliptical , and which become disky at larger radii . as a general rule , a galaxy region is classified as disky ( boxy ) , if @xmath119 ( @xmath120 ) for several consecutive isophotes while the other high - order parameters are not significantly different from zero .
exceptions are misaligned structures which also result in non - zero @xmath121 values .
a more detailed description of these intriguing structures is given in
[ sec : misdisk ] .
the isophotal codes for each galaxy are listed in column ( 5 ) in table [ tab : isoparameter ] . clearly , this encoding is somewhat subjective
. nevertheless , it is well suited to describe the complex and diverse isophotal behavior of the galaxies .
in addition to the isophotal code , the median ellipticity @xmath122 calculated over the radial interval from @xmath123 to @xmath124 is listed in column ( 8) of table [ tab : isoparameter ] .
note that these median values may not be very meaningful for galaxies that have rapidly changing ellipticities .
large variations may be real or may reflect the presence of dust , and are easily identified by their relatively large errors ( @xmath125 ) .
luminosity profiles along both the major and minor axes of the deconvolved images are determined and subsequently fitted with the nuker - law profile as described in detail in [ sec : models ] . in some cases a deviation from equation ( [ nukerlaw ] ) is visible in the inner @xmath94 as excess light , and pixels inside @xmath93
are ignored in the fitting procedure .
these ( largely unresolved ) ` nuclei ' are discussed in more detail in
[ sec : nuclei ] . in the cases where a central dust disk is present ,
we exclude any data inside the major axis radius of the dust disk from the fit .
the best fit parameters are listed in tables [ tab : nukermajor ] and [ tab : nukerminor ] for the major and minor axes , respectively .
plots of the luminosity profiles are presented in the appendix , with the best fit nuker law profiles overplotted ( solid lines ) . in cases where the luminosity profiles do not reveal a clear change in slope over the entire radial range ( ngc 4494 and ngc 3377 along the minor axis ) the observed luminosity profile
is fitted by a single power law : @xmath126 for these galaxies , the values listed for @xmath48 correspond to @xmath127 in tables [ tab : nukermajor ] and [ tab : nukerminor ] , and no values for @xmath46 , @xmath2 and @xmath47 are given . as discussed in
[ sec : intro ] , hst images have revealed that all early - type galaxies have luminosity profiles with central cusps .
more importantly , it was found that the distribution of cusp slopes is bimodal and in support of the dichotomy of early - type galaxies as inferred from their isophotal and kinematical structure . in their study of the central parameter relations of early - type galaxies , faber et al .
( 1997 ; hereafter f97 ) classified all galaxies with @xmath128 and @xmath129 as core galaxies .
all other galaxies were classified as power - law galaxies .
it is important to realize that such classification scheme suffers from distance effects .
galaxies at large distances will have break radii @xmath47 that are smaller than the resolution limit of the simulations .
consequently , the central cusp slopes of the observed luminosity profiles of these systems are to be compared to @xmath51 , and not @xmath2 , of less distant systems .
similarly , galaxies that are much closer than the main population of systems might reveal central luminosity profiles that deviate from the simple double - power law .
indeed , f97 pointed out that both m31 and m32 are poorly fit by the nuker - law profile , and that their classification would have been different had they been at the distance of virgo . with these caveats in mind , we now follow f97 and attempt to classify the galaxies in our unperturbed sample as either ` core ' or ` power - law ' . however , when resorting to the same classification scheme as used by f97 we are confronted with another shortcoming concerning luminosity profiles having small @xmath46 .
such profiles occur in our sample for two reasons : because of the distance - problem alluded to above , or because they intrinsically have luminosity profiles for which the slope changes smoothly with radius .
some galaxies in the latter category can still be fit by a nuker - law profile , while for others , even small @xmath46 nuker - law profiles fits can not accomodate all profile features . in all of these cases
we generally find best - fitting nuker - law profiles with @xmath130 .
however , @xmath2 is only a proper representation of the gradient in the luminosity profile at radii @xmath131 for which @xmath132 . for small @xmath46
, however , @xmath133 can be close to unity , even when @xmath131 is significantly smaller than @xmath47 , and @xmath2 therefore becomes a measure of the gradient of the luminosity profile at radii much smaller than the actual resolution limit .
since one has to be sceptical about basing a galaxy s classification on extrapolated quantities , we complement our classification scheme by introducing the parameter @xmath134_{r=0.1 '' } = { \gamma + \beta \left ( { r ' \over r_b } \right)^{\alpha } \over 1 + \left ( { r ' \over r_b } \right)^{\alpha}}\ ] ] with @xmath135 .
thus , @xmath136 is determined directly from the best - fit nuker - law , and is a measure of the gradient of the luminosity profile at @xmath137 .
the values of @xmath136 along the major and minor axes of the luminosity profiles are listed in tables [ tab : nukermajor ] and [ tab : nukerminor ] , respectively . in the upper panels of figure [ fig : gamma ]
we compare the cusp slopes of the luminosity profiles along the major and minor axes .
both @xmath2 and @xmath136 give very similar results along both axes , indicating that our method of using the actual observed pixel values inside @xmath89 to constrain the nuker - law fit yields a robust measure of the central cusp slope .
comparing @xmath2 to @xmath136 for both the major and minor axis luminosity profiles ( lower panels of figure [ fig : gamma ] ) yields a more complicated pixture .
galaxies with @xmath138 in general have @xmath139 . however , for @xmath140 about half of the galaxies have @xmath141 , and the other half has @xmath142 .
we therefore use the following classification scheme , based on both @xmath143 and @xmath144 and @xmath145 : core galaxies have @xmath146 , power - law galaxies have @xmath147 , and galaxies with @xmath148 and @xmath149 are provisionally refered to as ` intermediate ' galaxies . note that this classification scheme is more conservative in designating galaxies as core galaxies as the scheme used by f97 : in our sample , only @xmath150 percent of the galaxies classified as core galaxies according to the f97 scheme are classified similarly with our definition . the other @xmath151 percent
are classified as intermediate .
column ( 2 ) in table [ tab : isoparameter ] indicates the classification of our galaxies as either core ` @xmath152 ' , power - law ` @xmath153 ' , or intermediate ` ) ' .
a question mark indicates that the galaxy is not in the unperturbed sample , and that no attempt has been made to classify this galaxy .
figure [ fig : gammahisto ] presents histograms of the distributions of @xmath2 and @xmath136 . although the distribution of cusp slopes reveals a clear hint for bimodality , consistent with the results of f97 , we find less of a deficit of galaxies with @xmath154 .
most of the galaxies with @xmath2 in this range , and which are thus classified as power - law galaxies ( e.g. , ngc 5576 , ngc 5796 , and ngc 5831 ) , have small values of @xmath46 and no disky isophotes , similar to the intermediate galaxies .
the intermediate galaxies themselves follow the same distribution of @xmath2 as the _ core galaxies _ , but at the same time occupy the low-@xmath136 end of the _ power - law galaxies_. thus , whereas we confirm the bimodality in cusp slopes found by f97 , we find that @xmath155 percent of the galaxies in our unperturbed sample have luminosity profiles whose profile slopes change very gradually with radius ( reflected by the small value of @xmath46 ) , making a classification based on central cusp slope somewhat ambiguous ( see discussion in [ sec : cusp - slopes ] below ) .
as discussed in the introduction , the high resolution images obtained with the hst have strongly increased the detection rate of dust in early - type galaxies .
two different dust morphologies are distinguished : disks ( or tori ) and filaments ( indicated by ` d ' and ` f ' , respectively , in table [ tab : isoparameter ] ) .
the amount of filamentary dust present is indicated by a roman numeral , ranging from i ( small traces of dust that hardly influence the isophotal shapes ) to iii ( large amounts of dust that prevent a meaningful analysis of the isophotes and luminosity profiles ) .
types ii and iii are easily discerned from an inspection of the images by eye , whereas filamentary dust of class i is only detected after a more detailed investigation . a detailed discussion of the dust properties in our sample is presented in tran et al .
( 2000 ) . here
we merely list the main statistics . in 10 galaxies ( 15 percent )
we find a nuclear dust disk , with a diameter ranging from @xmath156 ( ngc 4648 ) to @xmath157 ( ngc 4233 ) .
most remarkably , 8 of these galaxies are classified as power - law galaxies , whereas the two remaining , ngc 4233 and ngc 4494 , have a dust disk that is too large to be able to classify the galaxy based on its luminosity profile .
filamentary dust is found in 19 cases ( 28 percent ) . in total
, we thus find evidence for dust in 43 percent of our galaxies .
van dokkum & franx ( 1995 ) , analyzing 64 early - type galaxies imaged with the hst before the refurbishment mission , found dust in 48 percent of the cases and pointed out that a significant fraction of the dust will be missed because of inclination effects .
they estimated that @xmath158 percent of the early type galaxies harbor significant amounts of dust .
our results are in good agreement with their estimates .
as mentioned in the introduction , early - type galaxies come in two different classes , core and power - law galaxies .
the latter are relatively faint ( @xmath159 ) , typically have disky isophotes , and are thought to be rotationally supported spheroidal systems .
core galaxies , on the other hand , are relatively bright systems , often with boxy isophotes , that are thought to be triaxial and henceforth pressure - supported .
we now investigate to what extent the galaxies in our sample obey this dichotomy . as discussed in
[ sec : isoph ] , the isophotal structures of the central regions are very complex , and we introduced an isophotal code ( hereafter ic ) to describe the observed morphologies .
the galaxies can roughly be divided into four classes : disky ( ` d ' , ics with @xmath160 but no @xmath161 ) , boxy ( ` b ' , ics with @xmath161 but no @xmath160 ) , disky - boxy ( ` db ' , ics with both @xmath160 and @xmath161 ) , and regular ( ` 0 ' , ics with no @xmath160 or @xmath161 ) . in table
[ tab : lumisocorr ] we correlate this classification based on the isophotal structure with that based on the luminosity profiles .
we only consider the 57 galaxies in our unperturbed sample . in general
, our sample nicely confirms the dichotomy that has emerged from previous studies ( e.g. , jaffe et al .
1994 ; lauer et al .
1995 ) : 20 of the 21 disky ( ` d ' ) galaxies are classified as power - laws , while 7 of the 9 core galaxies are classified as either boxy ( ` b ' ) or regular ( ` 0 ' ) . however , not _ all _ galaxies fall in either of these two classes . as we indicated in
[ sec : class ] , roughly twelve percent of our galaxies classify as ` intermediate ' based on their luminosity profiles .
remarkably , none of these galaxies reveal any sign of diskiness .
in fact , two of them classify as boxy , whereas the other four are regular .
as such , these intermediate galaxies are more similar to core galaxies than to power - law galaxies . of the in total 8 galaxies with both disky and boxy isophotes ( ` db ' ) , 7
classify as power - law galaxies . in total
, we thus find that 93 percent of all galaxies with disky isophotes are power - law galaxies .
if one takes into account that nuclear disks are much easier detectable in core galaxies than in power - law galaxies , it is clear that there is a very strong correlation between the diskiness and central cusp slope of early - type galaxies .
four of the in total 57 galaxies in our unperturbed sample deviate from these general trends : ngc 3078 and ngc 5576 are the only power - law galaxies ( out of 41 ) with boxy isophotes .
ngc 3078 is also the galaxy with the highest 1.4 ghz flux in our sample , and reveals a central dust disk with a diameter of @xmath162 .
the dust absorption combined with the presence of a strong agn makes the classification rather uncertain , and it is thus not inconceivable that ngc 3078 is in fact a core galaxy .
ngc 5576 , on the other hand , has no detectable dust or strong radio emission .
however , with @xmath163 and @xmath164 it is close to an intermediate galaxy .
ngc 3613 and ngc 4365 are the only two core galaxies with disky isophotes : ngc 3613 harbours a thin edge - on nuclear disk inside the central arcsec , and ngc 4365 contains a kinematically decoupled core with the characteristics of a cold , rotating system ( surma 1993 ) . in figure
[ fig : gammamag ] we plot @xmath165 as function of both the ( major axis ) break radius @xmath47 and the absolute @xmath166-band magnitude @xmath167 .
solid circles correspond to core galaxies , solid squares to power - law galaxies , and open circles to intermediate galaxies .
all 10 core galaxies have break radii in the ( relatively narrow ) interval @xmath168 , and are bright ( @xmath169 ) .
the power - law galaxies , on the other hand , typically have @xmath170 and @xmath171 .
this once again confirms the dichotomy amongst early - type galaxies inferred from previous studies .
but what about the intermediate galaxies ? according to their isophotal structure , one would tend to classify them as core galaxies ( see [ sec : isoshapes ] ) .
the same is true if one bases the classification on @xmath2 ( cf .
figure [ fig : gammahisto ] ) .
however , they also have properties that clearly differentiate them from core galaxies : their @xmath136-values are more similar to those of the power - law galaxies ( although they occupy the low-@xmath172 end of the distribution ) , their luminosity profiles have values of @xmath46 that are significantly lower than the typical value for core galaxies , and finally , their average absolute magnitudes seem to be intermediate between those of both the core and power - law galaxies ] : they can not be unambiguously classified as either one of those .
more detailed investigations of their properties are required to see if they make up a truly , physically distinct class of objects , or whether these galaxies are distinct from both the core and power - law galaxies only because of distance and/or resolution effects . in fourteen galaxies we find radial intervals with strong isophote twists , non - zero fourth order moments ( @xmath121 and/or @xmath40 ) and local ellipticity maxima .
a clear example is ngc 2699 where at @xmath173 from the center the isophotes reveal an isophote twist of @xmath174 , both @xmath121 and @xmath40 are positive , and there is a local maximum in ellipticity . in some cases
the orientation angle of the misaligned structures changes with radius , giving it the appearance of spiral arms ( e.g. , eso 443 - 39 , ngc 3595 , and ugc 4551 ) , wheras in other cases ( eso 378 - 20 , eso 443 - 39 , eso 447 - 30 , ngc 2950 and ugc 6062 ) misaligned structures are evident in two distinct radial intervals . a clear example is eso 447 - 30 with misaligned structures at both @xmath175 and @xmath176 . in each of these galaxies ,
the two misaligned structures are not only misaligned with the host galaxy , but also with each other . given the highly elongated nature of these structures and the fact that disks are common in elliptical galaxies , it is tempting to interpret these structures as stellar disks whose orientation axes are misaligned with those of the host galaxy .
however , disks in spheroidal systems can only be stable if their symmetry axis is perpendicular to the equatorial plane of the potential , which does not result in a projected misalignment as observed .
if , on the other hand , a stellar disk is embedded in a triaxial galaxy , projection effects can easily accommodate the observed misaligned morphologies .
furthermore , if the axis ratios of the triaxial host galaxy vary with radius , the projected angle of misalignment can vary with projected radius . such configuration , therefore , can explain the multiple misaligned structures observed .
a problem with this explanation , however , is that all 14 galaxies with misaligned structures are power - law galaxies which are generally thought to be spheroidal systems .
furthermore , if these structures are disks the observed fraction of disks embedded in power - law galaxies is too high ( see [ sec : disks ] below ) .
a more likely explanation , therefore , is that these misaligned structures correspond to bars .
this lends great support from the fact that isophotal structures , very similar to those seen here , have been observed in barred spirals and s0s ( e.g. , wozniak et al .
1995 ; friedli et al .
1996 ; erwin & sparke 1999 ) . in particular , the bars within bars found by these authors provide a natural explanation for the multiple misaligned structures found in some of our galaxies . indeed , both kormendy ( 1981 ) and wozniak et al .
( 1995 ) have previously noted the complex morphological structure of ngc 2950 ( one of the fourteen galaxies in our sample with misaligned structures ) and classified it as a double - barred galaxy .
the bar interpretation is further supported by the fact that ten of the fourteen galaxies in our sample with misaligned structures are classified as s0 or e / s0 ( see table [ tab : sample ] ) .
it is unclear , however , how to interpret the misaligned structures in the four galaxies classified as ellipticals , although it is not inconceivable that these galaxies may be misclassified s0s seen close to face - on .
detailed investigations into the kinematics of these galaxies are required to confirm that indeed these structures correspond to tumbling triaxial potentials . in total 29 out of the 57 galaxies ( 51 percent ) in our unperturbed sample
reveal disky isophotes over at least a small radial interval ( this includes the misaligned structures discussed in [ sec : misdisk ] ) .
this diskiness is most straightforwardly interpreted as due to an embedded stellar disk ( e.g. , scorza & bender 1995 , and references therein ) .
the diversity of disk structures in our sample is large , ranging from extremely thin and small disks in the very center ( i.e. , ngc 3613 ) up to very large disks spanning the entire radial interval imaged ( i.e. , ngc 4621 ) . in 18 galaxies the isophotes are disky in only a single radial interval , whereas the other 11 galaxies have disky isophotes in two radial intervals , separated by purely elliptical or boxy isophotes .
this `` double - diskiness '' may be due to a nuclear disk inside the central hole of a separate outer disk or ring ( i.e. , van den bosch & emsellem 1998 ) , reflect the projected surface brightness contribution of a single disk which is locally dwarfed by that of the spheroidal , or be due to a bar - within - a - bar structure ( see [ sec : misdisk ] ) .
the observed isophotal structure of an elliptical galaxy with an embedded stellar disk depends strongly on the ratio of the surface brightness distributions of the two components and on the inclination angle of the disk .
detailed disk - bulge decompositions ( i.e. , scorza & bender 1995 ; scorza & van den bosch 1998 ) are required to decompose the observed surface brightness in the separate components , which is beyond the scope of this paper . given the large fraction of early - type galaxies that reveal evidence for an embedded disk , and the fact that these disks are only visible if the disk is seen under a sufficiently large inclination angle , implies that the true fraction of early - type galaxies with embedded stellar disks is likely to be significantly larger than the @xmath177 percent found here . in our unperturbed sample ,
@xmath178 percent of all power - law galaxies have some amount of diskiness .
this is a surprisingly high fraction .
if we assume that all power - law galaxies have embedded disks , and that these disks are only detectable if their projected minor - to - major axis ratio is smaller than @xmath179 , one only expects to detect disks in @xmath180 percent of all power - law galaxies ( see rix & white 1990 for a detailed discussion on the detectability of embedded stellar disks ) .
the observed fraction of @xmath178 percent implies either of the following : ( i ) all disks with a projected axis ratio smaller than @xmath181 are detectable ( assuming that all power - law galaxies harbor embedded disks ) , ( ii ) power - law galaxies are always seen under high inclination angles , ( iii ) the steep central cusp owes to the projection of the highly flattened structure ( but see lauer et al . 1995 ; faber et al .
1997 ) , or ( iv ) in some cases the diskiness is due to highly elongated prolate or triaxial structures ( i.e. , bars ) rather than oblate disks .
we only consider the latter of these explanations viable .
it is supported by the fact that fourteen of our power law galaxies have their diskiness misaligned with the main galaxy . as we discussed in the previous section
, these structures are most likely due to bars in s0 galaxies .
if we exclude these fourteen galaxies , we find embedded disks in only @xmath182 percent of our power - law galaxies , which is in excellent agreement with the @xmath180 percent expected if all power - law galaxies harbor embedded disks , but which are only detectable if the disk s projected axis ratio is smaller than @xmath179 .
although the nuker - law in general provides a good fit to the observed surface brightness of the central regions of early - type galaxies , in several cases the very centers reveal excess light , associated with an increase in slope as the hst resolution limit is approached .
we follow lauer et al .
( 1995 ) and refer to these features as _
nuclei_. these unresolved nuclei may be due to non - thermal emission from agns or may reflect the presence of a nuclear star cluster .
table [ tab : isoparameter ] lists the degree of nucleation observed , ranging from very mild , indicated by ` i ' ( e.g. , ngc 2634 ) to nuclei that stand out strongly against the ` background ' nuker - law profile , indicated by ` iii ' ( e.g. , ngc 4482 ) . in order to be termed `
nucleated ' we require that the central steepening of the surface brightness occurs inside @xmath183 , and is visible from both the major and minor axes surface brightness profiles .
note that effects due to deconvolution as well as dust obscuration can produce artefacts in the luminosity profiles that are similar to our level i nucleation .
in order to guard against such artificial ` detections ' , we have been fairly conservative in assigning nucleation . in total ,
nuclei are detected in 9 out of the 67 galaxies ( 13 percent ) in our sample .
this is less than half the detection rate of lauer et al .
( 1995 ) , who detected nuclei in 16 out of 45 galaxies . except for the fact that we have been fairly conservative in assigning nuclei ,
it is unclear why our detection rate is so much smaller
. only 2 of the 16 nucleated galaxies in the sample of lauer et al .
are core galaxies . in both cases
the nucleation is interpreted as non - thermal emission from agns . in our sample ,
the two nucleated galaxies ( ngc 4168 and ngc 5077 ) not classified as power - law galaxies have been detected at 1.4 ghz ( see table [ tab : sample ] ) .
this is consistent with the idea that ` nucleation ' in core ( and intermediate ) galaxies is associated with an active galactic nucleus . for none of the 7 nucleated power - law galaxies
do we find any sign of non - thermal emission associated with an agn ( i.e. , none of these galaxies have been detected at 1.4 ghz ) .
this again is in good agreement with the results of lauer et al .
these results , however , are somewhat puzzling from an astrophysical point of view .
an often advertised interpretation of ( non - agn related ) nuclei is that they are stellar remnants of cannibalized , low - mass galaxies that have reached the center by means of dynamical friction .
since less luminous galaxies are in general denser than their more massive counterparts , one expects to find stellar nuclei predominantly in the less dense cores of the more massive galaxies , where tidal forces are less severe ( e.g. , kormendy 1984 ; faber et al .
the almost complete absence of stellar nuclei in core galaxies hints at some violent disruption process , most likely associated with massive black holes ( see merritt & cruz 2001 , and references therein ) .
however , this fails to explain why power - law galaxies do seem to harbor stellar nuclei .
these galaxies have massive black holes as well , and their intrinsically high stellar densities will only aid in disrupting any merger remnant that might reach the center of the potential well .
a more viable explanation is that these nuclei are the product of some dissipational formation mechanism .
the presence of highly flattened disk - structures in the majority of these power - law galaxies supports the view that dissipational processes have played an important role .
we have reported on the @xmath0-band images of the central regions of 67 early - type galaxies obtained with the wfpc2 aboard the hst .
this sample roughly doubles the number of early - type galaxies that have now been imaged at hst resolution , and complements similar data on the central regions of radio galaxies and the bulges of spiral galaxies . in this paper
we have presented the data , focusing in particular on the isophotal structures and luminosity profiles .
our sample strikingly confirms the complex morphologies of the stellar component in the central regions of early - type galaxies which have become apparent from previous studies with hst .
whereas the outer regions of early - type galaxies can in general be classified as either disky or boxy , the isophotal structure in the inner @xmath116 varies strongly with radius , such that a bimodal classification does not suffice to describe the bewildering variety of structures seen .
instead , we have used a simple letter coding to describe the diverse morphologies .
previous imaging surveys with hst have shown that a more suitable criterion for classifying the central regions of early - type galaxies is based on their luminosity profiles ( rather than their isophotal structure ) .
depending on the slope of the central cusp , galaxies are classified as either a core galaxy or a power - law galaxy .
we determined the luminosity profiles along the major and minor axes for 57 galaxies in our sample ( the unperturbed sample ) .
the remaining 10 galaxies do not allow a proper assessment of their luminosity profiles because of dust obscuration .
we have shown how luminosity profiles derived from an isophotal analysis of the data can produce relatively large systematic errors in the central region , owing to the sub - pixel interpolation used .
we therefore use the actual pixel values in the central @xmath89 to define the luminosity profiles .
we fit the luminosity profiles with a nuker - law profile ( equation [ [ nukerlaw ] ] ) , which has been shown to adequately describe the luminosity profiles of early - type galaxies in the inner @xmath184 . at radii
less than @xmath89 we fit this nuker - law model directly to the observed pixel values , while at larger radii we use the luminosity profile as determined from the isophotal analysis .
detailed modeling has shown that this method allows an accurate recovery of the actual luminosity profiles . in a similar study , based on a sample of 61 early - type galaxies imaged with the wfpc1 , faber et al .
( 1997 ) classified galaxies as core galaxies if the central cusp slope @xmath2 of the best fitting nuker law was less than @xmath185 .
galaxies with @xmath186 were classified as power - law galaxies . in our sample
we find a subset of galaxies with luminosity profiles that have a continuously changing slope , rather than a double power - law shape with a clear break radius . for these systems
@xmath2 of the best - fit nuker - law profile is smaller than @xmath185 , but it is only a measure of the gradient of the luminosity profile at radii much smaller than the resolution limit of the actual observations . at the resolution
limit the actual slope of the luminosity profile is steeper than @xmath185 , and it is unclear whether to classify these systems as core or power - law galaxies .
although the isophotal properties are similar to those of the core galaxies , there is a hint that the luminosities of these galaxies are intermediate between those of the core and power - law galaxies .
we therefore provisionally classify these galaxies as ` intermediate ' .
however , it is unclear at present whether these galaxies make up a physically distinct class or whether distance and/or resolution effects cause them to loose their distinct core and/or power - law characteristics .
f97 have pointed out that the distribution of central cusp slopes is clearly bimodal : whereas ( by definition ) core galaxies have @xmath128 , power - law galaxies are found to have @xmath187 .
the region @xmath188 is virtually devoid of galaxies .
this bimodality closely follows the dichotomy in other properties : the core galaxies make up the bright end of the luminosity function and typically have boxy isophotes , whereas the power - law galaxies are fainter and typically have disky isophotes .
we confirm this dichotomy with our sample : 93 percent of all galaxies that reveal disky isophotes are power - law galaxies , whereas 77 percent of all core galaxies have either boxy or purely elliptical isophotes .
furthermore , we confirm the bimodality of cusp slopes , but we do find a significant population of galaxies with @xmath189 . combined with the class of intermediate galaxies , which have cusp slopes at the resolution limit that also fall in this range , we thus conclude that the bimodality is not as strong as previously claimed .
finally , the statistical properties of our sample can be summarized as follows : * in 15 percent of the galaxies in our sample we detect dust disks with diameters that range from @xmath156 to @xmath157 . together with the 28 percent of the galaxies in which we detect filamentary dust , the total detection rate of dust is 43 percent , in good agreement with previous estimates based on hst data ( van dokkum & franx 1995 ) . * in 9 galaxies we find evidence for nuclei .
these are defined as ( largely ) unresolved , nuclear surface brightness spikes that cause a central upturn in the luminosity profiles .
our detection rate of nuclei ( 13 percent ) is substantially less than the 36 percent found by lauer et al .
( 1995 ) in their sample , but this is most likely a reflection of our more conservative approach to identifying nucleation . whereas the nuclei in power - law galaxies are most likely stellar , the nucleation seen in core galaxies is generally non - thermal emission from an active galactic nucleus . * in 14 galaxies we find structures that are misaligned with the principal axes of the host galaxy .
these structures are most likely associated with stellar bars in s0 galaxies . in 5 of these galaxies ,
multiple misaligned structures are apparent indicating bars within bars . * a large fraction ( 51 percent ) of the early - type galaxies in our unperturbed sample
reveal disky isophotes over at least a small radial interval , whereas this fraction increases to almost 70 percent when only considering power - law galaxies .
this is a remarkably large fraction : even if all power - law galaxies were to harbor an embedded stellar disk , one would not expect to detect these thin structures in more than @xmath190 percent of all cases because of inclination effects . if , however , we exclude the cases where the diskiness is misaligned with the main galaxy , and which are more likely associated with stellar bars
, we find that @xmath191 percent of all power - law galaxies reveal some amount of diskiness .
this is consistent with a picture in which _ all _ power - law galaxies harbor an embedded stellar disk .
we are grateful to the referee , tod r. lauer , for his insightful comments that greatly improved the analysis and presentation of the data in this paper .
the observations presented in this paper were obtained with the nasa / esa hubble space telescope , which is operated by aura , inc .
, under nasa contract nas 5 - 26555 .
we acknowledge the use of the leda database ( www-obs.univ-lyon1.fr ) .
fvdb was supported by nasa through hubble fellowship grant # hf-01102.11 - 97.a awarded by the space telescope science institute .
in the following figures we present the data for all 67 galaxies in our sample . the upper right - hand panels show contour plots of the @xmath192 field of view of the planetary camera .
the thick black bar in the upper - right corner corresponds to an angular size of @xmath193 .
contours are plotted at intervals of @xmath185 mag between @xmath194 mag arcsec@xmath72 and the surface brightness at @xmath195 ( in order to prevent central crowding of contours ) .
the upper - panels on the left show the observed luminosity profiles ( open circles are determined from the isophotal analysis , closed circles are the actual pixel values fitted and plotted as described in [ sec : models ] ) along the major ( upper curves ) and minor axis ( lower curves , offset by @xmath196 mag for clarity ) .
the thick solid lines are the best fit nuker - law profiles , the parameters of which are listed in tables [ tab : nukermajor ] and [ tab : nukerminor ] . in some cases data in the central regions
is excluded from the nuker - law profile fit because of nucleation and/or nuclear dust disks . in these cases
a vertical thin dashed line indicates the lower limit of the data range used .
the remaining panels plot the isophotal parameters for both the non - deconvolved image ( solid triangles ) and the deconvolved image ( open circles with errorbars ) .
for each galaxy we plot as function of radius the ellipticities ( @xmath31 ) , the position angles of the isophotes ( @xmath32 , measured from north to east ) , and the higher - order parameters @xmath33 and @xmath34 ( with @xmath35 ) .
some galaxies have their central regions saturated in the 700 sec exposure . in these cases
we plot the isophotal parameters of the 300 sec exposures in the radial regime where the 700 sec exposure shows saturation effects ( rather than those of the combined image ) . in three cases ( ngc 2950 , ngc 3377 , and ngc 4621 )
, the central regions of the 300 sec exposures are also saturated . for these galaxies ,
only the isophotal parameters are plotted down to the radius where saturation effects in the 300 sec exposure start .
kormendy , j. , dressler , a. , byun , y. i. , faber , s. m. , grillmair , c. , lauer , t. r. , richstone , d. , & tremaine , s. , 1994 , in dwarf galaxies , eds .
g. meylan & p. prugniel ( eso conference and workshop proceedings ) , p.147 lcccccccccc eso 378 - 20 & @xmath197 & @xmath198 & @xmath199 & s0a & @xmath200 & @xmath201 & @xmath202 & @xmath203 & & eso 437 - 15 & @xmath204 & @xmath205 & @xmath206 & s0 & @xmath207 & @xmath208 & @xmath209 & & @xmath210 & @xmath211 eso 443 - 39 & @xmath204 & @xmath212 & @xmath213 & s0 & @xmath214 & @xmath215 & @xmath216 & & & eso 447 - 30 & @xmath217 & @xmath218 & @xmath219 & s0 & @xmath220 & @xmath221 & @xmath222 & & & @xmath223 eso 507 - 27 & @xmath224 & @xmath225 & @xmath226 & s0 & @xmath227 & @xmath228 & @xmath229 & & & eso 580 - 26 & @xmath230 & @xmath231 & @xmath232 & s0 & @xmath233 & @xmath234 & @xmath235 & & @xmath236 & @xmath237 ic 875 & @xmath238 & @xmath239 & @xmath240 & s0 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & & mcg 11 - 14 - 25a & @xmath245 & @xmath246 & @xmath247 & e / s0 & @xmath248 & @xmath249 & @xmath250 & & & mcg 8 - 27 - 18 & @xmath251 & @xmath252 & @xmath253 & e / s0 & @xmath254 & @xmath255 & @xmath256 & & & ngc 2549 & @xmath257 & @xmath258 & @xmath259 & s0 & @xmath29 & @xmath260 & @xmath261 & @xmath262 & & ngc 2592 & @xmath263 & @xmath264 & @xmath265 & e & @xmath266 & @xmath267 & @xmath268 & & & ngc 2634 & @xmath269 & @xmath270 & @xmath271 & e & @xmath272 & @xmath273 & @xmath274 & @xmath275 & & ngc 2699 & @xmath276 & @xmath277 & @xmath278 & e & @xmath279 & @xmath280 & @xmath281 & & & ngc 2778 & @xmath282 & @xmath283 & @xmath284 & e & @xmath285 & @xmath286 & @xmath287 & @xmath288 & & ngc 2824 & @xmath289 & @xmath290 & @xmath291 & s0 & @xmath292 & @xmath234 & @xmath293 & @xmath294 & @xmath295 & @xmath296 ngc 2872 & @xmath297 & @xmath298 & @xmath299 & e & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 & @xmath305 ngc 2902 & @xmath306 & @xmath307 & @xmath308 & s0 & @xmath309 & @xmath310 & @xmath311 & & & ngc 2950 & @xmath312 & @xmath313 & @xmath314 & s0 & @xmath315 & @xmath316 & @xmath317 & @xmath318 & & ngc 2986 & @xmath319 & @xmath320 & @xmath321 & e & @xmath248 & @xmath322 & @xmath323 & @xmath324 & & ngc 3065 & @xmath325 & @xmath326 & @xmath327 & s0 & @xmath254 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & ngc 3078 & @xmath332 & @xmath333 & @xmath334 & e & @xmath335 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & ngc 3193 & @xmath340 & @xmath341 & @xmath342 & e & @xmath343 & @xmath344 & @xmath345 & @xmath346 & & ngc 3226 & @xmath347 & @xmath348 & @xmath349 & e & @xmath248 & @xmath350 & @xmath351 & @xmath352 & @xmath353 & @xmath354 ngc 3266 & @xmath347 & @xmath355 & @xmath356 & s0 & @xmath248 & @xmath357 & @xmath358 & & & ngc 3348 & @xmath359 & @xmath360 & @xmath226 & e & @xmath272 & @xmath361 & @xmath362 & @xmath363 & @xmath364 & ngc 3377 & @xmath365 & @xmath366 & @xmath367 & e & @xmath368 & @xmath369 & @xmath370 & @xmath371 & & ngc 3414 & @xmath372 & @xmath373 & @xmath374 & s0 & @xmath335 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & @xmath379 ngc 3595 & @xmath380 & @xmath381 & @xmath382 & e / s0 & @xmath200 & @xmath383 & @xmath384 & & & ngc 3613 & @xmath385 & @xmath386 & @xmath387 & e & @xmath388 & @xmath389 & @xmath390 & @xmath391 & & ngc 3640 & @xmath392 & @xmath393 & @xmath394 & e & @xmath395 & @xmath396 & @xmath397 & @xmath398 & & ngc 4121 & @xmath399 & @xmath400 & @xmath401 & e & @xmath248 & @xmath402 & @xmath287 & @xmath403 & & ngc 4125 & @xmath404 & @xmath405 & @xmath406 & e & @xmath266 & @xmath407 & @xmath408 & @xmath288 & & @xmath409 ngc 4128 & @xmath410 & @xmath411 & @xmath412 & s0 & @xmath413 & @xmath414 & @xmath415 & @xmath416 & & ngc 4168 & @xmath417 & @xmath418 & @xmath419 & e & @xmath241 & @xmath420 & @xmath421 & @xmath422 & @xmath423 & ngc 4233 & @xmath424 & @xmath425 & @xmath426 & s0 & @xmath200 & @xmath427 & @xmath428 & @xmath429 & @xmath430 & ngc 4291 & @xmath431 & @xmath432 & @xmath433 & e & @xmath434 & @xmath435 & @xmath436 & @xmath437 & & ngc 4365 & @xmath438 & @xmath439 & @xmath259 & e & @xmath285 & @xmath440 & @xmath441 & @xmath442 & & ngc 4474 & @xmath443 & @xmath444 & @xmath445 & s0 & @xmath446 & @xmath447 & @xmath448 & @xmath449 & & ngc 4478 & @xmath450 & @xmath451 & @xmath452 & e & @xmath266 & @xmath267 & @xmath453 & @xmath454 & & ngc 4482 & @xmath455 & @xmath456 & @xmath457 & e / s0 & @xmath458 & @xmath459 & @xmath460 & & & ngc 4494 & @xmath461 & @xmath462 & @xmath463 & e & @xmath254 & @xmath464 & @xmath465 & @xmath466 & & @xmath467 ngc 4503 & @xmath468 & @xmath469 & @xmath470 & s0a & @xmath471 & @xmath472 & @xmath473 & @xmath474 & & ngc 4564 & @xmath475 & @xmath476 & @xmath477 & e & @xmath388 & @xmath478 & @xmath479 & @xmath474 & & ngc 4589 & @xmath480 & @xmath481 & @xmath482 & e & @xmath241 & @xmath483 & @xmath484 & @xmath437 & @xmath485 & @xmath486 ngc 4621 & @xmath487 & @xmath462 & @xmath488 & e & @xmath285 & @xmath489 & @xmath490 & @xmath491 & & ngc 4648 & @xmath492 & @xmath493 & @xmath278 & e & @xmath220 & @xmath494 & @xmath495 & @xmath496 & & ngc 5017 & @xmath497 & @xmath498 & @xmath499 & e & @xmath285 & @xmath500 & @xmath501 & @xmath429 & & ngc 5077 & @xmath502 & @xmath503 & @xmath504 & e & @xmath505 & @xmath506 & @xmath507 & @xmath508 & @xmath509 & @xmath510 ngc 5173 & @xmath511 & @xmath512 & @xmath513 & e / s0 & @xmath272 & @xmath514 & @xmath515 & & @xmath516 & @xmath517 ngc 5198 & @xmath518 & @xmath519 & @xmath520 & e / s0 & @xmath434 & @xmath521 & @xmath522 & @xmath523 & @xmath524 & ngc 5283 & @xmath525 & @xmath526 & @xmath527 & s0 & @xmath395 & @xmath528 & @xmath529 & & @xmath530 & ngc 5308 & @xmath531 & @xmath532 & @xmath284 & s0a & @xmath533 & @xmath483 & @xmath534 & @xmath535 & & ngc 5370 & @xmath536 & @xmath537 & @xmath538 & s0 & @xmath39 & @xmath539 & @xmath540 & & & ngc 5557 & @xmath541 & @xmath542 & @xmath543 & e & @xmath544 & @xmath545 & @xmath546 & @xmath403 & & ngc 5576 & @xmath431 & @xmath547 & @xmath548 & e & @xmath233 & @xmath549 & @xmath550 & @xmath551 & & ngc 5796 & @xmath297 & @xmath552 & @xmath553 & e & @xmath300 & @xmath554 & @xmath555 & @xmath556 & @xmath557 & ngc 5812 & @xmath558 & @xmath559 & @xmath560 & e & @xmath248 & @xmath561 & @xmath562 & @xmath563 & & ngc 5813 & @xmath564 & @xmath565 & @xmath560 & e & @xmath285 & @xmath369 & @xmath566 & @xmath442 & @xmath567 & ngc 5831 & @xmath568 & @xmath569 & @xmath570 & e & @xmath571 & @xmath572 & @xmath573 & @xmath574 & & ngc 5846 & @xmath575 & @xmath576 & @xmath308 & e & @xmath279 & @xmath577 & @xmath578 & @xmath579 & @xmath580 & ngc 5898 & @xmath581 & @xmath582 & @xmath583 & e & @xmath584 & @xmath585 & @xmath586 & @xmath587 & & ngc 5903 & @xmath564 & @xmath588 & @xmath589 & e & @xmath220 & @xmath478 & @xmath590 & @xmath591 & @xmath592 & ngc 5982 & @xmath593 & @xmath594 & @xmath232 & e & @xmath595 & @xmath596 & @xmath597 & @xmath598 & & ngc 6278 & @xmath599 & @xmath237 & @xmath600 & s0 & @xmath601 & @xmath602 & @xmath603 & & & ugc 4551 & @xmath604 & @xmath605 & @xmath457 & s0 & @xmath606 & @xmath607 & @xmath578 & @xmath608 & & ugc 4587 & @xmath609 & @xmath610 & @xmath611 & s0 & @xmath214 & @xmath612 & @xmath613 & & & ugc 5467 & @xmath614 & @xmath400 & @xmath615 & s0 & @xmath272 & @xmath616 & @xmath617 & & @xmath618 & @xmath619 ugc 6062 & @xmath620 & @xmath621 & @xmath622 & s0 & @xmath368 & @xmath623 & @xmath624 & & & lccccccc eso 378 - 20 & @xmath153 & & & _ d0d _ & & 2 & @xmath625 eso 437 - 15 & ? & f & iii & _ xd _ & & & @xmath626 eso 443 - 39 & @xmath153 & & & _ 0d0d _ & & 2 & @xmath627 eso 447 - 30 & @xmath153 & & & _ 0d0d _ & & 2 & @xmath628 eso 507 - 27 & @xmath153 & & & _ dbd _ & ii & & @xmath629 eso 580 - 26 & ? & f & iii & _ x _ & & & @xmath630 ic 875 & @xmath153 & & & _ d0 _ & & & @xmath631 mcg 11 - 14 - 25a & ) & & & _ 0 _ & & & @xmath632 mcg 8 - 27 - 18 & @xmath153 & & & _ 0 _ & & & @xmath633 ngc 2549 & @xmath153 & & & _ dbd _ & ii & & @xmath634 ngc 2592 & @xmath153 & d & 0.6 & _ x0d _ & & 1 & @xmath635 ngc 2634 & @xmath153 & & & _ 0 _ & i & & @xmath636 ngc 2699 & @xmath153 & d & 0.6 & _ x0d0 _ & & 1 & @xmath637 ngc 2778 & @xmath153 & & & _ 0d0 _ & & & @xmath638 ngc 2824 & ? &
f & iii & _ x _ & & & @xmath639 ngc 2872 & @xmath153 & d & 0.8 & _ x0 _ & & & @xmath640 ngc 2902 & @xmath153 & f & ii & _ x0 _ & & & @xmath641 ngc 2950 & @xmath153 & & & _ d0d _ & & 2 & @xmath642 ngc 2986 & @xmath152 & & & _ ? 0 _ & & & @xmath643 ngc 3065 & @xmath153 & d & 0.4 & _ x0d0 _ & & & @xmath644 ngc 3078 & @xmath153 & d & 1.2 & _ xb _ & & & @xmath645 ngc 3193 & ) & & & _ 0 _ & & & @xmath646 ngc 3226 & ? & f & iii & _ x _ & & & @xmath647 ngc 3266 & @xmath153 & & & _ 0d0 _ & ii & 1 & @xmath648 ngc 3348 & @xmath152 & & & _ 0 _ & & & @xmath649 ngc 3377 & @xmath153 & f & i & _ d _ & & & @xmath650 ngc 3414 & @xmath153 & f & i & _ x0 _ & & & @xmath651 ngc 3595 & @xmath153 & & & _ dbd _ & & 1 & @xmath652 ngc 3613 & @xmath152 & & & _ db0 _ & & & @xmath653 ngc 3640 & ) & & & _ 0 _ & & & @xmath646 ngc 4121 &
@xmath153 & & & _ 0d0 _ & & 1 & @xmath642 ngc 4125 & ? & f & iii & _ x0 _ & & & @xmath654 ngc 4128 & @xmath153 & & & _ dbd _ & & & @xmath655 ngc 4168 & @xmath152 & f & ii & _ x0 _ & ii & & @xmath656 ngc 4233 & ? & d & 12.3 & _ x _ & & & @xmath657 ngc 4291 & @xmath152 & & & _ 0b _ & & & @xmath658 ngc 4365 & @xmath152 & & & _ ? d0 _ & & & @xmath659 ngc 4474 & @xmath153 & & & _ dbd _ & ii & & @xmath660 ngc 4478 & @xmath153 & & & _
db _ & & & @xmath661 ngc 4482 & @xmath153 & & & _ ? 0 _ & iii & & @xmath628 ngc 4494 & ? & d & 1.6 & _
x0 _ & & & @xmath662 ngc 4503 & @xmath153 & & & _ 0d0 _ & & 1 & @xmath663 ngc 4564 & @xmath153 & & & _ 0d _ & & & @xmath664 ngc 4589 & ) & f & ii & _ xb _ & & & @xmath665 ngc 4621 & @xmath153 & & & _
d _ & & & @xmath666 ngc 4648 & @xmath153 & d & 0.3 & _ 0d0 _ & & 1 & @xmath662 ngc 5017 & @xmath153 & d & 0.6 & _ x0 _ & & & @xmath667 ngc 5077 & ) & f & ii & _ xb _ & ii & & @xmath659 ngc 5173 & ? & f & iii & _ x _ & & & @xmath668 ngc 5198 & ) & & & _ 0 _ & & & @xmath669 ngc 5283 & ? & f & iii & _ x _ & & & @xmath670 ngc 5308 & @xmath153 & & & _ d0d _ & & & @xmath671 ngc 5370 & @xmath153 & & & _ 0d _ & & & @xmath672 ngc 5557 & ) & & & _ 0 _ & & & @xmath646 ngc 5576 & @xmath153 & & & _ 0b0 _ & & & @xmath659 ngc 5796 & @xmath153 & f & i & _ 0 _ & & & @xmath673 ngc 5812 & @xmath153 & d & 0.4 & _ x0 _ & & & @xmath674 ngc 5813 & @xmath152 & f & ii & _ x0 _ & & & @xmath633 ngc 5831 & @xmath153 & & & _ 0 _ & & & @xmath675 ngc 5846 & ? & f & iii & _ x0 _ & & & @xmath676 ngc 5898 & @xmath153 & f & i & _ 0 _ & & & @xmath677 ngc 5903 & @xmath152 & f & i & _ ? b _ & & & @xmath678 ngc 5982 & @xmath152 & & & _ 0b _ & & & @xmath679 ngc 6278 & @xmath153 & & & _ b0d _ & & 1 & @xmath680 ugc 4551 & @xmath153 & & & _ 0d _ & i & 1 & @xmath681 ugc 4587 & @xmath153 & f & i & _ 0 _ & & & @xmath682 ugc 6062 & @xmath153 & & & _ 0d0d _ & & 2 & @xmath683 lccccccc eso 378 - 20 & @xmath153 & 17.61 & 0.43 & 2.00 & 0.86 & 2.41 & 1.09 eso 437 - 15 & ? & & & & & & eso 443 - 39 & @xmath153 & 17.32 & 1.36 & 1.31 & 0.75 & 1.30 & 0.77 eso 447 - 30 & @xmath153 & 16.73 & 2.70 & 1.72 & 0.84 & 1.50 & 0.84 eso 507 - 27 & @xmath153 & 17.96 & 0.59 & 1.58 & 0.70 & 3.89 & 0.79 eso 580 - 26 & ? & & & & & & ic 875 & @xmath153 & 17.65 & 0.85 & 1.81 & 1.07 & 1.87 & 1.12 mcg 11 - 14 - 25a & ) & 16.59 & 0.69 & 2.13 & 0.00 & 0.56 & 0.49 mcg 8 - 27 - 18 & @xmath153 & 18.61 & 0.80 & 1.93 & 0.79 & 1.95 & 0.89 ngc 2549 & @xmath153 & 16.96 & 1.75 & 1.71 & 0.67 & 3.70 & 0.67 ngc 2592 & @xmath153 & 16.61 & 3.31 & 1.60 & 0.92 & 1.37 & 0.92 ngc 2634 & @xmath153 & 17.32 & 2.83 & 1.57 & 0.81 & 1.82 & 0.81 ngc 2699 & @xmath153 & 17.63 & 1.66 & 1.89 & 0.84 & 2.66 & 0.85 ngc 2778 & @xmath153 & 17.08 & 0.93 & 1.55 & 0.94 & 1.33 & 0.99 ngc 2824 & ? & & & & & & ngc 2872 & @xmath153 & 18.11 & 2.55 & 1.66 & 1.01 & 4.27 & 1.01 ngc 2902 & @xmath153 & 16.34 & 1.93 & 1.57 & 0.49 & 0.88 & 0.50 ngc 2950 & @xmath153 & 16.16 & 2.40 & 1.81 & 0.82 & 2.43 & 0.82 ngc 2986 & @xmath152 & 16.09 & 1.77 & 1.50 & 0.18 & 1.24 & 0.20 ngc 3065 & @xmath153 & 17.97 & 0.99 & 2.28 & 0.79 & 3.14 & 0.84 ngc 3078 & @xmath153 & 17.19 & 2.35 & 1.60 & 0.95 & 3.43 & 0.95 ngc 3193 & ) & 15.46 & 0.59 & 1.89 & 0.01 & 0.81 & 0.44 ngc 3226 & ? & & & & & & ngc 3266 & @xmath153 & 17.78 & 1.34 & 2.06 & 0.64 & 2.25 & 0.66 ngc 3348 & @xmath152 & 16.00 & 1.17 & 1.53 & 0.09 & 0.99 & 0.18 ngc 3377 & @xmath153 & 15.37 & 3.15 & 1.35 & 1.12 & 1.42 & 1.12 ngc 3414 & @xmath153 & 16.46 & 1.40 & 1.45 & 0.83 & 1.72 & 0.84 ngc 3595 & @xmath153 & 17.43 & 2.33 & 1.52 & 0.75 & 2.30 & 0.76 ngc 3613 & @xmath152 & 15.11 & 1.53 & 1.06 & 0.04 & 0.34 & 0.17 ngc 3640 & ) & 15.24 & 0.87 & 1.16 & 0.15 & 0.44 & 0.37 ngc 4121 & @xmath153 & 19.69 & 1.51 & 3.65 & 0.85 & 5.60 & 0.85 ngc 4125 & ? & & & & & & ngc 4128 & @xmath153 & 16.31 & 1.13 & 1.69 & 0.71 & 1.59 & 0.75 ngc 4168 & @xmath152 & 17.45 & 1.43 & 1.39 & 0.17 & 2.02 & 0.19 ngc 4233 & ? & & & & & & ngc 4291 & @xmath152 & 15.06 & 1.35 & 1.62 & 0.00 & 0.60 & 0.13 ngc 4365 & @xmath152 & 16.08 & 2.29 & 1.28 & 0.13 & 1.82 & 0.13 ngc 4474 & @xmath153 & 17.97 & 2.28 & 1.65 & 0.72 & 4.55 & 0.72 ngc 4478 & @xmath153 & 17.57 & 0.93 & 1.73 & 0.57 & 4.16 & 0.61 ngc 4482 & @xmath153 & 19.63 & 3.43 & 1.01 & 0.49 & 4.04 & 0.49 ngc 4494 & ?
& 15.21 & & 1.25 & & & ngc 4503 & @xmath153 & 16.53 & 1.77 & 1.30 & 0.64 & 1.65 & 0.65 ngc 4564 & @xmath153 & 15.89 & 1.43 & 1.27 & 0.80 & 1.28 & 0.81 ngc 4589 & ) & 15.64 & 0.82 & 1.44 & 0.26 & 0.71 & 0.46 ngc 4621 & @xmath153 & 16.57 & 2.53 & 1.40 & 0.96 & 3.75 & 0.96 ngc 4648 & @xmath153 & 15.87 & 3.73 & 1.54 & 0.92 & 1.02 & 0.92 ngc 5017 & @xmath153 & 16.92 & 2.97 & 1.59 & 1.12 & 1.48 & 1.12 ngc 5077 & ) & 16.49 & 1.09 & 1.67 & 0.23 & 1.61 & 0.30 ngc 5173 & ? & & & & & & ngc 5198 & ) & 14.79 & 2.61 & 1.13 & 0.23 & 0.16 & 0.42 ngc 5283 & ? & & & & & & ngc 5308 & @xmath153 & 15.55 & 0.39 & 1.27 & 0.82 & 0.83 & 0.96 ngc 5370 & @xmath153 & 17.84 & 0.78 & 1.50 & 0.62 & 1.63 & 0.71 ngc 5557 & ) & 16.18 & 0.80 & 1.77 & 0.14 & 1.21 & 0.33 ngc 5576 & @xmath153 & 15.79 & 0.79 & 1.73 & 0.36 & 1.41 & 0.51 ngc 5796 & @xmath153 & 16.19 & 0.83 & 1.67 & 0.41 & 1.06 & 0.56 ngc 5812 & @xmath153 & 16.61 & 1.25 & 1.67 & 0.59 & 1.84 & 0.62 ngc 5813 & @xmath152 & 15.77 & 1.90 & 1.33 & 0.00 & 0.73 & 0.03 ngc 5831 & @xmath153 & 16.90 & 0.47 & 1.84 & 0.33 & 1.78 & 0.64 ngc 5846 & ? & & & & & & ngc 5898 & @xmath153 & 16.32 & 1.23 & 1.57 & 0.41 & 1.24 & 0.46 ngc 5903 & @xmath152 & 16.84 & 1.80 & 1.48 & 0.13 & 1.59 & 0.14 ngc 5982 & @xmath152 & 15.60 & 0.99 & 1.52 & 0.00 & 0.74 & 0.18 ngc 6278 & @xmath153 & 15.69 & 0.76 & 1.62 & 0.55 & 0.60 & 0.77 ugc 4551 & @xmath153 & 17.06 & 2.19 & 2.16 & 0.51 & 2.26 & 0.51 ugc 4587 & @xmath153 & 15.48 & 1.04 & 1.24 & 0.72 & 0.19 & 0.90 ugc 6062 & @xmath153 & 18.35 & 0.90 & 1.81 & 0.80 & 2.75 & 0.85 lccccccc eso 378 - 20 & @xmath153 & 17.96 & 0.39 & 2.06 & 0.87 & 1.98 & 1.15 eso 437 - 15 & ? & & & & & & eso 443 - 39 & @xmath153 & 17.52 & 1.53 & 1.63 & 0.75 & 1.19 & 0.77 eso 447 - 30 & @xmath153 & 16.88 & 1.31 & 1.60 & 0.83 & 1.19 & 0.86 eso 507 - 27 & @xmath153 & 17.71 & 1.56 & 2.25 & 0.67 & 1.66 & 0.69 eso 580 - 26 & ? & & & & & & ic 875 & @xmath153 & 18.93 & 0.41 & 1.78 & 1.04 & 2.14 & 1.21 mcg 11 - 14 - 25a & ) & 16.69 & 0.70 & 2.21 & 0.00 & 0.56 & 0.51 mcg 8 - 27 - 18 & @xmath153 & 18.39 & 0.82 & 1.87 & 0.79 & 1.53 & 0.89 ngc 2549 & @xmath153 & 15.88 & 3.49 & 1.55 & 0.63 & 0.94 & 0.63 ngc 2592 & @xmath153 & 17.40 & 2.40 & 1.86 & 1.06 & 1.92 & 1.06 ngc 2634 & @xmath153 & 17.33 & 2.56 & 1.64 & 0.79 & 1.66 & 0.80 ngc 2699 & @xmath153 & 17.87 & 1.06 & 1.84 & 0.95 & 2.40 & 0.98 ngc 2778 & @xmath153 & 16.56 & 1.49 & 1.53 & 0.96 & 0.77 & 0.98 ngc 2824 & ? & & & & & & ngc 2872 & @xmath153 & 17.84 & 3.12 & 1.58 & 1.00 & 2.76 & 1.00 ngc 2902 & @xmath153 & 16.27 & 2.14 & 1.50 & 0.49 & 0.77 & 0.51 ngc 2950 & @xmath153 & 16.16 & 1.66 & 1.78 & 0.84 & 1.65 & 0.84 ngc 2986 & @xmath152 & 16.05 & 1.86 & 1.45 & 0.19 & 0.97 & 0.20 ngc 3065 & @xmath153 & 18.04 & 0.86 & 2.17 & 0.83 & 2.97 & 0.89 ngc 3078 & @xmath153 & 17.49 & 3.42 & 1.66 & 1.06 & 3.26 & 1.06 ngc 3193 & ) & 14.21 & 1.18 & 1.18 & 0.00 & 0.14 & 0.48 ngc 3226 & ? & & & & & & ngc 3266 & @xmath153 & 17.28 & 2.39 & 1.91 & 0.67 & 1.46 & 0.67 ngc 3348 & @xmath152 & 16.06 & 1.08 & 1.61 & 0.06 & 0.98 & 0.18 ngc 3377 & @xmath153 & 15.91 & & 1.29 & & & ngc 3414 & @xmath153 & 16.29 & 1.13 & 1.39 & 0.85 & 1.10 & 0.89 ngc 3595 & @xmath153 & 17.26 & 2.43 & 1.75 & 0.77 & 1.34 & 0.77 ngc 3613 & @xmath152 & 15.73 & 1.23 & 1.46 & 0.13 & 0.65 & 0.25 ngc 3640 & ) & 15.41 & 0.86 & 1.28 & 0.17 & 0.48 & 0.40 ngc 4121 & @xmath153 & 19.32 & 1.28 & 2.90 & 0.85 & 3.48 & 0.87 ngc 4125 & ? & & & & & & ngc 4128 & @xmath153 & 17.30 & 0.83 & 2.22 & 0.67 & 1.98 & 0.79 ngc 4168 & @xmath152 & 17.48 & 1.43 & 1.30 & 0.21 & 1.64 & 0.23 ngc 4233 & ? & & & & & & ngc 4291 & @xmath152 & 15.07 & 1.36 & 1.65 & 0.00 & 0.47 & 0.18 ngc 4365 & @xmath152 & 16.25 & 1.54 & 1.39 & 0.11 & 1.55 & 0.13 ngc 4474 & @xmath153 & 17.43 & 3.83 & 1.95 & 0.58 & 2.38 & 0.58 ngc 4478 & @xmath153 & 17.63 & 0.87 & 1.61 & 0.54 & 3.31 & 0.59 ngc 4482 & @xmath153 & 19.92 & 2.17 & 1.50 & 0.50 & 4.02 & 0.50 ngc 4494 & ?
& 15.58 & & 1.18 & & & ngc 4503 & @xmath153 & 16.89 & 1.21 & 1.60 & 0.63 & 1.70 & 0.66 ngc 4564 & @xmath153 & 16.90 & 1.13 & 1.93 & 0.78 & 2.10 & 0.82 ngc 4589 & ) & 15.64 & 0.71 & 1.45 & 0.18 & 0.54 & 0.47 ngc 4621 & @xmath153 & 17.52 & 0.69 & 1.49 & 0.96 & 4.81 & 1.00 ngc 4648 & @xmath153 & 16.50 & 3.16 & 1.75 & 1.03 & 1.29 & 1.03 ngc 5017 & @xmath153 & 17.75 & 1.81 & 1.68 & 1.19 & 2.12 & 1.19 ngc 5077 & ) & 16.48 & 1.15 & 1.74 & 0.25 & 1.19 & 0.33 ngc 5173 & ? & & & & & & ngc 5198 & ) & 14.81 & 2.04 & 1.15 & 0.21 & 0.14 & 0.52 ngc 5283 & ? & & & & & & ngc 5308 & @xmath153 & 18.39 & 0.85 & 2.18 & 0.86 & 3.95 & 0.92 ngc 5370 & @xmath153 & 18.61 & 1.26 & 2.51 & 0.64 & 2.40 & 0.68 ngc 5557 & ) & 15.74 & 1.11 & 1.54 & 0.18 & 0.67 & 0.33 ngc 5576 & @xmath153 & 15.57 & 0.91 & 1.76 & 0.36 & 0.89 & 0.53 ngc 5796 & @xmath153 & 16.35 & 0.82 & 1.71 & 0.40 & 1.05 & 0.57 ngc 5812 & @xmath153 & 16.69 & 1.22 & 1.73 & 0.58 & 1.91 & 0.62 ngc 5813 & @xmath152 & 15.88 & 1.61 & 1.41 & 0.00 & 0.74 & 0.05 ngc 5831 & @xmath153 & 15.16 & 0.99 & 1.34 & 0.39 & 0.27 & 0.65 ngc 5846 & ? & & & & & & ngc 5898 & @xmath153 & 16.63 & 0.90 & 1.61 & 0.38 & 1.35 & 0.49 ngc 5903 & @xmath152 & 16.66 & 2.30 & 1.35 & 0.12 & 0.98 & 0.13 ngc 5982 & @xmath152 & 15.26 & 1.91 & 1.42 & 0.05 & 0.46 & 0.12 ngc 6278 & @xmath153 & 16.31 & 1.11 & 1.96 & 0.65 & 0.86 & 0.76 ugc 4551 & @xmath153 & 17.39 & 1.95 & 2.57 & 0.49 & 2.48 & 0.49 ugc 4587 & @xmath153 & 15.97 & 0.72 & 1.47 & 0.63 & 0.23 & 0.92 ugc 6062 & @xmath153 & 18.46 & 1.05 & 2.05 & 0.80 & 2.16 & 0.85 | we present high resolution @xmath0-band images of the central regions of 67 early - type galaxies obtained with the wide field and planetary camera 2 ( wfpc2 ) aboard the hubble space telescope ( hst ) .
this homogeneously selected sample roughly doubles the number of early - type galaxies that have now been imaged at hst resolution , and complements similar data on the central regions of radio galaxies and the bulges of spiral galaxies .
our sample strikingly confirms the complex morphologies of the central regions of early - type galaxies which have become apparent from previous studies with hst .
in particular , we detect dust , either in the form of nuclear disks or with a filamentary distribution , in 43 percent of all galaxies , in good agreement with previous estimates .
in addition , we find evidence for embedded stellar disks in a remarkably large fraction of 51 percent . in 14 of those galaxies the disk - like structures are misaligned with the main galaxy , suggesting that they correspond to stellar bars in s0 galaxies .
we analyze the luminosity profiles of the galaxies in our sample , and classify galaxies according to their central cusp slope . to a large extent
we confirm the results from previous hst surveys in that early - type galaxies reveal a clear dichotomy : the bright ellipticals ( @xmath1 ) are generally boxy and have luminosity profiles that break from steep outer power - laws to shallow inner cusps ( referred to as ` core ' galaxies ) .
the fainter ellipticals , on the other hand , typically have disky isophotes and luminosity profiles that lack a clear break and have a steep central cusp ( referred to as ` power - law ' galaxies ) . the advantages and shortcomings of classification schemes utilizing the extrapolated central cusp slope @xmath2
are discussed , and it is shown that @xmath2 might be an inadequate representation for galaxies whose luminosity profile slope changes smoothly with radius rather than resembling a broken power - law . thus we introduce a new , alternative parameter , and show how this affects the classification .
in fact , we find evidence for an ` intermediate ' class of galaxies , that can not unambiguously be classified as either core or power - law galaxies , and which have central cusp slopes and absolute magnitudes intermediate between those of core and power - law galaxies .
it is unclear at present , however , whether these galaxies make up a physically distinct class or whether distance and/or resolution effects cause them to loose their distinct core or power - law characteristics . |
we start by analyzing the pure sc phase at @xmath54 , represented by @xmath55 .
the generally favored @xmath31 belongs to the 2d @xmath56 irreducible representation of the @xmath57 lattice point group @xcite .
the state can be written as a combination of @xmath58 , which gives @xmath59-wave intraband ( @xmath60 ) pairing , and @xmath61 , giving @xmath15-wave intraband pairing .
the @xmath62-wave solutions are degenerate at @xmath5 , but below @xmath5 the time - reversal symmetry breaking chiral combinations @xmath63 have the lowest energy .
this follows from a simple energy argument since the @xmath62 states have nodal quasiparticles , whereas the @xmath64 states are fully gapped , see fig . [
fig : syms ] .
the chiral @xmath65-wave state has a @xmath66 chern number , which can be viewed as the winding number for the intraband op around the brillouin zone center .
we can , alternatively , consider the symmetry of the intraband op around @xmath67 .
the @xmath65 state has @xmath68-wave symmetry around @xmath69 , but @xmath70-wave symmetry around @xmath71 @xcite , such that each valley contributes @xmath72 to @xmath73 .
the sign change between the valleys is dictated by the spin - singlet nature , which enforces even parity with respect to the zone center .
a finite @xmath16 adds the spin - triplet , odd - parity op @xmath74 to the intraband pairing . the existence of this spin - triplet component has dramatic consequences , due to the chirality and the @xmath75 phase shift of @xmath76 relative to @xmath60 .
adding @xmath76 to the @xmath59 state directly makes this state develop an imaginary part , such that it has @xmath77-wave symmetry around @xmath69 but the opposite @xmath78-wave chirality around @xmath71 , see fig .
[ fig : syms ] .
thus , for finite @xmath16 , the @xmath59 state becomes a time - reversal breaking mixed chirality sc state with opposite op winding in the two valleys , but @xmath79 when summed over the whole brillouin zone .
this mixed chirality state requires the combination of a spin - singlet ( @xmath80 ) and a spin - triplet ( @xmath81 ) component , since the spin - singlet pairing changes sign between @xmath69 and @xmath71 but the spin - triplet does not .
this is why a magnetic moment @xmath16 , with its accompanying spin - triplet @xmath76 , is necessary for generating mixed chirality @xmath0-wave superconductivity .
starting instead from the @xmath15-wave state just interchanges @xmath82 and @xmath83 . with the @xmath76 contribution , the @xmath59 state becomes fully gapped , as shown in fig .
[ fig : syms ] . quite the opposite happens for the @xmath65 state . at finite
@xmath16 it still has a @xmath66 winding .
however , the @xmath84 and @xmath76 intraband terms now start to cancel around @xmath71 , as seen in fig . [
fig : syms ] .
this results in very low - lying quasiparticle excitations around @xmath71 ( @xmath69 for @xmath85 ) as @xmath16 increases .
the quasiparticle energy spectrum shows that a phase transition must occur at some critical @xmath18 , between the chiral @xmath0-wave state favorable at @xmath54 and the mixed chirality @xmath0-wave state .
this preference for the mixed chirality state relies on a simple energy argument and is thus general and independent on a particular model . to find @xmath18 specifically for @xmath43 we minimize its free energy with respect to the three bond ops @xmath31 .
we plot in fig .
[ fig : pt](a ) the critical moment as a function of @xmath36 for the bare parameters , in order to directly compare with previous results for the @xmath19-@xmath20 model .
as seen , @xmath86 is small and also decreases towards half filling , even approaching zero for low doping .
this is highly advantageous for achieving the mixed chirality state , since the af moment is largest at half filling .
in fact , when comparing to the calculated magnetic moment in ref .
[ ] ( dashed line ) , we conclude that the mixed chirality state appears in a very large part of the phase diagram .
we also see little variation with @xmath87 , making the mixed chirality state robust .
having established the existence of a critical magnetic moment for entering the mixed chirality state , we now analyze the phase transition itself . for transparency we will from now on use the renormalized parameters @xmath20 and @xmath19 .
in fig . [
fig : pt](b ) we plot the bond ops as a function of @xmath16 .
the perfect @xmath65 solution at @xmath54 , with equal parts of @xmath59- and @xmath15-wave character ( inset ) , becomes a slightly imperfect @xmath65 state for @xmath88 , since the imaginary part of the bond ops is somewhat suppressed at finite @xmath16 .
however , as long as the imaginary part of the bond op is finite chirality is preserved . at @xmath18
there is a sudden jump in the bond ops , where the imaginary part disappears and the real part increases , such that the @xmath59-wave character of the bond op jumps from 0.56 to 1 with only a 0.6% change in the magnetization at @xmath18 .
this large discontinuous change of the op at @xmath18 is accompanied by a similarly discontinuous derivative in the free energy , strongly supporting a first order phase transition between the chiral @xmath0-wave state and the mixed chirality @xmath0-wave state .
the phase transition to the mixed chirality state is driven by minimizing the number of low - lying quasiparticle excitations .
the mixed chirality state is always fully gapped , with the gap increasing with @xmath16 , whereas the chiral state develops low - energy quasiparticle excitations at finite @xmath16 , as seen in fig .
[ fig : syms ] .
however , the phase transition takes place well before the chiral @xmath0-wave state develops zero - energy states .
thus , the system remains fully gapped in the whole af - sc coexistence phase , , as then the normal - state fermi surface only consists of the @xmath67 points .
] although it is possible that domain wall proliferation might mask the energy gap .
the fully gapped bulk energy spectrum is very interesting because of the topological nature of the phase transition . at @xmath54
, the chiral @xmath0-wave state belongs to symmetry class c , as it has full spin - rotation symmetry @xcite .
the c classification allows for @xmath89 different topological states in 2d . however , as soon as we add the af moment , only rotations around @xmath90 are left invariant .
this results in symmetry class a for finite @xmath16 , which also has @xmath89 topological classification @xcite .
the chiral @xmath0-wave state is therefore always classified by a @xmath91 winding number . on the other hand , the mixed chirality state at @xmath92 has opposite op windings in the two valleys .
thus , while it is topologically nontrivial in each valley , the topological invariant cancels when summed over the full brillouin zone .
the phase transition at @xmath18 is therefore a topological phase transition between two topologically distinct phases of matter .
topological phase transitions have widely been assumed to require the bulk energy gap to close , as that is the only generic way to change topological order @xcite .
the topological transition between the chiral and the mixed chirality @xmath0-wave states provides an explicit counterexample .
this is possible because the topological transition is the result of a first - order transition within the ginzburg - landau paradigm .
a defining property of topological states is the existence of edge modes , with the bulk - boundary correspondence equaling the number of edge states with the change of topological number at the edge @xcite . to investigate the edge band structure we solve self - consistently for the sc ops at every site in a thick ribbon . for @xmath93
the dirac valleys at @xmath94 each give rise to one state per edge , see fig .
[ fig : edge](a ) , which are copropagating in agreement with the change of topological number at the edge .
the figure shows the result for zigzag edges , but armchair edges behave very similarly @xcite . at @xmath95
the direction of the edge states at @xmath71 changes discontinuously due to the first - order transition , and the edge states instead become counterpropagating reflecting the mixed chirality , see fig .
[ fig : edge](b ) . reducing the width of the ribbon
allows the study of finite size effects .
we find @xmath18 to be significantly reduced in all but very thick ribbons .
for example , in a @xmath96 , @xmath97 ribbon , perfect chiral @xmath0-wave symmetry is reached already within @xmath98 unit cells from the edge .
still , @xmath99 in a ribbon 100 unit cells thick , compared to @xmath100 in the bulk .
thus , ribbons provide a route for inducing the mixed chirality state below the bulk @xmath18 .
for very thin ribbons we find that the phase transition is not sharply defined , instead there is gradual suppression of the imaginary part of the bond ops as @xmath16 increases .
reducing the ribbon thickness also hybridizes left and right edge states .
this is a scattering process within each dirac valley , which gaps the edge state spectra in both phases .
we find no notable difference between the thin ribbon energy gaps deep inside the chiral and the mixed chirality @xmath0-wave phases , see fig . [
fig : edge](c ) .
however , for @xmath101 low - lying bulk excitations exist at @xmath71 , resulting in a larger gap at @xmath71 just prior to the phase transition .
finally , we study the effect of disorder .
we find no influence of weak to moderate disorder on @xmath18 .
the two chiral @xmath0-wave edge states are topologically protected and also not sensitive to disorder @xcite .
the mixed chirality edge states are locally protected in each valley and thus only intervalley scattering can open a back - scattering channel .
intervalley scattering only exists for atomically sharp disorder , and can be relatively rare in clean samples .
still , we can demonstrate the sensitivity to intervalley scattering by using strong anderson disorder ; an atomically strongly fluctuating chemical potential @xmath102 , with @xmath103 distributed randomly within the interval @xmath104 $ ] .
we create disordered samples by using a @xmath105 long ( perpendicular to edges ) ribbon supercell , which reduces the 1d brillouin zone to @xmath106 $ ] . with the supercell momentum always preserved ,
intervalley scattering is only present for edge states centered at @xmath107 .
although this sounds as if it requires fine tuning , we find that strong disorder often locks the edge state to @xmath108 , as exemplified in fig . [
fig : edge](d ) .
while the clean system has its zero - energy crossing located away from @xmath108 , strong disorder gives a small gap ( @xmath109 of the sc energy gap ) in the edge state centered at @xmath108 . in summary ,
we have found a mixed chirality @xmath0-wave sc state in the af - sc coexistence region in strongly correlated honeycomb materials . at a finite af moment
there is a first - order topological phase transition from the chiral to the mixed chirality @xmath0-wave sc state .
the mixed chirality state hosts two counterpropagating edge modes , protected in the absence of intervalley scattering .
general rg and energy arguments make our results valid beyond the @xmath19-@xmath20 model . newly developed numerical approaches for
interaction - driven time - reversal broken superconductivity @xcite could offer precise treatments of an even wider range of materials .
we are very grateful to a .-
tremblay and w. wu for discussions related to this work and acknowledge financial support from the swedish research council ( vetenskapsrdet ) , the gran gustafsson foundation , the swedish foundation for strategic research ( ssf ) , and the palm labex paris - saclay .
this work has also benefited from discussions at cifar meetings in canada , nordita , a cargse summer school on topological phases , and kitp santa barbara , supported in part by the national science foundation under grant no .
nsf phy11 - 25915 . | we find a mixed chirality @xmath0-wave superconducting state in the coexistence region between antiferromagnetism and interaction - driven superconductivity in lightly doped honeycomb materials .
this state has a topological chiral @xmath1-wave symmetry in one dirac valley but @xmath2-wave symmetry in the other valley and hosts two counterpropagating edge states , always protected in the absence of intervalley scattering .
a first - order topological phase transition , with no bulk gap closing , separates the chiral @xmath0-wave state at small magnetic moments from the mixed chirality @xmath0-wave phase .
superconducting ( sc ) pairing driven by strong electron repulsion in two dimensions ( 2d ) has spin - singlet @xmath0-wave symmetry and appears close to , or even coexists with , an antiferromagnetic ( af ) phase in materials ranging from cuprates to heavy fermion compounds and organic superconductors @xcite . in materials with
three- and sixfold rotational lattice symmetries , the two @xmath0-wave states , @xmath3 and @xmath4 , are dictated to be degenerate at the transition temperature @xmath5 while at lower temperatures the chiral @xmath6 combinations are favored @xcite . the chiral @xmath0-wave state is fully gapped , topologically nontrivial with finite chern ( or winding ) number , and has two chiral edge states .
the sixfold symmetric honeycomb lattice near half filling adds further versatility by having two disjoint fermi surfaces , or dirac valleys , centered at the inequivalent brillouin zone corners @xmath7 .
recently , several honeycomb materials have been proposed to be chiral @xmath0-wave superconductors near half filling , including in@xmath8cu@xmath9vo@xmath10 @xcite , @xmath11-cu@xmath9v@xmath9o@xmath12 @xcite , srptas @xcite , mos@xmath9 @xcite , graphene and silicene @xcite , and ( 111 ) bilayer sriro@xmath8 @xcite .
two fermi surfaces allow for the tantalizing speculation of having @xmath13 symmetry in one valley but @xmath14 symmetry in the other @xcite ; a novel state with _ mixed chirality _ even in a translationally invariant system . however , as worked out in ref .
, the sign change of the @xmath15 component between the valleys requires it to be spin - triplet , which is incompatible with the distinctly spin - singlet mechanisms creating @xmath0-wave superconductivity . in this rapid communication
we show that mixed chirality @xmath0-wave superconductivity is present in the coexistence region between af and @xmath0-wave sc order in strongly correlated honeycomb materials . a finite af moment @xmath16 in a spin - singlet superconductor
is known to spontaneously generate a so - called @xmath17-triplet state .
we find that this spin - triplet component facilitates a phase transition at a critical @xmath18 from the chiral @xmath0-wave state to the mixed chirality @xmath0-wave state .
@xmath18 is well within the af - sc coexistence region previously reported for hole - doped honeycomb mott insulators , but where a mixed chirality state was never considered @xcite .
the mixed chirality state is topological in each dirac valley , although the topological number cancels in the full brillouin zone .
moreover , the phase transition between the chiral and mixed chirality @xmath0-wave phases occurs without the bulk energy gap closing , otherwise considered a necessity for topological phase transitions @xcite . at the phase transition
the two copropagating edge states of the chiral @xmath0-wave state are discontinuously transformed into two counterpropagating states , which are protected in the absence of intervalley scattering .
these findings establish both the existence and properties of the highly unconventional mixed chirality sc state . in the same way that the chiral @xmath0-wave sc state has many properties common with a quantum hall state @xcite ,
the mixed chirality @xmath0-wave state is similar to a quantum valley hall state @xcite .
more specifically , we model electron correlations in the limit of large on - site hubbard repulsion , which are well described within the @xmath19-@xmath20 model @xcite .
in@xmath8cu@xmath9vo@xmath10 @xcite and @xmath11-cu@xmath9v@xmath9o@xmath12 @xcite are two materials recently proposed to be well described by the @xmath19-@xmath20 model on the honeycomb lattice , but general arguments make our results also applicable to other chiral @xmath0-wave sc honeycomb compounds , as well as bilayer honeycomb materials @xcite .
we study the @xmath19-@xmath20 model within renormalized mean - field theory ( rmft ) @xcite .
even if rmft only provides a mean - field treatment , its sc state on the honeycomb lattice has been shown to agree with both quantum monte carlo ( qmc ) @xcite and functional renormalization group ( frg ) @xcite calculations .
the sc hamiltonian thus reads : @xmath21 with @xmath22 @xmath23 the annihilation operator on the honeycomb sublattice a ( b ) , using @xmath24 unit cells . for transparency
we work mainly with the renormalized parameters , with the hopping amplitude @xmath25 and the exchange coupling @xmath26 , where the prime indicates the bare values and @xmath27 are the statistical weighting factors handling the gutzwiller projection to single - occupancy states within rmft @xcite .
@xmath28 further renormalizes the kinetic energy and is assumed isotropic in space .
the chemical potential @xmath29 is set by fixing the filling fraction @xmath30 .
the mean - field sc order parameters ( ops ) @xmath31 @xmath32 on the three nearest - neighbor bonds @xmath33 can be expressed compactly as @xmath34 .
each op is independently determined by the self - consistency equations @xmath35 . as is well known from graphene @xcite , the normal - state band structure at small @xmath36
consists of two dirac valleys centered at the brillouin zone corners @xmath37 and @xmath38 , where @xmath39 is the length of the unit cell vectors . beyond the sc state
, it is also well known that the @xmath19-@xmath20 model displays af order close to half filling , with the mean - field order parameter @xmath40 and @xmath41 .
coexistence of sc and af orders on the honeycomb lattice has recently been found in both mean - field theory @xcite and beyond @xcite .
this coexistence region can naturally be described by the combined hamiltonian @xmath42 .
since we are primarily interested in the behavior of the sc state , and since the af order is not notably affected by the sc order @xcite , we can safely use a fixed @xmath16 ( but scanning possible values ) and solve @xmath43 self - consistently for @xmath31 . to proceed , we first fourier transform and then rewrite @xmath43 in the basis where the kinetic and magnetic terms are fully diagonal : , eq .
reduces to the hamiltonian used for the sc state on the honeycomb lattice , see e.g. ] @xmath44 here @xmath45 @xmath46 is the annihilation operator in the lower ( upper ) band with band dispersion @xmath47 , where @xmath48 and @xmath49 .
sc pairing is present through the intraband terms @xmath50 and an interband term @xmath51 expanding the partition function for @xmath43 to order @xmath52 gives the same spin - triplet term @xmath53 .
thus , a renormalization group ( rg ) flow generates the same induced pairing , making its appearance general and not only tied to the @xmath19-@xmath20 model .
the interband pairing is not important and we thus focus on the intraband pairing . |
caldero - chapoton - schiffler defined in @xcite categories arising from homotopy classes of paths between two vertices of a regular @xmath4-sided polygon .
independently , buan - marsh - reiten - reineke - todorov defined cluster categories as certain orbit categories of the bounded derived category of hereditary algebras , see @xcite .
the latter are algebras arising from oriented graphs @xmath5 with no oriented cycles .
when @xmath5 is an orientation of a dynkin graph of type @xmath6 , the category constructed in @xcite coincides with the one of @xcite . in this paper
we model a number of different orbit categories arising from orientations of tree graphs @xmath7 in geometric terms .
the description we propose here is based on the idea of doubling the set of oriented diagonals in a given regular polygon and combine the dynamics of these two sets in an appropriate way . among the additive categories we can model in this way we find cluster categories of type @xmath7 , as well as other triangulated categories of various calabi - yau dimension .
all these categories arise as mesh categories of a translation quiver whose vertices are single coloured oriented and paired diagonals in a polygon @xmath8 .
we will see that polygons of different sizes yield different categories and these are calabi - yau of dimension two only in few exceptional cases .
the advantage of our approach is that we can model the combinatorics of these various categories in terms of configurations of diagonals in @xmath8 . for a tree diagram @xmath7
let @xmath9 be cluster category of type @xmath7 defined as the orbit category @xmath10 .
then for @xmath11 and a regular @xmath4-gon @xmath8 , we construct an additive category @xmath12 of coloured oriented single and paired diagonals of @xmath8 . with this notation
our first main result can be stated as follows .
we have the following equivalences of additive categories : @xmath13 when @xmath7 is not of dynkin type we obtain an equivalence from the category @xmath14 associated to the infinite sided - polygon @xmath15 to the full subcategory of the cluster category @xmath9 with indecomposable objects in the transjective component of the ar - quiver of @xmath9 .
these equivalences enable us to investigate the combinatorics of the cluster category in geometric terms .
more precisely , we are able to describe all 833 cluster tilting sets of the cluster category of type @xmath16 @xmath17 , as cluster configuration of six single coloured oriented and paired diagonals in a heptagon
. the strategy will be to first determine two fundamental families of cluster configurations @xmath18 and @xmath19 and deduce the remaining cluster tilting sets using the rotation inside @xmath8 , induced from the auslander - reiten translation @xmath20 in @xmath17 , as well as a symmetry @xmath21 of our model . in a heptagon
* 350 different cluster configurations have one long paired diagonal , and they arise from @xmath18 through @xmath20 . * 483 other cluster configurations arise from @xmath19 through @xmath21 and @xmath20 .
this classification allows us to deduce the following result .
all 833 cluster tilting sets of @xmath17 can be expressed as configurations of six non - crossing coloured oriented single and paired diagonals inside two heptagons .
the previous results enable us to deduce geometrical moves describing the mutation process between cluster tilting objects in @xmath17 .
the geometrical moves we find extend the mutation process of cluster categories of type @xmath6 , as described by caldero - chapoton - schiffler in @xcite , to the setting of coloured oriented diagonals .
we will see that in many cases mutations inside @xmath17 correspond to flips of coloured oriented diagonals in @xmath8 . in the final part of the paper we use the previous results to categorify geometrically cluster algebras of type @xmath22 .
we also provide a geometric description of the mutation rule .
* relation to previous work : * geometrical models for cluster categories of other types have been investigated also in @xcite . moreover , with an appropriate paring of coloured oriented single and paired diagonals in an even sided polygon we recover the categorified geometric realisation of cluster algebras of type @xmath3 given by fomin - zelevinsky in @xcite .
the number of clusters in a cluster algebra of finite type was first computed in @xcite . under the bijection of @xcite one
deduces the number of cluster tilting sets in the corresponding cluster categories .
an explicit complete description of the cluster - tilting objects in the cluster category of type @xmath1 and @xmath22 however is new .
in addition , fomin - pylyavskyy used in @xcite polygons to describe the cluster algebra structure in certain rings of @xmath23-invariants .
more precisely , they construct invariants determined by tensor graphs associated to diagonals of polygons .
using a heptagon , fomin - pylyavskyy also model the cluster algebra structure of the homogeneous coordinate ring @xmath24 $ ] of the affine cone over the grassmannian @xmath25 of three dimensional subspaces in a seven dimensional complex vector space . by a result of scott , @xcite
, it is known that the ring @xmath24 $ ] is a cluster algebra type of @xmath1 .
the approach of @xcite however is different then the one we propose here , as it relies on relations satisfied by tensor graphs , called skein relations of tensor graphs .
* organisation of the article : * in section 2 , we state some preliminary results and definitions .
in particular , we state the fundamental properties of orbit categories and we remind the action of the shift functor on the auslander - reiten quiver of @xmath26 , for @xmath5 an orientation of a simply laced dynkin diagram . in section 3 , we construct the additive category @xmath12 associated to a regular ( n+3)-gon @xmath8 , where @xmath27
. the objects of @xmath12 will be single coloured oriented and paired diagonals of @xmath8 .
the morphism spaces are generated by minimal rotations in @xmath8 , modulo certain equivalence relations . in section 4
we prove the equivalences of additive categories stated in theorem [ eqofcat ] .
further equivalence of additive categories will also be discussed . in proposition [ isoinfty ] we show that there is an equivalence between the category @xmath14 associated to an infinite sided polygon and the full subcategory of the cluster category @xmath9 , whose indecomposable objects belong to the transjective component of the ar - quiver of @xmath9 . in section 5 ,
we describe the combinatorics of @xmath17 geometrically inside a heptagon @xmath8 . in theorem
[ count ] and theorem [ confthm ] we describe all cluster tilting sets of @xmath17 in terms of cluster configurations of coloured oriented diagonals of @xmath8 . in proposition [ int ] we also describe all @xmath28-spaces of @xmath17 using curves of coloured oriented diagonals of @xmath8 . finally , results concerning the mutation process of cluster configurations will be stated , see proposition [ 378 ] .
in section 6 we point out further applications of our work .
in particular we deduce a geometric additive categorification of cluster algebras of type @xmath22 .
moreover , we describe how our construction can be used to understand cluster tilting sets inside the cluster categories of type @xmath29 and @xmath30 , as well as cluster tilting sets in the transjective component of the ar - quiver of cluster categories associated to more general tree diagrams @xmath7 .
let @xmath31 be an algebraically closed field and let @xmath5 be an acyclic quiver .
let @xmath32 be the abelian category of @xmath31-finite dimensional right - modules over the path algebra @xmath33 .
let @xmath34 be the bounded derived category of @xmath32 endowed with the shift functor @xmath35 and the _ auslander - reiten translation _
@xmath36 characterised by @xmath37 for all @xmath38 .
we are interested in the orbit categories @xmath40 of @xmath39 , @xmath41 , generated by the action of cyclic group generated by the auto - equivalences @xmath42 .
the objects of @xmath40 are the same as the objects of @xmath39 and @xmath43 morphisms are composed in a natural way . when @xmath44 , @xmath45 is the _ cluster category of type @xmath5 _ defined in @xcite , and independently in @xcite in geometric terms for @xmath5 of type @xmath6 . in all other cases
@xmath46 is the _
@xmath47-repetitive cluster category _
studied by the author in @xcite for @xmath5 of type @xmath6 , and introduced by zhu in @xcite for @xmath5 an acyclic quiver .
like @xmath39 , the categories @xmath46 are krull - schmidt and have finite dimensional @xmath48-spaces .
the categories @xmath46 are triangulated categories , and the projection functor @xmath49 , @xmath50 is a triangle functor , see ( * ? ? ?
* theorem 1 ) .
the induced shift functor is again denoted by @xmath51 .
moreover , the categories @xmath46 have ar - triangles and the ar - translation @xmath20 is induced from @xmath39 .
the categories @xmath46 also have the calabi - yau property , i.e. @xmath52 as triangle functors , here we identify @xmath53 with the serre functor of @xmath46 . in particular , in @xmath54 we have that @xmath55 and @xmath56 , hence @xmath54 is calabi - yau of dimension @xmath57 . in @xmath46
we have that @xmath58 and @xmath59 in the above isomorphism of triangle functors , thus @xmath46 is said to be a calabi - yau category of fractional dimension @xmath60 .
notice that a calabi - yau category of fractional dimension @xmath60 is in general not a calabi - yau category of dimension 2 . in this paper @xmath61
, moreover we adopt the convention @xmath62 .
a _ stable translation quiver _ @xmath63 in the sense of riedtmann , @xcite , is a quiver @xmath64 without loops nor multiple edges , together with a bijective map @xmath65 called _ translation _ such that for all vertices @xmath66 in @xmath64 the set of starting points of arrows which end in @xmath66 is equal to the set of end points of arrows which start at @xmath67 .
for @xmath63 one defines the mesh category as the quotient category of the additive path category of @xmath64 by the mesh ideal , see for example @xcite .
in particular , the mesh category of @xmath63 is an additive category . in the next result ,
let @xmath68 be the repetitive quiver of @xmath5 , see ( * ? ? ?
* i,5.6 ) for a reminder on this construction .
let @xmath69 be the automorphism defined on the vertices @xmath70 of @xmath71 by @xmath72 , for @xmath73 , @xmath74 a vertex of @xmath5 .
* i.5.5)[thm presentation derived cat ] let @xmath33 be a finite dimensional hereditary @xmath31-algebra . if @xmath5 is an orientation of * a simply laced dynkin graph , @xmath75 is isomorphic ( as stable translation quiver ) to @xmath71 . * an affine graph , @xmath75 splits into components of the form @xmath76 and @xmath77 , for some @xmath78 . * a wild graph , the components of @xmath75 are of the form @xmath79 and @xmath80 . let @xmath81 be the full subcategory of @xmath39 of indecomposable objects .
* i,5.6)[happel2 ] let @xmath5 be an orientation of a simply laced dynkin graph .
the mesh category of @xmath82 is equivalent to @xmath83 .
a first important consequence of this result is that the ar - quiver of @xmath39 is independent of the orientation of @xmath5 . in this section
let @xmath5 be an orientation of a simply - laced dynkin graph .
below we point out some known facts about the induced action of @xmath51 and @xmath20 on @xmath68 taken from ( * ? ? ?
4 ) , see also ( * ? ? ?
4 ) . these considerations , together with theorem [ thm presentation derived cat ] , will enable us to determine the precise shape of the ar - quiver of various orbit categories investigated in the sequel .
the induced action of @xmath20 on @xmath71 is always an horizontal shift to the left .
the induced action of @xmath51 on @xmath84 coincides with a shift of @xmath85 units to the right , composed with a reflection along the horizontal central line of @xmath84 .
on @xmath86 the action of @xmath51 agrees with @xmath87 composed with an order two automorphism @xmath88 defined on @xmath89 when @xmath90 is odd .
while on @xmath91 the action of @xmath51 coincides with the action of @xmath92 where @xmath88 is an automorphism of order two defined on @xmath91 . moreover , @xmath51 acts as @xmath93 on @xmath94 , and as @xmath95 on @xmath96 .
the geometrical model for the cluster categories of type @xmath0 we propose in this paper is motivated from the following idea : glue together two copies of ar - quiver of @xmath98 .
let us remind the reader some facts about this category .
the previous discussion implies that @xmath99 , for @xmath100 .
moreover , @xmath101 can be modelled using diagonals in polygons as done in @xcite .
when @xmath102 , the quiver @xmath103 can entirely be modelled using oriented diagonals in a regular @xmath4-gon . in figure [ ara4 ]
an illustration of this construction is provided for @xmath104 . .
throughout the rest of the paper let @xmath7 be an orientation of a finite connected graph with @xmath105 vertices and three legs .
we assume the legs of @xmath7 to have @xmath106 , resp .
@xmath107 , resp . @xmath108 vertices and that one vertex of @xmath7 has three neighbours .
we say that a tree @xmath7 is _ symmetric _ if @xmath109 . unless specified otherwise , for @xmath110 let @xmath8 be a regular @xmath4-gon with vertices numbered in the clockwise order by the group @xmath111 . for vertices @xmath112 of @xmath8
we write @xmath113 if @xmath114 is between @xmath74 and @xmath31 in the clockwise order .
moreover , we denote by @xmath115 the unoriented diagonal of @xmath8 joining the vertices @xmath74 and @xmath114 and by @xmath116 $ ] the oriented diagonal of @xmath8 starting at @xmath74 and ending in @xmath114 .
we do not consider boundary segments as oriented diagonals .
we start describing the geometric construction leading to the modelling of a number of orbit categories of @xmath117 arising from orientations of symmetric trees @xmath7 . to begin
the construction we double the set of oriented diagonals of @xmath8 , and distinguish each set with colours using subscripts @xmath118 , @xmath119 e.g. @xmath120_r$ ] is the red diagonal linking the vertex 1 to 3 of @xmath8 . for every vertex @xmath74 of @xmath8 we form the following @xmath121 pairs of coloured oriented diagonals : @xmath122_p=&[i+2,i]_p:=\ { [ i , i+2]_r,[i+2,i]_b\}\\ [ i , i+3]_p=&[i+3,i]_p:=\ { [ i , i+3]_r,[i+3,i]_b\}\\ & \dots\\ [ i , i+r+2]_p=&[i+r+2,i]_p:=\ { [ i , i+r+2]_r,[i+r+2,i]_b\}.\end{aligned}\ ] ] when @xmath123 , we assume that there are no paired diagonals .
moreover , let us point out that @xmath116_p\neq[j , i]_p$ ] .
next we define a subset of @xmath124 paired , @xmath125 single red and @xmath126 single blue oriented diagonals of @xmath8 as follows : @xmath127_p , \dots,[i , i+r+2]_p , \\ & [ i , i+r+3]_r,\dots,[i , i+r+s+2]_r , \\ & [ i+r+3,i]_b , \dots,[i+r+t+2,i]_b,\hspace{0,3 cm } \textrm { $ i$ vertex of } \pi\bigg\}.\end{aligned}\ ] ] once coloured oriented diagonals are paired , they stop existing as single coloured oriented diagonals in @xmath128 .
consider the subset of elements of @xmath128 given by @xmath129_p , \dots,[1,r+3]_p , [ 1,r+4]_r,\dots,[1,r+s+3]_r , [ r+4,1]_b , \dots,[r+t+3,1]_b\}$ ] .
then on the one side elements of @xmath130 are in bijection with the vertices of @xmath7 . on the other side ,
elements of @xmath130 give rise to a triangulation of a region inside @xmath8 homotopic to a regular @xmath131-gon , resp . to a @xmath132-gon . in figure [ projdiag ]
an illustration of this situation is provided .
coloured oriented diagonals with the same label are identified .
the black vertices in the figure represent the vertices of @xmath7 .
the dotted lines are the edges of @xmath7 .
vertices on , and edges between , identified paired diagonals give rise to one vertex , and one edge , in @xmath7 .
[ ] [ ] [ 0.65]@xmath74 [ ] [ ] [ 0.65]@xmath133 [ ] [ ] [ 0.65]@xmath134 [ ] [ ] [ 0.65]@xmath135 [ ] [ ] [ 0.65]@xmath136 [ ] [ ] [ 0.65]@xmath137 [ ] [ ] [ 0.65]@xmath138 [ ] [ ] [ 0.65]@xmath139 [ ] [ ] [ 0.65]@xmath140 [ ] [ ] [ 0.65]@xmath141 [ ] [ ] [ 0.5]@xmath142 .,title="fig : " ] our next aim is to define two automorphisms : @xmath88 and @xmath20 , acting on the set of coloured oriented single an paired diagonals of @xmath8 associated to a tree @xmath7
. the first automorphism is induced from the graph automorphism of a symmetric tree , hence only defined on @xmath143 the definition of the second automorphism depends on the parity of the number of sides of @xmath8 .
. then we define @xmath145 as the automorphism of order two given by @xmath146_c\big):=\begin{cases } [ j , i]_b & \textrm{if } c = r , \\ [ j , i]_r & \textrm{if } c = b,\\ [ i , j]_p & \textrm{otherwise . } \end{cases}\ ] ] [ ] [ ] [ 0.5]@xmath147 [ ] [ ] [ 0.5]@xmath57 [ ] [ ] [ 0.5]@xmath148 [ ] [ ] [ 0.5]@xmath149 [ ] [ ] [ 0.5]@xmath150 [ ] [ ] [ 0.5]@xmath151 [ ] [ ] [ 0.5]@xmath152 [ ] [ ] [ 0.5]@xmath88 .,title="fig : " ] moreover , we define the automorphism @xmath153 as follows : if @xmath154 , then @xmath155_c):=[i-1,j-1]_c . $ ] if @xmath109 , then @xmath156_c):=\begin{cases } \rho^{(n+3)}\big([i-1,j-1]_c\big ) & \textrm { if } [ i , j]_c\in \pi_{r , t , t}\vert_1 \\ [ i-1,j-1]_c & \textrm { otherwise . } \end{cases}\ ] ] geometrically , the action of @xmath20 is given by the anticlockwise rotation through @xmath157 around the centre of @xmath8 , on all elements different then the diagonals in @xmath158 associated to a symmetric tree in a polygon with an odd number of sides .
on the latter the rotation is followed by the simultaneous change of colour and orientation .
[ quivegamma ] minimal clockwise rotations for unoriented diagonals have been introduced in @xcite with the aim of modelling irreducible morphisms in the cluster category @xmath159 . following the spirit of @xcite
we now define minimal rotations between diagonals of @xmath128 .
let @xmath160 be non - neighbouring vertices of @xmath8 and let @xmath161 .
let @xmath128 be the set of coloured oriented single and paired diagonals in @xmath8 associated to an asymmetric tree @xmath7 .
then the following three operations between diagonals in @xmath128 are called _ minimal clockwise rotation _ : @xmath162_c\rightarrow [ k , l+1]_c$ ] and @xmath162_c\rightarrow[k+1,l]_c$ ] and @xmath163_r\ar[rd]&&\\ & [ k , k+r+2]_p \ar[ru]\ar[rd ] & & [ k+1,k+r+3]_p . \\ & & [ k+r+3,k]_b\ar[ur]&&\\ } \ ] ] next , let @xmath164 be the set of coloured oriented single and paired diagonals in @xmath8 associated to a symmetric tree @xmath165 .
then minimal clockwise rotations are defined as before with the following adjustment : when @xmath162_c $ ] , @xmath166_p$ ] are in @xmath167 and @xmath8 is odd sided , then we also change simultaneously the colour an the orientation . in figure
[ d4 ] and figure [ are6 ] illustrations of the three minimal rotations described above can be found .
the next remark can be used to model orbit categories arising from orientations of tree graphs @xmath168 where one vertex has @xmath169 neighbours .
there are three minimal rotations linking the three types of coloured oriented single and paired diagonals of @xmath170 , namely the red , the blue and the paired ones . starting with oriented diagonals coloured in @xmath171 ways
, one can describe the @xmath169 minimal rotations linking the @xmath169 types of coloured oriented diagonals with a similar diagram as above .
let @xmath172 be the quiver whose vertices are the elements of @xmath128 .
an arrow between two vertices of @xmath172 is drawn whenever there is a minimal clockwise rotation linking them .
no arrow is drawn otherwise . concerning the shape of @xmath172 we remark that @xmath172 always lies on a cylinder , except when @xmath8 is odd - sided , and @xmath109 .
then we say that @xmath173 lies on a mbius strip , since the @xmath20-orbits of single coloured oriented diagonals are twice as long as the @xmath20-orbits of paired diagonals . in figure [ d4 ] the quiver @xmath174 of coloured oriented diagonals in a hexagon , as well as the quiver @xmath175 associated to a heptagon
are illustrated . in each quiver
we indicate in the last slice the identifications occurring . on both quivers
the action of @xmath20 is always given by a shift to the left .
[ ] [ ] [ 0.5]@xmath147 [ ] [ ] [ 0.5]@xmath57 [ ] [ ] [ 0.5]@xmath148 [ ] [ ] [ 0.5]@xmath149 [ ] [ ] [ 0.5]@xmath150 [ ] [ ] [ 0.5]@xmath151 corresponds to the ar - quiver of the orbit category @xmath176 . , title="fig : " ] [ ] [ ] [ 0.5]@xmath147 [ ] [ ] [ 0.5]@xmath57 [ ] [ ] [ 0.5]@xmath148 [ ] [ ] [ 0.5]@xmath149 [ ] [ ] [ 0.5]@xmath150 [ ] [ ] [ 0.5]@xmath151 [ ] [ ] [ 0.5]@xmath152 corresponds to the ar - quiver of the orbit category @xmath177 .
, title="fig : " ]
in this section we show that the construction of @xmath173 and @xmath172 allow us to model geometrically properties of a number of additive categories .
let @xmath88 be the automorphism of @xmath165 and let @xmath11 .
then we can show the following opening result .
[ propt ] the quiver @xmath178 is a stable translation quiver .
we have to show three things .
first , that @xmath172 is connected , has no loops , and is locally finite .
second , that for every vertex @xmath179 of @xmath172 the number of arrows going to @xmath179 equals the number of arrows leaving @xmath179 .
third , that the map @xmath20 is bijective .
all the above properties follow from the construction .
let us first consider the symmetric case .
then it is not hard to see that when @xmath8 is even sided , @xmath180 and @xmath181 since we consider oriented arcs .
thus , we deduce that @xmath182 and @xmath183 are stable translation quivers .
forming pairs of coloured oriented diagonals results in gluing these two quivers along @xmath106 disjoint @xmath20-orbits , and the above properties are preserved .
when @xmath8 is odd sided one can check that the modifications in the definition of minimal clockwise rotations and in the definition of @xmath20 are such that the resulting quiver has the claimed properties . finally , it is not hard to see that also @xmath172 associated to an asymmetric tree is a stable translation quiver .
[ isoar ] let @xmath184 be an additively finite krull - schmidt category .
let @xmath64 be a connected component of the ar - quiver of @xmath184 .
assume that @xmath185 , resp . @xmath186 .
then there is an isomorphism of stable translation quivers @xmath187 for @xmath172 associated to a regular @xmath4-gon .
the claim follows from the proof of lemma [ propt ] , since we saw that the following are isomorphisms of stable translation quivers : @xmath188 , @xmath189 for all integers @xmath190 and @xmath90 as above .
our next goal is to define a translation quiver @xmath191 obtained from @xmath192 associated to a symmetric tree @xmath165 after folding @xmath192 along its central line . for this consider
again the graph automorphism @xmath88 induced by the map simultaneously changing colour and orientation of the diagonals in @xmath164 .
then the vertices of @xmath191 are the @xmath88-orbits of vertices of @xmath173 , i.e. the pairs @xmath193_r , [ i+j , i]_b \}$ ] , for @xmath194 in vertices of @xmath8 .
the arrows in @xmath191 are always single and coincide with minimal clockwise rotation around a common vertex of @xmath8 linking pairs of coloured oriented diagonals .
the translation on @xmath191 is induced from the translation in @xmath173 and given by the anti clockwise rotation through @xmath157 around the center of @xmath8 . clearly , @xmath195 is a surjective map of stable translation quivers . as a corollary of theorem [ isoar ]
we obtain the geometrical modelling of cluster categories @xmath9 where @xmath7 is an arbitrary tree . since the number of connected components of the ar - quiver of @xmath196 varies with the shape of @xmath165
, we proceed considering two cases . in this section
we focus on tree graphs of dynkin type , in section [ gen ] the general case will be treated . in view of theorem [ isoe6 ] below
let @xmath12 be the additive category generated by the mesh category of @xmath172 , for @xmath197 .
then we can show the main result of this section .
[ eqofcat][isoe6 ] we have the following equivalences of additive categories @xmath198 since the full subcategories of indecomposable objects of the orbit categories we consider are equivalent to the mesh category of their ar - quiver , we only have to check that there is an isomorphism of stable translation quivers between the ar - quiver of the various orbit categories and the quivers of coloured oriented single and paired diagonals associated to @xmath8 .
this isomorphism then induces the claimed equivalences . to do so
, the strategy will be to compare the action of @xmath20 on @xmath172 with the actions of @xmath20 and of @xmath51 on @xmath71 where @xmath5 is an orientation of a simply laced dynkin diagram , as described in section [ action ] .
below we treat the case @xmath199 , the remaining two claims can be deduced with a similar reasoning . from the discussion of section [ action ] it follows that the ar - quiver of @xmath199 is isomorphic to the quotient graph @xmath200 . on the other side , by definition @xmath201
is the mesh category of @xmath202 associated to a 10-gon and @xmath203 .
[ eq ] we also deduce the following equivalences of additive categories @xmath204 we follow the proof of theorem [ eqofcat ] and observe that for the first claim we consider only paired oriented diagonals of @xmath8 . since @xmath116_p\neq[j , i]_p$ ] and
the map @xmath88 is the identity on paired oriented diagonals , we deduce that @xmath205 always lies on a cylinder . from lemma [ propt ]
we deduce that @xmath206 . from the discussion of section [ action ]
we deduce that @xmath207 which we recognise as the ar - quiver of @xmath208 .
the second and third claim follow in a similar fashion . for the last claim
we observe that for each vertex of @xmath8 the quiver @xmath209 has one red and one blue single coloured oriented diagonal .
these correspond to the exceptional vertices of a dynkin diagram of type @xmath210 .
the equivalences of theorem [ eqofcat ] and corollary [ eq ] allow us to define a shift functor @xmath51 on the categories associated to @xmath8 induced by the shift functor @xmath51 defined on the various orbit categories considered above , see also @xcite . in the following however , we will not use the triangulated structure of these categories .
let @xmath116_c$ ] , @xmath144 , be a coloured oriented single or paired diagonal of a heptagon @xmath8 and let @xmath115 be the underlying unoriented diagonal .
then we deduce the following useful results .
[ corproj ] there is a dense and full functor @xmath211_c&\mapsto(i , j)\end{aligned}\ ] ] from the cluster category @xmath17 of type @xmath1 to the cluster category @xmath212 of type @xmath213 .
consider the projection @xmath214 defined in subsection [ projections ] .
let @xmath64 be the stable translation quiver of unoriented diagonals of @xmath8 , as defined in caldero - chapoton - schiffler s paper @xcite .
then there is a projection @xmath215 , which maps @xmath193_r , [ i+j , i]_b ) \}$ ] to the unoriented diagonal @xmath216 of @xmath8 and pairs of arrows in @xmath217 to the one corresponding arrow in @xmath64 .
we get a surjective map of translation quivers @xmath218 this map then induces a dense and full functor @xmath219 after identifying @xmath17 , resp . @xmath220 , with the additive category generated by the mesh category of @xmath221 , resp .
@xmath64 . in section [ cluster e6 ] we will use the functor @xmath222 to describe all cluster tilting sets of @xmath17 as configurations of coloured oriented diagonals in @xmath8 .
there is a dense and full functor @xmath223_p&\mapsto(i , j)\end{aligned}\ ] ] from the category @xmath224 to the cluster category of type @xmath225
. follows from the proof of corollary [ eq ] .
the aim of this section is to link our category of coloured oriented single and paired diagonals to the cluster category of type @xmath226 , denoted by @xmath227 .
we recall that @xmath227 can be modelled geometrically in two equivalent ways .
one approach arises categorifying the model given by fomin - zelevinsky in ( * ? ? ?
3.16 ) defined in terms of unoriented diameters and pairs of diagonals in a regular @xmath228-gon .
the second approach uses tagged arcs in a once punctured disc , as described by schiffler in @xcite .
we can recover both descriptions from our model .
for the first one , we proceed as follows .
let @xmath229 be as before and define the quiver @xmath230 having as vertices the ( unordered ) triples of centrally symmetric diagonals of @xmath8 : @xmath231_p , [ i+2+n , i+n]_r , [ i+n , i+2+n]_b\},\\ & \{[i , i+3]_p , [ i+3+n , i+n]_r , [ i+n , i+3+n]_b\},\\ & \dots\\ & \{[i , i+n-1]_p , [ i+2n-1,i+n]_r , [ i+n , i+2n-1]_b\}\end{aligned}\ ] ] together with the oriented single diagonals @xmath232_r , [ n+1,i]_b$ ] of @xmath8 , for @xmath233 notice that @xmath232_r $ ] and @xmath234_b$ ] are the central oriented diagonals of @xmath235 the arrows of @xmath230 are induced by the minimal clockwise rotations between diagonals of @xmath236 , similarly for the translation map .
this allows us to obtain a surjective map of stable translation quivers @xmath237 . dropping the orientations of all coloured diagonals we also obtain the surjective map @xmath238 .
it follows that there is a dense and full functor @xmath239 from the mesh category of @xmath240 , @xmath241 , to the cluster category @xmath242 in addition , we observe that @xmath243 , where @xmath244 is the 2-repetitive cluster category of type @xmath226 defined in section [ def_orbitcat ] .
to recover schiffler s model we need to use a smaller punctured polygon and allow non - contractible loops .
defining @xmath245 and @xmath246 as before we obtain an isomorphism of stable translation quivers @xmath247 and the desired equivalence of categories : @xmath248 in figure [ arc_d ] we illustrate the quiver @xmath249 in a regular punctured square .
[ ] [ ] [ 0.5]@xmath147 [ ] [ ] [ 0.5]@xmath57 [ ] [ ] [ 0.5]@xmath148 [ ] [ ] [ 0.5]@xmath149 in a square.,title="fig : " ] when @xmath7 is not an orientation of a simply laced dynkin graph the ar - quiver of the cluster category @xmath9 splits into various connected components .
our next goal is to obtain an isomorphism between the irregular ( transjecctive ) component of the ar - quiver of @xmath9 and the quiver of coloured oriented single and paired diagonals . for this we extend the construction of section [ pairs ] and consider and infinite - sided polygon @xmath15
next , let @xmath250 be the subset of all coloured oriented single and paired diagonals of @xmath15 consisting of @xmath251 paired , @xmath107 red and @xmath108 blue single coloured oriented diagonals for every vertex of @xmath15 : @xmath127_p , \dots,[i , i+r+2]_p , \\ & [ i , i+r+3]_r,\dots,[i , i+r+s+2]_r , \\ & [ i+r+3,i]_b , \dots,[i+r+t+2,i]_b,\hspace{0,3 cm } \textrm { $ i$ vertex of } \pi^\infty\bigg\}.\end{aligned}\ ] ] then we extend the definition of minimal clockwise rotations of section [ rotations ] to this setting .
thus we obtain a quiver @xmath252 , whose vertices are the elements of @xmath250 and where we link two vertices with an arrow when there is a minimal clockwise rotation between them .
moreover , we define @xmath20 on the elements of @xmath250 as the anticlockwise rotation around the center of @xmath15 induced by the rotation though @xmath157 in an @xmath4-gon @xmath8 letting @xmath253 . in this way we turn @xmath254 into a stable translation quiver . to state the next result , let @xmath255 be the full subcategory of @xmath9 , consisting of @xmath20-shifts of indecomposable projective objects in @xmath9 .
moreover , denote the additive category generated by the mesh category of @xmath254 by @xmath14
. then we can show the following result .
[ isoinfty ] the functor @xmath256 is an equivalence of additive categories .
first we observe that @xmath257 .
next , let @xmath258 be the subset of @xmath250 consisting of the @xmath105 single and paired diagonals : @xmath259_p , \dots,[1,r+3]_p , [ 1,r+4]_r,\dots,[1,r+s+3]_r , [ r+4,1]_b , \dots,[r+t+3,1]_b \}$ ] .
then , we observe that since @xmath252 is a connected stable translation quiver @xmath260 induces an injective map between @xmath261 , for @xmath262 and @xmath263 between the indecomposable injective objects and @xmath51 of the projective objects in @xmath9 .
thus @xmath260 is full .
moreover , it is not hard to see that @xmath260 is dense and faithful , thus @xmath260 is indeed an equivalence of additive categories .
our next aim is to describe the combinatorics of the cluster category @xmath264 inside a heptagon . throughout the chapter
let @xmath8 be a regular heptagon and let @xmath265 .
moreover , let @xmath266 be the stable translation quiver having as vertices the unoriented diagonals of @xmath8 and arrows given by minimal clockwise rotations , see @xcite for details .
the support of @xmath267 in @xmath54 is called the _ front @xmath28-hammock of @xmath268_. the support of @xmath269 in @xmath54 is called the _ back @xmath28-hammock of @xmath268_. these hammocks can be deduced from the ar - quiver using mesh relations , or using starting and ending functions , see ( * ? ? ? * chap . 8) . for ar - quivers isomorphic to @xmath71 with @xmath5 an orientation of a dynkin graph ,
the support of @xmath270 has been described in detail in @xcite .
identify again @xmath17 , resp .
@xmath220 with the additive categories generated by the mesh categories of @xmath64 , resp
. @xmath266 . in the sequel
we view the hammocks inside @xmath64 or @xmath266 .
moreover , let @xmath271 , for coloured oriented diagonals @xmath272 @xmath273 in @xmath8 and indecomposable objects @xmath268 and @xmath274 in @xmath17 corresponding to @xmath275 and @xmath273 by the equivalence of theorem [ isoe6 ] . [ 2cy ] as @xmath17 is 2 calabi - yau : @xmath276 for all objects @xmath277 in @xmath17 .
therefore , the back and front @xmath28-hammocks in @xmath17 coincide for all objects . on the other hand , in @xmath278 the back and front hammocks are disjoint , as the category is not @xmath57-calabi - yau .
consider again the projection @xmath279 defined by @xmath280 as in corollary [ corproj ] .
for each @xmath275 we define two connected sub - quivers of @xmath64 , @xmath281 and @xmath282 , as follows .
the vertices of both @xmath281 and @xmath282 , are the vertices of @xmath64 in the @xmath28-hammock of @xmath275 in @xmath64 and in the preimage under @xmath283 of the @xmath28-hammock of @xmath284 in @xmath266 .
the arrows of @xmath281 and @xmath282 coincide with the arrows of @xmath64 .
then @xmath281 contains the vertex @xmath285 and will be called the _ front crossing _ of @xmath275 , @xmath282 contains @xmath286 and will be called the _ back crossing _ of @xmath275 . note that for all @xmath275 , the sub - quivers @xmath281 and @xmath282 are disjoint . in addition , all coloured oriented diagonals in @xmath8 _ crossing _
@xmath275 in an interior point of @xmath275 are vertices of @xmath287 and @xmath288
. see figure [ exts ] , were the vertices of inside the front and back crossings of @xmath275 are in heptagons with bold boundary , for @xmath275 a coloured oriented diagonal in the first slice of @xmath64 .
[ ] [ ] [ 0.75]a ) [ ] [ ] [ 0.75]b ) [ ] [ ] [ 0.75]c ) [ ] [ ] [ 0.75]d ) [ ] [ ] [ 0.75]e ) [ ] [ ] [ 0.75]f ) -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] -hammocks of the diagonals with vertices of @xmath289 and @xmath290 represented in heptagons with bold boundary.,title="fig : " ] in the next result , we assume that @xmath275 is a coloured oriented diagonal of @xmath8 in the first slice of @xmath64 .
this assumption can be dropped using @xmath20-shifts , or renumbering the vertices of @xmath8 .
moreover , we write @xmath291 to indicate the boundary of @xmath8 , and for two coloured oriented diagonals @xmath275 and
@xmath273 we say that @xmath273 _ enters _ the smaller region bounded by @xmath292 and @xmath291 if the arrow head of @xmath273 goes to a vertex of @xmath291 inside the region and different from the vertices joined by @xmath292 .
[ cuts ] let @xmath293 be coloured oriented diagonals of @xmath235 assume @xmath275 is in the first slice of @xmath64 , and that @xmath275 crosses @xmath294 . * if @xmath292 is a paired diagonal , then @xmath295 . *
if @xmath292 is a single diagonal , and @xmath294 enters the smaller region bounded by @xmath292 and @xmath291 , then @xmath295 .
if @xmath292 is paired , @xmath281 coincides with @xmath287 and @xmath282 coincides with @xmath288 , thus the vertices of @xmath281 and @xmath282 are all the oriented coloured diagonals of @xmath8 crossing @xmath292 .
if @xmath292 is a single coloured diagonal of @xmath8 , we need to distinguish between the diagonals inside @xmath296 and @xmath297 , @xmath298 .
then we observe that the coloured oriented diagonals in @xmath281 and @xmath282 are precisely the ones satisfying the assumptions of the proposition .
the aim of this section is to divide the @xmath28-hammocks in @xmath17 into curves .
the reason why we do this is because for each coloured oriented diagonal @xmath275 we want to find a uniform geometric description of the elements inside the @xmath28-hammock of @xmath275 . since the hammocks in @xmath17 are very big , this goal seems hopeless .
however , dividing the @xmath28-hammock of @xmath275 into smaller sets , allows us to describe the elements of each such set in geometric terms
. we will call these sets curves .
let @xmath268 be an indecomposable object of @xmath17 and let @xmath275 be the corresponding coloured oriented diagonal viewed as a vertex of @xmath64 . for @xmath299 , _ the curves _
@xmath300 of @xmath275 in @xmath64 are @xmath106 collections of oriented coloured diagonals having non - vanishing extensions with @xmath275 .
each collection @xmath301 has the shape of a curve in @xmath64 .
we begin defining the curves of @xmath275 , for @xmath275 in the first slice of @xmath64 . for all other vertices @xmath275 of @xmath64
, curves can be defined from the previous ones by @xmath20-shifts .
the first curve of @xmath302_r$ ] , denoted by @xmath303_r)$ ] , is defined as follows : @xmath304_r):=&\{[7,2+i ] _
c,\ , 0\leq i\leq 3,\ , c\in\{r , p\}\}\\ & \cup\{[5,7+i ] _ c,\ , 0\leq i\leq 3,\,c\in\{r , p\ } \}\\ & \cup\{[6,3]_r\}.\end{aligned}\ ] ] the second curve of @xmath302_r$ ] , denoted by @xmath305_r)$ ] , is defined as follows : @xmath306_r):=&\{[5-i,7 ] _
c,\ , 0\leq i\leq 3,\ , c\in\{b , p\}\}\\ & \cup \{[2,7+i ] _ c,\ , 0\leq i\leq 3,\,c\in\{b , p\ } \}\\ & \cup\{[1,4]_b\}.\end{aligned}\ ] ] by definition @xmath303_r)$ ] is obtained by the sequence of minimal clockwise rotations around the vertices 7 , 5 , 3 of @xmath8 starting in @xmath307_r)=[2,7]_r$ ] and ending with @xmath308_r$ ] .
dually , @xmath305_r)$ ] is obtained by a sequence of minimal anticlockwise rotations around the vertices 7 , 2 , 4 starting in @xmath309_r)=[5,7]_b$ ] and ending in @xmath310_b$ ] . by construction the @xmath28-hammock of @xmath302_r$ ] is @xmath303_r)\cup c_2([1,6]_r)$ ] . in figure [ hammocks](a )
the elements of @xmath303_r)$ ] are drawn in the upper half of @xmath64 , while the elements of @xmath305_r)$ ] are in the lower half .
next , we are going to associate four curves to @xmath311_r$ ] . the first curve of @xmath311_r$ ] , @xmath312_r)$ ] ,
is the set containing coloured oriented diagonals of @xmath8 obtained by a sequence of minimal clockwise rotations around the vertices 6 , 4 , 2 starting in @xmath313_r)$ ] and ending in @xmath314_r$ ] .
the third curve @xmath315_r)$ ] is obtained by a sequence of minimal anticlockwise rotations around the vertices 7 , 2 , 4 starting with @xmath316_r)$ ] and ending in @xmath310_b$ ] . moreover , @xmath317_r)$ ] coincides with @xmath303_r)$ ] , and @xmath318_r)$ ] coincides with @xmath305_r)$ ] . for @xmath319_r,[,1,5]_r\}$ ] , the curves of @xmath320 of @xmath321 are defined by @xmath322 , @xmath299 .
we are left with defining the curves of the paired diagonals @xmath120_p$ ] and @xmath310_p$ ] . for @xmath120_p$
] we have @xmath323_p)$ ] given by the set containing both single and paired coloured oriented diagonals obtained by a sequence of minimal rotations in the clockwise order around the vertices 2 , 7 , 5 starting in @xmath324_p)$ ] and ending in @xmath325_p$ ] .
similarly @xmath326_p)$ ] is obtained by a sequence of minimal anticlockwise rotations around the vertices 2 , 4 , 5 starting in @xmath327_p)$ ] and ending in @xmath328_p$ ] .
next , @xmath329_p)$ ] is obtained by a sequence of of minimal clockwise rotations starting in @xmath330_p ) $ ] and ending in @xmath325_p$ ] .
@xmath331_p)$ ] is obtained rotating in the anticlockwise order @xmath332_p)$ ] to @xmath333_p$ ] .
finally , @xmath334_p)=\tau^{-1}(c_1([1,3]_p))$ ] and @xmath335_p)=c_2([1,3]_p)$ ] . in figure [ hammocks](a)-(f )
we represent the various curves of @xmath275 , for@xmath275 be in the first ( and last ) slice of @xmath64 .
the numbers 1 , 2 , 3 , 4 indicate the starting term of the curves @xmath336 and the colours of the heptagons indicate the curves they intersect . [ ] [ ] [ 0.75](a ) [ ] [ ] [ 0.75](b ) [ ] [ ] [ 0.75](c ) [ ] [ ] [ 0.75](d ) [ ] [ ] [ 0.75](e ) [ ] [ ] [ 0.75]*1 * [ ] [ ] [ 0.75]*2 * [ ] [ ] [ 0.75]*3 * [ ] [ ] [ 0.75]*4 * [ ] [ ] [ 0.75](f ) .,title="fig : " ] .,title="fig : " ] .,title="fig : " ] .,title="fig : " ] .,title="fig : " ] .,title="fig : " ] [ int ] let @xmath337 be vertices of @xmath64 .
let @xmath300 be the curves of @xmath275 , @xmath338 .
then @xmath339 is equal to the number of curves of @xmath275 intersecting with @xmath273 in @xmath64 ( 0 up to 3 ) .
first , the @xmath28-hammocks in the ar - quiver of @xmath17 are invariant under @xmath20-shifts . after changing the image of the projective objects of @xmath340 in the equivalence of theorem [ isoe6 ]
we can assume that @xmath268 or @xmath274 corresponds to a diagonal in the first slice of @xmath64 .
by remark [ 2cy ] we can treat the cases where @xmath275 , or @xmath273 belongs to the first slice of @xmath64 in the same way .
second , the curves of @xmath275 are by construction such that their intersection points coincide with the vertices @xmath273 in @xmath64 for which @xmath341 . from the definition of morphisms in the mesh category of a stable translation quiver , it follows that the number of curves intersecting in @xmath273 is the dimension of @xmath342 let again @xmath275 be in the first slice of @xmath64 . in figure [ hammocks](a)-(f ) the dimension of the space @xmath339 is expressed by the numbers of colours filling the heptagon containing @xmath273 . if @xmath273 is in a white heptagon of @xmath64 then @xmath343 . more precisely , the two curves represented in figure [ hammocks](a ) and ( f ) never intersect , and @xmath295 for @xmath273 in a coloured heptagon , and @xmath275 in the first slice of @xmath64 . in figure [ hammocks](b ) and
( e ) the curve @xmath344 intersects @xmath345 in two vertices .
similarly , for @xmath346 and @xmath347 .
the curve @xmath345 intersects @xmath347 only once .
the five heptagons where two curves meet have two colours ( boundary and interior of the heptagon ) .
then @xmath348 for @xmath273 corresponding to one of these heptagons . in figure [ hammocks](c )
there are two heptagons where three curves meet .
they are drawn with three colours , and hence @xmath349 for @xmath273 corresponding to one of these two . moreover , in nine heptagons two curves meet , and they are drawn in two colours . finally , in figure [ hammocks](d ) two curves are drawn , and they intersect only in one vertex of @xmath64 .
let @xmath5 be an orientation of a simply - laced dynkin graph with @xmath90 vertices .
let @xmath350 be a set of pairwise non isomorphic indecomposable objects of @xmath54 .
if @xmath351 for all @xmath352 , then one says that @xmath184 is a _
cluster tilting set _ of @xmath353 .
a _ cluster tilting object _ in @xmath353 is the direct sum of all objects of a cluster tilting set in @xmath54 . observe that knowing a cluster tilting objects allows to determines a cluster tilting set and viceversa .
moreover , given a cluster tilting set @xmath184 , one says that @xmath354 , @xmath355 is an _ almost complete cluster tilting object _ if there is an indecomposable object @xmath356 in @xmath54 such that @xmath357 is a cluster tilting object of @xmath54 .
the object @xmath356 is called the _
complement _ of @xmath358 .
the _ mutation at @xmath74 _ of a cluster tilting object @xmath184 in @xmath54 , for @xmath359 , is the operation which replaces the indecomposable summand @xmath358 in @xmath360 with the complement @xmath356 of @xmath358 in @xmath361 the statements in the next theorem are shown in @xcite .
[ thmbmrrt ] let @xmath362 be a cluster tilting object in @xmath353 .
* each almost complete cluster tilting object @xmath363 in @xmath353 has exactly two complements , @xmath362 and @xmath364 . *
if @xmath362 and @xmath364 are complements of @xmath363 then @xmath365 on the other side , if @xmath366 , then there is an almost complete cluster tilting object @xmath363 such that @xmath362 and @xmath364 are complements of @xmath363 . after proposition 3.8 in @xcite and (
4.5 ) we know that there are 833 cluster tilting sets , hence cluster tilting objects , in @xmath17 . from the work of caldero - chapoton - schiffler , see @xcite , we know that cluster tilting objects of @xmath367 are in bijection with the formal direct sums of diagonals belonging to a maximal collection of non - crossing diagonals in a regular @xmath4-gon . in this context mutations corresponds to _ flips _ of diagonals .
more precisely , a flip replaces a diagonal @xmath368 in a given triangulation @xmath369 with the unique other diagonal @xmath370 crossing @xmath368 and completing @xmath371 to a new triangulation of the regular @xmath4-gon .
our next aim is to describe cluster tilting sets of @xmath17 as configurations of single and paired coloured oriented diagonals in @xmath235 a _ cluster configuration _ is a family of pairwise different coloured oriented diagonals of @xmath8 , @xmath372 , with the property that @xmath373 for all @xmath374 .
a coloured oriented diagonal @xmath370 is called _ complement _ of @xmath368 in @xmath375 if @xmath376 and @xmath377 obtained from @xmath375 after replacing @xmath378 by @xmath379 is a cluster configuration of @xmath8 .
consider two heptagons , and a long paired diagonal @xmath380_p$ ] , @xmath381 of @xmath8 .
our next goal is to complete @xmath382 to a set of coloured oriented diagonals inside the two heptagons , giving rise to a cluster configuration of @xmath8 .
for this we remark that @xmath382 divides each @xmath8 into the quadrilateral @xmath383 with boundary vertices @xmath384 , and the pentagon @xmath385 with boundary vertices @xmath386 , @xmath381 . in lemma [ cor ] below we will see that triangulating @xmath383 with a short paired diagonal , and each @xmath385 with single diagonals of the appropriate colour gives rise to cluster configurations .
[ cor ] let @xmath380_p$ ] , @xmath381 .
* for @xmath387 , triangulating each @xmath385 with single diagonals of the same colour , and @xmath383 with a short paired diagonal gives a cluster configuration @xmath388 of @xmath8 .
* all cluster configurations of @xmath8 containing @xmath389_p$ ] arise as @xmath390 , @xmath391 . since @xmath387 we can assume that the region in @xmath64 outside the @xmath28-hammock of @xmath382 has only blue diagonals below @xmath382 and only red diagonals above @xmath382 .
observes that the diagonals outside the @xmath28-hammock are precisely the diagonals involved in triangulations of the two copies of @xmath385 and @xmath383
. then chose a short paired diagonal @xmath392 triangulating @xmath383 with a short paired diagonal .
then triangulating a copy of @xmath385 with only single red diagonals , and triangulating the second copy of @xmath385 with only single blue diagonals yields a cluster configuration . with proposition [ cuts ]
we deduce that the arcs obtained in this way have no extension in each region above and below @xmath393 in @xmath64 .
one can then check that the @xmath28-hammocks in one region do not pass through the other region , nor though @xmath392 .
thus , to each red triangulation one can choose a blue triangulations of @xmath385 , and all choices are possible .
similarly , one can complete @xmath394 to a cluster configuration , where @xmath395 is the other short paired diagonal triangulating @xmath383 .
notice that there are no other possibilities to complete @xmath382 to a cluster configuration of @xmath8 .
next , there are 7 choices for @xmath382 in @xmath64 . for each choice of @xmath396
the associated cluster configurations are obtained from the previous by rotation though @xmath20 .
adjustment of the colours - orientations of the single diagonals triangulating @xmath385 are needed if @xmath382 is the first slice of @xmath64 .
notice that the two triangulations of @xmath385 can be different , and the color is uniquely determined by the position of @xmath382 in @xmath8 , resp . in @xmath64 . in the following we call the cluster configurations given by
a long paired diagonal and coloured oriented single and paired diagonals triangulating two copies of @xmath385 and @xmath383 , as describe in the first part of lemma [ cor ] , the _ first fundamental family _ of cluster configurations .
we denote this family by @xmath18 .
we saw in lemma [ cor ] that many cluster configurations correspond to two triangulations of @xmath8 .
our next goal is to define a second family of cluster configurations describing the remaining cluster tilting set of @xmath17 . for this
the following general observation is needed . for @xmath381 ,
consider the long single red diagonal @xmath397_r$ ] of @xmath8 .
then @xmath393 divides @xmath8 into the quadrilateral @xmath398 , and the pentagon @xmath399 .
let @xmath400 be a cluster configuration of @xmath8 containing @xmath393 .
then @xmath400 necessarily also contains one of the two short single diagonals triangulating @xmath383 , neighbouring @xmath393 in @xmath401 similarly for @xmath402_b$ ] .
more precisely , [ l1 ] let @xmath381 , @xmath397_r$ ] in @xmath8 , and @xmath400 be a cluster configuration containing @xmath393 . *
if @xmath387 , exactly one of @xmath403_r,[i+6,i+4]_r\}$ ] is in @xmath400 . * if @xmath404 , exactly one of @xmath403_r,\rho([i+6,i+4]_r)\}$ ] is in @xmath400 .
similarly for @xmath405 .
moreover , in each case the two diagonals are complements to each other .
let @xmath387 and consider the @xmath28-hammock of @xmath393 in @xmath64 . since @xmath393 is not in the first slice of @xmath64 the diagonals triangulating @xmath383 have the same color as @xmath393 .
then one can check that all @xmath28-hammocks of objects outside the @xmath28-hammock of @xmath393 , which are different from @xmath406_r$ ] and @xmath407_r$ ] , never contain single diagonals inside the quadrilateral @xmath383 in @xmath235 thus , by maximality we deduce that all cluster tilting sets containing @xmath393 necessarily also contain one of the diagonals inside @xmath383 .
taking one diagonal triangulating @xmath383 rules out the other , thus the two single diagonals triangulating @xmath383 are complements to each other . for @xmath404 ,
@xmath393 is in the first slice of @xmath64
. then one can proceed as before adjusting the colour of the diagonal triangulating @xmath383 . in view of the next result , we point out that the short single diagonals of lemma [ l1 ] , triangulating @xmath383 and neighbouring @xmath393 in @xmath64 , are displayed in filled light grey heptagons in figure [ tilt](a)-(n ) .
[ tilt_fig ] every six - tuple of diagonals of figure [ tilt](a)-(n ) determines a cluster configuration of @xmath8 . for each choice of a short single diagonal of lemma [ l1 ] triangulating @xmath383 and neighbouring @xmath393 in @xmath64 ,
the claim can be verified by checking that the diagonals in the highlighted heptagons have no extension among each other . in the following
, we refer to the collection of cluster configurations of figure [ tilt](a)-(n ) as the _ second fundamental family _ of cluster configurations of @xmath8 , and we denote this family by @xmath19 . [ ] [ ] [ 0.75](a ) [ ] [ ] [ 0.75](b ) [ ] [ ] [ 0.75](c ) [ ] [ ] [ 0.75](d ) [ ] [ ] [ 0.75](e ) [ ] [ ] [ 0.75](f ) [ ] [ ] [ 0.75](g ) [ ] [ ] [ 0.75](h ) [ ] [ ] [ 0.75](i ) [ ] [ ] [ 0.75](l ) [ ] [ ] [ 0.75](m ) [ ] [ ] [ 0.75](n ) 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] 0.42 .,title="fig : " ] we determine two symmetries in @xmath8 leading to cluster configurations .
one symmetry simply switches colours and orientations of the coloured oriented diagonals of a given cluster configuration .
the second one arises from a left - right symmetry of @xmath64 , and corresponds to a reflection in @xmath8 .
there are two reasons why these symmetries are important .
first , using these symmetries we can deduce all cluster configurations starting from the sets in @xmath19 .
second , knowing how a cluster configuration behaves under mutation , allows to understand how the symmetric ones behave .
let @xmath408 .
for @xmath381 let @xmath409 be the line in @xmath8 passing through @xmath74 and the middle point of @xmath410 . on all coloured oriented diagonals
different then @xmath411_c$ ] , let @xmath412 be the reflection in @xmath8 along @xmath409 followed by a switch of orientation .
otherwise , @xmath413_c):=\rho([i\pm1,i\mp1]_c).$ ] in figure [ g ] we illustrate on the left the cluster configuration ( g ) and on the right we show the cluster configuration obtained after applying @xmath414 to it . [ ] [ ] [ 0.75](g ) [ ] [ ] [ 0.75](g ) 0.42 -symmetric cluster configurations of @xmath8.,title="fig : " ] 0.42 -symmetric cluster configurations of @xmath8.,title="fig : " ] [ l23 ] let @xmath184 be a cluster configuration belonging to @xmath19 .
then @xmath415 , and @xmath416 are also cluster configurations .
apply the map @xmath88 , resp .
@xmath414 to the cluster configuration of lemma [ tilt_fig ] . because of the shape of the @xmath28-hammocks one indeed produces cluster configurations .
notice that the set @xmath415 of a cluster configuration @xmath184 is an elements of the @xmath20-orbit of @xmath184 , while @xmath417 does not belong to any @xmath20-orbit of a cluster configuration of @xmath19 ( nor of @xmath18 ) . from (
3.8 ) and ( * ? ? ?
4.5 ) we know that there are 833 cluster tilting sets in @xmath17 . in the next result
we give a complete geometric classification of all cluster tilting sets in @xmath17 in terms of cluster configurations of @xmath8 .
[ count ] in @xmath8 * 350 different cluster configurations have one long paired diagonal , and they arise from @xmath18 through @xmath20 . *
483 other cluster configurations arise from @xmath19 through @xmath21 and @xmath20 .
[ cor_count ] in @xmath8 : * 224 cluster configurations have precisely one short paired diagonal , * 175 cluster configurations have precisely two short paired diagonals , * 84 cluster configurations have no paired diagonals of @xmath8 .
all these cluster configurations are different . the first part of the claim follows from lemma [ cor ] .
in fact , we saw that for each long paired diagonal @xmath396 there are @xmath418 ways to triangulate one of the two pentagons @xmath385 with single coloured diagonals .
moreover , there are two ways to triangulate @xmath383 with short paired diagonals .
thus , each @xmath396 gives rise to @xmath419 different cluster configurations .
since there are @xmath152 choices for @xmath396 in @xmath8 , the first claim follows . for the second part of the claim the idea is to consider different cases , depending on the number of short paired diagonals leading to cluster configurations .
first case : the only paired diagonal of @xmath184 is a short one .
then @xmath184 arises from the collection highlighted in ( a),(b ) or ( c ) in figure [ tilt ] , up to @xmath20-shifts and the @xmath21-symmetry of lemma [ l23 ] .
moreover , after lemma [ l1 ] for each coloured oriented diagonal @xmath420_c$ ] , @xmath408 in @xmath184 there two possible choices of neighbouring short single diagonals in @xmath184 .
consequently , up to @xmath20-shifts , the there are 4 different cluster configurations arising from a collection of diagonals as in ( a ) .
similarly for ( b ) .
the collection in ( c ) gives rise to 8 different cluster configurations up to @xmath20-shifts , as there are 4 choices for short single diagonals , and further 4 arise by taking the @xmath21-symmetric case .
summing up , the cluster configurations in ( a ) , ( b ) , ( c ) give rise to 224 different cluster configurations .
second case : @xmath184 has exactly two short paired diagonals .
then one distinguishes further into ( d),(e ) and ( f ) which have single coloured oriented diagonals of the form @xmath421_c$ ] , @xmath408 . while the cluster configurations in ( g),(i),(h ) and ( l ) only contain short single diagonals of the form @xmath422_c$ ] ,
@xmath408 . proceeding as before , taking into account the symmetry @xmath21 of lemma [ l23 ] one obtains the claimed number . in the third case
we count the cluster configurations arising from @xmath20-shifts of the cluster configurations in ( m ) and ( n ) having no paired diagonals . as before we deduce that there are 28 the cluster configurations arising from @xmath20-shifts of ( m ) and 56 arising from @xmath20-shifts of the 4 cluster configurations in ( n ) .
together this gives 84 cluster configurations without paired diagonals .
in section [ longpaired ] we saw that cluster configurations in @xmath18 correspond to triangulations with coloured oriented diagonals of two copies of @xmath8 .
cluster configurations of @xmath19 are not as simple to describe . in theorem [ confthm ] below we can give a general statement concerning the geometry of cluster configurations in @xmath8 .
[ confthm ] all 833 cluster tilting sets of @xmath17 can be expressed as configurations of six non - crossing coloured oriented single and paired diagonals inside two heptagons .
first , given a cluster configurations in @xmath423 we divide the coloured oriented diagonals inside two heptagons by colour . paired diagonals appear in both heptagons .
then we observe that all cluster configurations in @xmath18 and @xmath19 are crossing free .
moreover , the symmetry @xmath21 produces new configurations of diagonals which are again crossing free .
taking @xmath20-shifts only rotates the entire configurations inside the two heptagons , occasionally switching colours and orientation according to the action of @xmath20 inside @xmath8 .
hence the crossing free property is preserved under @xmath20-shifts and the claim follows .
the cluster configuration @xmath424_r , [ 5,2]_r , [ 5,1]_p,[5,6]_p , [ 3,5]_b,[2,5]_b\}$ ] is expressed in two heptagons in the center of figure [ exchange graph ] , the numbering of the vertices of one heptagon is highlighted in the figure .
paired diagonals appear in both heptagons and have labels .
the converse statement of theorem [ confthm ] is not true , as configurations of non - crossing coloured diagonals different then cluster configurations of @xmath8 are not cluster tilting sets of @xmath17 . in this section
we will see that in many cases it is possible to deduce the mutation process in @xmath17 from the mutations process inside @xmath212 .
moreover , we can deduce the mutation process for the families in @xmath18 and @xmath19 and taking @xmath20-shifts extend it for the remaining cluster configurations . in the next definitions we indicate by @xmath425 the unoriented single diagonal corresponding to a coloured oriented diagonal @xmath3 of @xmath8 .
let @xmath426 be a paired oriented diagonal and let @xmath184 be a cluster configuration containing @xmath426 .
the paired diagonal @xmath427 is the flip - complement of @xmath426 , if @xmath428 and @xmath429 are related by a flip .
let @xmath430 be a single coloured oriented diagonal and let @xmath184 be a cluster configuration containing @xmath430 .
the single coloured oriented diagonal @xmath431 is the _ flip - complement _ of @xmath430 in @xmath184 , if @xmath432 and @xmath433 are related by a flip and @xmath434 is a cluster configuration .
notice that if a flip - complement exists then the colour and orientation is uniquely determined by lemma [ cor ] and lemma [ l1 ] . [
378 ] let @xmath435 be a long coloured oriented diagonal of @xmath8 giving rise to a cluster configuration .
* if @xmath435 is paired : single coloured diagonals triangulating @xmath385 , and paired diagonals triangulating @xmath383 have a flip - complement . *
if @xmath435 is single : single diagonals triangulating @xmath383 have a flip - complement .
let @xmath435 be a long coloured oriented diagonal dividing @xmath8 into the quadrilateral @xmath383 and the pentagon @xmath385 .
in lemma [ cor ] we saw that if @xmath435 is paired , triangulating @xmath383 with a paired diagonal , and two copies of @xmath385 with single oriented diagonals , always gives a cluster configuration . hence removing a single diagonal of a copy of @xmath385 or a paired diagonal triangulating @xmath383
can only be completed to a cluster configuration in two ways , namely with diagonals being flip - complement of each other .
if @xmath435 is single , the claim follows from lemma [ l1 ] .
further instances of the mutation process in @xmath17 can be described by flips of coloured oriented diagonals in @xmath8 , but not all mutations allow a description of this type .
this is unsurprising , as for example not all mutations in the cluster algebra @xmath436 $ ] can be described through plcker relations , see @xcite . with figure [ tilt ]
it is not hard to deduce all the remaining mutations occurring .
for example , the cluster configuration in ( a ) can be mutated to ( b),(c),(e),(l),(m ) and to the flip of a diagonal inside one light grey coloured heptagon .
similarly , we have a list for the other families .
some instances of the more complicated geometric exchanges can be found on the upper pentagon of figure [ exchange graph ] . in figure
[ exchange graph ] we display a part of the _ exchange graph _ of @xmath17 . for each heptagon appearing in the figure
the numbering of its vertices is as shown on the central heptagon .
the vertices of the graph correspond to cluster configurations , hence to cluster tilting sets of @xmath17 , edges are drawn when two cluster configurations are related by a single mutation . in the two central heptagons of figure [ exchange graph ] the configuration of @xmath424_r , [ 5,2]_r , [ 5,1]_p,[5,6]_p , [ 3,5]_b,[2,5]_b\}$ ] is displayed .
the @xmath437 neighbouring configurations are placed on the vertices of the two central pentagons sharing the vertex corresponding to @xmath184 .
these @xmath437 sets are obtained from @xmath184 through repeated flips of single diagonals , as described in proposition [ 378 ] .
the vertices of the left pentagon are obtained after mutating @xmath328_p$ ] in @xmath184 .
let @xmath438 be a cluster tilting object of @xmath353 , then @xmath439 is the _ cluster tilted algebra _ of type @xmath440 the quiver @xmath441 of @xmath439 has no loops nor 2-cycles and it encodes precisely the exchange matrix of the cluster associated to @xmath362 , see @xcite and @xcite . in @xmath367
the quiver @xmath441 can be read off from the triangulation @xmath362 , see @xcite .
the vertices of @xmath441 are the diagonals of the triangulation and an arrow between @xmath368 and @xmath442 is drawn , whenever @xmath368 and @xmath442 bound a common triangle .
the orientation of the arrow is @xmath443 , if @xmath442 is linked to @xmath368 by an anticlockwise rotation around the common vertex . extending the definition of @xmath441 to the case at hand ,
the quivers corresponding to the cluster tilting sets of figure [ exchange graph ] can be deduced .
one could also read off the quivers , and relations , directly from @xmath64 by determining the spaces @xmath444 . .
at each vertex of the graph , the diagonals with the same labels are identified . ]
cluster categories arising from valued quivers have been first categorified algebraically in @xcite .
the aim of this section is to use the geometric description of the cluster category @xmath17 to categorify geometrically the cluster algebra of type @xmath22 . since we know all cluster configurations of @xmath8 ,
see theorem [ count ] , we can deduce that there are only three types of @xmath88-symmetric cluster configurations in @xmath8 .
first , cluster configurations of @xmath8 projecting to triangulations of @xmath8 consisting of unoriented arcs through @xmath447 .
these arise from @xmath18 and from @xmath20-shifts of the configurations in figure [ tilt](d ) .
second , @xmath20-shifts of both , the cluster configuration in figure [ tilt](g ) and the @xmath21-symmetric configuration of figure [ tilt](g ) .
third , @xmath20-shifts of the configuration in figure [ tilt](h ) .
we call the first @xmath88-symmetric cluster configurations of type * t * , the second of type * c * and the third of type * l*. moreover , we refer to the double short diagonals in a configuration of type * c * as middle diagonals . in figure [ confcl ] configurations of type *
l * and * c * are illustrated .
the blue diagonals have opposite orientation with respect to the ones shown in the figure and are omitted .
the two @xmath88-symmetric cluster configurations of type * c * on the right side of the figure are related through the action of @xmath414 , already defined in section [ sigma ] .
let @xmath449 be a @xmath88-symmetric cluster configuration of @xmath8 and denote by @xmath450 the unique complement in @xmath17 of @xmath3 in @xmath449 .
then we observe that @xmath88-symmetric cluster configurations always have two @xmath88-orbits consisting of paired diagonals and two consist of single diagonals of opposite colour and orientation , thus we can assume @xmath451 . keeping the notation as above , we have the following result .
the claim follows from the mutation rule of @xmath17 and the symmetry of the configurations .
moreover , since there are always two paired diagonals and two @xmath88-orbits of single diagonals in @xmath449 we deduce that paired diagonals are exchanged with paired diagonals , and single diagonals with single diagonals .
we call the @xmath88-symmetric cluster configuration obtained from @xmath449 of proposition [ mutrhosym ] the _ mutation of @xmath449 at @xmath454 , resp . at @xmath455
_ , @xmath452 .
we refer to it by @xmath456 . proceeding as in the proof of theorem [ count ]
we deduce that there are 105 @xmath88-symmetric cluster configurations in @xmath8 : 84 are of type * t * , 14 of type * c * and 7 of type * l*. from ( * ? ? ?
3.8 ) we know that this number coincides with the number of clusters in the cluster algebra of type @xmath22 , @xmath457 .
in addition , the mutation rule in @xmath17 induces a unique mutation rule for @xmath88-symmetric cluster configurations of @xmath8 .
this can simply be seen performing the mutation case by case .
the mutation in @xmath17 on @xmath88-symmetric cluster configurations then agrees with the mutation of the cluster algebra @xmath457 , since the exchange graph associated to @xmath457 is regular of degree 4 , see ( * ? ? ?
* thm 1.15 ) .
thus there is essentially just one possible mutation .
the 84 @xmath88-symmetric cluster configurations of type * t * in @xmath17 are in 2:1 correspondence with the triangulations of @xmath8 , thus with the cluster tilting sets of the cluster category of type @xmath212 . geometrically the mutations of @xmath88-symmetric cluster configurations in @xmath8 are described by the following three moves .
let @xmath408 and let @xmath74 a vertex of @xmath8 .
we call a triangle bounded by coloured oriented diagonals in @xmath8 _ internal _ if all its edges are different then boundary edges .
l - c : : in a configuration of type * l * the diagonals @xmath459_c,\rho([i+3,i+1]_c)\}$ ] exchange with @xmath460_c,\rho([i-1,i-3]_c)\}$ ] in a configuration of type * c*. similarly @xmath461_p\leftrightarrow[i-3,i-1]_p$ ] . + c - t : : in a configuration of type * c * the middle diagonals @xmath462_c,\rho([i+2,i]_c)\}$ ] exchange with @xmath463_c,\rho([i , i-3]_c)\}$ ] bounding an internal triangle in a configuration of type * t*. similarly @xmath464_p\leftrightarrow[i , i+3]_p$ ] .
+ t - t : : all diagonals in a configuration of type * t * not bounding an internal triangle , are exchanged with the usual flip rule .
the orientation of the new diagonal is uniquely determined by the type of diagonal one exchanges .
more precisely , the orientation is such that paired diagonals are exchanges with paired diagonals , and single with single . in figure [ l - c ]
we illustrate a mutation between a @xmath88-symmetric cluster configuration of type * l * and of type * c * , as well as a mutation between a configuration of type * c * and type * t*. the diagonals in dotted lines are complements of each other and as before , paired diagonals are labelled .
the construction of @xmath465 and @xmath466 of section [ stabtrans ] was motivated by the following idea : glue two copies of the ar - quiver of @xmath467 along two disjoint @xmath20-orbits .
this resulted in pairing diagonals ( as in section [ pairs ] ) . in this context
one can ask . the cluster categories @xmath199 and @xmath199 have finitely many cluster tilting set and they can be described as cluster configurations of coloured oriented single and paired diagonals inside a 10-gon , resp .
a 16-gon , in a similar way as we did in section [ tilte6 ] . from these configurations
we can identify again a family ( denoted by @xmath18 previously ) of cluster configurations giving rise to triangulations of regions homotopic to a heptagon and a octagon , resp .
a heptagon and a nonagon , compare with section [ longpaired ] and lemma [ cor ] . in figure [ fig7 ]
a cluster tilting set of @xmath199 arising from a projection of the projective modules in @xmath468 is represented .
similarly , in figure [ fig8 ] a cluster tilting set of @xmath469 is represented .
when @xmath7 is not of dynkin type , @xmath9 has infinitely many cluster tilting sets . with our geometric approach , a number of cluster tilting sets of @xmath9 can be expressed as configurations of coloured oriented single and paired diagonals inside the @xmath4-gon @xmath8 , where @xmath11 . moreover ,
if one considers the projections @xmath476 and @xmath477 one can describe the @xmath28-hammocks in @xmath478 using the @xmath28-hammocks in @xmath479 , resp .
@xmath480 , as we did in subsection [ lift ] .
this enable us to express a number of cluster configurations of @xmath9 as triangulations of a @xmath481- and a @xmath132-gon .
finally , since in type @xmath2 all quivers obtained through fomin - zelevinsky quiver mutations are known , we expect that our model can be used to understand the quivers in the mutation class of an orientation of @xmath7 , as well as parts of the exchange graph of @xmath9 . for an initial choice of a triangulation of a regular @xmath484-gon ,
an explicit bijection between all ( unoriented ) diagonals of the polygon and the set @xmath483 of type @xmath485 was given in @xcite . from theorem [ eqofcat ] ,
together with ( * ? ? ?
4.1 ) it follows that there is a bijection between the vertices of @xmath486 and the set of almost positive roots @xmath483 of the root system of type @xmath1 , @xmath29 and @xmath30 .
describe geometrically the bijection between the coloured oriented single and paired diagonals in a 7-,10- , resp.16-gon and the almost positive roots of the root system of type @xmath1 , @xmath29 , resp .
@xmath30 .
* acknowledgments : * i would like to thank karin baur , giovanni felder and robert marsh for all the inspiring discussions we had and for the many very helpful comments .
i would also like to thank an anonymous referee for helpful suggestions .
height 2pt depth -1.6pt width 23pt , _ triangulated categories in the representation theory of finite - dimensional algebras _ , vol .
119 of london mathematical society lecture note series , cambridge university press , cambridge , 1988 .
height 2pt depth -1.6pt width 23pt , _ cluster algebras , quiver representations and triangulated categories _ , in triangulated categories , vol .
375 of london math .
lecture note ser .
, cambridge univ . press ,
cambridge , 2010 , pp . | in this paper we give a geometric - combinatorial description of the cluster categories of type @xmath0 .
in particular , we give an explicit geometric description of all cluster tilting objects in the cluster category of type @xmath1 .
the model we propose here arises from combining two polygons , and it generalises the description of the cluster category of type @xmath2 and @xmath3 . |
argo - ybj is a full coverage air shower detector located at the yangbajing cosmic ray laboratory ( tibet , p.r .
china , lat 30.1n , long 90.5e , 4300 m a.s.l . ) . due to its large field of view ( @xmath22 sr ) and its high duty cycle ( @xmath090@xmath3 ) it can monitor continously a large fraction of the sky , searching for gamma rays sources at energy e @xmath00.6 tev . the apparatus is composed of a central carpet of resistive plate chambers ( rpcs ) ( @xmath274@xmath4 78 m@xmath5 , @xmath293@xmath3 of active area ) enclosed by a guard ring with partial coverage , which allows to extend the instrumented area up to @xmath2100@xmath4110 m@xmath5 . the basic data acquisition element is a cluster ( [email protected] m@xmath5 ) , made of 12 rpcs ( [email protected] m@xmath5 ) .
each chamber is read by 80 strips of [email protected] cm@xmath5 ( the spatial pixel ) , logically organized in 10 independent pads of [email protected] cm@xmath5 which are individually acquired and represent the time pixel of the detector .
the full detector is composed of 153 clusters for a total active surface of @xmath26600 m@xmath5 @xcite
. operated in _ shower mode _ argo - ybj records all events with a number of fired pads n@xmath7 20 in the central carpet , detected in a time window of 420 ns .
the spatial coordinates and the time of any fired pad are then used to reconstruct the position of the shower core and the arrival direction of the primary@xcite .
the installation of the detector has been completed in autumn 2007 and since november 2007 argo - ybj is in stable data taking with a trigger rate of @xmath23.6 khz .
the angular resolution and the pointing accuracy of the detector have been evaluated by using the moon shadow , i.e. the deficit of cosmic rays in the moon direction.@xcite . according to the moon shadow data ,
the psf of the detector is gaussian for n@xmath8100 , while for lower multiplicities it can be described with an additional gaussian , which contributes for about 20% .
the obtained values of the radius @xmath9 containing @xmath271.5@xmath3 of the signal , are 2.59@xmath10 0.16@xmath1 , 1.30@xmath10 0.14 and 1.04@xmath10 0.12@xmath1 respectively for n@xmath1140 , 100 and 300 , in agreement with expectations from monte carlo simulations .
this measured angular resolution refers to cosmic rays - induced air showers .
the angular resolution for @xmath12-induced events has been evaluated by simulations and it is 10 - 30% smaller ( depending on n@xmath13 ) , due to the better defined time profile of the showers .
the data used in this analysis have been recorded from 2007 day 311 to 2009 day 89 for a total live time of 424 days . all events with a zenith angle @xmath1440@xmath1 and a number of hit pads on the central carpet n@xmath1540
have been considered .
the events with a large value of the @xmath16 resulting from the fitting procedure for the arrival direction determination , have been discarded from the analysis . with these events , five sky maps in celestial coordinates
are built , for different number of hit pads : n@xmath1540 , 60 , 100 , 200 , 300 .
the maps are produced with the healpix package ( hierarchical equal area isolatitude pixelization)@xcite .
each map has @xmath23.14 10@xmath17 pixels of equal area @xmath20.013 squared degrees .
the observed sky covers the declination band from -10@xmath1 to 70@xmath1 .
the background is evaluated with the _ time swapping _ method @xcite . for each detected event ,
10 `` fake '' events are generated by replacing the original arrival time with new ones , randomly selected from a buffer that spans about 4 hours of data taking .
for every n@xmath13 interval , the `` event '' map and the `` background map '' are then integrated over a circular area of radius @xmath9 , i.e. every bin is filled with the content of all bins whose center has an angular distance less than @xmath9 from its center .
the value of the smoothing radius , related to the angular resolution , is the angle that maximizes the signal to noise ratio from a crab - like source , according to simulations , i.e. @xmath9 = 1.3@xmath1 , 0.7@xmath1 and 0.5@xmath1 respectively for n@xmath1140 , 100 and 300 .
finally the background map is subtracted to the relative event map , obtaining the `` signal map '' , where for every bin the statistical significance of the excess is calculated .
fig 1 . shows the sky map for events with n@xmath1140 , obtained with a smoothing radius @xmath9=5@xmath1 , adding the data of the whole period .
the map shows two large hot spots in the region of the galactic anticenter , already reported by the milagro detector@xcite .
these regions have been interpreted as excesses of cosmic rays , but the origin is not yet understood . they could be related to the observed large scale anisotropy of cosmic rays due to the propagation of cosmic rays in the galactic magnetic field @xcite .
some authors discuss the possibility of the hot spots being due to a galactic nearby source as a supernova explosion in the recent past@xcite .
the two excesses ( @xmath0 10 standard deviations , corresponding to a flux increase of @xmath18 ) are observed by argo - ybj around the positions @xmath19120@xmath1 , @xmath2040@xmath1 and @xmath1960@xmath1 , @xmath20 - 5@xmath1 . in agreement with the milagro detection , even if the maximum of the second excess is slightly shifted towards lower declinations , probably because argo - ybj can observe with more efficiency the southern regions of the sky , being located at a smaller latitude than milagro .
the two hot spots are also present in the map corresponding to events with a larger number of pad , even if with less significance due to the reduced statistics .
the median primary energy @xmath21 corresponding to proton showers with n@xmath1140 ( 300 ) is @xmath2 2 ( 10 ) tev .
the deficit regions parallel to the excesses are due to a known effect of the analysis , that uses also the excess events to evaluate the background , artificially increasing the background .
the red dashed line represents the gaussian fit.,width=240 ] the analysis of these hot spots will be the subject of a forthcoming paper . in this work , we refer to them only since they affect the point gamma ray source search .
in fact the time swapping method applied over a time interval of 4 hours makes the analysis insensitive to features of scale larger than @xmath260@xmath1 , but all excesses of a smaller extension constitute a kind of `` noise '' for the point source search .
if a source is located inside an excess region , its significance will be overestimated , and vice versa . in order to renormalized the background eliminating all features of size 10 - 20@xmath1 , we multiplied the content of each bin of the background map by a correction factor f = e / b , where e and b are respectively the numbers of events of the event map and background map inside a circle of radius 8@xmath1 centered on that bin , excluding the events from the circle central region up to a radius of 3.5@xmath1 to avoid that a possible source affects the evaluation of e. fig .
2 shows the sky map after the correction .
the crab nebula and mrk421 are visible with a statistical significance respectively of 7.0 and 8.0 standard deviations .
the obtained crab energy spectrum is in agreement with the observations by other experiments@xcite .
mrk421 had many active states during 2008 , in particular during march and june , and has been deeply studied by argo - ybj in a dedicate paper@xcite .
excluding the bins around the crab and mrk421 positions , the distribution of the excesses well fits to a gauss distribution with mean value = [email protected] 10@xmath23 and r.m.s.=1.01@xmath22 5 10@xmath24 ( fig.3 ) . as a second step , we analised the maps contents at the positions of several known gamma rays sources .
we adopted two gamma ray source catalogues : \a ) the recently published list of bright sources observed by fermi at e@xmath0100 mev@xcite \b ) the list of sources seen at e@xmath0100 gev@xcite containing respectively 115 and 44 sources in the declination band corresponding to the argo - ybj field of view , of which 24 are contained in both lists .
merging the two catalogues , we got a list of 135 sources , 86 extragalactic and 49 galactic ( or unidentified but lying close to the galactic plane ) . for every source the radius @xmath25 of the observational window is set to @xmath9 , unless the source is known to be extended . in this case @xmath25=
@xmath26+@xmath27 ) where @xmath28 is equal to half the angular size .
.agns observation : for each source is given a ) the redshift , b ) the culmination zenith angle @xmath29 , c ) the hours of observation per day , d ) the flux above 0.6 tev . [ cols="^,^,^,^,^ " , ] the median energy @xmath21 corresponding to gamma rays showers with a given number of pads depends on the source spectrum and on the culmination zenith angle .
for n@xmath1540 the median energy ranges from @xmath20.6 tev ( for a source culminating at the zenith and with a steep spectrum like e@xmath23 ) up to @xmath22 tev ( for a source culminating at 20@xmath1 with a e@xmath30 spectrum ) .
4 shows the statistical significance of the observed sources , excluding the crab nebula and mrk421 , for n@xmath1540 .
no source shows a significance larger than 4 standard deviations , however the mean value of the distribution of the excesses for galactic sources is clearly shifted towards positive values ( [email protected] ) while the mean value for extragalactic sources is compatible with zero ( [email protected] ) .
we found 2 objects with a significance larger than 3 s.d for n@xmath860 , namely : \a ) mgroj1908 + 06 with 3.2 s.d .
( discovered by milagro@xcite , confirmed by hess@xcite and recently associated to the fermi pulsar 0fgl j1907 + 5 + 0602)@xcite ; \b ) the unidentified hess source hessj1841 - 055 , with 3.0 s.d .
; considered the number of trials ( 133 sources @xmath45 n@xmath13 intervals ) and the small significance of the excesses it is difficult to say if they are due to gamma rays or to a background fluctuation .
moreover the corresponding flux is larger than what expected from these source , in particular for hessj1841 - 055@xcite .
since agns are known to be variable on different time scales , we have monitored the time behaviour of a subset of 11 agns known to emit at e @xmath0 100 gev , and having a culmination zenith angle @xmath3120@xmath1 in our field of view ( see tab.1 ) .
the sources have been studied on time scales of 1 , 10 days and 30 days . for this analysis
we have considered the data taken in the period 2007 day 311 - 2009 day 89 . concerning the daily search , we found only one excess with a significance larger than 4 s.d .
( 4.3 s.d . ) , from the blazar 1es0229 + 200 on 2008 day 259 , for n@xmath8 40 .
however , considering the number of trials ( 11 sources @xmath4510 days @xmath4 5 n@xmath13 intervals ) this excess is consistent with a background fluctuation . concerning the 10 days analysis , the only excess larger than 4 s.d .
is due to mrk421 in the time intervals 2008 days 161 - 170 ( 4.6 s.d . ) during a strong x - ray flare . looking for 30 days excess , the search has been done shifting the 30 days interval in steps of 10 days .
also in this case we found several excesses from mrk421 with significances between 4 and 5 s.d .
, in particular in the intervals : 2008 days 1 - 30 , 71 - 100 , 81 - 110 , 91 - 120 , 141 - 170 , when several x - ray flares have been observed@xcite . for all sources except mrk421 (
for which we give the observed flux ) we calculate the upper limit to the flux at a confidence level of 3 standard deviations .
for each source we assumed a power law spectrum with a fixed spectral slope : dn / de = k e@xmath30 multiplied by an exponential factor e@xmath32 to take into account the absorption of gamma rays on the extragalactic background light ( ebl ) .
we evaluated @xmath33 interpolating the curves given by primack et al.@xcite for fixed redshifts in the range @xmath34=0.03 - 0.3 . lacking the absorption parameters outside this range , the upper limit for 3c66a ( @xmath34=0.444 ) and pg1553 + 113 ( @xmath34=0.36 )
have not been evaluated , while for m87 ( @xmath34=0.004 ) we assumed no absorption . from the number of observed events ( or the upper limit in case of no detection ) the corresponding gamma ray flux
is determined by a complete simulation process that evaluates the expected number of events from a source with a given spectrum and a given daily path in the sky .
tab.1 shows the observed integral flux above 0.6 tev for mrk421 and the upper limits obtained for the 8 agns with no detection , averaged over the whole observation period , for events with n@xmath1140 . to know the minimum flux observable by argo - ybj during a flare of a generic duration @xmath35 days
, one can multiply the upper limits given in the table by a factor @xmath36 , where @xmath37=424 is the total number of days considered in this analysis .
for a period of @xmath215 months argo - ybj monitored the gamma ray sky in the declination band from -10@xmath1 to + 70@xmath1 .
the observations of the background have revealed the existence of two regions with size of order 10@xmath130@xmath1 , where the cr flux is enhanced by approximately 0.1 % .
the origin of these medium scale anisotropies , previously observed by the milagro experiment , remains unexplained .
the data acquisition is currently going ahead with stable detection conditions .
studies to improve the detector sensitivity are in progress , both in the direction of increasing the angular resolution@xcite and of rejecting the cosmic rays background@xcite , implementing gamma - hadron separation algorithms .
aielli g. et al . , 2006 , nim a562 , 92 di sciascio g. et al . , 2007 , proc .
30th icrc , merida , mexico ( arxiv:0710.1945 ) di sciascio g. et al . , 2009 ,
31th icrc , lodz , poland gorski et al . , 2005 , apj 622 , 759 ( http://healpix.jpl.nasa.gov ) alexandreas d.e .
et al . , 1992 , nim a311 , 350 abdo a.a .
et al , 2008 , phys.rev.lett .
101 , 221101 zhang j.l .
et al . , 2009 ,
31th icrc , lodz , pol amenomori m. et al . , 2006 ,
science 314 , 439 drury l. & aharonian f. , 2008 , astrop.phys .
29 , 420 salvati m. & sacco b. , 2008 , astron .
485 , 2 , 527 martello d. et al . , 2009 ,
31th icrc , lodz , poland vernetto s. et al . , 2009 ,
31th icrc , lodz , poland abdo a.a .
et al . , 2009 ,
arxiv:0902.134v1 tevcat catalogue : http://tevcat.uchicago.edu abdo a.a .
, 2007 , apj lett .
664 , 91 djannati - atai a. et al . , 2007 , proc .
30th icrc , merida , mexico abdo a.a .
2009 , arxiv:0904.1018v1 aharonian f. et al . , 2008 , astron .
477 , 353 rxte / asm public data : http://xte.mit.edu/asmlc/asm.html primack j.r . ,
bullock j.s . , &
sommerville r.s .
, 2005 , aip conf .
745 , 23 dattoli m. et al . , 2009 ,
31th icrc , lodz , poland | a sky monitoring at gamma ray energy e @xmath0 0.6 tev has been performed by the full coverage extensive air shower detector argo - ybj , located in tibet at 4300 m of altitude .
we monitored 135 galactic and extragalactic gamma ray sources in the sky declination band from -10@xmath1 to + 70@xmath1 for 424 days , detecting the crab nebula and mrk421 with a significance respectively of 7.0 and 8.0 standard deviations . for a set of 11 agns known to emit in the tev energy range ,
the search has been performed in time scales of 1 , 10 and 30 days in order to study possible flaring activities .
significant emissions has been observed from mrk421 in the time scales of 10 and 30 days , during june and march 2008 , when the source had a strong activity also observed in the x - rays waveband . the analysis of the background has revealed the existence of a significant excess of the cr flux in two localized regions of angular size 10@xmath130@xmath1 , in agreement with previous indications .
gamma ray sources , agn , air showers |
the theory of strong interaction , quantum chromodynamics ( qcd ) , predicts that at sufficiently high energies nuclear matter transforms into a deconfined state where nucleons are no longer the basic constituents but rather quarks and gluons .
the event - by - event fluctuations @xcite of suitably chosen observables opens new possibilities to investigate the properties of the so called `` qcd phase diagram '' . in general
, one can classify fluctuations into two parts .
first , dynamical fluctuations which result from correlated particle production reflecting the underlying dynamics of the system . second , statistical fluctuations induced by the measurement process itself due to the finite event multiplicity . in case of independent particle production
the multiplicity fluctuations follow a poisson distribution . to quantify particle correlations the observable @xmath0was proposed in ref .
it is defined as @xmath3 = \dfrac{\langle n_{a}^{2 } \rangle}{{\langle n_{a } \rangle}^{2 } } + \dfrac{\langle n_{b}^{2 } \rangle}{{\langle n_{b } \rangle}^{2 } } - 2\dfrac{\langle n_{a}n_{b } \rangle}{\langle n_{a } \rangle\langle n_{b } \rangle } - \left(\dfrac{1}{\langle n_{a } \rangle } + \dfrac{1}{\langle n_{b } \rangle}\right),\ ] ] where @xmath4 and @xmath5 are the numbers of particle types a and b in a given event and @xmath6 denotes averaging over events .
the quantity @xmath0vanishes when the multiplicity distributions of both particle types are poissonian and independent from each other .
since the negative terms in the formula of @xmath0contain a correlation term , negative values of @xmath0 indicate the degree of correlation between particle species . in this article , we present the measurements of , and in pb - pb collisions at @xmath7=2.76 tev in alice .
the results are compared to the hijing @xcite and ampt @xcite models as well as to measurements at lower energies from cern - sps @xcite and rhic @xcite .
the present analysis uses about 13 million minimum bias pb - pb events at @xmath7=2.76 tev collected by the alice detector @xcite during the 2010 lhc run .
the subdetectors involved in the analysis were the time projection chamber ( tpc ) for the tracking and particle identification and the inner tracking system ( its ) for the vertexing .
moreover , the two forward v0 detectors were used for the centrality estimation . particle identification ( pid )
was based on the measurement of the specific ionisation energy loss , d@xmath1/d@xmath2 , in the tpc gas .
the charged particles reconstructed in the tpc in full azimuth and in the pseudo - rapidity range of @xmath80.8 were used in this analysis .
the momentum range was restricted to 0.2@xmath9@[email protected]
gev/@xmath11 in order to minimise the systematic uncertainties due to overlap of measured particle de / dx distributions .
when the d@xmath1/d@xmath2distributions of different particle species are well separated , a unique particle identification is possible .
however , this is not possible when the measured particle d@xmath1/d@xmath2distributions overlap . and
@xmath12 distributions for pions and protons within the momentum range of 0.3@xmath9@[email protected] gev/@xmath11.,width=415 ] the so called `` identity method '' @xcite proposes a solution to this misidentification problem .
the method follows a probabilistic approach using the inclusive d@xmath1/d@xmath2distributions measured in the tpc , and determines the moments of the multiplicity distributions by an unfolding procedure .
it requires two basic experimentally measurable event - by - event quantities , @xmath13 and @xmath12 , which are defined as @xmath14 , \quad \quad w_{j } \equiv \sum_{i=1}^{n(n ) } \omega_{j}(x_{i}),\ ] ] where @xmath2 stands for the d@xmath1/d@xmath2of a given track , @xmath15 is the d@xmath1/d@xmath2distribution of particle @xmath16 , @xmath17 is the inclusive d@xmath1/d@xmath2 distribution within a given event and @xmath18 is the number of tracks within the @xmath19th event . in other words , @xmath20 is a bayesian probability measure of being particle type @xmath16 for a given track .
thus , in case of perfect identification one expects @xmath21 , while this does not hold in case of misidentification .
the moments of the @xmath12 distributions @xmath22 can be easily constructed directly from experimental data .
the identity method provides the real moments @xmath23 of the particle multiplicity distributions by means of an unfolding procedure using the moments of these @xmath12 distributions .
examples for distributions of @xmath13 and @xmath12 in pb - pb data are shown in fig .
[ fig : wdist ] .
details of the derivation can be found in ref .
@xcite .
the identity method exploits the fits of inclusive d@xmath1/d@xmath2distributions for the calculation of @xmath13 .
since the overlap regions in the d@xmath1/d@xmath2distributions are also allowed , one needs very good fits of inclusive d@xmath1/d@xmath2spectra over the full momentum range covered in the analysis .
therefore , a good understanding of the detector response of the tpc is required . to this end ,
clean particle samples , which were retrieved from v0 decay particles ( pions from @xmath24 decays , protons from @xmath25 decays and electrons from photon conversions ) , were used .
[ fig : dedxdist ] left panel shows the determination of the detector response for pions , where a generalised gauss function was used for a better description of the response shape .
the detector response functions obtained in this way were later used as the fit functions in the inclusive d@xmath1/d@xmath2spectra fits as shown in fig .
[ fig : dedxdist ] right panel . [ cols="^,^,^,^ " , ]
in summary , dynamical fluctuations of identified particle ratios were measured as function of centrality in pb - pb collisions at @xmath7=2.76 tev with alice .
the identity method , which is planned to be used in further event - by - event analyses in alice , was applied to alice data successfully for the first time . the results for and are in qualitative agreement with models .
however , the negative values of observed in most peripheral collisions , which indicate a correlation between pions and protons , are not reproduced by the models . in all three cases , alice results in the most central events
indicate positive results consistent with poissonian expectation within 2@xmath26 which agree with the extrapolations based on the data at lower energies from cern - sps and rhic . as a next step , a more differential analysis of @xmath0 in terms of rapidity , @xmath27 , and charge sign dependence as well as
a study in p - p collisions is presently being prepared . | the study of event - by - event fluctuations of identified hadrons may reveal the degrees of freedom of the strongly interacting matter created in heavy - ion collisions and the underlying dynamics of the system .
the observable @xmath0 , which is defined in terms of the moments of identified - particle multiplicity distributions , is used to quantify the magnitude of the dynamical fluctuations in event - by - event measurements of particle ratios .
the alice detector at the lhc is well - suited for the study of @xmath0 , due to its excellent particle identification capabilities .
particle identification based on the measurement of the specific ionisation energy loss , d@xmath1/d@xmath2 , works well on a statistical basis but suffers from ambiguities when applied on an event - by - event level .
a novel experimental technique called the `` identity method '' is used to overcome such limitations .
the first results on identified particle ratio fluctuations in pb - pb collisions at in alice as a function of centrality are presented .
the alice results for the most peripheral events indicate an increasing correlation between pions and protons which is not reproduced by the hijing and ampt models .
on the other hand , for the most central events the alice results agree with the extrapolations based on the data at lower energies from cern - sps and rhic .
= 1 event - by - event fluctuations ; identified hadrons ; qcd phase transition ; heavy - ion collisions ; lhc |
for @xmath0 point vortices with cartesian coordinates @xmath1 and intensities @xmath2 , the hamiltonian equations of the motion are @xmath3 where the hamiltonian is @xmath4 for these equations , the poisson bracket is @xmath5 .
the system has three first integrals @xmath6 which are not involutive : @xmath7 however , two involutive integrals always exist , for example , @xmath8 and @xmath9 . with the help of these integrals , we can reduce the system by two degrees of freedom .
thus , in the case of three vortices the reduced system has one degree of freedom and it is integrable ( grbli , 1877 @xcite ; poincar , 1893 @xcite ) .
the four - vortex problem is reduced to a system with two degrees of freedom .
this system is not integrable ( s.ziglin , 1979 @xcite ) .
for @xmath0 point vortices moving on a sphere @xmath10 , with spherical coordinates @xmath11 and intensities @xmath2 , the hamiltonian equations can be written as ( bogomolov , 1977 @xcite ) @xmath12 with the poisson bracket @xmath13 and the hamiltonian @xmath14 here , @xmath15 is the sphere s radius , and @xmath16 is the angle between the vectors from the center of the sphere to the point vortices @xmath17 and @xmath18 so that @xmath19 beside the hamiltonian , the equations ( [ bmk - eq-6 ] ) have three independent non - involutive integrals @xmath20 the vector @xmath21 ( @xmath22 are the radius - vectors of the vortices ) with components @xmath23 is called the _ moment or center of vorticity_. the first integrals commute in the following manner : @xmath24 .
similar to the case of a plane , we can reduce the system by two degrees of freedom , using involutive integrals , for example , @xmath25 .
so , in the case of three vortices , we have a completely integrable system ( p.newton and r.kidambi , 1998 @xcite , 2000 @xcite ; borisov and lebedev , 1998 @xcite ) .
the four - vortex problem is reduced to a system with two degrees of freedom and , generally , is not integrable ( d.a . and a.a.bagrets @xcite ) .
to find new periodic solutions and apply the numerical analysis , we will do the most symmetric reduction of the system by two degrees of freedom .
we present here one of the possible ways to do such a reduction , based on new equations of motion for the mutual variables
. effective reduction of a four - vortex system was done by khanin @xcite ( for vortices of the same sign ) and by aref and pomphrey @xcite ( for equal vortices ) .
the generalized jacobi reduction by one degree of freedom in the point vortex dynamics was described in @xcite . in our study
, we used the universal algebraic method of effective reduction ( for an arbitrary number of vortices ) .
it is based on _ mutual variables representation _ of the motion equations of a system of point vortices @xcite , @xcite , @xcite .
as mutual variables , we introduce the following quantities . squared distances between the pairs of vortices @xmath26 and oriented areas of triangles @xmath27 , @xmath28 these variables are due to e.laura .
we put @xmath29 , @xmath30 .
then the momentum integral @xmath9 can be written as @xmath31 where @xmath0 is the number of the vortices .
writing the vortices coordinates in the complex form @xmath32 , we obtain the following representation for them : @xmath33 where @xmath34 is the angle between the vector from the @xmath35th vortex to the @xmath18th vortex and the positive direction of @xmath36 . by direct calculation
we can derive the following propositions , which describe the behavior of the reduced system of three and four vortices @xcite .
[ pro1 ] for three identical vortices , the evolution of the mutual distances @xmath37with fixed @xmath38 is described by a hamiltonian system with one degree of freedom . in terms of the canonical variables
@xmath39 , this system can be written as @xmath40 where @xmath41 , @xmath42 , @xmath43 .
[ pro2 ] for four identical vortices , the evolution of the mutual distances is described by a hamiltonian system with two degrees of freedom . in terms of the canonical variables
@xmath44 , this system is @xmath45 where @xmath46 the above canonical variables naturally follow from the lie - algebraic representation of the equations of motion .
according to , when @xmath47 are known , one needs to find the angles @xmath48 in order to determine the coordinates of the vortices .
clearly , only one of these angles is independent ( let it be @xmath49 ) , the remaining are obtained using the relations @xmath50 the evolution of @xmath51 is found by integrating the first order equation : @xmath52 there is an interesting geometric interpretation of the absolute motion of the periodic solutions of the reduced systems , .
[ pro3 ] let @xmath53 be a periodic solution @xmath37of period @xmath54 of the reduced system .
then @xmath55 there exists a frame of reference , rotating with the constant angular velocity @xmath56 about the center of vorticity , where each vortex moves along some closed curve @xmath57 @xmath58 the rotating velocity @xmath56 is given by ( accurate to @xmath59 , @xmath60 : @xmath61 @xmath62 if @xmath56 and @xmath63 are commensurable @xmath37i.e .
@xmath64 , @xmath65 then in the fixed frame of reference the vortices also move along closed curves ; @xmath66 if any of the curves @xmath67 transform into each other by rotation about the center of vorticity by an angle , commensurable to @xmath68 , then there is a ( rotating ) frame of reference where the corresponding vortices orbit one and the same curve .
proof in the right hand side of , there is a periodic function of period @xmath69 .
we have a convergent fourier expansion : @xmath70 integrating , and using , we find that all the angles @xmath51 depend on @xmath71 in the following way : @xmath72 where @xmath73 , while @xmath74 are @xmath69-periodic functions of time . substituting the above dependency into , we find that the positions of the vortices on the plane are given as follows : @xmath75 it follows that _ in the frame of reference , rotating about the center of vorticity with the angular velocity @xmath76 , all the vortices describe analytic closed curves , given by functions @xmath77_. the proofs of @xmath58 , @xmath62 , and @xmath66 , using the relations , are obvious .
@xmath78 without any modification , proposition [ pro3 ] is generalized to the case of @xmath0 vortices if we suppose that @xmath53 is a periodic solution of a reduced system with @xmath79 degrees of freedom to the @xmath0-vortex problem .
this section presents most well - known particular solutions for point vortices on a plane and on a sphere .
[ [ the - case - ofmathbbr2 ] ] _ the case of _
@xmath80 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 1 . collinear configurations ; 2 .
polygonal configurations ( in particular , polygonal configurations , embedded into each other ) .
an @xmath0-polygon ( thomson configuration @xcite ) is stable if @xmath81 , and unstable if @xmath82 ( the final result on the stability of the heptagon was obtained recently in @xcite ) ; 3 . non - symmetrical stationary configuration ( for @xmath83 ) @xcite . [ [ the - case - ofmathbbs2 ] ] _ the case of _
@xmath10 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 1 . collinear configuration .
+ the roots @xmath84 that define a collinear configuration , rotating with velocity @xmath76 , can be found as equilibriums of a system of @xmath85 particles on a circle with the hamiltonian @xmath86 + this system was obtained and studied by a.v.borisov and v.v.kozlov ( 1999 @xcite ) a special case with @xmath87 is called the dyson system .
it is proved that the system ( [ bmk - eq-10a ] ) is not integrable , but a _ quasi - integral _ exists , which gives a very good approximation of the behavior of systems at low energy .
analogs of thomson configurations @xcite .
+ in these configurations , the vortices are positioned at the same latitude @xmath88 in the vertices of a regular @xmath0-polygon and rotate about its center with the angular velocity @xmath89
recently , a new class of periodic solutions to the classical @xmath0-body problem has been obtained in celestial mechanics ( c.moore , r.montgomery , a.chenciner , c.sim , and others ) .
for example , they found a remarkable stable periodic orbit in the 3-body problem namely , a _ figure - of - eight _ @xcite . in this solution ,
three bodies follow each other along the same curve .
a long and ingenious analytical proof was done using the variational method based on the least action principle .
these solutions where all particles move along the same curve are called _ choreographies _ @xcite .
choreographies can be : * _ absolute _ ( in the fixed frame of reference ) or _ relative _ ( in rotating frames of reference ) ; * _ simple _ ( a single closed curve ) or _ complex _ ( more than one closed curve ) similar solutions exist for point vortex dynamics on a plane and on a sphere @xcite .
however , the least action principle can not be applied to an analytical proof in this case . using proposition [ pro3 ] and the known analytical solutions @xcite
, one can prove the following theorem for three - vortex systems : [ teo1 ] if , for the motion of three vortices of equal intensity , the constants of motion , @xmath9 and @xmath90 , satisfy the inequalities @xmath91 then such motion is a simple relative choreography ( see fig .
[ plane.eps ] ) .
note also that various properties of the motion of three vortices are discussed in @xcite and @xcite .
for example , the paper @xcite focuses on the study of stability of collinear configurations .
however , properties of the absolute motion have not actually been studied .
the four - vortex system has a remarkable symmetric analytical solution goryachev s solution @xcite ( see also @xcite ) where the vortices form a parallelogram as they move on a plane .
[ teo2 ] if , for the motion of four vortices , the vortices @xmath37of equal intensity@xmath92 form a centrosymmetrical configuration @xmath37a parallelogram@xmath92 , while the constants @xmath90 and @xmath9 satisfy the inequalities @xmath93 then the motion is a simple relative choreography ( see fig .
[ 4vortpar.eps ] )
. for the complete proofs , see @xcite .
let us show that for the system of four identical vortices , there exist choreographies , which do not coincide with goryachev s solution .
these are _ non - analytical choreographies_. these solutions can be obtained by variation of the hamiltonian near thomson s solution .
indeed , near the thomson configuration , the hamiltonian can be written in as @xmath94 where @xmath95 is a normal form @xmath96 and @xmath97 is the perturbation function ( a term of higher order ) .
on the level @xmath98 , there are precisely two nondegenerate periodic motions of the hamiltonian system with the hamiltonian @xmath95 : @xmath99 @xmath100 according to the lyapunov theorem , these periodic solutions are conserved under the perturbation @xmath97 and near thomson s solution there are two periodic solutions of the whole system @xmath101 : * solution ( [ nchor1 ] ) gives goryachev s solution , and * solution ( [ nchor2 ] ) gives a non - analytical relative choreography such as shown in fig .
[ trapez.eps ] . here
a solution is called non - analytical if it can not be expressed explicitly in terms of quadratures . as a rule
, this means that there are no explicit symmetries for these solutions , since all known symmetrical solutions can usually be found in terms of quadratures .
let us discuss in more detail , whether absolute choreographies exist in the three- and four - vortex problems . according to the above , any relative choreography ,
corresponding to a periodic solution ( of period @xmath69 ) of the reduced system ( see , ) closes in time @xmath102 . during this interval
the vortices pass through the same configurations @xmath103 times .
we denote the corresponding angular velocities of the frames of reference by @xmath104 . for three - vortex choreographies ( theorem [ teo1 ] )
, there exists a ( rotating ) frame of reference , where a choreography closes in the smallest possible time @xmath69 . for the case of four - vortex choreographies ( theorem [ teo2 ] )
, this time equals @xmath105 .
we denote and the corresponding angular velocities by @xmath106 and @xmath107 .
the angular velocities of other relative connected choreographies are now given by the following relations .
for three vortices : @xmath108 here @xmath109 and @xmath103 are coprime numbers . for goryachev s solution : @xmath110 where @xmath103 is odd , @xmath111 , the numbers @xmath18 and @xmath103 are coprime , while @xmath112 . here , the choreography that closes in @xmath113 corresponds to the velocity @xmath104 ( and the choreography that closes in @xmath114 corresponds to the velocity @xmath115 ) . and @xmath116 .
] and @xmath116 .
] figs .
4 and 5 show the angular velocities in terms of the energy parameter for three and four vortices , respectively .
the roots of @xmath117 correspond to the absolute choreographies shown in figs .
4a , b and 5a , b. it can be shown that there is a countable set of energy values such that choreographies described in theorems [ teo1 ] and [ teo2 ] are absolute and simple .
it can be shown , using the above - described methods and the reduction described in @xcite , that there are similar relative and absolute choreographies in the systems of three and four vortices on a sphere .
several examples of spherical choreographies of four vortices are given in fig .
6 ( they are quite similar to those shown in figs . 2 and [ trapez.eps ] ) . in @xcite and @xcite periodic ( in a fixed frame of reference ) solutions for different numbers of vortices
on a sphere are specified , which admit different discrete symmetry groups .
these motions are disconnected choreographies . using the group - theoretic methods
, one can construct a large number of disconnected choreographies in vortex dynamics on a sphere , generalizing those specified in @xcite and @xcite .
however , the choreographies that we found can not be obtained using these methods .
a.v.borisov , a.a.kilin , i.smamaev , _ reduction and chaotic behavior of point vortices on a plane and a sphere . _
submitted to proceedings for the aims firth international conference on dynamical systems and differential equations to be published in dcds
b , 2005 .
a. chenciner , j. gerver , r. montgomery , c. simo , _ simple choreographic motions of @xmath85 bodies : a preliminary study _ , geometry , mechanics , and dynamics ( volume dedicated to j. marsden , p. 287308 ) , springer , 2002 . | in this paper we describe new classes of periodic solutions for point vortices on a plane and a sphere .
they correspond to similar solutions ( so - called choreographies ) in celestial mechanics . * new periodic solutions for three or four identical vortices on a plane and a sphere*[multiblock footnote omitted ] a.v.borisov , i.s.mamaev , a.a.kilin + institute of computer sciences + udmurt state university + 426034 izhevsk , russia + e - mails : [email protected] , [email protected] , [email protected] + |
in the era of big data inferring features of complex systems , characterized by many degrees of freedom , is of crucial importance .
the high - dimensional setting , where the number of features to extract is not small compared to the number of available data , makes this task statistically or computationally hard .
one case of practical interest is the inference of the largest component ( eigenvector ) of correlation matrices .
consider @xmath0 independently drawn observations of @xmath1 interacting gaussian variables , _
i.e. _ such that the population covariance matrix @xmath2 is not the identity matrix .
if @xmath0 is much larger than @xmath1 the empirical covariance matrix @xmath3 computed from the @xmath0 observations converges to @xmath2 , and recovering the top eigenvector is easy .
the case where both @xmath4 are large ( sent to infinity at fixed ratio @xmath5 ) has received a lot of attention , both theoretically and practically @xcite . from a theoretical point of view , it has been shown , in the case of a covariance matrix @xmath2 with one ( or few compared to @xmath1 ) eigenvalues larger than unity , say , @xmath6 , that recovery is possible if @xmath7 is smaller than the critical value @xmath8 @xcite . for larger sampling noise ( @xmath9 ) , the top eigenvector of @xmath10 is essentially orthogonal to the top component of @xmath2 , and is therefore not informative .
it is reasonable to expect that the situation will improve in the presence of additional , prior information about the structure of the top component to be recovered , and that recovery will be possible even when @xmath7 is ( not too much ) larger than @xmath11 . that this is indeed the case has been rigorously shown when
all entries are nonnegative @xcite , and is supported by strong numerical evidence when the top component is known to have large entries ( finite as @xmath12 ) @xcite . in the present work , using techniques from statistical physics we propose explicit conjectures about the critical noise level and its dependence on the signal eigenvalue ( @xmath6 ) and on prior knowledge .
the framework is general and can be applied to any kind of entry - wise prior probability , _
i.e. _ factorized over the entries @xmath13 of the top component @xmath14 .
we show how rigorous results in the nonnegative case of @xcite are recovered , and present new results for the large entry prior . the motivation to consider the latter prior stems from computational biology , more precisely , from the study of coevolution between amino acids in protein families .
sequences of proteins diverged from a common ancestor widely differ across many organisms , while the protein structure and function are often very well conserved .
the constraints induced by structural and functional conservation manifest themselves as correlations between amino acids ( the @xmath1 variables , where @xmath1 is the protein length ) across the different organisms ( the @xmath0 observations ) .
recently , it was shown that the eigenmodes @xmath14 of the amino - acid covariance matrix corresponding to _ low _ eigenvalues were informative about three - dimensional contacts on the protein structure @xcite .
these modes show large entries on the protein sites and amino - acid types in contact ; as the other entries contain diffuse , non - structural signal @xcite , the components @xmath14 can not be thought of as being sparse .
the presence of large entries in structurally - informative components was empirically assessed through the so - called inverse participation ratio , @xmath15 ( for normalized @xmath14 ) , a quantity that remains finite for components with ( few ) large entries and otherwise vanishes for @xmath12 .
we hereafter use this quantity as a prior over the components to facilitate their recovery .
we consider the popular spiked covariance model , in which the entries of @xmath1-dimensional vectors , @xmath16 , are gaussian random variables with zero means and population covariance matrix @xmath17 .
all eigenvalues of @xmath17 but one are equal to unity , while the remaining eigenvalue is @xmath18 , with associated eigenvector @xmath19 . as usual
we choose @xmath20 but our results could be transposed to the case @xmath21 with minor modifications .
we draw @xmath0 independent samples @xmath22 , and define the sample covariance matrix @xmath23 , with entries @xmath24 .
the top eigenvector of @xmath23 is denoted by @xmath14 . in the double limit @xmath25 at fixed ratio @xmath5 , there exists a phase transition at a critical value of the sampling noise @xmath26 separating the _ high - noise regime _ , @xmath9 ,
in which @xmath14 and @xmath19 have asymptotically zero squared dot product , and the _ low - noise regime _
, @xmath27 , where the squared dot product between @xmath14 and @xmath19 is strictly positive with high probability @xcite .
the sample covariance matrix @xmath29 obeys a wishart distribution , determined by @xmath17 , @xmath1 and @xmath0 . using bayes formula we may write the likelihood ( density of probability ) for the normalized top component @xmath14 of @xmath23 as follows @xmath30 up to a normalization coefficient .
parameter @xmath31 above is equal to @xmath32 .
however , it is convenient to consider @xmath31 as a free parameter .
the @xmath33 limit corresponds to maximum likelihood inference , while working at low values of @xmath31 may be useful to ensure rapid mixing of monte carlo markov chain sampling of distribution @xmath34 , especially in the presence of prior information , see below .
we now assume that prior information over the population eigenvector @xmath19 is available under the form of a potential @xmath35 acting on the entries of @xmath14 .
the posterior distribution over the top component now reads @xmath36 up to a normalization coefficient .
three choices for the potential @xmath35 are shown in fig .
[ fig_pot ] .
motivated by previous works on protein sequence analysis , see introduction , we will hereafter mostly concentrate on @xmath37 , with @xmath38 ( fig .
[ fig_pot](a ) ) .
this potential favors the presence of large entries in the top component , but does not rule out the existence of many entries with small magnitude ( typically , of the order of @xmath39 ) .
it is therefore different from sparsity - enforcing potentials , such as @xmath40 in fig .
[ fig_pot](b ) .
exact results for the location of the transition in the nonnegative case ( fig . [ fig_pot](c ) ) were recently derived @xcite .
our formalism finds back those results , and can be applied to any potential @xmath35 as shown below .
corresponding to three prior information about the entries @xmath41 of the principal component : ( a ) large entries are present , ( b ) @xmath42 penalty favoring zero entries , ( c ) all entries are nonnegative .
, width=355 ]
we assume that the logarithm , divided by @xmath1 , of the normalization coefficient of @xmath34 in eq .
( [ rho2 ] ) , @xmath43 is concentrated around its expectation value @xmath44 $ ] in the @xmath45 limit , and compute the latter with the help of the replica method , a non rigorous technique commonly used in the statistical physics of disordered systems @xcite , see @xcite for recent applications to related high - dimensional inference problem .
we obtain @xmath46 where the extremum is computed over the order parameters @xmath47 $ ] , @xmath48 $ ] , @xmath49 $ ] , and the conjugated lagrange multipliers @xmath50 . here
, @xmath51 denotes the expectation over the @xmath34 distribution over @xmath14 in eq .
( [ rho2 ] ) .
the term @xmath52 in eq .
( [ logz ] ) depends on the prior potential @xmath35 and on the structure of the normalized population top component @xmath53 , more precisely , on how its entries scale with @xmath1 .
we now consider two cases of interest .
we assume first that the components of @xmath19 scales as @xmath54 , with @xmath55 finite , and denote by @xmath56 the distribution of the @xmath55 .
we focus on the nonnegative entry prior , for which @xmath57 for @xmath58 and 0 for @xmath59 .
we obtain @xmath60_z \ , \end{split}\ ] ] where erfc is the complementary error function , and @xmath61_z$ ] denotes the average of @xmath62 over the gaussian measure , @xmath63 . after some elementary algebra
, we obtain the expression for the overlap @xmath64_z^+}{[(x\,\tilde u+z)^2]_z^+}\ \text{with}\ x=\frac{\hat p}{\sqrt{2\hat q_1}}=\frac{(\gamma-1)\ , p}{\sqrt{r(1+(\gamma-1)p^2)}}\ ] ] where @xmath65_z^+=[\max(f(z),0)]_z$ ] .
these equations correspond to eqs .
( 7 ) , ( 8) , ( 9 ) , ( 21 ) and ( 23 ) of @xcite .
we now assume that @xmath53 has only @xmath66 ` large ' entries , @xmath67 , different from zero in the @xmath12 limit ( with finite @xmath66 ) , and that the other entries decay fast enough with @xmath1 , _
e.g. _ are of the order of @xmath68 . then @xmath69 while the above formula is valid for generic @xmath35 we restrict ourselves , in the remaining part of this article , to the potential @xmath37 .
furthermore , we assume that the @xmath66 finite entries of @xmath19 are all equal to @xmath70 ; the calculation can be easily extended to @xmath71 , or to the case of nonhomogeneous entries @xmath72 .
in addition we assume that ( a1 ) all @xmath13s take identical values in eq .
( [ logz ] ) ( homogeneous regime ) ; ( a2 ) @xmath14 has no large ( finite when @xmath12 ) entry @xmath13 on sites such that @xmath73 .
the validity of these assumptions will be discussed in the next section .
after some elementary algebra the extremization conditions reduce to the following set of @xmath74 coupled equations over @xmath75 and the first @xmath66 entries of @xmath14 : @xmath76 where @xmath77 note that the @xmath66 variables @xmath13s obey the same cubic equation and , hence , can take at most three different values as @xmath78 varies .
the equations above admit the solution @xmath79 , corresponding to the ` unaligned ' phase .
in addition , in some well - defined regions of the four - dimensional control parameter space @xmath80 solutions with @xmath81 may be found .
we give in section iv below results for three cases : ( a ) no prior ( @xmath82 ) ; ( b ) weak exploitation of many data with prior information ( small @xmath83 for finite @xmath84 ) ; ( c ) maximum _ a posteriori _ inference ( finite @xmath7 , and infinite @xmath31 and @xmath84 at fixed ratio @xmath85 ) .
we start with the case @xmath82 .
extremization conditions over the parameters in eq .
( [ logz ] ) give the value of the squared overlap @xmath86 between @xmath53 and @xmath14 for any @xmath31 .
we find : @xmath87 for @xmath88 whatever the value of @xmath31 , and , for @xmath27 , @xmath89 those expressions perfectly agree , in the @xmath33 limit , with known results for the spiked covariance model @xcite .
in addition our formalism gives access to the value of @xmath86 for finite @xmath31 .
note that , for @xmath27 , inference of the direction of @xmath53 is possible , _ i.e. _ @xmath90 , even for @xmath91 ( but larger than @xmath92 ) . at the critical noise , @xmath93 .
the above results imply that , in the absence of prior information ( @xmath82 ) , inference of the top component direction is possible at low @xmath94 provided the sampling noise @xmath7 is smaller than @xmath95 . in other words ,
when both @xmath31 and @xmath7 tend to zero with a fixed ratio @xmath96 , the aligned and not - aligned phases correspond , respectively , to @xmath97 , and @xmath98 .
we expect the critical ratio @xmath99 to be a drecreasing function of the prior strength @xmath84 , as stronger prior should facilitate the recovery of the large - entry top component @xmath19 .
resolution of eq .
( [ ppo ] ) gives the phase diagram shown in fig .
[ fig_pd ] .
several regimes can be identified , depending on @xmath84 : * for @xmath100 , the critical ratio @xmath99 remains unchanged , see fig .
[ fig_pd ] , and equal to @xmath101 .
as @xmath102 crosses this critical value the overlap @xmath103 continuously increases from 0 to a positive value , see fig .
[ fig_p2 ] .
* for @xmath104 , the aligned phase ( @xmath105 ) exist for @xmath106 , see dashed line in fig .
[ fig_pd ] . as @xmath102 crosses @xmath107 the squared overlap @xmath86 discontinuously jumps from 0 to @xmath108 , see fig .
[ fig_p2 ] . * for @xmath109 ,
the aligned phase ( @xmath105 ) is thermodynamically stable , meaning that the value @xmath110 of @xmath111 in eq .
( [ logz ] ) attached to this phase is larger than the one of the nonligned phase ( @xmath112 ) , @xmath113 , for @xmath114 , see full line in fig .
[ fig_pd ] .
the value of @xmath115 and of the overlap @xmath116 at the phase transition are the roots of the two coupled equations @xmath117 where the first equation actually implements the condition @xmath118 .
the phase transition is illustrated in the middle panel of fig .
[ fig_p2 ] for a specific value of @xmath84 .
the value of @xmath119 is defined from @xmath120 * for @xmath121 the prior strength is so strong that the inferred component @xmath14 has few large entries whatever the value of @xmath31 . for @xmath122 it is aligned ( @xmath105 ) with @xmath19 ( fig .
[ fig_p2 ] ) , while for @xmath123 , it is not , see fig .
[ fig_pd ] .
plane ; axis are rescaled by @xmath124 and @xmath125 factors , where @xmath66 is the number of nonzero components in @xmath126 .
dots locate the points @xmath127 , @xmath128 and @xmath129 , see text .
transitions between phases are shown by full lines .
the dashed lines show the limit of existence of the aligned phase , while the dot - dashed lines separate the regions with ( above ) and without ( below line ) homogeneity breaking among the @xmath66 large entries of @xmath14.,title="fig:",width=316 ] -.3 cm assumption ( a1 ) , see section [ poiuy ] , is trivially valid for @xmath130 , but is not necessarily correct for @xmath131 and strong prior strength , for which we expect that @xmath14 will condensate and one component , say , @xmath132 , will be larger than the other components , say , @xmath13 , with @xmath133 ( nonhomogeneous regime ) . the transition line between these two regimes
is identified upon imposing that the cubic equation over @xmath13 in eq .
( [ logz ] ) admits a two - fold degenerate solution @xmath41 , that is , @xmath134 .
the transition line is plotted in the phase diagram of fig .
[ fig_pd ] ( dot - dashed line ) , and ends up in the point of coordinate @xmath135 lying on the existence line ( dashed line ) .
$ ] , vs. control parameter @xmath136 for three prior strengths and @xmath130 .
, title="fig : " ] -.5 cm we now focus on map inference at finite sampling noise @xmath7 , whereas @xmath31 and @xmath84 are both sent to infinity , with a fixed ratio @xmath137 .
parameter @xmath138 , hereafter referred to as the slope , controls the relative magnitude of the @xmath10-dependent and prior terms in @xmath34 , see eq .
( [ rho2 ] ) , while the multiplicative factor @xmath7 is introduced in the definition of @xmath138 to compensate for the explicit dependence upon @xmath7 in @xmath34 . for simplicity we present results for @xmath130 only , the extension to larger @xmath66 being rather straightforward .
equations ( [ ppo ] ) admit the solution @xmath139 , and another solution , with @xmath140 as @xmath141 , with ratios @xmath142 , @xmath143 having finite limits .
after some simple algebra we obtain the following expresson for the slope as a function of the squared overlap for the latter solution : @xmath144 we show in fig .
[ fig_s1 ] the representative curve of @xmath86 vs. the slope @xmath138 , for @xmath7 below and above the critical noise level , @xmath8 , in the absence of prior . for @xmath27
the squared overlap is an increasing function of @xmath138 , starting from a non zero value for @xmath145 : the population eigenvector direction can be estimated without prior at low sampling noise @xcite , but the overlap is increased in the presence of prior . for @xmath146 a discontinuous jump is observed from @xmath112 to @xmath147 at a critical value of the slope , @xmath148 while the overlap further increases as @xmath138 exceeds @xmath149 .
remarkably , even for large sampling noise values , the presence of a sufficiently strong prior allows us to infer @xmath53 .
the value of @xmath149 as a function of the noise level @xmath7 is shown in fig .
[ fig_s2 ] ; for large noise levels we have @xmath150 .
, vs. slope @xmath138 for sampling noises @xmath151 ( top ) and 0.4 ( bottom ) .
note the presence of the discontinuous transition at @xmath152 in the latter case .
the randomly condensed solution appears for @xmath153 . here ,
@xmath154.,title="fig:",width=268 ] -.3 cm this aligned phase competes with a nonaligned , but condensed phase , in which assumption ( a2 ) , see section [ poiuy ] , is violated .
in other words , for @xmath138 and @xmath7 sufficiently large , @xmath14 has few large entries ( @xmath155 in the @xmath12 limit ) , but not along the directions @xmath78 corresponding to the large components of @xmath19 ; hence , @xmath87 . to describe this new phase we set @xmath72 to 0 in the expression for @xmath52 in eq .
( [ logz3 ] ) .
the corresponding optimization equations can be solved in the @xmath156 limit , with the result that the nonaligned , condensed regime exists for @xmath138 larger than @xmath157 \ .\ ] ] the value of @xmath158 as a function of the noise level @xmath7 is shown in fig .
[ fig_s2 ] . for small @xmath7 ( slightly above @xmath11 )
we observe that @xmath158 is larger than @xmath149 , as intuitively expected : it is favorable to condense @xmath14 along the direction of @xmath19 rather than any other direction .
it can be checked that the value of @xmath111 in eq .
( [ logz ] ) is higher for this phase than for the aligned condensed phase .
hence , as soon as @xmath138 exceeds @xmath158 the overlap @xmath103 vanishes .
( full line ) and @xmath158 ( dashed line ) as functions of the sampling noise @xmath7 .
insert : @xmath159 vs. @xmath7 .
the difference vanishes in @xmath160 . here , @xmath161 , _
i.e. _ @xmath162.,title="fig:",width=355 ] -2.6 cm for large noise levels , however , we have @xmath163 , which is asymptotically smaller than @xmath149 , see insert of fig . [ fig_s2 ] .
indeed , the threshold slopes @xmath149 and @xmath158 cross at a well - defined value of the noise , @xmath164 , which depends on the top eigenvalue @xmath6 .
we show in fig .
[ fig_s3 ] the behaviour of @xmath164 vs. @xmath6 , and compare it to the critical noise @xmath11 in the absence of prior .
we observe the presence of a region in the @xmath165 plane , above the critical line @xmath11 , where the direction of @xmath19 can be inferred thanks to the large - entry prior .
our replica symmetric theory predicts that the benefit of the prior does not extend to very large values of the signal eigenvalue @xmath6 and of the noise @xmath7 , see fig .
[ fig_s3 ] .
plane . the dashed line @xmath8 divides the plane into the weak noise region ( below line ) , where inference is possible withour prior , and the strong noise region ( above line ) .
the full line shows the value of @xmath164 , at which @xmath166 , as a function of @xmath6 . in between
the dashed and full lines , inference of the top component is possible in the presence of a prior with appropriate strength .
the two lines merge for @xmath167 , @xmath168 . here ,
@xmath130.,title="fig:",width=326 ] -1.9 cm
the non rigorous calculations done in this paper suggest that inference of the top component of the population covariance matrix is possible in the presence of prior information , even above the critical noise level of the spiked covariance model , in agreement with rigorous results for the nonnegative case @xcite and numerical investigations for the large - entry case @xcite .
many interesting questions have not been investigated in the present paper : how hard is the recovery problem from a computational point of view ?
are there ` dynamical ' phase transitions separating subregions in the aligned phase , where the top component can be recovered in polynomial time or not ?
if so how do these line compare to the ` static ' critical lines derived in this paper ?
in addition it would be interesting to investigate the validity of the replica - symmetric hypothesis used to derive the results above @xcite . though replica symmetry is generally expected to be correct for convex optimization problems what happens in nonconvex situations
is not clear .
for instance , inference of the top component with the nonnegative prior gives rise to a nonconvex optimization problem @xcite , but all rigorous results are exactly found back within our replica symmetric approach , see section iii.b . from this perspective it would be useful to investigate whether the present results are robust against replica symmetry breaking .
i am grateful to m.r .
mckay for the invitation to the asilomar 2016 conference , and to s. cocco and d. villamaina for useful discussions .
this work has benefited from the financial support of the cnrs inphyniti inferneuro project .
johnstone , _ high dimensional statistical inference and random matrices _ , proceedings of the international congress of mathematicians , madrid , spain , 307 - 333 , 2006 .
d.c . hoyle and m. rattray , _
principal - component - analysis eigenvalue spectra from data with symmetry breaking structure _ , phys .
e 69 , 026124 , 2004 . j. baik , g. ben arous g. and s. pch , _ phase transition of the largest eigenvalue for nonnull complex sample covariance matrices _ , ann .
33 , 1643 - 1697 , 2005 .
a. montanari and e. richard , _ non - negative principal component analysis : message passing algorithms and sharp asymptotics _
, ieee transactions on information theory 62 , 1458 - 1484 , 2016 .
d. villamaina and r. monasson , _ estimating the principal components of correlation matrices from all their empirical eigenvectors _ , europhys .
lett . 112 , 50001 , 2015 .
s. cocco , r. monasson and m. weigt , _ from principal component to direct coupling analysis of coevolution in proteins : low eigenvalue modes are needed for structure prediction _ ,
plos comp .
9 , e:1003176 , 2013 .
h. jacquin , a. gilson , e. shakhnovich , s. cocco , r. monasson , _ benchmarking inverse statistical approaches for protein structure and design with exactly solvable models _ , plos comp .
12 : e1004889 , 2016 .
a. engel and c.v .
den broeck , _ statistical mechanics of learning _ , cambridge university press , cambridge , england , 2001
. m. advani and s. ganguli , _ statistical mechanics of optimal convex inference in high dimensions _ , phys .
x 6 , 031034 , 2016 . y. nakanishi - ohno , t. obuchi , m. okada , y. kabashima , _ sparse approximation based on a random overcomplete basis _
, j. stat .
p063302 , 1 - 30 , 2016 f. krzakala , m. mzard , f. sausset , y. sun , l. zdeborova , _ probabilistic reconstruction in compressed sensing : algorithms , phase diagrams , and threshold achieving matrices _ , j. stat .
p08009 , 1 - 57 , 2012 . | the problem of infering the top component of a noisy sample covariance matrix with prior information about the distribution of its entries is considered , in the framework of the spiked covariance model . using the replica method of statistical physics the computation of the overlap between the top components of the sample and population covariance matrices
is formulated as an explicit optimization problem for any kind of entry - wise prior information .
the approach is illustrated on the case of top components including large entries , and the corresponding phase diagram is shown .
the calculation predicts that the maximal sampling noise level at which the recovery of the top population component remains possible is higher than its counterpart in the spiked covariance model with no prior information . random matrix theory , spiked covariance model , prior information , replica method , phase transitions |
a randomly branched tree is a finite connected set of bonds that contain no closed loops and which is embedded in a @xmath13 dimensional space . aside from their importance in statistical physics @xcite and the intriguing connection to relevant physical problems as percolation @xcite ,
randomly branched trees have received particular attention for being practically implicated in the modelling of branched @xcite , ring @xcite and supercoiled @xcite polymers . as customary in polymer physics @xcite , the behavior of randomly branched trees can be analyzed in terms of a small set of exponents describing how expectation values for observables characterising tree connectivities and conformations vary with the weight , @xmath4 , of the trees or the contour distance , @xmath14 , between nodes : @xmath15 here , @xmath16 denotes the average branch weight ; @xmath17 the average contour distance or length of paths on the tree ; @xmath18 the mean - square spatial distance between nodes with fixed contour distance ; @xmath19 the mean - square gyration radius of the trees ; @xmath20 the average number of intra - tree pair contacts ; @xmath21 the average number of pair contacts between different trees in melt . by construction , @xmath22 , and the relation @xmath23 is expected to hold in general @xcite .
exact values for the exponents are known only for a very few number of cases . for ideal non - interacting trees , the exponents @xmath24 and @xmath25 @xcite .
furthermore , @xmath26 where @xmath13 denotes the dimension of the embedding space . for interacting trees , the only known exact result @xcite is the value @xmath27 for self - avoiding trees in @xmath28 . on the other hand ,
numerical results @xcite as well as approximate theoretical calculations @xcite confirm that flory theories @xcite provide a useful framework for discussing the _ average _ behavior , eqs .
( [ eq : epsilon ] ) to ( [ eq : nu ] ) , of a wide range of interacting tree systems .
while being remarkably successful though , due to its simplicity flory theory is inevitably affected by serious known shortcomings and limitations @xcite .
they are , for instance , manifest in ( small ) deviations of predicted from observed or exactly know values for exponents as those defined in eqs .
( [ eq : epsilon ] ) to ( [ eq : gamma_c ] ) , and , importantly , it does not give any insight for the equally relevant exponent @xmath29 , eq .
( [ eq : beta ] ) . in the case of linear chains ,
these limitations are much more pronounced in the distribution functions for the corresponding observables which contain a wealth of additional information .
little seems to be known about the even larger range of configurational distribution functions for interacting trees . in the following , we present numerical results based on a detailed analysis of five different tree ensembles , which we have simulated in @xcite : ( i ) ideal randomly branching polymers , ( ii ) @xmath0 and @xmath1 melts of interacting randomly branching polymers , ( iii ) @xmath1 self - avoiding trees with annealed connectivity and ( iv ) @xmath1 self - avoiding trees with quenched ideal connectivity .
we analyse the distributions along the lines of known relations for ideal trees and linear chains and propose a coherent framework , including generalised fisher - pincus relations , relating most of the exponents characterising the distribution functions to each other and to the contact and flory exponents for interacting trees .
the paper is organised as follows : in section [ sec : theory ] we briefly summarise the theoretical background and list a number of useful results for ideal and interacting linear chains and trees . in section [ sec : modmethods ] we give a few details on the numerical methodologies employed for simulating the trees and analysing their connectivity .
finally , we present and discuss our results in sec .
[ sec : results ] and briefly conclude in sec .
[ sec : concls ] .
consider a tree of mass @xmath4 defined as @xmath30 monomers connected by @xmath4 bonds or kuhn segments .
for ideal trees , daoud and joanny @xcite calculated the partition function @xmath31 in the continuum approximation : @xmath32 where @xmath33 is the first modified bessel function of the first kind , @xmath34 the branching fugacity and @xmath35 . removing a randomly chosen bond ,
splits a branch of size @xmath36 from the remaining tree of size @xmath37 . for ideal trees , the probability distribution , @xmath2 , of branch sizes
is given by the kramers theorem @xcite @xmath38 and related to sum over all possible ways of splitting the tree . from this expression , it is possible to derive the following asymptotic relations : @xmath39 with @xmath40 , where @xmath41 is the scaling exponent describing the decay of @xmath2 in the large-@xmath4 limit , eq .
( [ eq : daoudjoanny_splitprob ] ) .
flory theories are formulated as a balance of an entropic elastic term and an interaction energy @xcite : @xmath42 in the standard case of self - avoiding walks , @xmath43 represents the entropic elasticity of a linear chain , while @xmath44 represents the two - body repulsion between segments , which dominates in good solvent . for interacting trees
, the elastic free energy takes the form @xcite : @xmath45 the expression reduces to the entropic elasticity of a linear chain for unbranched trees with quenched @xmath46 . the first term of eq .
( [ eq : fgutinprefactors ] ) is the usual elastic energy contribution for stretching a polymer of linear contour length @xmath14 at its ends @xcite .
the second term penalises deviations from the ideal branching statistics , which lead to longer paths and hence spatially more extended trees .
more formally , it is calculated from the partition function of an ideal branched polymer of @xmath4 bonds with @xmath14 bonds between two arbitrary fixed ends @xcite . for trees with quenched connectivity , eq .
( [ eq : fgutinprefactors ] ) has to be evaluated for the given mean path length @xmath14 and then minimised with respect to @xmath47 . for trees with annealed connectivity , eq .
( [ eq : fgutinprefactors ] ) needs to be minimised with respect to both , @xmath47 and @xmath14 . in this form ,
the theory predicts values for the exponents @xmath48 , @xmath49 , and @xmath50 for a wide range of tree systems @xcite as a function of the embedding dimension , @xmath13 , as well as relations between these exponents .
for example , optimising @xmath14 for annealed trees for a _ given _
asymptotic , @xmath51 , yields @xmath52 _ independently _ of the type of volume interactions causing the swelling in the first place .
plausibly , a fully extended system , @xmath53 , is predicted not to branch , @xmath54 , and to have a fully stretched stem , @xmath55 . for the radius of ideal randomly branched polymers , @xmath56 , one recovers @xmath57 and gaussian path statistics , @xmath58 . there is more to ( linear )
polymers than can be described by flory theory .
the number of self - avoiding walks is given by @xcite @xmath59 with a universal exponent @xmath60 and a non - universal constant @xmath61 characteristic of the employed lattice .
flory theory can neither predict the functional form of eq . (
[ eq : gamma_saw ] ) , nor the numerical value of @xmath60 , nor the related contact probability @xcite @xmath62 furthermore , flory theory incorrectly predicts that stretched chains exhibit essentially gaussian behavior with volume interactions becoming quickly negligible .
instead , the end - to - end distance distribution of self - avoiding walks is to an excellent approximation @xcite given by the redner - des cloizeaux ( rdc ) @xcite distribution @xmath63 with @xmath64 independent of @xmath4 . for small distances ,
( [ eq : rdc ] ) is dominated by the power law with the exponent @xmath65 given by the contact exponent eq .
( [ eq : theta_saw ] ) . for large distances
, the chain behaves like a string of @xmath66 blobs of size @xmath67 . for a given extension @xmath68
, the free energy of @xmath69 per blob implies that the exponents @xmath70 is given by @xcite @xmath71 interestingly , knowledge of the two exponents is sufficient to reconstruct the entire distribution function , because the constants @xmath72 and @xmath73 are determined by the conditions ( 1 ) that the distribution is normalized ( @xmath74 ) and ( 2 ) that the second moment was chosen as the scaling length ( @xmath75 ) : @xmath76 lattice trees are believed to fall into the same universality class as lattice animals @xcite with their number scaling similarly to eq .
( [ eq : gamma_saw ] ) : @xmath77 in particular , the tree and animal critical exponents in @xmath13-dimensions are related to the lee - yang edge singularity @xcite of the ising model in an imaginary magnetic field in @xmath78-dimensions , suggesting a relation between the entropy and the size @xmath79 interestingly , this relation suggests that it is possible to estimate the number of self - avoiding trees using flory theory .
in this section , we account very briefly for the algorithms used for generating equilibrated configurations of trees ( sec .
[ sec : treeensembles ] ) and the numerical schemes employed for their analysis ( sec .
[ sec : treesanalysis ] ) .
the reader interested in more technical details may look into our former works @xcite . a longer discussion is dedicated to how we extracted and extrapolated scaling exponents for rdc functions and contacts ( sec .
[ sec : finitesize ] ) .
quantitative details as well as tabulated values for single - tree statistics are presented in the supplemental material . to simulate randomly branching polymers with _
annealed _ connectivity , we employ a slightly modified version of the `` amoeba '' monte - carlo algorithm @xcite for trees on the cubic lattice with periodic boundary conditions . in the model , connected nodes occupy adjacent lattice sites .
as there is no bending energy term , the lattice constant equals the kuhn length , @xmath80 , of linear paths across ideal trees .
the functionality of nodes is restricted to the values @xmath81 ( a leaf or branch tip ) , @xmath82 ( linear chain section ) , and @xmath83 ( branch point ) .
we have studied ideal non - interacting trees , @xmath1 self - avoiding trees , and @xmath0 and @xmath1 melts of trees .
in addition , we have studied randomly branched trees with _ quenched ideal _ connectivity . for this ensemble ,
we have resorted to an equivalent off - lattice bead - spring model and studied it via molecular dynamics simulations . in this model ,
trees are represented by the same number of degrees of freedom , with bonds described as harmonic springs of average length equal to @xmath80 .
this allows a one - to - one mapping to and from the lattice model .
furthermore , beads interact with a repulsive soft potential whose strength is tuned in such a way , that the gyration radii of self - avoiding lattice trees remain invariant under the switch to the off - lattice model , if their quenched connectivities are drawn from the ensemble of self - avoiding on - lattice trees with annealed connectivity . for all our ensembles ,
the tree sizes are @xmath84 for ideal and self - avoiding trees @xcite and @xmath85 for melts of trees @xcite .
we have analysed tree connectivities using a variant of the `` burning '' algorithm for percolation clusters @xcite .
the algorithm works iteratively : each step consists in removing from the list of all nodes the ones with functionality @xmath86 and updating the functionalities and the indices of the remaining ones accordingly .
the algorithm stops when only one node ( the `` center '' of the tree ) remains in the list . in this way , by keeping track of the nodes which have been removed it is possible to obtain information about the mass and shape of branches
. the algorithm can be then generalized to detect the minimal path length @xmath87 between any pair of nodes @xmath88 and @xmath89 : it is in fact sufficient that both nodes `` survive '' the burning process . as discussed in detail in our former works
@xcite , extrapolation to the large-@xmath4 limit of scaling exponents is a delicate issue . in general , in fact , our data are affected by finite - size effects and extracted exponents are either ( i ) effective ( crossover ) exponents valid for the particular systems and system sizes we have studied or ( ii ) _ estimates _ of true , asymptotic exponents , which suffer from uncertainties related to the extrapolation to the asymptotic limit . _
extrapolating scaling exponents of distribution functions _ pairs of exponents ( @xmath90 ) , ( @xmath91 ) , and ( @xmath92 ) are obtained by best fits of data for distribution functions @xmath5 , @xmath93 and @xmath94 to the corresponding @xmath95-parameter redner - des cloizeaux functions ( eqs .
( [ eq : q_rdc_l ] ) , ( [ eq : q_rdc_path ] ) and ( [ eq : q_rdc_tree ] ) , respectively ) .
as shown in tables si - iii the values obtained from these fits display non negligible finite - size effects . then
, the search for extrapolated values has required two separate strategies . for exponents ( @xmath90 ) and ( @xmath92 ) , we follow a procedure similar to the one outlined first in @xcite and adopted later by us in @xcite .
it combines together the two following extrapolation schemes : 1 .
a fit of the data for @xmath96 and @xmath97 to the following 3-parameter fit functions : @xmath98 and @xmath99 and analogous expressions for @xmath100 and @xmath101 .
( [ eq : extrapolatethetalfuncts ] ) and ( [ eq : extrapolatethetatreefuncts ] ) correspond to a self - consistent linearisation of the 3 parameter fit @xmath102 around @xmath103 .
we have carried out a one - dimensional search for the value of @xmath104 for which the fits yield vanishing @xmath105 term . note that we have analyzed data for @xmath96 ( and @xmath100 ) in eq .
( [ eq : extrapolatethetalfuncts ] ) in the form `` @xmath106 _ vs. _ @xmath107 '' ( log - log ) , while for @xmath97 ( and @xmath101 ) we have used in eq .
( [ eq : extrapolatethetatreefuncts ] ) data in a log - linear representation , `` @xmath97 _ vs. _ @xmath107 . ''
these two different functional forms have been found to produce the best ( statistical significant ) fits .
2 . in the second method we fixed @xmath108 , and we calculated the corresponding @xmath95-parameter best fits to the same data .
results from the two fits are summarized separately in tables si and siii and their averages taken for our final estimates of scaling exponents ( see the corresponding boldfaced numbers ) . in the tables , we have also reported which ranges of @xmath4 have been considered for the best fits to the data . unfortunately , for @xmath109 and @xmath110 an analogous scheme can not be applied because we have data only for very limited ranges of @xmath6 .
then , our best estimates come from simply averaging single values together ( see boldfaced numbers in table sii ) . _ extrapolating scaling exponents of tree contacts _ tabulated values for intra - chain contacts , @xmath111 , and inter - chain contacts in tree melts , @xmath112 , are given in table siv .
corresponding extrapolated values of critical exponents @xmath113 and @xmath29 were obtained by the same methodology reported in our articles @xcite . for brevity ,
it was summarized in the caption of table siv .
otherwise , the interested reader can look into the above mentioned works for more details . in all cases ,
the quality of the fits is estimated by standard statistical analysis @xcite : normalized @xmath114-square test @xmath115 , where @xmath116 is the difference between the number of data points , @xmath117 , and the number of fit parameters , @xmath118 . when @xmath119 the fit is deemed to be reliable .
the corresponding @xmath120-values provide a quantitative indicator for the likelihood that @xmath121 should exceed the observed value , if the model were correct @xcite .
the results of all fits ( tables si - iv ) are reported together with the corresponding errors , @xmath122 and @xmath123 values .
unless otherwise said , all error bars for the estimated asymptotic values ( boldfaced numbers in tables si - iv ) are written in the form @xmath124(statistical error)@xmath124(systematic error ) , where the `` statistical error '' is the largest value obtained from the different fits @xcite and the `` systematic error '' is the spread between the single estimates . for brevity
, these are combined together into one single error bar in table [ tab : expsummary ] as : @xmath125 .
we begin with distribution functions for observables characterizing the tree connectivity : the distribution of branch weights ( sec .
[ sec : branchweightstatistics ] ) and path lengths ( sec .
[ sec : ppstat ] ) .
then , we turn to the conformational statistics of linear path on the tree ( sec . [ sec : confstatpaths ] ) and the distribution of internal distances ( sec .
[ sec : treestat ] ) .
( left - hand panels ) probability distribution functions for branch weight @xmath3 , @xmath2 , in trees of total mass @xmath4 .
( right - hand panels ) @xmath126-rescaled distributions as functions of @xmath127 .
in the top panel on the left , solid lines correspond to the exact analytical solution by daoud and joanny , eq .
( [ eq : daoudjoanny_splitprob ] ) .
left - hand and right - hand insets : corresponding differential fractal exponents @xmath128 and @xmath129 ( data were smoothed for better visualization ) are compared to the asymptotic scaling relation @xmath130 ( the regions within dashed lines , corresponding to the estimated values of `` @xmath131 '' as reported in refs .
, scaledwidth=49.0% ] in figure [ fig : pcutbranch ] we show the distribution , @xmath2 , of the weight , @xmath3 , of the branches generated by cutting randomly selected bonds in trees of size @xmath4 . obviously , it is possible to cut larger branches from larger trees , but independently of @xmath4 the vast majority of the branches is small .
this follows immediately from the fact that for all our systems a large fraction ( @xmath132 for ideal and melt of trees and @xmath133 for annealed self - avoiding trees
@xcite ) of the nodes is one - functional : cutting the bonds joining them to the tree generates branches of weight @xmath134 . to gain some intuition for the form of these distributions , it is useful to reconsider the case of ideal trees ( section [ sec : intro ] , eqs .
( [ eq : djpartfunct])-([eq : daoudjoanny_splitprob ] ) ) , where @xmath135 . in that case ,
@xmath136 , which simplifies to @xmath137 for small @xmath138 .
as expected , by plotting data for @xmath2 in log - log plots as a function of @xmath3 ( l.h . column of fig .
[ fig : pcutbranch ] ) we find good agreement with the expected power law for small @xmath3 , while plotting data as a function of @xmath127 ( r.h.s ) produces nearly perfect power law behavior over the entire range @xmath139 .
interestingly , we find similar power law behaviour for interacting trees , too .
we therefore tentatively generalize eq .
( [ eq : djpartfunct ] ) for the tree partition function to @xmath140 in fact , this form is compatible with the kramers theorem , eq .
( [ eq : kramers_splitprob ] ) , since @xmath141 .
the resulting average branch weight of @xmath142 implies @xmath143 .
as shown in the corresponding insets , the relation @xmath143 is well satisfied with values for @xmath144 taken from refs .
the numerical prefactor @xmath145 is related to the asymptotic branching probability ( see fig . 4 in ref .
@xcite and fig . 3 in ref .
@xcite ) , since in the limit @xmath146 , @xmath147 @xcite , and @xmath148 .
probability distribution functions , @xmath5 , for linear paths of length @xmath6 and average size @xmath17 in trees of total mass @xmath4 .
orange solid lines correspond to the one - dimensional redner - des cloizeaux function , eq .
( [ eq : q_rdc_l ] ) , with parameters @xmath96 and @xmath100 obtained from the best fits to data for @xmath149 ( ideal and self - avoiding trees ) and @xmath150 ( @xmath0 and @xmath1 melt of trees ) , see table si for detailed values of fit parameters .
inset : same plots in log - log scale .
, scaledwidth=49.0% ] as illustrated in fig .
[ fig : pathl_pdf ] , the measured path length distribution functions , @xmath5 , fall onto universal master curves , when plotted as a function of the rescaled path length @xmath151 : @xmath152 these master curves are well described by the one - dimensional redner - des cloizeaux ( rdc ) form ( orange lines in fig .
[ fig : pathl_pdf ] ) : @xmath153 the constants @xmath154 follow from the conditions that @xmath5 is normalized to @xmath155 and that the _ first _ moment , @xmath17 , is the only relevant scaling variable .
estimated values for @xmath156 obtained from best fits of eq .
( [ eq : q_rdc_l ] ) to data for specific values of @xmath4 and extrapolated values to large @xmath4 are summarized in table si . interestingly , we can give a physical interpretation of the observed ( effective ) exponents .
for small path lengths , results are not affected by the total tree size .
we thus expect to find @xmath157 segments at a contour distance @xmath158 from any node .
since @xmath159 , this suggests the scaling relationship @xmath160 to estimate the probability for observing very long paths , it is tempting to adjust the pincus - blob argument @xcite cited in the section [ sec : beyondflorysaw ] . a stretched tree should behave like a string of @xmath66 unperturbed trees of size @xmath161 , suggesting that @xmath162 the argument works well ( see bottom panel of table [ tab : expsummary ] , columns ( a ) and ( b ) ) when comparing the asymptotic values for @xmath100 in different ensembles to the corresponding numerical values for @xmath49 taken from @xcite .
interestingly , being only functions of @xmath49 , @xmath96 and @xmath100 can also be explicitly calculated in terms of results for @xmath49 from the flory theory @xcite , see top panel of table [ tab : expsummary ] .
interestingly , the flory theory gives a remarkable accurate prediction for most of the cases discussed .
probability distribution functions , @xmath163 , of spatial distances , @xmath164 , for linear paths of contour length @xmath6 .
orange solid lines correspond to theoretical predictions : for ideal trees data match the gaussian distribution , while melt of trees and self - avoiding trees are well described by the redner - des cloizeaux function , eq . ( [ eq : q_rdc_path ] ) , with parameters @xmath109 and @xmath110 obtained from the best fits to data for @xmath149 and @xmath165 ( self - avoiding trees ) and @xmath150 and @xmath166 ( @xmath0 and @xmath1 melt of trees ) , see table sii for detailed values of fit parameters .
insets : same plots in log - log representation . , scaledwidth=50.0% ] fig .
[ fig : path_e2edistance_distribution ] shows measured end - to - end vector distributions , @xmath9 , for paths of length @xmath6 on trees of mass @xmath4 .
the data superimpose , when expressed as functions of the scaled distances , @xmath167 : @xmath168 moreover , they are in excellent agreement with the rdc distribution ( eqs .
( [ eq : rdc ] ) , ( [ eq : rdc_c ] ) and ( [ eq : rdc_k ] ) and orange lines in fig .
[ fig : path_e2edistance_distribution ] ) : @xmath169 the shape of the rescaled distributions , and hence the characteristic exponents @xmath109 and @xmath110 , depend on the universality class .
not surprisingly , paths on ideal trees are well described by the gaussian distribution , i.e. @xmath170 and @xmath171 . fitted values for the other cases
are listed in table sii .
given the limited range of available path lengths @xmath6 , we have found no meaningful way to estimate asymptotic values .
we have then simply taken the average of the available fitted values ( see boldfaced numbers in table sii ) .
again , the observed values can be given a physical interpretation .
the exponent @xmath109 describes the reduction of the contact probability , eq .
( [ eq : theta_saw ] ) , relative to a nave gaussian estimate .
importantly , it is a genuinely novel exponent , _
i.e. _ it is _ independent _ from all other exponents discussed in refs .
@xcite and in this work . in the case of self - avoiding walks
, @xmath65 is related to the entropy exponent @xmath60 , eq .
( [ eq : theta_gamma_nu ] ) @xcite .
interestingly , grosberg and colleagues argued @xcite that the identical flory predictions of @xmath172 for self - avoiding walks and for the path statistics in melts of annealed lattice trees suggests a deeper analogy between the two problems @xcite .
using @xmath173 and @xmath174 for two and three - dimensional self - avoiding walks @xcite and eq .
( [ eq : theta_gamma_nu ] ) suggests @xmath175 and @xmath176 . in particular
, the @xmath1 value is in very good agreement with our finding @xmath177 ( tables [ tab : expsummary ] and sii ) , while the @xmath0 value appears smaller than the reported @xmath178 .
notice though , that these exponents were measured for path length @xmath179 and then finite - size effects may likely induce a bias on the final result . the exponent @xmath110 controls the non - linear path elasticity at large elongations .
the measured effective values can be compared to the fisher - pincus relation , eq .
( [ eq : t_saw ] ) , for self - avoiding walks @xcite @xmath180 where specific values for @xmath50 are taken from refs .
@xcite , see bottom panel of table [ tab : expsummary ] ( columns ( a ) and ( b ) ) . in general , agreement is overall good .
the only exception is for @xmath1 self - avoiding trees with quenched ideal statistics , which , again , may be ascribed to the limited range of path lengths of our simulated trees . as for @xmath96 and @xmath100 and being a function of @xmath50 only , specific values for @xmath110 can be also obtained by using the flory results @xcite for @xmath50 ( top panel of table [ tab : expsummary ] ) : again , for most of the cases , there exist fair agreement with numerical predictions . probability distribution functions , @xmath181 , of spatial distances , @xmath164 , between pairs of nodes .
orange solid lines correspond to the redner - des cloizeaux distribution function , eq . ( [ eq : q_rdc_tree ] ) , with parameters @xmath97 and @xmath101 obtained from the best fits to data with @xmath149 ( ideal and self - avoiding trees ) and @xmath150 ( @xmath0 and @xmath1 melt of trees ) , see table siii for detailed values of fit parameters .
insets : same plots in log - log scale . ,
scaledwidth=50.0% ] fig .
[ fig : monmondistspdf_comdistspdf ] ( left panels ) shows distributions @xmath7 of vectors @xmath8 connecting all tree nodes .
the data superimpose , when expressed as functions of the scaled distances , @xmath182 : @xmath183 again , the distributions are in excellent agreement with the rdc form ( eqs .
( [ eq : rdc ] ) , ( [ eq : rdc_c ] ) and ( [ eq : rdc_k ] and orange lines in fig .
[ fig : monmondistspdf_comdistspdf ] ) : @xmath184 the extracted exponents and corresponding extrapolations to @xmath185 are listed in table siii . in the following ,
we relate the characteristic exponents @xmath97 and @xmath101 to our previous results by using eqs .
( [ eq : q_rdc_path ] ) and ( [ eq : q_rdc_l ] ) together with the convolution identity @xmath186 which states that the local density can be calculated by adding up the contributions from paths of all possible length , @xmath187 .
the behavior of @xmath181 for large distances , @xmath188 , can be estimated from the contour distance @xmath189 , which makes the dominant contribution to particle pairs found at the spatial distance @xmath164 . combining the arguments of the compressed exponentials in eqs .
( [ eq : q_rdc_l ] ) and ( [ eq : q_rdc_path ] ) , this requires the minimization of @xmath190 and yields @xmath191 results from computer simulations for asymptotic exponents @xmath101 and @xmath48 @xcite support well this relation ( see bottom panel of table [ tab : expsummary ] , columns ( a ) and ( b ) ) . in the limit of small distances , @xmath192 , there are two possibilities . if there is no power law divergence in the small @xmath6 limit , the integral is dominated by contributions from long paths with @xmath193 , allowing to set the exponential term in eq .
( [ eq : q_rdc_path ] ) equal to one .
the only @xmath164-dependence comes through the explicit @xmath194 term and hence @xmath195 in the opposite limit , short paths dominate .
the apparent divergence of the integrand in the limit @xmath196 is removed by the exponential tail of eq .
( [ eq : q_rdc_path ] ) : paths with vanishing contour lengths up to @xmath197 do not contribute to the monomer density for finite spatial distances @xmath164 . for longer paths
, we can set the exponential to one . in this case
@xmath198 summarizing @xmath199 since @xmath200 for ideal trees and since we expect @xmath201 and @xmath202 for interacting trees .
table [ tab : expsummary ] compares the asymptotic results to theoretical predictions .
again , the general agreement is fairly good . finally , as for @xmath96 , @xmath100 and @xmath110 , specific values for @xmath97 and @xmath101
can be calculated by resorting to the flory results for @xmath48 ( top panel of table [ tab : expsummary ] ) . once again
, the predictions prove to be remarkably accurate . to conclude the section , in fig .
s1 in supplemental material we focus on distribution functions of monomers around the tree center of mass .
clearly , for the central node this distribution is gaussian .
on the other hand , distant nodes can not distinguish between the central node and the tree center of mass and their positions hence follow again a rdc distribution .
averaged over node identities , the monomer distribution around the tree center of mass seems to become gaussian for self - avoiding trees , but _ not _ for ideal trees or trees in melt .
interestingly , the latter effect was already noted a long time ago for _ ideal _ linear chains @xcite .
we turn then to the average number of self - contacts per tree , @xmath203 ( l.h . panel of fig .
[ fig : contacts ] and table siv ) .
consider an arbitrary pair of monomers .
the probability to find them in close contact scales as @xmath204 .
since there are @xmath205 different monomer pairs , we have @xmath206 this prediction compares extremely well with our numerical estimates for @xmath113 , see table [ tab : expsummary ] . note that the mean - field estimate @xmath207 holds only for ideal trees in @xmath208 dimensions .
the melt case is marginal in that we expect @xmath209 and thus @xmath210 : by using the estimated asymptotic values of @xmath48 in @xmath0 and @xmath1 @xcite and @xmath97 , the different values for @xmath113 compare well for all studied ensembles ( see bottom panel of table [ tab : expsummary ] , columns ( a ) and ( b ) ) . in all other cases , @xmath210 independently of @xmath48 , @xmath109 and @xmath97 , indicating that the local monomer density is finite and independent of tree size .
this is yet another illustration of the subtle cancellation of errors in flory arguments , which are built on the mean - field estimates of contact probabilities @xcite .
for tree melts , we have also considered the average number of contacts between nodes on _ different _ trees , @xmath112 ( r.h .
panels of fig .
[ fig : contacts ] and tables siv ) . in the melt ,
the average number of contacts per node @xmath211 is @xmath4-independent ( see also table siv ) , hence @xmath212 where @xmath213 is the exponent of the power - law correction to the large-@xmath4 behavior of @xmath214 : @xmath215 with @xmath216 and @xmath217 numerical prefactors .
@xmath213 can be calculated by considering the two leading terms in eq .
( [ eq : p_convolution ] ) for small @xmath164 s after substituting the upper bound of the integral with @xmath218 . since @xmath219 we get @xmath220 .
the average number of self - contacts per tree @xmath20 is proportional to the integral of the former expression from @xmath221 to some small cut - off spatial distance , or @xmath222 .
consequently , @xmath223 and @xmath224 in particular , eq .
( [ eq : betadef2 ] ) implies that @xmath225 . by employing the asymptotic values of @xmath48 @xcite and @xmath109 , eq .
( [ eq : betadef2 ] ) shows good agreement with direct estimates of @xmath29 values ( see bottom panel of table [ tab : expsummary ] , columns ( a ) and ( b ) ) .
+ & relation to & ideal trees , & @xmath0 melt of trees , & @xmath1 melt of trees , & @xmath1 self - avoiding trees , & @xmath1 self - avoiding trees , + & other exponents & annealed connect .
& annealed connect .
& annealed connect .
& annealed connect .
& quenched ideal connect .
+ & & & & & & + @xmath96 & @xmath226 & @xmath155 & @xmath227 & @xmath228 & @xmath229 & + & & & & & & + @xmath100 & @xmath230 & @xmath95 & @xmath231 & @xmath232 & @xmath233 & + & & & & & & + @xmath109 & & @xmath221 & @xmath234 & @xmath234 & @xmath234 & @xmath234 + & & & & & & + @xmath110 & @xmath235 & @xmath95 & @xmath236 & @xmath237 & @xmath238 & @xmath239 + & & & & & & + @xmath97 & @xmath240 & @xmath221 & @xmath221 & @xmath221 & @xmath241 & @xmath242 + & & & & & & + @xmath101 & @xmath243 & @xmath244 & @xmath95 & @xmath245 & @xmath246 & @xmath95 + & & & & & & + @xmath113 & @xmath247 & @xmath248 & @xmath155 & @xmath155 & @xmath155 & @xmath155 + & & & & & & + @xmath29 & @xmath249 & & @xmath250 & @xmath251 & & + & & & & & & + + + + & & & & & + & & & & & + & ( a ) & ( b ) & ( a ) & ( b ) & ( a ) & ( b ) & ( a ) & ( b ) & ( a ) & ( b ) + & & & & & & & & & & + @xmath96 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 & & + & & & & & & & & & & + @xmath100 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & @xmath264 & @xmath265 & @xmath266 & @xmath267 & & + & & & & & & & & & & + @xmath109 & @xmath221 & & @xmath268 & & @xmath269 & & @xmath270 & & @xmath271 & + & & & & & & & & & & + @xmath110 & @xmath95 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 + & & & & & & & & & & + @xmath97 & @xmath281 & @xmath221 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & @xmath288 & @xmath289 + & & & & & & & & & & + @xmath101 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath297 + & & & & & & & & & & + @xmath113 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 & @xmath305 & @xmath306 & @xmath307 & @xmath253 + & & & & & & & & & & + @xmath29 & & & @xmath308 & @xmath309 & @xmath310 & @xmath311 & & & & + & & & & & & & & & & +
in the present article , we have pursued our investigation of the conformational statistics of various types of lattice trees with volume interactions @xcite : @xmath0 and @xmath1 melts of lattice trees with annealed connectivity as well as @xmath1 self - avoiding lattice trees with annealed and with quenched ideal connectivity .
the well understood case of ideal , non - interacting lattice trees with annealed connectivity @xcite always serves as a useful reference .
here we have complemented the earlier analyses @xcite of the average behaviour by reporting results for distribution functions for observables characterising tree conformations and connectivities . in particular , we found that branch weight distributions follow a generalized kramers relation , eq .
( [ eq : kramers_splitprob ] ) , with @xmath312 and that path length and distance distributions are non - gaussian and closely follow redner - des cloizeaux @xcite distributions eqs .
( [ eq : pl],[eq : q_rdc_l ] ) , ( [ eq : pr_of_l],[eq : q_rdc_path ] ) and ( [ eq : pr],[eq : q_rdc_tree ] ) of the type @xmath313 which are fully characterised by pairs of additional exponents , @xmath65 and @xmath70 , summarised in table [ tab : expsummary ] .
the various exponents @xmath70 describe the compressed exponential large distance ( large path length ) behavior of the distribution .
our results suggest , that they obey generalized fisher - pincus @xcite relations , eqs .
( [ eq : tl ] ) , ( [ eq : tpath ] ) , and ( [ eq : ttree ] ) .
the exponents @xmath65 characterize small distance ( small path length ) power law behavior and are hence related to contact probabilities ( eqs .
( [ eq : gamma_c theta_tree ] ) and ( [ eq : betadef2 ] ) ) . we have related them to each other and the other tree exponents ( eqs .
( [ eq : theta_l ] ) and ( [ eq : thetatree ] ) ) .
the only exception is the exponent @xmath109 for the small distance behavior of the path end - to - end distance distribution .
the situation is similar to the well - known case of linear self - avoiding walks , where the corresponding exponent @xmath65 and the closure probability are related to the entropy exponent @xmath60 , which can not be predicted by flory theory .
additional work is required to corroborate the proposed relation @xcite between the self - avoiding walk exponents and those characterising tree melts . in conclusion ,
interacting randomly branching polymers exhibit an extremely rich behaviour and swell by a combination of modified branching and path stretching @xcite . as previously shown @xcite , the average behaviour @xcite in the various regimes and crossovers can be well described by a generalised flory theory .
our present results demonstrate , that this is not the case for distribution functions .
nonetheless , their non - gaussian functional form can be characterised by a small set of exponents , which are related to each other and the standard trees exponents .
the good agreement of the predicted relations with the numerical data suggests that we now dispose of a coherent framework for describing the connectivity and conformational statistics of interacting trees in a wide range of situations .
44 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) .
, , , , ( ) . , * * , ( ) . , ( ) . ,
_ _ ( , ) . , , , * * , ( ) . ,
_ _ ( , , ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . ,
* * , ( ) . ,
_ _ ( , , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( , , ) . , _ _ ( , , ) . , _ _ ( , , ) . ,
* * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . ,
* * , ( ) . ,
* * , ( ) . ,
* * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) .
, , , _ _ ( , ) . , , , * * , ( ) . ,
_ _ ( , ) . , , , , _ _ ( , , ) , ed . , * * , ( ) .
[ cols="^,^,^,^,^ " , ] & ideal trees & & & annealed connect . &
_ ideal _ connect .
+ & @xmath314 & @xmath314 & @xmath315 & @xmath314 & @xmath315 & @xmath314 & @xmath314 + & @xmath316 & @xmath317 & @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath322 + 5 & @xmath323 & @xmath324 & @xmath325 & @xmath326 & @xmath327 & @xmath321 & @xmath328 + 10 & @xmath329 & @xmath330 & @xmath331 & @xmath332 & @xmath333 & @xmath334 & @xmath335 + 20 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & @xmath340 & @xmath341 & + 30 & @xmath342 & @xmath343 & @xmath344 & @xmath345 & @xmath346 & @xmath347 & @xmath347 + 45 & @xmath348 & @xmath349 & @xmath350 & @xmath351 & @xmath352 & @xmath353 & + 75 & @xmath354 & @xmath355 & @xmath356 & @xmath357 & @xmath358 & @xmath359 & @xmath360 + 150 & @xmath361 & @xmath362 & @xmath363 & @xmath364 & @xmath365 & @xmath366 & + 230 & @xmath367 & @xmath368 & @xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath373 + 450 & @xmath374 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & @xmath379 & @xmath380 + 900 & @xmath381 & @xmath382 & @xmath383 & @xmath384 & @xmath385 & @xmath386 & @xmath387 + 1800 & @xmath388 & & & & & @xmath389 & @xmath390 + + & @xmath391 & @xmath391 & @xmath392 & @xmath391 & @xmath392 & @xmath391 & @xmath391 + & @xmath393 & @xmath394 & @xmath395 & @xmath396 & @xmath397 & & + @xmath122 & @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & &
+ @xmath403 & @xmath404 & @xmath405 & @xmath406 & @xmath407 & @xmath408 & & + & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath413 & & + & @xmath221 & @xmath221 & @xmath221 & @xmath221 & @xmath221 & @xmath221 & @xmath221 + @xmath122 & @xmath414 & @xmath415 & @xmath416 & @xmath417 & @xmath418 & @xmath419 & @xmath420 + @xmath403 & @xmath421 & @xmath422 & @xmath423 & @xmath424 & @xmath425 & @xmath426 & @xmath427 + & @xmath428 & @xmath429 & @xmath430 & @xmath431 & @xmath432 & @xmath433 & @xmath434 + & @xmath435 & @xmath436 & @xmath437 & @xmath438 & @xmath439 & @xmath440 & @xmath441 + & @xmath442 & @xmath443 & @xmath444 & @xmath445 & @xmath446 & @xmath447 & @xmath448 + | while flory theories @xcite provide an extremely useful framework for understanding the behavior of interacting , randomly branching polymers , the approach is inherently limited .
here we use a combination of scaling arguments and computer simulations to go beyond a gaussian description .
we analyse distributions functions for a wide variety of quantities characterising the tree connectivities and conformations for the four different statistical ensembles , which we have studied numerically in refs .
@xcite : ( a ) ideal randomly branching polymers , ( b ) @xmath0 and @xmath1 melts of interacting randomly branching polymers , ( c ) @xmath1 self - avoiding trees with annealed connectivity and ( d ) @xmath1 self - avoiding trees with quenched ideal connectivity .
in particular , we investigate the distributions ( i ) @xmath2 of the weight , @xmath3 , of branches cut from trees of mass @xmath4 by severing randomly chosen bonds ; ( ii ) @xmath5 of the contour distances , @xmath6 , between monomers ; ( iii ) @xmath7 of spatial distances , @xmath8 , between monomers , and ( iv ) @xmath9 of the end - to - end distance of paths of length @xmath6 . data for different tree sizes superimpose , when expressed as functions of suitably rescaled observables @xmath10 or @xmath11 . in particular , we observe a generalised kramers relation for the branch weight distributions ( i ) and find that all the other distributions ( ii - iv ) are of redner - des cloizeaux type , @xmath12 .
we propose a coherent framework , including generalised fisher - pincus relations , relating most of the rdc exponents to each other and to the contact and flory exponents for interacting trees . |
imagine a stationary droplet in quiescent conditions . for a real system
there are , of course , no velocities .
yet when this simple system is simulated with a lattice boltzmann method you will find a flow - field around the drop . even if you start with a no flow initial condition this velocity field will develop .
an example of these velocities in shown in figure [ fig1 ] ( a ) .
these velocities are know as `` spurious velocities '' .
the magnitude of the velocities depends on the details of the method , the radius of the drop , the surface tension and the viscosity .
there have been several studies@xcite of different methods that tell us about the dependence of the spurious velocities on these parameters . in the past the focus of the work has been on reducing the magnitude of these velocities but to the best of my knowledge no good understanding of their origin has been reached . in this article
i will explain why we see these spurious velocities at all and how , and at what cost , they can be avoided .
i will first introduce a simple lattice boltzmann method for model b dynamics and show the perfect approach to equilibrium which is free of any spurious currents .
then i extend this model to include hydrodynamics and we will see that spurious currents suddenly appear .
a simple examination shows why there should not be any spurious currents according to the continuum equations and i point out the terms through which discretization errors drive the spurious currents .
i introduced a thermodynamically consistent discretization of these terms which leads to a lattice boltzmann method that is free of spurious currents .
model b is a model describing the behavior of binary alloys .
we will consider a mixture of , say , @xmath0 and @xmath1 atoms .
the system is characterized by the order parameter @xmath2 which represents the difference in the densities of the two components , _
i.e. _ @xmath3 represents pure @xmath0 and @xmath4 represents pure @xmath1 .
the dynamics of this conserved order parameter @xmath2 is then given by @xmath5 where d is a diffusion coefficient and @xmath6 is the chemical potential .
the system is described by the free energy @xmath7 . for simplicity
we will consider a landau free energy expansion around the critical point which is given by @xmath8 to simulate this equation we use a bgk lattice boltzmann method given by @xmath9 \label{lbg}\ ] ] where the relaxation time @xmath10 may depend on @xmath2 and the set of discrete velocities @xmath11 will be determined later .
the behavior of the model will be determined by the moments of the equilibrium distribution @xmath12 .
we name these moments @xmath13 a taylor expansion to second order in the derivatives gives @xmath14\ ] ] we now take the first moment ( @xmath15 ) of this expansion and obtain as the equation of motion for the order parameter @xmath2 @xmath16 we have two independent ways to ensure that this equation be equivalent to ( [ order ] ) .
either we choose @xmath17 @xmath18 , @xmath19 or alternatively @xmath20 and @xmath21 note that the @xmath22 term is now third order . in two dimensions we can implement both of these approaches with a small velocity set of five velocities @xmath23 .
both approaches behave equivalently , as expected , but they differ in the ranges of stability as will be discussed elsewhere@xcite . both implementations of this simple diffusive lattice boltzmann algorithm are free from spurious currents .
we conclude that hydrodynamics is required to see spurious currents .
in order to simulate binary fluids we have to introduce a fluid velocity which will convect the order parameter @xmath2 . this fluid velocity will obey the navier - stokes equation which is coupled to the order parameter @xmath2 through a thermodynamics pressure tensor @xmath24 .
we will denote the fluid density as @xmath25 .
the equations of motion are@xcite @xmath26\end{aligned}\ ] ] where the pressure is @xmath27 , @xmath28 is the viscosity and @xmath29 the number of spatial dimensions . the pressure tensor @xmath24 is defined as @xmath30\mathbf{1 } + \kappa \nabla \phi\nabla\phi\ ] ] we can implement these continuum equations using lattice boltzmann . with ( [ lbg ] )
we now define @xmath31 . by imposing @xmath32 and @xmath33
we obtain the required order parameter equation ( [ orderc ] ) with either of the previous choices for @xmath34 and @xmath35 .
the navier - stokes equations are obtained with a usual one - component lattice boltzmann method which is coupled to the order parameter equation through a body force of @xmath36 . + ( a ) + ( b ) doing this leads to the well known spurious velocities as shown in figure [ fig1 ] .
we see that there are not only spurious currents but also the chemical potential @xmath6 is no longer constant . since the implementation of a simple model b did not lead to the spurious velocities it is reasonable to assume that the implementation of the additional lattice boltzmann equation for the total density and the momentum is the reason for the occurrence of the spurious velocities .
why should the continuous equations in equilibrium be consistent with a quiescent drop without spurious velocities ?
this is because the two driving terms @xmath37 and @xmath36 are related through @xmath38 this means that a constant chemical potential will lead to zero driving force in both equations .
when we now use this knowledge and replace the driving force @xmath39 with @xmath40 we find that the spurious velocities vanish to machine precision .
the density of the order - parameter in the surrounding fluid is @xmath41 and the density in the drop is @xmath42 ( cf figure [ fig1 ] ) .
the size of the maximum velocity is now @xmath43 .
the chemical potential has the value of @xmath44 and variations are smaller than @xmath45 .
so why do the spurious velocities appear in the first place ?
this is because the discretizations of @xmath36 and @xmath46 are very different .
we conclude that the different discretizations of the driving forces for the order parameter and momentum equations are the origin of spurious velocities for lattice boltzmann .
there is , however , a caveat .
the term @xmath40 in its discrete form is not a divergence of a scalar field which means that momentum is now only approximately conserved . to be able to see the absence of spurious velocities i included a tiny correction term in the definition of the momentum that ensures that the total momentum of the system does not change .
also my implementation of the algorithm turned out to be unstable so i added a small amount of numerical viscosity by multiplying a velocity by @xmath47 and adding 0.00025 times the velocity of the four nearest neighbors .
this rendered the simulations stable .
i proved that spurious velocities in one particular lattice boltzmann implementation are caused by non - compatible discretizations of the driving forces for the order - parameter and momentum equations .
the different discretization errors for the forces compete and drive the spurious currents .
i believe that the same argument holds for all lattice boltzmann methods that exhibit spurious velocities .
these spurious velocities can be avoided by ensuring that the discretizations of the driving forces are compatible . + * acknowledgement * this work was funded by epsrc gr / m56234 and gr / r67699 . | stationary droplets simulated by multi - phase lattice boltzmann methods lead to spurious velocities around them . in this article
i report the origin of these spurious velocities for one example and show how they can be avoided . |
the observable polarization field is described in terms of the two stokes parameters @xmath0 and @xmath1 with respect to a particular choice of axes about each direction on the sky .
we use spherical polar coordinates , with orthonormal basis vectors @xmath2 and @xmath3 .
the stokes parameters define a symmetric and trace - free ( stf ) rank two linear polarization tensor on the sphere @xmath4.\ ] ] a two dimensional stf tensor can be written as a sum of ` gradient ' and ` curl ' parts @xmath5 where @xmath6 is the covariant derivative on the sphere , angle brackets denote the stf part on the enclosed indices , and round brackets denote symmetrization .
the underlying scalar fields @xmath7 and @xmath8 describe electric and magnetic polarization respectively and are clearly non - local functions of the stokes parameters .
one can define scalar quantities which are local in the polarization by taking two covariant derivatives to form @xmath9 and @xmath10 which depend only on the electric and magnetic polarization respectively . for band limited data
one could consider taking these derivatives of the data to extract the e and b components .
the problem is that in the neighborhood of boundaries it becomes harder to measure the second derivatives , so the noise properties are not straightforward : taking the derivatives is effectively increasing the noise near the boundaries relative to the rest of the map . rather than taking second derivatives it it useful to work with integrals over the surface . as an example we focus here on the b polarization , since this is of most interest .
we define the surface integral @xmath11 where @xmath12 is a real window function defined over some patch @xmath13 of the observed portion of the sky . the factor of minus two is included to make our definition equivalent to that in ref .
@xcite . integrating by parts we have @xmath14 where @xmath15 is an stf tensor window function .
thus extraction of the b polarization amounts to measuring a well defined surface integral , _ and two line integrals_. it is these line integrals that encode the troublesome aspect of the e / b decomposition .
since the e / b decomposition is inherently non - local , it is quite natural to work in harmonic space .
the polarization tensor @xmath16 can be expanded over the whole sky in terms of stf tensor harmonics @xmath17 where @xmath18 and @xmath19 are the gradient and curl tensor harmonics of opposite parities defined in ref .
@xcite . from the orthogonality of the spherical harmonics over the full sphere
it follows that @xmath20 in a rotationally - invariant ensemble , the expectation values of the harmonic coefficients define the power spectrum : @xmath21 when we have data over a section of the sphere , the observed data can be encoded in a set of pseudo - multipoles @xmath22 and @xmath23 obtained by including a window function @xmath12 in the integral of eq .
: @xmath24 where the hermitian coupling matrices are given by @xmath25 the matrix @xmath26 controls the contamination with electric polarization and can be written as a line integral around the boundary of the cut defined by @xmath12 .
the matrices can be evaluated easily numerically in term of the harmonic coefficients of the window @xmath27 using @xmath28 { { { \begin{pmatrix}}l & l_1 & l_2 \\ m & -m_1 & m_2 { \end{pmatrix}}}}\end{gathered}\ ] ] where @xmath29 .
vectors are now denoted by bold roman font , e.g. @xmath30 has components @xmath31 , and matrices are denoted by bold italic font , e.g. @xmath32 have components @xmath33 . including the results equivalent to the above for the e polarization
, we then have the harmonic relations @xmath34 harmonic methods of e / b separation amount to ways of solving these equations for linear combinations of the observed @xmath35 , @xmath36 such as to give results that depend only on @xmath37 or @xmath30 . in principle
the matrices are formally infinite , though in practice we measure the components of @xmath38 up to some finite @xmath39 .
however the pseudo - harmonics to finite @xmath39 can still can contain contributions from the underlying fields at all @xmath40 unless the fields are band limited ( i.e. there exists some finite @xmath39 above which all the components of @xmath37 and @xmath30 are zero ) , or the coupling matrices are sufficiently localized .
thus in general the @xmath32 coupling matrices are rectangular , including the coupling to the @xmath41 and @xmath42 modes on all scales . for square coupling matrices
the symmetry of the spherical harmonics implies that @xmath43 for the matrix @xmath44 which satisfies @xmath45 .
this implies that eigenvalues of @xmath46 come in @xmath47 pairs : @xmath48 and hence @xmath49 .
these properties hold for all @xmath39 and are a direct result of our window on the sky being a real function . for a window function which is positive ( or zero ) everywhere
@xmath50 is positive semidefinite , and the eigenvalues are bounded between zero and one for window functions normalized to lie between zero and one . in the limit that @xmath51
the coupling matrix becomes a projection operator onto the section of sky observed ( from now on we assume the window function is taken to be zero or one everywhere ) , so @xmath52 this property is ensured by the completeness of the harmonics , and shows that @xmath53 for an eigenvector @xmath54 of @xmath50 , with @xmath55 , these relations imply that @xmath56 & = & ( 1-\lambda ) [ { \bm{w}}_- { \mathbf{e}}_\lambda].\end{aligned}\ ] ] it follows that @xmath57 is also an eigenvector of @xmath50 with eigenvalue @xmath58 , and we can define @xmath59 so that @xmath60 .
hence the eigenstructure of @xmath46 is given by @xmath61 thus two sets of eigenvectors of @xmath50 lie in the null space of @xmath46 : @xmath62 the set of vectors @xmath63 form a basis for the set of supported pure - b and pure - e modes and the @xmath64 form a basis for modes that are zero within the observed regions and therefore can not be measured .
the remaining modes with @xmath65 form a basis for a set of ambiguous modes . as @xmath66 the elements of the coupling matrix @xmath67 where @xmath68 is the coupling matrix for the scalar harmonics ( see ref @xcite ) .
completeness of the scalar harmonics implies @xmath68 is a projection matrix and hence has a fraction @xmath69 unit eigenvalues and @xmath70 zero eigenvalues , where @xmath71 is the fraction of the sky included in the cut .
thus we expect a fraction @xmath72 of the eigenvalues of @xmath50 to be unity and @xmath70 to be zero , with the other eigenvalues making up a fraction @xmath73 corresponding to the boundary to area ratio .
these results only apply in the limit that @xmath51 , however for finite @xmath74 we expect a large number of eigenvalues of @xmath50 close to zero or one , corresponding to modes which are either very well or very poorly supported over the observed area .
to measure the @xmath42 only we look for a matrix @xmath75 such that @xmath76 assuming such a @xmath77 can be found , we have then have a vector @xmath78 of pure - b modes @xmath79 that have no dependence on @xmath37 . the projection onto @xmath41
is generated similarly with @xmath80 where @xmath81 is the identity matrix .
the pure e and b modes are then given by @xmath82 where @xmath83 here we assume there exists some @xmath39 for which all components of @xmath30 , _ and @xmath37 _ , are negligible for @xmath84 . we can perform a singular value decomposition ( svd ) to write @xmath85 where @xmath86 the matrices @xmath87 and @xmath88 are column unitary , and @xmath89 is diagonal
. since we have observations over only part of the sky , many elements of @xmath89 will be close to zero , indicating modes that are not supported over the observed area .
the tilded variables are constructed by deleting the corresponding columns and rows of @xmath87 and @xmath90 , making @xmath91 a smaller square matrix .
the approximation can be made the same order as the numerical precision by choosing the threshold for the diagonal elements of @xmath89 small enough .
premultiplying by @xmath92 we then have @xmath93 where the hermitian matrix @xmath94 is defined by @xmath95 the vector @xmath96 contains essentially all the observable information about @xmath30 .
solving we have @xmath97 { { \begin{pmatrix}}{\tilde{\mathbf{e}}}\\ { \tilde{\mathbf{b}}}{\end{pmatrix}}}&\approx & ( 1-{\bm{h}}^2 ) { \tilde{{\bm{d}}}}{\tilde{{\bm{v}}}}^\dag { \mathbf{b}}\\ { \tilde{{\bm{u}}}}^\dag{\bm{r}}^\dag\left [ { \bm{i}}- { \tilde{{\bm{u}}}}{\tilde{{\bm{u}}}}^\dag \right ] { { \begin{pmatrix}}{\tilde{\mathbf{e}}}\\ { \tilde{\mathbf{b}}}{\end{pmatrix}}}&\approx & ( 1-{\bm{h}}^2 ) { \tilde{{\bm{d}}}}{\tilde{{\bm{v}}}}^\dag { \mathbf{e}},\end{aligned}\ ] ] which achieves the e / b separation .
it is in the form of a projection operator to remove the left - range of the coupling , followed by a reduction into the basis of partially supported modes by multiplication with @xmath92 . for band limited skies in the absence of noise
there is no information loss .
however for modes corresponding to non - zero eigenvalues of @xmath94 the signal to noise is decreased relative to doing no separation .
this is consistent with the understanding that the noise on the second derivatives of the data needed to perform direct e / b decomposition becomes larger as you near the boundary of the region .
note that this method is not useful for extracting large scale b modes from cmb polarization observations since even though b may be effectively band limited at low @xmath40 , there is also expected to be a e signal with a high band limit with @xmath98 .
vectors of harmonics are of size @xmath99 which makes the method computationally infeasible for @xmath100 without simplifying symmetries . on the full sky it is possible to impose a low band limit by convolution , however on the cut sky
this is not possible without mixing information from inside and outside the cut .
we now consider what happens to the above results as the band limit is removed . in the limit
@xmath51 we can use to show @xmath101 and from that @xmath102 diagonalizes @xmath50 with @xmath103 .
taking @xmath102 to have the related eigenvectors @xmath54 and @xmath59 in adjacent columns , it also follows from that @xmath104 is block diagonal , where the blocks are either zero or off - diagonal @xmath105 matrices with eigenvalues @xmath106 .
this implies that @xmath94 is block diagonal , with elements or zero , and hence that @xmath107 is diagonal , with elements zero or one .
thus each mode in @xmath108 can either be measured exactly ( for the zero eigenvalues of @xmath107 ) or is completely lost by the @xmath109 separation ( for the unit eigenvalues ) .
the lost modes that can not be separated correspond to the ` ambiguous ' modes discussed in ref .
since the zero eigenvalues of @xmath107 are determined by the null space of @xmath46 , in this limit @xmath109 separation amounts to projecting out the non - zero eigenvalues of @xmath46 , corresponding to the boundary terms . for non - band limited skies in the limit @xmath51 the method for projecting out the range of @xmath46 advocated in ref .
@xcite is therefore optimal for general sky patches , in addition to being optimal in the simple azimuthal case analysed in detail ( see also appendix [ azimuthal ] ) . in fig .
[ newmode ] we show two modes which for a band limit of @xmath110 are pure @xmath42 , but have non - zero projection into the range of @xmath46 .
it is clear that these are dominated by a line integral around the boundary , and as such they have significantly worse noise than the other modes due to the non - zero eigenvalue of @xmath107 .
the line integral is sensitive to e power on scales with @xmath111 .
as the band limit increases the eigenvalue of @xmath107 tends to one , and the line integrals can no longer be measured ( the signal to noise goes to zero ) .
+ for band limited skies one can use knowledge of the band limit to use the information from the modes with non - zero eigenvalues of @xmath46 . however in practice the @xmath41 polarization is effectively non - band limited as far as extracting the large scale @xmath42 signal is concerned .
thus one needs to find modes which lie in the null - space @xmath46 to extract pure b modes .
note that it is the non - band limitedness of @xmath41 which implies inevitable loss due to mode separation , even in the limit of zero noise .
[ ] [ ] [ 1.7]@xmath40 [ ] [ ] [ 1.7]@xmath112 in the case where the e signal is effectively non - band limited , but we observe only a finite number @xmath113 of pseudo - harmonics with @xmath114 , a pure @xmath42 mode @xmath115 can be extracted by finding a vector @xmath116 lying in the left null space of @xmath117 , where @xmath117 is in general an @xmath118 matrix , coupling in @xmath41 on all scales .
this requires that @xmath119 where here the @xmath50 is the hermitian square @xmath120 matrix , @xmath121 and @xmath122 is @xmath123 .
for a supported mode with @xmath124 this criterion is satisfied because @xmath125 and @xmath126 must be positive or zero .
note that @xmath127 is sufficient but not necessary for @xmath128 to vanish in the case of an azimuthal patch there is a left null space for finite @xmath39 even though there are no vectors satisfying @xmath129 ( see appendix [ azimuthal ] ) .
however in general , without including information up to the band limit of the data or special symmetries , this can not be done .
the idea here is to perform e / b separation for the low multipoles without including data up to the band limit ( which due to the @xmath130 scaling would be infeasible with current computers for general patches ) .
in general for finite @xmath113 there will be no fully supported modes , however for an eigenvector @xmath116 of @xmath50 with @xmath131 it follows that @xmath132 . in other words ,
approximately pure @xmath42 modes are simply the well supported modes .
note that @xmath133 is necessary but not sufficient since it does not ensure there is no coupling to power from @xmath41 on scales with @xmath134 in general .
the signal variance of @xmath115 is given by @xmath135 where @xmath136 the scalar contribution to the @xmath137 power spectrum is typically between @xmath138 times larger than @xmath139 for @xmath140 and levels of b detectable by planck ( see fig .
[ cls ] ) , and remains of the same order of magnitude up to @xmath141 .
taking @xmath142 the e contribution to the variance is @xmath143 , thus we require @xmath144 for clean separation of b. however there is little point removing e to levels much lower than the experimental noise , so for noise limited observations larger values of @xmath145 could be used .
[ ] [ ] [ 1.7]@xmath40 [ ] [ ] [ 1.7]@xmath146 in fig . [ wins ] we show a few window functions @xmath147 and @xmath148 for nearly pure b modes , using @xmath149 and the realistic cut discussed in the next section .
this low value of @xmath150 is not sufficient to extract many well supported modes ( there is only one with @xmath151 ) , however it allows us to compute the rectangular matrices @xmath50 and @xmath46 up to much higher @xmath40 in order to explicitly show the coupling of higher multipoles as a function of the choice of @xmath145 .
the expected behavior is demonstrated in the figure , with small values of @xmath145 effectively removing the coupling to e on all scales .
we now demonstrate explicitly the extraction of large scale b modes from a realistic non - symmetric sky cut , the ` kp2 ' cut used by the wmap analysis @xcite .
this excludes the galactic region , in addition to a number of other foreground sources , in total excluding about @xmath152 of the sky , and is shown as the solid area in fig .
[ polmap ] .
this may not be a good cut for polarized observations , but the cut is realistic in the sense that it does not have any artificial symmetries , and therefore is a reasonable test case for how b mode extraction can work in practice . as shown in fig .
[ cls ] the large scale reionization signal with @xmath153 is large and has high signal to noise , at least if the optical depth really is @xmath154 .
we use @xmath155 , which is computationally tractable ( @xmath156 of memory , few days of cpu time ) and sufficient to get most of the large scale reionization power .
the well supported mode with @xmath151 and maximal signal is shown in fig .
[ polmap ] . with more effort ( e.g. distributing the computation over a cluster or more efficient algorithms )
it may be possible to push to @xmath157 , possibly enough to extract essentially all of the separable b modes due to primordial gravitational waves .
this may be essential if the reionization turns out to be less significant , and of course resolving the shape of the magnetic power spectrum is a key check that the signal really is due to primordial tensor modes .
separation on smaller scales ( e.g. as a consistency check , and for studying the lensing signal ) would have to proceed on azimuthally symmetric cuts over clean regions of the sky with a number of cuts around foreground sources ( see appendix [ azimuthal ] ) , or rely on quadratic methods that work very well for estimating the separated power spectra on small scales . identifying the well supported modes
is straightforward , and requires diagonalization of the hermitian matrix @xmath50 , @xmath158 since we only need the well supported modes , in practice we only need to compute the eigenvectors with eigenvalues[multiblock footnote omitted ] between @xmath159 and @xmath160 , for some choice of @xmath145 and @xmath39 . for large @xmath39
there are @xmath161 such modes , where @xmath71 is the fraction of the sky included in the window . for less large @xmath39 a significant fraction of modes
will only be partially supported ; losing the subset of these modes that are not contaminated with @xmath41 is the price we pay for a computationally tractable analysis with manageable @xmath39 .
the well supported eigenvectors define a reduced column - orthogonal matrix @xmath162 which projects the cut - sky harmonics into the space of well supported nearly - pure b modes : @xmath163 a map of the nearly pure b modes can then be constructed using the harmonics @xmath164 .
note that though computing @xmath162 is time consuming for large @xmath39 , once this has been done for a particular cut , construction of b maps for different realizations is computationally fast . in fig .
[ maps ] we show the construction of a b map for a simple simulation without noise .
the recovered b map is not the same as the input map since we are only using @xmath155 , and some modes are lost due to e / b separation .
however it is clear than many of the large scale features are present in the recovered map , and that the method works usefully well . in general
there will be anisotropic and correlated noise on the polarization measurements .
the e / b separation methods we discussed give e / b clean separation of the signal , however any interpretation of the separated modes would have to carefully model the actual noise properties in the observation under consideration .
in general the e and b modes will have complicated noise correlations .
however in the idealized case of isotropic uncorrelated noise on the stokes parameters the noise properties are straightforward @xcite .
in particular the well supported e and b modes will also have uncorrelated isotropic noise @xmath165 the complications introduced by correlated anisotropic noise do not effect the _ signal _ e / b separation , which is purely determined by the geometry of the sky cut .
as the noise in a region inside the cut increases , some of the modes will become much more noisy . in the limit of infinite noise in this region ,
corresponding to an additional cut , these modes have zero signal to noise and can be removed .
this removes combinations of e and b modes in such a way that the number of pure b modes that can be measured decreases , and new ambiguous modes ( which are combinations of e and b with finite noise ) are generated .
[ planck ] shows how the e / b separation method performs with regards to the detection probability for magnetic polarization due to tensor modes , taking a simple model of the planck satellite as a test case ( assuming isotropic gaussian noise , that foregrounds can be subtracted accurately outside the cut region , and systematics are negligible ) .
we assume the primordial tensor modes are gaussian with scale invariant power spectrum , with @xmath166 being the variance of the transverse traceless part of the metric tensor , and that other cosmological parameters are well known .
the approximate method applied to the asymmetric cut does not perform quite as well as an optimal analysis with an azimuthally symmetric cut , but the difference is not that large equivalent to reducing a @xmath167 confidence detection to about @xmath168 .
the asymmetric cut is not expected to perform as well because it increases the area adjacent to a boundary and hence the number of ambiguous modes ( though the approximate result may be improved slightly by taking larger @xmath39 ) .
the plot also shows the result of applying the non - exact method ( retaining just the well supported modes ) to the azimuthally symmetric cut , and shows that the results are almost identical to performing the exact separation in this case . in the exact azimuthal analysis
the result is almost completely insensitive to @xmath39 , with @xmath150 results lying on top of those shown .
this is because of the small number of large signal to noise modes on large scales for a significant optical depth to reionization . in the asymmetric case much larger @xmath39
is needed to obtain well supported modes . for @xmath169
it is clear that planck should have a good chance of detecting tensor amplitudes @xmath170 , corresponding to an energy scale @xmath171 at horizon crossing during inflation .
the sensitivity to tensor magnetic modes however depends on the reionization optical depth , as shown in fig .
if the optical depth turns out to be at the lower end , the asymmetric cut with @xmath172 would not perform so well compared to the exact azimuthal case since there would be significant power on smaller scales , requiring @xmath173 to construct all the well supported modes .
[ ] [ ] [ 1.5]tensor amplitude @xmath166 [ ] [ ] [ 1.5]detection probability [ ] [ ] [ 1.5]tensor amplitude @xmath166 [ ] [ ] [ 1.5]detection probability
we have demonstrated explicitly that e / b separation is possible on large scales for realistic non - symmetric sky cuts . if the reionization optical depth is large almost all the detection significance for tensor modes would come from the very largest scales , where mode mixing is most important .
we showed that these modes can be separated in a computationally tractable way by retaining only the well supported modes .
the ambiguous modes that can not be separated will be dominated by the e signal if the noise is low , and whilst useful for analysis of the e polarization would not add significantly to a full likelihood analysis for magnetic polarization . with map - level separation one can also in principle study the full distribution of the b signal , for example testing for gaussianity .
analysing a pure b map is essentially the same as analysing a temperature map , and much more more straightforward and conceptually cleaner than doing a full joint analysis . for noise
limited observations , when the loss of ambiguous modes loses a significant amount of information , an analysis with e / b separated variables still provides a valuable cross - check on results from a full joint analysis , and may help to isolate systematics .
i thank ue - li pen for stimulating discussions , anthony challinor for pointing out an important error in an early draft , and sarah bridle for helpful suggestions .
i acknowledge use of the healpix @xcite and lapack packages .
for a circular boundary at constant latitude @xmath174 ( i.e. the boundary of an azimuthal patch ) , modes with different @xmath175 decouple , @xmath46 is block diagonal , and modes with different @xmath175 can be handled separately .
the @xmath46 surface boundary integral evaluates to @xcite @xmath176_\theta , \label{eq : wminusdcmp}\ ] ] where the vectors @xmath177 for @xmath178 and some arbitrary @xmath179 . in the @xmath180 limit
this is the spectral decomposition ( @xmath181 and @xmath182 are normalized and orthogonal ) , which follows from the results : @xmath183 for @xmath184 , and @xmath185 for @xmath186 . in this limit
the eigenvalues are @xmath187 , a special case of the general analysis given in section [ harmonics ] . for a single circular boundary and finite @xmath39
there are at most two non - zero eigenvalues per @xmath175 , with one for @xmath188 and none for @xmath189 .
the azimuthal case has the nice property that even if @xmath35 and @xmath36 are only available for finite @xmath39 one can project out exactly the cross - contamination on all scales .
the matrix @xmath46 truncated at a finite number of rows ( but retaining an unlimited number of columns ) can be written as @xmath190 where @xmath191 is a column unitary matrix with at most two columns per @xmath175 , and hence the range of @xmath46 can be projected out using @xmath192 so that @xmath193 losing at most two modes per @xmath175 ( per boundary ) .
the matrix @xmath191 can in practice be constructed for each @xmath175 by normalization and orthogonalization ( e.g. using a svd ) of the two vectors @xmath182 and @xmath181 ( which are no longer orthogonal or normalized for finite @xmath39 ) .
this projection is equivalent to the procedure of projecting out the non - zero eigenvalues of @xmath46 in ref .
@xcite , for @xmath194 , where @xmath113 is the number of boundaries .
if @xmath46 is smaller than it s range due to a finite @xmath39 ( i.e. for @xmath195 ) the zero eigenvalues @xmath46 do not correspond to separated modes .
two maps consisting of pure @xmath41 and pure @xmath42 can be constructed simply from @xmath196 if desired a map of the ambiguous modes can also be constructed from the remaining modes .
the separation is computationally trivial in the azimuthal case due to the separability in @xmath175 , and is possible up to the resolution of the experiment ( typically @xmath197 ) . for @xmath113 co - axial circular boundaries one looses up to @xmath198 modes due to the e / b separation , and the separation remains exact and separable in @xmath175 . for non - co - axial boundaries exact
separation is still possible , even though modes with different @xmath175 are mixed due to the rotation ( i.e. with the wigner - d matrices @xmath199 ) .
this is because the coupling matrix @xmath46 for the entire sky made up of a set of non - intersecting cuts can be written as a finite sum of coupling matrices @xmath46 each with finite range , and therefore itself has finite range which can be projected out exactly .
thus it is still possible to perform exact e / b separation at finite @xmath39 with azimuthal cuts containing a finite number of non - intersecting circular cuts around foreground sources . to extract a set of e and b modes for likelihood evaluation
one can proceed following ref .
we briefly review the method here with one minor enhancement .
first we diagonalize @xmath50 ( taking it to be square ) and discard the badly supported modes , to write @xmath200 so @xmath201 then we do the diagonalization @xmath202 and construct the matrix @xmath203 by deleting the columns of @xmath191 corresponding to non - zero diagonal elements of @xmath204 .
it follows from the hermiticity of @xmath46 that @xmath203 is made up of vectors in the left null - space of the mode - mixing matrix @xmath205 .
the pure e and b modes are then given by @xmath206 for isotropic uncorrelated noise these modes also have isotropic and uncorrelated noise .
this method is equivalent to that presented in ref .
@xcite but slightly faster as it replaces the svd of the @xmath207 coupling matrix with a faster diagonalization of a smaller hermitian matrix by using the hermiticity of @xmath46 .
note that as mentioned above one should only include modes with @xmath208 , though in practice for galactic cuts these modes are badly supported anyway .
the coupling matrices for an azimuthal cut can be computed efficiently using the results given in the appendix of ref . | the full sky cosmic microwave background polarization field can be decomposed into ` electric ' ( e ) and ` magnetic ' ( b ) components that are signatures of distinct physical processes .
we give a general construction that achieves separation of e and b modes on arbitrary sections of the sky at the expense of increasing the noise .
when e modes are present on all scales the separation of all of the b signal is no longer possible : there are inevitably ambiguous modes that can not be separated .
we discuss the practicality of performing e / b decomposition on large scales with realistic non - symmetric sky - cuts , and show that separation on large scales is possible by retaining only the well supported modes .
the large scale modes potentially contain a great deal of useful information , and e / b separation at the level of the map is essential for clean detection of b without confusion from cosmic variance due to the e signal .
we give simple matrix manipulations for creating pure e and b maps of the large scale signal for general sky cuts .
we demonstrate that the method works well in a realistic case and give estimates of the performance with data from the planck satellite . in the appendix
we discuss the simple analytic case of an azimuthally symmetric cut , and show that exact e / b separation is possible on an azimuthally symmetric cut with a finite number of non - intersecting circular cuts around foreground sources .
observations of fluctuations in the cosmic microwave background ( cmb ) provide information about primordial inhomogeneities in the universe
. one of the most interesting questions is whether there was a tensor ( gravitational wave ) component .
cmb polarization measurements offer a probe of this signal @xcite via the ` magnetic ' component of the polarization map .
this offers the opportunity to distinguish between different models of single field inflation , which generically predict a significant amplitude of gravitational waves , and other competing models which predict nearly zero amplitudes .
for example the ekpyrotic scenario generically predicts exponentially small amplitudes of tensor modes , as do inflation models where density perturbations originate from fluctuations in a second scalar field .
polarization of the cosmic microwave sky is produced by electron scattering , as photons decouple from the primordial plasma and during reionization .
gravitational waves produce ` magnetic ' ( b ) and ` electric ' ( e ) polarization components at a comparable level by anisotropic redshifting of the energy of photons .
magnetic polarization is not produced by linear scalar ( density ) perturbations , so detection of a magnetic component would provide strong direct evidence for the presence of a primordial gravitational wave ( tensor ) component .
there is a non - linear contribution to the magnetic signal from gravitational lensing of e polarization , though on the scales where the tensor mode signal is large the lensing signal is sufficiently small that it is negligible for observations up to planck sensitivity .
however ultimately the lensing signal ( and how well it can be subtracted ) may provide a limit on the amplitude of tensor modes that can be detected @xcite .
there are also a variety of other possible sources for tensor and vector modes generating a b mode signature @xcite , as well as numerous systematics @xcite .
inflationary models generically predict a gaussian spectrum of linear e and b modes , in which case it is possible to use the full likelihood function to constrain the tensor amplitude without separating into e and b modes . however if there are any departures from gaussianity , e.g. due to systematics , foregrounds , magnetic fields @xcite , defects @xcite or unexpected physics , this could give misleading answers .
so it is useful to have methods to separate out pure b modes for robust detections and isolation of unexpected features . for small
tensor amplitudes any mixture of e and b modes would be dominated by the scalar e signal , and a joint analysis for the tensor contribution would be only marginally more optimal than using only the pure b modes . on small scales
it is possible to provide excellent separation of the e and b mode power spectra using fast quadratic estimators @xcite .
these methods provide estimators which , when averaged over realizations , give zero if there really is no b signal .
however in any given realization , such as the one we observe , there will in general be a non - zero b signal .
this is a potential obstacle to clean e / b power spectrum separation on the large scales where cosmic variance is most important .
cmb observations by wmap have recently indicated a substantial optical depth to reionization @xcite .
if confirmed this would imply that detection of magnetic polarization could be achieved by observation of a relatively small number of high signal to noise b modes on large scales ( corresponding to the horizon size at reionization ) .
the large scales are just where cosmic variance is largest , and the effect of incompleteness of the map ( e.g. due to cuts around the galaxy and foreground sources ) makes the e and b modes harder to separate .
there is therefore clear motivation for methods to cleanly separate the e and b modes on the largest scales , avoiding problems with cosmic - variance mixing that arises from the use of quadratic estimators , and extracting the b modes without assumptions about their distribution . in this paper
we discuss various harmonic methods for performing e / b separation , equivalent to separation at the level of the map .
we start in section [ eb ] with a brief summary of the e / b decomposition . in section [ harmonics ]
we review the tensor harmonics and how the e / b mixing enters for observations over only part of the sky , and discuss the theoretical properties of the coupling matrices . in section [ separation ]
we show how to separate the modes for band limited and non - band limited fields with arbitrary sky cuts , and show that the method presented in earlier work @xcite is optimal for general cuts in the all - scales limit . for non - band limited fields
there are ambiguous modes than can not be separated .
we then show how the large scale reionization magnetic modes can be extracted from a realistic map by retaining only the well supported modes .
the method is computationally tractable and close to exact , and should allow robust detection of the large scale magnetic signal with future cmb polarization observations . in section [ planckex ]
we demonstrate the performance explicitly for a complicated sky cut geometry , and provide comparative estimates of the ability of the planck satellite to detect tensor modes using different methods . in the appendix we review some analytic results for azimuthal cuts from ref .
@xcite , show that exact separation is possible for non - intersecting combinations of azimuthal cuts , and give a slightly improved matrix method for extracting the pure e and b modes exactly . for a method similar to the general method presented here , but working explicitly in pixel space , see ref .
@xcite , and other related work in refs .
@xcite .
though we only discuss cmb polarization maps explicitly , the techniques could be applied to other data , for example shear distributions in weak lensing surveys .
indeed b modes in the lensing of pre - reionization gas may provide a good alternative way to detect tensor modes @xcite . |
the first direct detection of gravitational waves is a key science mission around the world , with many different approaches being advocated .
these include ground and space - based laser interferometers @xcite , and pulsar timing arrays ( collections of galactic millisecond pulsars ( msps ) ; @xcite ) , and it is with the latter of these that we will be concerned with here .
it is the exceptional stability of msps , with decade long observations providing timing measurements that show fractional instabilities similar to atomic clocks ( e.g. ) , that makes them key to the pursuit of a wide range of scientific endeavors .
for example , observations of the pulsar psr b1913 + 16 provided the first indirect detection of gravitational waves @xcite , whilst the double pulsar system psr j0737 - 3039a / b provides precise measurements of several ` post keplerian ' parameters allowing for additional stringent tests of general relativity @xcite .
current theoretical limits ( @xcite ) place the amplitude of a stochastic gravitational wave background ( gwb ) generated by coalescing black holes ( e.g. @xcite ) at only a factor 3 - 10 lower than current observational limits ( e.g. @xcite ) . in order to make the first tentative detections of these signals as much ancillary data will be needed as possible in order to constrain the other components present in the data .
dispersion measure ( dm ) variations are thought to be one of the largest components of noise in pulsar timing data ( e.g. @xcite ) , and many different methods exist to describe it ( e.g. @xcite , henceforth l13 ; lee et al . submitted 2013 ; @xcite ) . in the near future
, observations from lofar will allow precise measurements of dm in the direction of pulsars to be used in pta analysis .
including this information in subsequent analysis in order to constrain the dm signal and separate it from the gravitational waves will thus be critical . in this article
we describe how to include such dm measurements as prior information in pulsar timing analysis in order to constrain the signal realisation for the dm by modifying the existing bayesian techniques presented in l13 . in section [ section : bayes ]
we give a brief overview of our bayesian methodology , and in section [ section : models ] derive the likelihood that we use to include the additional dm measurements when analysing the simulated data in section [ section : sims ] .
finally we will provide some concluding remarks in section [ section : conclusion ] .
this research is the result of the common effort to directly detect gravitational waves using pulsar timing , known as the european pulsar timing array ( epta ) @xcite .
our method for performing pulsar timing analysis is built upon the principles of bayesian inference , which provides a consistent approach to the estimation of a set of parameters @xmath0 in a model or hypothesis @xmath1 given the data , @xmath2 .
bayes theorem states that : @xmath3 where @xmath4 is the posterior probability distribution of the parameters , @xmath5 is the likelihood , @xmath6 is the prior probability distribution , and @xmath7 is the bayesian evidence . in parameter estimation , the normalizing evidence factor
is usually ignored , since it is independent of the parameters @xmath0 .
inferences are therefore obtained by taking samples from the ( unnormalised ) posterior using , for example , standard markov chain monte carlo ( mcmc ) sampling methods .
an alternative to mcmc is the nested sampling approach @xcite , a monte - carlo method targeted at the efficient calculation of the evidence , that also produces posterior inferences as a by - product . in @xcite and @xcite this nested sampling framework
was built upon with the introduction of the multinest algorithm , which provides an efficient means of sampling from posteriors that may contain multiple modes and/or large ( curving ) degeneracies , and also calculates the evidence . since
its release multinest has been used successfully in a wide range of astrophysical problems , from detecting the sunyaev - zeldovich effect in galaxy clusters @xcite , to inferring the properties of a potential stochastic gravitational wave background in pulsar timing array data @xcite . in the following sections we make use of the multinest algorithm to obtain our estimates of the posterior probability distributions for both timing model , and stochastic parameters .
recently temponest ( l13 ) was introduced as a means of performing a simultaneous analysis of either the linear or non - linear timing model and additional stochastic parameters using multinest to efficiently explore this joint parameter space , whilst using tempo2 @xcite as an established means of evaluating the timing model at each point in that space .
we incorporate the likelihood developed in section [ section : models ] into temponest in order to perform the analysis described in section [ section : sims ] .
for any pulsar we can write the toas for the pulses as a sum of both a deterministic and a stochastic component : @xmath8 where @xmath9 represents the @xmath10 toas for a single pulsar , with @xmath11 and @xmath12 the deterministic and stochastic contributions to the total respectively , where any contributions to the latter will be modelled as random gaussian processes . writing the deterministic signal due to the timing model as @xmath13 , and the uncertainty associated with a particular toa as : @xmath14 where @xmath15 and @xmath16 represent the efac and equad parameters applied to toa @xmath17 respectively , we can write the probability that the data is described by the timing model parameters @xmath18 and white noise parameters @xmath15 and @xmath16 as : @xmath19 to include the dm variations , which we will denote @xmath20 , we begin by following the same process as in l13 . writing it in terms of its fourier coefficients
@xmath21 so that @xmath22 where @xmath23 denotes the fourier transform such that for frequency @xmath24 and time @xmath25 we will have both : @xmath26 and an equivalent cosine term . here the dispersion constant @xmath27 is given by : @xmath28 @xmath29 is the total observing timespan , @xmath30 is the observing frequency for the toa at barycentric arrival time @xmath25 , and @xmath24 the frequency of the signal to be sampled . defining the number of coefficients to be sampled by @xmath31
, we can then include the set of frequencies with values @xmath32 , where @xmath10 extends from 1 to @xmath31 . for typical pta data @xcite
show that for frequency independent spin noise , a low frequency cut off of @xmath33 is sufficient to accurately describe the expected long term variations present in the data , as the quadratic included in the timing model in the form of the spindown parameters acts as a proxy to lower frequency signals . for dm variations ,
however , these terms must be accounted for either by explicitly including these low frequencies in the model , or by including a quadratic in dm to act as a proxy , as with the red noise , defined as : @xmath34 with @xmath35 free parameters to be fitted , @xmath36 the barycentric arrival time for toa @xmath17 and @xmath37 elements of a vector : @xmath38 for a single pulsar the covariance matrix @xmath39 of the fourier coefficients @xmath21 will be diagonal , with components @xmath40 where there is no sum over @xmath17 , and the set of coefficients @xmath41 represent the theoretical power spectrum for the dm variations in the residuals . as discussed in @xcite , whilst eq [ eq : bprior ]
states that the fourier modes are orthogonal to one another , this does not mean that we assume they are orthogonal in the time domain where they are sampled , and it can be shown that this non - orthogonality is accounted for within the likelihood . instead , in bayesian terms
, eq . [ eq : bprior ] represents our prior knowledge of the power spectrum coefficients within the data .
we are therefore stating that , whilst we do not know the form the power spectrum will take , we know that the underlying fourier modes are still orthogonal by definition , regardless of how they are sampled in the time domain .
it is here then that , should one wish to fit a specific model to the power spectrum coefficients at the point of sampling , such as a broken , or single power law , the set of coefficients @xmath41 should be given by some function @xmath42 , where we sample from the parameters @xmath0 from which the power spectrum coefficients @xmath41 can then be derived .
we can then write the joint probability density of the timing model , white noise parameters , power spectrum coefficients and the signal realisation , pr@xmath43 , as : @xmath44 where @xmath45 represents any additional prior information regarding the dm signal realisation . henceforth we will consider @xmath45 to be given by a vector of measurements of the dm at some set of arbitrary times with associated measurement errors , which we will denote @xmath46 and @xmath47 respectively .
for our choice of @xmath48 we use an uninformative prior that is uniform in @xmath49 space , and draw our samples from the parameter @xmath50 instead of @xmath51 . given this choice of prior the conditional distributions that make up eq . [ eq : prob ]
can be written : @xmath52 \nonumber\end{aligned}\ ] ] where @xmath53 and represents the white noise errors in the residuals , and @xmath54 describes components of the dm model in addition to those contained in the fourier modes , such as the quadratic terms in the timing model which we have separated from the @xmath13 term for clarity .
we then also have : @xmath55 \\ \nonumber & \times & \frac{1}{\sqrt{\mathrm{det}\bmath{\psi}}}\exp\left[-\frac{1}{2}(\bmath{l_{dm}}-\bmath{f_{dm}}-\bmath{f_la})^t\bmath{\psi}^{-1}\right.\nonumber \\ & \times & \left.(\bmath{l_{dm}}-\bmath{f_{dm}}-\bmath{f_la})\right].\end{aligned}\ ] ] here @xmath56 is a matrix of fourier modes as in eq .
[ eq : fmatrix ] , however with points evaluated at the times that additional dm measurements were made , and @xmath57 is the diagonal noise matrix for the additional dm measurements with values @xmath47 .
we then marginalise over all fourier coefficients @xmath21 analytically in order to find the posterior for the remaining parameters alone . in order to perform the marginalisation over the fourier coefficients
@xmath21 , we first write the log of the likelihood in eq [ eq : prob ] , which , excluding the determinant terms , and denoting @xmath58 as @xmath59 , @xmath60 as @xmath61 , @xmath62 as @xmath63 and @xmath64 as @xmath65 is given by : @xmath66 taking the derivitive of @xmath67 with respect to @xmath21 gives us : @xmath68 which can be solved to give us the maximum likelihood vector of coefficients @xmath69 : @xmath70 re - expressing eq .
[ eq : logl ] in terms of @xmath69 : @xmath71 the 3rd term in this expression can then be integrated with respect to the @xmath72 elements in @xmath21 to give : @xmath73 \nonumber \\ & = & ( 2\pi)^m~\mathrm{det } ~ \bmath{\sigma}^{-\frac{1}{2}}.\end{aligned}\ ] ] our marginalised probability distribution for a set of gwb coefficients is then given as : @xmath74 , \nonumber\end{aligned}\ ] ] in much the same way in which we include an efac in the white noise matrix to alter the weighting of the data , we can include an efac in the matrix @xmath57 in order to alter the weight of the dm measurements in the event that the errors provided are under , or over estimated , or if the measurements are not consistent with the timing data .
as such we can simply define : @xmath75 in the simulations described in section [ section : sims ] however , we will not include this additional parameter .
we can include additional , frequency independent red ` spin ' noise in much the same way as the dm variations .
as before we define a matrix of fourier modes for a set of @xmath10 frequencies : @xmath76 and an equivalent cosine term .
these rows can then be appended to the fourier matrix in eq.[eq : fmatrix ] , which we will denote here @xmath77 to form a new matrix containing both the red noise and dm terms : @xmath78 similarly the matrix @xmath39 is extended to accommodate the new power spectrum coefficients required to describe the spin noise . the additional dm prior term is then kept the same : in forming the matrix @xmath79 we add the term @xmath80 to only the section of the matrix @xmath81 that corresponds to the autocorrelated terms of the dm modes , and similarly the vector @xmath82 is added only to the part of @xmath83 concerned with the dm fourier modes when forming @xmath65 . in many pulsar timing datasets , phase jumps
are fitted between different groups of observations , or there might be other parameters that are not of interest , such as a constant phase offset , resulting in a potentially significant increase in the number of parameters to be fit for in analysis .
if the specific values of such parameters are not of importance we can marginalise analytically over them , greatly reducing the dimensionality of the problem .
if we separate the timing model into a contribution from the set of parameters that we wish to parameterise @xmath13 and a contribution from the set of @xmath72 parameters that we plan to marginalise over analytically @xmath84 then we can write the probability that the data @xmath85 is described by the remaining parameters @xmath18 and any additional parameters @xmath86 we wish to include as : @xmath87 using a uniform prior on the @xmath72 @xmath88 parameters , we use the same approach as described in @xcite to perform this marginalisation process analytically . this results in equation [ eq : llikemargin ] , which is the expression we will be using in the subsequent analysis : @xmath89 , \end{aligned}\ ] ] where @xmath90 , @xmath91 and @xmath92 .
[ cols="<,^,^,^",options="header " , ] [ table : sim1 ] in order to check the efficacy of the method described in section [ section : models ] we simulate a dataset for the isolated pulsar j0030 + 0451 .
we begin by taking the injected parameter values given in table [ table : sim1 ] and generate a 5 year dataset with uneven sampling in the time domain , but with an average cadence of @xmath93 2 weeks .
we include two observing frequencies at 1440 and 2440 mhz where the higher frequency observations exist only for the latter @xmath94 of the observations , and where no multi - frequency data exists for any given observing epoch of duration @xmath93 2 weeks .
we then add variations in the dm that are described by a power law with functional form @xmath95 with a spectral index of @xmath96 .
note that we do not list injected values for the @xmath97 and @xmath98 parameters as these are used simply as proxies to the low frequency dm variations , and as such we do not know a priori what these values will take .
we then simulate discrete observations of the total dm signal that will act as our prior , @xmath46 .
we generate monthly samples that are scattered around the true signal with an rms of @xmath99@xmath100 pc .
the injected dm signal , simulated dm observations , and the final residuals obtained when subtracting the injected timing model parameters in table [ table : sim1 ] except the dm parameters , are shown in fig .
[ fig : res ] .
in addition to the timing model and dm parameters , we also include a red noise power law model in our analysis , of the same functional form as the dm spectrum .
we initially performed our analysis using a log uniform prior on the red noise amplitude , however as can be seen in fig .
[ fig : logred ] the signal is completely unconstrained below some upper limit . when using a log uniform prior this upper limit
will be dependent upon the lower limit chosen for the prior , and as such we instead use a uniform prior on the red noise amplitude , in order to obtain a robust upper limit on the signal . in fig .
[ figure : timingposteriors ] we show the one dimensional posteriors for the timing model and stochastic parameters given in table [ table : sim1 ] when including ( blue ) , and not including ( red ) , the simulated dm observations as additional prior information . for the timing model parameters the injected value , when known , is given by 0 on the @xmath101 axis , which is in units of the 1@xmath102 uncertainty returned by tempo2 when not including either the red noise or dm power law model components .
the clear result here is that the precision with which the timing model parameters have been recovered has improved significantly when including the additional prior information , between a factor @xmath103 .
comparing the posteriors for the amplitude of the red noise power law , when including the additional prior information the upper limit decreases by a factor of @xmath104 , demonstrating how critical such data will be in constraining gravitational wave signals in pulsar timing data .
@xmath105{signal.pdf } & % \includegraphics[width=80mm]{res.pdf } \\
\end{array}$ ] @xmath105{logredamp.pdf } & % \includegraphics[width=80mm]{logredindex.pdf } \\
\end{array}$ ] @xmath106{ra.pdf } & \includegraphics[width=50mm]{dec.pdf } & \includegraphics[width=50mm]{f0.pdf } \\ % % \hspace{-0.5 cm } \includegraphics[width=50mm]{f1.pdf } & \includegraphics[width=50mm]{pmra.pdf } & \includegraphics[width=50mm]{pmdec.pdf } \\ % % \hspace{-0.5 cm } \includegraphics[width=50mm]{dm.pdf } & \includegraphics[width=50mm]{dm1.pdf } & \includegraphics[width=50mm]{dm2.pdf } \\ % % \hspace{-0.5 cm } \includegraphics[width=50mm]{px.pdf } & \includegraphics[width=50mm]{redamp.pdf } & \includegraphics[width=50mm]{redspec.pdf } \\ % % \hspace{-0.5 cm } \includegraphics[width=50mm]{dmamp.pdf } & \includegraphics[width=50mm]{dmindex.pdf } & \\ % \end{array}$ ]
we have discussed a method of including discrete measurements of the dispersion measure in the direction of a pulsar as prior information in the analysis of that pulsar . by using an existing bayesian framework ,
this prior information can be simply folded into the analysis and used to constrain the dm signal realisation in pulsar timing data .
we have shown that this method can be applied where no multi - frequency data exists across much of the dataset , and does not require simultaneous multi - frequency data to be present for any observing epoch .
we have shown that , as expected , including this prior information can greatly increase both the precision of the timing model parameters recovered from the analysis , as well as increase the sensitivity to red noise in the data .
clearly the level of improvement in real data will be entirely dependent on the dataset in question , however the inclusion of such prior information will likely prove extremely useful both to test the validity of existing models for dm , and to improve constraints in future analysis .
many thanks to jason hessels for assistance in the production of this work . | here we present a bayesian method of including discrete measurements of dispersion measure due to the interstellar medium in the direction of a pulsar as prior information in the analysis of that pulsar .
we use a simple simulation to show the efficacy of this method , where the inclusion of the additional measurements results in both a significant increase in the precision with which the timing model parameters can be obtained , and an improved upper limit on the amplitude of any red noise in the dataset .
we show that this method can be applied where no multi - frequency data exists across much of the dataset , and where there is no simultaneous multi - frequency data for any given observing epoch .
including such information in the analysis of upcoming international pulsar timing array ( ipta ) and european pulsar timing array ( epta ) data releases could therefore prove invaluable in obtaining the most constraining limits on gravitational wave signals within those datasets .
[ firstpage ] methods : data analysis , pulsars : general , pulsars : individual |
it is well known that the burgers equation in one dimension could be linearized by the cole - hopf transformation in terms of the linear heat equation .
it allows one to solve the initial value problem for the burgers equation and to get exact solutions in the form of shock solitons and describe their scattering . in the present paper
we study the q - differential burgers type equation with quadratic dispersion and the cubic nonlinearity , and find its linearization in terms of the q - heat equation . in terms of the jackson s q - exponential function
we introduce the q - hermite and q - kampe - de feriet polynomials , representing moving poles solution for the q - burgers equation .
then we derive the operator solution of the initial value problem for the q - burgers equation in terms of the ivp for the q - heat equation .
we find solutions of our q - burgers type equation in the form of singular and regular q - shock solitons .
it turns out that static q - shock soliton solution shows remarkable self - similarity property in space coordinate @xmath1 .
the q - number corresponding to the ordinary number @xmath2 is defined as , @xcite , @xmath3where @xmath4 is a parameter , so that @xmath2 is the limit of @xmath5_q$ ] as @xmath6 .
a few examples of q - numbers are given here : @xmath7_q=0 $ ] , @xmath8_q=1 $ ] , @xmath9_q=1+q$ ] , @xmath10_q=1+q+q^2 $ ] . in terms of these q - numbers ,
the jackson q - exponential function is defined as @xmath11_q!}.\]]for @xmath12 it is entire function of @xmath1 and when @xmath6 it reduces to the standard exponential function @xmath13 . the q - exponential function can also be expressed in terms of infinite product @xmath14 when @xmath15 and @xmath16 when @xmath12 .
thus , the q - exponential function for @xmath15 has infinite set of poles at @xmath17 and for @xmath12 the infinite set of zeros at @xmath18 the q - derivative is defined as @xmath19 and when @xmath20 it reduces to the standard derivative @xmath21 . using the definition of the @xmath4-derivative one can easily see that @xmath22_qx^{n-1},\\ & & d^x_qe_q(ax)=ae_q(ax).\end{aligned}\ ] ]
we define the q - hermite polynomials according to the generating function @xmath23_q t x ) = \sum^{\infty}_{n=0 } h_n(x;q ) \frac{t^n}{[n]_q!}.\label{genfunc}\ ] ] from this generating function we have the special values @xmath24_q!}{[n]_q!},\\ & & h_{2n+1}(0;q ) = 0 , \end{aligned}\]]where @xmath5_q!=[1]_q[2]_q ... [n]_q$ ] , and the parity relation @xmath25 by q - differentiating the generating function ( [ genfunc ] ) according to @xmath1 and @xmath26 we have the recurrence relations correspondingly @xmath27_q [ n]_q h_{n-1}(x;q),\label{rec1}\ ] ] @xmath28_q \,x h_{n}(x;q ) - [ n]_q \,h_{n-1}(q x;q)\nonumber \\ - [ n]_q \,q^{\frac{n+1}{2 } } h_{n-1}(\sqrt{q } x;q).\label{rec2}\ ] ] using operator @xmath29 so that @xmath30 relation ( [ rec2 ] ) can be rewritten as @xmath28_q \,x h_{n}(x;q ) \nonumber\\ - [ n]_q ( m_q + q^{\frac{n+1}{2 } } m_{\sqrt{q } } ) h_{n-1}(x;q).\label{rec3}\ ] ] substituting ( [ rec1 ] ) to ( [ rec3 ] ) we get @xmath31_q \,x - \frac{m_q + q^{\frac{n+1}{2 } } m_{\sqrt{q}}}{[2]_q } d_x\right ) h_{n}(x;q)\label{rec4}\ ] ] by the recursion , starting from @xmath32 and @xmath33 we have next representation for the q - hermite polynomials @xmath34_q \,x - \frac{m_q + q^{\frac{k}{2 } } m_{\sqrt{q}}}{[2]_q } d_x\right)\cdot 1\label{rec5}\ ] ] we notice that the generating function and the form of our q - hermite polynomials are different from the known ones in the literature , @xcite , @xcite , @xcite , @xcite . + in the above expression the operator @xmath35\ ] ] is expressible in terms of the q - spherical means as @xmath36 f(x ) = \frac{1}{2}(f(q x ) + f(\frac{1}{q}x)).\ ] ] using definition , @xcite , @xmath37 which now we apply for operators , we should distinguish the direction of multiplication .
we consider two cases @xmath38 and @xmath39 then , we can rewrite ( [ rec5 ] ) shortly as @xmath40_q \,x - \frac { m_q \,d_x}{[2]_q } ) - q^{\frac{1}{2 } } \frac{m_{\sqrt{q}}\ , d_x}{[2]_q } \right)^n_{\sqrt{q}\ , > } \cdot1 \label{rec6}\nonumber \end{aligned}\ ] ] first few polynomials are @xmath41@xmath42_q \,x \nonumber,\end{aligned}\ ] ] @xmath43 ^ 2_q \ , x^2 - [ 2]_q\nonumber,\end{aligned}\ ] ] @xmath44 ^ 3_q \,x^3 - [ 2]^2_q [ 3]_q \,x\nonumber , \end{aligned}\ ] ] @xmath45 ^ 4_q \,x^4 - [ 2]^2_q [ 3]_q [ 4]_q\,x^2 + [ 2]_q [ 3]_q [ 2]_{q^2 } \nonumber .\end{aligned}\ ] ] when @xmath6 these polynomials reduce to the standard hermite polynomials . applying @xmath46 to both sides of ( [ rec4 ] ) and using recurrence formula ( [ rec1 ] )
we get q - differential equation for q - hermite polynomials @xmath47_q } d_x ( m_q + q^{\frac{n+1}{2 } } m_{\sqrt{q } } ) d_x h_n ( x;q ) \nonumber - [ 2]_q q x d_x h_n(x;q ) + [ 2]_q [ n]_q q h_n(x;q ) = 0 \nonumber .\end{aligned}\ ] ] we have next identity @xmath48_q } d^2_x \right ) e_q ( [ 2]_q x t ) = e_q ( -t^2 ) e_q ( [ 2]_q x t)\label{prop1}.\ ] ] by q- differentiating the q - exponential function in @xmath1 @xmath49_q x t ) = ( [ 2]t)^n e_q ( [ 2]_q x t),\ ] ] and combining then to the sum @xmath50_q ! }
d^{2n}_x e_q ( [ 2]_q x t ) = \sum_{n=0}^{\infty } \frac{[2]_q^n a^n t^{2n}}{[n]_q!}\,\ , e_q ( [ 2]_q x t),\ ] ] we have relation @xmath51_q x t ) = e_q ( [ 2]_q a t^2 ) e_q ( [ 2]_q x t).\ ] ] by choosing @xmath52_q$ ] we get @xmath48_q } d^2_x \right ) e_q ( [ 2]_q x t ) = e_q ( -t^2 ) e_q ( [ 2]_q x t).\ ] ] the next identity is valid @xmath53^n e_q \left(-\frac{1}{[2]_q } d^2_x \right ) x^n \label{prop2}.\ ] ] the right hand side of ( [ prop1 ] ) is the generating function for q - hermite polynomials . hence expanding both sides in @xmath26 we get the result .
@xmath54_q } \right ) x^{n+1}= \nonumber \\
\frac{1}{[2]_q}\left([2]_q \,x - \frac{(m_q + q^{\frac{n+1}{2 } } m_{\sqrt{q } } ) d_x}{[2]_q } \right ) e_q \left(-\frac{d^2_x}{[2]_q } \right ) x^{n}.\nonumber \label{prop3}\ ] ] we use ( [ prop2 ] ) and relation ( [ rec4 ] ) .
if function @xmath55 is analytic and expandable to power series @xmath56 then we have next q - hermite series @xmath48_q } d^2_x
\right ) f(x ) = \sum^\infty_{n=0 } a_n \frac{h_n(x;q)}{[2]_q^n}.\ ] ]
we define q - kampe - de feriet polynomials as @xmath57_q \sqrt{-\nu t}}\right ) , \ ] ] and from ( [ rec4 ] ) we have the next recursion formula @xmath58 by recursion it gives @xmath59 or by notation ( [ l ] ) @xmath60 first few polynomials are @xmath61 @xmath62 @xmath63_q \ , \nu t\nonumber,\end{aligned}\ ] ] @xmath64_q [ 3]_q \,\nu t \,x\nonumber , \end{aligned}\ ] ] @xmath65_q [ 4]_q\,\nu t \,x^2 + [ 2]_q [ 3]_q [ 2]_{q^2 } \nu^2 t^2\nonumber.\end{aligned}\ ] ]
we consider the q - heat equation @xmath66 solution of this equation expanded in terms of parameter @xmath67 @xmath68 ! } h_n(x , t;q),\ ] ] gives the set of q - kampe - de feriet polynomial solutions for the equation .
then we find time evolution of zeroes for these solutions in terms of zeroes @xmath69 of q - hermite polynomials , @xmath70 so that @xmath71 z_k(n , q ) \sqrt{-\nu t}.\label{zerosmotion}\ ] ] for n=2 we have two zeros determined by q - numbers , @xmath72_q } } \sqrt{-\nu t}\\,\ ] ] @xmath73_q } } \sqrt{-\nu t}\\,\ ] ] and moving in opposite directions according to ( [ zerosmotion ] ) .
for n=3 we have zeros determined by q - numbers , @xmath74_q!}}\sqrt{-\nu t } , \\
x_2(t ) & = & 0 , \\
x_3(t ) & = & \sqrt{{[3]_q!}}\sqrt{-\nu t } , \end{aligned}\ ] ] two of which are moving in opposite direction according to ( [ zerosmotion ] ) and one is in the rest .
following similar calculations as in proposition i we have next relation @xmath75 the right hand side of this expression is the plane wave type solution of the q - heat equation @xmath66 expanding both sides in power series in @xmath67 we get q - kampe de ferie polynomial solutions of this equation @xmath76 consider an arbitrary analytic function @xmath56 , then function @xmath77 is a time dependent solution of the q - heat equation ( [ qheat ] ) .
according to this we have the evolution operator for the q - heat equation as @xmath78 it allows us to solve the initial value problem @xmath79 in the form @xmath80
let us consider the q - cole - hopf transformation @xmath81 then @xmath82 satisfies the q - burgers type equation with quadratic dispersion and cubic nonlinearity @xmath83-\\ \\\frac{1}{2}\left[d_x\left(u(qx , t)u(x , t)\right)\right]+ \frac{1}{4 } \left[u(q^2x , t)-u(x , qt)\right]u(qx , t)u(x , t).\end{array}\ ] ] when @xmath0 it reduces to the standars burgers equation . substituting the operator solution ( [ ivp ] ) to ( [ c h ] ) we find operator solution for the q - burgers type equation in the form @xmath84this solution corresponds to the initial function @xmath85
thus , for arbitrary initial value problem for the q - burgers equation with @xmath86 we need to solve the initial value problem for the q - heat equation with initial function @xmath55 satisfying the first order q - differential equation @xmath87
as a solution of q - heat equation we choose first @xmath88then we find solution of the q - burgers equation as a constant @xmath89 if we choose @xmath90 then we have the q - shock soliton solution @xmath91 due to zeroes of the q - exponential function this expression admits singularities for some values of parameters @xmath92 and @xmath93 .
in fig.1 we plot the singular q - shock soliton for @xmath94 and @xmath95 at time @xmath96 .
however for some specific values of the parameters we found the regular q - shock soliton solution .
we introduce the q - hyperbolic function @xmath97 or @xmath98 then by using infinite product representation for q - exponential function we have @xmath99 from ( [ polesq]),([zerosq ] ) we find that zeroes of the first product are located on negative axis @xmath1 , while for the second product on the positive axis @xmath1
. therefore the function has no zeros for real @xmath1 and @xmath100 .
if @xmath94 , and @xmath101 , the time dependent factors in nominator and the denominator of ( [ qshock ] ) cancel each other and we have the stationary shock soliton @xmath102 due to above consideration this function has no singularity on real axis and we have regular q - shock soliton .
in fig.2 , fig.3 and fig.4 we plot the regular q - shock soliton for @xmath94 and @xmath101 at different ranges of @xmath1 .
it is remarkable fact that the structure of our shock soliton shows self - similarity property in space coordinate @xmath1 .
indeed at the ranges of parameter @xmath103 the structure of shock looks almost the same . for the set of arbitrary numbers
@xmath104 @xmath105 we have multi - shock solution in the form @xmath106 in general this solution admits several singularities . to have regular multi - shock solution we can consider the even number of terms @xmath107 with opposite wave numbers
when @xmath108 and @xmath94 , @xmath101,@xmath109,@xmath110 we have q - multi - shock soliton solution , @xmath111 in fig .
5 we plot @xmath108 case with values of the wave numbers @xmath94 , @xmath101,@xmath109,@xmath110 at @xmath96 . to have regular solution for any time @xmath26 and given base @xmath4
, we should choose proper numbers @xmath112 which are not in the form of power of @xmath4 .
this question is under the study now .
one of the authors ( sn ) was partially supported by national scholarship of the scientific and technological research council of turkey ( tubitak ) .
this work was supported partially by izmir institute of technology , turkey .
j. cigler and j. zeng , two curious @xmath4-analogues of hermite polynomials _ arxiv:0905.0228 _ , 2009 .
j. negro , the factorization method and hierarchies of q - oscillator hamiltonians centre de recherches mathematiques crm proceedings and lecture notes , volume 9 , 239 , 1996 . | by using jackson s q - exponential function we introduce the generating function , the recursive formulas and the second order q - differential equation for the q - hermite polynomials .
this allows us to solve the q - heat equation in terms of q - kampe de feriet polynomials with arbitrary n moving zeroes , and to find operator solution for the initial value problem for the q - heat equation . by the q - analog of the cole - hopf transformation we construct the q - burgers type nonlinear heat equation with quadratic dispersion and the cubic nonlinearity . in @xmath0
limit it reduces to the standard burgers equation .
exact solutions for the q - burgers equation in the form of moving poles , singular and regular q - shock soliton solutions are found . |
online auctions are becoming an increasingly important component of consumers shopping experience .
on ebay , for instance , several million items are offered for sale every day .
what makes online auctions popular forms of commerce is their availability of almost any kind of item , whether it be new or used , and their constant accessibility at any time of the day , from any geographical region in the world .
moreover , the auction mechanism often engages participants in a competitive environment and can result in advantages for both the buyers and the sellers [ @xcite ] . in this paper
we study online auctions from the point of view of the bidder seller network that they induce .
every time a bidder purchases from a seller , both bidder and seller are linked .
buying from a seller indicates that the bidder likes the product and trusts the seller thus , it establishes a relationship between bidder and seller .
many sellers list more than one auction ( i.e. , they sell multiple items across different auctions ) , so repeat transactions by the same bidder across different auctions of the same seller measure the strength of this relationship , that is , it measures the strength of a bidder s _ loyalty _ to a particular seller . studying loyalty in auction networks
much of the existing auction literature focuses on only the seller and the level of _ trust _ she signals to the bidders [ e.g. , @xcite ] . to that end , a seller s _ feedback score _
( i.e. , the number of positive ratings minus the number of negative ratings ) is often scrutinized [ e.g. , @xcite ] and it has been shown that higher feedback scores can lead to price - premiums for the seller [ see @xcite ; @xcite ] . in this paper we study a complementary determinant of a bidder s decision process : _
loyalty_. loyalty is different from trust .
trust is often associated with reliability or honesty ; and trust may be a necessary ( but not sufficient ) prerequisite for loyalty .
loyalty , however , is a stronger determinant of a bidder s decision process than trust .
loyalty refers to a state of being faithful or committed .
loyalty incorporates not only the level of confidence in the outcome of the transaction , but also satisfaction with the product , the price , and also with previous transactions by the same seller . moreover , loyal bidders are often willing to make an emotional investment or even a small sacrifice to strengthen a relationship .
this paper makes two contributions to the literature on online auctions : first , we propose a novel way to _ measure _ e - loyalty from the bipartite network of bidders and sellers ; then , we investigate the _ effect _ of e - loyalty on the outcome of an auction and the statistical challenges associated with it . more specifically , our goal is to understand and learn from loyalty networks . to that end , we first measure a seller s perceived loyalty by its induced bidder loyalty distribution .
then , borrowing ideas from _ functional data analysis _ , we capture key elements of that distribution using functional principal component analysis .
the resulting principal component scores capture different aspects of loyalty - strength , -skew and -variability .
we then investigate the impact of these loyalty scores on the outcome of an auction such as its final price .
we would like to point out that the goal of this paper is not to develop new auction theory ( i.e. , it is not our goal to develop a game - theoretic model under market equilibrium considerations ) .
rather , our goal is to mine a rich set of auction data for new patterns and knowledge . in that sense ,
our work is exploratory rather than confirmatory .
however , as it is often the case with exploratory work , we hope that our work will also inspire the development of new theory .
in particular , we hope that our work will bring the attention to the many statistical challenges associated with the study of online markets .
studying e - loyalty networks is challenging from a statistical point of view because of the asymmetric nature of the network .
just as in many offline markets , online auctions are dominated by few very large sellers ( `` megasellers '' ) .
megasellers have a large supply of products and thus account for a large number of all the transactions .
statistically , this dominance results in a clustering of the data and , as a result , a violation of standard ols model assumptions . in this paper
we investigate several remedies to this clustering effect via random effects models and weighted least squares .
however , our investigation shows that neither approach fully eases all problems .
we thus conclude that the data is too segmented to be captured by a single model and compare our analyses with the results of a data - clustering approach .
this paper has implications for future research in online markets .
many online markets are characterized by a few large `` players '' that dominate most of the interactions and many , many small players with occasional interactions .
this is often referred to as the `` long tail effect '' in online markets [ see , e.g. , @xcite ] . for instance , on ebay , megasellers dominate the marketplace .
the statistical implication is that repeat interactions by these megasellers are no longer independent and , hence , the assumptions of ols break down .
while this may not always create a problem , this research shows that , first , the conclusions from an ols regression are significantly different from models that account for the clustering induced by megasellers , and , second , that it is not at all obvious how to best account for this clustering . in particular , this research puts the spotlight on the findings from previous researchers [ e.g. , @xcite ; @xcite ; @xcite ] who , despite similar data - scenarios , rely their conclusions on the ols modeling assumptions .
[ see also @xcite who , in the context of trust and online auctions alone , count over 6 papers relying on ols modeling techniques . ]
this paper is organized as follows . in section [ sec2 ]
we introduce our data and we motivate the existence of auction networks . in section [ sec3 ]
we use seller bidder networks to derive several key measures of e - loyalty .
we investigate the effect of e - loyalty on the outcome of an auction in section [ sec4 ] and explore different modeling alternatives .
the paper concludes with final remarks in section [ sec5 ] .
in this section we describe the data for our study .
we start by describing the online auction mechanism in general and our data in particular .
then , we motivate the network structure induced by the auction mechanism and show several snapshots of our network data . in online auctions participants bid for products or services over the internet . while there are different types of auction mechanisms , one of the most popular types ( a variant of which
can also be found on http://ebay.com[ebay.com ] ) is the _ vickrey _ auction , in which the initial price starts low and is bid up successively .
online auctions have experienced a tremendous popularity recently , which can be attributed to several features : since the auction happens online , it is not bound by any temporal or geographical constraints , in stark contrast to its brick - and - mortar counterpart ( e.g. , at _
sotheby s _ ) .
it also fosters social interactions since it engages participants in a competition . as a result
, it attracts a large number of buyers and sellers which offers advantages for both sides : sellers find a large number of potential customers which often results in higher prices and lower costs . on the other hand
, buyers find a large variety of products which enables them to locate rare products and to choose between products with the lowest price .
one of the most well - known online auctions is ebay , but there are many more ( e.g. , ubid , prosper , swoopo or overstock ) , each of which offers a variety of different products and services .
the data for this research originates from ebay online auctions and we describe details of the data next .
@ld3.2d2.2d3.2@ * attribute * & & & + auction duration ( days ) & 3.50 & 3.00 & 4.43 + starting price ( usd ) & 3.77 & 3.33 & 5.64 + closing price ( usd ) & 6.61 & 4.25 & 9.15 + item quantity & 5.42 & 1.00 & 129.47 + bid count & 3.16 & 1.00 & 4.26 + size ( bead diameter ) & 6.41 & 6.00 & 3.35 + pieces ( # of beads per item ) & 124.30 & 48.00 & 343.79 + [email protected]@ * attribute * & & & + volume & 163.90 & 6.00 & 999.00 + conversion rate & 0.67 & 0.67 & 0.33 + seller feedback & 2054.00 & 264.00 & 12400.00 + we study the complete bidding records of _ swarovski _ fine beads for every single auction that was listed on ebay between april , 2007 , and january , 2008 .
( note that the data were obtained directly from ebay so we have a complete set of bidding records for that time frame . )
our data contains a total of 36,728 auctions out of which 25,314 transacted .
there are 365 unique sellers and 40,084 bidders out of which 19,462 made more than a single purchase .
each bidding record contains information on the auction format , the seller , the bidder , as well as on product details .
tables [ tab : tab 1][tab : tab 3 ] summarize this information .
table [ tab : tab 1 ] shows information about the auctions and the product sold in each auction .
we can see that the typical auction - length is 3 days .
the product sold in each auction ( packages of beads for crafts and artisanship ) is of relatively small value and , thus , both the average starting and closing prices are low .
while many ebay auctions sell only one item at a time ( e.g. , laptop or automobile auctions ) , auctions in the crafts category often feature multi - unit auctions , that is , the seller offers multiple counts of the same item and bidders can decide how many of these items they wish to purchase . in our data
the average item - quantity per auction is 5.42 .
auctions thrive under competition among bidders and while the average number of bids is slightly larger than 3 , the median is only 1 . as pointed out above , the items sold in these auctions are packages of swarovski beads .
the value of a bead is , in part , defined by its size , and the average diameter of our beads equals 6.41 millimeters .
another measure for the value of an item is the number of pieces per package ; we can see that there are on average over 124 beads in each package , but this number varies significantly from auction to auction .
[email protected]@ * attribute * & & & + volume & 3.62 & 1.00 & 14.29 + item quantity & 5.05 & 2.00 & 29.25 + bidder feedback & 228.10 & 70.00 & 559.53 + we are primarily interested in the bipartite network between bidders and sellers .
one main factor influencing this network is the size of the seller .
we can see ( table [ tab : tab 2 ] ) that the average seller - volume ( i.e. , number of auctions per seller ) is over 163 .
a seller s auction will only transact if ( at least one ) bidder places a bid . while low transaction rates ( or `` conversion rates '' ) are a problem for many ebay categories ( e.g. , automobiles ) , in our data
the average conversion rate is 67% per seller , which is considerably high .
one factor driving conversion rates is a seller s perceived level of trust .
trust is often measured using a seller s feedback rating computed as the sum of positive ( `` @xmath0 '' ) and negative ( `` @xmath1 '' ) ratings .
trust averages over 2000 in our data .
table [ tab : tab 3 ] shows the corresponding attributes of the bidders .
bidders win on average almost 4 auctions ( `` volume '' ) and , in every auction , they purchase on average over 5 items .
( recall the multi - unit auctions with several items per listing . )
the bidder feedback [ computed as the sum of positive ( `` @xmath0 '' ) and negative ( `` @xmath1 '' ) feedback ] captures a bidder s experience with the auction - process and its average is over 220 in our data , signaling highly experienced bidders .
interactions in an online auction result in a network linking its participants . bidders bidding on one auction
are linked to other bidders who bid on the same auction .
sellers selling a certain product are linked to other sellers selling the same product . in this study
we focus on the network _ between _ buyers and sellers . each time a bidder transacts with a particular seller ,
both are linked .
a seller can set up more than one auction , thus , repeat transactions ( i.e. , purchases ) measure the strength of this link .
for instance , a bidder transacting 10 times with the same seller has a stronger link compared to a bidder who transacts only twice . in our analyses ,
we only consider edges with link - strength of at least 4 .
that is , we disregard all bidder seller transactions with frequency less than 4 .
while there exists no recommended or ideal cut - off , our investigations suggest that results vary for smaller values but stabilize for link - strengths of 4 and higher . in that sense
, the network strength measures an important aspect about the relationship between buyers and sellers : customer loyalty .
we would like to emphasize that one can measure loyalty in different ways .
while one could count all the repeat _ bids _ a bidder places on auctions hosted by the same seller , we only count the number of _ winning bids _
( i.e. , the number of transactions ) . while both bids and winning bids indicate a relationship between buyers and sellers , a winning bid signals a much stronger commitment and is thus much more indicative of a buyer s loyalty . in this paper
we investigate loyalty relationships across auctions . studying cross - auction relationships
is rather rare in the literature on online auctions , and it has gained momentum only recently [ @xcite ; @xcite ; @xcite ; @xcite ] . in this work we consider network effects between auction participants and their impact on the outcome of an auction .
consider figure [ fig : net 2 ] which shows the top 10% of high volume _
sellers_. sellers are marked by white triangles , bidders are marked by red squares .
a ( black ) line between a seller and bidder denotes a transaction .
the width of the line is proportional to the number of transactions and hence measures the strength of a link .
we can see that some sellers interact with several hundred different bidders ( with 895 , on average ) ; we can also see that some sellers are `` exclusive '' in the sense that they are the only ones that transact with a set of bidders ( see , e.g. , at the margins of the network ) , while other sellers `` share '' are a common set of bidders .
serving bidders exclusively vs. sharing them with other sellers has huge implications on the outcome of the auction .
figure [ fig : net 4 ] shows another subset of the data . in this figure
we display only the top 10% of all _ bidders _ with the highest number of transactions .
we can see that many of these high - volume bidders transact with only one seller ( note the many the red triangles which are connected with only a single arc to the network ) and are hence very loyal to the same person .
figure [ fig : net 3 ] shows only new buyers ( i.e. , bidders who won an auction for the first time ) . this network exemplifies the market share of a seller with the effect of repeat buyers removed .
we can see that the market is dominated by few mega sellers , yet smaller sellers still attract some of the buyers .
we can identify 5 mega - sellers , 3 high - volume sellers , and many medium- and low - volume sellers .
since these are only first - time buyers , loyalty does not yet play a role in bidders decisions .
however , the fact that most first - time bidders `` converge '' to only a few mega - sellers suggests that this is a very difficult market for low - volume sellers to enter .
as pointed out above , bidder seller networks capture loyalty of participants . while most sellers and bidders are linked to one another , here we only focus on the sub - graphs created by each bidder seller pair .
next , we describe an innovative way to extract _ loyalty measures _ from these graphs .
our loyalty measures map the entire network of bidders and sellers into a few seller - specific numbers . for each seller ,
these numbers capture both the _ proportion of bidders _ loyal to that seller , as well as the _ degree of loyalty _ of each bidder .
we derive the measure in two steps .
first , we derive , for each seller , the _ loyalty distribution _ ; then , we summarize that distribution in a few numbers using functional principal component analysis . we describe each step in detail below .
note that there exists more than one way for extracting loyalty information from network data .
we chose the route of loyalty distributions since they capture the two most important elements of loyalty : the proportion of customers loyal to one s business , and the degree of their loyalty .
notice , in particular , that we do not try to dichotomize loyalty ( i.e. , categorize it into loyal vs. disloyal buyers ) : since we do not believe that loyalty can be turned on or off arbitrarily , we allow it to range on a continuous scale between 0 and 1 . this will allow us to quantify the impact of the _ shape _ of a seller s loyalty distribution on his or her bottom line .
for instance , it will allow us to answer whether sellers with _ pure _ loyalty ( i.e. , all buyers 100% loyal ) are better off compared to sellers with more variation among their customer base .
we would also like to caution that the resulting analysis is complex since we first have to characterize the infinite - dimensional loyalty distributions in a finite way , and subsequently interpret the resulting characterizations .
the resulting interpretations are more complex than , say , employing user - defined measures of loyalty ( e.g. , summary statistics such as the number of loyal buyers or the proportion of at least 70% loyal buyers ) .
while such user - defined measures are easy to interpret , there is no guarantee that they capture all of the relevant information .
( for instance , measuring the `` number of loyal bidders '' would first require us to define a cutoff at which we consider one buyer to be loyal and another one to be disloyal any such cutoff is necessarily arbitrary and would lead to a dichotomization which we are trying to avoid . )
rather than employing arbitrary , user - defined measures , we set out to let the data speak freely and first look for ways to summarize the information captured in the loyalty distributions in the most exhaustive way .
this will lead us to the notion of principal component loyalty scores and their interpretations .
we will elaborate on both aspects below .
consider the hypothetical seller
bidder network in figure [ fig : illustration ] . in that network
, we have 4 sellers ( labeled `` a , '' `` b , '' `` c '' and `` d '' ) and 10 bidders ( labeled 110 ) .
an arc between a seller
bidder pair denotes an interaction , and the width of the arc is proportional to the number of repeat - interactions between the pair .
consider bidder 1 who has a total of 10 interactions , all of which are with seller a ; we can say that bidder 1 is 100% loyal to seller a. this is similar for bidders 2 and 3 , who have a total of 8 and 6 interactions , respectively , all of which are , again , with seller a. in contrast , bidders 4 and 5 are only 80% and 70% loyal to seller a since , out of their total number of interactions ( both 10 ) , they share 2 with seller b and 3 with seller c , respectively .
all - in - all , seller a attracts mostly highly loyal bidders .
this is different for seller d who attracts mostly little loyal bidders , as he shares all of his bidders with either seller b or c. c + + + for each seller , we can summarize the proportion of loyal bidders and the degree of their loyalty in the associated _
loyalty distribution_. the loyalty distributions for sellers a d are displayed in the right panel of figure [ fig : illustration ] .
the @xmath2-axis denotes the degree of loyalty ( e.g. , 100% or 80% loyal ) , and the @xmath3-axis denotes the corresponding density .
we can see that the shape of all four distributions is very different ; while seller a s distribution is very left - skewed ( mostly high - loyal bidders ) , seller d s distribution is very right - skewed ( mostly little - loyal bidders ) .
the distributions of sellers b and c fall somewhat in between , yet they are still very distinct from one another . note that our definition of loyalty is similar to the concept of in- and out - degree analysis .
more precisely , we first measure the proportion of interactions for each buyer ( i.e. , the normalized distribution of out - degree ) . then , we measure the perceived loyalty of each seller , which can be viewed as the distribution of the weighted in - degree .
this definition of loyalty is very similar to the concept of brand - switching in marketing .
in essence , if we have a fixed number of brands ( sellers in our case ) and a pool of buyers ( i.e. , bidders ) , then we measure the switching - behavior from one brand to another .
while the loyalty distributions in figure [ fig : illustration ] capture all of the relevant information , we can not use them for further analysis ( especially modeling ) .
thus , our next step is to characterize each loyalty distribution by only a few numbers . to that end
, we employ a very flexible dimension reduction approach via functional data analysis . in order to investigate the effect of loyalty on the outcome of an auction ,
we first need to characterize a seller s loyalty distribution .
while one could characterize the distributions via summary statistics ( e.g. , mean , median or mode ) , figure [ fig : illustration ] suggests that loyalty is too heterogeneous and too dispersed .
therefore , we resort to a very flexible alternative via functional data analysis [ @xcite ] . by functional data
we mean a collection of continuous functional objects such as curves , shapes or images .
examples include measurements of individuals behavior over time , digitized 2- or 3-dimensional images of the brain , or recordings of 3- or even 4-dimensional movements of objects traveling through space and time . in our application , we regard each seller s loyalty distribution as a functional observation .
we capture similarities ( and differences ) across distributions via _ functional principal component analysis _
( fpca ) , a functional version of principal component analysis [ see @xcite ] .
in fact , while @xcite operate on the true probability distributions , these are not known in our case ; hence , we apply fpca to the observed ( empirical ) distribution function , which may introduce an extra level of estimation error .
functional principal component analysis is similar in nature to ordinary pca ; however , rather than operating on data - vectors , it operates on functional objects . in our context
, we take the observed loyalty distributions [ i.e. , the histograms figure [ fig : illustration](b ) ] as input . while one could also first smooth the observed histograms , we decided against it since the results were not substantially different .
ordinary pca operates on a set of data - vectors , say , @xmath4 , where each observation is a @xmath5-dimensional data - vector @xmath6 .
the goal of ordinary pca is to find a projection of @xmath4 into a new space which maximizes the variance along each component of the new space and at the same time renders the individual components of the new space orthogonal to one another . in other words , the goal of ordinary pca is to find a pc vector @xmath7 for which the principal component scores ( pcs ) @xmath8 maximize @xmath9 subject to @xmath10 this yields the first pc , @xmath11 .
in the next step we compute the second pc , @xmath12 , for which , similarly to above , the principal component scores @xmath13 maximize @xmath14 subject to @xmath15 and the _ additional constraint _ @xmath16
this second constraint ensures that the resulting principal components are orthogonal .
this process is repeated for the remaining pc , @xmath17 .
the functional version of pca is similar in nature , except that we now operate on a set of continuous curves rather than discrete vectors . as a consequence
, summation is replaced by integration .
more specifically , assume that we have a set of curves @xmath18 , each measured on a continuous scale indexed by @xmath19 .
the goal is now to find a corresponding set of pc curves , @xmath20 , that , as previously , maximize the variance along each component and are orthogonal to one another . in other words , we first find the pc function , @xmath21 , whose pcs @xmath22 maximize @xmath9 subject to @xmath23 similarly to the discrete case , the next step involves finding @xmath24 for which the pcs @xmath25 maximize @xmath14 subject to @xmath15 and the additional constraint @xmath26 in practice , the integrals in ( [ eq : fpc score])([eq : fpc contraint ] ) are approximated either by sampling the predictors , @xmath27 , on a fine grid or , alternatively , by finding a lower - dimensional expression for the pc functions @xmath20 with the help of a basis expansion . for instance
, let @xmath28 be a suitable basis expansion [ @xcite ] , then we can write @xmath29 for a set of basis coefficients @xmath30 . in that fashion , the integral in , for example , ( [ eq : fpc contraint ] ) becomes @xmath31 where @xmath32 .
for more details , see @xcite . in this work
we use the grid - approach .
-origin . ]
common practice is to choose only those eigenvectors that correspond to the largest eigenvalues , that is , those that explain most of the variation in @xmath18 . by discarding those eigenvectors that explain no or only a very small proportion of the variation
, we capture the most important characteristics of the observed data patterns without much loss of information .
in our context , the first 2 eigenvectors capture over 82% of the variation in loyalty distributions .
since our loyalty measures are based on their principal component representations , interpretation has to be done with care .
figure [ fig : pc loadings ] shows the first 2 principal components ( pcs ) .
the first pc ( top panel ) shows a growing trend and , in particular , it puts large negative weight on the lowest loyalty scores ( between 0 and 0.2 ) while putting positive weight on medium to high loyalty scores ( 0.4 and higher ) .
thus , we can say that the first pc contrasts the extremely disloyal distributions from the rest .
table [ pc_summaries ] ( first row ) confirms this notion : notice the large negative correlation with the minimum ; also , the large correlation with the skewness indicates that pc1 truly captures extremes in the loyalty distributions scores and shape .
we can conclude that pc1 distinguishes distributions of `` pure disloyalty '' from the rest .
the second pc has a different shape .
the second pc puts most ( positive ) weight on the highest loyalty scores ( between 0.8 and 1 ) ; it puts negative weight on scores at the medium and low scores ( between 0.4 and 0.6 ) and thus contrasts average loyalty from extremely high loyalty .
indeed , table [ pc_summaries ] ( second row ) shows that pc2 has a high positive correlation with the maximum and a high negative correlation with the median . in that sense
, it distinguishes the mediocre loyalty from the stars .
while the above interpretations help our understanding of the loyalty components , their overall impact is still hard to grasp , especially because every individual loyalty distribution will by nature of the principal component decomposition
comprise of a different _ mix _ between pc1 and pc2 .
moreover , as we apply fpca to observed densities ( i.e. , histograms ) , individual values of each density function must be heavily correlated .
this adds additional constraints on the pcs and their interpretations .
hence , in the following , we discuss five _ theoretical _ loyalty distributions and their corresponding representation via pc1 and pc2 .
[email protected]@ & & & & & + pc1 & 0.55 & -0.2 & 0.52 & -0.99 & 0.77 + pc2 & -0.78 & -0.05 & 0.81 & -0.02 & 0.63 + take a look at figure [ fig : theoretic loyalty ] .
it shows five plausible loyalty distributions as they may develop out of a bidder seller network .
we refer to these distributions as `` theoretic loyalty distributions '' and we can characterize them by their specific shapes . for instance , the first distribution is comprised of 100% loyal buyers and we hence refer to it as `` pure loyalty ; '' in contrast , the last distribution is comprised of 100% disloyal buyers and we hence name this distribution `` pure disloyalty ; '' the distribution in the center ( `` somewhat loyal '' ) is interesting since it is comprised mostly of buyers that exhibit some loyalty but do not purchase exclusively from only one seller
. table [ theoretical_loyalties ] shows the corresponding pc scores .
we can see that the theoretical distribution corresponding to _ pure loyalty _ scores very high on pc1 since it is very right - skewed and does not have any values lower than 0.9 ; in contrast , notice the pc1 scores for _ pure disloyalty _ : while it is the mirror image of _ pure loyalty _ , it scores ( in absolute terms ) higher than the former because it is not only very ( left- ) skewed , but its extremely small values weigh heavily ( and negatively ) with the first part of the pc1-shape , which is in contrast to the positive values of _ pure loyalty _ which do not receive as much weight .
as for pc2 , table [ theoretical_loyalties ] shows that _ pure loyalty _
scores even higher on that component as its values are extremely large , much larger than the typical ( median ) loyalty values . in contrast , _ pure disloyalty _ has very small pc2 values , as low scores are given very little weight by pc2 .
@lccd2.2d2.2d2.2@ & * pure lyty * & * strong lyty * & & & + pc1 & 0.56 & 0.47 & 0.32 & -0.04 & -0.64 + pc2 & 0.72 & 0.51 & -0.08 & 0.35 & -0.01 + we can make similar observations for the remaining theoretical loyalty distributions .
for instance , the distribution of _ somewhat loyal _ scores high on pc1 since it does not have many low values ; but it also only receives an average score on pc2 since it does not have many high values either . in the following , we will use these theoretical loyalty distributions to shed more light on the relationship between loyalty and the outcome of an auction .
our goal is to investigate the effect of loyalty on the outcome of an auction .
for instance , we would like to see whether sellers who attract exclusively high - loyal bidders elicit price - premiums , or whether more variability in buyers loyalty leads to a higher price .
to that end , we start out , in similar fashion to many previous studies on online auctions [ e.g. , @xcite ] , with an ordinary least squares ( ols ) modeling framework .
that is , we investigate a model of the form @xmath33 where @xmath34 is a matrix of covariates and @xmath35 follows the standard linear model assumptions . for the choice of the covariates , we are primarily interested in the effect of loyalty on the price of an auction ( i.e. , the first 2 pc scores from the previous section are our main interest ) .
however , we also want to control for factors other than loyalty which are also known to have an impact on price ; these factors include auction characteristics ( auction duration ) , item characteristics ( item quantity , size and pieces ) , seller characteristics ( seller feedback , i.e. , reputation and seller volume ) and auction competition ( number of bids , i.e. , bid - count ) .
we first investigate a standard ols model that relates these covariates to price .
however , we will show that an ols approach leads to violations of the model assumptions .
the reason lies in the asymmetry of the bidder seller network : the presence of high - volume sellers ( i.e. , seller nodes with extremely high degree ) biases the analysis and leads to wrong conclusions .
in particular , high - volume sellers have many repeat interactions which result in a strong clustering of the data and thus violate the i.i.d .
assumption of ols .
we investigate several remedies to this problem .
first , we investigate two `` standard '' remedies via random effect ( re ) models and weighted least squares ( wls ) .
our results show that although both remedies ease the problem , none removes it completely .
we thus argue that the data is too heterogeneous to be modeled within a single model and compare our results with that of a data - segmentation strategy .
many studies employ an ols modeling framework to investigate phenomena in online auctions such as the effect of the auction format , the impact of a seller s reputation , or the amount of competition [ e.g. , @xcite ; @xcite ; @xcite ] . however , one problem with an ols model approach is the presence of repeat observations on the same item .
for instance , if we want to study the effect of a seller s reputation ( measured by her feedback score ) , then repeat auctions by the same seller will severely overweight the effect of high - volume sellers in the ols model .
this problem is typically not addressed in the online auction literature .
we face a very similar problem when modeling the effect of e - loyalty .
@ld2.2cc@ * coefficient * & & * std .
error * & * @xmath36-value * + ( intercept ) & -1.64 & 0.06 & 0.00 + auction duration & 0.02 & 0.00 & 0.00 + log(item quantity @xmath0 1 ) & -0.04 & 0.04 & 0.30 + bid count & 0.06 & 0.00 & 0.00 + log(pieces ) & 0.42 & 0.00 & 0.00 + size & 0.07 & 0.00 & 0.00 + log(seller feedback @xmath0 1 ) & -0.00 & 0.01 & 0.52 + loyalty - pc1 & -0.17 & 0.05 & 0.00 + loyalty - pc2 & -1.00 & 0.07 & 0.00 + log(volume ) & 0.16 & 0.01 & 0.00 + [ 3pt ] aic & + @xmath37-squared & 0.77 + [ 6pt ] ( intercept ) & -0.58 & 0.22 & 0.00 + auction duration & 0.01 & 0.00 & 0.00 + log(item quantity @xmath0 1 ) & -0.21 & 0.07 & 0.00 + bid count & 0.05 & 0.00 & 0.00 + log(pieces ) & 0.27 & 0.00 & 0.00 + size & 0.03 & 0.00 & 0.00 + log(seller feedback @xmath0 1 ) & 0.07 & 0.04 & 0.11 + loyalty - pc1 & -0.40 & 0.22 & 0.07 + loyalty - pc2 & -0.15 & 0.22 & 0.51 + log(volume ) & 0.05 & 0.04 & 0.20 + [ 3pt ] aic & + @xmath37-squared & + [ 6pt ] ( intercept ) & -1.59 & 0.12 & 0.00 + auction duration & 0.01 & 0.00 & 0.01 + log(item quantity @xmath0 1 ) & 0.30 & 0.10 & 0.00 + bid count & 0.08 & 0.00 & 0.00 + log(pieces ) & 0.26 & 0.01 & 0.00 + size & 0.00 & 0.00 & 0.61 + log(seller feedback @xmath0 1 ) & -0.01 & 0.01 & 0.59 + loyalty - pc1 & -1.00 & 0.08 & 0.00 + loyalty - pc2 & 1.33 & 0.11 & 0.00 + log(volume ) & 0.24 & 0.01 & 0.00 + [ 3pt ] aic & + @xmath37-squared & 0.43 + for illustration , take the ols regression model in the top panel of table [ effect of theoretical loyalties ] . in this model
we estimate the dependency of ( log-)price on loyalty ( measured by pc1 and pc2 ) , controlling for all other factors described above .
note that this model appears to fit the data very well ( @xmath38-squared @xmath39 77% ) .
however , it is curious to see that seller feedback has a negative sign and is statistically insignificant .
this contradicts previous findings which found that an increased level of trust leads to price premiums [ @xcite ; @xcite ; @xcite ] .
figure [ fig : res2 ols ] shows the residuals corresponding to the above model .
the top half shows the residuals plotted against seller - volume ; the bottom half shows the residual distribution . for each type of graph we present 4 different views : one graph ( left graphs in first and third panel ) gives an overview ; the other graphs zoom in by seller volume ( low , medium and high volume , respectively ) .
notice that the residuals are rather skewed : a large proportion of residuals are negative ( see , e.g. , top left graph ) , implying that our model over - estimates price effects of loyalty .
moreover , we can also see that the residual - variation increases for larger seller volumes .
if we zoom in on both the low - volume and medium - volume sellers , we can see that the true effect of model misspecification is confounded with seller volume : while price effects of low - volume sellers are underestimated ( note the positive - skew in the residual distribution for low - volume sellers ) , the effects are overestimated for medium - volume sellers ( negative - skew ) ; only high volume sellers appear to be captured well by the model .
thus , the ols regression model blends low volume and medium volume sellers but represents neither of them adequately .
we have seen in the previous section that an ols approach does not result in a model that can be interpreted without concerns .
we thus investigate two alternate models , a random effects ( re ) model and weighted least squares ( wls ) .
random effects models are often employed when there are repeat observations on the same subject or when the data is clustered [ e.g. , @xcite ] .
since we have many repeat auctions by the same seller , adding a random , seller - specific effect to the model in ( [ eq : regression model ] ) lends itself as a natural remedy for ols .
while re models have become popular only recently with the advent of powerful computing and efficient algorithms , wls has been around for a longer time as a possible solution for heteroscedasticity [ @xcite ] .
while the principle of wls is powerful , it assumes that the matrix of weights is known ( or at least known up to a parameter value ) , which reduces its practical value . in our context , we use weights that are inversely proportional to the residual variance in each cluster .
we will now compare both approaches and see if they result in more plausible models for e - loyalty.=-1 table [ effect of theoretical loyalties ] ( second and third panels ) shows the results of the re and wls models , respectively .
we can see that wls results in a very poor model fit ( both in terms of @xmath38-squared and aic ) . while the re model results in much better model fit ( compared to both the wls and the ols model ) , it is curious that seller feedback is insignificant , similar to the ols model above . in fact , it is quite curious that none of the seller - related variables ( feedback , loyalty or volume ) are significant in the re model .
this finding suggests that none of the actions taken by the seller affect the outcome of an auction , which contradicts both common practitioner knowledge as well as previous research on the topic [ @xcite ; @xcite ; @xcite ] .
figure [ fig : res2 random ] shows the residuals of the re model .
we can see that the magnitude of the residuals has decreased , suggesting a better model fit .
this is expected as the random effects account for seller - specific variation due to individual selling strategies ( e.g. , seller - specific auction parameters or product descriptions ) , which all may lead to differences in final price .
but we can also see that the re model still suffers from heteroscedasticity ( much larger residual variance for high volume sellers compared to low volume sellers ) .
figure [ fig : res2 wls ] shows the corresponding residuals of the wls approach . while we would have expected that wls tames the heteroscedasticity somewhat , it appears that model fit has become worse .
( this is also supported by the much poorer values of @xmath38-squared and aic . )
one possible reason is that weights have to be chosen by the user ( inversely proportional to seller volume , in our case ) , which may not result in the most appropriate weighting of the data .
none of the proposed modeling alternatives so far have lead to models with reasonable residuals or economically defendable conclusions .
in fact , we have seen that the model fit differs systematically by the seller volume .
we take this as evidence that the data may be segmented into different _
seller volume clusters_. we have seen earlier ( e.g. , figures [ fig : net 2 ] and [ fig : net 4 ] ) that sellers of different magnitude exhibit quite different effects on bidders
. we will thus now first cluster the data and then model each data - segment separately .
we first cluster the data by seller volume ( low , medium and high ) and then apply ols regression within each segment , resulting in three different regression models , one for each segment .
we select the clusters with the objective of minimizing the residuals mean squared errors within each cluster .
this results in the following three segments : low volume sellers40 transactions or less ; medium volume sellers40350 transactions ; high volume sellers more than 350 transactions .
figure [ fig : res cluster ] shows the residuals of the resulting three models .
we can see that the model fit is much better compared to the previous modeling approaches . in each segment
the magnitude of the residuals is very small , all residuals scatter around the origin , and we also no longer find evidence for heteroscedasticity in any of the three segments .
in fact , the model fit statistics ( see table [ reg : cluster ] ) suggest that the segmentation approach leads to a much better representation of the data compared to either ols , re or wls models
. table [ reg : cluster ] shows the parameter estimates for each segment .
we can see that the relationship between loyalty , trust and price varies from segment to segment .
in fact , while for the low volume sellers the significance of all seller - related variables ( feedback , loyalty or volume ) is low , both feedback and volume are much more significant than loyalty .
( note the much smaller @xmath5-values of seller feedback and volume . )
this suggests that while seller - related actions may not play much of a role for low volume sellers ( such as rookie sellers and sellers that are new to the market ) , trust is much more important compared to loyalty .
this makes sense as low volume sellers have not much of a chance to establish a loyal customer base due to the infrequency of their transactions .
this is different for medium volume sellers . for medium volume sellers , loyalty and volume
are more significant than feedback .
this suggests that with increasing frequency of transactions , repeat transactions ( i.e. , loyalty ) have a more dominant effect on a seller s bottom line .
this effect is even more pronounced for high volume sellers .
this suggests that high volume sellers are most affected by the actions of repeat customers .
it is also interesting that both feedback and loyalty are significant for high volume sellers .
this suggests that in the presence of two sellers with the same reputation , buyers `` act with memory '' and return to repeat their previous shopping experience . in order to precisely quantify the effect of loyalty in each segment , consider table [ tblols ] . in that table , we present , for each of the 5 theoretical loyalty distributions from figure [ fig : theoretic loyalty ] , their corresponding combined effect on the regression model .
that is , we compute the combined effect of pc1 and pc2 , holding all other variables in the model constant .
[email protected]@ * coefficient * & & * std .
error * & * @xmath36-value * + ( intercept ) & -0.51 & 0.36 & 0.16 + auction duration & 0.04 & 0.02 & 0.10 + log(item quantity @xmath0 1 ) & -0.06 & 0.10 & 0.50 + bid count & 0.12 & 0.01 & 0.00 + log(pieces ) & 0.15 & 0.06 & 0.02 + size & 0.06 & 0.03 & 0.05 + log(seller feedback @xmath0 1 ) & 0.08 & 0.04 & 0.07 + loyalty - pc1 & -0.22 & 0.16 & 0.18 + loyalty - pc2 & -0.03 & 0.13 & 0.82 + log(volume ) & -0.10 & 0.06 & 0.07 + [ 3pt ] aic & & & + @xmath37-squared & 0.76 & & + [ 6pt ] ( intercept ) & -0.49 & 0.38 & 0.20 + auction duration & -0.05 & 0.01 & 0.00 + log(item quantity @xmath0 1 ) & -0.11 & 0.19 & 0.58 + bid count & 0.17 & 0.01 & 0.00 + log(pieces ) & 0.03 & 0.01 & 0.05 + size & 0.01 & 0.01 & 0.58 + log(seller feedback @xmath0 1 ) & -0.01 & 0.02 & 0.65 + loyalty - pc1 & -0.24 & 0.12 & 0.04 + loyalty - pc2 & 0.28 & 0.19 & 0.15 + log(volume ) & 0.33 & 0.08 & 0.00 + [ 3pt ] aic & + @xmath37-squared & 0.75 + [ 6pt ] ( intercept ) & 1.68 & 0.10 & 0.00 + auction duration & 0.02 & 0.00 & 0.00 + log(item quantity @xmath0 1 ) & -0.25 & 0.04 & 0.00 + bid count & 0.05 & 0.00 & 0.00 + log(pieces ) & 0.39 & 0.00 & 0.00 + size & 0.06 & 0.00 & 0.00 + log(seller feedback @xmath0 1 ) & 0.42 & 0.01 & 0.00 + loyalty - pc1 & 7.39 & 0.16 & 0.00 + loyalty - pc2 & -9.41 & 0.17 & 0.00 + log(volume ) & -0.78 & 0.02 & 0.00 + [ 3pt ] aic & + @xmath37-squared & 0.83 + @ld2.2d2.2d2.2d2.2d2.2@ & & & & & + cluster 1 ( low ) & -0.14 & -0.12 & -0.07 & 0.00 & 0.14 + cluster 2 ( medium ) & 0.07 & 0.03 & -0.10 & 0.11 & 0.15 + cluster 3 ( high ) & -2.60 & -1.29 & 3.14 & -3.61 & -4.62 + we can see that in clusters 1 and 2 , the effect of loyalty is considerably small , consistent with the small ( and insignificant ) coefficients for low and medium volume sellers in table [ reg : cluster ] . for cluster 3 ,
it is interesting that only the distribution corresponding to _ somewhat loyal _ buyers results in a positive price effect .
in fact , we can see that extreme loyalty ( i.e. , the distributions for both _ pure loyalty _ and _ pure disloyalty _ ) has negative implications for a seller s bottom line .
while the effect of disloyal bidders is easier to explain ( disloyal bidders may `` shop around '' more actively in the search for lower prices and , as a result , drive down a seller s revenue ) , the negative effect of purely loyal bidders may be due to the fact that a bidder who exclusively interacts with the same seller may form an opinion about that seller s `` going price '' which results in a less competitive auction process ( and thus renders the transaction into a fixed - price transaction ) .
thus , our results show that the effect of loyalty is surprisingly `` nonlinear '' in that a mix of somewhat loyal bidders results in the most competitive auction environment and thus the highest price for the seller .
another way of quantifying the impact of loyalty is via the difference between pure loyalty and pure disloyalty .
notice that the difference in estimated coefficients equals ( @xmath40 2 , which ( as the response is on the log - scale ) implies that , all else equal , sellers with a purely loyal customer base extract price premiums 200% higher compared to sellers with a purely disloyal customer base .
in this paper we investigate loyalty of online transactions . loyalty is an important element to many business models , and it is especially difficult to manage in the online domain where consumers are offered different choices that are often only a mouse - click away .
we study loyalty in online auctions .
we derive online loyalty from the network of sellers and bidders and find that while bidder s loyalty can have a strong impact on the outcome of an auction , the magnitude of its impact varies depending on the size of the seller .
we want to point out that while we find that loyalty has a strong effect on price , we do not determine the cause of loyalty .
a buyer s loyalty can have many different causes such as a high - quality product , a speedy delivery , or an otherwise seamless service . while loyalty could also be caused by price itself ( i.e. , a buyer returning to the same seller because of a low price ) , it is unlikely in our setting due to the auction process .
recall that in an auction the price is not fixed .
thus , a seller offering a top notch product and an outstanding service will sooner or later see an increase in bidders and , as a result , more competition and thus a higher price for her product .
thus , loyalty is unlikely to be caused merely by bargain sellers.=1 also , we want to emphasize that while we find many repeat transactions between the same seller bidder pair in our data , the frequency of these repeat interactions may depend on the type of product and the buyer s demand for this product . in our case ( beads , i.e. , arts and crafts ) ,
buyers have frequently re - occurring demand for the same product and , hence , the chances that a buyer will seek out the same seller rise drastically . on the other hand , if we were to consider the market for a product in which repeat transactions are less common ( such as computers , digital cameras , automobiles , etc . ) , our loyalty networks would likely not be as dense
. nevertheless , it would be equally important for sellers to understand what factors drive consumers to spend money and we believe that loyalty networks are one way to address that question .
there are several statistical challenges when studying loyalty networks .
first , deriving quality measures from the observed networks requires a method that can capture both the intensity as well as the size of loyalty .
we accomplish this using ideas from functional data analysis .
second , modeling the effect of loyalty is complicated by the extreme skew of loyalty networks .
our analysis shows that many different approaches can lead to model misspecification and , as a consequence , to economically wrong conclusions .
similar problems likely exist in other studies on online markets ( e.g. , those that study seller feedback or reputation where one also records repeat observations on the same seller ) .
our analysis leads us to conclude that the data is too segmented to be treated by a single model and thus propose a data - clustering approach .
another statistical challenge revolves around _ sampling _ bidder seller networks .
as pointed out earlier , we have the complete set of bidding records for a certain product ( swarovski beads , in this case ) for a certain period of time ( 6 months ) . as a result , we have the complete bidder seller network for this product , for this time frame . while sampling would be an alternative , it would result in an incomplete network ( since we would no longer observe all nodes / arcs ) . as a result
, we would no longer be able to compute loyalty without error , which would bring up an interesting statistical problem .
but we caution that sampling would have to be done very carefully . while one could , at least in theory , sample randomly across all different ebay categories , it would bring up several problems .
the biggest problem is that we would now be attempting to compare loyalty across different product types .
for instance , we would be comparing , say , a bidder s loyalty for purchasing beads ( a very low price , low stake item ) with that of purchasing digital cameras , computers , or even automobiles ( all of which are high price and high stakes ) , which would be conceptually very questionable .
we also want to mention that we treat the bidder seller network as _ static _ over time .
our data spans a time - frame of only 6 months and we assume that loyalty is static over this time - frame .
this assumption is not too unrealistic as many marketing models consider loyalty to be static over much longer time frames [ @xcite ; @xcite ; @xcite ] . while incorporating a temporal dimension ( e.g. , by using a network with a sliding window or via down - weighting older interactions ) would be an intriguing statistical challenge , it is not quite clear how to choose the width of the window or the size of the weights . moreover , we also explicitly tested for learning effects by buyers over time and could not find any strong statistical evidence for it . and finally , in this work we address one specific kind of network dependence , namely , that between buyers and sellers .
we argue that the lack of independence among observations on the same sellers leads to a clustering - effect and we investigate several remedies to this challenge .
however , the dependence structure may in fact be far more complex .
as bidders are linked to sellers which , in turn , are linked again to other bidders , the true dependence structure among the observations may be far more complex .
this may call for innovative statistical methodology and we hope to have sparked some new ideas with our work .
bailey , j. , gao , g. , jank , w. , lin , m. , lucas , h. c. and viswanathan , s. ( 2008 ) .
the long tail is longer than you think : the surprisingly large extent of online sales by small volume sellers .
technical report , rh smith school of business , university of maryland .
available at ssrn : http://ssrn.com/abstract=1132723 .
donkers , b. , verhoef , p. c. and de jong , m. ( 2003 ) .
predicting customer lifetime value in multi - service industries .
technical report , erim report series reference no .
ers-2003 - 038-mkt .
available at ssrn : http://ssrn.com/abstract=411666 .
jank , w. and zhang , s. ( 2008 ) .
an automated and data - driven bidding strategy for online auctions . technical report , rh smith school of business , univ .
available at ssrn : http://ssrn.com/abstract=1427212 . | creating a loyal customer base is one of the most important , and at the same time , most difficult tasks a company faces .
creating loyalty online ( e - loyalty ) is especially difficult since customers can `` switch '' to a competitor with the click of a mouse . in this paper
we investigate e - loyalty in online auctions . using a unique data set of over 30,000 auctions from one of the main consumer - to - consumer online auction houses , we propose a novel measure of e - loyalty via the associated network of transactions between bidders and sellers . using a bipartite network of bidder and seller nodes ,
two nodes are linked when a bidder purchases from a seller and the number of repeat - purchases determines the strength of that link .
we employ ideas from functional principal component analysis to derive , from this network , the loyalty distribution which measures the perceived loyalty of every individual seller , and associated loyalty scores which summarize this distribution in a parsimonious way .
we then investigate the effect of loyalty on the outcome of an auction . in doing so ,
we are confronted with several statistical challenges in that standard statistical models lead to a misrepresentation of the data and a violation of the model assumptions .
the reason is that loyalty networks result in an extreme clustering of the data , with few high - volume sellers accounting for most of the individual transactions .
we investigate several remedies to the clustering problem and conclude that loyalty networks consist of very distinct segments that can best be understood individually . . |
binarity has long been suggested as a mechanism to shape bipolar planetary nebulae ( pne ) ( e.g. * ? ? ?
* ; * ? ? ?
observations have confirmed that a significant number of pne indeed have binary central stars @xcite .
since binarity precedes the formation of the pn , it is paramount to detect it along previous phases of stellar evolution before the pn formation , particularly at the asymptotic giant branch ( agb ) phase .
an unbiased comparison among the binarity occurrence rates during the main sequence , agb and pn phases can help to reinforce a causality relationship between binarity and the formation of aspherical pne and to assess the evolution of binary systems ( e.g. * ? ? ?
* ; * ? ? ?
* ) binary or multiple systems including agb stars have been often observed as symbiotic systems .
they are identified by their spectra , which includes features characteristic of the red giant as well as emission lines arising from the wind - driven atmosphere of the giant , which is ionized by the uv photons of the secondary white dwarf ( wd ) .
the latest catalogue of symbiotic systems contains about two hundred objects , including confirmed and suspected objects @xcite .
the discovery of hot companions of agb stars is somewhat straightforward where they compose symbiotic systems , but the detection of low- and intermediate - mass main - sequence ( hereafter ms ) companions is not simple .
the detection of the secondary in direct images is difficult because the high brightness contrast between them hampers the detection of the secondary , except in the cases where the components are well resolved @xcite . indeed , the recent advent of new generation adaptive optics systems has allowed the detection of late - type , close companions of agb stars @xcite , but this method still remains restricted to a few near objects @xcite .
other methods to detect binary agb stars include asymmetries in their circumstellar envelopes @xcite , proper - motion variations @xcite , and the identification of features attributed to the secondary in the visual spectrum @xcite . as a rule , the shorter the wavelength , the higher the relative contribution of the hot component to the spectrum because the flux emitted by an agb star decreases abruptly beyond @xmath22800 .
therefore , uv space observatories ( _ fuse _ , _ galex _ , _ hst _ ) have greatly increased the possibilities to detect ms companions of agb stars .
@xcite have carried out a program in quest of binary agb stars based on an imaging survey obtained by the _ galex _ observatory ( _ galaxy evolution explorer , _ * ? ? ?
they selected 25 agb stars showing m5 or later spectral type classified during the mission as `` bright star '' and `` high - background '' , with the additional criterium that they should exhibit the `` multiplicity '' flag in the catalogue .
these were considered as promising conditions to detect companions with spectral type earlier than g0 in the two _ galex _ photometric bands : far - uv ( @xmath3 ) and near - uv ( @xmath4 ) .
indeed , uv counterparts were detected in most of the sources and for a significant fraction of them the uv emission seemed to be in excess , i.e. it probably results from a companion undetected at visual wavelengths .
these results have undoubtfully made a major contribution in increasing the number of known binary agb stars , but these searches have mostly been biased towards the most promising candidates and/or been restricted to symbiotic systems . in this paper
we address the problem of detecting ms companions of agb stars using _ galex _ uv data .
the issues explored in this study will pave the way for future unbiased determinations of the occurrence of binarity among agb stars .
the main scopes of the present paper are : ( 1 ) to establish the detection limits of the _ galex _ survey to detect ms companions of agb binary stars ( sect .
[ detect ] and [ method ] ) ; ( 2 ) to derive criteria for selecting binary agb star candidates , to be eventually confirmed by other techniques , e.g. radial velocity ( rv ) studies ( sect . [ results ] ) ; and ( 3 ) to assess the stellar properties of these companions , mainly their effective temperature ( sect .
[ discussion ] ) .
the detection of the ms secondary companion of an agb star depends on the contrast between its flux and that of the primary .
the detectability of a ms intermediate or early spectral type companion using uv photometry is feasible because , despite the high luminosity of the primary , its flux bluewards of 2800 is usually negligible when compared to the flux emitted by the secondary .
for instance , @xcite estimated a secondary - to - primary _ galex _ near - uv flux contrast ratio @xmath510 for a ms secondary star with spectral type earlier than g0 ( or @xmath6 k ) . besides the flux contrast between the two components , the detectability depends also on the distance to the system as it determines the uv flux .
this issue was extensively addressed by @xcite for hot objects .
assuming theoretical atmosphere models , they conclude that _ galex _ could detect all wd in the galactic halo along their constant - luminosity phase ( i.e. wd s with @xmath7 k and radii down to @xmath8 ) up to a distance of 20 kpc . concerning _ galex _
detections of ms stars , @xcite focused on objects with @xmath9 k , thus excluding intermediate - mass stars .
_ galex _ observation strategy was mainly organized into three different modes , namely ais , mis and dis , mostly different by their varying exposure times of 100 s , 1500 s , and 30,000 s , respectively .
these modes refer to specific areas in the sky called _ tiles .
_ most observations were made in the ais mode to cover large areas in the sky .
for each entry the _ galex _ catalogue gives , among other information , the magnitude and flux density in the near- and far - uv bands @xcite .
an analysis of sources registered in different mode tiles shows that there are significant differences in sensitivity among them : sources detected with a signal - to - noise ratio @xmath10 of 5 have fluxes 10 @xmath11jy @xmath12 20 @xmath11jy in ais tiles , going down to 5 @xmath11jy @xmath12 10 @xmath11jy in mis tiles given their longer integration time . in order to calculate the limiting distances of ms stars detectable by _ galex
_ , we adopt the ms visual absolute magnitudes given by @xcite and all the theoretical atmosphere models provided by @xcite for solar metallicity stars and spectral type later than b2v , i.e. for intermediate - mass stars .
the grid of @xcite models is equally paced in @xmath13 by 0.5 dex intervals , but unequally in @xmath14 . in all cases
the difference between the @xmath13 values given by @xcite and the value of the adopted theoretical model never exceeded 0.2 dex .
the relative differences in temperature were @xmath15 2.3% .
absolute ( i.e. at a distance of 10 parsecs ) flux densities in the @xmath16 band were calculated by convolving the theoretical spectrum with the @xmath16 filter curve given by @xcite using equation : @xmath17 where @xmath18 is the name of the photometric band ( @xmath16 in this case ) , @xmath19 is the theoretical flux density of the star , and @xmath20 is the filter response curve .
@xmath19 is `` scaled '' in order to match its flux density at @xmath21 with the value corresponding to its absolute visual magnitude @xmath22 by @xcite .
the visual zero - magnitude density flux used in this calibration is also given by @xcite : @xmath23 erg s@xmath24@xmath25@xmath24 . finally , _
near- and far - uv flux densities are calculated from this scaled theoretical spectrum using the same equation [ fx ] , but for the near- and far - uv filter curves given by @xcite .
mag for 10 pc , @xmath26 mag for 100 pc , and @xmath27 mag for 1 and 10 kpc .
the horizontal dashed lines mark the range of ais ( i.e. _ galex _ observations with exposure time of @xmath28 s ) and mis ( @xmath29 s ) detection limits . ]
figure [ fig - detect ] shows the flux density of ms stars in the near- and far - uv bands as a function of the stellar effective temperature for distances in the range from 10 pc up to 10 kpc .
the interstellar extinction decreases the detection radius of the _ galex _ survey . to calculate this effect
, we derive the extinction @xmath30 in the @xmath18 band ( _ x = fuv , nuv _ ) as follows : @xmath31 where @xmath32 is the colour excess in the direction of the star ( extracted from the _ galex _ catalogue ) , and @xmath33 is the extinction coefficient in the _ x _ band .
the behaviour of the interstellar extinction coefficients was exhaustively studied by @xcite who analysed thousands of stars detected by _ galex _ ( and other surveys ) in several directions in the galaxy . in the present study
we adopt an uv extinction law that corresponds to the average total - to - selective coefficient determined by @xcite : @xmath34 and @xmath35 .
therefore the _ extinction - corrected _ ultraviolet flux density @xmath36 can be calculated from equation [ anuv ] and the corresponding extinction coefficients : @xmath37 where @xmath38 is the _ observed _ flux density in the near- and far - uv bands .
equation [ fnuv ] was applied to the theoretical flux densities shown in fig .
[ fig - detect ] to simulate the effect of the interstellar extinction on near- and far - uv flux densities . because the extinction varies from source to source
depending on their distance @xmath39 and line of sight through the galaxy , we simulate various values of @xmath40 in figure [ fig - detect ] from @xmath41 mag for sources at @xmath42 pc to @xmath43 mag for @xmath44 kpc ( which corresponds to an average local extinction of 1 mag kpc@xmath24 , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the results are shown in the same fig .
[ fig - detect ] , where we applied @xmath41 mag to the 100 pc and @xmath43 mag to the 1 and 10 kpc curves . assuming @xmath45 ,
the near - uv ( far - uv ) flux densities would decrease by 19% and 88% ( 13% and 75% ) for @xmath41 mag and @xmath43 mag , respectively .
although a realistic estimation of the extinction should take into account the distance and the galactic coordinates of the object , this simulation allows us to estimate the detection limits of ms stars with _
galex_. ms stars with @xmath46 k ( spectral type g6 or earlier ) could be detected in the near - uv band up to a distance of @xmath47 kpc in ais tiles .
contrary to naive expectations , only hotter stars with @xmath48 k ( spectral type f2 or earlier , i.e. mainly ms stars of spectral types o , b and a ) could be detected in the far - uv band due to its lower sensitivity .
the latter stars are less frequently companions of agb stars because of the steep slope of the initial - mass function .
the method used here for estimating the uv excess of agb stars has much in common with that proposed by @xcite , and has been partially described in sect .
[ detect ] .
firstly , @xmath49 and @xmath16 magnitudes taken from _ the guide star catalogue , gsc2.3 _
@xcite and/or _ hipparcos _
@xcite , and @xmath50 , @xmath51 and @xmath52 ( 2mass , * ? ? ?
* ) magnitudes are corrected for interstellar extinction using the @xmath32 values listed in the _ galex _ catalogue .
the standard values of @xmath53 and @xmath54 of 0.26 , 0.15 and 0.09 for the @xmath50 , @xmath51 and @xmath52 bands , respectively , are adopted @xcite . since @xmath55 mag in most cases ,
the assumption of an alternative reddening law would not change the results significantly .
secondly , the extinction - corrected flux densities corresponding to these five photometric bands are calculated using equation [ fx ] .
these are then fitted with the least - squared method to the theoretical spectral library by @xcite assuming [ fe / h]=0 .
no photometric bands redwards of @xmath52 are considered in these fits because agb stars may have circumstellar dust envelopes that produce an infrared excess in this spectral region .
the average magnitude residuals considering the 5 bands is typically less than 0.2 mag , which is considerably smaller than the amplitude of variability of the stars .
once the best - fit spectrum has been found , its flux density in the _ galex _ near - uv and far - uv bands is calculated using equation [ fx ] .
this latter step deserves a warning .
since the agb stellar spectrum in this spectral region is very steep and the _ galex _ photometric bands are broad , the flux densities @xmath56 and @xmath57 obtained by convolving the theoretical spectrum with the filter curve are dominated by the stellar contribution on the longest wavelength side of the band . as a result ,
the flux densities differ notably , about two orders of magnitude , from the value of the flux at the effective wavelength @xmath58 and @xmath59 , respectively . finally , the ratios @xmath60 and @xmath61 of the extinction - corrected ( using equation [ fnuv ] ) _ galex _ @xmath56 and @xmath57 to the predicted @xmath56 and @xmath57 are computed .
in this section we describe the results of the application of this method to various samples of agb stars .
firstly , we test it with the same stars analysed by ( * ? ? ?
* ; * ? ? ?
[ results - sahai ] ) ; second , a sample of confirmed binary stars discovered using kinematical criteria is studied ( sect . [
results - confirmed ] ) ; finally , the method is applied to a volume - limited sample of nearby ( @xmath62 kpc ) agb stars ( sect .
[ results - nearby ] ) . [ cols="<,^,^,^,^,^,^,^,^,^ " , ] radial velocity monitoring of spectroscopic binaries is a widely used technique to derive their orbital parameters .
@xcite comment that only 1.1% of the stars in the ninth catalogue of spectroscopic binary orbits @xcite are m - type giants .
indeed , most of the binary systems known to contain a red giant are symbiotic , whereas systems with a non - degenerate companion are scarce .
@xcite used a series of radial velocity measurements obtained with the coravel spectrovelocimeter @xcite to determine the binarity status of a number of stars .
binarity was flagged as _ orb _ when the set of kinematic measurements allowed the determination of the orbit , _ orb : _ when the orbit was poorly determined , _ sb _ when binarity was confirmed spectroscopically , but no orbit could be computed from the data , _ sb ? _ when the binarity is only suspected , and _ non - sb _ when binarity is discarded .
our sample of confirmed binary agb stars was extracted from the work by @xcite , with the following main additional criteria : ( a ) the star should be classified as _ orb _ , _ orb : _ or _ sb _ ; ( b ) concerning variability , the star must be classified as mira ( a _ long - period variable _ or lpv ) , sr- ( _ semi - regular _ ) or l - type ( _ slow irregular variable _ ) . the former criterium is devised to select confirmed agb binaries and the latter intends to discard m - stars belonging to the rgb phase .
we also added some objects ( hd62898 , hd108907 , hd130144 , hd150450 , hd156014 and hd187372 ) that are associated with x - ray _ rosat _ sources , as suggested by @xcite .
stars with large x - ray offsets were not included because of their doubtful association .
finally , symbiotic systems , which do not represent the main scope of this paper , have been also rejected .
two stars ( hd16058 and hd42995 ) were not surveyed by _
galex_. our final list , as shown in table [ table - famaey ] , contains 12 confirmed binary stars .
their properties were obtained similarly as those of the sample analysed in sect .
[ results - sahai ] .
table [ table - famaey ] shows that the discrepancy between the _ galex _ measured and predicted @xmath56 persists , but on average @xmath60 is higher than in the sample analysed in sect .
[ results - sahai ] . except in the case of hip58545 ( hd104216 = fr cam ) ,
@xmath60 is generally comparable to the highest values in sahai s sample , ( y gem and @xmath63 ceti ) , i.e. @xmath64 .
two general properties can be derived from this sample of confirmed agb stars : @xmath65 ( except in the case of fr cam ) and the detection of a far - uv counterpart ( except hip60998 = hd108907 = 4 dra ) . after having tested the method proposed by @xcite to investigate the uv excess associated with confirmed binary agb stars ,
we extend it to a volume - limited sample of agb stars .
there are different lists of nearby agb stars in the literature that could be used to assamble a volume - limited sample of agb stars @xcite
. however , all these `` old '' lists suffer from large uncertainties in the determination of the stellar distances @xcite .
this limitation has been overcome with the advent of _ hipparcos _ , which allowed the determination of visual photometry , position , proper motion and annual parallax with accuracy starting at @xmath66 _ mas _ ( milli - arcsec ) for magnitudes @xmath67 up to @xmath68 _ mas _ near the limiting magnitude @xmath69 of the survey @xcite .
thus , since @xmath70 increases with the apparent magnitude and @xmath71 decreases with the distance , the relative uncertainty @xmath72 for a star generally increases with its distance .
red giant stars were selected from the list of _ hipparcos _ sources classified by @xcite as m- , c- ,
s- and oh / ir stars . in order to remove rgb stars from that list , we selected only stars with luminosity higher than the tip of the rgb ( @xmath73 , * ? ? ? * ) obtained by @xcite .
about 95% of the agb stars within 500 pc selected according to this criterium have apparent visual magnitudes @xmath74 , and consequently @xmath75 .
beyond this distance , the distance uncertainties based on _ hipparcos _ data are larger than those reported by other formerly proposed distance scales .
therefore , the volume - limited sample of agb stars in the solar neighbourhood studied in this paper is limited to a distance of 500 pc . after applying the above criteria ,
the sample is reduced to 90 agb stars , from which we additionally discarded those that ( 1 ) are classified in the literature as symbiotic ( 1 object , r aqr = hip117054 ) , ( 2 ) were not observed by _
( 31 objects ) , and ( 3 ) were not _ detected _ by _ galex _ ( 5 objects ) .
our final sample consists of 53 agb stars of various types within 500 pc , with _ galex _ counterparts and @xmath76 ( table [ table - ita ] ) .
according to figure [ fig - detect ] , the limiting temperatures for the detection of ms stars at a distance of 500 pc in _ galex _ data would be @xmath77 k for the near - uv and @xmath78 k , for the far - uv .
lccrcccccc name & d & @xmath14 & @xmath13 & @xmath32 & galex-@xmath57 & galex-@xmath56 & predicted- & @xmath61 & @xmath60 + & ( pc ) & ( k ) & & & ( @xmath11jy ) & ( @xmath11jy ) & @xmath56(@xmath11jy ) & & + + hip5914 & 385 & 3500 & 0.00 & 0.110 & ... & [email protected] & 15.1 & ... & [email protected] + hip14930 & 322 & 3500 & 0.00 & 0.022 & [email protected] & [email protected] & 27.7 & ( [email protected])@xmath80 & [email protected] + hip43811 & 342 & 3350 & @xmath81 & 0.023 & 7.8 & 8.4 & 112.4 & @xmath82 & 0.1 + hip52009 & 208 & 3500 & 0.00 & 0.051 & 6.1 & 221.6 & 61.9 & @xmath83 & 3.6 + hip52577 & 380 & 3500 & 0.00 & 0.016 & [email protected] & [email protected] & 21.0 & ( [email protected])@xmath80 & [email protected] + hip62223 & 321 & 3350 & @xmath81 & 0.023 & ... & [email protected] & 191.8 & ... & [email protected] + hip63152 & 431 & 3350 & @xmath81 & 0.033 & ... & 2.8 & 65.8 & ... & 0.04 + hip91703 & 358 & 3200 & 0.28 & 0.224 & ... & 78.3 & 108.9 & ... & 0.7 + hip95154 & 386 & 3500 & 0.00 & 0.094 & ... & 49.9 & 20.9 & ... & 2.4 + hip117245 & 275 & 3500 & 0.00 & 0.056 & 10.7 & 331.8 & 43.2 & @xmath84 & 7.7 + + hip1236 & 337 & 3000 & 0.60 & 0.015 & ... & [email protected] & 119.5 & ... & 0.3 + hip2086 & 483 & 3500 & 3.00 & 0.023 & 40.9 & 969.3 & 114.0 & @xmath85 & 8.5 + hip9234 & 174 & 2500 & 0.28 & 0.117 & ... & [email protected] & 191.8 & ... & [email protected] + hip13384 & 368 & 3350 & 0.28 & 0.048 & [email protected] & [email protected] & 153.9 & @xmath86 & [email protected] + hip17881 & 260 & 3200 & 0.28 & 0.151 & 47.7 & [email protected] & 336.3 & @xmath87 & [email protected] + hip20075 & 463 & 3500 & 1.50 & 0.046 & 96.1 & 1863.1 & 76.6 & @xmath88 & 24.3 + hip26169 & 333 & 3500 & 3.00 & 0.122 & [email protected]&[email protected] & 256.0 & ( [email protected])@xmath89 &
[email protected] + hip28041 & 437 & 2800 & 0.60 & 1.112 & ... & 62931.8 & 381.5 & ... & 164.9 + hip28874 & 203 & 3200 & 0.28 & 0.038 & 65.9 & 482.2 & 298.3 & @xmath90 & 1.6 + hip38792 & 341 & 4500 & 3.00 & 1.022 & ... & @xmath91 & @xmath92 & ... & 49.6 + hip41028 & 450 & 2800 & 0.60 & 0.038 & [email protected] & [email protected] & 62.2 & ( [email protected])@xmath93 & 1.5 + hip53809 & 261 & 2500 & @xmath94 & 0.050 & ... & [email protected] & 66.6 & ... & 0.8 + hip53940 & 448 & 3500 & 2.00 & 0.138 & 97.3 & 1944.6 & 105.8 & @xmath95 & 18.4 + hip57607 & 242 & 3200 & @xmath81 & 0.113 & ... & [email protected] & 317.7 & ... & 2.8 + hip58225 & 356 & 3200 & 0.60 & 0.028 & 4.0 & [email protected] & 165.8 & @xmath96 & [email protected] + hip62611 & 405 & 3500 & 1.50 & 0.060 & 55.8 & 692.0 & 51.9 &
@xmath84 & 13.3 + hip64569 & 143 & 3000 & -0.70 & 0.034 & ... & 86.9 & 208.9 & ... & 0.4 + hip72208 & 279 & 3350 & @xmath81 & 0.021 & [email protected] & 6833.6@xmath7910,818.5 & 290.4 & ( [email protected])@xmath97 & [email protected] + hip73213 & 403 & 3200 & 0.60 & 0.076 & ... & 335.9 & 248.8 & ... & 1.4 + hip76075 & 305 & 3500 & 2.00 & 0.189 & 90.6 & [email protected] & 140.2 & @xmath83 & [email protected] + hip80488 & 235 & 3000 & @xmath81 & 0.059 & ... & 58.2 & 102.7 & ... & 0.6 + hip80704 & 109 & 3200 & 0.60 & 0.009 & 46.3 & 3097.7 & 1878.4 & @xmath98 & 1.6 + hip98438 & 312 & 3750 & 3.00 & 0.368 & ... & 26,058.7 & 1093.0 & ... & 23.8 + hip98608 & 146 & 3350 & 0.87 & 0.048 & 395.8 & 4423.3 & 927.5 & @xmath99 & 4.8 + hip99082 & 197 & 2500 & @xmath100 & 0.109 & ... & [email protected] & 126.1 & ... & [email protected] + hip99920 & 318 & 3500 & 2.00 & 0.072 & 147.3 & 7618.4 & 117.5 & @xmath101 & 64.8 + hip100935 & 211 & 3000 & @xmath100 & 0.071 & ... & [email protected] & 216.1 & ... & [email protected] + hip102978 & 258 & 4000 & 0.50 & 0.083 & ... & [email protected] & 181.4 & ... & [email protected] + hip104451 & 188 & 3200 & 0.28 & 0.523 & ... & [email protected] & 1405.7 & ... & [email protected] + hip106642 & 175 & 3500 & 2.50 & 0.556 & ... & 55,342.9 & 1214.7 & ... & 45.6 + hip110428 & 250 & 3000 & @xmath81 & 0.157 & ... & 101.1 & 90.2 & ... & 1.1 + hip110736 & 446 & 2500 & @xmath94 & 0.012 & ... & [email protected] & 15.7 & ... & [email protected] + hip114917 & 472 & 3000 & 0.60 & 0.035 & ... & 121.4 & 156.1 & ... & 0.8 + + hip47886 & 347 & 2800 & @xmath100 & 0.015 & 2.5 & [email protected] & 31.6 & @xmath102 & [email protected] + + hip1728 & 270 & 3350 & 0.60 & 0.024 & 24.1 & 1488.4 & 464.1 & @xmath103 & 3.2 + hip1901 & 386 & 2500 & @xmath94 & 0.091 & ... & 50.1 & 12.5 & ... & 4.0 + hip10687 & 448 & 3000 & 0.28 & 0.091 & ... & [email protected] & 88.6 & ... & [email protected] + hip19853 & 450 & 3500 & 1.00 & 0.405 & ... & [email protected] & 46.2 & ... & [email protected] + hip22667 & 200 & 3750 & 3.00 & 0.505 & 13,420.7 & 83,351.5 & 2236.7 & @xmath104 & 37.3 + hip36288 & 253 & 3000 & 0.60 & 0.094 & 9.0 & [email protected] & 273.7 & @xmath105 & [email protected] + hip45058 & 143 & 2800 & 0.28 & 0.021 & 8.0 & 393.9 & 717.5 & @xmath106 & 0.5 + hip77619 & 293 & 3200 & @xmath100 & 0.021 & [email protected] & [email protected] & 111.0 & @xmath107 & [email protected] + hip113131 & 418 & 3500 & 3.00 & 0.091 & 42.4 & 1113.6 & 158.4 & @xmath108 & 7.0 + excess ratio _ versus _ the temperature obtained from the near - uv ( _ top panel _ ) and the far - uv ( _ bottom panel _ ) excess .
the different samples of agb stars are shown using symbols with blue ( ( * ? ? ?
* ; * ? ? ?
* table [ table - sahai ] ) ) , green ( ( * ? ? ? * table [ table - famaey ] ) ) , and red ( distance - limited agb stars , table [ table - ita ] ) points .
when multiple uv observations are available , only the temperatures corresponding to the maximum and minimum values of the fluxes are shown.,width=302 ] the near - uv excess of the sample , expressed as @xmath60 , spans over a wide range , from 0.1 to @xmath109165 . over 70% of the stars show low near - uv excess , @xmath110 , and only 12 stars ( 23% ) show near - uv excess similar to confirmed binary agb stars , i.e. @xmath111 .
another significant statistical differences between this sample and that of confirmed agb binaries are that only 50% of the volume - limited sample ( 27 out of 53 stars ) have a far - uv counterpart , while @xmath61 spans over a wider range of values , from @xmath112 to @xmath113 .
the uv excess observed in agb stars can be used to assess the properties of their putative companions , such as the temperature . assuming that the secondary component is a ms star
, there is a close relationship between its intrinsic flux and the effective temperature . in this section
we obtain the temperature of the companion star using two methods : ( 1 ) the near - uv and the far - uv flux in excess of the agb star and ( 2 ) the near - to - far uv excess flux ratio . once the distance to the the star is known , the uv flux density that would be emitted by a ms companion star is calculated using the method described in sect .
[ method ] .
thus , the secondary s flux density is calculated as the difference between the observed _ galex _ and the predicted ( theoretical ) flux density , listed in tables [ table - sahai ] , [ table - famaey ] and [ table - ita ] . the temperature of the secondary star is given by the intersection of the horizontal line corresponding to the source s uv excess and the flux - density curve corresponding to the distance of the star in the @xmath114 _ versus _ @xmath115 diagram ( fig .
[ fig - detect ] ) .
the method is carried out separately in the near- and far - uv _ galex _ bands .
the second method is based on the @xmath116 excess ratio .
this quantity is compared with the prediction of a theoretical model of ms stars , as discussed in sect .
[ detect ] , being thus similar to the temperature commonly determined from colour indices .
table [ table - dobles ] lists the temperature of the agb companions obtained with the two methods described above .
in general , we observe the following inequality : @xmath117 ( see also fig . [ fig - txt ] ) .
this discrepancy suggests that the uv excess can not be ascribed to a single stellar spectrum .
figure [ fig - temp ] shows the temperature of the secondary members of the three samples obtained with the first method , using the near- and far - uv excesses .
the reduced number of stars closer than 100 parsecs ( only @xmath63 ceti and hd89758 = @xmath11 uma ) is a natural limitation caused by the relative low density of agb stars in the solar neighbourhood . on the other hand , the relative larger number of objects near the distance upper limit of 500 pc results from the fact that , assuming a constant density of stars , the number of objects at a distance @xmath39 increases with @xmath118 . on average , temperatures obtained from the far - uv excess flux are higher than those determined from the near - uv band .
there is also a selective distribution of temperatures when the three distinct samples are compared , but this is mostly due to selection effects . the sample studied by @xcite and @xcite are biased towards uv bright objects , whilst that described in sect .
[ results - nearby ] is a distance - limited sample .
@xcite used the second method to determine the temperature of the secondary , i.e. based on the near - to - far uv flux density ratio and atmosphere models by @xcite . table [ table - dobles ] shows a good agreement between their results and ours .
for instance , y gem has one of the highest @xmath116 ratio and @xcite argue that its companion might have a blackbody temperature as high as 38,000 k , which is beyond the spectral range considered in our study , restricted to intermediate - mass ms stars .
our results , obtained with the same method , confirms this conclusion .
we note that some uv sources are associated with x - ray sources , but no clear relationship seems to exist among them , neither with the amount of far - uv excess , and/or the temperature of the secondary . .
colour code refers to different samples as in figure [ fig - txt ] .
filled symbols represent sources with @xmath65 and empty symbols @xmath119 . in the case where multiple uv observations are available
, the maximum and minimum values of the fluxes are represented.,width=302 ]
one of the major concerns when fitting photometric or spectroscopic data of agb stars is their variability .
this effect acts differently along the spectrum , changing its shape as pulsation changes the stellar temperature . as commented in sect .
[ results - sahai ] , mira- and sr - type variables often show a peak - to - peak variation of @xmath120 magnitudes ( a factor @xmath121 in flux ) in the @xmath122 band @xcite .
the majority of the galex measurements were taken only once or twice , which is clearly insufficient to sample the whole range of the uv fluxes along the cycle . therefore , in the cases where one or just a few multiple photometric measurements of a given source were taken , the actual amplitude of variability might be larger than the measured interval of fluxes .
the majority of the confirmed binaries listed in table [ table - famaey ] show a variation in flux of 10 - 20% , whereas other systems , like @xmath63 ceti and y gem for example , show relative flux variations over one order of magnitude .
this suggests that the variability of the primary might affect the uv flux emitted by its companion .
this effect is more evident in the far - uv band , where the predicted flux of the primary is over 5 orders of magnitude smaller than the values detected by galex .
therefore , besides the intrinsic variability of the primary s atmosphere , other effects might play a role in the large uv amplitude observed in some binary agb stars .
for instance , it is well known that some agb stars may exhibit large circumstellar dust envelopes ( cde ) that can extend up to 1000 stellar radii .
the extended atmosphere of the agb star and its cde may block part of the uv radiation emitted by the companion .
an accurate estimate of this effect depends on various factors , most of them poorly known , such as the orbital parameters of the system , including the inclination angle , the extinction law of the dust ( which depends on the chemical composition ) , the density law , and the opacity of the cde .
figure [ fig - temp ] shows that the variation of 1.3 dex in the near - uv flux of @xmath63 ceti causes a change of 1000 k in the determination of its temperature .
the effect on the far - uv band is even more important : the variation of 1.2 dex in flux of y gem modifies its temperature by 1800 k. another possiblity would be the uv emission produced by an accretion flow onto the secondary or an accretion disk around it .
this would produce stochastic or orbital - locked variations in the uv flux @xcite .
lccccccl name & hipparcos & @xmath123 & @xmath124 & @xmath125 & far - uv & @xmath60 & comments + & & ( k ) & ( k ) & ( k ) & ( y / n ) & & + + rwboo & hip71802 & 5760@xmath795810 & 6640 & 8090@xmath798220 & y & [email protected] & companion : @xmath126 k and @xmath127 + aacam & hip35045 & 5860@xmath795930 & 6790 & 8000@xmath798170 & y & [email protected] & companion : @xmath126 k@xmath128 + veri & hip19004 & 5340 & 6690 & 9880 & y & 3.5 & companion : @xmath129 k@xmath128 + ruma & hip52546 & 5670 & 7120 & 9630 & y & 3.2 & companion : @xmath130 k@xmath128 + ygem & hip37438 & 7010@xmath797830 & 9310@xmath7911,140 & @xmath131 & y & [email protected] & companion : @xmath132 k@xmath133 + @xmath63cet & hip10826 & 5530@xmath796530 & 7550@xmath798160 & @xmath131 & y & [email protected] & symbiotic , wd secondary@xmath134 + + @xmath11uma & hip50801 & 5850 & 6790 & 8200 & y & 37.7 & spectroscopic binary@xmath135 + frcam & hip58545 & 5920@xmath795960 & 7000 & 8200@xmath798270 & y & [email protected] & spectroscopic binary@xmath135 + 4dra & hip60998 & 6350 & @xmath136 & @xmath136 & n & 24.7 & hot compact companion@xmath137 + bycvn & hip62355 & 6500 & 7840 & 9630 & y & 200.1 & spectroscopic binary@xmath135 + ekboo & hip72208 & 6670@xmath797090 & 8130@xmath798160 & 9770@xmath7914,870 & y & [email protected] & x - ray source@xmath138 + rrumi & hip73199 & 6070@xmath796150 & 7340@xmath797360 & 8770@xmath798910 & y & [email protected] & x - ray source@xmath138 + ffboo & hip74253 & 6250@xmath796260 & 7000@xmath797110 & 7780@xmath798090 & y & [email protected] & spectroscopic binary@xmath135 + 42her & hip81497 & 6010@xmath796070 & 6520@xmath796580 & 7053@xmath797420 & y & [email protected] & x - ray source@xmath138 + @xmath139hera & hip84345 & 7280 & 8020 & 8710 & y & 26.1 & @xmath139herb ( g5iii+a9iv - v ) at 4.7@xmath140 + v980her & hip88563 & 5890 & 6670 & 7990 & y & [email protected] & non - symbiotic@xmath135 + hr7547 & hip97372 & 6400@xmath796440 & 7160@xmath797190 & 7930@xmath798070 & y & [email protected] & x - ray source@xmath138 + ptpeg & hip110346 & 6040 & 6900 & 7960 & y & 68.7 & non - symbiotic@xmath135 + + tcet & hip1728 & 6060 & 6570 & 7290 & y & 3.2 & upper limit on uv excess@xmath141 + hd2268 & hip2086 & 6350 & 7080 & 7840 & y & 8.5 & + rreri & hip13384 & 5910@xmath795980 & 6820@xmath796830 & @xmath131 & y & [email protected] & + twhor & hip14930 & 5730@xmath795970 & 6640@xmath796730 & 7710@xmath798840 & y & [email protected] & binary@xmath142 , with no obvious bow shock@xmath143 + sscep & hip17881 & 5870@xmath796070 & 6660 & 7620@xmath797980 & y & [email protected] & + v1139tau & hip19853 & 6380@xmath796440 & ... & ... & n & [email protected] & undetected _ hipparcos _ binary@xmath144 + gzeri & hip20075 & 6560 & 7250 & 7960 & y & 24.3 & + @xmath63ori & hip22667 & 7750 & 8200 & 8600 & y & 37.3 & spec .
binary with wd companion@xmath145 + wxmen & hip26169 & 6480@xmath796520 & 7160@xmath797220 & 7800@xmath798010 & y & [email protected] & unresolved _ hipparcos _ `` problem star''@xmath144 + uori & hip28041 & 8690 & ... & ... & n & 164.9 & jet observed at 43 ghz sio @xmath146 @xmath147 + slep & hip28874 & 5620 & 6630 & 8510 & y & 1.6 & binary candidate@xmath148 + ylyn & hip36288 & 5610@xmath795710 & 6400 & 7430@xmath797640 & y & [email protected] & binary@xmath149 + pxpup & hip38792 & 8530 & ... & ... & n & 49.6 & + zcnc & hip41028 & 5610 & 6630@xmath796670 & 8510@xmath798640 & y & 1.5 & + xcnc & hip43811 & 4980 & 6470 & 14,030 & y & 0.1 & suspected binary@xmath135 + rscnc & hip45058 & 5400 & 6210 & 7420 & y & 0.5 & asymmetric cse@xmath150 + rlmi & hip47886 & 5080@xmath795390 & 6310 & 7790@xmath798500 & y & [email protected] & jet observed at 43 ghz sio @xmath146 @xmath147 + uhya & hip52009 & 5440 & 6290 & 7600 & y & 3.6 & uv emission from a detached shell@xmath151 + vyuma & hip52577 & 5530@xmath795770 & 6490@xmath796540 & 7580@xmath798390 & y & [email protected] & binary@xmath152 + v361vel & hip53940 & 6560 & 7240 & 7940 & y & 18.4 & + zuma & hip58225 & 5730@xmath795770 & 6390 & 7170@xmath797290 & y & [email protected] & + svcrv & hip62611 & 6080 & 7070 & 8220 & y & 13.3 & unresolved _ hipparcos _ `` problem star''@xmath144 + ekboo & hip72208 & 6750@xmath797100 & 8130@xmath798160 & 9720@xmath7912040 & y & [email protected] & double star , separation 0.2@xmath153 + gglib & hip76075 & 6170@xmath796200 & 7050 & @xmath154 & y & [email protected] & + sther & hip77619 & 5710@xmath795760 & 7340@xmath797390 & 8320@xmath798520 & y & [email protected] & unresolved _ hipparcos _ `` problem star''@xmath144 + gher & hip80704 & 5770 & 6400 & 7230 & y & 1.6 & unresolved _ hipparcos _ `` problem star''@xmath144 + 13sge & hip98438 & 7590 & ... & ... & n & 23.8 & double star , separation 23@xmath79113@xmath153 + nupav & hip98608 & 6000 & 7050 & 8270 & y & 4.8 & + v4434sgr & hip99920 & 7080 & 7180 & 7390 & y & 64.8 & double star , separation 30@xmath7946@xmath153 + @xmath155cap & hip102978 & 7020@xmath797030 & ... & ... & n & [email protected] & + wcyg & hip106642 & 7340 & ... & ... & n & 45.6 & _ hipparcos _ suspected double@xmath156 + sgru & hip110736 & 5350@xmath796180 & ... & ... & n & [email protected] & + hrpeg & hip113131 & 6300 & 7020 & 7780 & y & 7.0 & unresolved _ hipparcos _ `` problem star''@xmath144 + 19psc & hip117245 & 5680 & 6450 & 7690 & y & 7.7 & unresolved _ hipparcos _ `` problem star''@xmath144 + one of the objectives of this paper is to define a procedure for selecting potential binary candidates . table [ table - famaey ] lists 12 confirmed binary agb stars , discovered after the monitoring of their radial velocities @xcite .
all of them have been detected in the near - uv and all of them , but hd108907 , have also a far - uv counterpart .
this is especially significant because whilst @xmath60 varies between @xmath157 , the excess in the far - uv band is much larger , @xmath158 .
this means that , as the near - uv flux has relative contributions of the primary and the secondary , the far - uv flux is originated mostly in the secondary .
therefore it can be proposed that the detection of an agb star in the _ galex _ far - uv band is probe of binarity .
a second criterium to select binary agb candidates is the near - uv excess , expressed as @xmath60 .
although disentangling the various effects that contribute for the flux variability ( and as a consequence @xmath60 ) would require a modelization of each individual binary system , it is possible to establish a general statistical criterium for the majority of the objects .
a @xmath122-band amplitude of 3 mag ( a factor 16 in flux ) has been observed in the light curve of several types of variable agb stars @xcite , which suggests that near - uv excesses in the approximate interval @xmath159 can be partially explained by the agb variability .
accounting for this intrinsic agb variability , we propose @xmath111 as an additional criterium for selecting binary agb candidates . the lower limit of this quotient is uncertain , and in any case binarity should be confirmed only after a careful monitoring of the radial velocity .
the assumption of far - uv detection _ and/or _ @xmath65 as a general criterium to select binary agb stars would select the following agb stars of this study as binaries : all the objects studied by @xcite shown in table [ table - sahai ] ; all the agb stars selected from @xcite listed in table [ table - famaey ] ; 34 objects among the 53 listed in table [ table - ita ] .
these are all listed in table [ table - dobles ] , as well as the @xmath123 , @xmath124 and @xmath125 companion temperatures determined with the methods described in sect .
[ temperature ] .
the incidence of binary candidates in a sample limited to the distance of 500 pc can be estimated as follows . as explained in sect .
[ results - nearby ] , among the 58 agb sources surveyed by _
galex _ not known to be symbiotic systems , 5 were not detected in the survey and 53 were detected in the far- and/or near - uv bands . among these , 34 stars have been selected as binary agb candidates .
therefore , the occurrence of candidates is 34/58 or 59% , higher than some previous studies focused on other samples : 8% of spectroscopic binaries in field red giants @xcite and 26% in three open clusters @xcite .
additional work , e.g. based on kinematic data , should be devoted to confirm the candidates selected in the present study , which eventually may decrease the incidence of confirmed agb binary systems .
this work has focused on the search for main - sequence ( ms ) companions of agb stars detected in the galex survey .
this study extends the method originally proposed by @xcite : the uv excess of agb stars is due to the presence of a companion , which can be detected above the emission of the primary if its temperature is @xmath160 6000 k and the flux is within the detection limits of the galex survey .
after analysing the characteristics of the uv emission of a sample of _ bona - fide _ binary agb stars detected with galex , it can be concluded that : 1 . within a distance of 500 pc ,
galex detection limits of ms stars are @xmath1605000 k and @xmath1616500 k for the near - uv and far - uv bands , respectively .
2 . although the uv flux originates mostly from the secondary star , other extrinsic factors might affect its intensity , causing variability , for instance , the intrinsic variability of the primary and the opacity of a common envelope of gas and dust or accretion of material onto the secondary or onto an accretion disk around it .
3 . the galex near - uv excess of confirmed binary agb stars varied within @xmath64 and @xmath162 .
a possible criterium for binarity is : ( 1 ) detection in the far - uv galex band _ and/or _ ( 2 ) @xmath65 .
5 . the uv excess of the candidates to binary agb stars can not be fitted with a single - temperature companion .
the uv emission of the secondary is possibly affected by the extended atmosphere of the primary and/or its circumstellar envelope or it may reveal additional sources of uv emission in accretion processes .
ro thanks the so paulo research foundation ( fapesp ) for the grants # 2010/18835 - 3 and # 2015/00890 - 1 .
mag acknowledges support of the grant aya 2014 - 57280-p , co - funded with feder funds . this research has made use of the simbad database , operated at cds , strasbourg , france .
this research has made use of nasa s astrophysics data system .
the guide star catalogue - ii is a joint project of the space telescope science institute and the osservatorio astronomico di torino .
space telescope science institute is operated by the association of universities for research in astronomy , for the national aeronautics and space administration under contract nas5 - 26555 .
the participation of the osservatorio astronomico di torino is supported by the italian council for research in astronomy .
additional support is provided by european southern observatory , space telescope european coordinating facility , the international gemini project and the european space agency astrophysics division .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
mayer , a. , jorissen , a. , kerschbaum , f. , ottensamer , r. , nowotny , w. , cox , n.l.j . ,
aringer , b. , blommaert , j.a.d.l . ,
decin , l. , van eck , s. , gail , h .-
p . , groenewegen , m.a.t . ,
kornfeld , k. , mecina , m. , posch , th . ,
vandenbussche , b. & waelkens , c. , 2013 , a&a 549 , a69 samus , n.n . ,
durlevich , o.v . ,
goranskij , v.p . ,
kazarovets , e.v . , kireeva , n.n .
, pastukhova , e.n . ,
zharova , a.v . , 2007 - 2015 , _ general catalogue of variable stars _
online edition available at www.sai.msu.su/gcvs/gcvs skrutskie , r.m . ,
cutri , r. stiening , m.d .
weinberg , s. schneider , j.m .
carpenter , c. beichman , r. capps , t. chester , j. elias , j. huchra , j. liebert , c. lonsdale , d.g .
monet , s. price , p. seitzer , t. jarrett , j.d .
kirkpatrick , j. gizis , e. howard , t. evans , j. fowler , l. fullmer , r. hurt , r. light , e.l .
kopan , k.a .
marsh , h.l .
mccallon , r. tam , s. van dyk , and s. wheelock , 2006 , aj , 131 , 1163 | although the majority of known binary asymptotic giant branch ( agb ) stars are symbiotic systems ( i.e. with a wd as a secondary star ) , main - sequence companions of agb stars can be more numerous , even though they are more difficult to find because the primary high luminosity hampers the detection of the companion at visual wavelengths .
however , in the ultraviolet the flux emitted by a secondary with @xmath0 k may prevail over that of the primary , and then it can be used to search for candidates to binary agb stars . in this work ,
theoretical atmosphere models are used to calculate the uv excess in the _ galex _ near- and far - uv bands due to a main - sequence companion .
after analysing a sample of confirmed binary agb stars , we propose as a criterium for binarity : ( 1 ) the detection of the agb star in the _ galex _ far - uv band and/or ( 2 ) a _ galex _ near - uv observed - to - predicted flux ratio @xmath1 .
these criteria have been applied to a volume - limited sample of agb stars within 500 pc of the sun ; 34 out of the sample of 58 agb stars ( @xmath260% ) fulfill them , implying to have a ms companion of spectral type earlier than k0 .
the excess in the _ galex _ near- and far - uv bands can not be attributed to a single temperature companion star , thus suggesting that the uv emission of the secondary might be absorbed by the extended atmosphere and circumstellar envelope of the primary or that uv emission is produced in accretion flows .
[ firstpage ] stars : agb and post - agb binaries : general circumstellar matter ultraviolet : stars |
tunneling is a supreme quantum effect .
every introductory text@xcite on quantum mechanics gives the paradigm example of a particle tunneling through a one - dimensional potential barrier despite having a total energy less than the barrier height .
indeed , the reader typically works through a number of excercises , all involving one - dimensional potential barriers of one form or another modelling several key physical phenomena ranging from atom transfer reactions to the decay of alpha particles@xcite . however , one seldom encounters coupled multidimenisonal tunneling in such texts since an analytical solution of the schrdinger equation in such cases is not possible .
interestingly , the richness and complexity of the tunneling phenomenon manifest themselves in full glory in the case of multidimensional systems@xcite .
thus , for instance , the usual one - dimensional expectation of increasing tunneling splittings as one approaches the barrier top from below is not necessarily true as soon as one couples another bound degree of freedom to the tunneling coordinate . in the context of molecular reaction dynamics ,
multidimensional tunneling can result in strong mode - specificity and fluctuations in the reaction rates@xcite .
in fact , a proper description of tunneling of electrons and hydrogen atoms is absolutely essential@xcite even in molecular systems as large as enzymes and proteins .
although one usually assumes tunneling effects to be significant in molecules involving light atom transfers it is worth pointing out that neglecting the tunneling of even a heavy atom like carbon is the difference between a reaction occuring or not occuring . in particular
, one can underestimate rates by nearly hundred orders of magnitude@xcite .
interestingly , and perhaps paradoxically , several penetrating insights into the nature and mechanism of multidimensional barrier tunneling have been obtained from a phase space perspective@xcite .
the contributions by creagh , shudo and ikeda , and takahashi in the present volume provide a detailed account of the latest advances in the phase space based understanding of multidimensional barrier tunneling . what happens if there are no coordinate space barriers ?
in other words , in situations wherein there are no static energetic barriers separating reactants " from the products " does one still have to be concerned about quantum tunneling ?
one such model potential is shown in fig .
[ fig1 ] which will be discussed in the next section .
here we have the notion of reactants and products in a very general sense .
so , for instance , in the context of a conformational reaction they might correspond to the several near - degenerate conformations of a specific molecule . naively one might expect that tunneling has no consequences in such cases .
however , studies over last several decades@xcite have revealed that things are not so straightforward . despite the lack of static barriers ,
the dynamics of the system can generate barriers and quantum tunneling can occur through such dynamical barriers@xcite .
this , of course , immediately implies that dynamical tunneling is a very rich and subtle phenomenon since the nature and number of barriers can vary appreciably with changes in the nature of the dynamics over the timescales of interest .
this would also seem to imply that deciphering the mechanism of dynamical tunneling is a hopeless task as opposed to the static potential barrier case wherein elegant approximations to the tunneling rate and splittings can be written down .
however , recent studies have clearly established that even in the case of dynamical tunneling it is possible to obtain very accurate approximations to the splittings and rate . in particular , it is now clear that unambiguous identification of the local dynamical barriers is possible only by a detailed study of the structure of the underlying classical phase space .
the general picture that has emerged is that dynamical tunneling connects two or more classically disconnected regions in the phase space . more importantly , and perhaps ironically , the dynamical tunneling splittings and rates are extremely sensitive to the various phase space structures like nonlinear resonances@xcite , chaos@xcite and partial barriers@xcite .
it is crucial to note that although purely quantum approaches can be formulated for obtaining the tunneling splittings , any mechanistic understanding requires a detailed understanding of the phase space topology . in this sense
, the phenomenon of dynamical tunneling gets intimately linked to issues related to phase space transport .
thus , one now has the concept of resonance - assisted tunneling ( rat ) and chaos - assisted tunneling ( cat ) and realistic systems typically involve both the mechanisms .
since the appearance of the first book@xcite on the topic of interest more than a decade ago , there have been several beautiful experimental studies@xcite that have revealed various aspects of the phenomenon of dynamical tunneling .
the most recent one by chaudhury _
_ realizes@xcite the paradigmatic kicked top model using cold @xmath0cs atoms and clearly demonstrate the dynamical tunneling occuring in the underlying phase space .
interestingly , good correspondence between the quantum dynamics and classical phase space structures is found despite the system being in a deep quantum regime . as another example , i mention the experimental observation@xcite by flling _
_ of second order co - tunneling of interacting ultracold rubidium atoms in a double well trap .
the similarities between this system and the studies on dynamical tunneling using molecular effective hamiltonians is striking .
in particular , the description of the cold atom study in terms of superexchange ( qualitative and quantitative ) is reminiscent of the early work by stuchebrukhov and marcus@xcite on understanding the role of dynamical tunneling in the phenomenon of intramolecular energy flow .
further details on the experimental realizations can be found in this volume in the articles by steck and raizen , and hensinger .
an earlier review@xcite provides extensive references to the experimental manifestations of dynamical tunneling in molecular systems in terms of spectroscopic signatures .
undoubtedly , in the coming years , one can expect several other experimental studies which will lead to a deeper understanding of dynamical tunneling and raise many intriguing issues related to the subject of classical - quantum correspondence . as remarked earlier
, it seems ironic that a pure quantum effect like tunneling should bear the marks of the underlying classical phase space structures .
however , it is useful to to recall the statement by heller that tunneling is only meaningful with classical dynamics as the baseline .
thus , insights into the nature of the classical dynamics translates into a deeper mechanistic insight into the corresponding quantum dynamics . indeed
, one way of thinking about classical - quantum correspondence is that classical mechanics is providing us with the best possible basis " to describe the quantum evolution .
the wide range of contributions in this volume are a testimony to the richness of the phenomenon of dynamical tunneling and the utility of such a classical - quantum correspondence perspective . in this article
i focus on the specific field of quantum control and show as to how dynamical tunneling can lead to useful insights into the control mechanism@xcite .
the hope is that more such studies will eventually result in control strategies which are firmly rooted in the intutive classical world , yet accounting for the classically forbidden pathways and mechanisms in a consistent fashion .
this is a tall order , and some may even argue as an unnecessary effort in these days of fast computers and smart and efficient algorithms to solve fairly high dimensional quantum dynamics .
however , in this context , it is useful to remember the following which was written by born , heisenberg , and jordan nearly eighty years ago@xcite _ the starting point of our theoretical approach was the conviction that the difficulties that have been encountered at every step in quantum theory in the last few years could be surmounted only by establishing a mathematical system for the mechanics of atomic and electronic motions , which would have a unity and simplicity comparable with the system of classical mechanics ... further developement of the theory , an important task will lie in the closer investigation of the nature of this correspondence and in the description of the manner in which symbolic quantum geometry goes over into visualizable classical geometry . _
" the above remark was made in an era when computers were nonexistent .
nevertheless , it is remarkably prescient since even in the present era one realizes the sheer difficulty in implementing an all - quantum dynamical study on even relatively small molecules@xcite . in any case , it is not entirely unreasonable to argue that large scale quantum dynamical studies will still require some form of an implicit classical - quantum correspondence approach to grasp the underlying mechanistic details . with the above remark in mind
i start things off by revisiting the original paper@xcite by davis and heller since , in my opinion , it is ideal from the pedagogical point of view .
three decades ago , davis and heller in their pioneering study@xcite gave a clear example of dynamical tunneling .
a short recount of this work including the famous plots of the classical trajectories and the associated quantum eigenstates can be found in heller s article in this volume .
however , i revisit this model here in order to bring forth a couple of important points that seem to have been overlooked in subsequent works .
first , the existence of another class of eigenstate pairs , called as circulating states@xcite , which can exert considerable influence on the usual tunneling doublets at higher energies .
second , a remark in the original paper@xcite which can be considered as a harbinger for chaos - assisted tunneling .
as shown below , there are features in the original model that are worth studying in some detail even after three decades since the original paper was published .
the hamiltonian of choice is the two degrees of freedom ( 2dof , in what follows the acronymn dof stands for degrees of freedom ) barbanis - like model@xcite @xmath1 with the labels ` s ' and ` u ' denoting the symmetric and unsymmetric stretch modes respectively .
the above hamiltonian has also been studied@xcite in great detail to uncover the correspondence between classical stability of the motion and quantum spectral features , wavefunctions , and energy transfer .
the potential is symmetric with respect to @xmath2 as shown in fig .
[ fig1 ] but there is no potential barrier .
davis and heller used the parameter values @xmath3 , and @xmath4 for which the dissociation energy @xmath5 . note that the masses are taken to be unity and one is working in units such that @xmath6 .
the key observation by davis and heller was that despite the lack of any potential barriers several bound eigenstates came in symmetric - antisymmetric pairs @xmath7 and @xmath8 with energy splittings much smaller than the fundamental frequencies _
i.e. , _ @xmath9 . in fig .
[ fig1 ] the various splittings between adjacent eigenstates are shown and it is clear that several tunneling " pairs appear above a certain threshold energy .
poincar surface of sections .
note that large scale chaos appears for @xmath10 .
the formation of a separatrix and two classically disjoint regular regions can be seen at @xmath11 due to the symmetric mode periodic orbit becoming unstable .
the regular regions almost vanish near the dissociation energy . in the bottom row
some of the chaotic orbits have been suppressed , for clarity of the figure , by showing them in gray.,width=566,height=377 ] how can one understand the onset of such near degeneracies in the system ?
the crucial insight that davis and heller provided was that the appearance of such doublets is correlated with the large scale changes in the classical phase space .
the nature of the phase space with increasing total energy is shown in fig .
[ fig2 ] using the @xmath12 poincar surface of section .
such a surface of section , following standard methods , is constructed by recording the points , with momentum @xmath13 , of the intersection of a trajectory with the the @xmath14 plane in the phase space .
such a procedure for several trajectories with specific total energy generates a typical surface of section as shown in fig .
[ fig2 ] and clearly indicates the global nature of the phase space .
it is clear from the figure that the phase space for @xmath15 is mostly regular while higher energy phase spaces exhibit mixed regular - chaotic dynamics .
one of the most prominent change happens when the symmetric stretch periodic orbit @xmath16 becomes unstable around @xmath17 . in fig .
[ fig2 ] the consequence of such a bifurcation can be clearly seen for @xmath11 as the formation of two classically disjoint regular regions .
in fact , the two regular regions signal a @xmath18:@xmath18 resonance between the stretching modes .
the crucial point to observe here is that classical trajectories initiated in one of the regular regions can not evolve into the other regular region . with increasing energy
the classically disjoint regular regions move further apart and almost vanish near the dissociation energy .
the result presented in fig .
[ fig1 ] in fact closely mirrors the topological changes , shown in fig .
[ fig2 ] , in the phase space . thus , in fig .
[ fig1 ] a sequence of eigenstates with very small splittings begins right around the energy at which the symmetric periodic orbit becomes unstable .
the unsymmetric stretch , however , becomes unstable at a higher energy @xmath19 and one can observe in fig .
[ fig1 ] another sequence that seemingly begins near this point .
note that at higher energies it is not easy to identify any more sequences , but small splittings are still observed .
the important thing to note here is that not only do the splittings but the individual eigenstates also correlate with the changes in the phase space . in fig .
[ fig3 ] we show a set of four eigenstates to illustrate an important point - as soon as the @xmath18:@xmath18 resonance manifests itself in the phase space , the tunneling doublets start to form and an integrable normal form approximation is insufficient to account for the splittings .
note that the phase space is mostly regular at the energies of interest .
a much more detailed analysis can be found in the paper by farrelly and uzer@xcite .
we begin by noting@xcite that the pair of eigenstates ( counting from the zero - point ) @xmath20 and @xmath21 are split by about @xmath22 whereas the pair @xmath23 and @xmath24 are separated by about @xmath25 . in both cases the splitting is smaller than the fundamental frequencies , which are of order unity .
note that the latter pair of states appears to be a part of the sequence in fig .
[ fig1 ] that starts right after the bifurcation of the symmetric stretch periodic orbit .
since the original mode frequencies are nonresonant , one can obtain a normal form approximation@xcite to the original hamiltonian and see if the obtained splittings can be explained satisfactorily .
in other words , the observed @xmath26 are coming from the perturbation that couples both the modes and in such a case there is no reason for classifying them as tunneling doublets .
however , from the coordinate space representations of the states shown in fig .
[ fig3 ] one observes that there is an important difference between the two pairs of states .
the pair @xmath27 seems to have a perturbed nodal structure and hence one can approximately assign the states using the zeroth - order quantum numbers @xmath28 . inspecting the figure leads to the assignment @xmath29 and @xmath30 for states @xmath20 and @xmath21 respectively .
if the above arguments are correct then the splitting @xmath31 should be obtainable from the normal form hamiltonian . since the theory of normal forms
is explained in detail in several textbooks@xcite , i will provide a brief derivation below .
begin by using the unperturbed harmonic action - angle variables @xmath32 defined via the canonical transformation ( similar set for @xmath33 mode ) : @xmath34 to express the original hamiltonian as @xmath35 where @xmath36 .
since @xmath37 is purely oscillatory the angle average @xmath38 and thus to @xmath39 the normal form hamiltonian can be identified with the zeroth - order @xmath40 above .
the first nontrivial correction arises at @xmath41 and can be obtained using the @xmath39 generating function @xmath42 where @xmath43 are the coefficients in the fourier expansion of the oscillatory part of @xmath44 .
one now obtains the @xmath41 correction to the hamiltonian as @xmath45 with the bar denoting angle averaging of the poisson bracket involving @xmath46 and @xmath37 . performing the calculations the normal form at @xmath41
is obtained as @xmath47 with @xmath48\ ] ] the primitive bohr - sommerfeld quantization @xmath49 yields the quantum eigenvalues perturbatively to @xmath41 .
the procedure can be repeated to obtain the normal form hamiltonian at higher orders .
for instance , farrelly and uzer have@xcite computed the normal form out to @xmath50 and used pad resummation techniques to improve in cases when the zeroth - order frequencies are near - resonant . for our qualitative discussions ,
the @xmath41 normal form is sufficient . : @xmath18 resonance is just starting to appear .
the upper panel shows the @xmath51 coordinate space representations and the lower panels show the corresponding husimi distributions in the @xmath12 surface of section of the phase space .
the first two states ( @xmath52 ) can be assigned approximate zeroth - order quantum numbers @xmath28 whereas the last two states ( @xmath53 ) show perturbed nodal features .
clear difference in the phase space nature of the eigenstates can be seen .
see text for details.,width=604,height=453 ] ) of the davis - heller system involving three states . as in the previous figure , the @xmath51 and phase space husimi representations
are also shown .
the first and the last states appear to be strongly mixed states .
the middle state seems relatively cleaner .
does chaos play a role in this case ?
detailed discussions in the text.,width=566,height=377 ] notice that the normal form hamiltonian above is integrable since it is ignorable in the angle variables . indeed , using the normal form and the approximate assignments of the eigenstates shown in fig . [ fig3 ] one finds @xmath54 which is in fair agreement with the actual numerical value .
thus , a large part of the splitting @xmath31 can be explained classically . on the other hand , although the states @xmath23 and @xmath24 seem to have an identifiable nodal structure ( cf .
[ fig3 ] ) , it is clear that they are significantly perturbed .
nevertheless , persisting with an approach based on counting the nodes , states @xmath23 and @xmath24 can be assigned as @xmath55 and @xmath56 respectively .
using the normal form one estimates @xmath57 and this is about a factor of two larger than the numerically computed value .
thus , the splitting in this case is not accounted for solely by classical considerations .
one might argue that a higher order normal form might lead to better agreement , but the phase space husimi distributions of the eigenstates shown in fig .
[ fig3 ] suggests otherwise .
the husimi disributions , when compared to the classical phase spaces shown in fig .
[ fig2 ] , clearly show that the pair @xmath58 are localized in the newly created @xmath18:@xmath18 resonance zone .
hence , this pair of states is directly influenced by the nonlinear resonance and the splitting between them can not be accurately described by the normal form hamiltonian .
indeed , as discussed by farrelly and uzer@xcite , in this instance one needs to consider a resonant hamiltonian which explicitly takes the @xmath18:@xmath18 resonance into account .
this provides a clear link between dynamical tunneling , appearance of closely spaced doublets and creation of new , in this case a nonlinear resonance , phase space structures . the choice of states in fig .
[ fig3 ] is different from what is usually shown as the standard example for dynamical tunneling pairs .
however , in the discussion above the states were chosen intentionally with the purpose of illustrating the onset of near - degeneracy due to the formation of a nonlinear resonance . in a typical situation involving near - integrable phase spaces
there are several such resonances ranging from low orders to fairly high orders .
the importance of a specific resonance depends sensitively on the effective value of the planck s constant@xcite .
indeed , detailed studies@xcite have shown that excellent agreement with numerically computed splittings can be obtained if proper care is taken to include the various resonances .
clear and striking examples in this context can be found in the contributions by schlagheck _
et al . _ and bcker _ et al . _
in this volume .
to finish the discussion of the davis - heller system i show an example which involves states forming tunneling pairs that are fairly complicated both in terms of their coordinate space representations as well as their phase space husimi distributions .
the example involves three states @xmath59 , @xmath60 , and @xmath61 around @xmath62 , which is rather close to the dissociation energy . in the original work@xcite davis and
heller noted that the splittings seem to increase by an order of magnitude in this high energy region .
notably , they commented that `` _ perhaps the degree of irregularity between the regular regions plays some part _ '' . in fig .
[ fig4 ] the variation of energy levels with the coupling parameter @xmath63 is shown .
one can immediately see that the three states are right at the center of an avoided crossing .
the coordinate space representations of the states shows extensive mixing for states @xmath59 and @xmath61 while the state @xmath60 seems to be cleaner .
the husimi distributions for the respective states conveys the same message .
comparing to the phase space sections shown in fig .
[ fig2 ] it is clear that the husimis for state @xmath59 and @xmath61 seem to be ignoring the classical regular - chaotic divison - a clear indication of the quantum nature of the mixing . although it is not possible to strictly assign these states as chaotic , a closer inspection does show substantial husimi contribution in the border between the regular and chaotic regions .
interestingly , the pairwise splitting between the states is nearly the same but localized linear combinations of any two states exhibits two - level dynamics .
thus , the situation here is not of the generic@xcite chaos - assisted tunneling one wherein one of the states is chaotic and interacts with the other two regular states .
nevertheless , linear combinations of the three states reveal ( not shown here ) that they are mixed with each other .
a look at the coordinate and phase space representations of states @xmath59 and @xmath61 in fig .
[ fig4 ] reveals that a different kind of state is causing the three - way interaction .
it turns out that in the davis - heller model there is another class of symmetry - related pairs that appear when the unsymmetric mode becomes unstable . in the original work they were refered to as circulating " states which displayed much larger splitting then the so - called local mode pairs .
the case shown in fig .
[ fig4 ] involves one of the circulating pairs interacting with the usual local mode doublet leading to the complicated three - way interaction .
the subtle nature of this interaction is evident from the coordinate space representation of state @xmath60 which exhibits a broken symmetry . interestingly , and as far as i can tell ,
there has been very little understanding of such three - state interactions in the davis - heller system and further studies are needed to shed some light on the phase space nature of the relevant eigenstates .
in the previous section the intimate connection between dynamical tunneling and phase space structures was introduced .
the importance of a nonlinear resonance and a hint of the role played by the chaos ( cf .
[ fig4 ] ) is evident .
several other contributions in this volume discuss the importance of various phase space strucures using different models , both continuous hamiltonians and discrete maps . in the molecular context ,
various mode - mode resonances play a critical role in the process of intramolecular vibrational energy redistribution ( ivr)@xcite . the phenomenon of ivr is at the heart of chemical reaction dynamics and it is now well established that molecules at high levels of excitation display all the richness , complexity and subtelty that is expected from nonlinear dynamics of multidimensional systems@xcite . in this
context , dynamical tunneling is an important agent@xcite of ivr and state mixing for a certain class of initial states ( akin to the so called noon states@xcite ) which are typically prepared by the experiments . the importance of the anharmonic resonances to ivr and the regimes wherein dynamical tunneling , mediated by these resonances , is expected to be crucial is described in some detail in this volume by leitner .
the phase space perspective on leitner s viewpoint can be found in a recent review@xcite ( see also heller s contribution in the present volume ) . in this regard
it is interesting to note that there has been a renaissance of sorts in chemical dynamics with researchers critically examining the validity of the two pillars of reaction rate theory@xcite - transition state theory ( tst ) and the rice - ramsperger - kassel - marcus ( rrkm ) theory .
since both theories have classical dynamics at their foundation , advances in our understanding of nonlinear dynamics and continuing efforts to characterize the phase space structure of systems with three or more degrees of freedom are beginning to yield crucial mechanistic insights into the dynamics@xcite . at the same time
, rapid advances in experimental techniques and theoretical understanding of the reaction mechanisms has led researchers to focus on the issue of controlling the dynamics of molecules .
what implications might dynamical tunneling have on our efforts to control the atomic and molecular dynamics ? in the rest of this article
i focus on this issue and use two seemingly simple and well studied systems as examples to highlight the role of dynamical tunneling in the context of coherent control .
both examples are in the context of periodically driven systems and i refer the reader to the work of flatt and holthaus@xcite for an exposition of the close quantum - classical correspondence in such systems . historically , an early indication that dynamical tunneling could be sensitive to the chaos in the underlying phase space came from the study of strongly driven double well potential by lin and ballentine@xcite .
the model hamiltonian in this case can be written down as @xmath64 with @xmath65 being the frequency of the monochromatic field ( henceforth refered to as the driving field ) and the unperturbed part @xmath66 is the hamiltonian corresponding to a double well potential with two symmetric minima at @xmath67 and a maximum at @xmath68 .
following the original work@xcite , the parameters of the unperturbed system ( assuming atomic units ) are taken to be @xmath69 , @xmath70 , and @xmath71 for which the potential has a barrier height @xmath72 and supports about eight tunneling doublets .
as is well known , in the absence of the driving field , a wavepacket prepared in the left well can coherently tunnel into the right well with the time scale for tunneling being inversely proportional to the tunnel splitting .
this unperturbed scenario , however , is significantly altered in the presence of a strong driving field . in the presence of a strong field
the phase space of the system exhibits large scale chaos coexisting with two symmetry - related regular regions .
lin and ballentine observed that a coherent state localized in one of the regular region tunnels to the other symmetry - related regular region on timescales which are orders of magnitude smaller than in the unperturbed case .
it was suspected that the extensive chaos in the system might be assisting the tunneling process . .
a coherent state localized in the left regular island tunnels ( indicated by a green arrow ) over to the right regular island . the chaotic regions ( gray )
have been suppressed for clarity .
( b ) monitoring the survival probability of the initial state to determine the timescale of tunneling ( @xmath73 @xmath74 in this case ) .
snapshots of the evolving husimi distributions are also shown at specific intervals , with green ( gray ) indicating maxima of the distributions .
( c ) the decay time determined as in ( b ) for a variety of field strengths with the initial state localized in the left island .
note the fluctuations in the decay time over several orders of magnitude despite very similar nature of the phase spaces over the entire range of the driving field strength .
the arrows highlight some of the plateau regions which are crucial for bichromatic control.,width=604,height=529 ] in order to illustrate the tunneling process fig .
[ fig5]a shows the stroboscopic surface of section for the case of strong driving with @xmath75 .
note the extensive chaos and the two regular islands ( left and right ) in the phase space .
a coherent state @xmath76 is placed in the center of the left island and time evolved using the floquet approach which is ideally suited for time - periodic driven systems . in this instance one is interested in the time at which the coherent state localized on the left tunnels over to the regular region on the right . in order to obtain this information
it is necessary to compute the survival probability of the initial coherent state .
briefly , floquet states \{@xmath77 } are eigenstates of the hermitian operator @xmath78 and form a complete orthonormal basis .
an arbitrary time - evolved state @xmath79 can be expressed as @xmath80 with @xmath81 being the quasienergy associated with the floquet state @xmath77 .
the expansion coefficients @xmath82 are independent of time and given by @xmath83 yielding the expansion @xmath84 measuring time in units of field period ( @xmath85 ) and owing to the periodicity of the floquet states , @xmath86 , the above equation simplifies to @xmath87 with @xmath88 and integer @xmath89 . the time evolution operator @xmath90 is determined by successive application of the one - period time evolution operator @xmath91 _ i.e. _ , @xmath92^k.\ ] ] this allows us to express the survival probability of the initial coherent state in terms of floquet states as @xmath93 where the overlap intensities are denoted by @xmath94 . in order to determine the `` lifetime '' ( @xmath95 ) of the initial coherent state
, we monitor the time at which @xmath96 goes to zero ( minimum ) at the first instance .
in other words , it is the time when the coherent state leaves its initial position _
i.e _ , the left regular island of the phase space for the first time . in fig .
[ fig5]b the @xmath96 for the initial state of interest is shown along with the snapshots of husimi distribution at specific times . clearly , the initial state tunnels in about @xmath97 , which suggests that chaos assists the dynamical tunneling process . however , the issue is subtle and highlighted in fig .
[ fig5]c which shows the decay time plot for a range of driving field strengths for an intial state localized in the left regular island .
it is important to note that the gross features of the phase space are quite similar over the entire range .
however , fig .
[ fig5]c exhibits strong fluctuations over several orders of magnitude and this implies that a direct association of the decay time with the extent of chaos in the phase space is not entirely correct .
( black ) and for @xmath98 ( blue ) and @xmath99 ( red ) as a function of @xmath100 .
the local phase structure in the vicinity of the initial coherent state and the survival probability for fixed @xmath101 ( indicated by green line in ( a ) ) corresponding to red shifted , fundamental and blue shifted field frequency is shown in panels ( b ) and ( c ) respectively .
the @xmath102:@xmath18 field - matter resonance in case of @xmath99 is clearly visible and correlates with the very short decay time as compared to the other two cases .
note the different time axis scale in the first survival probability plot.,width=529,height=340 ] the above discussion and results summarized in fig .
[ fig5 ] bring up the following key question .
what is the mechanism by which the initial state @xmath103 decays out of the regular region ? in turn , this is precisely the question that modern theories of dynamical tunneling strive to answer . according to the theory of resonance - assisted tunneling@xcite , the mechanism is possibily one wherein @xmath103 couples to the chaotic sea via one or several nonlinear resonances provided certain conditions are satisfied .
specifically , the local structure of the phase space surrounding the regular region is expected to play a critical role .
the theoretical underpinnings of rat along with several illuminating examples can be found in the contribution by schlagheck _
and here we suggest a simple numerical example which points to the importance of the local phase space structure around @xmath103 . preliminary evidence for the role of field - matter nonlinear resonances in controlling the decay of @xmath103 is given in fig .
[ fig6 ] , which shows the effect of changing the driving field frequency @xmath65 on the local phase space structures and the decay times . from fig .
[ fig6 ] it is apparent that detuning @xmath65 by @xmath104 leads to a significant change in the local phase space structure and the decay time of the initially localized state . in fig .
[ fig6](b ) , corresponding to the field frequency @xmath105 , a prominent @xmath102:@xmath18 field - matter resonance is observed .
it is plausible that the the decay time is only a few hundred field periods in this case due to assistance from the nonlinear resonance .
however , the decay time increases for @xmath106 and becomes even larger by an order of magnitude for @xmath107 , due to absence of the @xmath102:@xmath18 resonance .
this indicates that decay dynamics is highly sensitive to the changes in local phase space structure of the left regular island .
therefore , taking into account the relatively large order resonances , since @xmath6 , is unavoidable in order to understand the decay time plot in fig .
[ fig5 ] and for smaller values of the effective planck constant one expects a more complicated behavior .
note , however that the very high order island chain visible in the last case in fig .
[ fig6 ] is unable to assist the decay .
this is where we believe that an extensive @xmath108-scaling study will help in gaining a deeper understanding of the decay mechanism .
such an extensive calculation can be found , for example , in the recent work@xcite by mouchet , eltschka , and schlagheck on the driven pendulum .
there is sufficient evidence@xcite in the driven pendulum system for a mechanism in which nonlinear resonances play a central role in coupling initial states localized in regular phase space regions to the chaotic sea .
one might be able to provide a clear qualitative and quantitative explanation for the results in fig .
[ fig5]c based on the recent advances . and primary field strength ( a ) @xmath109 and ( b ) @xmath110 . as in the earlier plots ,
the chaotic regions have been suppressed ( in gray ) for clarity .
note the breaking of the symmetry in both cases .
however , significant suppression of the deacy of the state localized in the left regular island happens in case ( b ) only.,width=566,height=340 ] this brings us to the second important issue - what is the role of chaos in this dynamical tunneling process ?
an earlier work@xcite by utermann , dittrich and hnggi on the driven double well system showed that there is indeed a strong correlation between the splittings of the floquet states and the overlaps of their husimi distributions with the chaotic regions in the phase space .
however , a different perspective yields clear insights into the role of chaos with potential implications for coherent control . in order to highlight this perspective i start with a simple question : _ is it possible to control the decay of the localized initial state with an appropriate choice of a control field ? _ it is important to note that the lin - ballentine system parameters imply that one is dealing with a multilevel control scenario and there has been a lot of activity over the last few years to formulate control schemes involving multiple levels in both atomic and molecular systems . the driven double well system has been a particular favorite in this regard , more so in recent times due to increased focus on the physics of trapped bose - einstein condensates@xcite .
more specifically , several studies have explored the possibility of controlling various atomic and molecular phenomenon using bichromatic fields with the relative phase between the fields providing an additional control parameter . in the present context ,
for example , sangouard _ et al .
_ exploited the physics of adiabatic passage to show that an appropriate combination of @xmath111 field leads to suppression of tunneling in the driven double well model@xcite .
the choice of relative phase between the two fields allowed them to localize the initial state in one or the other well .
however , the parameter regimes in the work by sangouard _ et al . _
correspond to the underlying phase space being near - integrable and hence a minimal role of the chaotic sea .
more relevant to the mixed regular - chaotic phase space case presented in fig .
[ fig5 ] is an earlier work@xcite by farrelly and milligan wherein it was demonstrated that one can suppress the tunneling dynamics in a driven double well system using a @xmath111 bichromatic field . in other words
the original hamiltonian of eq .
[ linbalham ] is modified as following @xmath112 with @xmath113 being the same as in eq .
[ lbh0ham ] and the additional @xmath114-field is taken to be the control field . moreover , modulating the turn - on time of the control field can trap the wavepacket in the left or right well of the double well potential .
hence , it was argued@xcite that the tunneling dynamics in a driven double well can be controlled at will for specific choices of the control field parameters @xmath115 .
note that in the presence of control field _
i.e. , _ @xmath116 with the relative phase @xmath117 the hamiltonian in eq .
[ farmilham ] transforms under symmetry operations as @xmath118 \nonumber \\ &
= & h_{0}(x , p)-x[-\lambda_{1}\cos(\omega_{f}t)+\lambda_{2}\cos(2\omega_{f}t ) ] \\ & \neq & h(x , p;t ) .
\nonumber\end{aligned}\ ] ] similarly , except at @xmath119 , the discrete symmetry of the hamiltonian is broken under the influence of the additional @xmath114-field .
farrelly and milligan thus argued@xcite that the control field with strength smaller than the driving field will lead to localization due to the breaking of the generalized symmetry of the hamiltonian and the floquet states .
the impact of a small symmetry breaking control field with @xmath120 can be clearly seen in fig .
[ fig7 ] in terms of the changes in the classical phase space structures . however , there is a subtlety which is not obvious upon inspecting the phase spaces shown in fig .
[ fig7 ] for two different but close values of the driving field strength . in case of @xmath109 corresponding to fig .
[ fig7](a ) , the control field is unable to suppress the decay of the initial state @xmath103 localized in the left regular region .
on the other hand , fig .
[ fig7](b ) corresponds to @xmath75 and computations show that the control field is able to suppress the decay of @xmath103 to an appreciable extent .
thus , although in both cases the control field breaks the symmetry and the resulting phase spaces show very similar structures , the extent of control exerted by the @xmath114-field is drastically different . as a function of the field parameters @xmath121 .
the initial state in every case is localized in the left regular region in the classical phase space .
notice the convoluted form of the landscape with the regions of low probability indicating little to no control ( red thick arrows ) .
the green lines ( thin arrows ) indicate regions where a high degree of control can be achieved.,width=604,height=377 ] the discussions above and the results summarized in fig .
[ fig7 ] and fig .
[ fig6 ] clearly indicate that the decay of @xmath103 out of the left regular region can be very different and deserves to be understood in greater detail .
in fact , computations show that such cases of complete lack of control are present for other values of @xmath100 as well .
this is confirmed by inspecting fig .
[ fig8 ] which shows the control landscape for the bichromatically driven double well in the specific case of @xmath117 .
other choices for @xmath122 also show similarly convoluted landscapes .
there are several ways of presenting a control landscape and in fig .
[ fig8 ] the time - smoothed survival probability ( cf .
eq . [ survprob ] ) @xmath123 associated with the initial state @xmath103 is used to map the landscape as a function of the field strengths @xmath121 . note that the choice of @xmath124 to represent the landscape
is made for convenience ; the decay time is a better choice which requires considerable effort but the gross qualitative features of the control landscape do not change upon using @xmath124 .
large ( small ) values of @xmath124 indicate that the decay dynamics is suppressed ( enhanced ) .
it is clear from fig .
[ fig8 ] that the landscape comprises of regions of control interspersed with regions exhibiting lack of control
. such highly convoluted features of control landscape are a consequence of the simple bichromatic choice for the control field and the nonlinear nature of the corresponding classical dynamics . from a control point of view , there are regions on the landscape for which a monotonic increase of @xmath125 leads to increasing control .
interestingly , the lone example of control illustrated in farrelly and milligan s paper@xcite happens to be located on one of the prominent hills on the control landscape ( shown as a green dot in fig .
[ fig8 ] ) . a striking feature that can be seen in fig .
[ fig8 ] is the deep valley around @xmath109 which signals an almost complete lack of control even for significantly large strengths of the @xmath114-field .
the valleys in fig .
[ fig8 ] correspond precisely to the plateaus seen in the decay time plot shown in fig .
[ fig6 ] for @xmath126 ( red arrows ) .
it is crucial to note that this wall of no control " is robust even upon varying the relative phase @xmath122 between the driving and the control field . .
the primary driving field strength is fixed at @xmath127 .
six states that have appreciable overlap with @xmath103 are highlighted by circles .
( b ) overlap intensity @xmath128 for @xmath109 and @xmath129 indicates multilevel interactions involving the states shown in ( a ) .
husimi distribution function of the floquet states regulating the decay of @xmath103 are also shown .
notice that the nature of delocalized states ( magenta ) does not change much with @xmath125.,width=529,height=453 ] insights into the lack of control for driving field strength @xmath109 ( and other values as well which are not discussed here ) can be obtained by studying the variation of the floquet quasienergies with @xmath125 , the control field strength . in fig .
[ fig9 ] the results of such a computation are shown and it is immediately clear that even in the presence of @xmath114-control field six states contribute to the decay of @xmath103 over the entire range of @xmath125 .
this is further confirmed in fig .
[ fig9](b ) where the plot of the overlap intensities shows multiple floquet states participating nearly equally in the decay dynamics of the initial state .
the final clue comes from inspecting the husimi distributions shown in fig .
[ fig9 ] , highlighting the phase space delocalized nature of some of the participating states . despite the symmetry of the tunneling doublets
being broken due to the bichromatic field , two or more of the participating states are extensively delocalized in the chaotic reigons of the phase space . moreover
, the participation of the chaotic floquet states persists even for larger values of @xmath125 .
hence , using the symmetry breaking property of the @xmath114-field for control purposes is not very effective when chaotic states are participating in the dynamics .
_ therefore , the lack of control , signaled by plateaus in fig .
[ fig6 ] and the valleys in fig .
[ fig8 ] , is due to the dominant participation by chaotic states _ i.e. , _ chaos - assisted tunneling . _
the plateaus arise due to the fact that the coupling between the localized states and the delocalized states vary very little with increasing control field strength @xmath125 - something that is evident from the floquet level motions shown in fig .
[ fig9 ] and established earlier by tomsovic and ullmo in their seminal work on chaos - assisted tunneling in coupled quartic oscillators@xcite .
it is important to note that for @xmath126 the chaotic states , as opposed to the regular states , do not have a definite parity .
consequently , the presence of the @xmath114-field does not have a major influence on the chaotic states .
thus , if one or more chaotic states are already influencing the dynamics of @xmath103 at @xmath126 then the bichromatic control is expected to be difficult .
an earlier study@xcite by latka _
_ on the bichromatically driven pendulum system also suggested that the ability to control the dynamics is strongly linked to the existence of chaotic states . the model problem in this section and the results point to a direct role of chaos - assisted pathways in the failure of an attempt to bichromatically control the dynamics .
however , it is not yet clear if control strategies involving more general fields would exhibit similar characteristics .
there is some evidence in the literature which indicates that quantum optimal control landscapes might be highly convoluted if the underlying classical phase space exhibits large scale chaos@xcite .
nevertheless , further studies need to be done and the resulting insights are expected to be crucial in any effort to control the dynamics of multilevel systems . the driven morse oscillator system has served as a paradigm model for understanding the dissociation dynamics of diatomic molecules .
studies spanning nearly three decades have explored the physics of this system in exquisite detail .
consequently , a great deal is known about the mechanism of dissociation both from the quantum and classical perspectives .
indeed , the focus of researchers nowadays is to control , either suppress or enhance , the dissociation dynamics and various suggestions have been put forward . in addition , one hopes that the ability to control a single vibrational mode dynamics can lead to a better understanding of the complications that arise in the case of polyatomic molecular systems wherein several vibrational modes are coupled at the energies of interest .
several important insights have originated from classical - quantum correspondence studies which have established that molecular dissociation , in analogy to multiphoton ionization of atoms , occurs due to the system gaining energy by diffusing through the chaotic regions of the phase space .
for example , an important experimental study by dietrich and corkum has shown@xcite , amongst other things , the validity of the chaotic dissociation mechanism .
thus , the formation of the chaotic regions due to the overlap@xcite of nonlinear resonances ( field - matter ) , hierarchical structures@xcite near the regular - chaotic borders acting as partial barriers , and their effects on quantum transport@xcite have been studied in a series of elegant papers@xcite . in the context of this present article
an interesting question is as follows .
since a detailed mechanistic understanding of the role of various phase space structures of the driven morse system is known , is it possible to design local phase space barriers to effect control over the dissociation dynamics ?
in particular , the central question here is whether the local phase space barriers are also able to suppress the quantum dissociation dynamics .
there is an obvious connection between the above question and the theme of this volume - quantum mechanics can shortcircuit " the classical phase space barriers due to the phenomenon of dynamical tunneling .
thus , such phase space barriers might be very effective in controlling the classical dissociation dynamics but might fail completely when it comes to controlling the quantum dissociation dynamics .
there is a catch here , however , since there is also the possibility that the cantori barrier in the classical phase space may be even more restrictive in the quantum case .
thus , creation or existence of a phase space barrier invariably leads to subtle competition between classical transport and quantum dynamical tunneling through the barrier . as expected
, the delicate balance between the classical and quantum mechanism is determined by the effective planck constant of the system of interest .
an earlier detailed review@xcite by radons , geisel and rubner is highly reccomended for a nice introduction to the subject of classical - quantum correspondence perspective on phase space transport through kolmogoroff - arnold - moser ( kam ) and cantori barriers . in the driven morse oscillator case , brown and wyatt showed@xcite that the cantori barriers do leave their imprint on the quantum dissociation dynamics and act as even stronger barriers as compared to the classical system . maitra and heller in their study@xcite on transport through cantori in the whisker map have clearly highlighted the classical versus quantum competition . and
frequency @xmath130 connects state @xmath131 and @xmath132 via a two - photon resonance transition .
the location of cantori in presence of the field with @xmath130 are indicated as dotted curves in the figure.,width=529,height=415 ] from the above discussion it is apparent that any approach to control the dissociation dynamics by recreating local phase space barriers will face the subtle classical - quantum competition .
in fact , it is tempting to think that every quantum control algorithm works by creating local phase space dynamical barriers and the efficiency of the control is decided by the classical - quantum competition .
however , at this point of time there is very little work towards making such a connection and the above statement is , at best , a conjecture . for the purpose of this article
, i turn to the driven morse oscillator system to provide an example for the importance of resonance - assisted tunneling in controlling the dissociation dynamics .
the model system is inspired from the early work@xcite by wyatt and brown ( see also the work by breuer and holthaus@xcite ) and the hamiltonian can be written as @xmath133 in the dipole approximation .
the zeroth - order hamiltonian @xmath134}^2 , \label{morsepot}\ ] ] represents the morse oscillator modeling the anharmonic vibrations of a diatomic molecule . in the above
, @xmath135 is the dipole moment function , @xmath100 is the strength of the laser field , @xmath65 is the driving field frequency , and @xmath136 is the reduced mass of the diatomic molecule with @xmath137 and @xmath138 being the two atomic masses . in eq .
[ morsepot ] , @xmath139 is the dissociation energy , @xmath140 is the range of the potential and @xmath141 is the equilibrium bond length of the molecule . rather than attempting to provide a general account as to how rat might interfere with the process of control ,
i feel that it is best to illustrate with a realistic molecular example .
hopefully , the generality of the arguments will become apparent later on .
for the present purpose i choose the diatomic molecule hydrogen fluoride ( hf ) as the specific example .
any diatomic molecule could have been chosen but hf is studied here due to the fact that brown and wyatt have already discussed the role of cantori barriers to the dissociation dynamics in some detail .
the morse oscillator parameters for hydrogen fluoride are @xmath142 and @xmath143 .
these parameters correspond to ground electronic state of the hf molecule supporting @xmath144 bound states .
note that atomic units are used for both the molecular and field parameters with time being measured in units of the field period @xmath145 .
the only difference between the present work and that of brown and wyatt has to do with the field - matter coupling .
brown and wyatt use the dipole function @xmath146 with @xmath147 and @xmath148 , obtained from _ ab - initio _ data on hf . here a linear approximation for @xmath135 @xmath149 is employed with @xmath150 in case of hf .
there are quantitative differences in the dissociation probabilities due to the linearization approximation but the main qualitative features remain intact despite the linearization approximation . in fig .
[ fig10 ] the morse potential for hf is shown along with a summary of the key features such as the energy region of interest , the quantum state whose dissociation is to be controlled , and the classical phase space structures that might play an important role in the dissociation dynamics .
the driving field parameters are chosen as @xmath151 , same as in the earlier work@xcite , and the field strength corresponds to about @xmath152 tw/@xmath153 w/@xmath154 . as shown in fig .
[ fig10 ] ( inset ) , the focus is on understanding and controlling the dissociation dynamics of the @xmath131 excited morse oscillator eigenstate .
there are several reasons for such a choice and i mention two of the most important reasons .
firstly , the earlier study@xcite has established that @xmath131 of hf happens to be in an energy regime wherein two specific cantori barriers in the classical phase space affect the dissociation dynamics . moreover , for driving frequency @xmath130 , two of the zeroth - order eigenstates @xmath131 and @xmath132 have unperturbed energies such that @xmath155 and hence corresponds to a two - photon resonant situation .
secondly , for a field strength of @xmath152 tw/@xmath154 , the dissociation probability of the ground vibrational state is negligible .
far stronger field strengths are required to dissociate the @xmath156 state and ionization process starts to compete with the dissociation at such high intensities .
thus , in order to to illustrate the role of phase space barriers in the dissociation dynamics without such additional complications , the specific initial state @xmath131 is chosen .
incidentally , such a scenario is quite feasible since a suitably chirped laser field can populate the @xmath131 state very efficiently from the initial ground state @xmath156 and one imagines coming in with the monochromatic laser to dissociate the molecule .
the monochromatically driven morse system studied here has a dimensionality such that one can conveniently visualize the phase space in the original cartesian @xmath157 variables .
however , since the focus is on suppressing dissociation by creating robust kam tori in the phase space , action - angle variables @xmath158 which are canonically conjugate to @xmath157 are convenient and a natural representation to work with .
the action - angle variables @xmath158 of the unperturbed morse oscillator , appropriate for the bound regions , are given by@xcite @xmath159.\end{aligned}\ ] ] [ actionangle ] in the above equations , @xmath160 denotes the dimensionless bound state energy , and @xmath161 for @xmath162 , @xmath163 for @xmath164 . in terms of the action - angle variables it is possible to express the cartesian @xmath157 as follows [ xpactang_chap3 ] @xmath165\\ p&=&\frac{-(2md_0)^{1/2 } [ e_{0}(j)(1-e_{0}(j))]^{1/2}\sin \theta}{1+\sqrt{e_{0}(j)}\cos \theta},\end{aligned}\ ] ] where @xmath166 . substituting for @xmath157 in terms of @xmath158 , the unperturbed morse oscillator hamiltonian in eq .
[ morsepot ] is transformed into @xmath167 where @xmath168 is the harmonic frequency at the minimum . the zeroth - order nonlinear frequency
is easily obtained as @xmath169 and with increasing excitation _
i.e. , _ increasing action ( quanta ) @xmath170 , the non - linear frequency @xmath171 decreases monotonically and eventually vanishes , signaling the onset of unbound dynamics leading to dissociation . and
frequency @xmath130 corresponding to a on - resonance situation . the @xmath172-photon resonance is clearly seen in the phase space and marked as @xmath172:@xmath18 resonance in dark green .
the initial morse eigenstate @xmath131 ( thick dashed line ) is situated rather close to a cantorus ( indicated ) , @xmath173 with @xmath174 being the golden ratio .
the @xmath131 state is connected to the @xmath132 morse eigenstate via the @xmath172:@xmath18 resonance .
note that the states @xmath175 , and @xmath176 are symmetrically located about the resonance with @xmath177 being localized in the resonance.,width=604,height=415 ] the driven system can now be expressed in terms of the variables @xmath158 as @xmath178\cos(\omega_{f}t ) .
\label{prelim_hamact}\ ] ] in addition , since @xmath179 is a periodic function of @xmath180 , one has the fourier expansion @xmath181.\ ] ] as a consequence the driven hamiltonian in eq .
[ prelim_hamact ] can be written as @xmath182 where the matter - field interaction term is denoted as @xmath183\cos(\omega_f t ) , \label{pertact } \nonumber\ ] ] the fourier coefficients @xmath184 and @xmath185 are known analytically and given by [ potfour ] @xmath186 , \label{eqn_four1}\\ v_{n}(j)&=&\frac{(-1)^{n+1}}{\alpha n}{\left[\frac{\sqrt{d_{0}e_{0}(j)}}{d_{0}+ \sqrt{d_{0}^{2}-d_{0}e_{0}(j)}}\right]}^{n } \label{eqn_four2}.\end{aligned}\ ] ] the stroboscopic surface of section in the @xmath158 variables is shown in fig .
[ fig11 ] and is a typical mixed regular - chaotic phase space .
a few important points are worth noting at this stage .
firstly , the initial state of interest is located close to a cantorus with @xmath187 with @xmath174 being the golden ratio .
the importance of this cantorus to the resulting dissociation dynamics of the @xmath131 state was the central focus of the work by brown and wyatt@xcite .
in particular , the extensive stickiness around this region can be clearly seen and hence one expects nontrivial influence on the classical dissociation dynamics as well .
secondly , a prominent @xmath65:@xmath188:@xmath18 nonlinear resonance is also observed in the phase space and represents the classical analog of the quantum @xmath172-photon resonance condition .
interestingly , the area of this resonance is about @xmath108 and , therefore , can support one quantum state .
it turns out that the husimi density of the morse state @xmath177 is localized inside the @xmath172:@xmath18 resonance island .
in addition , the states @xmath131 and @xmath132 are nearly symmetrically located about the @xmath172:@xmath18 resonance . a way to see
this is to use secular pertubation theory on the driven morse hamiltonian .
one can show that in the vicinity of the @xmath172:@xmath18 resonance ( cf .
[ fig11 ] ) an effective pendulum hamiltonian @xmath189 is obtained with @xmath190 , @xmath191 and @xmath192 .
the resonant action @xmath193 \approx
12.6,\ ] ] for the parameters used in this work . using action values @xmath194 ( quantum state @xmath131 ) and @xmath195 ( quantum state @xmath196 ) the energy difference is calculated as @xmath197 in other words , the states @xmath131 and @xmath132 are nearly symmetrical with respect to the state @xmath177 , which is localized in the @xmath172:@xmath18 resonance . therefore , the nonzero coupling @xmath198 will efficiently connect the states @xmath131 and @xmath132 .
moreover , for the given parameters , using the definition of @xmath199 in terms of fourier coefficient @xmath200 , the strength of the resonance is estimated as @xmath201 , clearly a fairly strong resonance .
consequently , the situation in fig . [ fig11 ] is a perfect example where rat can play a crucial role in the dissociation dynamics .
indeed quantum computations ( not shown here ) show that there is a rabi - type cycling of the probabilities between the three morse states .
i now turn to the issue of _ selectively _ controlling the dissociation dynamics of the initial morse eigenstate @xmath131 by creating local barriers in the phase space shown in fig .
[ fig11 ] , bearing in mind the possibility of quantum dynamical tunneling interfering with the control process .
if the classical mechanism of chaotic diffusion leading to dissociation holds in the quantum domain as well then a simple way of controlling the dissociation is to create a local phase space barrier between the state of interest and the chaotic region . in a recent work@xcite
, huang , chandre and uzer provided the theory for recreating local phase space barriers for time - dependent systems and showed that such barriers indeed suppress the ionization of a driven atomic system .
however , huang _ et al .
_ were only concerned with the classical ionization process .
thus , potential complications due to dynamical tunneling were not addressed in their study .
the driven morse system studied here presents an ideal system to understand the interplay of quantum and classical dissociation mechanisms . in what follows ,
i provide a brief introduction to the methodology with an explicit expression for the classical control field needed to recreate an invariant kam barrier , preferably an invariant torus with sufficiently irrational frequency @xmath202 . to start with
, the nonautonomous hamiltonian is mapped into an autonomous one by considering @xmath203 as an additional angle - action pair . denoting the action and angle variables by @xmath204 and @xmath205 , the original driven system hamiltonian ( see eq .
[ hamact2 ] ) can be expressed as @xmath206 with @xmath207 .
note that for a fixed driving field strength @xmath100 and the value of @xmath208 corresponding to a diatomic molecule , @xmath209 is also fixed .
moreover , for physically meaningful values of @xmath208 for most diatoms and typical field strengths far below the ionization threshold one always has @xmath210 . in the absence of the driving field ( @xmath211 ) ,
the zeroth - order hamiltonian is integrable and the phase space is foliated with invariant tori labeled by the action @xmath212 corresponding to the frequency @xmath213 .
however , in the presence of the driving field ( @xmath214 ) the field - matter interaction renders the system nonintegrable with a mixed regular - chaotic phase space .
more specifically , for field strengths near or above a critical value @xmath215 one generally observes a large scale destruction of the field - free invariant tori leading to significant chaos and hence the onset of dissociation .
the critical value @xmath215 itself is clearly dependent on the specific molecule and the initial state of interest .
the aim of the local control method is to rebuild a nonresonant torus @xmath216 , @xmath217 with integer @xmath218 , which has been destroyed due to the interaction with the field . assuming that the destruction of @xmath219 is responsible for the significant dissociation observed for some initial state of interest
, the hope is that locally recreating the @xmath219 will suppress the dissociation _
i.e. , _ @xmath219 acts as a local barrier to dissociation .
ideally , one would like to recreate the local barrier by using a second field ( appropriately called as the control field ) which is much weaker and distinct from the primary driving field .
following huang@xcite _ et al .
_ such a control field @xmath220 can be analytically derived and has the form @xmath221 where @xmath222 and @xmath223 being a linear operator defined by @xmath224 the classical control hamiltonian can now be written down as @xmath225 in case of the driven morse system the control field can be obtained analytically and to leading order is given by @xmath226 where @xmath227 and it can be shown that @xmath228 .
\nonumber\end{aligned}\ ] ] i skip the somewhat tedious derivation of the result above and refer to the original literature@xcite as well as a recent thesis@xcite for details .
note that in the above @xmath202 is the frequency of the invariant torus that is to be recreated corresponding to the unperturbed action @xmath229 and to @xmath230 is located at @xmath231 , assuming the validity of the perturbative treatment . the leading order control field in eq .
[ control2 ] is typically weaker than the driving field and has been shown@xcite to be quite effective in off - resonant cases in suppressing the classical dissociation .
however , in order to study the effect of the control field on the quantum dissociation probabilities , it is necessary to make some simplifications .
one of the main reasons for employing the simplified control fields via the procedure given below has to do with the fact that the classical action - angle variables do not have a direct quantum counterpart@xcite . essentially , the dominant fourier modes @xmath232 of eq . [ control2 ] are identified and one performs the mapping @xmath233 yielding the simplified control hamiltonian @xmath234 if more than one dominant fourier modes are present then they will appear as additional terms in equation [ qcham ] .
note that the above simplified form is equivalent to assuming that the control field is polychromatic in nature , which need not be true in general .
nevertheless , a qualitative understanding of the role of the various fourier modes towards local phase space control is still obtained .
more importantly , and as shown next , in the on - resonant case of interest here , the simplified control field already suggests the central role played by rat . in fig .
[ fig12 ] a summary of the efforts to control the dissociation of the initial @xmath131 state is shown .
specifically , fig .
[ fig12](a ) and ( b ) show the phase spaces where two different kam barriers , @xmath235 and @xmath236 respectively are recreated .
the control hamiltonian in both instances , in cartesian variables , has the following form @xmath237 with @xmath238 tw/@xmath239 .
thus , as desired , the control field strengths are an order of magnitude smaller than the driving field strength .
the simplified control hamiltonian reflects the dominance of the @xmath240 fourier mode of the leading order control field in eq .
[ control2 ] and obtained using the method outlined before .
also , note that the control field comes with a relative phase @xmath241 with respect to the driving field . as expected , from the line of thinking presented in the previous section , fig .
[ fig12](c ) and ( d ) show that the kam barriers indeed suppress the classical dissociation significantly .
_ however , the quantum dissociation in both cases increases slightly ! _ clearly , the recreated kam barriers are ineffective and suggests that the quantum dissociation mechanism is somehow bypassing the kam barriers .
a clue to the surprising quantum results comes from comparing the phase spaces in fig .
[ fig11 ] and fig .
[ fig12 ] which show the uncontrolled and controlled cases respectively .
although , the kam barriers seem to have reduced the extent of stochasticity , the @xmath172:@xmath18 resonance is intact and appears to occupy slightly larger area in the phase space . thus , this certainly indicates that a significant amount of the quantum dissociation is occuring due to the rat mechanism involving the three morse states @xmath175 , and @xmath176 as discussed before .
in particular , the @xmath177 state must still be actively providing a route to couple the initial state to the chaotic region via rat . how can one test the veracity of such an explanation ?
one way is to scale the planck constant down from @xmath6 and monitor the quantum dissociation process . reduced @xmath108 implies that the @xmath172:@xmath18 island can support several states as well as the fact that other higher order resonances now become relevant to the rat mechanism .
such a study is not presented here but one would anticipate that the dissociation mechanism can be understood based on the theory of rat that already exists ( see contributions by schlagheck _
et al . _ and bcker _ et al . _
, for example ) .
another way is to directly interfere locally with the resonance and see if the quantum dissociation is actually suppressed . locally interfering with a specific phase space structure , keeping the gross features unchanged ,
is not necessarily a straightforward approach .
however , the tools in the previous section allow for such local interference and i present the results below . with @xmath242 ) designed to recreate specific kam barriers ( a ) @xmath235 and ( b ) @xmath236 .
note that in both cases the dominant @xmath240 fourier amplitude of the leading order control field of eq .
[ control2 ] have been utilized and the desired kam barriers are clearly seen ( thick gray ) .
the effect of the barriers seen in ( a ) and ( b ) on the classical ( gray circles ) and quantum ( gray squares ) dissociation probabilities are shown in ( c ) and ( d ) respectively .
the uncontrolled results are indicated by the corresponding black symbols .
interstingly , the classical dissociation is suppressed but the quantum dissociation is slightly enhanced.,width=529,height=377 ] kam barrier . in ( a ) two dominant fourier modes @xmath240 and @xmath243 of the leading order control term in eq .
[ control2 ] are retained . in
( b ) the simplified control field as in eq .
[ qcham ] is used with an effective field strength estimated using @xmath240 and @xmath243 .
notice that the desired kam barrier is recreated in ( a ) but not in ( b ) ( thick gray line indicates the expected location ) . in ( c )
the classical dissociation probabilities of the uncontrolled ( black circles ) , control using field ( a ) ( open circles ) , and control using field ( b ) ( gray circles ) are shown .
the quantum results are shown in ( d ) for the uncontrolled ( black squares ) and control using field ( b ) ( gray squares ) are shown .
see text for discussion.,width=529,height=377 ] the control field in eq .
[ control2 ] corresponding to the case of fig .
[ fig12](b ) _
i.e. , _ recreating the @xmath236 kam barrier , turns out to be well approximated by @xmath244 the above hamiltonian comes about due to the fact@xcite that two fourier modes @xmath240 and @xmath243 are significant in this case .
note that the specific kam barrier of interest has a frequency between that of the @xmath245 cantorus ( around which the initial state is localized , cf .
[ fig11 ] ) and the @xmath172:@xmath18 nonlinear resonance . from a perturbative viewpoint ,
creation of this kam barrier using eq .
[ appcontham ] is not expected to be easy due to the proximity to the @xmath172:@xmath18 resonance and the fact that the @xmath243 fourier component is nothing but the @xmath172:@xmath18 resonance .
nevertheless , fig .
[ fig13](a ) shows that the specific kam barrier is restored and , compared to fig .
[ fig11 ] , the controlled phase space does exhibit reduced amount of chaos .
consisitently , fig .
[ fig13](c ) shows that the classical dissociation is suppressed by nearly a factor of two as in the case shown in fig .
[ fig12](b ) wherein only the @xmath240 component of the control hamiltonian was retained . as mentioned earlier , in order to calculate the quantum dissociation probability the control hamiltonian in eq .
[ appcontham ] needs to be mapped into a from as in eq .
[ qcham ] .
since two fourier modes need to be taken into account , the effective control field strength is given by @xmath246 such a procedure yields @xmath247 and thus the control field , still less intense than the primary field , comes with a relative phase of zero .
interestingly , as shown in fig .
[ fig13](b ) , the resulting simplified control hamiltonian fails to create the desired barrier .
moreover , the phase space also exhibits increased stochasticity and as a consequence the classical dissociation is enhanced ( cf .
[ fig13](c ) ) . however , fig .
[ fig13](b ) reveals an interesting feature - the @xmath172:@xmath18 resonance is severly perturbed .
this perturbation is a consequence of including the @xmath243 fourier component into the effective control hamiltonian .
_ the key result , however , is shown in fig .
[ fig13](d ) where one observes that the quantum dissociation probability is reduced significantly_. it seems like the quantum dynamics feels the barrier when there is none ! the surprising and counterintutive results summarized in fig .
[ fig12 ] and fig .
[ fig13 ] can be rationalized by a single phenomenon - dynamical ( resonance assisted in this case ) tunneling .
the main clue comes from the observation that quantum suppression happens as soon as the @xmath172:@xmath18 resonance is perturbed .
although the results here are shown with a specific example , the phenomenon is general .
indeed computations ( not published ) for different sets of parameters have supported the viewpoint expressed above .
interestingly , the work of huang _
focused on classical suppression of ionization by creating local phase space barriers in case of the driven one - dimensional hydrogen atom@xcite . around the same time
highlighted@xcite the importance of the rat mechanism in order to obtain accurate decay lifetimes of localized wavepackets in the same system .
work is currently underway to see if the conclusions made in this section hold in the driven atomic system as well _ i.e. , _ whether the attempt to suppress the ionization by creating local phase space barriers is foiled by the phenomenon of rat .
the two examples discussed in this work illustrate a key point - _ dynamical tunneling plays a nontrivial role in the process of quantum control .
_ in the first example of the driven quartic double well it is clear that bichromatic control fails in regimes where chaos - assisted tunneling is important . in the second example of a driven morse oscillator it is apparent that efforts to control by building phase space kam barriers fail when resonance - assisted tunneling is possible . in both instances the competition between classical and quantum mechanisms
is brought to the forefront .
although the two examples shown here represent the failure of specific control schemes due to dynamical tunneling , one must not take this to be a general conclusion .
it is quite possible that some other control schemes might owe their efficiency to the phenonmenon of dynamical tunneling itself .
further classical - quantum correspondence studies on control with more general driving fields are required in order to confirm ( or refute ) the conclusions presented in this chapter . at the same time the two examples presented here are certainly not the the last word ; establishing the role of dynamical tunneling in quantum coherent control requires one to step into the murky world of three or more degrees of freedom systems@xcite .
i mention two model systems , currently being studied in our group , in order to stress upon some of the key issues that might crop up in such high dimensional systems .
for example , two coupled morse oscillators which are driven by a monochromatic field already presents a number of challenges both from the technical as well as conceptual viewpoints .
the technical challenge arises due to the fact that dimensionality constraints do not allow one to visualize the global phase space structures as easily as done in this chapter .
one approach is to use the method of local frequency analysis@xcite to construct the arnold web _ i.e. , _ the network of nonlinear resonances that regulate the multidimensional phase space transport . in a previous work@xcite , involving a time - independent hamiltonian system , the utility of such an approach and the validity of the rat mechanism has been established .
however , lifting " quantum dynamics onto the arnold web is an intriguing possibility which is still an open issue . on the conceptual side
there are several issues with multidimensional systems .
i mention a few of them here .
firstly , even at the classical level one has the possibility of transport like arnold diffusion@xcite which is genuinely a three or more dof effect and has no counterpart in systems with less than three dofs .
note that arnold diffusion is typically a very long time process and is notoriously difficult to observe in realistic physical systems@xcite .
moreover , arguments can be made for the irrelevance of arnold diffusion ( or some similar process ) in quantum systems due to the finiteness of the planck constant .
secondly , an interesting competition occurs in systems such as the driven coupled morse oscillators . even in the absence of the field
the dynamics is nonintegrable and one can be in a regime where the modes are exchanging energy but none of the modes gain enough energy to dissociate . on the other hand , in the absence of mode - mode coupling , a weak enough field can excite the system without leading to dissociation
. however , in the presence of such a weak field and the mode - mode coupling one can have significant dissociation of a specific vibrational mode . clearly , there is nontrivial competition between transport due to mode - mode resonances , field - mode resonances and the chaotic regions@xcite .
selective control of such driven coupled systems is an active research area@xcite today and the lessons learnt from the two examples suggest that dynamical tunneling in one form or another can play a central role . in the context of studying the potential competition between arnold diffusion and dynamical tunneling ,
i should mention the driven coupled quartic oscillator system with the hamiltonian : @xmath248 in the absence of the field @xmath249 the hamiltonian reduces to the case originally studied by tomsovic , bohigas and ullmo wherein the existence of chaos - assisted tunneling was established in exquisite detail@xcite . at the same time , with @xmath250 the system represents one of the few examples for which the phenomenon of arnold diffusion has been investigated over a number of years .
nearly a decade ago , demikhovskii , izrailev , and malyshev studied@xcite a variant of the above hamiltonian to uncover the fingerprint of arnold diffusion on the quantum eigenstates and dynamics .
an interesting question , amongst many others , is this : _ will the fluctuations in the chaos - assisted tunneling splittings for @xmath249 , observed for varying @xmath108 , survive in the presence of the field ? _ a related topic which i have not touched upon in this article has to do with the control of ivr using weak external fields . based on the insights gained from studies done until now on field - free ivr
( see also the contribution by leitner in this volume ) , it is natural to expect that dynamical tunneling could play spoilsport for certain class of initial states that are prepared experimentally .
the issue , however , is far more subtle and more studies in this direction might shed light on the mechanism by which quantum optimal control methods work .
for instance , recent proposals on quantum control by takami and fujisaki@xcite and the so called quantized ulam conjecture " by gruebele and wolynes@xcite take advantage ( implicitly ) of the system having a completely chaotic phase space .
dynamical tunneling is not an issue in such cases .
however , in more generic instances of systems with mixed regular - chaotic phase space , a quantitative and qualitative understanding of dynamical tunneling becomes imperative .
given the level of detail at which one is now capable of studying the clasically forbidden processes , as reflected by the varied contributions in this volume , i expect exciting progress in this direction .
i think that this is an appropriate forum to acknowledge the genesis of the current and past research of mine on dynamical tunneling . in this context
i am grateful to greg ezra , whose first suggestion for my postdoc work came in the form of a list of some of the key papers on dynamical tunneling .
we never got around to work on dynamical tunneling _ per se _ but those references came in handy nearly a decade later .
it is also a real pleasure to thank peter schlagheck for several inspiring discussions on tunneling in general and dynamical tunneling in particular
. theoretical study of intramolecular vibrational energy relaxation of acteylinic ch vibration for @xmath251 and @xmath172 in large polyatomic molecules ( cx@xmath252)@xmath252ycch , where x@xmath253h or d and y@xmath253c or si . | contribution to the edited volume _ dynamical tunneling : theory and experiment _ , editors srihari keshavamurthy and peter schlagheck , crc press , taylor & francis , boca raton , 2011 . |
in this appendix , we investigate the effect of matter dominated era ( mde ) on the evolution of holographic dark energy .
we assume that there are a period of the inflation followed by the radiation dominated era ( rde ) , a slow transition from matter dominated era to dark energy dominated era ( mde+dde ) of which scale factors are given by @xmath119 respectively , where @xmath120 is a constant .
we set the transition from the radiation dominated era ( rde ) to the matter dominated era ( mde ) happens at @xmath121 , the equipartition time .
that is , @xmath122 , where @xmath123 is matter energy density .
the last phase consists of the matter dominated era ( mde ) ( @xmath124 ) followed by a dark energy dominated era ( dde ) ( @xmath125 ) .
the inflection point @xmath50 is determined by the value in which the second derivative of @xmath19 vanishes during the third phase : @xmath126 ( red thick line ) and @xmath44 ( blue dashed line ) of the dark energy as a function of time @xmath21 for the evolution in eq .
( [ rhmatter ] ) , which includes the matter dominated era . @xmath99 and @xmath100 .
[ figomega2],scaledwidth=50.0% ] using the scale factors in eq .
( [ rtnew ] ) , it is easy to derive the event horizon size @xmath13 and dark energy density @xmath2 by following the procedure in the main text . after some straightforward calculation
we obtain @xmath13 for each phase ; @xmath127,$ & $ t_{eq } \leq t , $ ~{\text ( mde+dde)}.\\ \end{tabular } \right.\end{aligned}\ ] ] during the calculation , from the continuity condition of @xmath13 between rde and mde , we have obtained = ^-1 and used this @xmath120 for following calculations . in generic holographic dark energy models , dark matter is independent of dark energy , and we need an parameter describing the nature of dark matter .
we choose the observed equipartition time @xmath128 for the parameter .
[ figomega2 ] shows the results for @xmath129 .
@xmath65 becomes smaller compared to the case in fig .
3 , because the matter energy density decreases slowly ( @xmath130 ) than the radiation energy density ( @xmath131 ) .
the other parameters are the same as those of fig .
3 . as assumed in the main text ,
including mde in our consideration does not significantly change the results .
compared to the case without mde ( fig .
[ figomega ] ) , @xmath132 curve is more flat and @xmath133 years is later
. these results are more consistent with observations , while @xmath134 is slightly smaller than the observed value .
since our holographic dark energy density changes about @xmath135 times in scale from the inflation to the present , this level of coincidence is interesting , considering the approximations we have used . to check the accuracy of our calculation using the friedmann equation , we plot the total energy density and @xmath136 in fig .
[ figcheck ] .
the graph shows the level of accuracy mentioned above . in the case with mde considered in this appendix , due to the freedom of @xmath138 , there was no guarantee that the inflation with @xmath139 could solve the cosmic coincidence problem .
however , interestingly , it turns out that even in this case the required @xmath65 value is similar to that of the case without mde .
this is due to the fact that the observed initial dark matter density is much smaller than that of radiation . even in the worst case
that mde started just after the reheating of the inflation and there was no rde , eq .
( [ rholambda ] ) with @xmath140 gives a value @xmath141 . | we show that holographic dark energy could explain why the current dark energy density is so small , if there was an inflation with a sufficient expansion in the early universe .
it is also suggested that an inflation with the number of e - folds @xmath0 may solve the cosmic coincidence problem in this context . assuming the inflation and the power - law acceleration phase today we obtain approximate formulas for the event horizon size of the universe and dark energy density as functions of time .
a simple numerical study exploiting the formula well reproduces the observed evolution of dark energy .
this nontrivial match between the theory and the observational data supports both inflation and holographic dark energy models . the type ia supernova ( sn ia ) observations @xcite strongly suggest that the current universe is in an accelerating phase , which can be explained by dark energy ( a generalization of the cosmological constant ) having pressure @xmath1 and density @xmath2 such that @xmath3 .
there are various dark energy models rely on exotic materials such as quintessence @xcite , @xmath4-essence @xcite , phantom @xcite , and chaplygin gas @xcite .
being one of the most important unsolved puzzles in modern physics , the cosmological constant problem consists of three sub - problems ; why the cosmological constant is so small , nonzero , and comparable to the critical density at the present . in this paper
we show that , in the holographic dark energy model , an inflation with a sufficient expansion explain why the current dark energy density is so small .
we also suggest that the last problem , the cosmic coincidence problem , could be solved , if there was an inflation with a specific expansion .
note that , in many other dark energy models , it is not easy to explain the current ratio of dark energy density to matter energy density , because usually dark energy density and matter energy density reduce at different rates @xcite for a long cosmological time scale .
it is well known @xcite that a simple combination of the reduced planck mass @xmath5 and the hubble parameter @xmath6 , gives a value @xmath7 comparable to the observed dark energy density @xmath8 @xcite .
this interesting coincidence , on one hand , is of the cosmic coincidence problem and , on the other hand , motivated holographic dark energy models .
the holographic dark energy models are based on the holographic principle proposed by t hooft and susskind @xcite , claiming that all of the information in a volume can be described by the physics at the boundary of the volume . with the base on the principle , cohen et al @xcite proposed a relation between an uv cutoff ( @xmath9 ) and an ir cutoff ( @xmath10 ) by considering that the total energy in a region of size @xmath10 can not be larger than the mass of a black hole of that size . saturating the bound , one can obtain [ holodark ] _ = , where @xmath11 is a constant .
hsu @xcite pointed out that for @xmath12 , the holographic dark energy behaves like matter rather than dark energy .
many attempts @xcite have been made to overcome this ir cutoff problem , for example , by using non - minimal coupling to a scalar field @xcite or an interaction between dark energy and dark matter @xcite .
li @xcite suggested that an ansatz for the holographic dark energy density [ holodark2 ] _
= , would give a correct accelerating universe , where the future event horizon ( @xmath13 ) is used instead of the hubble horizon as the ir cutoff @xmath10 . to solve the coincidence problem many attempts
have been done @xcite .
an interaction of dark matter @xcite with dark energy was introduced in @xcite . in @xcite inflation at the gut scale with the minimal number of e - folds @xmath14 was suggested as a solution . in this paper
we suggest a solution similar to the later .
one motivation to study the cosmic coincidence problem in the context of inflationary cosmology is that if there was no inflation , there could be no ` now ' ( @xmath15 years ) for the ` why now ' question .
according to astronomical observations and cosmological theory there are at least two inflationary periods in the history of the universe . as is well known , the first inflation at the early universe with @xmath16 is need to solve the problems of the standard big - bang cosmology .
this inflation is often assumed to be related to vacuum energy of a scalar field ( inflaton ) .
the second inflation ( re - inflation ) is a period of an accelerated expansion today due to dark energy .
( usually , the first inflation is related to a phase transition of the inflaton and has a different origin from that of the re - inflation due to dark energy . in this paper
we assume this case . )
thus , we assume that in the universe there are the inflaton , holographic dark energy , radiation and matter ( mostly , cold dark matter ) .
we also assume that after reheating inflaton energy decays to radiation perfectly . during the first inflation holographic dark energy
is diluted exponentially . in this work
we suggest that if there is holographic dark energy in the universe , the first inflation with @xmath0 leads to onset of the second inflation at the time @xmath17 years as observed , and , hence , the inflation solves the cosmic coincidence problem in the context of holographic dark energy . in this paper
we consider the flat ( @xmath18 ) friedmann universe which is favored by observations @xcite and described by the metric ds^2=-dt^2+r^2(t)d^2 , where @xmath19 is the scale factor . in the holographic dark energy model
a typical length scale of the system with the horizon is given by the future event horizon [ rh ] r_hr(t)_t^ = r(t)_t^ , which is a key quantity .
it is a subtle task to obtain an explicit form for @xmath20 , because @xmath20 depends on the whole history of the universe after @xmath21 . to tackle this problem we divide the history of the universe into two phases ; the inflation ( phase 1 ) is followed by phase 2 which are consecutive radiation dominated era ( rde ; @xmath22 ) and dark energy dominated power - law accelerating era ( dde ; @xmath23 ) , respectively . for simplicity
, we ignore the matter dominated era ( mde ) as often done in an order of magnitude estimate in cosmology .
( in appendix , we perform a similar calculation with mde .
the main results are similar . )
\1 ) inflation phase ( @xmath24 ) + the inflation starts at @xmath25 and ends at @xmath26 .
the scale factor evolves in this phase as follows r(t)=
r_i e^h_i ( t - t_i ) , where @xmath27 is the initial scale factor at @xmath25 and @xmath28 is the hubble parameter with the energy scale @xmath29 of the inflation .
hence , the number of e - folds of expansion @xmath30 .
\2 ) power - law expansion phase ( @xmath31 ) + this phase consists of rde ( @xmath32 ) followed by dde ( @xmath33 ) .
the universe starts to accelerate at an inflection point @xmath34 , i.e. , @xmath35 .
we assume that the scale factor evolves in this phase as [ rt2 ] r(t)= r_i e^n()^ ( ) ^2n , where @xmath36 is a constant .
the scale factor @xmath19 grows as @xmath37 during the rde and as @xmath38 during the dde later .
@xmath19 of this form gives a smooth transition from rde to dde .
note that @xmath19 for each era is well - known and can be derived from the friedmann equation depending on the dominant energy source .
the power - law acceleration is a generic feature of dde if @xmath39 .
( alternatively , one can divide this phase into rde and dde and choose the scale factor as @xmath40 and @xmath41 for rde and dde , respectively .
this choice gives almost the same results except for a slightly decreasing @xmath13 as @xmath42 .
thus , we can use the specific form in eq .
( [ rt2 ] ) without loss of generality . )
since observational data favor @xmath43 @xcite and @xmath44 close to @xmath45 , the power index @xmath46 is much larger than @xmath47 . if we choose @xmath48 , then @xmath49 .
the inflection point @xmath50 is determined by the value at which the second derivative of @xmath19 vanishes : [ ta : alpha ] = ( ) ^2 . from @xmath19
we obtain @xmath20 using eq .
( [ rh ] ) . during the inflationary phase ( phase 1 ) : ( t ) & & _ t^t_f + _ t_f^ & = & + c(t_f ) , where @xmath51 is a constant dependent on @xmath26 .
a finite @xmath51 implies a finite @xmath20 and , hence , the existence of dde .
this constant should be determined by the initial condition at @xmath52 .
thus the distance to the future horizon @xmath20 during the phase 1 is [ rhinflation ] r_h(t)&=&r(t ) i_1(t)&= & + ( r_i e^n c(t_f)- ) e^h_i(t - t_f ) . to determine the value of @xmath51
, we use an initial condition @xmath53 for @xmath13 : [ rhti ] r_h(t_i)=+ ( r_i e^n c(t_f)- ) e^-n . inserting eq .
( [ rhti ] ) into eq .
( [ rhinflation ] ) we can rewrite eq .
( [ rhinflation ] ) as [ rhsimple ] r_h(t)= ( 1 + a e^h_i(t - t_i ) ) , where @xmath54 is a dimensionless constant which depends on the initial condition at @xmath25 , given by ah_i r_h(t_i)-1 . therefore ,
if @xmath55 , i.e. , @xmath56 , the event horizon grows exponentially during the inflation . at the same time @xmath2 decreases exponentially
this is also noted in ref .
@xcite , where the correction to the inflation due to holographic dark energy was investigated .
it is a reasonable assumption that dark energy density at @xmath52 is comparable to other energy densities , that is , @xmath57 , or @xmath58 . if not , we need either fine tuning or a special mechanism to make initial dark energy density parameter @xmath59 be much smaller than 1 at the extremely early universe , which is implausible .
this can be also seen from the following relation @xcite ; h r_h = , which holographic dark energy model should satisfy all the time .
it is @xmath60 for @xmath59 not too much smaller than 1 .
therefore , @xmath61 is a plausible initial condition . at @xmath62 , [ rh : f ] r_h(t_f ) = ( 1 + ae^n ) .
now consider the phase 2 .
using @xmath19 in eq .
( [ rt2 ] ) , it is straightforward to obtain the following relations , [ i2 t ] i _
2(t ) & & _ t^ & = & ( ) ^1 - 2n . therefore , during the phase 2 the event horizon at @xmath21 is at the distance [ rh:2 ] r_h(t ) = ( + ) .
now , the horizon distance at @xmath62 is [ rhtf ] r_h(t_f)=. comparing eq .
( [ rh : f ] ) with eq .
( [ rhtf ] ) , we have [ alpha : n ] = ( 1+ae^n ) .
therefore , we have [ alpha : a ] = ^-1 1 . inserting this into eq .
( [ rh:2 ] ) we obtain @xmath63 now we have approximate analytical formulas for @xmath20 for the whole history of the universe since the inflationary era .
note that @xmath20 is a monotonically increasing function of time .
we next consider the behaviors of @xmath20 for @xmath64 . in this case
, we have [ rhrde ] r_h(t ) t_f(-1n-1/2 ) , which is proportional to @xmath19 . from eq .
( [ ta : alpha ] ) the inflection point @xmath50 is given by the initial conditions @xmath54 and @xmath65 to be [ tatf ] ( ) ^2 0.095 ( ) ^2 .
this relation is interesting and informative .
the ratio of the two time scale @xmath50 and @xmath26 is related to the initial condition . for @xmath66 and @xmath67 ,
this ratio explain why dark energy dominates so lately .
eq .
( [ tatf ] ) implies that @xmath50 and , hence , evolution of the universe is more sensitive to @xmath65 than to @xmath54 or @xmath68 . from now on all quantities are given in natural units ; @xmath69 . for @xmath70 ( gut scale inflation ) and @xmath71 , a reasonable value for inflation to solve the problems of the big - bang cosmology , this equation gives observed @xmath72 . in this way
, the holographic dark energy model could solve the cosmic coincidence problem .
interestingly , eq .
( [ tatf ] ) gives a lower bound for the energy scale of the inflation .
the usual bound @xmath73 for inflation returns @xmath74 and hence @xmath75 .
this can rule out low energy scale inflation models . on the other hand , an obvious condition @xmath76 returns @xmath77 , where @xmath78 is the planck time . to determine the true value of @xmath65 we need to go beyond the approximation used in this work .
let us explain more physically how our model could solve the coincidence problem .
using eq .
( [ rhsimple ] ) we obtain @xmath79 , which means that the event horizon expands exponentially during the inflation . at the same time
the dark energy density @xmath80 rapidly decreases ( see fig .
[ rho ] ) ; _ ( t_f)= _ ( t_i)()^2 e^-2n , where we used @xmath81 and @xmath82 .
this is the dark energy density just after the inflation .
after the inflation , dark energy is sub - dominant , i.e. , @xmath83 , and behaves like matter with a constant equation of state @xcite [ omega ] _
= - ( 1+)- . in this case
@xmath84 , while the radiation energy density , [ rhor ] _ r ( t ) m_i^4()^4 , decreases more rapidly than the dark energy density .
( from eq .
( [ rhrde ] ) one can also see that @xmath85 during the rde ) . here
we assume an instant reheating after the inflation for simplicity .
therefore , during the rde [ rholambda ] _ ( t)_(t_f)()^2 _ ( t_f ) ( ) = m^4_i e^-2n ( ) , which should be comparable to @xmath86 at @xmath50 , where @xmath87 is the observed energy scale of the universe at the inflection point @xmath50 . from the above relation ,
the required e - folds is [ napprox ] n- lnln ( ) 64.5 , which is slightly larger than the minimal @xmath65 for the inflation to solve the many problems of the standard big bang cosmology .
here we have used @xmath88 .
this result is comparable with heuristic arguments of li @xcite .
hence , we see again that an inflation with @xmath0 could solve the cosmic coincidence problem in a self - consistent manner in the holographic dark energy context . to be more concrete we perform a numerical study using the analytic formulas to fit parameters for the inflation such as @xmath65 and @xmath29 onto the observed cosmological parameters such as @xmath89 and @xmath90 . once we know @xmath20 and @xmath19 , it is easy to obtain @xmath91 and @xmath92 by using eqs .
( [ rhor ] ) and ( [ rholambda ] ) for a given @xmath65 .
we choose reasonable values @xmath93 , @xmath94 and @xmath49 .
we will show later that our results are not so sensitive to the value of @xmath54 or @xmath68 as long as @xmath95 . from eqs .
( [ rhrde ] ) and ( [ tatf ] ) one can see that @xmath50 and [ omega ] _
( t ) are sensitive to @xmath65 .
( red thick line ) and the scale factor @xmath19 ( blue dashed line ) as functions of time @xmath21 for @xmath96 , @xmath49 , and @xmath93 .
@xmath20 as well as @xmath19 grows exponentially during the inflation .
all quantities are given in natural units , where @xmath69 .
[ rhfig ] , scaledwidth=50.0% ] ( red thick line ) and the radiation density @xmath97 ( blue dashed line ) as a function of time @xmath21 for the evolution shown in fig .
[ rhfig ] .
@xmath98 is comparable to the observed value .
we ignore the matter dominated era for simplicity .
[ rho],scaledwidth=50.0% ] ( red thick line ) from eq .
( [ omega ] ) and @xmath44 ( blue dashed line ) from eq .
( [ omega ] ) of the dark energy as a function of time @xmath21 for the evolution shown in fig .
[ rhfig ] .
@xmath99 and @xmath100 are comparable to the observed values .
[ figomega],scaledwidth=50.0% ] for @xmath96 our theory gives @xmath99 , @xmath101 , and @xmath98 , which are comparable with current observations .
note that this is not a fine tuning of @xmath65 .
since the expansion during the inflation is a history already happened , @xmath65 is obviously a fixed value .
thus , @xmath65 value is , like other cosmological parameters , something which should be predicted by a theory and then be verified by observations .
one can assert fine - tuning only when a required parameter value is unnatural .
our model @xmath102 a value @xmath103 which satisfies all known observational constraints and is consistent with inflation theory .
thus , the possibility of determining @xmath65 is not necessary a flaw but a possible merit of our theory .
although we can not rule out @xmath104 , interestingly , there is an asserted upper bound , @xmath105 from the holographic principle @xcite and from the density perturbation generation @xcite .
if this upper bound is correct , one can say the holographic dark energy can solve the cosmic coincidence problem . on the other hand
, @xmath106 is smaller than the observed value .
this discrepancy can be attributed to approximations we used such as an instant reheating after the inflation and ignoring the matter dominated era .
if we choose @xmath107 instead of @xmath49 , @xmath108 gives the same results .
thus , the results are not sensitive to @xmath68 .
we do not need a fine tuning for @xmath54 too .
for example , if we choose @xmath109 , then we need @xmath110 to reproduce the observed universe and @xmath111 for @xmath112 . as mentioned above a natural value for this dimensionless quantity @xmath54 without fine tuning
is @xmath60 .
let us recall the inputs and the outputs in our theory .
we have assumed that there are inflationary era , rde , and dde in the observed evolution of our universe and used the typical forms of @xmath19 for these phases . with reasonable input values for @xmath113 (
our results are not sensitive to these values ) and @xmath0 , we have obtained output values for current density parameter @xmath89 and equation of state @xmath114 for dark energy , which are comparable with observed values .
this could solve the cosmic coincidence problem .
note that assuming dde without an appropriate inflation does not automatically explain why @xmath115 .
considering the long time scale involved ( @xmath116 years ) ) and the difference between time dependency of the dark energy density and that of the matter density , it is remarkable that with the parameter @xmath65 , and the reasonable assumptions , our analysis reproduces the observed universe with the correct order of magnitude as shown in the figures .
this indicates that the holographic dark energy models with @xmath43 are promising candidates for a correct dark energy model . in summary
, we show that an inflation with a sufficient expansion make the current holographic dark energy density exponentially small .
it is also possible that an inflation of @xmath0 could solve the cosmic coincidence problem without introducing an interaction with dark matter or modifying gravity .
the holographic dark energy models have an intrinsic advantage over non - holographic models in that it does not need fine tuning of parameters or an @xmath117 mechanism to cancel the zero - point energy of the vacuum , simply because it has no @xmath118 zero - point vacuum energy from the start .
quantum field theory over - counts the independent physical degrees of freedom inside the volume .
furthermore , as suggested in this paper , the cosmic coincidence problem could be also solved if there was an inflation with @xmath0 .
all these results support not only the inflation theory but also the holographic dark energy models with @xmath43 . |
butcher & oemler s ( 1978,1984 ) discovery that fraction of blue galaxies in clusters increases with redshift first highlighted the significance and strength of the evolution of the cluster galaxy population over the past 4 - 5 gyr .
the butcher - oemler effect has since been confirmed by many authors ( rakos , schombert 1995 ; couch et al . 1994,1998 ; margoniner , de carvalho 2000 ; margoniner et al . 2001 ; ellingson et al . 2001 ;
kodama & bower 2001 ; goto et al . 2003a , but also see andreon et al .
1999,2004 ) .
high redshift clusters ( @xmath11 ) are also observed to have significantly larger fractions of star - forming galaxies than local clusters ( dressler & gunn 1992 ; postman , lubin , & oke 1998 ; dressler et al .
1999 ; poggianti et al . 1999 ; postman , lubin , & oke 2001 ) .
blue and star - forming galaxies are clearly more common in intermediate redshift clusters than in their local counterparts .
furthermore , the population fraction of s0 galaxies is lower in high redshift clusters ( dressler et al . 1997 ; van dokkum et al . 1998 ; fasano et al . 2000 ; jones , smail , & couch 2000 ; fabricant et al . 2000 ; also see andreon et al .
1998 ; postman et al . 2004 ) and van dokkum et al . (
1998 ) have identified a significantly larger fraction of merging / interacting galaxies in the cluster ms1054 - 0321 at @xmath0 .
these observations suggest that environmental processes are at least partly responsible for altering the morphology and star formation rates ( sfr ) of cluster galaxies as a function of redshift .
there is likely more than one physical mechanism involved given the large dynamic range in density over which the morphology - density relation extends ( postman et al .
1984 ; goto et al .
however , it has been difficult to specify which specific physical processes are the dominant ones given a limited amount data at key epochs over the past half a hubble time . the cluster galaxy luminosity function ( lf ) is a fundamental constraint on the mass assembly history of galaxy clusters as it is related to the cluster galaxy mass function through the mass - to - light ratio .
precise and accurate measurements of cluster lfs as a functions of local galaxy density , clustocentric radius , galaxy structure and color have the potential to provide insight into the role of environmental processes in determining the properties of the present - day galaxy populations .
the lf is , therefore , central to many cosmological issues ( koo & kron 1992 ; ostriker 1993 ; binggeli et al .
1998 ) . in the local universe ,
the cluster galaxy lfs have been studied in considerable detail at optical and near - infrared wavelengths ( e.g. , lugger 1986 ; trentham 1998 ; garilli et al .
1999 ; paollilo et al .
2001 ; martnez , zandivarez , merch ' an , & domnguez 2002 ; cuesta - bolao & serna 2003;de propris et al .
2003;lin et al . 2004 ) .
goto et al . ( 2002b ) derived the composite luminosity function of 204 clusters from the `` sdss cut & enhance galaxy cluster catalog '' ( goto et al .
2002a ) in the redshift range from @xmath12 to @xmath13 in the five sdss bands @xmath14 , @xmath15 , @xmath16 , @xmath17 and @xmath18 , and found that the faint - end slope of the luminosity function becomes flatter toward the redder wavebands , consistent with the hypothesis that the cluster luminosity function has two distinct underlying populations ; a population of bright ellipticals with a gaussian - like luminosity distribution that dominate the bright end , and a population of faint blue star - forming galaxies with a steep power - law like luminosity distribution , that dominate the faint - end . indeed
, several studies find that cluster lfs are better described by a sum of two functions ; a gaussian for bright galaxies and a steep schechter function for faint galaxies ( driver et al .
1994 ; molinari , chincarini , moretti , & de grandi 1998 ; parolin et al .
2003 ; mercurio et al .
2003 ; dahl ' en et al . 2004 ) , consistent with the above picture .
de propris et al . ( 1998 ) derived the @xmath19-band luminosity function of the coma cluster galaxies , providing further support to the hypothesis of distinct populations : intrinsically bright red galaxies and intrinsically faint blue dwarf galaxies .
the dwarf - to - giant ratio also appears to depend on environment .
several studies ( driver et al . 1998 ; phillips et al .
1998 ; andreon 2001 ; dahl ' en et al . 2004 ) find that dwarf galaxies are more common in lower density environments in local clusters .
it is important to extend these studies to higher redshifts as the evolution of the above characteristics can provide useful discrimination amongst structure formation and evolution models . to date , however , only a handful of clusters at @xmath20 have been studied in detail ( stanford et al .
1997 ; benitez et al . 1999 ; rosati et al . 1999 ; tanaka et al .
2000 ; haines et al . 2001 ; van dokkum et al . 2001 ; lidman et al .
in addition , most of these studies relied on photometric redshifts to determine cluster membership .
the evolution of the @xmath21-band luminosity function in the redshift range @xmath22-@xmath23 has been studied by de propris et al .
the evolution of the characteristic magnitude @xmath24 of the galaxies , is found to be consistent with a passively evolving population of galaxies formed at @xmath25 .
the data are not deep enough , however , to constrain the faint - end slope which was fixed at @xmath26 in the above study ( and adopted from the value derived in the @xmath19-band for the coma cluster ) .
nakata et al .
( 2001 ) extended the study to @xmath27 by deriving the @xmath21 band luminosity function for the cluster around the radio galaxy 3c324 .
they also chose to fix the faint - end slope at @xmath26 and found a result consistent with what is expected for a passively evolving stellar population formed at @xmath25 .
toft et al .
( 2004 ) studied the @xmath21-band lf of galaxy cluster rdcs j1252.9 - 2927 at @xmath28 .
they found @xmath29 mag of evolution and a similar shape for the faint - end tail .
the main difficulties in studying cluster galaxy populations at @xmath20 are , of course , the fainter apparent magnitude of the distant cluster galaxies and contamination by non - cluster member galaxies . with the advent of the advanced camera surveys ( acs ) on board the _ hubble space telescope ( hst ) _
, we now have a chance to step forward . as a part of guaranteed - time - observations ( gto )
, we have observed the galaxy cluster , ms1054 - 0321 at @xmath0 . a large companion spectroscopic campaign yielded redshifts for 143 cluster members , eliminating the uncertainties associated with photometric redshift estimation or statistical background subtraction ( driver et al . 1994
; bernstein et al . 1995 ; valotto et al . 2001 ) .
the superb sensitivity and angular resolution of the acs allows us to classify galaxies morphologically to the depth of @xmath30 mag .
we present in this paper our acs - based lf and its dependence on clustocentric position , environment , and galaxy properties for the galaxy cluster ms1054 - 0321 at @xmath0 .
the paper is organized as follows : in section [ data ] , we describe the observations ; in section [ analysis ] , we describe the details of the analysis ; in section [ results ] , we present the results ; in section [ discussion ] , we discuss the physical implications of our results ; and in section [ conclusion ] , we summarize our work and findings . unless otherwise stated , we adopt the best - fit wmap cosmology : @xmath31 ( bennett et al .
2003 ) , giving a scale of 7.62 kpc per arcsec at @xmath32 .
with a redshift of @xmath33 , ms1054 - 0321 is the most distant cluster in the extended medium sensitivity survey ( emss ) x - ray selected cluster sample ( gioia et al .
it has a rest - frame x - ray luminosity of @xmath34(2 - 10kev)=2.2@xmath35 erg s@xmath36(donahue et al .
1998 ) , which makes it one of the brightest x - ray clusters known .
recent observations with xmm ( gioia et al . 2004 ) find that the x - ray temperature is 7.2@xmath37 kev , in excellent agreement with tozzi et al .
( 2003 ) , vikhlinin et al .
( 2002 ) and donahue ( 2004 ; priv .
ms1054 - 0321 has an abell richness class of 3 and a line - of - sight velocity dispersion of @xmath38 km s@xmath36 ( tran et al .
the weak - lensing mass is estimated to be [email protected] @xmath40 ( jee et al .
a significant fraction of cluster galaxies , 17% , are classified as merger / peculiar on the basis of double nuclei ( separation @xmath41 kpc ) , tidal tails , and distorted morphologies ( van dokkum et al .
interestingly , many of the merging galaxies are red , bulge - dominated galaxies with no detected nebular line emission .
the fraction of blue galaxies in the cluster , calculated in the same way as the original work by butcher & oemler ( 1978,1984 ) , is [email protected] ( van dokkum et al . 2000 ) , comparable with the mean value determined for clusters at @xmath42 ( e.g. , goto et al .
2003a ) .
we have observed ms1054 - 0321 in the @xmath1 , @xmath2 and @xmath3 bandpasses ( hereafter @xmath43 , @xmath44 , and @xmath45 , respectively ) with the wide field channel ( wfc ) of acs as a part of the gto program 9290 .
a mosaic observation was constructed using a @xmath46 wfc pointing pattern , with 1 , 2 , and 2 orbits of integration in @xmath43 , @xmath44 and @xmath45 bands , respectively , at each of the four pointings .
there was nearly 1 arcmin of overlap between pointings ; thus , the core of cluster was imaged for a total of 8 orbits in @xmath44 and @xmath45 .
the field of view of the mosaic is 36.51 square arcminutes covering an approximately square geometry with a projected proper spatial dimension of 2.7 mpc at @xmath0 .
the data were processed with the `` apsis '' pipeline described by blakeslee et al .
we use the sextractor mag_auto value output by the pipeline for our photometric measure .
this magnitude is intended to give a precise estimate of total magnitudes for galaxies using kron s first moment algorithm ( blakeslee et al .
we calibrate our photometry to the ab system using photometric zero points of 26.48 ( @xmath43 ) , 25.65 ( @xmath44 ) and 24.86 ( @xmath45 ) .
these are uncertain at the @xmath470.02 mag level , which has no effect on our conclusions .
we adopt a galactic reddening for this field of @xmath48 , and @xmath49 mag based on the schlegel et al .
( 1998 ) dust maps .
multi - object spectroscopy of the galaxies in region was carried out on the keck telescope using the low resolution imaging spectrograph ( oke et al .
this survey has yielded redshifts for over 300 objects , with 143 of these being confirmed as cluster members ( van dokkum et al . 2000 ; tran et al . in prep . ) .
spectroscopic redshifts are measured for 76% of the galaxies brighter than @xmath50 ( see section [ correction ] ) .
the availability of this extensive spectroscopic database provides a significant advantage in studying the galaxy cluster ms1054 - 0321 . to obtain a large number of spectroscopic memberships in galaxy clusters at @xmath51 , a large amount of telescope time
is usually required , and thus , most of the previous studies had to rely on the photometric redshift estimation or the statistical background subtraction , both of which were the main source of bias in determining the slope of the lf at the faint end ( driver et al .
1994 ; bernstein et al . 1995 ; valotto et al .
we exclusively use the spectroscopically confirmed cluster members to construct the lf in ms1054 - 0321 .
we regard the 143 galaxies with @xmath52 as members of ms1054 - 0321 .
this corresponds to a rest - frame spread of @xmath53 km s@xmath36 about the mean cluster redshift . after correcting to the cluster restframe and subtracting the contribution to the observed velocity dispersion from the redshift measurement errors ( @xmath54 km s@xmath36 )
, we find a line - of - sight cluster velocity dispersion of 1131@xmath395 km s@xmath36 ( measured as a sigma of biweighted mean with the jackknife error estimate ) .
the derived line - of - sight velocity dispersion is consistent with previously published numbers ( e.g. , tran et al .
1999 ; goto et al .
2004b ) .
incompleteness in measuring spectroscopic redshifts ( or targeting galaxies for observation ) as a function of apparent magnitude must be compensated for in order to extract unbiased results .
[ fig : radec ] shows the sky distribution of objects with spectroscopy ( circles ) and all extended objects in the imaging data ( small dots ) .
since we have only performed spectroscopic observations in the region defined by the solid lines , we use objects within this region to define and quantify the incompleteness ( see van dokkum et al .
1999,2000 ; postman et al .
2004 ; tran et al . in prep .
for more details of the spectroscopic observations ) .
point sources are excluded from incompleteness calculations using sextractor s stellarity index ( class_star @xmath55 ) , which assigns a numerical value close to 1 if the object is a point source and values closer to 0 if it is an extended object .
a slight change in the criterion of the separation between extended and point sources ( _ e.g. , _ using a class_star value of 0.8 instead of 0.5 as the threshold ) does not affect our results . in fig .
[ fig : incompleteness ] , we show the fraction of galaxies that have a spectroscopic redshift among all galaxies in the imaging data as a function of apparent @xmath44 magnitude . as can be seen in fig .
[ fig : incompleteness ] , the completeness is almost independent of magnitude in the range of @xmath56 , indicating that the spectroscopic survey is not strongly biased and has a high degree of completeness .
regardless , when we measure a lf in the following sections , we correct for the incompleteness by weighting the observed counts by the inverse of the completeness . in the following analyses ,
we divide our cluster galaxy sample using various criteria in morphology , environment , and color . when measuring lfs of these subsamples , the incompleteness correction is re - computed using each subsample of galaxies , and then applied to the lfs of each subsample accordingly . as an example
, we plot the spectroscopic completeness for blue ( @xmath57 ) and red ( @xmath5 ) subsamples in dashed and dotted lines in fig.[fig : incompleteness ] .
it can be seen that the completeness is not strongly dependent on @xmath58 color .
when we construct a lf in different passbands ( @xmath43 , @xmath44 and @xmath45 ) , we correct the counts using the incompleteness as a function of magnitude in that passband . because different physical mechanism that act upon galaxies in dense environments have efficiencies that vary as a function of density or position , we quantify the lf as a function of these variables in a hope of understanding the underlying processes ( see a review in treu et al .
2003 ) .
for each galaxy , we measure a projected ( angular ) distance to the 5th nearest galaxy using the cluster members ( @xmath52 ) .
the number of galaxies ( @xmath59 ) within the distance is divided by the circular surface area with the radius of the distance to the 5th nearest galaxy to obtain the local galaxy density , @xmath60 ( mpc@xmath9 ) . when the projected area touches the boundary of the data ( fig .
[ fig : radec ] ) , we calculate the density by dividing the area within the data region . this parameterization is similar to the one used in previous work ( e.g. , goto et al .
2003b , c ; tanaka et al .
although our local galaxy density is a two - dimensional surface density , the contamination from both foreground and background galaxies is zero because we are using spectroscopic information . however , because we do not have spectroscopic redshifts for all the members , our measured density is an underestimate .
we use this density only for internal comparisons .
we have chosen the 5th nearest to measure local galaxy density , but we also have confirmed that using up to the 10th nearest neighbor does not change the main results described below . to assess dependencies of the lf on position within the cluster , we measure the projected distance to the cluster center , @xmath61 , for all spectroscopically confirmed member galaxies .
we use the position of the brightest cluster galaxy ( j2000 ra : 10h 56 m 59.99s , dec : -03d 37 m 36.1s ) as the center of the cluster . using the centroid of the x - ray surface brightness distribution ( neumann & arnaud 2000 ) does not significantly change the results . the virial radius of the cluster is known to be @xmath621 mpc ( tran et al . 1999 ) . in fig.[fig : distance_density ] , we compare the clustocentric distance with the local galaxy density . the solid lines connect medians .
the error bars are based on the rms in each bin . as expected
, there is a fairly good correlation between these two parameters .
note that clustocentric distance of @xmath630.9 mpc corresponds to 30 - 40 mpc@xmath9 in local galaxy density .
we use a non - linear least squares method to determine the best - fit schechter ( 1976 ) lf parameters . the lf normalization ( @xmath64 ) , faint - end slope
( @xmath65 ) , and characteristic magnitude ( m@xmath66 ) are fit simultaneously .
the input data for the fitting procedure is a binned , apparent magnitude distribution for the spectroscopically confirmed cluster members , where the number of galaxies in each bin has been corrected for incompleteness in the spectrscopic survey .
we use a magnitude bin size of 0.5 mag .
the values of the best fit lf parameters do not strongly depend on our choice of the bin size in the range @xmath67 .
the incompleteness correction is determined empirically by dividing the number of objects with measured redshifts in a given apparent magnitude bin by the total number of galaxies in that same bin .
the total number of galaxies in a bin is taken from our sextractor catalog .
the corrected number counts as a function of apparent magnitude are then converted to the corresponding absolute magnitude using a distance modulus of @xmath68 .
we do not apply k - corrections for the internal ms1054 lf comparisons .
however , whenever we compare our results to those from other studies , a k - correction is applied and the details of how the k - correction is determined are given in section [ comaprion_literature ] .
finally , we divide the corrected number counts by the effective area of the survey ( in units of projected mpc@xmath69 ; see fig.[fig : radec ] ) .
all fits are weighted by the inverse square of the error in the galaxy number in each bin and these errors are determined from poisson statistics using the original ( uncorrected ) number counts .
the best - fit schechter parameters for all samples are summarized in table [ tab : lf_parameter ] .
with the spectroscopic membership information and parameters to describe the environment of each galaxy in hand , it is straight - forward to construct a lf of the member galaxies .
it is desirable to normalize the absolute magnitude to @xmath70 for external comparison purposes , and we thus adopt the distance modulus of @xmath68 at @xmath0 . in the upper panel of fig . [
fig : lf_3mag ] , we show the lfs of the spectroscopic members .
the incompleteness in the spectroscopic sample is corrected for by using the prescriptions described in section [ correction ] .
the error bars assume poisson statistics .
the overall normalization is shown in mpc@xmath9 , obtained by dividing the corrected counts in each bin by the surface area of the survey .
the dotted , dashed , and solid lines are lfs in @xmath71 and @xmath45 filters , respectively .
the best - fit schechter functions ( schechter 1976 ) are shown in the lower panel of fig .
[ fig : lf_3mag ]
. fits to the luminosity function are performed using a non - liner least square method applied to the corrected binned number counts , where each bin is weighted by its error .
the best - fit parameters are ( @xmath72 ) = ( @[email protected] , @[email protected] ) , ( @xmath74 ) = ( @[email protected] , @[email protected] ) , and ( @xmath75 ) = ( @[email protected] , @[email protected] ) .
these results are also summarized in table [ tab : lf_parameter ] .
our lf parameters are , of course , less accurate than those derived from composite lfs from many galaxy clusters ( e.g. , goto et al
. 2002b ) .
nevertheless , characteristic features of cluster lfs can be clearly recognized . in the literature ,
lfs in bluer bands tend to have steeper faint - end slope , which is clearly seen in fig .
[ fig : lf_3mag ] . at @xmath0
, @xmath43 filter corresponds to the rest - frame @xmath76 light .
the difference between the lf in @xmath43 and the lfs in @xmath77 is a nice distinction between star - forming and old - passive galaxies .
we compare our lfs with those in the literature in section [ discussion ] .
we begin to explore the dependence of the lf on environment by looking for differences in the galaxy luminosity distribution as a function of clustocentric distance . in the upper panel of fig .
[ fig : lf_distance ] , we show the lf in the inner ( @xmath78 mpc ) and outer ( @xmath79 0.9 mpc ) regions of ms1054 - 0321 . the threshold of 0.9 mpc is chosen to correspond to the distance where galaxy properties start to undergo significant changes inwards towards the core .
the cluster core region with @xmath80 mpc has a higher density of bright galaxies and a flatter faint - end slope compared to the lf at @xmath81 mpc .
we fit a schechter function to the observed data and show the results in the lower panel of fig .
[ fig : lf_distance ] along with the errors on each data point .
these error bars assume poisson statistics .
the best - fit schechter parameters are ( @xmath82,@xmath65 ) = ( @[email protected] , @[email protected] ) , ( @[email protected] , @[email protected] ) for galaxies with @xmath81 and @xmath80 , respectively ( see also table [ tab : lf_parameter ] ) .
a steeper faint - end slope is detected for galaxies in the outer part ( @xmath81 mpc ) as compared with the faint - end slope for galaxies in the inner cluster region .
these trends are in general agreement with the findings in the composite lfs of galaxy clusters in the literature ( e.g. , goto et al
. 2002b ) .
the difference in lfs in different environments suggests that the environment indeed plays a role in determining the luminosities of cluster galaxies . in the upper panel of fig .
[ fig : lf_density ] , we plot the lf of the cluster partitioned by the local galaxy density of 30 mpc@xmath9 .
the solid line is for galaxies in regions with @xmath83 30 mpc@xmath9 and the dashed line is for galaxies in regions with @xmath84 30 mpc@xmath9 .
again , the criterion of @xmath85 30 mpc@xmath9 is chosen to correspond to the environment where galaxy properties start to exhibit substantial change relative to the field galaxy population ( see section [ sec : gdr ] ) .
the lf of galaxies in dense regions has a flatter faint - end slope than in sparse environments , consistent with the results obtained as a function of clustocentric radius .
note that we do not normalize the lfs in fig .
[ fig : lf_density ] because the sampling volume of each galaxy is different as a consequence of the way the local galaxy density is measured ( section [ density ] ) . in the lower panel ,
we fit a schechter function to each lf .
the best - fit schechter parameters are ( @xmath82,@xmath65 ) = ( @[email protected] , @[email protected] ) , ( @[email protected] , @[email protected] ) for galaxies with @xmath86 mpc@xmath9 and @xmath87 mpc@xmath9 , respectively .
these results are also given in table [ tab : lf_parameter ] .
although the errors on the parameters are substantial , the parameter values are consistent with the interpretation that the galaxy lf in dense regions has a flatter faint - end slope .
the @xmath44 filter is particularly sensitive to different stellar populations at @xmath0 because it includes the 4000 break . in particular
, the @xmath58 color does a reasonable job of distinguishing blue , star - forming galaxies from red , passive galaxies in the color - magnitude relation ( homeier et al . in prep . ) .
[ fig : cmd ] shows the color - magnitude relation for the known cluster members ( filled circles ) and for all extended objects in the field ( small dots ) .
the @xmath88 separates well the galaxies in the red - sequence from the other galaxies .
a detailed analysis of acs - based color - magnitude relation in ms1054 - 0321 will be discussed in blakeslee et al .
the upper panel of fig .
[ fig : lf_iz ] shows the lfs of galaxies with ( @xmath57 ) and ( @xmath89 ) by the dashed and solid lines , respectively .
the corresponding best - fit schechter functions are shown in the lower panel of fig .
[ fig : lf_iz ] with best - fit parameters of ( @xmath82,@xmath65 ) = ( @[email protected],@[email protected] ) , ( @[email protected] , @[email protected] ) for red and blue samples , respectively .
one potentially interesting difference is that the lf of red galaxies shows a deficit at the faint end @xmath90 .
similar deficits of faint red galaxies in high redshift galaxy samples have been reported in the literature ( e.g. , kajisawa et al .
2000 ; nakata et al . 2001
; kodama et al . 2004 ; de lucia et al . 2004 ) .
such a deficit may signal the presence of galaxy population that has yet to evolve into a faint , red phase .
we discuss this hypothesis and its implication further in section [ discussion ] .
in addition to providing superb quality photometry , a key advantage provided by the high angular resolution provided by hst and the acs , in particular , is the ability to classify galaxy morphology out to redshifts as high as @xmath0 .
one of us ( m.p . ) has visually classified the morphology of ms1054 - 0321 galaxies with @xmath91 ( see postman et al .
2004 for further details ) .
we use these classifications here to examine how the luminosity distribution depends on morphology . in fig . [
fig : lf_morph ] , we present a lf of early - type galaxies ( e+s0 or @xmath92 ) and that of late - type galaxies ( sa or later ; @xmath93 ) .
our best - fit schechter parameters are ( @xmath82,@xmath65 ) = ( @[email protected] , @[email protected] ) , ( @[email protected] , @[email protected] ) for early- and late - type galaxies , respectively .
the best - fit schechter functions are shown in the lower panel of the figure . to highlight some subtle differences
, we plot the ratio of morphologically classified galaxies as a function of absolute magnitude in fig .
[ fig : morph_ratio ] with slightly finer bins .
there may be a gradual decline of early - type fractions toward fainter absolute magnitudes , suggesting that the cluster lf is dominated by early - type galaxies .
note that at the bright - end the fraction of early - type galaxies is somewhat higher than previous estimates ( 44% ; van dokkum et al .
this is due to a difference in morphological classification process as we attempt to assign a hubble type to galaxies involved in mergers , if at all possible , to facilitate comparisons with work at lower redshift ( see postman et al .
2004 for more details in morphological classification ) . in this section ,
we investigate the giant - to - dwarf ratio ( gdr ) as functions of color , morphology , clustocentric radius , and local projected density .
we calculate the gdr as a number ratio of giant galaxies ( @xmath94 ) to dwarf galaxies ( @xmath95 ) using the expression : @xmath96 the criterion between dwarf and giant at @xmath97 is chosen to correspond approximately to the start of the deviation between the faint ends of the color - selected lfs in fig .
[ fig : lf_iz ] . due to the large distance to the cluster , we can not sample the dwarf population as faint as was probed in previous studies ( @xmath98 mag ; e.g. , ferguson & sandage 1991 ; secker & harris 1996 ) .
therefore , we take the inverse of the commonly used dwarf - to - giant ratio in order to magnify the subtle change in the parameter . in fig.[fig : gdr_iz ] , we show the gdr as a function of @xmath58 color . the trend seen in fig .
[ fig : lf_iz ] is much clearer here in the sense that the gdr is much higher for red galaxies , and it gradually declines as the galaxy color becomes bluer . in fig .
[ fig : gdr_morph ] , we show the gdr as a function of morphological type .
confirming the trend seen in fig .
[ fig : lf_morph ] , the gdr is higher for early - type ( @xmath92 ) galaxies .
these two trends are consistent with the hypothesis that the bright end of the cluster lf is dominated by red or early - type galaxies and the faint end is dominated by blue or late - type galaxies . in previous sections , we have shown that galaxies in the cluster core ( or the densest regions ) show different characteristics from galaxies in the cluster outskirts ( or the sparsest regions ) .
an important question to ask now is where ( in what environment ) these changes happen . in fig .
[ fig : gdr_radius ] , we show the gdr as a function of clustocentric distance .
the gdr is higher for galaxies in the inner regions of the cluster , consistent with the results shown in fig .
[ fig : lf_distance ] .
in addition , fig .
[ fig : gdr_radius ] shows that the gdr rises steeply at @xmath80 mpc , and flattens outside this radius .
similarly in fig . [
fig : gdr_density ] , we show the gdr as a function of the local galaxy density ( @xmath60 ) defined in section [ density ] . the same trend can be seen here .
the gdr rises with increasing local galaxy density . at @xmath99 mpc@xmath9 ,
the gdr is consistent with a constant .
if ms1054 - 0321 were roughly spherical , the de - projected rise in the gdr at small radii would be even steeper . however , ms1054 - 0321 is far from spherical - as evidenced by the distributions of the x - ray gas ( donahue et al .
1998 ; neumann & arnaud 2000 ; jeltema et al .
2001 ) , the galaxies and the inferred mass ( luppino & kaiser 1997 ; van dokkum et al . 1999,2000 ; hoekstra et al .
2000 ; jee et al .
we begin our discussion by comparing the derived lf of ms1054 - 0321 with that for cluster galaxies in the local universe .
since the majority of previous work has measured the lfs in the rest - frame @xmath100-band , we convert our @xmath82 to the rest - frame @xmath100-band for a fair comparison .
the conversion is reliable since the shift between the sampled rest - frame at @xmath0 in the @xmath44 filter is quite close to the @xmath101 @xmath102band spectral coverage . as most of the cluster galaxies have passive seds ( blakeslee et al .
2003 ) , we use an elliptical galaxy sed template from bentez et al .
( 2004 ) in order to compute the transformation to the @xmath102band .
the transformed characteristic magnitude for ms1054 - 0321 in the rest - frame @xmath102band is @xmath103 ( @xmath70 ) .
while not all cluster galaxies have the sed of the elliptical galaxy , the uncertainty introduced by using an elliptical sed for the rest - frame conversion is small because ( i ) the majority of the galaxies in clusters do exhibit elliptical seds and ( ii ) a spiral sed ( sbc in bentez et al .
2004 ) yields a transformation that differs by only 0.126 mag , comparable to the statistical error .
postman et al .
( 2001 ) measured @xmath104 for three clusters with a median redshift of 0.859 ( all magnitudes we quote in this section are converted to @xmath70 and are on the vega magnitude system ) , in excellent agreement with our ms1054 - 0321 results . at low redshift , colless ( 1989 ) measured ( @xmath105 ) = ( @xmath7320.12 , @xmath731.24 ) using 14 nearby rich clusters .
lumsden et al .
( 1997 ) found ( @xmath105 ) = ( @[email protected] , @[email protected] ) at @xmath106 .
valotto et al .
( 1997 ) found ( @xmath105 ) = ( @[email protected] , @[email protected] ) at @xmath107 .
rauzy , adami , & mazure(1998 ) studied 28 rich clusters with a mean redshift of 0.07 , and found ( @xmath105 ) = ( @[email protected] , @[email protected] ) .
the transformation from @xmath100 to @xmath108 is @xmath109 ( blair & gilmore 1982;cunow 1993 ) .
since a typical @xmath110 color is 0.96 for early - type galaxies ( fukugita , shimasaku , & ichikawa 1995 ) , we use @xmath111 . applying this conversion , we find 0.46 - 0.71 mag of evolution in @xmath112 in clusters between @xmath101 and 0.83 .
we summarize these comparisons in fig . [
fig : m_b ] . the results from dahl ' en et al .
( 2002 ) appear to have a larger scatter because we plot each cluster as an individual point .
however , some caution is required in the interpretation of any lf evolution because ms1054 - 0321 is extremely rich with @xmath34(2 - 10kev)=2.2@xmath35 erg s@xmath36 , and therefore , is perhaps not a progenitor of the well - studied low redshift clusters .
on the other hand , there is significant disagreement between studies on the faint - end slope , @xmath65 .
we find a much flatter faint - end slope . as our study
is limited to a single cluster some of the disagreement may be associated with peculiarities inherent to ms1054 - 0321 .
however , the determination of the faint - end slope depends critically on a good background subtraction and reliable completeness corrections .
hence , some of the difference may also reflect the fact that our analyses are based exclusively on confirmed cluster members .
if the deficit of faint red galaxies at high redshift turns out to be real , then some of the difference between our slope and the mean faint end lf slope found at lower redshift may also be related to this under - represented population component at @xmath113 .
the flat slope in ms1054 - 0321 may thus be suggestive of a population of faint red galaxies that is going to be created or accreted into clusters between @xmath114 .
a similar lack of faint galaxies at high redshift was reported by kajisawa et al .
( 2000 ) and nakata et al .
( 2001 ) in the 3c 324 field , and by de lucia et al .
( 2004 ) in the eso distant cluster survey at @xmath115 and 0.8 .
the physics behind the origin of such a galaxy population remains to be explored .
a hint comes from drinkwater et al.(2003 ) who finds dwarf galaxies without any disk , possibly tidally stripped , in the fornax cluster ( also see bekki et al . 2003 ; mieske et al .
2004 ) . however , kodama et al .
( 2004 ) reported the lack of faint red galaxies in the high redshift field regions using the subaru - xmm deep data , and thus , the creation of faint galaxies may not be cluster specific phenomena .
the dependence of the lf on color and morphology in ms1054 - 0321 are shown in figures [ fig : lf_iz ] , [ fig : lf_morph ] , and [ fig : morph_ratio ] .
in the local universe , yagi et al . ( 2002 ) studied lfs of 10 nearby clusters and finds that the lf of galaxies with exponential profiles has a steeper faint - end slope than that for galaxies with @xmath116 profiles .
similarly , boyce et al .
( 2001 ) studied the lf of abell 868 ( @xmath117 ) and finds that the presence of dwarf irregular galaxies causes the faint - end slope of the cluster lf to steepen .
goto et al . ( 2002b ) used 204 galaxy clusters found in the sloan digital sky survey ( goto et al .
2002a ) to construct composite lfs in @xmath118 and @xmath119 filters .
when they divide lfs by morphology and color , the faint - end of the lfs were steeper for late - type ( or bluer ) galaxies .
we also find that the faint - end slope of the lf in ms1054 - 0321 is always steeper for blue ( @xmath57 ) galaxies .
these findings are consistent with the hypothesis that the bright end of the cluster lf is dominated by red , early - type galaxies and the faint - end by blue , late - type galaxies .
considering that field galaxies at around @xmath20 have very different properties from their local counterpart ( kajisawa & yamada 2001 ; cross et al .
2004 ) , it is revealing that these typical characteristics of cluster lfs are already firmly established in a high redshift galaxy cluster at @xmath0 , suggesting that this massive cluster most likely began to form at redshifts considerably greater than unity . a study on a large
, statistically complete sample of high redshift clusters is clearly needed to assess whether the presence of similar lf features is wide spread in clusters that existed 6 8 gyr ago . to gain a better picture of how the cluster environment impacts galaxy - scale stellar systems
, we should compare our ms1054 - 0321 lf results to those from low - density regions at similar redshifts .
postman , lubin , & oke ( 2001 ) proposed that the evolution of the @xmath100-band lfs of field galaxies are consistent with the form @xmath120 , where @xmath121 .
our cluster @xmath122 roughly agrees with the equation within the errors .
cross et al .
( 2004 ) have measured lfs of morphologically selected field e / s0 galaxies at @xmath123 and find @xmath124 with @xmath125 . in the subaru
deep survey , kashikawa et al .
( 2003 ) derives ( @xmath126)=(@xmath127 , @xmath128 ) for field galaxies at @xmath51 .
their @xmath122 is slightly fainter than the value we obtained for the cluster lf .
the @xmath65 is much steeper than the cluster value we found .
these differences are comparable with the differences we find when we compare lf of galaxies in the inner region of the cluster with those in the outskirts of the cluster , i.e. , lower density regions have a fainter @xmath129 and steeper faint - end tail . again , these results are consistent with the hypothesis that cluster regions are dominated by the bright galaxies , and faint galaxies are more numerous in the field regions . in figs . [
fig : gdr_radius ] and [ fig : gdr_density ] , we demonstrate that the gdr starts to change at around @xmath130 mpc or @xmath131 mpc@xmath9 . in fig .
[ fig : radec ] , the crosses denote the projected positions on the sky of cluster members in this transition region with @xmath132 mpc@xmath9 . as can be seen , this density regime corresponds to the outskirts of the cluster .
this environment coincides with that where the relative morphological population fractions and star formation rate ( sfr ) of galaxies starts to exhibit significant change , at least in @xmath7 surveys .
for example , goto et al . ( 2003b ) studied the morphology - density relation using a large sample from the sdss to find that galaxy morphology starts to change around the virial radius ( girardi et al .
1998 ) , which is quite similar to the environments we found ( see also postman & geller 1984 ) .
tanaka et al.(2004 ) studied the environmental dependence of galaxy sfr using a similar sample from the sdss and found that the sfr starts to change at a similar environment .
the passive spiral galaxies spiral galaxies with very little or no star formation activity are also found to preferentially exist in the same environment ( goto et al .
2003c ; yamauchi & goto 2004 ) .
the galaxies in ms1054 - 0321 appear to follow similar relationship between color and density / clustocentric radius .
[ fig : iz_distance ] and [ fig : iz_density ] show the @xmath58 color of cluster galaxies as a function of @xmath61 and @xmath60 .
the solid line connects median values in each bin .
the galaxy colors start to change significantly at around @xmath630.9 mpc or @xmath8510 - 40 mpc@xmath9 , which is the same environment where the gdr starts to change .
postman et al .
( 2004 ) find that the morphological composition of galaxies starts to change at a similar environment at @xmath51 as well .
a substantial volume of evidence thus suggests there is at least one physical mechanism that begins to become important at around @xmath630.9 mpc or @xmath8510 - 40 mpc@xmath9 and plays a role in altering the luminosity , sfr , and morphology of cluster galaxies .
various physical mechanisms have been proposed to explain this cluster galaxy transformation .
these mechanisms include ram - pressure stripping of cold gas by the intra - cluster medium ( gunn & gott 1972 ; farouki & shapiro 1980 ; kent 1981 ; fujita & nagashima 1999 ; abadi , moore & bower 1999 ; quilis , moore & bower 2000 ; fujita & goto 2004 ) ; galaxy harassment via high speed impulsive encounters ( moore et al .
1996 , 1999 ; fujita 1998 ) ; cluster tidal forces ( byrd & valtonen 1990 ; valluri 1993 ; fujita 1998 ; gnedin 2003a , b ) which distort galaxies as they come close to the centre ; interaction / merging of galaxies ( icke 1985 ; lavery & henry 1988 ; mamon 1992 ; makino & hut 1997 ; bekki 1998 ; finoguenov et al .
2003 ) ; evaporation of the cold gas in disc galaxies via heat conduction from the surrounding hot icm ( cowie & songaila 1977 ; fujita 2004 ) ; and a gradual decline in the sfr of a galaxy due to the stripping of hot halo gas ( strangulation or suffocation ; larson , tinsley & caldwell 1980 ; bekki et al .
2002 ; kodama et al .
2001b ; finoguenov et al .
2003 ) . among these processes , ram - pressure stripping and the evaporation of the cold gas
may be disfavored by our results . since
0.9 mpc is almost the virial radius , the icm density will be too low for these mechanisms to work with any substantial efficiency ( kodama et al . 2001b ; treu et al .
2003 ; also see mamon et al .
2004 ; moore et al .
galaxy - galaxy merging would be consistent with both of the above results .
however , for galaxy - galaxy merging to be an efficient mechanism the relative velocity of the merging galaxies must be comparable with their internal velocity dispersion ( makino & hut 1997 ) . thus , such mergers would have to take place in infalling , low velocity dispersion groups or at an epoch when the cluster mass was still comparable with that on group mass scales .
all we can conclude for now is that merging remains a viable explanation , but not necessarily the only explanation , for the observed trends seen in ms1054 - 0321 as a function of density and position .
in an effort to gain further insight into the physical processes associated with cluster galaxy formation and evolution , we have explored correlations between the global parameters of the galaxy luminosity function and the properties of galaxies in the cluster ms1054 - 0321 at @xmath0 .
our study is based on 143 spectroscopically confirmed cluster members imaged with the acs in the @xmath133 and @xmath45 filters . due to the wide field of view and high resolution imaging capability of the acs
, we were able to study cluster lfs as a function of environment , color , and morphology .
our key findings are summarized below . 1 .
the faint - end slope of the lfs is steeper in the bluest band ( fig.[fig : lf_3mag ] ) .
the lf of galaxies in the inner part of the cluster ( @xmath80 mpc ) has a flatter faint - end slope than that in the outer regions ( @xmath81 mpc ; fig.[fig : lf_distance ] ) .
the corresponding trend can be seen when the lf is measured as a function of the local galaxy density ( fig .
[ fig : lf_density ] ) .
the fraction of early - type galaxies is largest at the bright end of the lf and gradually decreases toward fainter magnitudes ( figs .
[ fig : lf_morph ] and [ fig : morph_ratio ] ) .
4 . the trends seen in ( 1 ) , ( 2 ) , and ( 3 ) are generally in agreement with observations of local galaxy clusters showing that the cluster core regions are dominated by bright early - type galaxies .
hence , a lf similar to what is seen at low redshift is well established in ms1054 - 0321 , suggesting that , on average , the galaxies in this massive cluster began their formation at redshifts considerably greater than unity .
when the lf is partitioned by @xmath58 color , we find a marginally significant ( @xmath134 ) deficit of faint red galaxies ( @xmath5 , @xmath6 ) .
the deficit is inferred as well when we analyze the lf as a function of morphology a possible lack of early - type galaxies in the faintest bin ( fig .
[ fig : lf_morph ] ) .
if real , these trends may suggest that faint , red , early - type galaxies have not been yet created in ms1054 - 0321 .
a deficit of such galaxies has not been seen in low-@xmath119 clusters .
the giant - to - dwarf ratio exhibits a strong radial dependence beginning at around the virial radius or a strong density dependence when the density increases above @xmath13530 mpc@xmath9 .
the ratio is higher in the inner parts ( or higher density regions ) of the cluster ( figs .
[ fig : gdr_radius ] and [ fig : gdr_density ] ) .
this environment roughly corresponds to the same location where galaxy color and the morphology start to vary significantly with position and density ( figs .
[ fig : iz_distance ] and [ fig : iz_density ] ) .
a physical process effective at this environment , such as galaxy merging , may be responsible for the cluster galaxy evolution in these transition regions .
we thank the anonymous referee for providing constructive and insightful comments that have greatly improved the clarity of this paper .
we also thank pieter g. van dokkum for very useful comments .
the acs was developed under nasa contract nas 5 - 32865 , and this research is supported by nasa grant nag5 - 7697 .
some of the results in this paper were based on computations performed on equipment generously provided through a grant from sun microsystems , inc . .
the number ratio of galaxies with spectroscopic redshift with respect to all the galaxies in the spectroscopic area of the imaging data is plotted against @xmath44 mag .
the error bars assume poisson statistics .
the bin size is 1.0 mag .
the dashed and dotted lines are for blue ( @xmath57 ) and red ( @xmath5 ) samples , respectively . ] , @xmath136 and @xmath137 filters in ab magnitude , respectively .
the incompleteness in the spectroscopic sample is corrected using the method described in section [ correction ] .
the lower panel shows best - fit schechter functions , whose parameters are presented in table [ tab : lf_parameter ] .
, title="fig : " ] , @xmath136 and @xmath137 filters in ab magnitude , respectively .
the incompleteness in the spectroscopic sample is corrected using the method described in section [ correction ] .
the lower panel shows best - fit schechter functions , whose parameters are presented in table [ tab : lf_parameter ] .
, title="fig : " ] , respectively .
since both the high density and low density samples are specially spread over the entire sky area , the same ( over all ) incompleteness corrections are applied to the both lfs .
, title="fig : " ] , respectively .
since both the high density and low density samples are specially spread over the entire sky area , the same ( over all ) incompleteness corrections are applied to the both lfs .
, title="fig : " ] color of 0.5 .
the corrections for the incompleteness of spectroscopic observation is corrected for each subsample of galaxies .
, title="fig : " ] color of 0.5 .
the corrections for the incompleteness of spectroscopic observation is corrected for each subsample of galaxies .
, title="fig : " ] : e / s0 , @xmath138:late type .
corrections for incompleteness are applied separately for each morphological sample .
, title="fig : " ] : e / s0 , @xmath138:late type .
corrections for incompleteness are applied separately for each morphological sample .
, title="fig : " ] | we present new measurements of the galaxy luminosity function ( lf ) and its dependence on local galaxy density , color , morphology , and clustocentric radius for the massive @xmath0 cluster ms1054 - 0321 .
our analyses are based on imaging performed with the advanced camera for surveys ( acs ) onboard the hubble space telescope ( hst ) in the @xmath1 , @xmath2 and @xmath3 passbands and extensive spectroscopic data obtained with the keck low - resolution imaging spectrograph .
our main results are based on a spectroscopically selected sample of 143 cluster members with morphological classifications derived from the acs observations .
our three primary findings are ( 1 ) the faint - end slope of the lf is steepest in the bluest filter , ( 2 ) the lf in the inner part of the cluster ( or highest density regions ) has a flatter faint - end slope , and ( 3 ) the fraction of early - type galaxies is higher at the bright end of the lf , and gradually decreases toward fainter magnitudes .
these characteristics are consistent with those in local galaxy clusters , indicating that , at least in massive clusters , the common characteristics of cluster lfs are established at @xmath0 .
these results provide additional support for the hypothesis that the formation of galaxies in ms1054 - 0321 began at redshifts considerably greater than unity .
we also find a 2@xmath4 deficit of intrinsically faint , red galaxies ( @xmath5 , @xmath6 ) in this cluster .
although the significance is marginal , this trend may suggest that faint , red galaxies ( which are common in @xmath7 rich clusters ) have not yet been created in this cluster at @xmath0 .
the giant - to - dwarf ratio in ms1054 - 0321 starts to increase inwards of the virial radius or when @xmath8 mpc@xmath9 , coinciding with the environment where the galaxy star formation rate and the morphology - density relation start to appear
. a physical process that begins to become effective at around the virial radius or @xmath1030 mpc@xmath9 may thus be responsible for the evolution of color and luminosity of cluster galaxies . |
the hamiltonian of a two - level quantum emitter ( qe ) interacting with a quantized mode of an optical cavity can be written as : @xmath78 where @xmath79 and @xmath80 are fermionic creation and annihilation operators for the qe while @xmath81 and @xmath82 are bosonic creation and annihilation operators for the cavity .
@xmath83 is the qe frequency , @xmath84 is the cavity frequency , and @xmath12 is the qe - cavity coupling strength .
we consider up to one excitation in the system so that within the rotating - wave approximation the dynamics involved only the states @xmath85 .
as no coherent coupling occurs between @xmath86 and the other states , it is sufficient to study the dynamics within the basis formed by @xmath87 , in which the hamiltonian reads @xmath88 where @xmath89 is the detuning between the qe and the cavity .
the density matrix can be written as @xmath90 its evolution is described by the following linbladt master equation @xcite : @xmath91 \label{master}\ ] ] @xmath92 = \frac{i}{\hbar}[\hat{\rho},\hat{h } ] + \hat{\mathcal{l}}_{\text{qe } } + \hat{\mathcal{l}}_{\text{cav } } + \hat{\mathcal{l}}_{\text{deph}}\ ] ] where the dissipative terms describing respectively the qe decay , the cavity decay and pure dephasing reads in the @xmath87 basis @xmath93 = -\gamma \begin{pmatrix } \rho_{ee } & \rho_{ec}/2\\ \rho_{ce}/2 & 0 \end{pmatrix}\ ] ] @xmath94 = -\kappa \begin{pmatrix } 0 & \rho_{ec}/2\\ \rho_{ce}/2 & \rho_{cc } \end{pmatrix}\ ] ] @xmath95 = -\gamma^ * \begin{pmatrix } 0 & \rho_{ec}/2 \\ \rho_{ce}/2 & 0 \end{pmatrix}\ ] ] an initial excitation of the qe is assumed , i.e. @xmath96 . the evolution of the density matrix is then computed using @xmath97
in order to conveniently express the two - time correlators , we make use of the non - equilibrium green s function formalism @xcite . here the lesser , greater and retarded green s function are respectively defined by : @xmath98 @xmath99 @xmath100 , \ ] ] where @xmath101 is a time difference .
note that for simplicity we have dropped in the above definitions the factor @xmath102 involved in standard definitions @xcite .
the retarded and lesser self - energies describing the dissipative terms are expressed as : @xmath103 @xmath104 for such time - independent hamiltonian the retarded green s function depends only of one variable .
it can be expressed in angular frequencies as : @xmath105 the dyson s equation for the retarded green s function reads : @xmath106 @xmath107 from kadanoff - baym equations , which describe the equations of motion of the green s functions @xcite , one can easily derive the following relation in the case of markovian self - energies for @xmath108 : @xmath109 from this relation we can extract the cavity correlations @xmath110 hence the calculation of the two - time correlators is splitted into the calculation of two one - time operators , namely the density matrix @xmath111 on one hand and the retarded green s function @xmath112 on the other hand .
@xmath112 can be computed by fourier transforming eq .
[ gr ] , or by solving its equation of motion which reads @xmath113 \hat{g}^r(\tau ) .
\label{eomgr}\ ] ]
in the following discuss three limiting cases of dissipative cavity quantum electrodynamics ( cavity - qed ) and derive the degree of photon indistinguishability in each of these cases .
we assume a perfect resonance ( i. e. @xmath114 ) between the qe and the cavity .
coherent - coupling regime between the qe and the cavity occurs if @xmath115 . in the limit @xmath116
, we derive below an analytical expression for the indistinguishability . in this limit ,
the coherent part of the dynamics ( i.e. rabi oscillations ) is much faster than the incoherent part ( i.e. population and phase decay ) .
an approximate solution of the dynamics is obtained by decoupling these two timescales .
the density matrix then reads : in the bad cavity limit ( i.e. @xmath42 ) , we can eliminate the cavity population ( i.e. @xmath133 ) in the above rate equations , leading to @xmath134 on the other hand , approximation on the retarded green s function involved in eq .
[ cc ] can also be derived . for @xmath135
, one can adiabatically eliminate the off - diagonal term @xmath136 in eq .
[ eomgr ] : @xmath137 so that the second term in eq .
[ cc ] reads for @xmath135 @xmath138 on the other hand , the first term in eq .
[ cc ] is given by @xmath139 for @xmath135 , only the former term is not vanishing : @xmath140 the last expression indicates that in this regime the cavity correlations do indeed follow the qe correlations , with : @xmath141 @xmath142 as the integral of the correlations ( eq . 2 of the paper ) is dominated by delay times such as @xmath135 , one finds for the indistinguishability @xmath45 in the limit @xmath143 , we show below that the cavity correlations are dominated by the terms diagonal in the cavity mode . in this limit , the projection of the retarded green s function on the cavity mode reads @xmath144 where @xmath51 is the linewidth of the cavity - like eigenstate which can be calculated from from eq .
[ gr ] : @xmath145 in eq .
[ eomgr ] , we can adiabatically eliminate @xmath146 for @xmath147 , which gives @xmath148 and thus providing an upper limit for the second term in eq .
[ cc ] : @xmath149 on the other hand , eq .
[ cohm ] provides an upper bound for @xmath150 : @xmath151 which combined with eq .
[ grccgc ] gives @xmath152 eqs .
[ ceec ] and [ cccc ] show that in this regime @xmath153 dominates over @xmath154 for @xmath147 : @xmath155 we can hence express the indistinguishability as if the cavity acts as an effective emitter : @xmath156 by solving the rate equations for populations ( eq . [ coupled ] ) , and plugging in @xmath50 in the above expression , we find after some algebra @xmath157 | we investigate theoretically the generation of indistinguishable single photons from a strongly dissipative quantum system placed inside an optical cavity .
the degree of indistinguishability of photons emitted by the cavity is calculated as a function of the emitter - cavity coupling strength and the cavity linewidth . for a quantum emitter
subject to strong pure dephasing , our calculations reveal that an unconventional regime of high indistinguishability can be reached for moderate emitter - cavity coupling strengths and high quality factor cavities . in this regime , the broad spectrum of the dissipative quantum system is funneled into the narrow lineshape of the cavity .
the associated efficiency is found to greatly surpass spectral filtering effects .
our findings open the path towards on - chip scalable indistinguishable - photon emitting devices operating at room temperature .
indistinguishable single photons are the building blocks of various optically - based quantum information applications such as linear optical quantum computing @xcite , boson sampling @xcite , quantum teleportation @xcite or quantum networks @xcite .
indistinguishable photons are usually generated either using parametric down conversion @xcite , or alternatively directly from a single two - level quantum emitter such as atoms , color centers , quantum dots or organic molecules @xcite .
parametric down conversion is presently the most mature technology available , but the usual low efficiency of the nonlinear processes is a severe limitation to the scalability of such sources . on the other hand , sources based on single solid - state quantum systems
have been greatly developped in the last decade , as they hold the promise to combine indistinguishable , on - demand , energy - efficient , electrically drivable and scalable characteristics . however , except at cryogenics temperature , solid - state systems emitting single - photons are subject to strong pure dephasing processes @xcite , making them at first view inappropriate for quantum applications requiring photon indistinguishability .
a two - level quantum emitter ( qe ) coupled only to vacuum fluctuations should emit perfectly indistinguishable photons .
however , as soon as pure dephasing of the qe occurs , the degree of indistinguishability of the emitted photons is reduced to @xcite @xmath0 where @xmath1 is the population decay rate , @xmath2 the pure dephasing rate , and @xmath3 the total dephasing rate . for solid - state
qe emitting photons at room temperature such as color centers , quantum dots or organic molecules , pure dephasing rates are typically several orders of magnitude larger than the population decay rate ( typically ranging from 3 to 6 orders of magnitude ) @xcite . hence the intrinsic indistinguishability given by eq .
[ itls ] is almost zero . a possible way to enhance
the indistinguishability is to spectrally filter the emitted photons a posteriori .
however , this linear - filter strategy leads to a very low efficiency .
engineering of both efficiency and indistinguishability are possible by placing the dissipative qe in an optical cavity @xcite .
a usual strategy is then to use the purcell effect to enhance the spontaneous emission , as in eq .
[ itls ] an increase in @xmath4 results in an increase of @xmath5 .
however , reaching purcell factors larger than @xmath6 for room - temperature solid - state systems appears to be well beyond the present experimental state of the art . in this letter
we propose a realistic and robust way to generate highly indistinguishable photons from strongly dissipative qe ( i.e. for @xmath7 ) .
the idea is to exploit a cavity - quantum - electrodynamics ( cavity - qed ) regime of low cavity linewidth and moderate cavity - emitter coupling , in which the broad spectrum of the dissipative qe is funneled into the narrow emission line of the cavity . in this regime ,
high indistinguishability is predicted together with efficiencies orders of magnitude higher than spectral filtering .
insights into the full quantum calculation are gained by semiclassical derivations of indistinguishability in limiting cases of dissipative cavity qed . as depicted in fig.[schema ]
, we consider a two - level qe system @xmath8 coupled to a cavity mode whose fock states are denoted @xmath9 .
all the dissipative terms are assumed to be described within the markov approximation @xcite .
the relevant parameters are : the qe decay rate @xmath4 ( which may include radiative as well as non - radiative components ) , the cavity decay rate @xmath10 , the pure dephasing rate @xmath11 ; @xmath12 is the dipolar coupling between the qe and the cavity mode ( see fig.[schema ] ) .
the emitter - cavity detuning is set to zero ( i.e. perfect resonance ) . for simplicity , we assume an instantaneous excitation of the qe , so that only one quantum of excitation can be transferred to the cavity . within the rotating - wave approximation
, it is therefore sufficient to investigate the dissipative quantum dynamics in the two - dimensional hilbert space formed by @xmath13 .
the degree of indistinguishability of photons can be defined by the probability of two - photon interference in a hong - ou - mandel experiment @xcite . for a single - photon emitter ,
this indistinguishability figure of merit reads @xcite : @xmath14 where @xmath15 are the ladder operators of the em mode in which the photons are emitted .
this equation imposes the necessary condition for perfectly indistinguishable photons that time correlations of the em field decay the same way as the intensity , i.e. that photons are fourier - transform limited .
the calculation of the above quantities can be separated into two steps ( see the supplemental material @xcite ) .
first , we calculate the evolution of the density matrix @xmath16 by solving the lindblad equation . @xmath17 where @xmath18 is the total liouvillian of the system @xcite . secondly
, we calculate the retarded green s function , which reads in the @xmath13 basis : @xmath19 the two - time correlator of the cavity field can be expressed as a product of the retarded propagator @xmath20 and the density matrix @xcite : @xmath21 in fig .
[ imap ] , we report calculations for strongly dissipative emitters verifying @xmath22 .
this is a typical value for a solid - state qe at room temperature , and the results are qualitatively similar for any strongly dissipative emitters verifying @xmath7 . without any cavity , the degree of indistinguishability would then amount to @xmath23 according to eq .
[ itls ] . in fig .
[ imap](a ) , @xmath5 is plotted as a function of the coupling @xmath12 and the cavity linewidth @xmath10 while @xmath4 and @xmath11 are fixed .
two regions of high indistinguishability are found , which are discussed below .
the one in the upper - right corner corresponds to very large couplings @xmath12 and broad cavities such as @xmath24 and @xmath25 , which is extremely challenging to reach experimentally for strongly dissipative emitters ( i.e. for large values of @xmath11 ) .
the other region of high indistinguishability , in the lower - left corner , appears for good cavities for @xmath26 together with moderate or small coupling values @xmath12 .
as these values are within experimental reach , this unconventional regime is highly attractive for the generation of indistinguishable photons . ) , ( b ) efficiency @xmath27 and ( c ) funneling ratio @xmath28 , plotted as a function of the cavity linewidth @xmath29 and the emitter - cavity coupling strength @xmath30 for a fixed ratio @xmath31 .
the funneling ratio is defined as @xmath32 .
solid lines delimit three different regimes discussed in the text : coherent coupling ( `` 1 '' ) , incoherent coupling and bad cavity regime ( `` 2 '' ) , and incoherent coupling and good cavity regime ( `` 3 '' ) . , title="fig:",scaledwidth=43.0% ] ) , ( b ) efficiency @xmath27 and ( c ) funneling ratio @xmath28 , plotted as a function of the cavity linewidth @xmath29 and the emitter - cavity coupling strength @xmath30 for a fixed ratio @xmath31 . the funneling ratio is defined as @xmath32 .
solid lines delimit three different regimes discussed in the text : coherent coupling ( `` 1 '' ) , incoherent coupling and bad cavity regime ( `` 2 '' ) , and incoherent coupling and good cavity regime ( `` 3 '' ) . ,
title="fig:",scaledwidth=43.0% ] ) , ( b ) efficiency @xmath27 and ( c ) funneling ratio @xmath28 , plotted as a function of the cavity linewidth @xmath29 and the emitter - cavity coupling strength @xmath30 for a fixed ratio @xmath31 .
the funneling ratio is defined as @xmath32 .
solid lines delimit three different regimes discussed in the text : coherent coupling ( `` 1 '' ) , incoherent coupling and bad cavity regime ( `` 2 '' ) , and incoherent coupling and good cavity regime ( `` 3 '' ) .
, title="fig:",scaledwidth=43.0% ] to get a physical insight into the calculated indistinguishability , we divide the ( @xmath10,@xmath12 ) space into three regimes labelled from `` 1 '' to `` 3 '' in fig .
[ imap ] and study the corresponding limiting cases below .
first , we can distinguish between the qe - cavity coherent regime occuring for @xmath33 and the incoherent regime @xmath34 . in the coherent regime ,
labelled `` 1 '' in fig .
[ imap ] , the dynamics consists of damped rabi oscillations which evolves into an incoherent population of the two polariton modes ( i.e. mixed qe - cavity state ) . in the limit where @xmath35 , the indistinguishability degree reads @xcite : @xmath36 as this expression is independent of @xmath12 , increasing @xmath12 alone does not allow to reach arbitrarily large indistinguishability , as seen on fig .
[ imap ] . on the contrary
, nearly - perfect indistinguishability occurs in the coherent regime only if @xmath37 . in the incoherent limit ( @xmath38 )
, the dynamics of the system can be described in terms of incoherent population transfer with an effective transfer rate between the qe and the cavity given by @xcite : @xmath39 within this incoherent regime , we can further define a bad - cavity regime for @xmath40 ( labelled `` 2 '' in fig .
[ imap ] ) and a good - cavity regime for @xmath41 ( labelled `` 3 '' in fig . [ imap ] ) @xcite . in the bad - cavity limit @xmath42
, the cavity can be adiabatically eliminated , and its sole effect is to add a new channel of irreversible radiative emission at a rate @xmath43 .
reabsorption by the qe is then negligible .
the dynamics of the coupled system can then be described by the one of an effective qe with a decay rate @xmath44 .
applying eq .
[ itls ] to this effective qe leads to an indistinguishability of @xmath45 within this regime of incoherently - coupled bad cavity , the usual strategy to increase indistinguishability is basically to maximize @xmath43 .
this can be done by increasing @xmath12 and/or minimizing @xmath10 . however , from eq .
[ ibc ] , near - unity indistinguishability requires @xmath46 and consequently @xmath47.it is found that the coupling strength @xmath12 has to exceed @xmath11 by nearly one order of magnitude in order to reach an indistinguishability value of @xmath48 .
reaching such coupling is technologically extremely challenging for solid - state emitters under ambient temperature . on the other hand , the incoherent good - cavity regime ( labelled `` 3 '' in fig . [ imap ] )
occurs for @xmath41 . in this regime , the cavity can store the photons within a time scale comparable to or longer than the qe dephasing time .
the cavity itself then acts as an effective emitter incoherently pumped by the qe @xcite , so that the cavity field correlations read @xmath49 where @xmath50 is the population of the cavity mode and @xmath51 is the linewidth of the cavity - like eigenstate . from eq .
[ retardedgf ] , one can derive that @xmath52 , which is the sum of the cavity decay rate @xmath10 into em modes plus the incoherent reabsorption rate @xmath43 between the cavity and the qe . by solving the population rate equations and by plugging in the resulting cavity dynamics @xmath50 in eqs .
[ corrgc ] and eq .
[ iformula ] , one finds an indistinguishability of @xcite : @xmath53 consequently , large indistinguishability occurs in this regime for @xmath26 and @xmath54 ( i.e. @xmath55 ) , in agreement with the full calculation shown on fig .
[ imap](a ) .
this can be understood by noting that two ingredients are involved in the degradation of indistinguishability in this good - cavity regime where the cavity acts as the effective emitter .
the first point is that the initial incoherent feeding of the cavity occurs on a time scale @xmath56 , producing a time uncertainty in the population of the cavity .
hence @xmath10 has to be kept small compared to @xmath4 in order to prevent such time - jittering effect , in analogy to the incoherent pumping of qe via high energy states @xcite .
the second point is that , after the initial filling of the cavity , incoherent exchange processes between the qe and the cavity can still occur .
however , back and forth incoherent hopping between the cavity and the qe leads to the dephasing of the photons emitted by the cavity . to prevent such detrimental hopping ,
@xmath54 is required .
we now discuss the efficiency of the photon emission from the cavity mode , i.e. the probability to have emission by the cavity mode per initial excitation of the qe .
the efficiency of photon emission in the cavity mode is given by @xmath57 in fig .
[ imap](b ) the efficiency @xmath27 is plotted as a function of the cavity linewidth @xmath10 and the emitter - cavity coupling strength @xmath12 .
near - unitary ( i.e. on - demand ) efficiencies are obtained in the upper - right corner . in the weak - coupling regime , we find @xmath58 efficiencies larger than 0.5 are typically obtained for @xmath59 and @xmath60 .
this is compatible with high indistinguishability in the region of high @xmath12 and high @xmath10 values ( i.e. right - upper corner in fig .
[ imap ] ) , but not in the good cavity regime ( i.e. region `` 3 '' in fig .
[ imap ] ) . nevertheless ,
as discussed in the following , the product of efficiency and indistinguishability @xmath61 in the good cavity regime can still be way above the one obtained by any linear spectral filtering technique .
let us consider a linear spectral filter , with a narrow spectral range @xmath62 , through which the spectrum of the broad qe is sent .
we assume @xmath63 .
the output efficiency is bounded by @xmath64 . due to the fourier - transform condition ,
the corresponding indistinguishability is bounded by @xmath65 .
hence the efficiency - indistinguishability product for spectral filtering can not exceed @xmath66 . in order to compare @xmath67 in the present cavity - qed scheme with the upper limit for spectral filtering
, we define a cavity - funneling factor @xmath68 such that @xmath28 values larger than unity necessarly indicate a spectral cavity - funneling effect .
@xmath28 indicates the minimum enhancement ratio of @xmath67 with respect to any spectral - filtering effect . in practice
this enhancement will be larger since light emitted from a cavity can be very efficiently collected @xcite , in contrast to free - space spontaneous emission . in fig .
[ imap](c ) , the funneling @xmath28 is plotted in the same parameter range ( @xmath10,@xmath12 ) as previous plots . only the values satisfying the cavity - funneling condition of @xmath69
are shown .
it appears clearly that almost - perfect indistinguishability in the good cavity regime is compatible with cavity funneling . in fig .
[ comp ] , @xmath5 , @xmath27 and @xmath28 are plotted as a function of @xmath10 for a fixed value of @xmath12 .
the full calculation is found to be in good agreement with the above formulae for the incoherent regime .
it illustrates the necessary trade - off between indistinguishability and efficiency in the good - cavity regime , where a clear maximum of the funneling factor occurs .
the large calculated values for @xmath28 are signatures of a very efficient redirection of the qe spectrum into the unperturbed cavity spectrum of linewidth @xmath10 .
( full calculation in black solid line ) , efficiency @xmath27 ( blue solid line ) and cavity funneling ratio @xmath28 ( red solid line ) for a fixed emitter - cavity coupling of @xmath70 .
the full calculation for indistinguishability is compared with analytic expressions in two different limiting cases : incoherently - coupled good cavity regime ( dotted line , eq .
[ igc ] ) and coherently - coupled bad cavity regime ( dashed line , eq . [ ibc ] ) . the analytical expression for @xmath27 ( dotted line , eq .
[ betaweak ] ) overlaps perfectly with the full calculation.,scaledwidth=48.0% ] finally , we propose two experimental realizations of this unconventional regime at room temperature .
we first consider a single self - assembled quantum dot coupled to a photonic crystal cavity .
state of the art photonic crystal cavities can provide @xmath71 = 120 @xmath72ev and @xmath73=20 @xmath72ev @xcite .
assuming @xmath74 = 60 @xmath72ev and @xmath75 = 7 mev for an inas / gaas qd at 300k @xcite , we predict @xmath5=0.72 , @xmath27=0.088 and @xmath28=7.3 . secondly , we consider a single silicon vacancy ( siv ) center in a nano - diamond coupled to a fiber cavity .
for siv at 300k , we take @xmath4 = 2@xmath76@xmath77 160 mhz and @xmath11 = 2@xmath76@xmath77 550 ghz @xcite .
coupling siv with a fiber cavity with @xmath12=2@[email protected] ghz and @xmath10=2@xmath76@xmath7730 mhz is within experimental reach @xcite , for which our calculation predicts @xmath5=0.81 , @xmath27=0.035 and @xmath28=99 .
these predicted degrees of indistinguishability at room temperature are comparable with state of the art values obtained from low temperature single - photon sources under incoherent pumping @xcite , with efficiencies far beyond any spectral filtering technique . in summary , for strongly dissipative emitters
we predict an unconventional regime of high indistinguishability in which the broad spectrum of the quantum emitter is funneled into a narrow cavity resonance . for typical room - temperature quantum emitters
, the associated efficiency can surpass any spectral filtering schemes by orders of magnitude .
this strategy opens the road towards the generation of indistinguishable single photons from solid - state quantum emitters under ambient temperature .
this work has been funded by the european union s seventh framework programme ( fp7 ) under grant agreement no .
618078 ( wasps ) .
jason smith and john rarity are acknowledged for fruitfuil discussions .
53 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , * * , ( ) .
, , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) .
, , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) .
, , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , ,
* * , ( ) . , , , , , , , , , , , ( ) . , , , , , ,
, * * , ( ) . ,
* * , ( ) .
, , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , , , , * * ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , , * * , ( ) . , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , ,
* * , ( ) . , , , , , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , , ,
* * , ( ) . , , , * * , ( ) .
, , , * * , ( ) . .
, , , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , ,
* * , ( ) .
* supplemental materials * |
we discuss resonance recombination for quarks and show that it is compatible with quark and hadron distributions in local thermal equilibrium .
we then calculate realistic heavy quark phase space distributions in heavy ion collisions using langevin simulations with non - perturbative @xmath0-matrix interactions in hydrodynamic backgrounds .
we hadronize the heavy quarks on the critical hypersurface given by hydrodynamics after constructing a criterion for the relative recombination and fragmentation contributions .
we discuss the influence of recombination and flow on the resulting heavy meson and single electron @xmath1 and elliptic flow .
we will also comment on the effect of diffusion of open heavy flavor mesons in the hadronic phase .
quark recombination has been identified as an important mechanism of hadronization in high energy nuclear collisions @xcite .
evidence for recombination can be seen in the large baryon / meson ratios and the quark - number scaling of elliptic flow @xmath2 at the relativistic heavy ion collider ( rhic ) @xcite .
first - generation recombination models that could explain these phenomena at intermediate transverse momenta @xmath3 gev/@xmath4 were based on so - called instantaneous quark coalescence .
they lack explicit energy conservation and serious concerns arise if they are applied to low @xmath5 where the bulk of hadrons are produced .
hence it is important to understand how quark recombination in local kinetic equilibrium can be described .
a promising formalism was put forward in @xcite based on a boltzmann equation for quarks and antiquarks scattering through resonances which resemble mesons . this resonance recombination model ( rrm )
conserves 4-momentum and exhibits detailed balance .
therefore , in the long - time limit of the boltzmann dynamics the mesons should acquire local thermal equilibrium with temperature and flow fields given by those of the equilibrated quark distributions .
we have numerically confirmed this assertion both for simple blastwaves with constant hadronization time @xcite , and for the non - trivial hadronization hypersurface of a hydrodynamic evolution of the quark phase @xcite .
hence resonance recombination preserves local thermal equilibrium for arbitrary flow fields and for any realistic hypersurface .
an important corollary is the fact that any quark - number or kinetic energy scaling at _ low _ hadron-@xmath5 is not due to quark recombination .
local equilibration would not allow any microscopic information about hadronization to be detected , and we explicitely show in ref .
@xcite how a realistic flow field together with a reasonable freeze - out hierarchy for hadrons can mimic quark number- and kinetic energy scaling at rhic at _ low _ @xmath5 to good accuracy .
heavy charm ( @xmath4 ) and bottom ( @xmath6 ) quarks have been widely discussed as ideal internal probes for quark gluon plasma , see @xcite and references therein .
the large masses increase their thermalization times , and the degree of thermalization in heavy ion collisions encodes valuable information about their interactions with quark gluon plasma ( qgp ) . here
we report on our effort to set up a formalism that is consistently based on the notion of strongly coupled qgp , both for the diffusion of heavy quarks through qgp , and their hadronization through an appropriate superposition of recombination and fragmentation @xcite .
the latter provides an opportunity to extend the resonance recombination formalism to quark distributions away from local thermal equilibrium .
the dynamics of the heavy quarks in a background medium can be described by a relativistic implementation of the langevin equation @xcite .
we use the ideal , 2 + 1-d , boost - invariant hydro code azhydro @xcite to model the locally equilibrated background qgp medium .
the drag and diffusion coefficients of the heavy quarks in the medium are calculated from non - perturbative @xmath0-matrix results on heavy - light quark interactions @xcite .
the @xmath0-matrix approach exhibits feshbach resonances both in the color - singlet and color - triplet channel and leads to resonant interactions up to @xmath7 .
the resonant relaxation rates are substantially enhanced compared to the corresponding perturbative elastic rates .
the initial heavy quark momentum spectra are taken from perturbative calculations cross - checked with measured @xmath8 semi - leptonic decay spectra @xcite , while the spatial distribution is determined by the binary collision density . - and @xmath9-meson nuclear modification factors for semi - central au+au collisions at rhic compared to the @xmath4 and @xmath6 quark @xmath1 from which they originate.,width=283 ] we employ a test particle method for our langevin simulation .
we have checked that in the limit of very large ( unphysical ) relaxation rates our charm quark distributions approach the equilibrium given by the hydrodynamic background .
figure [ fig:1 ] shows the nuclear modification factor @xmath1 for @xmath4 and @xmath6 quarks on the hadronization hypersurface ( given by azhydro ) calculated from our langevin with non - perturbative @xmath0-matrix coefficients for semi - central collisions at rhic .
we observe about 60% quenching for charm quarks at large momenta @xmath10 which is reflected in an excess at small @xmath10 due to conservation of heavy quarks .
suppression for @xmath6 is significantly smaller .
hadronization of heavy quarks should proceed through coalescence with light quarks from the medium if heavy - light scattering rates are large enough to allow for such interactions in the color - singlet channel during the pertinent time interval ( e.g. the duration of the mixed phase ) . on the other hand ,
if such heavy - light interactions are rare , e.g. at large @xmath10 , independent fragmentation of the heavy quark should occur . for a given heavy quark momentum in its local fluid rest frame we estimate the scattering rate @xmath11 in the color singlet channel in the @xmath0-matrix approach at @xmath12 .
we approximate the coalescence probability through @xmath13 ( or @xmath14 if the product exceeds one ) . in practice
we boost the rates into the lab frame and apply them as a function of heavy quark @xmath10 with @xmath15 fm/@xmath4 ( motivated by the fact that this time interval is well below the duration of the mixed phase in azhydro ) which leads to @xmath16 at vanishing @xmath10 .
note that this approach treats the interactions of heavy quarks in the medium and their hadronization using the same non - pertrubative dynamics .
it also ensures that @xmath17 for very large @xmath10 .
we evaluate @xmath18 for each test particle and apply fragmentation or resonance recombination accordingly .
-meson @xmath5-spectrum for semi - central au+au collisions at top rhic energy .
the total spectrum is normalized to one test - particle .
( b ) : the same three contributions to the elliptic flow @xmath2.,title="fig:",width=283 ] -meson @xmath5-spectrum for semi - central au+au collisions at top rhic energy .
the total spectrum is normalized to one test - particle .
( b ) : the same three contributions to the elliptic flow @xmath2.,title="fig:",width=283 ] figure [ fig:2 ] shows spectra and elliptic flow @xmath2 for @xmath19 mesons for semi - central collisions at rhic .
fragmentation and recombination contributions are shown together with the total .
figure [ fig:1 ] provides @xmath1 for both @xmath19 and @xmath9 mesons .
we notice that resonance recombination provides a significant enhancement of both radial and elliptic flow .
the former can be seen from the prominent flow `` bump '' developing in @xmath1 for @xmath19 mesons .
resonance recombination acts as an additional interaction of heavy quarks with the medium , making a noticeable contribution to further equilibration . in figure [ fig:3 ]
we display @xmath1 and @xmath2 of electrons from semi - leptonic decays of @xmath19 and @xmath9 mesons , compared to experimental data .
the comparison with data is favorable if one keeps in mind two facts : azhydro is tuned to kinetic freeze - out and exhibits too little radial flow around @xmath12 .
this becomes obvious if the results with azhydro are compared to those using a fireball parameterization with the correct radial flow at @xmath12 ( see @xcite for details ) .
secondly , we expect an increase in elliptic flow of about 20 - 30% in the hadronic phase .
we have recently calculated relaxation rates of charm in a hadron gas and have found that at @xmath12 the results tend to be close to those of the @xmath0-matrix calculation in qgp @xcite .
we will elaborate on both points further in a forthcoming publication . from semi - leptonic decays of @xmath19 and @xmath9 mesons for central ( @xmath20 fm ) and semi - central ( @xmath21 fm )
au+au collisions at rhic . using either azhydro or a parameterized fireball as the background medium .
we compare to data from phenix @xcite and star @xcite .
( b ) electron @xmath2 from the same source.,title="fig:",width=283 ] from semi - leptonic decays of @xmath19 and @xmath9 mesons for central ( @xmath20 fm ) and semi - central ( @xmath21 fm ) au+au collisions at rhic .
using either azhydro or a parameterized fireball as the background medium .
we compare to data from phenix @xcite and star @xcite .
( b ) electron @xmath2 from the same source.,title="fig:",width=283 ] this work was supported by the u.s
. national science foundation ( nsf ) through career grant phy-0847538 and grant phy-0969394 , by the a .- v .-
humboldt foundation , by the riken / bnl research center and doe grant de - ac02 - 98ch10886 , and by the jet collaboration and doe grant de - fg02 - 10er41682 . |
|
so coupling for ultracold fermi atoms has been successfully demonstrated in ultracold fermi @xmath3 and @xmath4 gases @xcite at about the same time , in which the raman dressing scheme is based on coupling two magnetic sub - levels of the ground state manifold with two counter propagating raman lasers @xcite .
the system considered in this work is depicted in fig .
1(a ) , where the raman dressed @xmath3 fermi gases are loaded in a bilayer geometry , which can be readily realized by adding a two - dimension double - well optical lattice @xcite .
note that the optical lattice is spin - independent and can induce tunneling without spin - flip .
same two atomic internal spin states @xmath5 and @xmath6 of the ground state in @xmath7-th layer are selected to be labeled as spin - up ( @xmath8 ) and down ( @xmath9 ) states , where @xmath10 refers to an individual layer .
the atom move along the @xmath11 axis within a layer and two layers are separated by a distance @xmath12 .
the tunneling amplitude @xmath13\psi_{2,\sigma}(z)$ ] with @xmath14 can be regulated by changing the intensity or relative phase of laser standing waves that engineer the double - well lattice @xcite .
the positive and negative signs label the detuning from raman resonance to spin - up and spin - down respectively . without loss of generality ,
we assume them to be real and @xmath15 .
the four - level topology has been schematically shown in fig .
1(b ) , where the blue solid lines represent the inter - layer tunneling and the green dashed circles denote the raman - assisted intra - layer interaction with momentum transfer @xmath16 .
the inter - layer spin - flip tunnelings are negligible under current experimental conditions , therefore they are neglected in fig .
four atomic states couple with each other in a cyclic manner with no momentum transferred during a closed loop transition .
the hamiltonian for bilayer fermi gases in the presence of so coupling is given by @xmath17 $ ] , with the single - particle component @xmath18 and the interacting component @xmath19 describing the @xmath20-wave contact interaction between the two spin states in @xmath7-th layer .
the single - particle hamiltonian is written as @xmath21 with @xmath22^{t}$ ] being the atomic annihilation operators , @xmath23 , where @xmath24 denotes kinetic energy and @xmath25 is the chemical potential , @xmath26 is mass of an atom , @xmath27 means the detuning of the raman process , and @xmath28 is the trapping potential ( for convenience , we set @xmath29 ) .
the constants @xmath30 and @xmath16 represent the coupling strength and photon recoil momentum of the two - photon raman coupling , respectively . after applying a local gauge transformation @xmath31 , \\ & \psi_{j,\downarrow}(x)=\frac{1}{\sqrt{2}}[e^{-ik_{0}x/2}\phi_{j,\uparrow}(x)+ie^{-ik_{0}x/2}\phi_{j,\downarrow}(x ) ] , \\
\end{split}\ ] ] the single - particle hamiltonian becomes @xmath32 where @xmath33^{t}$ ] is a single layer field operator , and the single layer term is @xmath34 where we define the chemical potential @xmath35 and @xmath36 , @xmath37 are the pauli matrices acting on the spins .
the effective spin - orbit coupling constant @xmath38 and effective zeeman field ( @xmath39 , @xmath40 ) are introduced . for convenience ,
the recoil momentum @xmath41 and recoil energy @xmath42 are taken as natural momentum and energy units .
notice that the interaction hamiltonian @xmath43 is invariant under this gauge transformation @xcite .
we first consider homogeneous fermi gases to give the simplest qualitative picture of the bilayer system . at this point ,
the physics of the single - particle hamiltonian is simple in momentum space using a fourier decomposition .
the four - band structure of the bilayer system is illustrated in fig .
1(c , d ) .
obviously , the in - plane zeeman field @xmath44 leads to a fermi surface without inversion symmetry and gives a so - called fulde - ferrell - larkin - ovchinnikov ( fflo ) state with a finite total momentum pairing @xcite , but can not open a band gap at the dirac point , as shown in fig .
for the realization of topological phase , a perpendicular zeeman field is needed to break the time - reversal ( tr ) symmetry and open a topological band gap . in fig .
1(c ) , we report that the topological band gaps occur at the dirac point @xmath45 induced by increasing @xmath46 , which are similar to the single - layer mechanism .
the superfluid pairing is between atoms with opposite momentum @xmath47 and @xmath48 without in - plane zeeman field @xmath44 .
a critical observation is that when the zeeman field is larger than the tunneling effect @xmath49 , the increasing zeeman field removes the level crossing and open a gap between the two middle branches , as shown by the red curves in the fig .
1(c , d ) .
we will give a detailed discussion in the mean - field formalism .
we consider the most simplified form of interactions with superfluid pairing formed between atoms with opposite momentum ( @xmath47 , @xmath48 ) . within the mean - field approximation ,
the interaction term can be rewritten as @xmath50 with the order parameters @xmath51 .
one can easily write the mean - field bdg hamiltonian , @xmath52 where the nambu spinor basis is chosen as @xmath53^{t}$ ] .
the elementary excitations can be found by solving the bdg equation @xmath54 , where @xmath55 ( @xmath56 ) are the eigenvalues of the above @xmath57 bdg hamiltonian and corresponding wave functions are assumed as @xmath58 , where the label @xmath59 represent the particle as well as hole branches . without loss of generality , we set @xmath60 throughout the work .
the spectrum of @xmath61 consists of eight bands given by @xmath62 where the @xmath63 and @xmath64 with @xmath65 and @xmath66 . by solving @xmath67
, we obtain all the gap close conditions in analytic forms : @xmath68 with @xmath69 , @xmath70 .
the solutions for the zeeman field are @xmath71 here we just consider the parallel case @xmath72 . if there is no tunneling between two layers , the critical zeeman field reduces to the well - known @xmath73 , which has been discussed in detail in earlier works @xcite . the system will be in a conventional superfluid at @xmath74 and in a topological superfluid at @xmath75 . unlike the conventional phase transition in a single layer system , the tunneling shifts the chemical potential and gives two critical zeeman fields : @xmath76 and @xmath77 ( or exchange to assure @xmath78 ) .
the system undergoes a tqpt when the zeeman field cross these two critical values , where the single - particle excitation gap vanishes , representing a topological phase transition . to see the phase transitions more clearly
, we give the spin population at @xmath79 , as shown in fig .
2(b ) . here
we define the spin vector expectation value along the @xmath80-axis as @xmath81 we find that @xmath82 changes discontinuously when the zeeman field crosses the critical values @xmath83 $ ] , which implies the change of the topology of the spin texture .
@xmath84 if @xmath85 , which corresponds to the trivial state , while @xmath86 for @xmath87 and @xmath88 for @xmath89 denote diverse topological states , as show in fig .
therefore we give a classification of topological superfluid phases according to the strength of perpendicular zeeman field @xmath46 : @xmath90 where the last condition @xmath91 in the topological regimes ensures the bulk quasi - particle excitations are gapped to protect the zero - energy mfs .
the region for topological superfluid depends not only on the chemical potential , pairing strength , but also on the tunneling strength .
it provides more control knobs for tuning the topological phase transition .
especially when the tunneling can be comparable to the chemical potential , the perpendicular zeeman field threshold can be dramatically lowered .
these new features may open a possibility for the experimental realization of topological superfluid under small zeeman field . to better understand the transition from one state to another defined by equation ( [ equ : gapclosecondition ] ) , it is necessary to observe the close and reopen of the excitation gap @xmath92 , which change the topology of fermi surface .
the phase diagram in the plane of the order parameter and the perpendicular zeeman field is presented in fig .
the graph is colored according to the energy gap @xmath92 , and white dotted curves mark contours of @xmath93 .
there are three different topological regions determined by the behavior of the energy gap .
the system is topological trivial at first , as increasing the zeeman field with a fixed @xmath94 , the band gap may first close and then reopen , signifying the transition from non - topological to @xmath95 superfluid ( @xmath96 $ ] ) , and finally undergoes a topological phase transition from @xmath95 to @xmath97 phase at the second gapless point @xmath98 . in fig .
2(b - f ) , we plot the band gap and energy spectrums to illustrate the tqpt when @xmath99 .
it should be emphasized that the range of region : @xmath95 is highly dependent on the value of order parameter @xmath94 , the increasing of the intensity of @xmath94 rapidly shrinks this region . as have discussed the evolution of the single - particle hamiltonian as a function of @xmath46 , here we can further more rigorously demonstrate the topological phase transition with the change of the chemical potential @xmath25 .
we can sketch the phase diagram in the fig .
3(a ) , which shows perfect inversion symmetry ( @xmath100 ) in the @xmath101 plane . for a small order parameter ,
the important feature is that there have four topological transition points along the change of the chemical potential .
similar to the case in fig .
2 , by tuning through the transition point , the quasi - particle excitation gap closes and open again , thus we can discriminate five regions in accordance with the topological condition ( [ equ : gapclosecondition ] ) with central area being @xmath97 .
we can write down the scaling forms of the critical chemical potentials @xmath102 which can be easily got from the gap closing condition ( 9 ) .
it should be pointed out that the range of central region @xmath97 gets shrink by increasing @xmath94 and turn into trivial phase above a threshold : @xmath103 . by further increasing the intensity of the order parameter beyond @xmath104 , the inequality @xmath105 holds at any @xmath25 ,
thus there is only trivial phase region in this situation .
the evolvement of band gap @xmath92 is plotted as the change of chemical potential in fig .
there are five phases and separated by the critical values of the chemical potential @xmath106 with @xmath107 .
it can be better understood by observing change of the topology of fermi surface in the single - particle dispersion ( shown in fig .
there have three gaps between branches under the condition @xmath49 .
when @xmath25 lies in the lowest gap , the system exhibits a single pair of fermi points as desired .
the increase of @xmath25 tunes the system from @xmath95 to @xmath97 then back to @xmath95 by crossing the critical points between gaps , as shown in fig .3(b ) . to confirm the topological phase transition and observe the appearance of zero - energy mf mode under some appropriate parameter condition
, we take into account the trapping potential that is necessary to prevent the atoms from escaping . within the lda
, the chemical potential can be thought of as a position dependent function : @xmath108 , that continuously decreases away from the center of the trap .
therefore , the critical zeeman field can be accordingly redefined as @xmath109 both the local pairing gap and chemical potential decrease away from the trap center . at a small field @xmath46 , @xmath110 for any position and
the whole fermi cloud is in the conventional superfluid .
meanwhile , the condition @xmath111 may be always satisfied at an even large zeeman field .
we can tune the system through an intermediate mixed phase in which topological trivial and non - trivial phase can co - exist .
the mixed phases which appeare naturally in the parabolic trap potential , are illustrated in fig .
4(a1,b1,c1 ) .
we first consider setting the central chemical potential @xmath112 at a point above all zeeman gaps that four transition points are all covered by the position - dependent chemical potential ( shown in fig .
there are five districts along with continuously decrease of the chemical potential from the middle of the trap towards the flanks .
if we lower down the chemical potential , less transition points are covered and less topological phases can be distinguished ( see fig .
4(b1,c1 ) ) .
all we discussed just depend on the starting point of the chemical potential .
the number of the majorana zero - energy modes is decided by the amount of spatial interfaces between the different phases discussed above .
the mf is a half of an ordinary fermion and must always come in pairs .
each of the paired states , localized in real space , can be hardly pushed away from @xmath113 by a local perturbation , giving rise to the intrinsic topological stability enjoyed by mfs .
here we use the numerical method and solve self - consistent bdg equation in real space to explore the mf physics beyond the lda . in the presence of a trap @xmath114 , the bdg equation in real space can be written as @xmath115 , where @xmath116 @xmath117^{t}$ ] are the nambu spinor wave functions corresponding to the quasi - particle excitation energy @xmath118 .
the single particle hamiltonian in real - space becomes @xmath119 .
the eigenfunctions satisfy the normalization condition @xmath120=\delta_{\eta,\eta^{'}}.\end{aligned}\ ] ] the order parameter @xmath121 and the chemical potential @xmath25 can be determined by the self - consistency equation @xmath122\ ] ] and the number equation @xmath123 with @xmath124\ ] ] is the local density of @xmath125 fermions in @xmath7-th layer , @xmath126 is the quasi - particle fermi - dirac distribution at the temperature @xmath127 .
similar to the intrinsic symmetry built in the bdg hamiltonian in momentum space , @xmath128 is invariant under the particle - hole transformation : @xmath129 and @xmath130 . to remove
this redundancy , a factor of @xmath131 has appeared in the above expressions . due to @xmath132 , we will immediately have @xmath133 , a zero - energy quasi - particle being its own antiparticle , exactly the defining feature of a mf .
it is straightforward to show from the bdg hamiltonian that the wave function of mfs should satisfy either @xmath134 or @xmath135 .
we self - consistently solve the gap equation and the number equation for different total number of particles .
@xmath136 is set to assure the trap oscillation frequency is much smaller than the recoil frequency .
commonly , we use a dimensionless interaction parameter , @xmath137 , to characterize the interaction strength , where @xmath138 is the zero - temperature center density of an ideal fermi gas .
the typical interaction strength is about @xmath139 and we take @xmath140 in our calculations . we have been taken @xmath141 harmonic oscillators as the expansion functions and find that @xmath142 is large enough to ensure the accuracy of the calculations . by varying the zeeman field @xmath46 while keeping the chemical potential fixed in the middle of the gap , this way has been proved to be a conventional tool in experiments . alternatively , changing the number of particles while keeping the zeeman field fixed
will have the same effect .
4(a2,b2,c2 ) show the mean - field results of energy spectrum with different total particle number .
the emergence of mfs can be clearly revealed by the behavior of the energy spectrum . at a large total particle number @xmath143 , eight majorana zero - energy modes associated with the presence of phase transitions ,
appear in the middle of the gap ( seen in fig .
the number of mfs changes as the total number of atoms varies .
we could have six ( fig .
4(b2 ) ) or four ( fig .
4(c2 ) ) and even less mfs corresponding to what we have discussed in fig .
4(b1,c1 ) . in fig . 5
, we give a calculation of @xmath144 for a local uniform cell at position @xmath145 with the local chemical potential @xmath146 and order parameter @xmath147 .
the local uniform cell would be in the @xmath148 ( @xmath149 ) state if @xmath150 ( @xmath151 ) and the fermi cloud is in the conventional superfluid with @xmath152 for any other position @xmath145 .
these areas are highlighted by different colors . at a total particle number @xmath143 , there are six districts of topological state and therefore we can find four pairs of mfs over the whole system . by decreasing the total number , the conventional area at the trap center
gets shrunk and eventually disappears at about @xmath153 , three pairs of mfs get preserved .
it is obvious that less mfs remain if we future decrease the total number ( see fig .
5(b , c ) ) .
zero - energy modes just locate around the place with @xmath154 , corresponding to the steep slope of the order parameter .
the agreement between bdg and lda result is reasonable .
both theories predict multi - shell structures .
their spatial variation clearly identifies that .
in this work , we investigate the topological properties of bilayer fermi gases in the presence of so coupling . two diverse topological phases with @xmath155 are observed and connected by the tqrt , which have not been discovered previously .
the topological structure of such system , whether it can be topological trivial or topological non - trivial , depend on not only the combined effect of so coupling and the zeeman field , but also the inter - layer tunneling , thus offering more tunability in experiments .
we give the reliable results by solving the self - consistent bdg equation within a harmonic trap and predict that one can manipulate numerous majorana zero energy modes associated with interfaces between diverse phases by reforming the phase geometric construction of the system .
our results also can be applicable to two dimensional topological superfluids .
it would be interesting to examine the chern number that denotes the topology of the matter , and can be used to further test the concept behind our studies .
the bilayer fermi gases with so coupling described here offer an practically realizable platform for further experiments to investigate the topological superfluid and realize and manipulate mfs , providing new insights into quantum computation .
we work in a complete basis of single - particle wave functions in the harmonic trap @xmath156 with energy @xmath157 , @xmath158 here @xmath159 is the hermite polynomial , @xmath160 is the characteristic harmonic oscillator length .
note that spin - orbit coupling working at these basis becomes @xmath161 .
then we expand the normalized wave functions as @xmath162 , @xmath163 . according to the orthonormality relation @xmath164
, we reduce the bdg equation to a matrix diagonalization problem , @xmath165 where the matrix elements are @xmath166 due to the normalization of the quasi - particle wavefunctions , the coefficients of the eigenstate must satisfy the condition @xmath167 .
the same procedure also reduces the order parameter to @xmath168 and the local - density equations @xmath169 the above equations must be solved together by means of an iterative procedure , which starts from trial functions ( @xmath121 , @xmath25 ) and converges to the self - consistent solution .
all sums in the equations are limited by an energy cutoff @xmath170 .
this cutoff is required in order to cure the ultraviolet divergences . during the iteration , the order parameters @xmath121
are updated , and the chemical potential @xmath25 is also adjusted slightly to enforce the number - conversation condition . the self - consistent iterative procedure is over until final convergence is reached .
99 hasan , m. z. & kane , c. l. _ colloquium _ : topological insulators . _ rev .
* 82 * , 3045 ( 2010 ) .
zhang , s .- c .
topological insulators and superconductors .
* 83 * , 1057 ( 2011 ) .
jiang , l. , kitagawa , t. , alicea , j. , akhmerov , a. r. , pekker , d. , refael , g. , cirac , i. , demler , e. , lukin , m. d. & zoller , p. majorana fermions in equilibrium and in driven cold - atom quantum wires .
lett . _ * 106 * , 220402 ( 2011 ) .
brouwer , p. w. enter the majorana fermion .
_ science _ * 336 * , 989 - 990 ( 2012 ) .
alicea , j. new directions in the pursuit of majorana fermions in solid state systems .
phys . _ * 75 * , 076501 ( 2012 ) .
elliott , s. r. & franz , m. _ colloquium _ : majorana fermions in nuclear , particle , and solid - state physics . _
phys . _ * 87 * , 137 ( 2015 ) .
kitaev , a. y. unpaired majorana fermions in quantum wires .
_ physics - uspekhi _ * 44 * , 131 ( 2001 ) .
nayak , c. , simon , s. h. , stern , a. , freedman , m. & das sarma , s. non - abelian anyons and topological quantum computation .
* 80 * , 1083 ( 2008 ) .
deng , d .-
, & duan , l .- m .
fault - tolerant quantum random - number generator certified by majorana fermions . _ phys .
a _ * 88 * , 012323 ( 2013 ) .
lutchtn , r. m. , sau , j. d. & das sarma , s. majorana fermions and a topological phase transition in semiconductor - superconductor heterostructures .
* 105 * , 077001 ( 2010 ) .
oreg , y. , refael , g. & von oppen , f. helical liquids and majorana bound states in quantum wires .
_ * 105 * , 177002 ( 2010 ) .
alicea , j. , oreg , y. , refael , g. von oppen , f. & fisher , m. p. a. non - abelian statistics and topological quantum information processing in 1d wire networks .
phys . _ * 7 * , 412 ( 2011 ) .
stoudenmire , e. m. , alicea , j. , starykh , o. a. & fisher , m. p. a. interaction effects in topological superconducting wires supporting majorana fermions .
_ phys . rev .
b _ * 84 * , 014503 ( 2011 ) .
stanescu , t. d. , lutchyn , r. m. & das sarma , s. majorana fermions in semiconductor nanowires . _ phys .
b _ * 84 * , 144522 ( 2011 ) . liu , c .- x . , qi , x .-
l . , dai , x. , fang , z. & zhang , s. c. quantum anomalous hall effect in @xmath171 quantum wells .
_ * 101 * , 146802 ( 2008 ) .
fu , l. & kane , c. l. superconducting proximity effect and majorana fermions at the surface of a topological insulator .
* 100 * , 096407 ( 2008 ) .
chen , y. l. , chu , j .- h .
, analytis , j. g. , liu , z. k. , igarashi , k. , kuo , h .- h . , qi , x .-
, mo , s. k. , moore , r. g. , lu , d. h. , hashimoto , m. , sasagawa , t. , zhang , s. c. , fisher , i. r. , hussain , z. & shen , z. x. massive dirac fermion on the surface of a magnetically doped topological insulator .
_ science _ * 329 * , 659 ( 2010 ) .
sau , j. d. , lutchyn , r. m. , tewari , s. & das sarma , s. generic new platform for topological quantum computation using semiconductor heterostructures .
lett . _ * 104 * , 040502 ( 2010 ) .
alicea , j. majorana fermions in a tunable semiconductor device .
b _ * 81 * , 125318 ( 2010 ) .
tewari , s. , das sarma , s. , nayak , c. , zhang , c. & zoller , p. quantum computation using vortices and majorana zero modes of a @xmath172 superfluid of fermionic cold atoms .
_ * 98 * , 010506 ( 2007 ) .
vyasanakere , j. p. , zhang , s. & shenoy , v. b. bcs - bec crossover induced by a synthetic non - abelian gauge field .
b _ * 84 * , 014512 ( 2011 ) . liu , x .-
j . , jiang , l. , pu , h. & hu , h. probing majorana fermions in spin - orbit - coupled atomic fermi gases .
a _ * 85 * , 021603(r ) ( 2012 ) .
gong , m. , chen , g. , jia , s. & zhang , c. searching for majorana fermions in 2d spin - orbit coupled fermi superfluids at finite temperature . _
* 109 * , 105302 ( 2012 ) .
fu , z. , huang , l. , meng , z. , wang , p. , zhang , l. , zhang , s. , zhai , h. , zhang , p. & zhang , j. production of feshbach molecules induced by spin - orbit coupling in fermi gases .
* 10 * , 1038 ( 2013 ) .
zhang , y. , chen , g. & zhang , c. tunable spin - orbit coupling and quantum phase transitions in a trapped bose - einstein condensate .
_ * 3 * , 1937 ( 2013 ) .
zhai , h. spin - orbit coupled quantum gases .
b _ * 26 * , 1230001 ( 2012 ) .
qu , c. zheng , z. , gong , m. , xu , y. , mao , l. , zou , x. , guo , g. & zhang , c. topological superfluids with finite - momentum pairing and majorana fermions .
commun . _ * 4 * , 2710 ( 2013 ) .
zhang , w. & yi , w. topological fulde - ferrell - larkin - ovchinnikov states in spin - orbit - coupled fermi gases .
commun . _ * 4 * , 2711 ( 2013 ) .
hamner , c. , qu , c. , zhang , y. , chang , j. , gong , m. , zhang , c. & engels , p. dicke - type phase transition in a spin - orbit - coupled bose - einstein condensate .
commun . _ * 5 * , 4023 ( 2014 ) .
bhler , a. , lang , n. , kraus , c. v. , mller , g. , huber , s. d. & bchler , h. p. majorana modes and @xmath173-wave superfluids for fermionic atoms in optical lattices .
_ * 5 * , 4504 ( 2014 ) .
lin , y .- j . , compton , r. l. , jimnez - garcia , k. , porto , j. v. & spielman , i. b. synthetic magnetic fields for ultracold neutral atoms .
_ nature _ * 462 * , 628 - 632 ( 2009 ) .
lin , y .- j .
, jimnez - garcia , k. & spielman , i. b. spin - orbit - coupled bose - einstein condensates . _ nature _ * 471 * , 83 - 86 ( 2011 ) .
sau , j. d. , sensarma , r. , powell , s. , spielman , i. b. & das sarma , s. chiral rashba spin textures in ultracold fermi gases .
b _ * 83 * , 140510 ( r ) ( 2011 ) .
wang , p. , yu , z .- q . ,
fu , z. , miao , j. , huang , l. , chai , s. , zhai , h. & zhang , j. spin - orbit coupled degenerate fermi gases . _
lett . _ * 109 * , 095301 ( 2012 ) .
cheuk , l. w. , sommer , a. t. , hadzibabic , z. , yefsah , t. , bakr , w. s. & zwierlein , m. w. spin - injection spectroscopy of a spin - orbit coupled fermi gas .
* 109 * , 095302 ( 2012 ) . ho ,
zhang , s. bose - einstein condensates with spin - orbit interaction . _
lett . _ * 107 * , 150403 ( 2011 ) .
wang , c. , gao , c. jian , c .- m . &
zhai , h. spin - orbit coupled spinor bose - einstein condensates . _
lett . _ * 105 * , 160403 ( 2010 ) .
wen , l. , sun , q. , wang , h. q. , ji , a. c. & liu , w. m. ground state of spin-1 bose - einstein condensates with spin - orbit coupling in a zeeman field .
a _ * 86 * , 043602 ( 2012 ) .
liu , c .- f .
, yu , y .- m . , gou , s .- c . &
liu , w. m. vortex chain in anisotropic spin - orbit - coupled spin-1 bose - einstein condensates . _
a _ * 87 * , 063630 ( 2013 ) .
peterson , m. r. , papi , z. & das sarma , s. fractional quantum hall effects in bilayers in the presence of interlayer tunneling and charge imbalance .
b _ * 82 * , 235312 ( 2010 ) . sui , m. , chen , g. , ma , l. , shan , w .- y . ,
tian , d. , watanabe , k. , taniguchi , t. , jin , x. yao , w. , xiao , d. & zhang , y. gate - tunable topological valley transport in bilayer graphene .
phys . _ * 11 * , 1027 ( 2015 ) .
milburn , g. j. , corney , j. , wright , e. m. & walls , d. f. quantum dynamics of an atomic bose - einstein condensate in a double - well potential .
a _ * 55 * , 4318 ( 1997 ) .
smerzi , a. , fantoni , s. , giovanazzi , s. & shenoy , s. r. quantum coherent atomic tunneling between two trapped bose - einstein condensate . _
lett . _ * 79 * , 4950 ( 1997 ) .
giovanazzi , s. smerzi , a. & fantoni , s. josephson effects in dilute bose - einstein condensates .
lett . _ * 84 * , 4521 ( 2000 ) .
albiez , m. , gati , r. , flling , j. , hunsmann , s. , cristiani , m. & oberthaler , m. k. direct observation of tunneling and nonlinear self - trapping in a single bosonic josephson junction .
lett . _ * 95 * , 010402 ( 2005 ) .
levy , s. , lahoud , e. , shomrouni , i. & steinhauer , j. the a.c . and d.c .
josephson effects in a bose - einstein condensate .
_ nature _ * 449 * , 579 - 583 ( 2007 ) .
leblanc , l. j. , bardon , a. b. , mckeever , j. , extavour , m. h. t. , jervis , d. & thywissen , j. h. dynamics of a tunable superfluid junction .
lett . _ * 106 * , 025302 ( 2011 ) .
sebby - strabley , j. , anderlini , m. , jessen , p. s. & porto , j. v. lattice of double wells for manipulating pairs of cold atoms .
a _ * 73 * , 033605 ( 2006 ) .
xu , b. , wu , h .- y .
, zhang , y .- c . , wang , h .- t . &
liu , w .- m .
excitation spectra and hard - core thermodynamics of bosonic atoms in optical superlattices .
a _ * 91 * , 013617 ( 2015 ) .
soliton - induced majorana fermions in a one - dimensional atomic topological superfluid .
a _ * 91 * , 023610 ( 2015 ) .
we thank an - chun ji and qing sun for their early contribution to this project and are grateful to kashifammar yasir and xiao - xi lv for valuable discussions .
this work was supported by the nsfc under grants nos .
11434015 , 61227902 , 61378017 , nkbrsfc under grants nos . 2012cb821305 , sklqoqod under grants no .
kf201403 , sprpcas under grants no .
xdb01020300 .
* figure 1 the experimental setup for spin - orbit coupled bilayer fermi gases . *
( * a * ) schematics diagram for the bilayer system of @xmath3 fermi gases .
a double well optical lattice that has two local minima in a unit cell , is formed by the interference of two pairs of counter - propagating laser beams . in this case , there is a resulting double - well periodic potential along @xmath174-axis , which can allows the inter - well tunnelling @xmath175 . within a unit cell ,
the barrier height and relative depth of the double well can be regulated by changing the intensity or relative phase of laser standing waves .
two raman lasers counter - propagate along @xmath11-axis are used to generate so coupling with the recoil momentum @xmath176 , where @xmath177 and @xmath178 is the angle between two raman lasers .
( * b * ) effective coupling between four atomic levels @xmath179 ( @xmath180 ) in a double - well trap , where @xmath175 denotes the inter - layer tunneling without spin flipping , and @xmath30 represents the intra - layer spin - orbit coupling induced by the raman lasers .
( * c*,*d * ) show the evolution of single - particle energy spectra as a function of the zeeman field @xmath46 . both without ( * * c**1)-(**c**3 ) and with ( * * d**1)-(**d**3 ) in - plane zeeman field are considered . without in - plane zeeman field ,
the band structure is symmetric around @xmath79 , which supports the superfluid pairing between atoms with opposite momentum .
the band structure becomes asymmetric and a finite total momentum pairing may occurs when @xmath181 .
the colored energy curves correspond to different zeeman field @xmath46 .
it is worth noting that the crossover between @xmath182 and @xmath183 disappears and open a gap between the two .
when @xmath49 .
the tunneling is given by @xmath184 , where @xmath185 is the recoil energy , chosen as the natural energy unit .
* figure 2 mean - field topological phase diagram at a given chemical potential . *
( * a * ) topological phase diagram on the @xmath94-@xmath46 plane with ( @xmath186 , @xmath187 ) : we can distinguish three different topological regions from the behavior of the energy gap @xmath92 .
the white dotted lines represent phase boundaries where the energy gap equal zero , and the colors ranging from dark to red describe the values of @xmath92 .
( * b * ) plot of the energy gap @xmath92 and the spin polarization @xmath188 at zero momentum with respect to the zeeman field @xmath46 . here , the order parameter is chosen as @xmath189 , the tunneling between two layers as @xmath190 and the chemical potential as @xmath186 .
there is a jump in the spin polarization at @xmath79 , occurring precisely at the phase interface , where the topology of the fermi gases changes .
the phases are labelled with different colors : topological trivial superfluid ( * trivial * , blue ) with @xmath191 , a new type of topological superfluid ( @xmath192 , gray ) with @xmath193 , and normal topological superfluid ( @xmath194 , pink ) with @xmath195 .
the red triangles mark the critical values of zeeman field @xmath196 .
( * c*)-(*f * ) show the quasi - particle spectrum in momentum space at different zeeman field @xmath46 . as the zeeman field increases ,
the gap closes and opens when @xmath197 , and then re - closes at the point of @xmath198 .
the system evolves from a conventional superfluid to a @xmath192 superfluid and finally to a @xmath194 state .
other parameters are the same as in ( * b * ) .
* figure 3 phase diagram and the evolution of spin - orbit coupled bilayer fermi gases with the increasing chemical potential . *
( * a * ) topological phase diagram along with the energy gap , is shown on @xmath94-@xmath25 plane . with increasing the strength @xmath94 of the pairing potential
, the area of @xmath194 gets smaller and finally disappear above a threshold , @xmath199 . if we move on and reach the limit @xmath200 , the full topological trivial occurs in the absence of topological transition .
( * b * ) the energy gap @xmath92 is plotted as a function of the chemical potential @xmath25 with @xmath189 .
the colored regions are determined by gap closing equation : @xmath196 , as we have used in fig .2 .
there are five phase areas indicated by different colors and separated by the critical values of the chemical potential @xmath106 with @xmath201 , marked by the red triangles .
the zeeman field is fixed at @xmath202 , and the tunneling strength is taken as @xmath203 .
( * c * ) an illustration of the topological phase transition at different chemical potentials for small @xmath94 .
the single particle energy dispersion is drawn by solid lines and the dashed line is the chemical potential .
it is found that the system enter topological phase when the chemical potential is tuned to lie inside a zeeman gap .
* figure 4 bilayer fermi gases with synthetic spin - orbit coupling in a harmonic trap . *
the left panel gives a rough configuration of bilayer fermi gases in a harmonic trap for different chemical potential within the local - density approximation ( lda ) .
mfs just appear around the interface between different regions , approximately where the chemical potential satisfy the critical gap closing equation : @xmath196 .
when the chemical potential at the center of the trap @xmath25 is set to be so large that all four branches can be filled , as shown in ( * * a**1 ) .
we will find four pairs of @xmath106 from the center to the edge of the trap .
( * * b**1 ) and ( * * c**1 ) give the results with lower chemical potential , less pairs of @xmath106 can be detected in the trap .
the right panel shows the quasi - particle excitation spectrum @xmath118 ( in units of @xmath204 ) , calculated within the self - consistent bdg approach , at @xmath143 , @xmath205 and @xmath206 , respectively .
the number of majorana zero modes changes as the total number of atoms varies .
we therefore could have eight ( * * a**2 ) , six ( * * b**2 ) and four ( * * c**2 ) majorana fermions , corresponding to what we have discussed in the left panel . here
@xmath207 , @xmath208 and @xmath209 .
* figure 5 spatial dependence of the critical zeeman fields @xmath210 , @xmath211 and the superfluid order parameter @xmath147 at different phases . * in accord with fig .
4 , we present @xmath210 , @xmath211 and @xmath147 for a fermi gas of ( * a * ) @xmath143 , ( * b * ) @xmath205 , and ( * c * ) @xmath206 fermions in trap at zero temperature .
the symmetry around @xmath212 originates from the harmonic trapping geometry .
the phases distinguished by the local topological phase condition , are highlighted with different colours .
the dots in each subgraphs label majorana fermions associated with interfaces between different phases , where the zeeman filed cross the lines . by decreasing @xmath213 , the number of majorana zero modes changes with the deformation of topological phase structure . from ( * a * ) to ( * b * ) , the * trivial * area in the bottom of the trap shrinks and vanishes with disappearance of two majorana fermions .
therefore , @xmath192 occurs at the centre of the well . from ( * b * ) to ( * c * ) , similar to the above case , @xmath192 in the middle of the trap disappears and the number of majorana fermions becomes 2 pairs , corresponding to what we have shown in fig . | topological superfluid , new quantum matter that possesses gapless exotic excitations known as majorana fermions , has attracted extensive attention recently .
these excitations , which can encode topological qubits , could be crucial ingredients for fault - tolerant quantum computation .
however , creating and manipulating multiple majorana fermions remain an ongoing challenge .
loading a topologically protected system in multi - layer structures would be a natural and simple way to achieve this goal .
here we investigate the system of bilayer fermi gases with spin - orbit coupling and show that the topological condition is significantly influenced by the inter - layer tunneling , yielding two novel topological phases , which support more majorana fermions .
we demonstrate the existence of such novel topological phases and associated multiple majorana fermions using bilayer fermi gases trapped inside a harmonic potential .
this research pave a new way for generating multiple majorana fermions and would be a significant step towards topological quantum computation .
beijing national laboratory for condensed matter physics , institute of physics , chinese academy of sciences , beijing 100190 , china @xmath0e - mail : [email protected] topological state of matter @xcite , as an active new frontier in physics , has been attracting considerable interest both experimentally and theoretically .
the so - called majorana fermions ( mfs ) @xcite , accompanied with topological states , have a good prospect for applications in topological quantum computation , quantum memory and quantum random - number generation @xcite since they are intrinsically immune to decoherence caused by local perturbations .
due to such potential applications , the realization of topological states in a well - controlled environment is highly desirable .
recently , several theoretically predictions and experimentally observations of topological states have been reported in a variety of systems , for example , inas wires and banded carbon nanotubes engineered by spin - orbit coupling of electrons @xcite , the surface of topological insulators @xcite , ultracold atomic gases with strong non - abelian synthetic gauge field @xcite and so on .
spin - orbit ( so ) coupling plays a key role in all systems .
in contrast to the other physical contexts , ultracold atomic gases with more experimental controls over its parameters , such as high controllability in interatomic interaction , geometry , purity , etc , offer an exceptional clean platform for exploring the topological superfluidity .
recent experimental progresses in the synthetic so coupling of ultracold bose gases @xcite and fermi gases @xcite have stimulated a lot of interest in exploring exotic topological quantum properties of ultracold atoms . in particular , by deforming the fermi surface and opening a topological band gap , so coupled fermi gases become topological and support majorana zero - energy excitations in the presence of a large perpendicular zeeman field .
however , based on these experiments , the existence of a topological band gap can only lead to a single topological phase and a pair of majorana zero modes associated with topological quantum phase transition ( tqpt ) .
there is a strong requirement in the realm of quantum information and quantum memory for proposing a advanced system , which results in more types of topological phases , to create and manipulate multiple mfs . to achieve this
, we give a realistic scenario based upon engineering a topologically protected system in a bilayer or multi - layer structures .
as we know , some intriguing quantum effects , generated by layers s extra degrees of freedom , have been widely studied in many multi - layer condensed matter systems including fractional quantum hall states @xcite and topological valley transport in bilayer graphene @xcite .
interesting quantum effects such as atomic josephson effects and macroscopic quantum self - trapping @xcite have also been studied in ultracold bose gases . however , for ultracold fermi gases , most of the investigations have been developed mainly in the field of single - layer topological system@xcite , very little is known regarding the bilayer fermi gases with so coupling
. it would be interesting to examine the possibility for realizing and manipulating multiple mfs by loading so coupled fermi gases in a bilayer structure , providing a peculiar insight into the topological states . in this article
, we concentrate on the potential topological properties of bilayer fermi gases with so coupling and predict the existence of multiple majorana zero states at the interfaces between distinct phases . the bilayer geometry can be readily realized by adding a two - dimensional double - well optical lattice , and the atoms in each layer are affected by the same raman so coupling , while coupled with each other via the inter - layer tunneling . notably different from previous mean field studies , by tunning the inter - layer tunneling , traditional topological phase condition induced by the collective effect of so coupling and zeeman field is shifted and causes two critical transition points [ @xmath1,@xmath2 .
as a consequence , the topological phase region becomes separated into two parts : topo- and topo- phases .
here we provide a physical picture to classify the two different topological phases and discover that in suitable parameter regions , the system may undergo a unique tqpt from a topo- state to a topo- state , which has not been previously identified .
the region for topological superfluid depends on not only magnitude of the zeeman field , the chemical potential , the so coupling strength , but also on the inter - layer tunneling , thus provides more knobs in experiments .
furthermore , we give the self - consistent bogoliubov - de gennes ( bdg ) results of the system trapped in a harmonic potential and discuss the resulting multiple majorana zero - energy modes in great detail .
the number of exotic majorana zero modes depends on the number of interfaces between distinct phases that form in different areas of the trap , which are consistent with analytical local - density approximation ( lda ) expressions .
such so coupled bilayer fermi gases offer an opportunity to create and manipulate multi - mfs and a advanced multi - layer experimental setup may be therefore improved to measure more mfs . |
among the central questions in galactic structure and parameters are the solar motion , solar galactocentric radius @xmath13 and the local circular velocity @xmath2 of our galaxy .
galactic rotation curves are found to be generally quite flat over a vast range of galactocentric radii @xcite . as common for galaxies with exponential discs
there is some evidence for a radial trend of the galactic circular velocity near the sun , but it is very moderate @xcite , so that the circular velocity at solar galactocentric radius @xmath0 characterises well the entire potential . initially local kinematic data from stellar samples were the main source to extract galactic parameters including @xmath2 , e.g. by the use of the oort constants @xcite .
the kinematic heat of stellar populations requires large sample sizes , so that studies determining the local standard of rest ( lsr ) are still primarily based on stars , but some classical strategies like the position determination of the galactic centre pioneered by @xcite reached their limits by the constraints from geometric parallaxes , by the number of available luminous standard candles , and by the magnitude requirements of stellar spectroscopy .
apart from some more recent attempts to use luminous stars ( e.g. * ? ? ?
* ) , most current evidence on @xmath0 and @xmath2 derives from modelling streams in the galactic halo ( see e.g. * ? ? ?
* ; * ? ? ?
* ) and from radio observations of the @xmath14 terminal velocity ( see * ? ? ?
* ) , the galactic centre , molecular clouds and masers ( e.g. * ? ? ?
while the first branch relies strongly on assumptions such as distance scale and shape of the galactic potential ( cf .
the discussion in * ? ? ?
* ) , there are further complications like the failure of tidal streams to delineate stellar orbits @xcite .
radio observations @xcite have accurately determined the proper motion of the radio source sgr @xmath15 , which is identified with the central black hole of the milky way ( for a discussion of possible uncertainties see also * ? ? ?
it tightly constrains the ratio of the solar speed in the azimuthal direction @xmath1 to @xmath0 , but further information is required to obtain both quantities , letting aside the need for an independent measurement .
recently parallaxes to objects in the central galactic regions have become available ( see the discussion in * ? ? ?
* ) , and values for the galactic circular speed have been derived from @xmath14 motions and from masers @xcite . despite the decent errors in the determined kinematics of masers , the small sample sizes impose considerable systematic uncertainties : they are not on pure circular orbits and more importantly they are intimately connected to the intense star formation in spiral arms , where the kinematic distortions are largest . as the youngest strategy ,
studies of orbits in the galactic centre have gained high precision , but weakly constrain @xmath13 due to a strong degeneracy with the black hole mass @xcite .
hence the values for @xmath0 and @xmath1 remain under debate .
an independent determination of galactic parameters is facilitated by the new large spectroscopic surveys like rave @xcite and segue @xcite .
i will show that already now the stellar samples , which so far have been primarily used for the exploration of substructure @xcite give results for @xmath2 competitive with radio observations .
common ways to determine galactic parameters from the motion of stars require a significant bundle of assumptions and modelling efforts to evaluate the asymmetric drift in a subpopulation compared to the velocity dispersion and other measurements and assumptions , e.g. detailed angular momentum and energy distributions , the validity of the theoretical approximations , etc . on the contrary
a good measurement strategy for galactic parameters should have the following properties : * do not rely on dispersions and other quantities that require accurate knowledge of measurement errors . * do not rely on kinematically cold objects in the disc plane , which are particularly prone to perturbations from galactic structure . * do not rely on any specific models with hidden assumptions and parameters and require as few assumptions as possible . in an attempt to approximate these conditions
this note concentrates on the use of kinematically hot ( thick disc , halo ) populations .
the use of mean motions and the assumption of approximate axisymmetry in our galaxy will be sufficient to fix the galactocentric radius of the sun as well as the solar motion and galactic circular speed . in section [ sec : p7general ]
i lay out the general method before describing the sample selection and treatment in section [ sec : p7sample ] .
the latter comprises a discussion of proper motion systematics , a discussion of distances in subsection [ sec : p7dist ] and a discussion of line - of - sight velocities and the vertical motion of the sun . in section [ sec : p7globrot ]
i lay out the radial velocity based rotation measurement using segue , discuss the radial velocity of the sun in subsection [ sec : p7lsr ] and shortly discuss possible trends . from [ sec : p7galparam ] on i present three methods to to extract galactic parameters from stellar samples , the first two methods relying on the radial velocity based rotation measurement and the third on the mean direction of stellar motions . in section [ sec : p7conclude ] i summarise the results .
before i start the dirty work of data analysis it seems appropriate to concisely lay out the general definitions and ideas . this work is based on comparisons between velocities in the heliocentric frame and galactocentric velocities . in the heliocentric frame
i define the velocities @xmath16 in a cartesian coordinate system as the components pointing at the sun towards the galactic centre , in the direction of rotation and vertically towards the galactic north pole .
the galactocentric velocities are defined in cylindrical coordinates around the galactic centre with @xmath17 pointing again towards the galactic central axis , in the direction of rotation and vertically out of the plane , as illustrated in .
@xmath18 are the three components of the solar motion relative to the local standard of rest ( i.e. the circular orbit at the local galactocentric radius @xmath13 ) , for which i assume the values from @xcite if not stated otherwise .
the total azimuthal velocity of the sun in the galactic frame is written @xmath19 where @xmath2 denotes the circular velocity in the disc plane of the milky way at @xmath13 .
the mean rotation speed of a subsample of stars will be named @xmath4 and by the asymmetric drift is generally smaller than @xmath2 . for convenience
local cartesian coordinates are defined left - handed @xmath20 pointing outwards , in the direction of rotation and upwards with the sun at the origin .
commonly studies on the rotation of galactic components suffer from the uncertainty in the solar azimuthal velocity that directly translates into a systematic uncertainty of stellar azimuthal velocities .
there is a way out : as already discussed in @xcite and evident from , galactic rotation leaves its imprint in both the heliocentric @xmath3 and @xmath21 velocities .
the direction of the rotational component of motion @xmath22 turns with the angle @xmath10 between the lines galactic centre sun and galactic centre
star to align partly with the radial component @xmath3 of the local cartesian frame . from this systematic streaming motion in heliocentric @xmath3 velocities
we can directly infer the galactic rotation of a stellar sample .
accounting for rotation the mean motions of stars in the heliocentric frame are : @xmath23 with the rotation speed of the population @xmath4 and the galactic angle @xmath10 .
@xmath24 are the velocity components of the solar motion in the galactic rest frame radially towards the galactic centre , azimuthally in the direction of disc rotation and vertically out of the plane . observing the change of heliocentric @xmath3 velocity in a sufficiently extended sample hence estimates the rotation of a component @xmath4 , once we know the angle @xmath10 .
similarly the observed heliocentric azimuthal velocity @xmath21 shrinks towards larger @xmath25 .
prima facie it might be tempting to use this term . yet , all it expresses is a slowing of heliocentric azimuthal velocities . in general @xmath4 is not a real constant , but a function of altitude @xmath26 , galactocentric radius @xmath27 , metallicities , etc .
in particular the kinematically hotter disc populations at higher altitudes above the plane have a larger asymmetric drift ( see e.g. * ? ? ?
* ) , or vice versa slower rotation , seconded by an increase in the share of the nearly non - rotating halo . pointing out from the sun s position in the mid - plane , @xmath28 correlates via the typical distance range with altitude @xmath26 .
in contrast to @xmath3 , where this bias is of second order , it is of first order in @xmath21 and prohibits any naive estimate of @xmath4 from azimuthal velocities .
a similar uncertainty derives from the possibility for variations of @xmath4 in @xmath27 even at the same altitude .
the radial velocities are better behaved .
sample changes along the baseline @xmath29 are not only a second order effect , but are generally more robust : the greatest concern is a bias in the derived @xmath9 , which can be prevented by fixing its value by previous knowledge or by simply having a sufficient sky coverage that allows to balance both sides of the galactic centre and includes sufficient numbers of stars at low @xmath25 .
if the integrity of @xmath9 is ensured , sample changes along the baseline just result in different statistical weights of the involved populations .
the value @xmath4 derived from heliocentric radial motions is by construction independent from the assumed solar azimuthal velocity and ( in an extended sample ) is only weakly impacted by ( if held fixed ) a mis - estimate of @xmath9 .
this contrasts to direct measurements of @xmath22 that rely fully on the total solar azimuthal velocity @xmath1 - a short - come now turning into a virtue : @xmath1 and @xmath0 can be varied to force an agreement between the rotation estimates from @xmath4 and @xmath22 .
as the position - dependent statistical weights in @xmath4 are known , one can estimate a weighted average of @xmath22 exactly on these weights .
the sole assumption of approximate axisymmetry of the galactic disc ( but not necessarily in the sample selection ) is then sufficient to compare @xmath4 and @xmath22 and measure galactic parameters without the least concern about sample selection , velocity trends , etc .
the comparison gives a relation between @xmath1 and @xmath0 , which can be combined with independent data like the proper motion of sgr @xmath15 to provide a superior estimate of galactic parameters .
however , @xmath0 alone can be measured with this simple method alone , as @xmath4 is roughly proportional to @xmath0 ( see section [ sec : p7galparam ] or ) , so that any error in @xmath0 shears the estimate of @xmath1 for fast rotating populations against their slower counterparts .
any reader not interested into the details of sample selection , distance analysis and the proper motions may skip this section and turn directly towards section [ sec : p7lsr ] .
all data used in this study were taken from the seventh and eighth data release ( hereafter dr7 and dr8 * ? ? ?
* ; * ? ? ?
* ) of the sloan digital sky survey ( sdss * ? ? ?
* ) . the stellar spectra stem from the sloan extension for galactic understanding and exploration ( segue , * ?
an evaluation of the performance of their parameter pipeline ( sspp ) can be found in @xcite .
this work has no interest in specific metallicity distributions , but in having kinematically unbiased samples that include as many stars as possible .
hence the sample drawn from dr8 consists of a raw dataset of @xmath30 stars from all segue target categories and the photometric and reddening standard stars that do not include any proper motion cuts and have clean photometry .
in particular i use the categories of f turnoff / sub - dwarf , low metallicity , f / g , g dwarf , k dwarf , m sub - dwarf stars from segue1 , as well as ms turnoff , low metallicity and the reddening and photometric standard stars from all samples .
the categories were selected by the target bitmasks of the database .
some of these categories overlap with kinematically biased selections and the survey descriptions are not clear about a possibility of direct ( cross - selection of stars ) or indirect ( non - selection of stars that are targeted in a biased category ) bias .
i made sure that the presented results are unchanged when selecting stars by the unique target flags in the sspp . also checks on leaving out e.g.
k dwarfs that might overlap with the biased k giants did not provide evidence for a notable bias by the segue target selection .
the older dr7 forms a subset of @xmath31 stars . to ensure sufficient quality of the involved proper motions , i select only stars with a match in the proper motion identifications ( @xmath32 ) , the presence of at least @xmath33 detections ( @xmath34 ) and a good position determination with @xmath35 as a compromise between @xcite and currently used selection schemes .
i further require the formal errors on the proper motions to fulfil @xmath36 , removing the tail of uncertain proper motion determinations and require line - of - sight velocity errors below @xmath37 .
sample homogeneity is not a concern in this study , but to avoid issues with reliability of the isochrones in colour regions with insufficient coverage to check them by statistics , i select @xmath38 , which tosses out a handful of stars especially on the red side .
i tested that narrower colour cuts would have no significant effect on the presented results . for our purpose
it is most important to have a good discrimination between dwarf stars and other categories and to dispose of as many evolved stars above the main sequence as possible .
hence only stars are used that have values for gravity and metallicity determined by the pipeline .
the sample must be restricted to dwarf stars , as @xcite showed that stars with intermediate gravities in the pipeline are an indeterminable blend of dwarf , subgiant and giant stars , which renders their distance scatter and residual biases detrimental to any investigation of their kinematics .
following @xcite i adopt a sloping gravity cut generally tighter than the usually applied @xmath39 , demanding @xmath40 and setting a constant limit of @xmath41 for @xmath42 and @xmath43 for @xmath44 .
it is important to point out that this selection could especially on the metal - poor end still be irresponsibly lenient for studies of detailed stellar kinematics , as it will comprise a serious contamination with misidentified ( sub-)giants and does not bias sufficiently against the brighter turn - off stars . indeed
the distance corrections described below react to a tighter gravity cut in the metal - poor regime by up to @xmath45 . relying only in mean velocities the preference in this work
is leaned towards larger sample numbers and for assurance i checked that a tighter selection widens the error margins , but does not alter my results . quite frequently studies on segue use a cut in the signal to noise ratio @xmath46 to ensure reasonable parameters .
since the determination of stellar parameters is difficult in those noisy spectra , the detection of a mild drift in the derived quantities towards higher @xmath47 does not surprise , motivating a cut at @xmath48 where results become more stable .
de - reddening is done by subtraction of the reddening vectors given in the dr8 catalogue from the observed magnitudes . to limit uncertainties and avoid problems with complicated three dimensional behaviour of the reddening ,
i remove all objects with an estimated @xmath49 magnitude extinction @xmath50 .
this strips all stars in the low latitude fields from the sample .
i tested that the exclusion does not have a significant impact on my results apart from mildly increasing the derived formal errors , but maintain this cut for the sake of sample purity .
i exclude all stars flagged for spectral anomalies apart from suspected carbon enhancement .
the internal consistency of the distance estimator against closer distance bins starts eroding for @xmath51 .
hence i use this distance limit instead of the more common @xmath52 . for a precise measurement of galactic rotation
i am reliant on a decent control of systematic errors on proper motions .
unbiased noise is of no concern as proper motion errors translate linearly into velocities ( see equ . 20 in
* ) and hence mean velocities ( on which this paper relies ) are only affected by the mean value of proper motion errors . after a short description of proper motion systematics i will measure them in equatorial coordinates in section [ sec : p7pmeq ] and develop terms that are more natural to a contamination problem and galactic rotation in section [ sec : p7pmgal ] .
@xcite reported significant problems with the dr8 astrometry implying also problems with proper motions .
fortunately the revised astrometry has become available during refereeing stage of this paper and i will exclusively use the new data .
it should be noted that also dr7 had major problems , e.g. the entire @xmath53 @xmath54 was clearly erroneous / contaminated with large effects on kinematics .
its @xmath55 share of the total sample would bring down my later measurement of galactic rotation by as much as @xmath56 .
most problems with sloan proper motions have been adressed in the revised astrometry , but the large impact calls for an investigation of residual biases .
two main sources of systematic terms can be identified : astrometric `` frame - dragging '' and chromatic aberrations . despite the efforts to clean up the galaxy sample used for astrometry from residual contamination with galactic stars , the sample may still be subject to some astrometric `` frame - dragging '' . qualitatively those stars shift the frames in the direction of galactic rotation or respectively their motion relative to the sun , underestimating the real motion in the sample . as an alternative explanation @xcite traced the significant declination dependent net proper motions on quasars back to chromatic aberration by earth s atmosphere @xcite : the angle at which the stellar light passes through our atmosphere strongly depends on the declination and this should give rise to colour - dependent offsets .
further , deformations in the telescope can introduce minor distortions of the focal plane .
l|ccc & @xmath57 _ ra & @xmath57 _ dec & @xmath58 + f_i , ra 10 ^ 3 & @xmath59 & 0.18 & 0.16 + f_i , dec10 ^ 3 & -0.59 & @xmath60 & 0.50 + c_i & @xmath61 & @xmath62 & 0.033 while the existence of these aberrations appears plausible , i tested proper motions on the cleaner @xcite quasar sample .
i fitted a simple linear function in right ascension and declination to the proper motions from the revised dr8 astronomy : @xmath63[eq : p7radec ] where @xmath64 denotes the two possible proper motion components in right ascension ( ra ) and declination ( dec ) , @xmath65 are the coefficients , @xmath66 are free fitting constants . using the combined fit diminishes the impact from omitted variable bias that might arise from the particular sample geometry ( the explaining variables are mildly correlated ) .
the fit parameters are shown in table [ tab : qparradec ] .
there is a clear and significant detection of systematic proper motions of order @xmath67 .
i also checked the same fit on the older dr8 astrometry , where the derived terms were even slightly smaller .
the large term on @xmath68 can be expected from chromatic aberration . however , the strong effect on the right ascension is more difficult to explain by atmospheric influences , but may still come from specific distortions on the telescope or if the observations were made in a very particular manner that correlates the airmass of a stellar observation significantly to right ascension .
the sample size of just about @xmath69 quasars limits the possibility to dissect into subgroups .
however , cutting in observed colours there is no detectable difference between quasars at @xmath70 and @xmath71 implying that in the colour range interesting for the stellar sample the bias is quite consistent .
derived values go only adrift for @xmath72 , which bears no importance to the stellar sample .
the strong deviations on the right ascension motivate tests for astrometric `` frame '' dragging by stellar contamination . in the following
i develop a very crude analytic approximation for a proper motion bias in the @xmath73-plane as it could arise from galactic streaming - note also that this toy is not used in any other way throughout the paper apart from the proper motion corrections : candidates for contamination of the galaxy sample should preferentially populate a certain magnitude range and also have preferred colours resembling the galaxies used .
it hence seems appropriate to assume for simplification some dominant distance for stars contaminating the sample , placed at @xmath74 .
the outcomes will anyway not be critically affected by this distance ( which can alter the shape of the systematic bias once the stars are sufficiently remote for the small angle approximation to break down ) . on this shell
i assume the stars to rotate dependent on their altitude @xmath26 around the galactic centre with @xmath75 .
the proper motion can be written down from the heliocentric velocities from equation ( [ eq : p7rot ] ) subtracting the reflex motions of the solar local standard of rest ( lsr ) values from @xcite and assuming a total solar azimuthal velocity according to the iau recommendations at @xmath76 . using this the simple galactic streaming terms are : @xmath77 with @xmath78 where @xmath79 does the unit conversion , @xmath3 and @xmath21 are the heliocentric velocities from streaming minus the assumed motion of the sun and @xmath80 are the fit parameters .
the terms connected to @xmath81 give a crude estimate for the effects from halo streaming .
.fit parameters for systematic streaming in the quasar proper motions in dr8 with and without cut for @xmath82 as described in the text .
[ cols="<,^,^,^,^",options="header " , ] the derived values for dr8 are presented in table [ tab : qpar ] .
the old , erroneous dr8 and dr7 produce similar results . slicing the sample in colour i could not detect any significant changes , just the bluemost @xmath83 of the quasars appear to be different .
still , cutting for @xmath82 does not evoke notable consequences @xmath83 , as well as removing the @xmath45 of objects with the largest proper motions does not have any impact .
the coefficients @xmath84 and @xmath81 may be understood roughly as contamination fractions , but not on accurate terms because the distance dependence of proper motions effects a strong dependence of the fractions on the assumed distance of stellar contaminants .
the sign in the expected trends is not predictable a priori , as the @xcite sample may harbour itself some residual stellar contamination that competes with possible astrometry problems . in both cases the main rotation term
@xmath84 is significantly negative hinting to some more misclassified stars affecting astrometry as compared to the quasar sample . on the other hand the coefficients @xmath81 that resemble terms expected for the galactic halo
are slightly positive , which might be interpreted as a tiny surplus of halo stars hiding in the quasar sample .
i rather understand it as an omitted variable bias : neither do i know the exact functional shapes and distance distributions of the contamination , nor do i have precise terms at hand to cope with the possible chromatic aberrations .
exact terms for contamination would demand a complete modelling of the entire measurement process including a reasonable model to the milky way , other galaxies and the exact sloan selection function .
if i fit alone either @xmath81 or @xmath84 , both variables turn out negative and still their counterparts are negative when i fit them with this naively derived value .
so one could argue that @xmath84 is getting excessively negative being balanced by a slightly positive @xmath81 to cope with these uncertainties . in the end
i can not achieve more than a rather phenomenological correction without being able to determine the true cause for the observed trends .
the naive approximation still fulfils its purpose to parametrise the influence on galactic rotation measurements . throughout the following paper
i will use the terms for @xmath85 to correct stelar proper motions independent of colour , since no significant colour dependence was detected . for proper motion errors ,
which influence the distance determinations , i use the pipeline values and checked that using the values from @xcite for dr7 instead does not significantly affect the presented analysis . in this work
i make use of the statistical distance correction method developed by @xcite .
the approach relies on the velocity correlations over galactic angles that are induced by systematic distance errors and entails a better precision and robustness than classical methods .
the method can find the average distance in any sample of a couple of hundred stars .
the only thing required as first input is a smooth distance or respectively absolute magnitude calibration that has a roughly correct shape in parameters like colours and metallicity .
the shape has some importance as erroneous assumptions would result in an increased distance scatter on the sample bins and thus in residual systematic trends for different subpopulations .
the use of statistical distance corrections is mandatory as there are in all cases expected offsets by reddening uncertainties , age uncertainties , the uncertain binary fraction , mild differences by evolutionary differences on the main sequence , systematic metallicity offsets and by the helium enhancement problem , where the isochrones appear to be too faint at a fixed colour .
primary stellar distances are derived from a dense grid of basti isochrones @xcite that was kindly computed by s. cassisi for the purpose of our age determinations in @xcite .
i use the @xmath86 isochrones which seems an appropriate compromise between the suspected age of the first disc populations ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the age of halo stars . for metal - poor stars with @xmath87
i account for alpha enhancement by raising the effective @xmath88 by @xmath89 as suggested by @xcite or @xcite . in analogy to the observations of @xcite i let the alpha enhancement go linearly to zero towards solar metallicity : @xmath90 some of the stars are located bluewards of the turn - off point at this age . to cope with these objects , i search the point at which the old isochrone is @xmath91 brighter than the corresponding @xmath92 isochrone and extrapolate bluewards parallel to the main sequence defined by the @xmath92 isochrones .
i also experimented with using the turnoff point , i.e. the blue - most point of the old isochrone , but this introduces larger shot noise in distances that would hamper the distance corrections via the method of @xcite . comparing the expected @xmath93-band absolute magnitude of each star to the measured reddening corrected @xmath94 ,
i then infer its expected distance modulus and hence distance @xmath95 . to derive a correction field , i bin the sample in metallicity and @xmath96 colour .
i also attempted binning the sample simultaneously in distance as well , but found no additional benefits , probably because the bulk of the sample is further away than @xmath97 .
i also excluded high reddening regions so that variations of reddening with distance are comparatively small .
there is a trade - off between resolution of the smallest possible structures in the correction field and the introduction of poisson noise by lower effective numbers of stars contributing to the estimate at a certain point in the observed space .
the strategy is to select every 30th star in the sample and estimate the distance shift for its 700 closest neighbours ( i checked that the bin size has no significant effect ) using an euclidean metric on the parameter space : @xmath98 for non - halo or respectively metal - rich stars , which offer less accurate statistics by their smaller heliocentric velocities , this results in some significant scatter , which is reduced by smoothing the distance corrections via a weighted logarithmic mean : @xmath99 where @xmath95 is the a priori distance estimate from isochrone interpolation .
the first sum runs over all stellar parameter sets @xmath100 around which the best - fit distance correction @xmath101 or respectively the distance mis - estimate factor @xmath65 has been evaluated . the second index @xmath102 runs over used parameters ( here metallicity and colour ) , to evaluate the parameter differences @xmath103 between the star in question and the mean values of all evaluation subsets and smooth via the exponential kernel with the smoothing lengths @xmath104 .
i chose @xmath105 and @xmath106 .
when binning the sample a second time for measuring galactic parameters , i fit the distance for a second time to diminish the danger of remaining systematic biases .
anyway there is no evident trend , e.g. with distance , in the second step and the remaining corrections are on the noise level .
@xcite showed that in the presence of significant intrinsic distance scatter the method a slight mean distance underestimate occurs , which is caused by stars with distance overestimates populating the edges of the fitting baseline and hence acquiring larger weight .
this bias was found at @xmath107 , where @xmath65 denotes the relative distance error and @xmath108 its variance .
i confirmed the validity of the prefactor @xmath109 for the sample in use by applying a range of gaussian broadenings to the first guess isochrone distances .
though the prefactor is known , it is not straight forward to estimate the values of @xmath110 dependent on metallicity and colour . on the red side
the stars are relatively firmly placed on the main sequence , the reddening vector runs quite parallel to the main sequence , and hence the colour and metallicity errors dominate justifying a moderate error estimate @xmath111 . on the blue
end the placement of the stars on the turn - off becomes increasingly uncertain and especially on the metal - rich side one must expect significant spread in ages and hence absolute magnitudes as well .
i hence adopt a mild increase in distance depending on colour @xmath112 which is larger on the blue side and increases mildly with metallicity .
i limit the relative distance increase to a maximum of @xmath113 , which corresponds to a distance scatter @xmath114 .
shows the finally adopted absolute magnitudes ( upper panel ) and the correction field for the isochrone distances ( lower panel ) in the colour - metallicity plane . along the blue metal - poor stars i require moderate distance stretching
this is the turn off region , where the @xmath86 isochrones might be a little bit too young and get shifted upwards in addition to the known suspicion of isochrones being too faint . on the other hand a bit of contamination by subgiants
is more likely because of the reduced difference in surface gravity and weaker absorption lines .
the problems with gravities for the most metal - poor stars are indicated by the absolute magnitudes bending back , i.e. the most metal - poor stars being made brighter again at the same colour .
carbon stars constitute another bias having stronger absorption in the g - band , making them too `` red '' in the colour assignment , which brings forth a brightness underestimate .
in principle the method explores the structure around the turn off and the dominant ages among these populations , but owing to the scope of this work i defer this to a later paper . at higher metallicities the turn off line evidently bends upwards , i.e. to redder colours as expected from stellar models . on the other hand
there is evidence for a younger metal - rich population at blue colours , indicated by the absence of distance under - estimates contrasting with their metal - poor antipodes .
some vertically aligned features might point to minor systematics in the pipeline metallicities ( e.g. under - estimating a star s metallicity makes us underestimate its distance ) .
the little spike near the turn - off around @xmath115 could be an unlucky statistical fluctuation .
if it was real and not caused by a problem in the pipeline it would indicate a rather sharp transition in ages with relatively young stars at slightly higher metallicities opposing the very old population in the halo regime . throughout the range of g and k
dwarfs there is a quite constant need for mildly larger distances than envisaged by naive use of the stellar models . to test my approach for consistency ,
i experimented with the @xcite ( a7 ) relation as well and found that as expected the distance corrections robustly produce an almost identical final outcome despite the highly different input .
despite all efforts put into the segue parameter pipeline , it can not be granted that we can trust the radial velocities to be free of any systematic errors of order @xmath116 . to shed light on this problem ,
let us examine the vertical velocity component .
a good estimate of vertical motion can be derived from the galactic polar cones , where proper motions and hence distance problems have almost no impact on the measurement .
luckily segue has both the galactic north and south pole and i use @xmath117 , while checking that a stricter cut at @xmath118 does not significantly alter the findings .
these polar cones display a surprisingly low @xmath119 .
this is @xmath120 off the old value from hipparcos and the gcs .
while this may of course trace back to some unnoticed stream in the solar neighbourhood , it seems appropriate to check for consistency .
@xmath121 stars towards the north pole show @xmath122 , while @xmath123 stars towards the south pole give @xmath124 .
allowance for a variation in the line of sight velocities : @xmath125 estimates @xmath126 and @xmath127 .
the tighter cut in latitude gives the same line of sight velocity deviation , but @xmath128 .
binning in metallicity brings me to the limits of statistical significance , but reveals a rather robust uptrend in the estimated @xmath129 offset from near to negligible at solar metallicity to of order @xmath130 near @xmath131 \sim -2 $ ] .
it is tempting to argue in accordance with @xcite that all those trends are reaction of the galactic disc to some satellite interaction , but such reasoning is contradicted by more of the signal coming from the halo stars , which can not participate in such a response .
an explanation by streams conflicts with the broad metallicity range concerned that would require a conspiracy of nature
. a simpler explanation might be a warp in the disc , but again this conflicts with the metallicity trend .
the easiest explanation for the deviation is a residual error in the pipeline . to avail this and stay as close as possible to the pipeline values , i use henceforth a minimal correction that keeps the discussed trend just below significance : @xmath132 allowing the corrections to vary between @xmath133 and @xmath134 .
shows the systematic motion for the sample projected into and binned in the galactic plane . at each bin
i evaluate the mean motion in the heliocentric frame ( left panel ) and the mean motion in the galactocentric frame ( right ) .
while i would expect some minor systematic motion even in the galactocentric frame by spiral structure and maybe influence by the bar , such contributions will be relatively small as most of this sample is at high altitudes , where the stars with their large random velocities experience less important changes of motion by the relatively shallow potential troughs of the spiral pattern .
some of the residual structure will likely be blurred out by poisson noise and distance uncertainties , so that it is no surprise not to see any appreciable signal . as a good sign
there are no direction - dependent shifts , which would be expected under systematic distance errors , which make especially the azimuthal velocity component cross over into the radial velocity determinations . in the left panel
, however , we see the very prominent rotational streaming of the galactic disc , which is central to the following investigations .
the strong influence of rotational streaming on the observed radial motion makes any determination unreliable that does not explicitly take it into account .
before i turn to its positive application , i lay out a simple procedure to deal with this effect as a contaminant and measure the solar radial lsr velocity . in the literature varying offsets of the solar radial motion of up to more than @xmath11 from the value derived from local samples have been claimed .
besides possible velocity cross - overs , e.g. the motion in @xmath21 translating into a vertical motion ( for an example see fig . 7 in * ? ? ?
* ) invoked by systematic distance errors @xcite , neglected sample rotation can be the reason .
in any sample that is not demonstrably rotation - free and is not symmetric in galactic longitude the streaming motion offsets the mean heliocentric radial motion and hence the inferred solar motion .
the error grows with the remoteness of the observed stars .
shows the heliocentric @xmath3 velocities for all stars ( red crosses ) from dr8 and stars with @xmath135 ( light blue points ) against @xmath136 .
i set @xmath137 , and the value for @xmath1 accordingly to @xmath138 .
the metal - rich subsample displays a nearly linear relationship of the mean @xmath3 velocity versus @xmath136 as expected for rotational streaming from equation ( [ eq : p7rot ] ) .
this streaming motion is far larger than the expected offset caused by the solar lsr motion @xmath139 @xcite .
even the mild asymmetry in @xmath10 distorts a naive estimate of @xmath9 in disregard of rotation to @xmath140 for metal - rich objects with @xmath135 and @xmath141 for the entire sample .
consistent with the rotational bias the deviation from the standard lsr value is significantly larger for the disc sample .
to rid my results from this bias , i use the estimator @xmath142 where the free parameters @xmath4 and @xmath143 are the effective rotation and the reflex of the solar motion , @xmath144 describes noise from both velocity dispersion and observational scatter that should have zero mean independent of @xmath10 .
i obtain @xmath145 for the metal - rich subsample and @xmath146 for all stars .
the formal errors are mildly underestimated due to the non - gaussianity of the @xmath3 velocity distribution and a non - weighted fit .
this differs from the @xcite estimate at marginal significance .
yet there are larger uncontrolled systematic biases : this fit implies that the effective rotation term carries away all the observed bias on @xmath9 .
while the detailed rotation pattern is not of interest here , the measurement will still be biased if one encounters a larger fraction of slow rotators on one side than on the other .
e.g. a different number of halo stars or unmatched differences of rotation with galactocentric radius can distort @xmath4 , which correlates with @xmath9 . in this sample @xmath9 shrinks by @xmath147 for every @xmath11 decrease in @xmath4 .
in the present case such effects are below detectability , as can be seen from the lower panel in , where the binned means ( green error bars ) show no significant deviation from the linear relationship .
further , any remaining systematic distance errors can lead to systematic velocity shifts .
the first error can be controlled by selecting only stars with small @xmath10 , the second by selecting stars near the galactic centre and anticentre directions . in principle
this could be tested by geometric cuts on the sample , however , there are no data near the galactic plane . exploring
the full parameter space of possible cuts reveals some expected fluctuations of order @xmath148 , but no striking systematics .
for example @xmath149 gives @xmath150 in concordance with the full sample .
more importantly , cuts for a smaller distance range , which is more reliable concerning the proper motion uncertainties favour @xmath151 .
the difference is caused by the @xmath152 most remote stars displaying a rather extreme @xmath153 .
e.g. cutting for @xmath154 i get @xmath155 , again consistent with @xmath156 for @xmath149 , demonstrating that this drift in distance is not an effect of some rotation error . a similar rise in @xmath9 to up to @xmath157
can be achieved by increasingly tight cuts on the signal to noise ratio , which indirectly selects a shorter mean distance . a global line - of - sight velocity increment
@xmath158 increases the estimate for @xmath9 by about @xmath147 .
it is interesting to investigate into the question of a trend in radial velocities with galactocentric radius @xmath27 as suggested in @xcite . to this purpose
i estimate the equation @xmath159 where @xmath4 is the effective rotation , @xmath27 the galactocentric radius , @xmath160 is the effective ( heliocentric ) radial velocity gradient on the sample , and @xmath161 is the noise term from velocity dispersion .
for convenience i limit @xmath162 as before and get on the full distance range @xmath163 at @xmath164 . on the shorter distance range of @xmath165
i get @xmath166 and @xmath167 , delivering no sensible gradient at all . restricting @xmath135
i get @xmath168 .
this residual trend is caused by the @xmath169 stars with @xmath44 measuring @xmath170 . while this may finally be the desired signal , it is puzzling to have no more metal - poor counterpart .
hence this is more likely either be a statistical fluctuation or some flaw in the sample .
parameters and distances for stars with super - solar metallicity in segue are likely problematic , particularly as i do not have sufficient numbers to constrain their mean distance very well .
it has to be noted that this sample is at rather high altitudes and is dominantly in the outer disc , so that velocity trends in objects of the inner thin disc might not show up .
hence the missing detection of an appreciable trend ( and exclusion of a significant deviation from circularity ) does not disprove @xcite .
more of a problem for their result should be the larger value of @xmath9 , which likely reduces or diminishes their estimated gradient , as their analysis mainly relies on assumption of the old value and does not have sufficient data points in the outer disc to provide a decent fix .
after all , when accounting for the systematic uncertainties in the data , i can not detect any reliable hint for a motion of the solar neighbourhood against some more global galactic standard of rest . combining these estimates with the result from the mean direction of motion ( see section [ sec : p7direct ] ) of @xmath171 with a systematic uncertainty of @xmath172 and the reader may either use this value or a lower @xmath173 , taking into account the value to the older gcs .
so far i exclusively looked at the global rotation pattern and its influence on @xmath9 , but the new rotation estimator can do more : any measurement relying on observed azimuthal velocities does not provide the real rotation of the component in question , but only a value relative to the sun , from which we conclude its azimuthal speed by discounting for the galactic circular velocity plus the solar azimuthal lsr motion ( plus a small part of the solar radial motion when @xmath174 ) . on the contrary
the radial velocity part of ( [ eq : p7rot ] ) uses the velocity differences at a range of galactic positions and gives the `` real '' rotation of the component without significant influence by assumptions on the motion of the sun .
the measurement is in most cases less precise than the azimuthal measurement , but is nevertheless interesting by its different set of assumptions .
consider a possible presence of stellar streams : they can distort azimuthal velocity measurements soliciting an independent source of information : in general a stream will then show up in highly different values for the two rotational velocity indicators .
the price one has to pay for the radial velocity estimator is a larger formal error , the need for extended samples , a different dependence on the assumed @xmath0 ( which is usually a bonus ) and some stronger vulnerability to distortions from axisymmetry , e.g. influence by the galactic bar .
as discussed by @xcite , dissecting a sample by kinematics is detrimental for this kind of kinematic analysis , but one can divide the galaxy into its components by a selection in metallicity . shows the rotation measurement from the radial velocity estimator when i dissect the sample into bins of @xmath175 objects in metallicity and move half a bin in each step ( large red error bars ) . the two plateaus of disc and halo rotation with the transition region in between show up nicely .
the plateau of disc rotation stays about @xmath176 below the circular speed of the disc due to asymmetric drift .
clearly there is no significant average net rotation in the metal - poor galactic halo . despite the far smaller formal error , this would be more difficult to state on firm grounds by using the average azimuthal velocity alone ( blue error bars ) , which does depend strongly on the solar galactocentric motion .
in fact the azimuthal velocity estimator has been shifted onto the @xmath3 estimator s curve by matching @xmath177 , which turns us to the topic of the next subsection .
more important than mutual confirmation of results one can play off the two rotation measurements ( from @xmath3 and @xmath22 ) against each other .
as stated before , only @xmath22 depends on the solar azimuthal velocity , so that forcing both rotation estimators into agreement we obtain an estimate for the total solar azimuthal velocity in the galactic rest frame @xmath1 . by subtraction of the solar lsr motion
this directly estimates the galactic circular speed @xmath2 near the solar radius without any need to invoke complicated models .
the apparent difference caused by a wrong @xmath1 is demonstrated in the top panel of .
for this plot i lowered @xmath1 from the optimal value by about @xmath37 .
comparison with the previous figure reveals that the red bars are unchanged as expected , while the azimuthal velocity based measurements are pushed upwards into stark disagreement . to bring the two rotation measurements on common ground one has to apply weights to the estimate of the mean azimuthal velocity @xmath178 for each subpopulation : in a least squares estimate on the rotation speed @xmath4 each star appropriates a weight proportional to its distance from the population
mean on the baseline @xmath179 . as the value @xmath180 invoked by rotation varies linearly on this baseline , each star attains the weight @xmath181 when measuring rotation . introducing a simultaneously weighted mean for the mean azimuthal velocity @xmath182 where the sum runs over all stars @xmath100 , any structural differences between the two estimators
are eliminated , since every population enters both estimators with equal weights .
this measurement of the solar velocity does thus not require any modelling of galactic populations or any other deeper understanding at all .
the procedure is then very simple : * assume @xmath0 and a first guess solar velocity vector @xmath183 * estimate @xmath4 and @xmath184 . *
reduce @xmath185 by the difference @xmath186 and recalculate @xmath4 , @xmath184 until convergence .
the only remaining uncertainty ( apart from systematic measurement errors ) can arise if the geometry of the sample allows for a distortion in the radial velocity estimator .
i hence recommend controlling the value of @xmath9 , which is well - behaved for this sample , and in other case fixing it to a predetermined value . in the latter case
the weights of @xmath184 must be changed to @xmath187 .
the weighted averages on the azimuthal velocities are drawn with the green line in .
generally the weighted values in the disc regime are a bit lower , because the extreme ends on @xmath188 are populated by remote stars at higher altitudes and at larger asymmetric drift outweighing other effects like the polewards lines of sight reaching higher altitudes . via the galactic angle @xmath10 the assumed solar galactocentric radius enters the derivation .
as depicted in the larger @xmath0 , the smaller the derived @xmath189 will be for stars on our side of the galaxy .
this results in a larger rotation speed @xmath4 required for the same observed pattern and so finally increases the estimate of @xmath1 .
fortunately the rotation speed of galactic components is lower than the solar value and further in contrast to the linear oort constants the dependence on @xmath10 is not linear in distance , which helps towards a relationship flatter than a mere proportionality between @xmath2 and @xmath190 .
this is shown with the red error bars in . for the practical calculation
i bin the sample in distance and move a mask of @xmath175 stars width in steps of @xmath191 stars .
this stabilises the results against scatter from the coarse binning , the error is corrected for the induced autocorrelation .
the result is then drawn from a weighted average of the distance bins .
the binning in distance is mandatory in this sample not mainly because of reddening uncertainties , but because of the previously described anomaly in the mean radial velocities with distance .
this way contaminations are kept at a minimum level . by using the proper motion of sgr @xmath15 ( determined as @xmath192 by * ? ? ?
* ) as a second constraint the galactocentric radius can be fixed to @xmath193 and @xmath194 . for the resulting circular
speed one has to subtract about @xmath195 .
one can even gain an estimate for the galactocentric radius that is fully independent of any other results .
for this purpose we have to recapitulate again the impact of @xmath0 on the derived @xmath1 for different populations , shown in : the measurement of @xmath1 compares the weighted azimuthal velocity in any population to the derived absolute rotation @xmath4 .
changing the galactocentric radius @xmath0 makes the galactic coordinate system bend differently and hence gives only minor changes to @xmath22 , but strongly affects @xmath188 . this way @xmath4 is nearly proportional to @xmath0 .
a weakly rotating population has a far smaller value of @xmath4 and hence experiences a far smaller change than fast rotating disc stars .
if we now underestimate @xmath0 , the estimates for @xmath4 are compressed compared to the @xmath22 based estimator as demonstrated in the bottom panel of . in other words
the estimates of @xmath1 for disc stars get sheared against the estimates for halo stars .
however , for all populations one measures the same quantity , i.e. the azimuthal speed of the sun , which can not change between the samples . a deviation in @xmath0 is hence easily detected by dissecting the sample in metallicity , which is ( as can be seen from figure [ fig : p7rotmet ] ) a very good proxy for rotation .
i cut the sample again into bins of @xmath175 stars now in metallicity instead of distance and evaluate the trend @xmath1 @xmath196 of @xmath1 against the mean azimuthal speed between the bins by fitting the linear equation @xmath197 with @xmath198 as additional free parameter .
from the red error bars in one can see that already on the quite modest sample size and extension of the @xmath199 usable stars in dr8 i get a useful , independent estimate of the solar galactocentric radius @xmath200 . as in the first measurement i used the absolute value of rotation , while here the unit - less slope @xmath196 is employed ,
the two results are formally independent and can be combined to @xmath6 in excellent agreement with @xcite .
this translates to @xmath7 . working in the lsr value of @xcite
i hence obtain @xmath8 . to assess the systematic uncertainties i vary some of the assumptions .
first i checked that the choice of bin size does not change the results more than the general noise induced by bin changes as long as the bins are not unreasonably small .
galactic rotation at fixed @xmath0 is increased by about @xmath120 , when i globally augment distances by @xmath45 and would be about @xmath201 lower without the @xcite distance correction . increasing the distance scatter by applying an additional @xmath45 raises @xmath1 by @xmath202 at fixed radius , application of a @xmath203 error lowers it by @xmath204 .
the mild lowering is expected , since the strong distance scatter should give a global distance underestimate of @xmath205 ( consistently the rotation measurement from @xmath3 velocities drops by @xmath206 ) .
it is obvious that such a scatter has a minor effect , as the relation between @xmath180 and @xmath188 is linear and hence at small changes the mean of @xmath180 is only mildly affected .
this is a benefit from operating remote from the galactic centre where the changes in galactic angle become more rapid .
the mean azimuthal velocity is mostly affected by distance scatter via the bias on average distance derived from our statistical method .
tests reveal an expected systematic bias of order @xmath116 .
similarly the slope of the derived rotation speed @xmath207 depends mostly on distance . here
the dependence is larger : it reacts by @xmath208 , i.e. about @xmath209 standard deviations to a forced increase of @xmath45 in distance and by @xmath210 to a forced decrease of @xmath45 .
removing the statistical distance correction would lift values by about @xmath211 .
increased distance scatter without distance correction increases the value by @xmath212 for an additional scatter of @xmath83 and lowers it by @xmath213 for a scatter of @xmath203 .
gravity cuts have minor influence on my results : the mean rotation rate is lowered by @xmath214 when i lift the minimum gravity to @xmath215 and at the same time the trend in azimuthal velocity rises by @xmath216 .
tighter cuts are not feasible for the trend , as i lose all metal poor stars . replacing the sloping cut by a fixed @xmath39 ,
did not significantly alter the results ether . while the distance estimator strongly hints to some contamination of the dwarf sample , the observed deviation in @xmath0 is still not significant and likely results partly from random scatter , as many metal - poor objects are lost by this tighter cut .
the robustness of the result against changes in the gravity selection points to the statistical distance estimator being able to cope with the different contamination by adjusting the mean distances .
the largest uncertainty appears to arise from the proper motion determinations .
the correction i apply raises rotational velocities by about @xmath217 and in light of the unsatisfactory physical reasoning as discussed in section [ sec : p7pm ] it is uncertain how well the correction works . a little reassurance is given by the second estimator , its unitless slope rises by only @xmath218 upon removal of the proper motion correction .
the applied line - of - sight velocity correction lowers @xmath1 by about @xmath219 and lowers the unitless slope by just @xmath220 .
the better robustness of the second estimator for @xmath0 reflects the fact that any spurious rotation affects both halo and disc rotation in a similar way and hence their difference is more robust than the mean .
further i find that a change of @xmath9 in the calculation of galactocentric azimuthal velocities ( a minor bias is expected in the lopsided sample ) is irrelevant to the estimated quantities .
however , if the rotation estimates themselves are done with fixed @xmath9 instead of using it as a free parameter , the estimates agree fully when setting @xmath221 , and @xmath1 falls by about @xmath201 for every @xmath116 reduction in @xmath9 .
depending on how much integrity we ascribe to the astrometry in the sample this sets an interesting relation between @xmath9 and @xmath0 .
similarly @xmath9 can be determined by demanding the estimate for @xmath1 to be consistent in galactic longitude @xmath222 .
not surprisingly the favoured value is again @xmath12 .
i further tested that there are no trends of the rotation estimate ( with fixed @xmath9 ) with colour , distance or latitude .
the estimator of @xmath9 from section [ sec : p7lsr ] depends weakly on distance .
@xmath9 increases by about @xmath223 for a global @xmath83 increase in distances and decreases by about the same amount for a distance increase .
its counterpart @xmath4 changes by around @xmath224 with some fluctuations as different distance cuts can affect the sample composition .
the proper motion correction is responsible for about @xmath217 of the measured @xmath4 and @xmath225 of @xmath9 , while the line - of - sight velocity correction acts in the opposite direction , lowering @xmath4 by @xmath120 and @xmath9 by @xmath226 . as a third strategy for fixing galactic parameters
i suggest the direction of motion .
all that needs to be done is to compare the expected galactic angle @xmath10 to the angle that the galactocentric mean motion in the subsamples has against the local azimuth .
the upper panel in shows the angle @xmath227 of the mean velocity for a binned sample of all stars with @xmath228 ( above that value i expect sufficient rotation ) assuming a solar azimuthal velocity @xmath177 , @xmath229 and plotting only bins with more than @xmath230 stars .
the lines show positions of constant galactic angle @xmath10 . for easier fits
i note that the ratio @xmath231 behaves as @xmath232 as long as we stay away from @xmath233 ( avoiding the arctangent gives better convergence when starting far from the optimal parameters ) .
it is obvious that this kind of statistic critically depends on the choice of @xmath9 and the total azimuthal velocity of the sun .
formally the agreement between the position angle and the velocity angle is optimised by fitting @xmath234 @xmath235 to @xmath236 , where @xmath237 does the correction from the first guess @xmath238 to the parametrised solar radial velocity @xmath9 and @xmath239 and @xmath240 are the coordinates of the local cartesian frame in the radial ( outwards ) and azimuthal direction . while i can fit @xmath9 directly to these data ,
this is not possible for @xmath241 as the fit would then converge to the wrong global minimum : let @xmath241 and @xmath0 go to infinity and the fit becomes perfect with all ratios / angles zeroed . in this light
i vary @xmath241 externally and allow for @xmath9 and @xmath0 as free fit parameters .
all bins were weighted by the inverse number of stars they contain and i only accepted bins with more than @xmath230 objects .
i note that velocity dispersions potentially exert a ( predictable ) distortion on @xmath236 by its non - linearity , but found this effect to be very small on the current sample .
it is not surprising that the result for the solar radial velocity is @xmath242 in excellent agreement with the direct measurement from stellar @xmath3 velocities and with section [ sec : p7systerr ] .
the result for @xmath0 is shown in the lower panel of , where i plot the new datapoints from angle fitting in addition to the previously discussed data .
the formal error on @xmath0 at fixed @xmath1 is a small @xmath243 , but their relation largely resembles that of sgr @xmath5 .
i would need a larger sample extension and/or significantly more stars to compete with sgr @xmath5 , which will be achieved with gaia .
for now , formal inclusion of the third constraint hence does not change the final results and just reduces the formal errors a bit .
nevertheless the potential of this estimator for larger samples is evident .
the residuals of the fit will likely outline perturbations of the galactic potential and substructure with data from gaia , but in the present sample i could not detect any structure . adding stars with halo metallicities to the test
just drives up the error bars by a factor of @xmath244 because of the higher dispersion , while no rotating stars are added .
@xmath0 stays the same while @xmath9 shifts to @xmath245 . restricting the disc sample to a distance closer than @xmath246 increases @xmath0 and @xmath9 by insignificant @xmath247 and @xmath248 .
the value for @xmath9 depends on the position of the galactic centre .
if our coordinate system was off - centre by @xmath249 degree in galactic longitude , @xmath9 could shift down by about @xmath224 .
the most important outcomes of this work are three modelling - free and simple estimators for the solar azimuthal velocity and hence the local circular speed @xmath2 and for the solar galactocentric radius @xmath0 .
on this course i developed the idea that in a spatially extended sample the absolute rotation of stellar components can be measured from systematic streaming in the heliocentric radial direction .
the stars on one side of the galactic centre show an opposite heliocentric @xmath3 velocity to those on the other side .
this value has a lower formal precision than the classically used azimuthal velocity , but can boost accuracy by its relative independence from assumptions about the velocity of the sun .
the rotation in any extended sample severely affects determinations of the local standard of rest : @xmath9 and @xmath250 are frequently determined via simple sample averages in each component .
the rotation bias in @xmath9 gets more important with increasing distance and impacts all presently available big surveys , since they are lopsided , i.e. asymmetric in galactic longitude . for segue
dwarfs it amounts to @xmath251 .
accounting for rotation the otherwise observed difference between disc and halo stars disappears and combining this classic method with the estimate from the mean direction of motion described below , i find @xmath252 with an additional systematic uncertainty of about @xmath148 .
the value is @xmath253 larger than from the geneva - copenhagen survey , but still within the error margin .
while the gcs is clearly affected by stellar streams , the presented value may be distorted by the problematic sloan proper motions and possibly residual distance errors .
while there could of course be some interesting physics involved , a systematic difference of @xmath254 in the average @xmath255 motion between cones towards the galactic north and south poles points to a systematic error in the line - of - sight velocities by @xmath206 .
this is not implausible in light of the adhoc shift of @xmath256 applied in @xcite .
the correction reconciles @xmath250 to reasonable agreement with hipparcos and the geneva - copenhagen survey @xcite albeit at a lower value of about @xmath217 .
comparison of the absolute rotation measure @xmath4 based on heliocentric @xmath3 velocities to the mean azimuthal velocities @xmath22 in a sample delivers the solar azimuthal velocity @xmath1 .
this measurement is correlated with the assumed galactocentric radius @xmath0 . combining this relation with another datum like the proper motion of sgr
@xmath5 one can determine both @xmath0 and @xmath1 .
for dr8 i obtain @xmath193 and @xmath194 . by dissecting the sample via metallicity into slow and fast rotating subgroups
i can independently infer the galactocentric radius from their comparison : a larger @xmath0 reduces @xmath10 and hence nearly proportionally increases @xmath4 .
thus fast rotators experience a larger absolute change in the rotation speed @xmath4 , putting their value of @xmath1 at odds with that from the slow rotators when @xmath0 is wrong .
enforcing consistency provides @xmath200 , and in combination with the simple rotation measure and the proper motion of sgr @xmath5 i get @xmath6 and @xmath7 in excellent agreement with the values from @xcite or @xcite .
the circular speed @xmath257 using the lsr value of @xmath258 from @xcite is then @xmath8 .
the third approach uses just the direction of motion in the galaxy to estimate @xmath0 .
assuming near axisymmetry again the direction of motion in a sample should always point in the direction of the azimuth .
fitting the galactic angle throughout the plane to the angle of motion provides @xmath0 .
this measurement displays a strong dependence on the solar radial motion apart from the dependence on @xmath1 .
fortunately @xmath9 and @xmath0 are not degenerate and i obtain again a relatively large @xmath242 , confirming the previous conventional analysis , but with an even smaller formal error . as
the resulting relationship between @xmath1 and @xmath0 is only weakly inclined against the result from sgr @xmath5 , this approach will be of relevance for larger samples and here just provides some reassurance .
the laid out strategies are extremely simple .
they do not rely on any modelling with possible hidden assumptions , so that possible biases are readily understood .
they rely on the sole assumption of near axisymmetry of the galactic disc .
they are designed for thick disc stars that reside high above the galactic plane and are by their position and high random energy far less affected by perturbations of the galactic potential , like the spiral pattern .
their modelling independence is supported by the fact that by construction none of the estimators depends on the detailed changes of sample composition with position in the galaxy or on the radial dependence of @xmath2 .
still some uncertainties must be pointed out : the methods are vulnerable to large - scale distortions from axisymmetry in the galactic disc . the bar region should be avoided , and there may be a residual signal from spiral arms .
i could not detect significant structures connected to this , so the consequences should be rather small , particularly as i stay out of the region dominated by the bar and as the sample is sufficiently extended not to cover just one side of a spiral arm .
the currently available data add to the possible biases by their uncertain distances , radial velocities and especially astrometry . the segue proper motions display a distinctive systematic pattern on quasar samples .
while i use the quasar catalogues for corrections , the solution remains unsatisfactory : i could not quantify separately the physical reasons , i.e. chromatic aberrations , astrometric `` frame - dragging '' by stars mistaken for galaxies or even focal plane distortions of the telescopes , and also there might be some undetected dependence on the stellar colours .
further some minor uncertainty in parameters is caused by possible biases in line - of - sight velocities .
i made extensive use of the distance corrections developed by @xcite . besides that this project would have been futile without the accuracy achieved by the statistical corrections , the outcome is prone to all the weaknesses of that method . especially streams and wrong assumptions about the velocity ellipsoid can induce systematic distance errors of order @xmath259 .
the contamination in the sample will vary to some extent with galactic coordinates , an effect that i did not model .
this can lead to a bias , because the distribution of weights in the rotation estimators and the distance estimator is different . the accuracy could be far better could i use more stars and had i better control of the systematic errors by another dataset .
part of this will be resolved by use of additional samples like rave .
fortunately the estimators show very different behaviour on the different biases : the proper motion problem affects almost exclusively the first estimator for rotation with a correction to @xmath1 of @xmath260 , while the difference between fast and slow rotators is nearly untouched .
vice versa , the difference estimator reacts strongly to distance changes , while the first estimator does not - a @xmath45 change shifts it by @xmath206 . despite their benign distribution
the systematic uncertainties on @xmath1 and @xmath0 may come close to the formal errors .
if the reader should take only one point from this note then let it be this : with the advent of the large surveys stellar samples are regaining their place as a primary source to obtain not only the local standard of rest , but global galactic parameters .
already a sample of @xmath199 stars from segue provides formal accuracies for the galactocentric radius and the solar azimuthal velocity that are competitive with any other known approach and without any need for modelling .
it is a pleasure to acknowledge useful discussions with j. binney , p. mcmillan and m. asplund on early drafts .
i thank d. weinberg for fruitful discussions and r. dong for kind provision of the quasar samples and valuable discussions on sloan proper motions .
i acknowledge financial and material support from max - planck - gesellschaft .
support for this work was partly provided by nasa through hubble fellowship grant @xmath261 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 . | i present three methods to determine the distance to the galactic centre @xmath0 , the solar azimuthal velocity in the galactic rest frame @xmath1 and hence the local circular speed @xmath2 at @xmath0 . these simple , model - independent strategies reduce the set of assumptions to near axisymmetry of the disc and are designed for kinematically hot stars , which are less affected by spiral arms and other effects .
the first two methods use the position - dependent rotational streaming in the heliocentric radial velocities ( @xmath3 ) .
the resulting rotation estimate @xmath4 from @xmath3 velocities does not depend on @xmath1 .
the first approach compares this with rotation from the galactic azimuthal velocities to constrain @xmath1 at an assumed @xmath0 .
both @xmath1 and @xmath0 can be determined using the proper motion of sgr @xmath5 as a second constraint .
the second strategy makes use of @xmath4 being roughly proportional to @xmath0 .
therefore a wrong @xmath0 can be detected by an unphysical trend of @xmath1 with the intrinsic rotation of different populations .
from these two strategies i estimate @xmath6 and @xmath7 for a stellar sample from segue , or respectively @xmath8 .
the result is consistent with the third estimator , where i use the angle of the mean motion of stars , which should follow the geometry of the galactic disc .
this method also gives the solar radial motion with high accuracy .
the rotation effect on @xmath3 velocities must not be neglected when measuring the solar radial velocity @xmath9 .
it biases @xmath9 in any extended sample that is lop - sided in position angle @xmath10 by of order @xmath11 . combining different methods i find @xmath12 , moderately higher than previous results from the geneva - copenhagen survey .
[ firstpage ] stars : kinematics and dynamics - galaxy : fundamental parameters - kinematics and dynamics - disc - halo - solar neighbourhood |
* 1.0 . * the kazhdan lusztig correspondence @xcite between a vertex - operator algebra ( a conformal field theory ) and a quantum group ,
although functorial , is not necessarily an equivalence ( nor , strictly speaking , does it necessarily involve the representation category of a _ quantum group _ ; this often has to be replaced by a quasitensor category @xcite that is not the representation category of any quantum group ) .
`` deviations '' from the equivalence typically occur in _ rational _ conformal field theories @xcite .
a class of conformal field theory models where the kazhdan lusztig correspondence is an equivalence is given by the @xmath2 logarithmic models .
this remarkable property may be attributed to the fact that the @xmath2 models are not logarithmic _ extensions _ of minimal conformal field theory models , simply because the latter are nonexistent . in the @xmath2 model ,
the representation category is nonsemisimple , and establishing the kazhdan lusztig correspondence does not therefore require a special `` semisimplification '' of the representation category of the quantum group . for recent advances in the study of vertex - operator algebras with nonsemisimple representation categories ,
see @xcite and the references therein and in @xcite . in the @xmath2 logarithmic models ,
the vertex - operator algebra is the triplet @xmath0-algebra @xmath1 `` based on the @xmath9 operator '' @xcite , and the kazhdan lusztig - dual quantum group is @xmath4 , the restricted quantum @xmath10 at @xmath5 @xcite .
the @xmath1 algebra and the @xmath4 quantum group have the same number ( @xmath11 ) of irreducible representations and identical fusion rings ( the grothendieck ring for the quantum group ) .
the results in @xcite suggest that the quasitensor categories of @xmath1- and @xmath4-representations are in fact equivalent . in this paper
, we prove the equivalence in the most explicit case of @xmath7 .
the proof extends to the general @xmath12 , but requires more technical details , which will be given elsewhere .
the general strategy is to construct a bimodule ( a right @xmath4-module and a left @xmath1-module ) that decomposes into projective modules from each side and therefore establishes the equivalence of the two representation categories . for @xmath7 ,
this construction is quite simple because the relevant vertex operators are `` almost local '' with respect to each other . by the equivalence , which we assume to hold for any @xmath12
, the study of a category of @xmath1-representations can be replaced by the study of the category of finite - dimensional @xmath4-representations .
these are not difficult to classify using the standard means @xcite ; we show that apart from the projective modules , all indecomposable representations are enumerated in terms of indecomposable representations of the kronecker quiver and can easily be constructed explicitly ; there are four series labeled by @xmath13 and two series labeled by @xmath14 . in accordance with the equivalence
, this classification also applies to @xmath1-representations ; finding their classification directly would be a much more difficult problem .
the equivalence of categories leads to other identifications between `` conformal field theory data '' and `` quantum - group data , '' the most prominent one being the identification of the respective _
centers_. the center @xmath15 of a conformal field theory model
endomorphisms of the identity functor in the representation category of the corresponding vertex - operator algebra @xmath16contains `` highly nonlocal '' operators ( the ishibashi states ) . a basis in the center
is naturally constructed in terms of conformal blocks on a torus . in the semisimple case ,
@xmath15 coincides with the space of conformal field theory characters , but in logarithmic models , it is not exhausted by the characters ( cf .
this has been studied recently within different approaches @xcite .
the general approach to finding the missing `` extended characters '' via the construction of pseudo - traces for a class of vertex - operator algebras with nonsemisimple representation category was proposed in @xcite .
the extended characters , along with the usual ones , are solutions of certain differential equations @xcite .
when the kazhdan lusztig correspondence establishes the equivalence between representation categories of a vertex - operator algebra and a quantum group , @xmath15 is canonically identified with the _ center @xmath17 of the quantum group_. moreover , in the @xmath2 models at least , this identification between @xmath15 and @xmath17 extends to modular - group representations : the @xmath18-representation on @xmath15 induced by the @xmath19 and @xmath20 transformations is equivalent @xcite to the @xmath18-action on the quantum group center @xmath17 , defined in quite general terms of ribbon braided tensor categories @xcite .
the actual identification of @xmath15 with @xmath17 yields a remarkable relation @xmath21 where @xmath22 is the ribbon element ( belonging to the quantum group center ) and @xmath23 is the zero - mode of the energy - momentum tensor .
furthermore , the @xmath8-matrix associated with the kazhdan lusztig - dual quantum group @xmath24 can be identified with the braiding matrix in conformal field theory , its action on vertex operators following from the construction of @xmath24 in terms of screening and contour - removal operators .
itself is _ not _ quasitriangular ( does not have a universal @xmath8-matrix ) , at least in the known examples of logarithmic @xmath2 @xcite and @xmath25 models . but the universal @xmath8-matrix exists for a somewhat larger quantum group , generated by the generators of @xmath24 _ and _ the square root @xmath26 of the cartan generator @xmath27 in @xmath24 , i.e. , @xmath28 . in conformal field theory ,
the counterpart of the ambiguity in defining @xmath29 is the ambiguity in choosing the direction of contours along which operators are permuted .
this leads to _ two _ braiding matrices @xmath30 and @xmath31 , well known in ( rational ) conformal field theories @xcite . ]
[ sec : ures ] let @xmath4 be the restricted quantum @xmath10 at @xmath5 , generated by @xmath32 , @xmath33 , and @xmath27 with the relations @xmath34 and the hopf - algebra structure given by @xmath35={\mbox{\footnotesize$\displaystyle\frac{k - k^{-1}}{{\mathfrak{q}}-{\mathfrak{q}}^{-1}}$}},\\ \delta(e)={\boldsymbol{1}}\otimes e+e\otimes k,\quad \delta(f)=k^{-1}\otimes f+f\otimes{\boldsymbol{1}},\quad \delta(k)=k\otimes k,\\ \epsilon(e)=\epsilon(f)=0,\quad\epsilon(k)=1,\\ s(e)=-ek^{-1},\quad s(f)=-kf,\quad s(k)=k^{-1}. \end{gathered}\ ] ] this quantum group has @xmath11 inequivalent irreducible representations , labeled as @xmath36 and @xmath37 , @xmath38 .
the module @xmath39 ( with @xmath40 ) has the highest weight @xmath41 .
let @xmath3 be the category of finite - dimensional @xmath42-graded @xmath4-modules at @xmath5 . in the standard way
, it is endowed with the structure of a ribbon braided tensor category ( the ribbon element required for this can be found in accordance with the standard recipe from the universal @xmath8-matrix found in @xcite for a quantum group that contains @xmath4 as a subalgebra ; it turns out that this results in the ribbon element not for the larger quantum group but just for @xmath4 ) . * . the @xmath1 algebra . * for a scalar field @xmath43 with the ope @xmath44 the virasoro algebra with central charge @xmath45 is generated by the energy - momentum tensor ( with normal ordering understood ) @xmath46 where @xmath47 , @xmath48 , and @xmath49 .
there are the screening operators @xmath50 and the @xmath0-algebra @xmath1 @xcite is generated by the three currents @xmath51,\quad w^+(z)\!=\![s_+,w^0(z)]\ ] ] ( which are virasoro primaries of conformal dimension @xmath52 ) .
there are @xmath11 irreducible modules of @xmath1 , labeled as @xmath53 ( `` singlet , '' associated with the @xmath54 operator ) and @xmath55 ( `` doublet , '' associated with the @xmath56 operator ) , @xmath38 , with the characters given by @xcite @xmath57 \chi_{{\mathfrak{x}}^{-}_{s}}^{\phantom{y}}(q)&= { \mbox{\small$\displaystyle\frac{1}{\eta(q)}$ } } \bigl({\mbox{\footnotesize$\displaystyle\frac{s}{p}$}}\,\theta_{s , p}(q ) - 2\,\theta'_{s , p}(q)\bigr ) , \end{aligned}\qquad 1{\,{\leqslant}\,}s{\,{\leqslant}\,}p,\ ] ] where @xmath58 and we set @xmath59 and @xmath60 . the fusion for @xmath61 was directly calculated in @xcite ( also see @xcite for @xmath1 ) .
a suitable choice of the representation category for @xmath1 is the category @xmath62 .
let @xmath63 denote the category @xmath62 of representations of the @xmath1 algebra , i.e. , the category of modules in which adjoint eigenspaces of @xmath23 are finite - dimensional and the spectrum of @xmath23 is bounded from one side .
this category is a braided ribbon quasitensor category ; its quasitensor structure follows from operator product expansions in conformal field theory , the braiding is derived from the monodromy of correlation functions , and the ribbon structure is given by the operator @xmath64 .
[ thm : equv ] the categories @xmath65 and @xmath66 are equivalent as ribbon braided quasitensor categories . as noted above , the proof of this fact is greatly simplified by the existence , for @xmath7 , of the symplectic - fermion fields that are `` almost local '' with respect to each other . in the general case ,
the proof requires working out some technical details , but there is little doubt that the following statement holds . for each integer @xmath6 , the categories @xmath67 and @xmath68 are equivalent as ribbon braided quasitensor categories .
assuming the equivalence for all @xmath12 , we can therefore describe the category of @xmath1-representations by describing the category of @xmath4-representations , which is a sufficiently simple problem ; to a large degree , it amounts to describing indecomposable representations of the kronecker quiver .
we now give a list of indecomposable modules and then state the result in [ thm : main ] . * .
indecomposable modules .
* the basic fact is that those spaces @xmath69 of extensions between the irreducible representations that are nonzero are two - dimensional ( see [ lem : ext - irrep ] ) ; in terms of the bases @xmath70 and @xmath71 , @xmath72 , chosen in the respective spaces @xmath73@xmath74 and @xmath75 , we can construct four families of indecomposable modules as follows . in the diagrams below
, @xmath76 denotes the extension by @xmath77 , with @xmath78 being an irreducible subquotient and @xmath79 an irreducible submodule .
@xmath80 : : for @xmath81 , @xmath82 , and integer @xmath83 , the module @xmath84 is defined as + where @xmath85 is the number of @xmath86 modules . a basis and the @xmath4-action on @xmath84
are explicitly described in [ app : modw - base ] in the example of @xmath87 .
@xmath88 : : for @xmath81 , @xmath89 , and integer @xmath83 , the module @xmath90 is defined as + + where @xmath85 is the number of @xmath91 modules . a basis and the @xmath4-action on @xmath90 are explicitly described in [ app : modm - base ] in the example of @xmath92 . @xmath93 : : for @xmath94 , @xmath82 , integer @xmath95 , and @xmath96 , the module @xmath97 is defined as + where @xmath85 is the total number of the @xmath86 modules and @xmath98 . a basis and the @xmath4-action on @xmath97
are explicitly described in [ app : modn - base ] in the example of @xmath99 . @xmath100
: : for @xmath81 and @xmath89 , the projective module @xmath101 is defined in [ app : proj - mod - base ] ; it has the following structure of subquotients ( see [ sec : proj - mod ] below ) : @xmath102_{x^a_1 } \ar@/_/[dr]^{x^a_2}&\\ & \stackrel{{\mathscr{x}}^{-a}_{p - s}}{\bullet}\ar@/^/[dr]_{x^a_2 } & & \stackrel{{\mathscr{x}}^{-a}_{p - s}}{\bullet}\ar@/_/[dl]^{x^a_1}\\ & & \stackrel{{\mathscr{x}}^a_{s}}{\bullet } & } \ ] ] the following theorem shows that the modules listed above exhaust the finite - dimensional indecomposable @xmath4-modules .
[ thm : main ] 1 . the category @xmath3 has the decomposition @xmath103 where each @xmath104 is a full subcategory .
2 . each of the subcategories @xmath105 and @xmath106 is semisimple and contains precisely one irreducible module , @xmath107 and @xmath108 respectively .
each subcategory @xmath104 , @xmath81 , contains precisely two irreducible modules @xmath109 and @xmath110 and the following set of indecomposable modules : * the projective modules @xmath111 and @xmath112 ; * three series of indecomposable modules : * * the modules @xmath113 and @xmath114 for all integer @xmath83 , * * the modules @xmath115 and @xmath116 for all integer @xmath83 , * * the modules @xmath117 and @xmath118 for all @xmath96 and integer @xmath95 .
+ this exhausts the list of indecomposable modules in @xmath104 .
next , for each subcategory @xmath104 with @xmath119 , the vector space @xmath120 is an associative algebra with respect to the yoneda product . in the following theorem
, we describe the algebraic structure of @xmath121 .
[ ext - alg ] the algebra @xmath121 is generated by @xmath122 , @xmath123 , with the defining relations @xmath124 we note that the derived category of representations of @xmath121 is equivalent to the derived category of @xmath104 ( and hence to the derived category of a full subcategory of @xmath1-representations ) .
we set @xmath125 with @xmath7 in sec .
[ sec : equiv ] and @xmath126 elsewhere . in
what follows , @xmath127 denotes the set @xmath128 .
this paper is organized as follows .
the proof of the classification result in [ thm : main ] is given in sec .
[ sec : proof ] . in sec .
[ sec : ext - proof ] , we prove [ ext - alg ] .
the equivalence of the @xmath0-algebra and quantum - group representation categories is proved in sec .
[ sec : equiv ] for @xmath7 .
explicit constructions of some modules are given in appendix [ app : mod ] .
the necessary facts about quivers are recalled in appendix [ app : quivers ] .
we prove [ thm : main ] by straightforward investigation of the structure of the category @xmath3 .
we fix an integer @xmath6 .
the strategy of the proof is as follows . in [ sec : cat - decomp ] , using the casimir operator , we decompose the category @xmath3 into a direct sum of full subcategories @xmath104 . in
[ lem : ext - irrep ] , we then calculate @xmath69 between irreducible representations in each @xmath104 . in [ sec : proj - mod ] , we construct projective modules in each full subcategory @xmath104 .
we next note that all projective modules in @xmath3 are injective modules .
this information suffices to ensure that indecomposable modules with semisimple length @xmath129 are projective modules and that there are no modules with semisimple length @xmath130 or more .
therefore , to complete the proof of [ thm : main ] , it remains to classify indecomposable modules with semisimple length @xmath131 .
we do this in [ sec : semisimple-2 ] using a correspondence between modules with semisimple length @xmath131 and indecomposable representations of the kronecker quiver . in setting the notation and recalling the basic facts about @xmath132 needed below , we largely follow @xcite . the hopf algebra @xmath4 is generated by @xmath32 , @xmath33 , and @xmath27 with the relations given in [ sec : ures ] . in all @xmath4-modules , we choose a basis such that the generator @xmath27 acts diagonally ( the action of @xmath27 in any module in @xmath3 is diagonalizable because @xmath134 ) .
let @xmath17 denote the center of @xmath4 ( it is @xmath135-dimensional @xcite ) .
the irreducible @xmath4-modules are labeled by their highest weights @xmath136 , where @xmath137 .
we also parameterize the same highest weights as @xmath138 , where @xmath82 and @xmath38
. then for @xmath38 , the irreducible module with the highest weight @xmath139 is denoted by @xmath39 .
the dimension-@xmath140 module @xmath39 is spanned by elements @xmath141 , @xmath142 , where @xmath143 is the highest - weight vector and the algebra action is given by @xmath144[s - n]{\mathsf{a}}_{n - 1}^{\pm},\\ f { \mathsf{a}}_n^{\pm } & = { \mathsf{a}}_{n + 1}^{\pm},\end{aligned}\ ] ] where we set @xmath145 and use the standard notation @xmath146 = { \mbox{\footnotesize$\displaystyle\frac{{\mathfrak{q}}^n-{\mathfrak{q}}^{-n}}{{\mathfrak{q}}-{\mathfrak{q}}^{-1}}$}}.\ ] ] @xmath147 is the trivial module .
let @xmath148 and @xmath149 be left @xmath4-modules .
we say that a short exact sequence of @xmath4-modules @xmath150 , or equivalently , @xmath151 is an _ extension _ of @xmath149 by @xmath148 , and we let @xmath152 denote the set of equivalence classes ( see , e.g. , @xcite ) of extensions of @xmath149 by @xmath148 .
let @xmath153 be a @xmath4-module .
we define a _
semisimple filtration _ of @xmath153 as a tower of submodules @xmath154 such that each quotient @xmath155 is semisimple . the number @xmath156 is called the _ length _ of the filtration . in the set of semisimple filtrations of @xmath153 , there exists a filtration with the minimum length @xmath157 .
we call @xmath157 the _ semisimple length _ of @xmath153 . evidently , an indecomposable module has the semisimple length not less than @xmath131 .
any semisimple module has semisimple length @xmath158 .
the recall the casimir element @xmath159 which satisfies the minimal polynomial relation ( e.g. , see @xcite ) @xmath160 where @xmath161 and hence generates a @xmath11-dimensional subalgebra in @xmath17 .
this relation yields a decomposition of the category @xmath3 into the direct sum of full subcategories @xmath162 such that @xmath163 acts nilpotently on objects in @xmath162 . because @xmath164 for @xmath165 , there are @xmath166 full subcategories @xmath104 for @xmath167 .
this proves part ( 1 ) of theorem [ thm : main ] .
each subcategory @xmath104 for @xmath168 contains precisely two irreducible modules @xmath109 and @xmath110 , because the casimir element acts by multiplication with @xmath169 on precisely these two irreducible modules .
the irreducible modules @xmath107 and @xmath108 corresponding to the respective eigenvalues @xmath170 and @xmath171 belong to the respective categories @xmath106 and @xmath105 . to describe all isomorphism classes of indecomposable representations , we first describe extensions between irreducible representations .
inspection shows that @xmath172@xmath173 and that all other @xmath69 between irreducible representations in @xmath104 vanish .
we thus have the following lemma .
[ lem : ext - irrep ] for @xmath174 and @xmath175 , there are vector - space isomorphisms @xmath176 we also see that part ( 2 ) and the statement in [ thm : main ] about irreducible modules of @xmath104 hold . the @xmath177 space in this lemma is the fundamental representation of the @xmath10 lie algebra whereby @xmath4 is extended to lusztig s quantum group `` with divided powers . ''
this is in fact another manifestation of the equivalence of the @xmath4 and @xmath1 representation categories : the occurrence of the @xmath10 lie algebra is well known in the @xmath2 conformal field theory models , where it is an outer automorphism of the _ triplet _
@xmath0-algebra @xmath1 and hence acts on some of the @xmath1-representations , and on extensions of two irreducible representations in particular . for each @xmath38 and @xmath82 , let @xmath178 denote the verma module with the highest weight @xmath179 , and let @xmath180 be the contragredient verma module with the highest weight @xmath181 ( it is contragredient to the verma module @xmath182 ) . and the contragredient verma @xmath180 modules are particular cases of the respective @xmath99 modules with @xmath183 and @xmath184 . ]
we note that @xmath185 . for each @xmath1860@xmath187
, we choose the bases @xmath70 and @xmath71 , @xmath72 , in the respective spaces @xmath73@xmath188 and @xmath189 such that @xmath190 corresponds to the verma module @xmath191 , @xmath192 corresponds to the contragredient verma module @xmath193 , @xmath194 corresponds to @xmath195 , and @xmath196 corresponds to @xmath197 .
[ rem : gluing ] in terms of the action of the @xmath4 generators @xmath32 and @xmath33 , the extension @xmath198 , giving the verma module , is such that @xmath33 sends the _
lowest_-weight vector in the subquotient @xmath109 to the _ highest_-weight vector in the submodule @xmath110 ; similarly , the extension determined by @xmath192 , giving the contragredient verma module , is such that @xmath32 sends the highest - weight vector in @xmath109 to the lowest - weight vector in @xmath110 .
the extensions determined by @xmath199 , @xmath72 , are described similarly .
we next describe projective modules in each subcategory @xmath104 for @xmath200 .
to construct these modules , we use that any indecomposable projective module can be obtained as a projective cover of an irreducible module .
[ prop : proj ] 1 .
projective modules of the subcategories @xmath105 and @xmath201 are the respective irreducible modules @xmath107 and @xmath108 .
2 . for each @xmath140
, @xmath202 , projective modules of the subcategory @xmath104 are @xmath111 and @xmath112 see .
part ( 1 ) is clear because the subcategories @xmath105 and @xmath201 are semisimple . to prove part ( 2 ) ,
we construct a projective cover of each irreducible module .
first , we construct a nontrivial extension of @xmath109 by the maximal number of irreducible modules @xmath110 , that is , by @xmath203 , which is @xmath204 the indecomposable module @xmath205 is described explicitly in [ app : modm - base ] .
next , to find the projective cover of @xmath109 , we extend each summand in the submodule @xmath206 by @xmath207 .
the compatibility with the @xmath4-algebra relations ( with @xmath208 and @xmath209 in particular ) leads to an extension corresponding to the module @xmath111 , which is the projective cover of @xmath109 .
this module is described explicitly in [ app : proj - mod - base ] .
a similar procedure gives the projective cover @xmath112 of @xmath110 .
the modules @xmath210 and @xmath112 are described in [ app : modm - base ] and [ app : proj - mod - base ] respectively .
therefore , in each subcategory @xmath104 with @xmath211 , we have two inequivalent projective modules @xmath111 and @xmath112 , which cover the respective irreducible modules . to show that these projective covers exhaust all indecomposable projective modules in the category @xmath3 , we recall that the left regular representation @xmath212 of @xmath4 is a direct sum over all projective modules with the multiplicity of each given by the dimension of its simple quotient . the dimension of the simple quotient of @xmath213 is @xmath140 ( see [ app : proj - mod - base ] ) and @xmath214 , and we therefore have @xmath215 but counting the dimension of the right - hand side gives @xmath216 , and therefore the above inclusion is an equality , and part ( 2 ) follows .
it also follows from [ prop : proj ] that the subquotient structure of the projective modules is the one shown in .
next , the `` dual '' procedure of constructing the injective modules in each subcategory @xmath104 shows that they in fact coincide with the projective modules ( injective modules are contragredient to projective modules , but the module obtained by inverting all arrows in is evidently isomorphic to the original one ) .
we therefore have the following proposition .
[ prop : proj - inj ] the projective modules in @xmath3 are injective . 1 .
there are no modules in @xmath3 with semisimple length greater than @xmath129 .
the only indecomposable modules with semisimple length @xmath129 are the projective modules @xmath213 with @xmath217 . for @xmath119 ,
let @xmath153 be a module such that @xmath218 in contains the simple summand @xmath219 .
this means that there is a nonvanishing mapping @xmath220 that covers @xmath219 .
there are two possibilities for the mapping @xmath220 : ( i ) it is an embedding or ( ii ) it has a nonvanishing kernel . in case ( i ) , we recall that the projective module is at the same time injective and is therefore a direct summand in any module into which it is embedded . in case
( ii ) , the kernel necessarily contains the submodule @xmath219 of @xmath213 . but
this means that the subquotient @xmath219 of @xmath153 actually belongs to a direct summand in @xmath153 with the semisimple length less than or equal to @xmath131 . to complete the proof of part ( 3 ) in [ thm : main ] , it remains to classify finite - dimensional modules with the semisimple length @xmath221 . for @xmath168 ,
let @xmath222 be the full subcategory of @xmath104 consisting of @xmath4-modules with semisimple length @xmath223 .
obviously , any module in @xmath222 can be obtained either by * the extension of a direct sum of the irreducible modules @xmath109 by a direct sum of the irreducible modules @xmath110 via a direct sum of @xmath224 or by * the extension of a direct sum of the irreducible modules @xmath110 by a direct sum of the irreducible modules @xmath109 via a direct sum of @xmath225 .
for @xmath226 , let @xmath227 denote the module defined by the extension @xmath228 for @xmath229 , we define the modules @xmath230 similarly .
we also set @xmath231 and @xmath232 . , and @xmath233 in * 1 .
* are particular cases of the @xmath234 modules . ]
we define two full subcategories @xmath235 and @xmath236 of the category @xmath222 as follows .
an object of @xmath235 is either a semisimple module or a module @xmath153 such that @xmath237 for some @xmath238 , where @xmath239 is the maximal semisimple submodule ( the socle ) ; in other words , an object of @xmath235 is the @xmath234 module for some @xmath240 .
objects of @xmath236 are defined similarly with @xmath241 .
we note that @xmath242 .
we now reduce the classification of modules with semisimple length @xmath131 to the classification of indecomposable representations of the kronecker quiver @xmath243 .
the reader is referred to @xcite and appendix [ app : quivers ] for the necessary facts about quivers .
[ lemma : mod - kronecker ] each of the categories @xmath235 and @xmath236 is equivalent to the category @xmath244 of representations of the kronecker quiver @xmath243 .
the lemma is almost obvious if we note that two maps @xmath245 and @xmath246 in @xmath247_{\varepsilon } \ar@/^/[r]^{\bar\varepsilon } & * { \;{\mathscr{m}}^+_{s}(2 ) } } \ ] ] make up a quiver in the category of @xmath4-representations ; taking the functors of @xmath248 to each of the two categories then establishes the equivalence . in somewhat more formal terms
, the equivalence , e.g. , between the categories @xmath235 and @xmath244 is given by the functor @xmath249 that acts on objects as @xmath250 where @xmath251 and , for two linearly independent homomorphisms @xmath252 , the two linear maps @xmath253 are defined as @xmath254 for each @xmath255 , and with the natural action on morphisms . the existence of a functor @xmath256 such that both @xmath257 and @xmath258 are the identity functors is evident from the definitions of the categories @xmath235 and @xmath244
. propositions [ lemma : mod - kronecker ] and [ prop : repr - kron ] immediately imply the desired classification of finite - dimensional @xmath4-modules with semisimple length @xmath259 , thus completing the proof of [ thm : main ] .
the strategy of the proof is as follows . in [ sec : resolutions ] , we first find the projective and injective resolutions of irreducible representations . in
[ lem : extn - irrep ] , we then calculate @xmath260 between irreducible representations in each @xmath104 and in [ sec : ext - alg ] , study the algebra @xmath261 .
inspection shows that the mappings defined in the following lemma give rise to a resolution .
[ lem : proj - res - irrep ] for each @xmath202 , the module @xmath86 has the projective resolution @xmath262 where for even @xmath85 , the @xmath85th term is given by @xmath263 with the boundary morphism given by the throughout mapping in @xmath264 and for odd @xmath85 , the @xmath85th term and the boundary morphism @xmath265 are given by changing @xmath266 to @xmath267 and @xmath140 to @xmath268 .
the statement dual to [ lem : proj - res - irrep ] is as follows . for each @xmath140 , @xmath202 ,
the module @xmath86 has the injective resolution @xmath269 where for even @xmath85 , the @xmath85th term is given by @xmath270 with the coboundary morphism given by the throughout mapping in @xmath271 and for odd @xmath85 , the @xmath85th term and the coboundary morphism @xmath272 are given by changing @xmath266 to @xmath267 and @xmath140 to @xmath268 .
we now calculate @xmath260 between irreducible modules .
[ lem : extn - irrep ] for @xmath273 and @xmath82 , there are the vector - space isomorphisms @xmath274 applying the contravariant functor @xmath275 to the projective resolution of @xmath86 gives the cochain complex @xmath276 where for even @xmath85 , the @xmath277th term is given by @xmath278 and for odd @xmath85 , the @xmath277th term is given by @xmath279 it follows that coboundary morphisms @xmath280 , @xmath281 .
we then have @xmath282 to calculate @xmath283 , we apply the functor @xmath284 to .
we next note that the lusztig quantum group acts on the projective resolution of @xmath285 , and we have the following lemma .
[ lemma : ext - sl2 ] the vector spaces @xmath286 and @xmath283 are irreducible @xmath10-modules . next , for each subcategory @xmath104 with @xmath119 , we consider the total @xmath288-group @xmath289 by [ lem : extn - irrep ] , we have the vector - space isomorphism @xmath290 we write @xmath261 for @xmath121 because its algebraic structure is actually independent of @xmath140 .
the graded vector space @xmath261 is equipped with an associative algebra structure given by the yoneda product . in
what follows , abusing the notation , we just write a representative of an element in @xmath260 instead of the respective equivalence class .
we recall that a basis @xmath291 , @xmath72 , in @xmath292 can be chosen such that @xmath293 [ lemma : ext - alg - gen ] the algebra @xmath261 is generated by @xmath122 , @xmath72 .
we first consider @xmath294 for even @xmath85 . by [ lemma : ext - sl2 ]
, @xmath294 is an irreducible @xmath10-module .
the highest - weight vector @xmath295 of this module is given by @xmath296 this cohomology class is nonzero because its restriction to the subalgebra @xmath297 $ ] of @xmath4 is nonzero , as can be directly verified .
because of the @xmath10 action , it therefore follows that @xmath294 is generated by the @xmath298 . for @xmath299 and odd @xmath85 ,
the argument is similar .
taking the yoneda product of the above short exact sequences , we obtain the following exact sequences in the space @xmath300 : @xmath301 and the exact sequences @xmath302 in @xmath303 . [ lemma : ext - alg - rel ] @xmath304 .
these relations follow from the existence of the projective modules . explicitly , the argument for @xmath305 can be expounded as follows .
we consider the short exact sequence @xmath306 which gives rise to the long exact @xmath288-sequence @xmath307 where @xmath308 and @xmath309 are the connecting homomorphisms , given by the yoneda product with the given short exact sequence up to a sign .
we then consider the short exact sequence @xmath310 given by @xmath311 the extension @xmath312 of @xmath205 defines the extension of the submodule @xmath313 as the exact sequence @xmath314 given by @xmath315 as a consequence of the identity @xmath316 , we have @xmath317 in @xmath318 or , equivalently , @xmath319 that @xmath320 is established similarly with the help of the short exact sequence @xmath321 . from [ lemma : ext - alg - gen ] and [ lemma : ext - alg - rel
] , it follows that the algebra @xmath261 is generated by @xmath298 , @xmath72 , and has the defining relations @xmath322 and @xmath323 .
we also set @xmath324 , @xmath123 , because the yoneda product is only defined for two exact sequences if the end of one sequence coincides with the beginning the other .
this completes the proof of [ ext - alg ] .
we now prove the equivalence of the @xmath1 and @xmath4 representation categories for @xmath7 .
the proof considerably simplifies compared with the case of general @xmath12 because the @xmath1-generators are explicitly constructed in terms of symplectic fermions @xcite . extending the proof to general @xmath12 requires an appropriate generalization of the symplectic fermions , achieved by introducing a multiplet of first - order conformal fields carrying an @xmath4-action .
the implication of the equivalence is that the pictures in * 1 . *
equally well describe @xmath1-modules . for @xmath7 , the @xmath0-algebra @xmath61 @xcite ( see * 1 .
* ) is conveniently described in terms of the so - called symplectic fermions @xcite with the ope ( in the holomorphic sector , which we consider in what follows ) @xmath325 the associated virasoro algebra with central charge @xmath326 is generated by the energy - momentum tensor ( with normal ordering understood ) @xmath327 and the @xmath61 algebra is generated by @xmath328 and the two currents given by @xmath329 the @xmath330 model contains two sectors of fields : generated by @xmath331 and @xmath332 from the operator @xmath333 ( periodic boundary conditions ) and from the so - called twist operator @xmath334 ( antiperiodic boundary conditions ) ; the subscript @xmath335 is for two _ quantum - group _ components . in the sector of periodic boundary conditions , we have @xmath336 in the sector of antiperiodic boundary conditions , we have @xmath337 the commutation relation between the modes in the sector of periodic boundary conditions are ( with @xmath338 $ ] denoting supercommutator , in this case anticommutator ) @xmath339=\delta_{m+n,0 } , \quad [ \theta_n,\theta_m]=0 , \quad [ \bar\theta_n,\bar\theta_m]=0,\ ] ] and the only nonzero commutation relations involving @xmath340 and @xmath341 are , it follows that the anticommutator @xmath342 $ ] is given by the increment of the ope @xmath343 as @xmath344 is continued via @xmath345 , that is , @xmath346 .
this is independent of @xmath347 , and therefore @xmath348=1 $ ] , @xmath349=0 $ ] . ]
@xmath350=1 , \quad [ \bar\theta_0,\theta^*_0]=1.\ ] ] another description of the model ( as in * 1 . * ) is given in terms of a free bosonic field that is a solution of the `` riccati equation '' @xmath351 this equation has two solutions @xmath352 and @xmath353 , which are scalar fields with the opes @xmath354 and are related by @xmath355 where @xmath356 the symplectic fermions can then be described in either of the two mutually nonlocal pictures , in terms of either @xmath357 or @xmath358 . in the @xmath357-picture , we have @xmath359 and in the @xmath358-picture , @xmath360 a special role is played by the zero modes @xmath361 in the @xmath357-picture , @xmath362 is a screening operator and @xmath363 is a ( nonlocal ) contour - removal operator ; in the @xmath358-picture , @xmath362 is a contour - removal operator and @xmath363 is a screening operator .
in the @xmath358-picture , we recover the construction in with @xmath364 and @xmath365 .
each of the bosonized pictures allows a component of the twist operator to be expressed locally .
starting with the @xmath357-picture for definiteness , we set @xmath366 the second component is then given by @xmath367 where @xmath368 denotes a contour running from @xmath369 to @xmath344 . in the @xmath358-picture , we then have @xmath370 the two components are vectors in the two - dimensional representation of the quantum group , see below . as usual ,
the screening and contour removal operators generate a quantum group . in the @xmath2 logarithmic conformal field
theory models , it is just @xmath4 at @xmath5
. we have @xmath371 for the @xmath330 model , with the quantum group generators explicitly given by @xmath372 with @xmath373 , where @xmath374 the zero modes @xmath340 and @xmath341 can then be considered functions on the quantum group .
we also have @xmath375 we recall from @xcite that the algebra @xmath61 has four irreducible representations , @xmath376 and @xmath377 .
the @xmath377 are steinberg modules ( irreducible , verma , and projective simultaneously ) .
we also need the projective modules @xmath378 with the structure of subquotients described as @xmath379\ar[dr]&\\ & \stackrel{{\mathfrak{x}}^{\mp}_{1}}{\bullet}\ar[dr ] & & \stackrel{{\mathfrak{x}}^{\mp}_{1}}{\bullet}\ar[dl]\\ & & \stackrel{{\mathfrak{x}}^{\pm}_{1}}{\bullet } & } \ ] ] the field corresponding to the highest - weight vector of @xmath380 is @xmath381 ( the `` logarithmic partner '' of the operator @xmath333 ) . to construct the @xmath61-representations in terms of the symplectic fermions ,
we introduce the vacuum vector @xmath382 corresponding to the operator @xmath333 ( on which the symplectic fermions act with integer modes , see ) ; it is equivalently defined by the highest - weight conditions @xmath383 where we simultaneously indicate the action of the @xmath27 generator .
let @xmath384 be the module generated from @xmath382 by the integer modes @xmath385 and @xmath386 .
the integer - moded module over the symplectic fermions is then @xmath387 as a @xmath61-module , @xmath388 decomposes into a direct sum of two projective modules , @xmath380 , generated from @xmath389 , and @xmath390 , generated from @xmath391 . as a @xmath4-module
, @xmath388 decomposes into an infinite sum of projective modules . for any state @xmath392 , where @xmath393 is a homogeneous even - order polynomial in @xmath394 and @xmath395 , the projective @xmath4-module @xmath396 ( see ) is spanned by @xmath397@xmath398 .
similarly , for any state @xmath392 , where @xmath393 is a homogeneous odd - order polynomial in @xmath394 and @xmath395 , the projective @xmath4-module @xmath399 is spanned by @xmath400@xmath401@xmath402 .
next , let @xmath403 be the vector corresponding to the operator @xmath404 , on which the symplectic fermions act with half - integer modes , see .
it is equivalently defined by the highest - weight conditions @xmath405 additionally , it satisfies the relations @xmath406 the vectors @xmath403 are two basis vectors in the two - dimensional representation of @xmath4 , with the screening operators @xmath362 and @xmath363 acting as in , that is , @xmath407 we let @xmath408 denote the module over the symplectic fermions generated from @xmath403 by half - integer modes . as a @xmath61-module ,
each of @xmath408 decomposes into two steinberg @xmath61-modules @xmath409 and @xmath410 .
as a @xmath4-module , the direct sum of @xmath61-modules @xmath411 decomposes into an infinite sum of steinberg modules .
the steinberg module @xmath412 is spanned by @xmath413@xmath414 for any homogeneous even - order polynomial @xmath393 in @xmath394 and @xmath395 ; similarly , the steinberg module @xmath415 is spanned by @xmath416 for any homogeneous odd - order polynomial @xmath393 in @xmath417 and @xmath418 .
let @xmath419 denote the module @xmath420 as a @xmath61-module , it decomposes into a sum of _ all _ the projective modules @xmath421 , @xmath422 , @xmath409 , and @xmath410 with the multiplicities given by the dimensions of the corresponding quantum - group modules , @xmath423 the fact that @xmath419 is simultaneously a @xmath61-module and a @xmath4-module and decomposes with respect to the projective modules in each case allows establishing the equivalence of the representation categories in the standard way .
we first construct a functor @xmath424 ; it is actually more convenient to first define the contravariant functor @xmath425 which gives a @xmath61-module for any @xmath4-module @xmath426 , and then compose it with the functor @xmath427 that replaces each @xmath61-module with its contragredient one .
thus , @xmath428 is a covariant functor .
we next construct the functor @xmath429 similarly . for any @xmath61-module @xmath430
, we set @xmath431 ( which is a @xmath4-module ) , where @xmath432 is the functor of taking the contragredient module . the functors @xmath433 and @xmath434 send irreducible representations into irreducible , and it therefore immediately follows that @xmath435 the natural braiding in conformal field theory , determined by contour manipulations in the complex plane , coincides with the braiding determined by the quantum - group @xmath8-matrix .
this is illustrated below with several characteristic examples .
we first note that , as already remarked in the introduction , the @xmath4 quantum group at @xmath5 is not quasitriangular , but is a subalgebra in a larger quantum group generated by @xmath32 , @xmath33 , and @xmath26 such that @xmath28 , which _ is _
quasitriangular @xcite . for @xmath7 , the universal @xmath8-matrix for this larger quantum group becomes @xmath436 where we choose @xmath437 to correspond to the braiding matrix @xmath30 .
in view of , the action of @xmath8 on vertex operators is defined , which allows identifying @xmath8 with the @xmath30 matrix .
viewed in this capacity , the @xmath8-matrix is rewritten as @xmath438 this implies the braided relations @xmath439 for any two fields @xmath440 and @xmath441 , where @xmath344 and @xmath347 are real , with @xmath442 ( the relations are then analytically continued from the real axis ) .
in particular , we have @xmath443 these relations can be illustrated by the procedure of permuting operators in the complex plane . in the first relation in , in the @xmath357-picture
( see ) , the operator @xmath331 is nonlocal , @xmath444 , which we represent with the integration contour from infinity to its position , @xmath445 and taking it half the circumference ( clockwise ) around @xmath347 changes the integration contour , resulting in the additional term @xmath446 .
no such term occurs in the second relation in , represented as @xmath447 similarly , for the field @xmath381 , which is the logarithmic partner of @xmath333 , it follows from that @xmath448 and for the twist operator , we have @xmath449 the ribbon structure of @xmath4 is given by the central element @xmath450 taking into account , we obtain the element @xmath451 which acts on the space of states as @xmath452
by the kazhdan lusztig equivalence , the structure of indecomposable modules pictured in * 1 .
* simultaneously describes the structure of indecomposable @xmath1-modules .
the projective resolution in is also applicable to the @xmath1 algebra , and may therefore be used in conformal field theory as well as the known ( felder - type ) resolutions are used in rational conformal field theories . as an application of the algebra @xmath261 described in [ ext - alg ] , we note the following . for a subcategory @xmath104 in the category @xmath3 ( considered as
either the @xmath1- or @xmath4-representation category ) , the derived category of @xmath104 is equivalent to the derived category of coherent sheaves on a noncommutative extension of @xmath453 @xcite .
the algebra @xmath261 is then the coordinate ring of this noncommutative extension .
in addition to establishing the equivalence of categories , the bimodule @xmath419 constructed in has another application , to the construction of the _ full _ conformal field theory @xcite . given @xmath419 ( a module over the quantum group and the @xmath1 algebra assigned to the holomorphic sector ) and its second copy @xmath454 where the @xmath1-algebra action is replaced with its action in the antiholomorphic sector , the standard construction of the full conformal field theory is to take the tensor product @xmath455 over the quantum group @xmath456 .
other possibilities are to take @xmath457 , where @xmath458 is any @xmath24-bimodule with an associative product @xmath459 consistent with the bimodule structure .
the bimodule @xmath419 may also be used to obtain ( the space dual to ) the quantum group center as `` extended characters '' in logarithmic conformal field theory models .
the `` extended characters '' can be obtained by taking the trace of @xmath460 over the bimodule @xmath419 ( over its vertex - operator - algebra component ) , which should give a version of the general construction in @xcite .
we are grateful to s. ariki , t. miwa and i.v .
tyutin for useful remarks and to j. fuchs for valuable discussions and comments on the literature .
part of the paper was written during our stay at kyoto university , and we are grateful to t. miwa for the kind hospitality extended to us . this paper was supported in part by the rfbr grants 04 - 01 - 00303 , 05 - 02 - 17451 , and lss 1578.2003.2 and the rfbr jsps grant 05 - 01 - 02934yaf_a .
let @xmath140 be an integer @xmath217 and @xmath465 .
the module @xmath87 ( see * 1 . * ) has three subquotients : @xmath86 ( `` left '' ) , @xmath91 ( `` bottom '' ) , and @xmath86 ( `` right '' ) .
the basis can be chosen as @xmath466 where @xmath467 is the basis corresponding to the right @xmath86 module , @xmath468 to the left @xmath86 , and @xmath469 to the bottom @xmath91 module , with the @xmath4-action given by @xmath470[p - s - k]{\mathsf{x}}_{k-1 } , & \quad 0{\,{\leqslant}\,}k&{\,{\leqslant}\,}p{-}s{-}1 \quad(\text{with}\quad{\mathsf{x}}_{-1}\equiv0 ) , \kern-60pt\end{aligned}\ ] ] @xmath471[s - n]{\mathsf{a}}_{n-1 } , & 1{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1,\\ { \mathsf{x}}_{p - s-1 } , & n=0,\\ \end{cases}\\ e{\mathsf{b}}_n&=a[n][s - n]{\mathsf{b}}_{n-1 } , \quad 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1\quad(\text{with}\quad{\mathsf{b}}_{-1}\equiv0),\\ \intertext{and } f{\mathsf{a}}_n&={\mathsf{a}}_{n+1 } , \quad 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1 \quad(\text{with}\quad{\mathsf{a}}_s\equiv0),\\ f{\mathsf{b}}_n&= \begin{cases } { \mathsf{b}}_{n+1 } , & 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}2,\\ { \mathsf{x}}_0 , & n = s{-}1 , \end{cases}\\ f{\mathsf{x}}_k&={\mathsf{x}}_{k+1 } , \quad 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1 \quad(\text{with}\quad{\mathsf{x}}_{p - s}\equiv0).\end{aligned}\ ] ] let @xmath140 be an integer @xmath217 and @xmath465 .
the module @xmath92 , which is the extension @xmath473_{x^+_1 } \ar@/_/[dr]^{x^+_2}&&\\ \stackrel{{\mathscr{x}}^{-}_{p - s}}{\bullet } & & \stackrel{{\mathscr{x}}^{-}_{p - s}}{\bullet } & } \ ] ] has the basis @xmath474 where @xmath467 is the basis corresponding to the top module @xmath86 in , + @xmath469 to the left @xmath91 , and @xmath475 to the right @xmath91 , with the @xmath4-action given by @xmath476[p - s - k]{\mathsf{x}}_{k-1 } , & \quad e{\mathsf{y}}_k&=-a[k][p - s - k]{\mathsf{y}}_{k-1},\quad & 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1\\ * & & & ( \text{with}\quad{\mathsf{x}}_{-1}\equiv{\mathsf{y}}_{-1}\equiv0 ) , & \ ] ] @xmath471[s - n]{\mathsf{a}}_{n-1 } , & 1{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1,\\ { \mathsf{x}}_{p - s-1 } , & n=0,\\ \end{cases}\\ \intertext{and } f{\mathsf{x}}_k&={\mathsf{x}}_{k+1 } , \quad f{\mathsf{y}}_k={\mathsf{y}}_{k+1 } , \quad 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1 & \quad(\text{with}\quad{\mathsf{x}}_{p - s}\equiv{\mathsf{y}}_{p - s}\equiv0),\\ f{\mathsf{a}}_n&= \begin{cases } { \mathsf{a}}_{n+1 } , & 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}2,\\ { \mathsf{y}}_0 , & n = s{-}1 .
\end{cases}\end{aligned}\ ] ] let @xmath140 be an integer @xmath217 , @xmath465 , and @xmath477 . for @xmath478 , the diagram for the @xmath99 module in * 1 .
* takes the form @xmath479^{z_1x^a_1 } \ar@/_3ex/[dr]_{z_2x^a_2}&\\ & \stackrel{{\mathscr{x}}^{-a}_{p - s}}{\bullet } } \ ] ] in terms of the respective bases @xmath467 and @xmath469 in @xmath86 and @xmath91 , the @xmath99 module has the basis @xmath480 with the @xmath4-action given by @xmath481[s - n]{\mathsf{a}}_{n-1 } , & 1{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1,\\ z_2 { \mathsf{x}}_{p - s-1 } , & n=0,\\ \end{cases } \\
e{\mathsf{x}}_k&=-a[k][p - s - k]{\mathsf{x}}_{k-1 } , \quad 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1,\\ \intertext{where we set $ { \mathsf{x}}_{-1}=0 $ , and } f{\mathsf{a}}_n&= \begin{cases } { \mathsf{a}}_{n+1 } , & 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}2,\\ z_1 { \mathsf{x}}_0 , & n = s{-}1,\\ \end{cases } \\ f{\mathsf{x}}_k&={\mathsf{x}}_{k+1 } , \quad 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1,\end{aligned}\ ] ] where we set @xmath482 . that this defines a @xmath4-module
is verified by direct calculation .
let @xmath140 be an integer @xmath217 and @xmath465 .
the projective module @xmath484 has the basis @xmath485 where @xmath468 is the basis corresponding to the top module @xmath86 in , + @xmath467 to the bottom @xmath86 , @xmath469 to the right @xmath91 , and @xmath475 to the left module @xmath91 , with the @xmath4-action given by @xmath486[p - s - k]{\mathsf{x}}_{k-1 } , & \quad 0{\,{\leqslant}\,}k&{\,{\leqslant}\,}p{-}s{-}1 \quad(\text{with}\quad{\mathsf{x}}_{-1}\equiv0 ) , \kern-60pt\end{aligned}\ ] ] @xmath487[p - s - k]{\mathsf{y}}_{k-1 } , & 1{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1,\\ { \mathsf{a}}_{s-1 } , & k=0,\\ \end{cases } \\
e{\mathsf{a}}_n&=a[n][s - n]{\mathsf{a}}_{n-1 } , \quad 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1\quad(\text{with}\quad{\mathsf{a}}_{-1}\equiv0),\\ e{\mathsf{b}}_n&= \begin{cases } a[n][s - n]{\mathsf{b}}_{n-1}+{\mathsf{a}}_{n-1 } , & 1{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1,\\ { \mathsf{x}}_{p - s-1 } , & n=0,\\ \end{cases } \\ \intertext{and } f{\mathsf{x}}_k&= \begin{cases } { \mathsf{x}}_{k+1 } , & 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}2,\\ { \mathsf{a}}_0 , & k = p{-}s{-}1,\\ \end{cases } \\ f{\mathsf{y}}_k&={\mathsf{y}}_{k+1 } , \quad 0{\,{\leqslant}\,}k{\,{\leqslant}\,}p{-}s{-}1 \quad(\text{with}\quad{\mathsf{y}}_{p - s}\equiv0),\\ f{\mathsf{a}}_n&={\mathsf{a}}_{n+1 } , \quad 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}1 \quad(\text{with}\quad{\mathsf{a}}_s\equiv0),\\ f{\mathsf{b}}_n&= \begin{cases } { \mathsf{b}}_{n+1 } , & 0{\,{\leqslant}\,}n{\,{\leqslant}\,}s{-}2,\\ { \mathsf{y}}_0 , & n = s{-}1 .
\end{cases}\end{aligned}\ ] ]
we here recall basic notions about quivers @xcite .
a quiver @xmath488 is an oriented graph , that is , a quadruple @xmath489 , consisting of a finite set @xmath490 of vertices , a finite set @xmath148 of oriented edges ( arrows ) , and two maps @xmath140 and @xmath491 from @xmath148 to @xmath490 .
an oriented edge @xmath492 starts at the vertex @xmath493 and terminates at @xmath494 .
a _ representation _ of a quiver @xmath488 ( over @xmath495 ) is a collection of finite - dimensional vector spaces @xmath496 over @xmath495 , one for each vertex @xmath497 , and @xmath495-linear maps @xmath498 , one for each oriented edge @xmath499 .
the _ dimension _ of a representation @xmath500 of @xmath488 is an element of @xmath501 $ ] given by the dimensions of @xmath496 , @xmath502 : @xmath503 .
a _ morphism _ from a representation @xmath500 of a quiver @xmath488 to another representation @xmath504 of @xmath488 is an @xmath490-graded @xmath495-linear map @xmath505 satisfying @xmath506 for each oriented edge @xmath499 .
this gives the category of representations of the quiver @xmath488 , to be denoted by @xmath507 in what follows .
if a quiver @xmath488 has no oriented cycles , isomorphism classes of simple objects in @xmath507 are in a one - to - one correspondence with vertices of @xmath488 .
the simple object corresponding to a vertex @xmath502 is given by the vector spaces @xmath508 a quiver is said to be of _ finite _ type if the underlying nonoriented graph is a dynkin graph of finite type .
similarly , a quiver is said to be of _ affine _ type if the underlying nonoriented graph is a dynkin graph of affine type .
a quiver is said to be of _ simply laced _
type if it does not have a pair of vertices connected by more than one arrow .
[ [ subsec : rep - kronecker ] ] the category @xmath244 of representations of the kronecker quiver ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the kronecker quiver @xmath243 has two vertices connected by two edges , which are directed in the same way , that is , @xmath509 , where @xmath510 and @xmath511 : @xmath512_g \ar@/^/[r]^f & { \stackrel{1}{\bullet } } } \ ] ] a representation @xmath500 of @xmath243 is a quadruple @xmath513 consisting of two @xmath495-linear spaces @xmath514 and @xmath515 and two linear maps @xmath516 .
the dimension of @xmath500 is given by @xmath517 .
simple objects in the category @xmath244 are given by the two representations @xmath518 and @xmath519 .
we now recall the classification of indecomposable representations of the kronecker quiver @xmath243 , summarized in [ prop : repr - kron ] below .
there is a correspondence between indecomposable representations of a quiver and the set @xmath520 of positive roots of the lie algebra corresponding to the dynkin graph associated with the quiver .
this correspondence is one - to - one for a quiver of simply laced finite type @xcite .
namely , a representation @xmath500 of a quiver @xmath488 is indecomposable if and only if @xmath521 and , conversely , for every @xmath522 , there is , up to an isomorphism , a unique indecomposable representation @xmath500 of the quiver @xmath488 such that @xmath523 . for quivers of affine type , the bijection between isomorphism classes of indecomposable representations and positive roots of the corresponding affine lie algebra holds only for positive real roots @xmath524 @xcite . for each positive imaginary root @xmath525
there exists an uncountable set @xmath526 of nonisomorphic indecomposable representations corresponding to @xmath527 . actually , the set @xmath526 is independent of @xmath525 @xcite .
if @xmath528 $ ] is not in @xmath529 , then the set of indecomposable representations with the dimension @xmath527 is empty .
the nonoriented graph associated with the kronecker quiver @xmath243 is the extended dynkin graph @xmath530 .
it is well known that @xmath531 is a positive real root of @xmath530 if @xmath532 or @xmath533 , and @xmath531 is a positive imaginary root if @xmath534 , where @xmath535 .
in particular , @xmath536 and @xmath537 are the simple roots and @xmath538 is the first imaginary root .
the simple roots @xmath539 and @xmath540 correspond to the respective simple objects @xmath518 and @xmath519 in the category @xmath244 .
the other real roots are in a one - to - one correspondence with indecomposable representations @xmath541 and @xmath542 , where @xmath8 and @xmath543 are the _ rectangular _ matrices @xmath544 the first imaginary root @xmath545 corresponds to the family @xmath546 of indecomposable representations @xmath547 for some @xmath548 , parameterized by @xmath453 .
but the set @xmath526 for a positive imaginary root @xmath527 is independent of @xmath527 , and therefore @xmath549 for any imaginary root @xmath527 .
we summarize these results in the following well - known proposition ( see , e.g. ,
@xcite ) , first obtained in @xcite . 1 . if @xmath550 , then the set of indecomposable representations of @xmath243 with the dimension @xmath527 is empty .
2 . if @xmath551 , then an indecomposable representation of @xmath243 with the dimension @xmath527 is either the representation @xmath552 if @xmath532 or @xmath553 if @xmath533 , @xmath554 , 3 . if @xmath555 , then @xmath549 and indecomposable representations have the dimension @xmath556 for some @xmath554 . 99 d. kazhdan and g. lusztig , _ tensor structures arising from affine lie algebras , _ i , j. amer .
( 1993 ) 905947 ; ii , j. amer .
( 1993 ) 9491011 ; iii , j. amer . math . soc . 7 ( 1994 ) 335381 ;
iv , j. amer . math .
soc . 7 ( 1994 ) 383453 .
feigin , a.m. gainutdinov , a.m. semikhatov , and i.yu .
tipunin , _ modular group representations and fusion in logarithmic conformal field theories and in the quantum group center _ , commun
. math .
phys . , to appear [ hep - th/0504093 ] .
v. lyubashenko , _ invariants of @xmath129-manifolds and projective representations of mapping class groups via quantum groups at roots of unity _ , commun .
( 1995 ) 467516 [ hep - th/9405167 ] ; _ modular properties of ribbon abelian categories _ , in : _
2nd gauss symposium .
conference a : mathematics and theoretical physics _
, m. behara et al . , eds ( de gruyter , berlin 1995 ) , 529579 [ hep - th/9405168 ] . _ modular transformations for tensor categories _
, j. pure applied algebra 98 ( 1995 ) 279327 . | to study the representation category of the triplet @xmath0-algebra @xmath1 that is the symmetry of the @xmath2 logarithmic conformal field theory model , we propose the equivalent category @xmath3 of finite - dimensional representations of the restricted quantum group @xmath4 at @xmath5 .
we fully describe the category @xmath3 by classifying all indecomposable representations .
these are exhausted by projective modules and three series of representations that are essentially described by indecomposable representations of the kronecker quiver .
the equivalence of the @xmath1- and @xmath4-representation categories is conjectured for all @xmath6 and proved for @xmath7 , the implications including the identifications of the quantum - group center with the logarithmic conformal field theory center and of the universal @xmath8-matrix with the braiding matrix . |
it was more than a decade ago when a whispering - gallery - mode ( wgm ) optical resonator has been used to experimentally prove the feasibility of a label - free biosensor@xcite .
in his paper of 2002@xcite , vollmer and colleagues measured the resonance - wavelength shift induced by a layer of proteins on an handcrafted silica microsphere . since then , many other authors have elaborated this concept , improving its scalability@xcite , the sensitivity of the resonator@xcite , the quality factor@xcite , the integration of the sensors in complex and automated system@xcite and the biological functionalization technique for the specific recognition of the target biomolecule@xcite .
we recently demonstrated the possibility to integrate on - chip a monolithic free - standing disk resonator with a vertically - coupled bus waveguide@xcite .
our further works showed how the vertical coupling architecture is particularly suited to couple integrated high quality - factor ( q ) wedge resonators , where the sidewall of the resonator is tilted with respect to the surface of the substrate@xcite .
the very high q is a key aspect for the realization of sensors with the highest resolution . in this sense
, wedge resonators could be good candidates for the realization of hi - performance integrated sensors . in this paper
, we investigate the advantages of these structures from the point of view of their application as refractive index sensors . in the first section ,
we experimentally characterize the optical sensing parameters of a wedge resonator and compare them to those of a disk resonator with similar dimensions . in order to analyze the effects of the shape of the structures under test from a sensor point of view
, we pose the basis and the model for their analysis .
finally , we solve numerically the optimization problem by means of finite - element - analysis .
we find the critical geometrical parameters for achieving best sensing performances and we compare the results to that of the measured samples .
wedge resonators are attractive because of their demonstrated feasibility of obtaining on - chip ultra - high quality - factors , which is particularly appealing for many applications@xcite .
this class of devices are wgm optical resonators lithographically defined by means of an isotropic etching step .
therefore , their sidewalls are not vertical , and the angle with the substrate can be varied by tuning the etching parameters . )
.,scaledwidth=90.0% ] in fig .
[ fig:1](a ) a simple scheme of a wedge wgm resonator is shown , with the definitions of radius ( @xmath1 ) , thickness ( @xmath2 ) and inclination ( @xmath3 ) .
afm images of some of these structures ( fig .
[ fig:1](b ) ) show that a trapezoidal shape is a simplicistic model for the real profile of the wedges . in this work we used the logistic function @xmath4^{-1}\ ] ] to model the sidewalls , where and @xmath5 is the radial coordinate , with origin in the center of the disk . on our samples
this function gives best fits to the profile measurements , as can be seen in fig .
[ fig:1](b ) .
the resonant structures herein analyzed were intended to be operated in the nir - visible spectrum .
thus the chosen dielectrics are : * silicon nitride ( sin@xmath6 ) for the core , for its high refractive index and transparency ; * borophosphosilicate glass ( bpsg ) for the lower cladding , to allow for a cost effective fabrication of the monolithic vertical coupling to bus waveguide@xcite ; * water for the upper cladding ( label - free experiments are actually carried out in buffer solutions , which usually have a refractive index very similar to that of water @xmath7% ) . in order to compare the sensing properties of the commonly used disk resonators ( i.e. those with a vertical sidewall ) versus wedge resonators , we have run two fabrication processes .
these two runs have in common the same pecvd deposition steps and same lithography masks , but they differ in the etching procedure during the resonator definition .
the resulting devices are @xmath8 nm thick , with a radius @xmath9 m and @xmath10 m for disk and wedge respectively .
the disk edge inclination is @xmath11 , while the wedge inclination is @xmath12 .
both resonators are vertically coupled to an integrated silicon oxynitride bus waveguide .
details on the fabrication process are reported in @xcite .
laser light with wavelength close to @xmath13 nm can be coupled in these two structures .
depending on the horizontal displacement of the bus waveguide with respect to the edge of the microresonators the coupling changes and different resonance families can be excited . in this work
this parameter has been adjusted to excite the first and second radial quasi - te family modes .
we measured the spectral position and quality factor of these resonances as a function of the refractive index of the liquid in contact with the sensor surface .
as explained and motivated in section [ sec : theory ] , from this data we calculated the bulk sensitivity @xmath14 and the bulk figure of merit @xmath15 : @xmath16 where @xmath17 , @xmath18 and @xmath19 are the resonance wavelength , linewidth and quality factor , respectively , and @xmath20 is the refractive index of the sensing liquid . during the experiments ,
the temperature of the samples was controlled with a peltier element , and a drop of the liquid to be measured was dispensed with an eppendorf femtojet . to change the refractive index of the sensing liquid we prepared several water - glucose solutions , with concentrations spanning from @xmath21 to @xmath22w / w , which provide a refractive index variation of up to @xmath23riu .
immediately before and after the dispensing , the spectrum of the resonator has been measured using a swept laser at wavelengths in the range @xmath24 nm . thereupon , the surface of the sensor was then rinsed with excess of pure water and dryed with a nitrogen flow for the next measurement .
nm.,height=181 ] in fig .
[ fig : exp](a ) we show an example of the measured resonance shift and we compare the sensing performances of the disk and wedge resonators in fig .
[ fig : exp](b ) .
we notice that , compared to the commonly used first radial resonance of the disk resonator ( @xmath25nm / riu ) , the bulk sensitivity is enhanced by either using the second radial family ( @xmath26nm / riu ) or reducing the inclination ( @xmath27nm / riu ) .
this is reasonable , since both choices have the effect of reducing the confinement of the wgm and , therefore , of increasing the intereaction of the mode with the sensing liquid .
however , the wgms of the second radial family exhibit lower quality factors .
we measured a quality factor @xmath28 and @xmath29 for the second radial families of disk and wedge , respectively , to be compared with @xmath30 and @xmath31 for the first radial families .
this is the reason why the highest bulk fom is achieved by the wedge s first radial wgm family .
compared to the disk , which exibits a fom@xmath32riu@xmath33 , the wedge performs almost twice better , with a fom@xmath34riu@xmath33 . in order to understand such differences
, we decided to analyze by numerical simulation the influence of the geometry of the wedge resoantor on the device performance .
the problem of the optimization of the sensor s parameters can be adressed by different strategies .
for example , in a recent paper the author s aim was to maximize the sensitivity to analyte while minimizing the thermal sensitivity@xcite .
a different approach is used in @xcite , where the optimal parameters for minimizing the analyte detection limit are extensively studied starting from the measurement of the amplitude , position and linewidth of the resonance . in this paper
, we use an approach similar to that of @xcite , whereby we analyze both the sensitivity and quality factor of the supported resonant modes as a function of geometrical parameters of the resonator .
thus , to optimize the sensor performances , one should ideally know the sensitivity and the quality factor of the wgms supported by the possible resonator configurations .
the sensitivity ( [ eq : sb_simple ] ) can be calculated from the optical mode profile @xmath35 . by using a perturbative approach@xcite
, it can be shown that the wavelength sensitivity @xmath36 of the @xmath37-th resonant mode to a variation of the refractive index in the domain @xmath38 can be expressed as @xmath39 here the two integrals are taken in the sensing volume @xmath38 and in the whole model , respectively , and @xmath40 is the unperturbed refractive index of the homogeneous medium contained in @xmath38 . in the same way , in both integrals the dielectric constant @xmath41 and the electric field @xmath42 are meant to be unperturbed , i.e. when no analyte is present . as it can be seen , the sensitivity density is proportional to the electric field energy density @xmath43 . due to the complex gemetrical structure ,
finite element method ( fem ) modeling of the resonator has been used to obtain the mode profiles @xmath42 .
we assume that the mode profile in the cavity is not significantly altered by the presence of the waveguide .
thus , the model used in this work can ignore the presence of the bus waveguide .
the advantage of this choice is that the cylindrical symmetry of the structure is not broken , and a simple and fast axisymmetric 2.5d simulation can be performed .
when light is confined in a waveguide or in a cavity , it is quite common to obtain evanescent tails which extend from the surface of the structure for tens to hundreds of nanometers@xcite .
when used as label - free sensors , the volume of interaction between the analyte and the wgm evanescent field is very small , and mainly constrained by the thickness of the layer of captured analyte . in the case of nanometric - sized molecules ( like proteins ) this means that most of the evanescent tail is unperturbed and not used to produce a signal . for a fair comparison between different structures ,
it is very helpful to introduce the superficial sensitivity @xmath44 defined as @xmath45 where @xmath46 is the area of the bottom surface of a thin layer of volume @xmath38 and thickness @xmath47 .
the approximation in ( [ eq : sensitivity_sup ] ) is valid for layers @xmath38 much thinner than the extension of the evanescent field ( @xmath48 nm ) .
please note the double partial derivative in ( [ eq : sensitivity_sup ] ) : this is different from the definition usually adopted @xmath49 ( see for example @xcite ) which is not independent from the refractive index of the analyte assumed in the volume @xmath38 .
the definition that we adopted is independent from the refractive index @xmath40 and can be used for direct comparison between different structures or models .
calculating the quality factor is a more subtle issue . by taking into account the losses of the structure ,
the hermiticity of the eigenvalue problem is broken and the solutions are complex - valued . from the complex effective refractive index or from the complex eigenfrequency @xmath50 of the resonant mode , the quality factor can be calculated as @xmath51 where @xmath37 is the subscript for the @xmath37-th resonant mode and @xmath52 and @xmath53 are the real and imaginary part of @xmath54 , respectively . in our model
we account for radiative losses , absorption losses and coupling losses .
the firsts are modelled implementing perfect matched layers ( pml ) at the boundaries of the simulation@xcite , the seconds are added in both water and core material as imaginary part of the refractive index , while the last are simply considered by dividing by two the quality factor ( critical coupling regime ) .
we neglected other sources of losses according to our previous ( unpublished ) measurements on similar resonators . to evaluate the bulk and superficial sensing performances and choose the optimal set of parameters , we use two figures - of - merit ( fom ) calculated from the sensitivities and quality factors of every resonance : @xmath55 from now on ,
all the subscripts @xmath37 used to label the different modes are dropped for simplicity of reading and , if not differently specified , @xmath38 and @xmath46 are meant to be the whole sensing volume / area above the parts of the sensor exposed to analytes .
this definition of figure of merit is equivalent to that used in surface plasmon resonance ( spr ) label - free sensors@xcite . with this figure of merit
it is possible to directly compare the performances of sensors exploiting different techniques .
a table which reviews the fom of wgm resonators , photonic crystals ( phc ) and spr sensors is reported in @xcite .
in addition , we also introduce the superficial figure of merit @xmath56 , which aims to compare the performances of surface sensors , used for example in label - free detection .
if the resolution of the resonance wavelength measurement is limited by the resonance linewidth itself and not from other experimental parameters , one can estimate the achievable limit of detection ( lod ) as @xmath57 where @xmath58 accounts for an enhanced resolution given by a proper resonance fitting procedure .
this parameter @xmath58 depends on experimental details and usually lays in the range @xmath59@xcite .
using a commercial fem solver , we have modelled the geometry shown in fig . [ fig:1 ] and we solved for the electric field s eigenfunctions supported by the structure .
we varied the value of the wedge inclination from @xmath60@xmath61 to @xmath62 , and the wedge thickness from @xmath63 nm to @xmath64 nm .
for every solution we calculated the bulk and superficial sensitivity , @xmath14 and @xmath65 , the quality factor , @xmath19 , and the two figures of merit , @xmath66 .
the other geometrical parameters were kept fixed , and their values are reported in tab .
[ tab:1 ] .
.parameters used in the optimization study . [ cols="^,^,<",options="header " , ] nm . ] in fig .
[ fig : modes ] we show the first radial te and tm modes for three different wedge angles and same wedge thickness @xmath67 nm . the difference in the confinement and distribution of the electric field is evident , particularly when comparing the two polarizations . slightly more difficult to notice , but still very interesting , is the fact that the mode reaches its highest confinement for intermediate wedge angles ( roughly between @xmath68@xmath69 and @xmath70 , depending on the other parameters ) . for greater angles
, the field extends mainly from the top surface of the resonator , while for smaller angles the field can sense the area above the external sidewall . in fig .
[ fig:2 ] we report the 2d contour - plot summarising the optimization analysis of the bulk and surface fom for the first radial te and tm modes . and thickness @xmath2 .
the plots report the bulk ( a , b ) and superficial ( c , d ) figure of merit @xmath66 . the red circles in ( a ) mark the properties of the samples characterized in section [ sec : experiments ] .
] for what concerns the bulk sensing ( sensor used as refractometer for liquids ) , it is interesting to notice that a wide set of parameters can be used to achieve good performances .
once that the thickness has been optimized ( i.e. @xmath71 nm for te polarization and @xmath72 nm for tm polarization ) , the value of the inclination has little impact on the bulk fom , at least for @xmath73 . in this configuration
we get @xmath74 for te polarization and @xmath75 for tm polarization .
conversely , the choice of the optimal set of parameters for a sensor used for surface sensing ( e.g. a label - free biosensor ) is much more strict .
the best fom is achieved only with a vertical wedge angle ( i.e. a disk resonator ) both for te ( @xmath76 ) and tm polarization ( @xmath77 ) .
interesting is the fact that the tm polarization can provide higher fom if the resonator is properly optimized . for bulk
sensing the enhancement can be higher than @xmath78 , while for surface sensing the enhancement can exceed @xmath79 . regarding the experimental results of section [ sec : experiments ] , the first observation is that the thicknesses of the wedge and disk resonators were not optimal for sensing applications . for a resonator with @xmath80 nm , the modes with
the highest fom ( both bulk and surface ) are the first two radial te modes , which was also observed experimentally . in fig .
[ fig:2](a ) , the red circles label the parameters of the measured sensors on the bulk fom map .
nm.,width=340 ] to compare the experimental results with the simulations , fig .
[ fig:3 ] reports the bulk fom together with the bulk sensitivity and quality factor of the two aforementioned te modes .
the figure qualitatively confirms the experimental results : for this thickness the wedge resonator can provide a higher bulk figure of merit with respect to a disk resonator .
similarly , the bulk sensitivity is higher for @xmath81@xmath61 than for @xmath82@xmath61 and even higher for the second mode in both structures .
finally , a note should be done about the accuracy of the quality factor in the simulations : in our model we did nt take into account possible variations in the scattering losses due to the roughness of the resonator .
in particular , the two compared structures were defined with two different etching processes , which are known to lead to different final roughnesses .
this can explain why the real quality factor , as well as the bulk fom , are actually lower than expected in the case @xmath82 .
the same does not apply to the case of the bulk sensitivity , since @xmath83 does not depend on the roughness of the structure .
the results of our analysis clearly show that the control on the resonator thickness is the most important parameter to achieve optimal sensing .
the use of vertical coupling is thus necessary to have this freedom in the fabrication without imposing limitations on the geometry of the bus waveguide@xcite .
this technique , in fact , relies on two deposition and lithography steps , where the thicnkess of both deposited layers can be changed independently according to the results of the optimization analysis .
in this work we tested the effectiveness of vertically - coupled integrated wedge resonators for refractometric sensing applications , with a particular focus to label - free biosensing .
these devices are very appealing because of the very high quality factor that they can exhibit . nevertheless , the quality factor is not the only concern when analyzing the device performances .
the figures of merit of interest for sensing applications are also affected by the field overlap with the analyte .
we experimentally verified that under certain conditions a wedge resonator can perform better compared to a disk resonator with similar dimensions . at the same time , however , we also showed , by means of numerical analysis , that optimal design parameters can be achieved by changing the thickness of a normal disk resonator ( inclination @xmath84 ) .
we also showed that the tm - polarization modes can lead to higher performances with respect to te modes , provided an adequate resonator thickness is used to reduce the radiation losses . in the modelling we used two figures of merit ( bulk and surface ) that permit an easy and fair comparison between sensors with different geometries , even between sensors relying on different physical principles ( e.g. spr or phc ) .
the bulk fom shows that our structures perform at least one order of magnitude better than spr devices for bulk refractometric sensing , which are in the order of @xmath85 @xcite .
in addition , we introduced the surface fom , that can be used to compare different structures with special focus on surface sensing ( like the label - free method ) . to summarize ,
the use of wedge resonators in place of conventional disk resonators is not suggested , at least for resonators similar to the ones described by tab .
[ tab:1 ] .
the isotropic etching step is still suggested to obtain better surface quality , but cares have to be taken in order to keep the inclination @xmath86 .
the analysis here presented shows that the optimization of the resonator thickness is critical to achieve the best performances . in this regard
, the use of vertical coupling has emerged to be particularly appealing : not only it avoids any interaction with analyte in the coupling region during sensing , but in addition it gives control on the resonator thickness and geometry without posing limitation to the waveguide .
this works was supported by the fp7 eu project `` symphony '' and by fu - pat through the `` naomi '' project .
the authors thanks dr .
romain guider for the fruitful discussions . | in this paper we report on the measurement and modelling of enhanced optical refractometric sensors based on whispering - gallery - modes .
the devices under test are optical microresonators made of silicon nitride on silicon oxide . in our approach , these microresonators are vertically coupled to a buried waveguide with the aim of creating integrated and cost - effective devices .
the optimization analysis is a delicate balance of resonance quality factor and evanescent field overlap with the sorrounding environment to analyze . by numerical simulations
we show that the microdisk thickness is critical to yield high figure of merit for the sensor , while edge inclination is less important .
we also show that figures of merit as high as @xmath0 are feasible . 10 c. ciminelli , c. m. campanella , f. dellolio , c. e. campanella , and m. n. armenise , `` label - free optical resonant sensors for biochemical applications , '' progress in quantum electronics * 37 * , 51107 ( 2013 ) .
f. vollmer , d. braun , a. libchaber , m. khoshsima , i. teraoka , and s. arnold , `` protein detection by optical shift of a resonant microcavity , '' applied physics letters * 80 * , 4057 ( 2002 ) .
k. de vos , j. girones , t. claes , y. de koninck , s. popelka , e. schacht , r. baets , and p. bienstman , `` multiplexed antibody detection with an array of silicon - on - insulator microring resonators , '' ieee photonics journal * 1 * , 225235 ( 2009 ) .
a. l. washburn , m. s. luchansky , a. l. bowman , and r. c. bailey , `` quantitative , label - free detection of five protein biomarkers using multiplexed arrays of silicon photonic microring resonators . ''
analytical chemistry * 82 * , 6972 ( 2010 ) .
o. gaathon , j. culic - viskota , m. mihnev , i. teraoka , and s. arnold , `` enhancing sensitivity of a whispering gallery mode biosensor by subwavelength confinement , '' applied physics letters * 89 * , 223901 ( 2006 ) .
d. k. armani , t. j. kippenberg , s. m. spillane , and k. j. vahala , `` ultra - high - q toroid microcavity on a chip . ''
nature * 421 * , 9258 ( 2003 ) .
h. lee , t. chen , j. li , k. y. yang , s. jeon , o. painter , and k. j. vahala , `` chemically etched ultrahigh - q wedge - resonator on a silicon chip , '' nature photonics * 6 * , 369373 ( 2012 ) .
m. iqbal , m. a. gleeson , b. spaugh , f. tybor , w. g. gunn , m. hochberg , t. baehr - jones , r. c. bailey , and l. c. gunn , `` label - free biosensor arrays based on silicon ring resonators and high - speed optical scanning instrumentation , '' ieee journal of selected topics in quantum electronics * 16 * , 654661 ( 2010 ) .
l. pasquardini , s. berneschi , a. barucci , f. cosi , r. dallapiccola , m. insinna , l. lunelli , g. n. conti , c. pederzolli , s. salvadori , and s. soria , `` whispering gallery mode aptasensors for detection of blood proteins . ''
journal of biophotonics * 6 * , 17887 ( 2013 ) .
m. ghulinyan , r. guider , g. pucker , and l. pavesi , `` monolithic whispering - gallery mode resonators with vertically coupled integrated bus waveguides , '' ieee photonics technology letters * 23 * , 11661168 ( 2011 ) .
f. ramiro - manzano , n. prtljaga , l. pavesi , g. pucker , and m. ghulinyan , `` a fully integrated high - q whispering - gallery wedge resonator , '' optics express * 20 * , 22934 ( 2012 ) . c. delezoide , i. ledoux - rak , and c. t. nguyen , `` general approach for the sensitivity analysis and optimization of integrated optical evanescent - wave sensors , '' journal of the optical society of america b * 31 * , 851 ( 2014 ) .
m. r. foreman , w .- l .
jin , and f. vollmer , `` optimizing detection limits in whispering gallery mode biosensing . ''
optics express * 22 * , 5491511 ( 2014 ) . c. ciminelli , f. dellolio , d. conteduca , c. campanella , and m. armenise , `` high performance soi microring resonator for biochemical sensing , '' optics & laser technology * 59 * , 6067 ( 2014 ) .
s. johnson , m. ibanescu , m. skorobogatiy , o. weisberg , j. joannopoulos , and y. fink , `` perturbation theory for maxwell s equations with shifting material boundaries , '' physical review e * 65 * , 066611 ( 2002 ) .
i. teraoka , s. arnold , and f. vollmer , `` perturbation approach to resonance shifts of whispering - gallery modes in a dielectric microsphere as a probe of a surrounding medium , '' journal of the optical society of america b * 20 * , 1937 ( 2003 ) .
e. rigo , f. j. aparicio , m. r. vanacharla , s. larcheri , r. guider , b. han , g. pucker , and l. pavesi , `` evanescent - field excitation and collection approach for waveguide based photonic luminescent biosensors , '' applied physics b * 114 * , 537544 ( 2013 ) .
f. j. aparicio rebollo , e. froner , e. rigo , d. gandolfi , m. scarpa , b. han , m. ghulinyan , g. pucker , and l. pavesi , `` silicon oxynitride waveguides as evanescent field based luminescent biosensors , '' journal of physics d : applied physics * _ in press _ * ( 2014 ) .
m. i. cheema and a. g. kirk , `` accurate determination of the quality factor and tunneling distance of axisymmetric resonators for biosensing applications . ''
optics express * 21 * , 872435 ( 2013 ) .
y. shen , j. zhou , t. liu , y. tao , r. jiang , m. liu , g. xiao , j. zhu , z .- k .
zhou , x. wang , c. jin , and j. wang , `` plasmonic gold mushroom arrays with refractive index sensing figures of merit approaching the theoretical limit . ''
nature communications * 4 * , 2381 ( 2013 ) .
l. huang , h. tian , d. yang , j. zhou , q. liu , p. zhang , and y. ji , `` optimization of figure of merit in label - free biochemical sensors by designing a ring defect coupled resonator , '' optics communications ( 2014 ) .
f. vollmer and l. yang , `` review label - free detection with high - q microcavities : a review of biosensing mechanisms for integrated devices , '' nanophotonics * 1 * , 267291 ( 2012 ) . |
we would like to thank giorgio parisi for continuous interactions and advises , the `` infn sezione di roma i '' for the use of the ape100 computer and the `` universit di roma la sapienza '' for kind hospitality .
we thank also m. a. virasoro and r. zecchina for useful discussions .
w. l. mcmillan , j. phys .
c * 17 * , 3179 ( 1984 ) .
a. j. bray , m. a. moore , in _ heidelberg colloquium on glassy dynamics _ , lecture notes in physics * 275 * , j. l. van hemmen and i. morgenstern ( ed .
springer , berlin ) .
d. s. fisher , d. a. huse , phys .
* 56 * , 1601 ( 1986 ) and phys .
b * 38 * 386 ( 1988 ) .
e. marinari , g. parisi , j. j. ruiz - lorenzo and f. ritort , phys .
lett . * 76 * , 843 ( 1996 ) .
e. marinari , g. parisi and j. j. ruiz - lorenzo , `` numerical simulations of spin glass systems '' in _ spin glasses and random fields _ , edited by a. p. young .
world scientific ( singapore , 1997 ) .
e. marinari , g. parisi , f. ricci - tersenghi and j. j. ruiz - lorenzo , j. phys .
a * 31 * , l481 ( 1998 ) .
g. parisi , f. ricci - tersenghi and j. j. ruiz - lorenzo , phys .
b * 57 * , 13617 ( 1998 ) . | we perform an accurate test of ultrametricity in the aging dynamics of the three dimensional edwards - anderson spin glass .
our method consists in considering the evolution in parallel of two identical systems constrained to have fixed overlap .
this turns out to be a particularly efficient way to study the geometrical relations between configurations at distant large times .
our findings strongly hint towards dynamical ultrametricity in spin glasses , while this is absent in simpler aging systems with domain growth dynamics .
a recently developed theory of linear response in glassy systems allows to infer that dynamical ultrametricity implies the same property at the level of equilibrium states . 2 the dispute about the nature of the spin glass phase of finite dimensional spin glasses
has lasted by now almost twenty years , after that the droplet model @xcite has challenged the predictions of mean field theory ( mft ) @xcite . in the last years a large collection of numerical data have tested , with positive answer , the applicability of spin glass mean field picture to finite dimensional ( @xmath0=3,4 ) systems @xcite .
however some of these numerical evidences have been recently re - interpreted as finite volume effects by the authors of ref .
@xcite and so many questions on the low - temperature phase of finite dimensional spin glasses still remain unaswered .
one of the most characteristic aspects of mean field theory is the prediction of ultrametricity ( um ) @xcite . at low temperature
the ergodicity is broken and many low free energy states are present .
ultrametricity implies that the distances between these states verify an inequality stronger than the triangular one ( see below ) .
however this property has been rather elusive to a direct probe .
the best evidences in favor of this property have been given ( to our knowledge ) in refs .
@xcite . in the former
it was studied the equilibrium of a 4@xmath0 spin glass model using small samples . in the latter very - low - energy configurations of a 3@xmath0 model
were used , again for relatively small samples . in both cases an extrapolation to large volumes was needed in order to check ultrametricity . in this paper we present results for the 3@xmath0 spin glass model with a method that allows us to reach much larger sizes .
another fundamental aspect of mft is the prediction of slow dynamics and aging @xcite .
this is a non stationary asymptotic regime , following a quench from a high temperature , which persists forever in infinite systems . in this out - of - equilibrium regime
, the equilibrium property of ultrametricity has a dynamical counterpart in ultrametric relations among time dependent autocorrelation functions .
recent results of linear response theory succeeded to relate in a unique way properties of statics and dynamics , relying on the recently introduced hypothesis of stochastic stability @xcite .
this property states the continuity of the average correlation functions under weak random perturbation of the hamiltonian .
the validity of the property in three and four dimensional spin glasses has been numerically verified in ref .
@xcite .
a remarkable consequence of stochastic stability is that dynamical ultrametricity implies the static one @xcite . in this paper
we investigate the possibility for dynamical um in the three dimensional edwards - anderson ( ea ) model . for comparison
we also perform the same test on models with domain coarsening off - equilibrium dynamics , where um should not be expected .
the high evidence we can achieve is based on a new dynamical method where we evolve in parallel two identical systems ( replicas for short ) with fixed value of the mutual overlap .
this is analogous to a conserved - order - parameter dynamics in a ferromagnetic system and it is similar to the one already used in ref .
@xcite to study the equilibrium behavior .
the ea model is defined by the hamiltonian @xmath1 where the spins are ising variables , the sum spans the nearest neighbors pairs on a cubic lattice , and the couplings @xmath2 are normally distributed quenched independent random variables .
we define the overlap among two spin configurations @xmath3 and @xmath4 as @xmath5 , which is directly related to the hamming distance @xmath6 through @xmath7 . the mean field equilibrium solution of this model has the ultrametric property at low temperature ; for each three configuration @xmath3 , @xmath8 and @xmath9 chosen with boltzmann probability , the following inequalities hold : @xmath10 which are much stronger than the usual triangular one @xmath11 .
moreover in ref .
@xcite has been shown that , if some kind of ultrametricity holds in the low - temperature phase of the ea model , it must be of the same kind of that present in the mean - field solution . in the off - equilibrium dynamical solution a relation analogous to eq.([sum ] ) holds for the two - time auto - correlation functions : @xmath12 .
taken three large times @xmath13 , one finds that @xmath14 the precise statement is that the relation among the correlations _ should tend _ to the one of eq.([dum ] ) in the infinite time limit @xcite . in the simulations
the relation among the three correlations is plagued by strong finite time effect , so that a direct verification would be difficult with the present computer resources .
we therefore decided to probe a relation which for long times is consequence of , and equivalent to , eq.([dum ] ) .
we consider the dynamics of two replicas of the system with the same disorder , @xmath3 and @xmath8 , which evolve in parallel with different thermal noises and constrained at each time to have a fixed mutual overlap @xmath15 .
the dynamics follows a quench at time zero and we measured the auto- and cross - correlation functions @xmath16 note that @xmath17 for ising variables , while the constraint implies @xmath18 .
we would like to argue , with a hand waving argument , that the relation in eq.([dum ] ) entails the following ultrametric constraint on the cross - correlation function if the value of @xmath15 is between @xmath19 and the value of the edwards - anderson parameter : @xmath20 during the relaxation the free energy of the system decreases monotonically towards its equilibrium value .
this will be higher or equal to the one of the unconstrained system at the same temperature .
the basic observation is that , if @xmath15 is one of the values allowed for the overlap among equilibrium states , _
i.e. _ @xmath21 $ ] , the equilibrium free - energy of the constrained system should coincide with the one of the unconstrained one @xcite .
eq.([dum ] ) expresses the fact that for large times , the directions in which the system can go without increasing the free - energy must be compatible with ultrametricity
. the constrained system can lower its free - energy down to the equilibrium value only if all the possible correlations one can form ( auto and cross ) verify for long times um inequalities . as two of the 4 correlations that can be formed with the configurations @xmath22 , @xmath23 , @xmath24 and @xmath25
are fixed to @xmath15 , eq.([cum ] ) should follow . a formal argument leading to the same conclusions
can be formulated @xcite , extending to constrained systems the correspondence among statics and dynamics mentioned above . in order to see how restrictive relation in eq.([cum ] ) is , let us discuss what one can expect on a general ground for the relation among @xmath26 and @xmath27 in a relaxational system . in fig .
[ area ] we display the set of allowed values for @xmath28 and @xmath29 ( shaded area ) , simply assuming the triangular relation and a monotonic decrease of both functions when the time argument s difference increases .
the set of values allowed by the um relation is represented with the bold dashed line .
= 0.9 in order to check whether eq.([cum ] ) holds in the 3@xmath0 ea model , we have simulated , by monte carlo method ( metropolis update ) , two coupled systems with a soft constraint @xmath30 , imposed modifying the boltzmann weight with a gaussian of width proportional to @xmath31 , such that the weight of a configuration would be @xmath32 $ ] .
the value of the @xmath33 parameter must be appropriately tuned : a too small value would not force enough the systems and their overlap will be systematically different from the one we fixed ( @xmath15 ) . on the other hand a too large value would render movements too unprobable and the dynamics would evolve very slowly . in the whole set of runs we fixed @xmath34 for the ea model and @xmath35 for the other models .
the choice has been made with the aim of maximizing the monte carlo acceptance rate , avoiding the systematic errors just described .
we have checked ( see fig . [ conv ] ) that , for these values of @xmath33 , @xmath36 tends to the desired value @xmath15 .
as we study the behavior of the model in the aging regime , we do not need to reach thermalization and we can simulate large sizes ( @xmath37 ) . for such volume ,
finite size effects do not affect the dynamical regime we study .
the critical temperature of a single uncoupled model is @xmath38 @xcite and we simulate the system in the spin glass phase at @xmath39 . the large size , the low temperature and the starting configuration , which is randomly chosen ,
ensure that the system stays in the aging regime all along the simulation which can be as long as @xmath40 mcs .
these high performances have been achieved using the parallel computer ape100
@xcite .
all the correlation functions we measure are extensive and self - averaging quantities , so we do not need a large number of disorder realizations .
their fluctuations are small thanks to the large volume used and we average the results on a quite small number of samples ( @xmath41 ) .
the error on the data is always calculated as the sample - to - sample fluctuation .
= 0.9 the first quantities we have studied are the overlap of the two replicas and the internal energy as a function of time .
as announced , we see in fig .
[ conv ] that the overlap converges towards the value @xmath15 we have fixed . the `` soft '' way of imposing the constraint is evident in the behavior of the overlap , which is not equal to @xmath15 during all the run , but clearly converges to it . in the forecoming analysis
we will use only the data obtained in the time range where the overlap is statistically compatible with @xmath15 , that is @xmath42 .
the data in the inset of fig .
[ conv ] show the difference between the energy of the constrained system and that of a free system evolving at the same temperature after a quench at time zero .
as we expected , the energy difference tends to zero , as it should if the value @xmath15 is allowed at equilibrium , while for @xmath43 the energy difference goes to a finite value . in fig .
[ sg ] we show our main result : plotting the cross - correlation as a function of the auto - correlation we obtain , for relatively long times , that the data points are quite close to the um bound .
a perfectly ultrametric system would stay on the line @xmath44 as long as @xmath45 and then would follow the line @xmath46 ( both lines are plotted in fig .
[ sg ] ) . the data for the 3@xmath0 ea model are clearly converging to the um bound : we remind the reader that the data can not leave the shaded area of fig .
[ area ] , that is they can not cross any of the boundary lines reported in fig .
[ sg ] , and so we expect that they naturally converge to the um bound . to understand better how probing are our data , we have also simulated two models where um is not expected to hold : the 2@xmath0 ferromagnetic ising model , and its site diluted version in 3@xmath0 .
the off - equilibrium dynamics of both models shows a domain coarsening regime where the correlation function depends on both times as @xmath47 , a scaling form incompatible with ultrametricity .
= 0.9 we have simulated a 2@xmath0 pure ferromagnetic ising model of linear size @xmath48 at a temperature @xmath49 ( @xmath50 ) .
we choose a very large size and small waiting times ( @xmath51 ) in order to avoid that the system gets out from the aging regime we are interested in .
these times are sufficiently large to be close to the scaling regime .
the data are averaged on @xmath52 different noise realizations .
= 1.05 the data from the pure ferromagnetic system are plotted in fig .
[ ferro_dilu]a in the usual @xmath53 versus @xmath54 plot .
we also report the bounds of the allowed region ( the shaded region of fig .
[ area ] ) .
note that the data are far away from the um bound and they seem to converge to some @xmath55-independent curve .
maybe one can think that the comparison of a spin glass with the pure ferromagnet is not enough .
so we have simulated also a 3@xmath0 site - diluted ferromagnetic ising model , which has a coarsening dynamics similar to that of the pure model , but much slower @xcite and complicated by interface pinning .
we have simulated two samples ( each one consisting of a pair of interacting systems ) of linear size @xmath56 and spin concentration @xmath57 .
the temperature is well deep in the frozen phase , @xmath58 , the critical temperature being @xmath59 @xcite . in fig .
[ ferro_dilu]b we see that the behavior of the data from the diluted ferromagnetic model may resemble that of the ea model , because it seems to be somehow close to the um bound .
= 0.95 looking carefully at the figure we note , however , that the @xmath55-dependence of the data in fig .
[ sg ] and in fig .
[ ferro_dilu]b are opposite . in fact increasing the waiting time , the value of the auto - correlation function at the point where the data leave the horizontal line decreases in the spin glass case , while it increases in the diluted ferromagnetic case . to make this effect clearer we zoomed the region of fig .
[ sg ] and [ ferro_dilu]b near the horizontal line ( see fig .
[ zoom ] ) . summarizing ,
we have used a new numerical method to test ultrametricity in short range spin glasses .
we find evidence that for long times the ultrametric equality between two time correlations become fulfilled .
as we already stressed the property of stochastic stability implies then static ultrametricity .
the behavior of spin glasses is strikingly different from the behavior of ordered and disordered models with domain coarsening , where we find incompatibility with ultrametricity . |
light elements such as h , he were mainly synthesized in the big - bang , whereas heavier elements were synthesized primarily in many different stellar nucleosynthesis after the formation of the first generation of stars in the galaxy .
the solar system was formed from interstellar media ( ism ) , which compositions have been provided from the stellar nucleosynthesis . the solar system abundance is therefore an important record of these stellar nucleosyntheses and the galactic chemical evolution ( gce ) .
the solar system abundance gives several evidence that two neutron capture reaction processes have happened before the solar system formation .
the first evidence is pairs of two abundance peaks near three - neutron magic numbers of n = 50 , 82 , 126 .
these two peaks are corresponding to the s- and r - processes , which originated from the positions of the two nucleosynthesis flows in the nuclear chart @xcite .
the second evidence for the s - process is an empirical relation , @xmath4 @xmath5 constant , where @xmath6 and @xmath7 are the neutron capture cross - section and the solar abundance for the pure @xmath1-nuclei @xcite .
this can be understood by a concept of `` steady flow '' . in this way
, the evidence observed in the solar abundances are correlated with the fundamental mechanisms of the nucleosyntheses .
we here focus attention to origin of @xmath0-process isotopes ( @xmath0-nuclei ) .
they have the following features .
first , they can not be synthesized by the neutron - capture reactions because they are on the neutron - deficient side from the @xmath8-stability line in the nuclear chart ( see fig .
[ fig : chart ] ) .
second , their isotopic fractions are small ( typically 0.1%-1% ) .
_ suggested that the @xmath0-nuclei may be synthesized by ( @xmath2,n ) or ( p,@xmath2 ) reactions about 50 years ago @xcite .
arnould ( 1976 ) proposed the @xmath0-process in pre - supernova phases , and woosley and howard ( 1978 ) proposed the @xmath2-process in supernovae ( sne ) . in these pioneering works , the @xmath0-nuclei
are synthesized mainly by photodisintegration reactions from pre - existing nuclei affected by the @xmath1-process in earlier evolutionary states of progenitors .
woosley and howard ( 1978 ) pointed out that the anti - correlation between the photodisintegration reaction rates and the solar abundances for the p - nuclei , which is a piece of evidence of the @xmath2-process .
after these works , detailed calculations reproduced the relative solar abundances of most @xmath0-nuclei within a factor of 3 @xcite . nevertheless , their origin has long been discussed with many possible nuclear reactions over the last 50 years , and their astrophysical sites have not been identified uniquely .
the proposed nuclear processes are the rapid proton capture reactions in novae or type i x - ray bursts in neutron stars ( rp - process ) @xcite , the proton - induced reactions by galactic cosmic rays @xcite , the @xmath2-process in core - collapse sne @xcite the explosive nucleosynthesis in type ia sne @xcite , supernova - driven supercritical accretion disks @xcite , and the neutrino - induced reactions in sn explosions ( @xmath9-process ) @xcite .
the origin of the p - nuclei is crucial to our understanding of how the solar - system material formed and evolved . in our previous paper @xcite
, we reported two empirical scaling laws found in the solar system abundances , which are a piece of evidence that the most probable origin of the @xmath0-nuclei is the sn @xmath2-process .
the empirical laws lead a novel concept of `` the universality of the @xmath2-process '' that each sn @xmath2-process arising from different conditions in individual sn explosions should reproduce n(s)/n(p ) @xmath10 constant over a wide range of atomic number @xcite .
the purpose of this paper is to report detailed analyses of these scalings .
we present the mechanism of this universality in core - collapse sn @xmath2-process model calculations .
the @xmath0-nuclei are on the neutron - deficient site from the @xmath8 stability line as shown in fig . [
fig : chart ] .
typical @xmath0-nuclei are even - even isotopes , of which both proton and neutron numbers are even .
they are isolated in the nuclear chart , around which isotopes are unstable .
there are twenty - two pairs of a @xmath0-nucleus and an @xmath1-nucleus that is heavier than the @xmath0-nucleus by two neutrons . in most cases ,
the @xmath1-nuclei are pure @xmath1-nuclei that are dominantly synthesized by the @xmath1-process since these @xmath1-nuclei are shielded by stable isobars against @xmath11-decay after freezeout of the @xmath12-process .
nine elements have two @xmath0-nuclei .
we here define the first and second @xmath0-nuclei that are lighter than the @xmath1-nucleus by two and four neutrons , respectively .
figure [ fig : chart ] shows a partial nuclear chart as a typical example .
@xmath13ba is a pure @xmath1-nucleus shielded by a stable isobar @xmath13xe against the @xmath8-decay after the freezeout of the @xmath12-process .
the stable isotopes , @xmath14ba and @xmath15ba , are the first and second @xmath0-nuclei , respectively .
it is noted that @xmath16sm is the 2nd p - nucleus and @xmath17sm is the @xmath1-nucleus but the first @xmath0-nucleus @xmath18sm is unstable .
there are thirty - five @xmath0-nuclei : twenty - two first @xmath0-nuclei , ten second @xmath0-nuclei , and other three @xmath0-nuclei .
the other three @xmath0-nuclei are odd-@xmath19 isotopes of @xmath20sn , @xmath21la , and @xmath22ta and beyond of the scope of following discussion of the solar abundance ratios . here
we discuss the isotope abundance ratios of a @xmath0-nucleus and an @xmath1-nucleus with the same atomic number . taking the abundance ratios of the s - nucleus to the p - nucleus , n@xmath23/n@xmath24(p ) , where n@xmath24(s ) and n@xmath24(p ) are the solar isotope abundances of the @xmath1- and @xmath0-nuclei , respectively .
for example , the solar isotope abundances of a @xmath0-nucleus @xmath14ba and a @xmath1-nucleus @xmath13ba are 0.101% and 2.417% , respectively , and the isotope abundance ratio of n@xmath24(@xmath13ba)/n@xmath24(@xmath14ba ) is 23.93 .
we reported the first scaling that n@xmath24(@xmath1)/n@xmath24(@xmath0 ) ratios are almost constant over a wide region of atomic number @xcite .
we also reported another empirical scaling between two p - nuclei with the same atomic number as shown in fig .
[ fig : sp_solar]@xcite .
nine elements have two p - nuclei . as shown in fig .
[ fig : pp_solar ] , the n@xmath24(1st p)/n@xmath24(2nd p ) ratios are almost constant over a wide region of the atomic number except a large deviation of er , of which the reason will be discussed later .
the n@xmath24(s)/n@xmath24(1st p ) and n@xmath24(1st p)/n@xmath24(2nd p ) ratios show clear correlation @xcite , but a question whether the ratios of n@xmath24(s)/n@xmath24(2nd p ) can be described by the first scaling or not has been remained . here
we present both ratios of n@xmath23/n@xmath24(1st p ) and n@xmath23/n@xmath24(2nd p ) in fig .
[ fig : sp_solar ] .
there is a clear correlation between them .
the ratios are almost constant over a wide range of the atomic number except for some deviations , of which the reason will be discussed later .
the ratios in the region of 49 @xmath25 z @xmath25 72 except ce and er is centered around their average of 23.2 .
in addition , most ratios over a wide region are constrained at a constant value of n@xmath24(s)/n@xmath24(p ) @xmath10 23 within a factor of 2 .
we summarize the isotope abundances and their ratios in table 1 .
the first scaling shows a strong correlation between p- and s - nuclei with the same atomic number , which indicates that the origin of the p - nuclei is strongly correlated with the s - nuclei .
this is consistent with the previous theoretical calculations that the p - nuclei are produced by the @xmath2-process ( @xmath0-process ) in sn explosions @xcite . in the @xmath2-process models ,
pre - existing nuclei in massive stars are affected by the weak s - process during pre - supernova evolutionary stages .
these pre - existing nuclei originate from early generations of stars through interstellar media .
the @xmath0-nuclei are produced from them by photodisintegration reactions such as ( @xmath2,n ) reactions in a huge photon bath at extremely high temperatures in sn explosions .
the previous calculations indicated that the @xmath0-nuclei are produced via two nuclear reaction paths .
the first is direct ( @xmath2,n ) reactions from heavy isotopes .
the second is the ec/@xmath26-decay from the neutron - deficient unstable nuclei after the freezeout of the @xmath2-process .
these neutron - deficient nuclei are first transmuted by successive photodisintegration reactions as ( @xmath2,n ) reactions from heavier isotopes and subsequently transmuted by downflows of ( @xmath2,p ) and ( @xmath2,@xmath27 ) reactions .
the first scaling suggests that the former reaction path are likely to play a role more important than the latter reaction path .
the contribution of these two nucleosynthesis paths is quantitatively discussed later .
the charged particle reactions in the rp - process @xcite and proton - induced reactions by the cosmic rays @xcite change the proton number of the seed nuclei . in the @xmath9-process , the charged current interaction that has a contribution larger than the neutral current interaction also changes the proton number @xcite .
therefore , the scaling does not emerge from the dominant charged particle processes and the @xmath9-process .
the first scaling is , thus , a piece of evidence that the sn @xmath2-process is the most probable origin of the @xmath0-nuclei .
figure [ fig : abundances ] shows the solar abundances of the @xmath0-nuclei and @xmath1-nuclei that are members of the first scaling .
the correlation between the @xmath0- and @xmath1-nuclei can be observed in this figure .
seeger et al .
( 1965 ) pointed out that the empirical relation , @xmath4 @xmath5 constant for the pure @xmath1-nuclei .
we would like to stress that the abundances of the @xmath1-nuclei that are the members of the first scaling , @xmath7 , is almost constant in the two mass regions of 50 @xmath28 @xmath19 @xmath28 82 and 82 @xmath28 @xmath19 , respectively .
the abundance pattern of the @xmath0-nuclei shows a tendency similar to that of the @xmath1-nuclei .
the abundances of the @xmath0-nuclei are almost constant in the two mass regions of 50 @xmath28 @xmath19 @xmath28 82 , and 82 @xmath28 n , respectively .
there are some enhancements of the @xmath0-nuclei : for example @xmath29sn , @xmath30hg .
their partner @xmath1-nuclei , @xmath31sn and @xmath32hg , also show enhancements .
these enhancements show the abundances of the @xmath0-nuclei are proportional to those of the @xmath1-nuclei with the same atomic number .
these are consistent with the first scaling that the n@xmath24(s)/n@xmath24(p ) ratios are almost constant over a wide range of the atomic number .
figure [ fig : sp_solar ] shows some deviations from the average value of 23 .
these deviations can be explained by contributions of other nucleosynthesis processes or other astrophysical origin .
+ + * ce , er , and w * : the three large deviations for ce , er and w can be explained by an exceptional contribution from the r - process because the heavier isotopes in the pairs , @xmath33ce , @xmath34er , and @xmath35w , are not shielded against @xmath11-decay after the freezeout of the r - process .
the r - process contributions to @xmath34er and @xmath35w are larger than the s - process contributions in theoretical calculations @xcite . in addition
the effect of the neutron magic number n=82 should contribute to the deviation for ce .
+ + * gd * : the small deviation for gd can be explained by a contribution of a weak branch of the @xmath1-process . a unstable nucleus @xmath36sm is known as a branching point of the @xmath1-process , where the neutron capture reaction and @xmath8-decay compete .
the @xmath0-nucleus @xmath37gd is produced by a following @xmath1-process branch @xmath38 as the result , the n@xmath24(@xmath39gd)/n@xmath24(@xmath37gd ) ratio decreases from the n(@xmath39gd)/n(@xmath37gd ) ratio at sne .
+ + * kr * : the deviation of n@xmath24(@xmath40kr)/n@xmath24(@xmath41kr ) may originate from a weak branch of the @xmath1-process . as shown in fig .
[ fig : kr ] , @xmath42se is a branching point and @xmath40kr is synthesized by a weak branch of the @xmath1-process .
@xmath43kr is located on the main flow of the @xmath1-process and the isotope abundance of @xmath43kr is about 4 times larger than that of @xmath40kr . if we take a ratio n@xmath24(@xmath43kr)/n@xmath24(@xmath41kr ) instead of n@xmath24(@xmath40kr)/n@xmath24(@xmath41kr ) , it is about 33 .
+ + * mo and ru * : four isotopes of @xmath44mo and @xmath45ru located near the neutron magic number of @xmath19=50 show large deviations .
isotope abundances of typical @xmath0-nuclei in a mass region around mo and ru are 0.3 - 1% , whereas the isotope abundances of @xmath44mo and @xmath45ru are 14.84 , 9.25 , 5.54 , 1.87% , respectively .
thus , the ratios of mo and ru are lower than the average value of 23 by an order of magnitude .
previous @xmath2-process ( or @xmath0-process ) model calculations tried to reproduce the relative abundances of all the @xmath0-nuclei .
however , the calculated results showed the underproduction of mo and ru . @xcite .
the large deviations in the n@xmath24(s)/n@xmath24(p ) ratios are consistent with the previous calculated results and suggest that their origin may be different with the other @xmath0-nuclei .
+ + * sm * : the n@xmath24(@xmath17sm)/n@xmath24(@xmath16sm ) ratio shows large deviation like those of mo and ru .
the @xmath0-nucleus @xmath16sm is on the neutron magic number of @xmath19=82 .
this situation is similar to mo and ru and the large deviation suggests that the origin of @xmath16sm might be same as those of mo and ru .
+ + * os and pt * : the uncertainties for @xmath46os and @xmath47pt are about 50 and 100% , respectively@xcite
. these deviations may originate from the large uncertainties of the measured abundances .
the measurements of their abundances with a high precision are desired .
+ + figure [ fig : sp_solar_mod ] shows the corrected n@xmath24(s)/n@xmath24(p ) ratios .
the open and filled circles are the original n@xmath24(s)/n@xmath24(p ) ratios and the corrected ratios , respectively .
the @xmath1-process abundances of @xmath33ce , @xmath48er , and @xmath35w are taken from calculated results in a stellar @xmath1-process model @xcite .
the @xmath1-process abundance for @xmath37gd is taken from a calculated result based on neutron capture reaction rates measured at a n_tof facility @xcite .
for kr , we take n@xmath24(@xmath43kr)/n@xmath24(@xmath41kr ) instead of n@xmath24(@xmath40kr)/n@xmath24(@xmath41kr ) . to focus our attention to the @xmath2-process , we omit the ratios of mo , ru , and sm .
most corrected ratios are centered around 23 within a factor of 2 ( see fig . [
fig : sp_solar_mod ] ) .
it is noted that the ratio of in decreases after the correction of the @xmath12-process contribution .
@xmath20 in does not shielded against the @xmath8-decay and produced by the @xmath12-process as ce , er , and w and thus the n(@xmath20in)/n(@xmath49 in ) ratio decreases .
this suggests that @xmath49 in may have contaminations of other processes as the @xmath1-process .
the universality of nucleosynthesis processes is an important concept for understanding the stellar nucleosyntheses .
the solar system was formed from interstellar media ( ism ) originated from many stellar nucleosynthesis episodes in the galaxy .
the stellar nucleosynthesis environments such as mass , metallicity and explosion energy are different and hence the abundance distributions of the synthesized nuclides may be different .
however , astronomical observations for very metal - deficient stars have reported the `` universal '' abundance distributions for z @xmath3 56 @xcite , which are in agreement with the abundance distribution of the r - process nuclides in the solar system .
these facts suggest a uniform site and/or uniform conditions for the synthesis of the r - process nuclei .
otsuki et al .
could success to explain the universality of the r - process by a neutrino - energized wind model @xcite .
the p- and s - nuclei were produced in different stellar environments .
thus , the mass distribution of the synthesized nuclei may depend on the astrophysical conditions .
nevertheless , the observed @xmath50(s)/@xmath50(p ) ratios in the solar system are almost constant over a wide range of the atomic number .
this leads to a novel concept , `` the universality of the @xmath2-process '' , that the n(@xmath1)/n(@xmath0 ) ratios of nuclides produced by individual @xmath2-processes are constant over a wide region , respectively .
we would like to stress that the concept of the universality of the @xmath2-process has been derived from only the analysis of the solar abundances independent of nucleosynthesis calculations .
the situation of the @xmath2-process is similar with that of the s - process .
it is known that @xmath4 is roughly constant for the s - process over a large range in nuclear mass , where @xmath6 and @xmath7 are the neutron capture cross - section and the solar abundance @xcite . in this case , one could infer a `` universality of the s - process '' that arose from the conditions being such that a steady neutron capture flow was achieved .
the difference of the @xmath4 for a@xmath2890 and a@xmath390 then pointed to two different s - process components .
smaller deviations from constant @xmath4 signal interesting nuclear phenomena as weak branch .
also , if we could observe abundances of neighboring s - only isotopes in stars of various ages , one would expect their abundance ratio to be inversely proportional to the cross section ratio , just as they are in the solar system .
the universality of the @xmath2-process suggests that one could find roughly constant n(s)/n(p ) ratios with observation of abundances of p- and s - nuclei in stars .
the universality of the @xmath2-process suggests three possible mechanisms as the reason why the scaling appears in the solar system .
first , a single supernova occurred nearby the solar system before the solar system formation and strongly affects the solar system materials .
second , the @xmath2-process occurs only under a specific uniform astrophysical condition in the galaxy and thus the mass distributions of materials produced by individual @xmath2-processes are almost same .
third , the @xmath2-processes under various astrophysical conditions can occur but n(s)/n(p ) ratios are almost constant over a wide region of the atomic number independent of the astrophysical conditions . in the last case , a question why the scaling holds for materials produced by individual @xmath2-processes independent of the astrophysical conditions has been remained .
the universality is , thus , essential for understanding the nature of the @xmath2-process and to constrain the nucleosynthesis site of the @xmath2-process .
the solar abundances provide two empirical scalings , which are a piece of evidence that the most probable origin of the @xmath0-nuclei is @xmath2-process in sne .
a question we have to ask here is whether a standard @xmath2-process model in core - collapse sne can reproduce the scalings . in previous studies @xcite ,
calculated results were presented by a form of f(x)/f@xmath51(x ) , where f(x ) is the ratio of the calculated abundance to the solar abundance for an isotope x , and f@xmath51 is the average of f(x ) .
however , these results have not been presented by a form of n(s)/n(p ) .
we present the calculated n(s)/n(1st p ) ratios under various astrophysical conditions .
a massive star evolves from the main sequence to the core - collapse stage and explodes @xcite .
the solar abundances are adopted as the initial composition in the massive star and is affected by the weak @xmath1-process in evolutional states .
the mass distribution of the seed nuclei affected by the weak @xmath1-process are used as the initial composition at the sn explosions .
most of the reaction rates are taken from a common astrophysical nuclear data library of reaclib @xcite .
figure [ fig : sp_cal ] shows the results in four different models .
they are a standard model ( 1 ) with @xmath52 = 25 @xmath53 , @xmath54 = @xmath55 and @xmath56 = 10@xmath57 erg , a heavier progenitor mass model ( 2 ) with @xmath52 = 40 @xmath53 , @xmath54 = @xmath55 and @xmath56 = 10@xmath57 erg , a metal - deficient model ( 3 ) with @xmath52 = 25 @xmath58 , @xmath54 = 0.05 @xmath55 and @xmath56 = 10@xmath57 erg , and a higher energy explosion model ( 4 ) with @xmath52 = 25 @xmath58 , @xmath54 = @xmath55 and @xmath56 = 20 @xmath59 10@xmath57 erg .
the calculated ratios in the four models are almost constant over a wide region of @xmath54 @xmath3 40 , respectively .
we have presented that there are three possibilities as the principle of the universality of the @xmath2-process .
the calculated results show that even should the @xmath2-process occur under the various astrophysical conditions , the n(s)/n(p ) ratios in individual nucleosynthesis episodes result in an almost constant value over a wide range independent of the sn conditions assumed .
our calculated results have verified the universality of the @xmath2-process . however
, the reason how the scalings hold for the different nucleosynthesis episodes has been an open question .
here we discuss the mechanism of the universality of the @xmath2-process in detailed calculations .
the supernova explosion is a dynamical phenomenon and physical conditions , for example temperature or density , in individual layers of a progenitor are different .
thus , the physical condition of each @xmath2-process layer is a key for understanding the universality of the @xmath2-process .
it is known that the mass regions of the @xmath2-process layers depend on the astrophysical conditions and are shifted to the outer layers as the explosion energy or the mass of the progenitor increases @xcite . in the present calculations , the mass regions in the @xmath2-process models are 1.97 - 2.8 @xmath60 for @xmath61 ergs and 2.87 - 4.63 @xmath53 for 20 @xmath59 @xmath61 ergs , respectively .
the density and temperature of the layers at the peak of the explosion are ( 1.7 - 3.5 ) @xmath59 10@xmath62 @xmath63 and ( 0.13 - 0.95 ) @xmath59 10@xmath64 g/@xmath65 , respectively , for the model ( 1 ) , ( 1.7 - 3.5 ) @xmath59 10@xmath62 @xmath63 and ( 0.12 - 0.90 ) @xmath59 10@xmath64 g/@xmath65 for ( 2 ) , ( 1.6 - 3.5 ) @xmath59 10@xmath62 @xmath63 and ( 0.14 - 2.1 ) @xmath59 10@xmath64 g/@xmath65 for ( 3 ) , and ( 1.7 - 3.5 ) @xmath59 10@xmath62 @xmath63 and ( 0.03 - 0.21 ) @xmath59 10@xmath64 g/@xmath65 for ( 4 ) .
although the mass regions are different , the temperature ranges are approximately equal to each other . the shift also is confirmed in our calculations and contributes to the universality .
the @xmath0-nuclei are dominantly produced by sne , whereas the @xmath1-nuclei are continuously produced by the main @xmath1-process in asymptotic giant branch ( agb ) stars @xcite .
the question we have to ask here is why the abundances of the @xmath0-nuclei is proportional to the @xmath1-nuclei in the solar system .
a weak @xmath1-process in evolutional stages of massive stars gives a hint to answer this question .
it is well known that the weak @xmath1-process produces @xmath1-nuclei of @xmath66 @xmath28 90 from iron seeds .
however , we focus our attention to the weak @xmath1-process effect to the heavy nuclei of @xmath66 @xmath3 90 .
the pre - existing nuclei in the mass region of a @xmath3 90 are also irradiated by successive neutrons in the weak @xmath1-process and thereby their abundance pattern is changed .
we calculate a mass distribution by the weak @xmath1-process in evolutional stages of a massive star with @xmath52 = 25 @xmath58 , @xmath54 = @xmath55 .
the open and filled circles in fig .
[ fig : weak_s ] show the solar abundances as the initial abundances adopted in the massive star and the abundances of the s - processed seed nuclei , respectively .
the abundance distribution of the mass region of @xmath66 @xmath3 90 is locally changed to that of the main @xmath1-process in the agb stars .
for example , there are two @xmath12-process abundance peaks around @xmath66 = 130 and 195 and a @xmath12-process hill around @xmath66=150 ( open circles in fig .
[ fig : weak_s ] ) but these peaks and hill clearly disappear after the weak @xmath1-process ( filled circles in fig .
[ fig : weak_s ] ) . to clarify the effect of the neutron irradiation , we present the abundance ratios of the @xmath1-processed seeds to the solar abundances , n(s)/n@xmath24(s ) , for the twenty - two @xmath1-nuclei in fig .
[ fig : ss ] .
these @xmath1-nuclei are the members of the first scaling .
the ratios are almost constant over the wide region of @xmath66 @xmath3 90 ; namely , the mass distribution of the @xmath1-processed nuclei of @xmath66 @xmath3 90 is approximately equal to the agb @xmath1-process one .
the abundances of the @xmath0-nuclei are proportional to those of the seed nuclei at the sn explosions .
therefore the abundance change by the weak @xmath1-process is the reason for the proportionality between the @xmath0- and @xmath1-nuclei .
the n@xmath24(s)/n@xmath24(p ) ratios in the mass region of @xmath66 @xmath28 90 are lower than the average value of 23 ( see fig . [
fig : sp_solar ] ) .
this is understood by taking into account the enhancement of the s - processed seed in the mass region of @xmath66 @xmath28 90 .
the abundance ratios of n(s)/n@xmath24(s ) of @xmath66 @xmath28 90 are larger than those of @xmath66 @xmath3 90 as shown in fig .
[ fig : ss ] and the abundances of the p - nuclei are proportional to those of the massive star @xmath1-nuclei . therefore , the n@xmath24(s)/n@xmath24(p ) ratios of @xmath66 @xmath28 90 is lower than those of the higher mass region .
it is noted that there are three deviations for ce , er , w in fig .
[ fig : ss ] .
the solar abundances of @xmath33ce , @xmath34er , and @xmath35w are contributed from both the @xmath1- and @xmath12-processes since they are not shielded against @xmath11-decay .
the partial abundances originated from the @xmath12-process disappear after the weak @xmath1-process and hence the n(s)/n(p ) ratios decrease . here
we study effects of two nucleosynthesis paths in the @xmath2-process @xcite .
the first path is direct ( @xmath2,n ) , ( @xmath2,p ) , ( @xmath2,@xmath27 ) reactions from heavier isotopes .
the ( @xmath2,n ) reactions are dominant near the @xmath8 stability line .
the second path is @xmath26 decay after the freezeout of the @xmath2-process .
the seed nuclei are transmuted to light isotopes by successive photodisintegration reactions .
this reaction flow reaches to neutron - deficient region where the ( @xmath2,@xmath27 ) or ( @xmath2,p ) reaction rate is larger than the ( @xmath2,n ) reaction rate .
the neutron - deficient nuclei are subsequently transmuted to lighter elements by downflow of the ( @xmath2,@xmath27 ) and ( @xmath2,p ) reactions .
after the freezeout of the @xmath2-process , the nuclei on the neutron - deficient side decay to the stable isotopes .
woosley and howard ( 1978 ) reported the anti - correlation between the photodisintegration reaction rates and the solar abundances of the @xmath0-nuclei , and the ( @xmath2,n ) reactions have essential role for the synthesis of the @xmath0-nuclei .
the first scaling suggests that the contribution of the first path is dominant .
however , the @xmath67 decay from the neutron - deficient nuclei may break the scaling since the mass distribution on the neutron - deficient side flow depends on nuclear properties such as particle separation energies . here
we study an effect of the second path to the final n(s)/n(p ) ratios .
figure [ fig : paths ] ( a ) shows the n(s)/n(p ) ratios before and after the @xmath8-decay of the neutron - deficient nuclei on the second path .
the ratio before the @xmath8-decay is corresponding to that contributed from only the first path .
figure [ fig : paths ] ( b ) shows percentages of the @xmath0-nuclei populated from only the first path to the final abundance due to both the paths .
these n(s)/n(p ) ratios and percentages are results using a model of @xmath54 = @xmath55 , @xmath52 = 25 @xmath53 , @xmath56 = @xmath61 erg . in the mass region of 58 @xmath28 @xmath54 @xmath28 82 corresponding to 82 @xmath28 @xmath19 , the second path flow is located on the neutron - deficient side from the @xmath8 stability line by mass unit @xmath68 @xmath10 2 - 6 , typically .
thus , the @xmath0-nuclei in this mass region are produced by the two reaction paths .
the calculated results show that the n(s)/n(p ) ratios before the @xmath26 decay are almost constant in this mass region ( filled circles in fig .
[ fig : paths ] ( a ) ) . however , the percentages of the first path scatters over a wide range of 5 - 100% ( see fig .
[ fig : paths ] ( b ) ) .
this indicates that the contributions of the second path to the individual @xmath0-nuclei are different .
in fact the variance of the final n(s)/n(p ) ratios after the @xmath8-decay ( open circles in fig .
[ fig : paths ] ( a ) ) is larger than that before the @xmath26 decay ( filled circles ) .
however , the final n(s)/n(p ) ratios are still almost constant because both the p- and s - nuclei are populated by @xmath8-decay .
therefore the second path does not drastically change the ratios in this mass region .
in the mass region of 40 @xmath28 @xmath54 @xmath28 58 corresponding to 50 @xmath28 n @xmath28 82 , the nucleosynthesis flow mostly proceeds along the @xmath8 stability line . a similar result
was previously reported @xcite . the ratio of ( @xmath2,n ) reaction rates to ( n,@xmath2 ) reaction rates are important for the position of the second path in the nuclear chart . since the neutron separation energy increases with decreasing the proton member , the ( @xmath2,n ) reaction rates in lower mass regions are lower than those in higher mass regions .
therefore , the position of the second path in the lower mass regions is close to the @xmath8-stability line .
the second path proceed mostly on the @xmath8-stability line in the mass region of @xmath19 @xmath28 82 .
the n(s)/n(p ) ratios before and after @xmath8-decay show almost constant values , respectively .
the n(s)/n(p ) ratios before the @xmath69-decay ( filled circles ) are lower than those after the @xmath69-decay ( open circles ) .
this is because the second path locates almost on the @xmath8-stability line but some neutron - deficient isotopes are still produced .
the @xmath8-decay from the neutron - deficient isotopes populates the @xmath1-nuclei , whereas the @xmath0-nuclei are dominantly produced via the first path ( see fig .
[ fig : paths ] ( b ) ) . therefore the @xmath69-decay increases the n(s)/n(p ) ratios . in the mass region of 32 @xmath28 @xmath54 @xmath28 40
, there are only three elements of se , kr and sr .
the ratios of three elements scatter .
the ratio before the @xmath8-decay are approximately equal to that after the @xmath8-decay since the second path locates almost on the @xmath8-stability line and the @xmath0-nuclei are dominantly produced via the first path . in this section ,
we have discussed the contributions of the two nucleosynthesis paths . in most cases in the mass region
40 @xmath28 @xmath54
@xmath28 58 , the @xmath8-decay of the second path nuclei populates only the @xmath1-nuclei in the pairs and hence the n(s)/n(p ) ratios increase after the @xmath8-decay .
we would like to stress that the final ratios in the two mass regions of 40 @xmath28 @xmath54 @xmath28 58 and 58 @xmath28 @xmath54 @xmath28 82 arr almost constant . in this way
, the second path contributes the manifestation of the first scaling that the ratios are almost constant over a wide range of the atomic number .
therefore both the contributions of the first and second paths have essential role for the first scaling .
it should be noted that ( @xmath2,n ) reactions in the two mass region of @xmath19 @xmath28 50 and 50 @xmath28 @xmath19 @xmath28 82 have dominant role and experimental measurements of ( @xmath2,n ) reaction rates on neutron - deficient isotopes ( for example , mohr et al .
2000 ) are of importance for the quantitative calculation of the @xmath2-process and the scalings .
the calculated n(s)/n(p ) ratios in fig .
[ fig : sp_cal ] show some deviations .
we here discuss the reason of two deviations for dy and er in viewpoint of the nuclear data input for the nucleosynthesis flow .
figure [ fig : second_paths ] shows a typical second path flow at the freezeout of the @xmath2-process . in this mass region ,
the neutron - deficient nuclei are mainly transmuted to lighter elements by ( @xmath2,@xmath27 ) reactions .
after the freezeout , the nuclei on the second path decay to the @xmath0- and/or @xmath1-nuclei . in most cases , both the @xmath0- and
@xmath1-nuclei have the contributions from the @xmath8-decay and the final n(s)/n(p ) ratios are not drastically changed from the ratios before the @xmath8-decay as a case of yb .
the n(s)/n(p ) ratios for dy calculated by the various stellar models are systematically larger than those of other @xmath0-nuclei ( see fig . [
fig : sp_cal ] ) .
this can be understood by effects of the @xmath8-decay from the second path .
the @xmath1-nucleus @xmath70dy is strongly populated by the @xmath8-decay but the @xmath0-nucleus @xmath71dy is dominantly produced by the first path ( see fig . [
fig : paths ] ( b ) ) . as the result , the n(@xmath70dy)/n(@xmath71dy )
ratio increases after the @xmath8-decay .
in contrast , the calculated ratios of er are systematically lower than the average value .
the @xmath0-nucleus @xmath72er is populated by the @xmath8-decay but the @xmath1-nucleus @xmath34er is not populated ( see fig .
[ fig : paths ] ( b ) ) . as the result ,
the n(@xmath34er)/n(@xmath72er ) ratio decreases after the @xmath8-decay .
the populations by the @xmath8-decay are determined by the position and width of the second path , which depend strongly on the ( @xmath2,@xmath27 ) reaction rates .
thus , the final n(s)/n(p ) ratios are sensitive to the ( @xmath2,@xmath27 ) reaction rates on neutron - deficient isotopes .
rauscher ( 2006 ) calculated photodisintegration reaction rates with various nuclear models and presented branching points where the ( @xmath2,@xmath27 ) or ( @xmath2,p ) reaction rate competes against the ( @xmath2,n ) reaction rate .
the results showed that the branching points depend on the nuclear models .
the discussion mentioned above shows that the actual ( @xmath2,@xmath27 ) reaction rates on hf and yb isotopes may be lower than the reaction rates used in the present calculations and the second path may be located far from the @xmath8-stability line .
if so , both the p- and s - nuclei of dy and er are populated by the @xmath8-decay as the case of yb and the final n(s)/n(p ) ratios are not largely changed from those before the @xmath8-decay .
thus , the reaction rates based on experimental data with high accuracy are desired .
the ( @xmath2,@xmath27 ) reaction rates have been systematically studied using @xmath27 particle induced reactions @xcite .
a new generation of @xmath2-ray sources of a laser compton scattering ( lcs ) @xmath2-rays in an energy range of mev has been developed @xcite and widely used for studies of nuclear astrophysics @xcite .
the lcs sources are powerful tools for measuring nuclear properties as photo - induced reaction rates because of the sharp energy edge in their energy spectra and the tunable energy
. the measurements of ( @xmath2,@xmath27 ) reaction rates on the @xmath0-nuclei using the lcs @xmath2-rays will contribute to the improvement of @xmath2-process models @xcite .
the first scaling leads to the constant n(s)/n(p ) ratios against the atomic number in individual @xmath2-processes but their absolute values are considered to depend strongly on astrophysical conditions . however , most n(s)/n(p ) ratios calculated in the four different sn models are not only constant but also centered on a specific value of 3 within a factor of 3 ( see figs .
[ fig : sp_cal ] ) .
these results suggest an extended universality that the n(s)/n(p ) ratios produced by individual sn @xmath2-processes are centered on the specific value of 3 .
the universality of the first scaling is important for a study of a frequency of the @xmath1-process episodes in the agb stars and massive stars .
figure [ fig : ism ] shows a schematic chart for the galactic chemical evolution ( gce ) .
the @xmath1-nuclei in the solar system originate mainly from the main @xmath1-process in the agb stars @xcite .
the abundance , n@xmath24(s ) , is proportional to the abundance synthesized by the individual main @xmath1-processes and the frequency of the formation of the agb stars , whereas n@xmath24(p ) is proportional to the @xmath2-process abundance and the frequency of sne .
using the estimated frequency of sne and the calculated abundances of the @xmath0- and @xmath1-nuclei in single nucleosyntheses , we can estimate the frequency of the agb @xmath1-process relative to the @xmath2-process .
the extended universality gives a ratio of @xmath1-nucleus abundance originated from the agb stars to that from the massive stars .
the extended universality shows that the n@xmath73(s)/n@xmath73(p ) ratios produced by a single sn is almost equal to 3 in the mass region of a@xmath390 .
the @xmath1-process abundances in the solar system are contributed from both the agb stars and massive stars associated with sne ; namely , n@xmath24(s ) = n@xmath74(s)+n@xmath73(s ) .
the @xmath0-nuclei originate dominantly from sne ; namely , n@xmath73(p ) = n@xmath24(p ) . using the first scaling of n@xmath24(s)/n@xmath24(p ) = 23 and the extended universality of n@xmath73(s)/n@xmath73(p ) = 3
, we obtain a result that the ratio of n@xmath74(s)/n@xmath73(s ) at the solar system is about 6.7 in the mass region of a @xmath3 90 .
the verification of the extended universality of the @xmath2-process is essential for the discussion mentioned above .
recent progress in spectroscopic studies of metal - poor stars has enabled isotope separation of several heavy elements such as eu by measurements of hyperfine splitting and isotope shifts @xcite .
the observation of @xmath49 in is most promising among all the @xmath0-nuclei in extra - solar objects because this element has the following two advantages .
the first is that this element has only two odd-@xmath19 stable isotopes , a @xmath49 in @xmath0-nucleus , and a @xmath20 in @xmath1-nucleus .
the second is that the @xmath49 in @xmath0-nucleus has an isotopic fraction as large as 4.29 % in the solar system .
in contrast , most @xmath0-isotopes are even - even nuclei and have many isotopes .
there are three other odd-@xmath54 or odd-@xmath19 @xmath0-nuclei ; @xmath20sn , @xmath21la and @xmath22ta , but their isotopic fractions are small ( 0.012 @xmath5 0.34 % ) .
the absorption line of in ( 4511.31@xmath75 ) was detected in the solar photosphere @xcite .
recently , the in @xmath76 line at 4511 @xmath75 in sun - like stars were detected @xcite .
if the indium isotopic fractions could be observed in metal - deficient r - process enhanced stars , it would be valuable evidence for the universality of the @xmath2-process .
the calculated n(@xmath20in)/n(@xmath49 in ) ratios in the four models are 3.1 @xmath5 8.5 corresponding to the @xmath49 in fraction of 11% @xmath5 24% .
the fraction of the @xmath0-isotope in material affected strongly by a single sn should be enhanced , although the enhancement may be small because of the mixing of materials from other layers .
we have presented that the weak @xmath1-process before sn explosions is important for the proportionality between the @xmath0- and @xmath1-nuclei in the solar abundances .
the three large deviations of n@xmath24(s)/n@xmath24(p ) for ce , er , w give a hint to understand the weak @xmath1-process .
the ratios of the three elements are larger by an order of magnitude than those of the other elements ( see fig .
[ fig : sp_solar ] ) .
these deviation originate from enhancements of heavier isotopes,@xmath33ce , @xmath34er and @xmath35w , in their pairs .
these three isotopes are not shielded against the @xmath8 decay after the freezeout of the @xmath12-process and the contributions of the @xmath12-process for @xmath34er and @xmath35w are larger than those of the @xmath1-process @xcite .
the solar abundances can be expressed by @xmath77 the abundances of typical @xmath1-nuclei are expressed by n@xmath24(z , n ) = n@xmath78(z , n ) , but the abundances of the three isotopes are expressed by n@xmath24(z , n ) = n@xmath78(z , n)+n@xmath79(z , n ) .
the first scaling indicates that the abundance of the @xmath0-nucleus , n@xmath80(z , n-2 ) or n@xmath80(z , n-4 ) , is proportional to the heavier isotope abundance of n@xmath78(z , n)+n@xmath79(z , n ) at sn explosions ; namely n@xmath80(z , n-2 ) = @xmath27@xmath81(n@xmath78(z , n)+n@xmath79(z , n ) ) , where @xmath27 is about 1/23 at the solar system .
the abundances of the @xmath0-nuclei can be expressed by n@xmath80(z , n-2 ) = n@xmath82(z , n-2 ) + n@xmath83(z , n-2 ) , where n@xmath82(z , n-2 ) = @xmath27@xmath81n@xmath79(z , n ) , and n@xmath83(z , n-2 ) = @xmath27@xmath81n@xmath78(z , n ) .
the weak @xmath1-process occurs in the massive stars before the sne and the mass distribution of pre - existing heavy elements is changed to that of the @xmath1-process ( see fig .
[ fig : ss ] ) .
the abundance n@xmath78(z , n)+n@xmath79(z , n ) is changed to n@xmath78(z , n ) and the @xmath0-nucleus abundance n@xmath82(z , n-2 ) originated from n@xmath79(z , n ) vanishes ; for example , n@xmath80(@xmath72er ) = 1/23@xmath81n@xmath78(@xmath34er ) . because the solar abundances of @xmath33ce , @xmath84er , and @xmath35w are contributed from both the @xmath12- and @xmath1-processes , the n@xmath24(s)/n@xmath24(p ) ratios in the three elements are expressed by ( n@xmath78(z , n)+n@xmath79(z , n))/n@xmath80(z , n-2 ) , which are larger than n@xmath78(z , n)/n@xmath80(z , n-2 ) of neighboring @xmath0-nuclei by a factor of ( n@xmath78(z , n)+n@xmath79(z , n))/n@xmath78(z , n ) . in this way
, the three large enhancements in n(s)/n(p ) ratios are observed in the solar abundances .
in contrast , if the weak @xmath1-process does not occur before the @xmath2-process , the abundances of the @xmath0-nuclei in the three elements are also proportional to those of heavier isotopes .
therefore , the three deviations for ce , er and w are a piece of evidence that the weak @xmath1-process in the massive stars occurs before the @xmath2-process . a question why the abundance of @xmath72er is an order of magnitude larger than those of neighboring @xmath0-nuclei as shown in fig .
[ fig : abundances ] has been known about fifty years ago @xcite .
takahashi and yokoi ( 1983 ) suggested that a stable isotope @xmath85dy becomes to unstable to @xmath86-decay in a typical @xmath1-process temperature because of an atomic effect of ionized ions and @xmath72er may be synthesized by a weak branch of the @xmath1-process , @xmath87 the @xmath8-decay rate of @xmath85dy depends on thermodynamic condition of the @xmath1-process , and hence the abundance ratio of @xmath7(@xmath34er)/@xmath7(@xmath72er ) is sensitive to the temperature .
however , the evaluation of @xmath7(@xmath34er)/@xmath7(@xmath72er ) seems to be difficult , because @xmath72er is produced by the @xmath1- and @xmath2-processes and @xmath34er is produced by the @xmath1- and @xmath12-processes as shown in fig .
[ fig : er_ho ] .
we here propose a new method to evaluate the @xmath1-process abundances of @xmath88er using the empirical scalings ; n(s)/n(p ) = 23 and n(1st p)/n(2nd p ) = 1 .
because the second @xmath0-nucleus @xmath89er is dominantly synthesized by the @xmath2-process , we can evaluate the @xmath0-process abundance of @xmath72er and the seed @xmath1-nucleus abundance of @xmath34er using the scalings .
the evaluated ratio , n@xmath90(@xmath34er)/ n@xmath90(@xmath72er ) , is 2.2 at the solar system
. arlandini _
et al_. ( 1999 ) calculated the component of all @xmath1-nuclei using a stellar model for low - mass agb stars with @xmath91c burning .
the calculated ratio n@xmath90(@xmath34er)/ n@xmath90(@xmath72er ) is 3.8 , which is consistent with the present evaluated ratio of 2.2 within a factor of 2 .
therefore , the empirical scalings support the @xmath1-process calculations by arlandini et al .
our discussion shows that the two scalings are a piece of evidence that twenty - seven @xmath0-nuclei are dominantly produced by the sn @xmath2-process .
another model discussed in the literature @xcite is @xmath2-process in the outer layers of exploding white dwarf stars .
those outer layers should have been enriched in s - process nuclei as the the ashes of he shell burning settled on them in the star s agb phase .
if the burning front passing through them during the explosion tends to give the right set of conditions , a @xmath2-process might occur . though the p - isotope yields from exploding white dwarf stars are highly uncertain , they can be enormous and may be significant contributors to the solar system s supply of the p - isotopes .
the question whether the @xmath2-process in the white dwarf stars can satisfy the scaling laws has been remained .
there are thirty - five @xmath0-nuclei and the origin of the other eight @xmath0-nuclei has remained as an open question .
the other @xmath0-nuclei are @xmath44mo , @xmath45ru , @xmath16sm , @xmath20sn , @xmath21la , and @xmath22ta .
the first scaling at the solar abundances shows large deviation for mo and ru isotopes ( see fig .
[ fig : sp_solar ] ) .
the isotope abundances of these nuclei are about ten times larger than those of typical @xmath0-nuclei .
it has been well known that the @xmath2-process calculations underproduce the abundances of these isotopes , @xmath44mo and @xmath45ru .
recently , as the origin of these isotopes , nucleosynthesis in early neutrino wind in core - collapse sne ( @xmath9@xmath0-process ) have been studied @xcite .
burbidge et al .
( 1957 ) pointed out that the mass distribution of the @xmath0-nuclei show two peaks near two neutron magic numbers of n=50 , 82 and the @xmath0-nuclei may be synthesized by ( p,@xmath2 ) and ( @xmath2,n ) reactions .
the abundance pattern around mo region is similar to that of sm . as shown in fig .
[ fig : abundances ] , the abundances of @xmath44mo is larger than those of lighter @xmath0-isotopes , @xmath92sr and @xmath41kr , and the abundance of @xmath16sm is also larger than those of lighter @xmath0-nuclei .
in contrast , the abundances of the seed @xmath1-nuclei , @xmath93mo , @xmath94ru and @xmath17sm , are consistent with those of the same mass region : namely , their abundances are lower than those of @xmath1-nuclei in the lighter mass regions beyond the neutron magic numbers .
this fact suggests a possibility that these five @xmath0-nuclei , @xmath44mo , @xmath45ru and @xmath16sm might be synthesized mainly from @xmath1-nuclei located in lighter mass regions beyond the neutron magic number by particle - induced reactions such as ( p,@xmath2),(@xmath27,@xmath2 ) or ( n,@xmath2 ) reaction .
recently , the proton capture experiments have been carried out up to @xmath66 @xmath5 120 @xcite .
however , the reaction rates in a heavy mass region of @xmath66 @xmath5 140 have not been measured .
these rates are of importance for understanding the origin of the p - nuclei near both the neutron magic numbers of @xmath19 = 50 and 82 .
recent progress of meteorite science give crucial hints about the origin of mo and sm .
yin et al .
( 2002 ) measured isotopic fractions of mo in primitive meteorites with high accuracy and isotopic fractions of the @xmath0-nucleus @xmath95mo and the @xmath12-nucleus @xmath94mo show different anomalies in comparison with the solar abundances .
yin et al . concluded that there are at least three patterns of the isotopic anomalies of the mo isotopes .
this result suggests that the nucleosynthesis site of @xmath95mo is not correlated with the @xmath12-process site .
isotopic anomalies of sm in primitive meteorites were also measured and the results showed that nucleosynthesis sites supplying the p- and r - isotopes were disconnected or only weakly connected @xcite .
it is of importance measurements of the isotope abundances of both the elements , sm and mo , in individual meteoritic samples , which can answer the question whether the nucleosynthesis sites of sm and mo are same or not . as the origin of @xmath20sn , the @xmath2- , @xmath1- and @xmath12-processes were proposed .
@xmath20sn may be synthesized through a @xmath8 unstable isomer in @xmath20 in after the freezeout of the @xmath12-process and by a weak branch of the @xmath1-process by @xmath96 however , previous studies reported that @xmath20sn can not be produced enough by the @xmath1-process @xcite .
mmeth et al . suggested that the population of the isomer in @xmath49cd should follow by thermal - equilibrium conditions in the s - process environment @xmath97=12 - 30 kev .
in such the case , the population ratio of the isomer depend only on the temperature of the environment .
here we would like to point out that recent studies of the @xmath1-process suggest a possibility that the s - process may occur at the low temperature @xmath97 @xmath28 10 kev .
in such the case , the population of the excited states in @xmath49cd dose not follow by the thermal - equilibrium conditions and thus the @xmath20sn abundance depends on a partial neutron capture reaction cross - section to the isomer of @xmath49cd . the neutron capture reaction cross - section to the @xmath49cd isomer
has never measured with high accuracy at any energy and the measurement is desired .
two isotopes , @xmath21la and @xmath22ta , are classified to a group of odd - odd nuclei .
these two isotopes have the following unique features .
first their isotope abundances are small , 0.090% ( for @xmath21la ) and 0.012% ( @xmath22ta ) .
second they are shielded against both of @xmath67-decay and @xmath86-decay by stable isobars and thus they are produced only by direct nuclear reactions , for example ( @xmath2,n ) reactions from heavy isotopes or ( @xmath98,e@xmath99 ) reactions from isobars .
@xmath21la : as the origin of @xmath21la , the cosmic - ray process @xcite and the @xmath9-process @xcite were proposed .
a result that about 90% in the solar abundance of @xmath21la can be explained by the @xmath9 process was reported @xcite and the contribution of a key charged current reaction to @xmath21la was quantitatively evaluated @xcite .
a study of primitive meteorites provides a hint about the origin of @xmath21la .
shen and lee ( 2003 ) reported the isotope abundance anomalies of several elements as ti and la in ca - al - rich inclusions , which is considered to originated from the early solar nebula .
the abundance ratios of @xmath21la/@xmath100la is correlated with that of @xmath101ti/@xmath102ti @xcite .
the reason why @xmath21la is correlated with @xmath101ti has been an open question .
@xmath22ta : the origin of @xmath22ta is one of hot topics in the nuclear astrophysics since @xmath22ta has a unique feature that the meta - stable state in @xmath22ta is the only isomer ( half - life of @xmath103 10@xmath104 yr ) existing in the solar system , while the ground state is unstable state against @xmath8 -decay ( half - life of 8 hr ) . as the origin of @xmath22ta , the @xmath1-process @xcite , and the @xmath2-process @xcite , the @xmath9-process @xcite and the cosmic - ray process @xcite were suggested .
a recent @xmath9-process calculation overproduces the solar abundance of @xmath22ta relative to @xmath105o , which based on a new experimental result using a @xmath22hf(@xmath106he , t)@xmath22ta reaction to evaluate the charged current reaction rate @xcite . in explosive nucleosynthesis such as the @xmath2- and @xmath9-processes , the population ratio of the meta - stable isomer to the @xmath8-unstable ground state at freezeout of a nucleosynthesis process is an important factor . the transition probability from the isomer to the ground state
was measured using real photons @xcite and the population ratio by a ( @xmath2,n ) reaction was also measured @xcite . to estimate exactly the residual abundance of @xmath22ta ,
a time - dependent calculation is required . however , for the estimation of the isomer population , a thermal equilibrium approximation @xcite are widely used , wherein the population ratio of the isomer to the ground state can be calculated by the maxwell - boltzmann population at critical temperature @xmath107 that is the lowest temperature of the thermal equilibrium region .
it should be noted that exited states located above the isomer are also populated in high temperature environments and may decay to the isomer .
on other words , the excited states above the isomer also contribute the residual abundance of @xmath22ta .
mohr et al . ( 2007 ) calculated the temperature of @xmath107 based on measured transition probability @xcite and concluded @xmath107 is 40.4 kev which leads to @xmath108 of 0.35 .
here we discussed the population of the isomer under the thermal equilibrium approximation .
a population of an excited state is followed to the maxwell - boltzmann population presented by @xmath109 in a temperature range of @xmath110 = 0.1 - 1.0 , all excited states lower than a few hundred kev energy are populated ( see fig .
[ fig:180ta ] ) .
after the freezeout , the each excited state decays to the ground state or the isomer .
the final population ratio of the isomer to the total is presented by @xmath111 where @xmath112 and @xmath113 indicate levels decaying to the isomer and the ground state , respectively .
we calculate the contribution of the excited levels presented in fig .
[ fig:180ta ] , since the contribution of the higher levels is lower than 0.1% . in previous calculations , the residual ratio of 0.3 - 0.5
have been used @xcite .
however , the present calculated ratios are 0.06 , 0.13 , 0.20 , and 0.26 for the assumed critical temperature of @xmath114 = 0.20 , 0.25 , 0.30 , and 0.35 , respectively .
this result shows that the residual abundance of @xmath22ta should decrease against the previous estimations if the thermal equilibrium is a good approximation .
we present two empirical scaling laws concerning the @xmath0- and @xmath1-nuclei with the same atomic number in the solar system abundances . the first @xmath0-nucleus and the second @xmath0-nucleus are lighter than the @xmath1-nucleus by two and four neutrons , respectively
the first scaling is the correlation between the @xmath0- and @xmath1-nuclei .
the corrected n@xmath23/n@xmath24(1st p ) and n@xmath23/n@xmath24(2nd p ) ratios are almost constant over a wide region of the atomic number and are mostly centered around 23 within a factor of 2 .
these scalings are a piece of evidence that the @xmath2-process is the most probable origin of the twenty - seven @xmath0-nuclei .
the scalings in the solar system lead to a novel concept `` the universality of the @xmath2-process '' that the abundance ratios n(s)/n(p ) produced by individual sn @xmath2-processes are almost constant over a wide region of the atomic number .
the @xmath2-process calculations of core - collapse sne under various astrophysical conditions support the universality of the @xmath2-process .
three mechanisms contribute to the universality .
the first mechanism is the weak @xmath1-process .
it is well known that light elements of @xmath66 @xmath28 90 are produced by successive neutron capture reactions from iron seeds in the weak @xmath1-process .
however , pre - exiting heavy elements of @xmath66 @xmath3 90 are also irradiated by neutrons and their mass distribution is changed locally to that of the main @xmath1-process in asymptotic giant branch stars . therefore , the mass distribution patterns of the seed nuclei at the supernova explosions are almost identical .
the second mechanism is the shifts of the mass regions of the @xmath2-process layers to keep their peak temperature .
the @xmath2-process layers are shifted to outer layers as the explosion energy or mass increases . inside the @xmath2-process layers ,
the seed nuclei are completely destroyed by photodisintegration reactions , whereas the @xmath0-nuclei outside the @xmath2-process layers are not synthesized because of their low temperature .
therefore , the physical conditions of the @xmath2-process layers are almost identical .
the third mechanism is independence from two nucleosynthesis flows .
the @xmath0-nuclei are produced by the two flows : direct ( @xmath2,n ) reactions from heavier isotopes , and @xmath8-decay after downflow by ( @xmath2,p ) and ( @xmath2,@xmath27 ) reactions from heavier elements which may break the scalings .
we calculated the contributions of both the flows for each @xmath0-nucleus .
the ( @xmath2,n ) reactions are dominant in the light mass region of @xmath19 @xmath28 82 .
the @xmath8-decay after the downflow contributes to the @xmath0-nucleus abundances in the heavy mass region of @xmath19 @xmath3 82 but does not drastically change the abundance ratios because both the p- and s - nuclei are populated by @xmath8-decay in most cases . in this way ,
the scalings hold for individual @xmath2-processes .
in addition , we find that large deviations of n@xmath24(s)/n@xmath24(p ) for ce , er , and w in the solar system are a piece of evidence that the weak @xmath1-process actually occurred before sne .
the present calculated results suggest an extended universality that the n(s)/n(p ) ratios produced in individual sne may be constant around a specific value of 3 independent of the stellar conditions . with the extended universality of n@xmath73(s)/n@xmath73(p )
= 3 and the first scaling of n@xmath24(s)/n@xmath24(p ) = 23 , we estimate that the ratio of the @xmath1-process abundance contribution from the agb stars to the massive stars associated with sne is about 6.7 for the @xmath1-nuclei of @xmath66 @xmath3 90 in the solar system .
an astronomical observation with indium isotope separation is useful for the verification of the extended universality
. this observation can give a crucial impact on the identification of the @xmath2-process site and gce for @xmath0- and @xmath1-nuclei .
aoki , k. , hosono , k. , hadame , t. , munenaga , h. , kinoshita , k. , toda , m. , amano , s. , miyamoto , s. , mochizuki , t. , aoki , m. , and li , d. 2004 , nucl .
inst . method .
a 516 , 228 andreasen , r. & sharma , m. 2006 , science , 314 , 806 aoki , w. , ryan , s. g. , iwamoto , n. , beers , t. c. , norris , j. e. , ando , h. , kajino , t. , mathews , g. j. , fujimoto , m. 2003 , apj , 592 , l67 arlandini , c. kppeler , f. , wisshak , k. , gallino , r. , lugaro , m. , busso , m. , straniero , o. 1999 , apj , 525 , 886 arnould , m. 1976 , a&a , 46 , 117 arnould , m. & goriely , s. 2003 , phys .
384 , 1 audouze , j. 1970 , a&a , 8 , 436 belic , d , arlandini , c. , besserer , j. , boer , j. de , carroll , j. j. , enders , j. , hartmann , t. , kappeler , f. , kaiser , h. , kneissl , u. , loewe , m. , maier , h. j. , maser , h. , mohr , p. , neumann - cosel , p. von , nord , a. , pitz , h. h. , richter , a. , schumann , m. , volz , s. , & zilges , a. 1999 , phys .
83 , 5242 bievre , p. de . &
taylor , p.d.p .
1993 , int . j. mass spectrom .
ion proc . , 123 , 149 burbidge , e. m. , burbidge , g. r. , fowler , w. a. , & hoyle , f. 1957 , rev .
29 , 548 busso , m. , gallino , r. & wasserburg , c.j . 1999 ,
araa , 37 , 239 byelikov , a. et al . , 2007 ,
98 , 082501 cayrel , r. , _ et al .
_ 2001 , nature , 409 , 91 fujimoto , s. , hashimoto , m. , koike , o. , arai , k. , & matsuba , r. 2003 , apj , 585 , 418 frhlich , c. et al .
2006 , phys .
96 , 142502 gallino , r. et al . , 1998 ,
apj , 497 , 388 goko , s. , utsunomiya , h. , goriely , s. , makinaga , a. , kaihori , t. , hohara , s. , akimune , h. , yamagata , t. , lui , y .- w . ,
toyokawa , h. , koning , a. j. , hilaire , s. 2006 , phys .
96 , 192501 goldberg , l. , muller , e.a . &
aller , l.h .
1960 , apjs , 5 , 1 gonzalez , g. 2006 , mnras , 371 , 781 goriely , s. , arnould , m. , borzov , i. , rayet , m. 2001 , a&a , 375 , l35 gyrky , gy . ,
kiss , g.g .
, elekes , z. , flp , zs . , somorjai , e. , palumbo , a. , g"orres , j. , lee , h. y. , rapp , w. , wiescher , m. , " ozkan , n. , g"uray , r.t . ,
efe , g. , rauscher , t. 2006 , phys .
c 74 , 025805 hayakawa , t. , iwamoto , n. , shizuma , t. , kajino , t. , umeda , h. & nomoto , k. 2004 , phys .
93 , 161102 hayakawa , t. , iwamoto , n. , shizuma , t. , kajino , t. , umeda , h. , 2006 , apjl , 648 , l47 hayakawa , t. , et al .
2006 , phys .
c 74 , 065802 heger , a. , _ et al . _ , 2005 , phys .
b , 606 , 258 hoffman , r.d .
, woosley , s.e .
, fuller , g.m . ,
meyer , b.s .
, 1996 , apj , 460 , 478 .
honda , s et al .
2004 , apj , 607 , 474 howard , w.m . , meyer , b.s . & woosley , s.e .
1991 , apj , 373 , l5 iwamoto , n. , umeda , h. & nomoto , k. 2004 , international symposium on origin of matter and evolution of galaxies , world scientific , p.493 kppeler , f. , gallino , r. , busoo , m. , picchio , g. , & raiteri , c.m .
1990 , apj , 354 , 630 kiss , g.g . ,
fulop , zs . , gyurky , gy . , mate , z. , somorjai , e. , galaviz , d. , kretschmer , a. , sonnabend , k. , zilges , a. 2006 , eur .
j. a27 , 197 kusakabe , m. , iwamoto , n. & nomoto , k. 2005 , nucl .
a , 758 , 459c lambert , d.l . , mallia , e.a . ,
& warner , b. 1969 , mon . not .
, 142 , 71 lambert , d.l . &
prieto , c.a .
, 2002 , mon .
not . r. astron .
335 , 325 litvinenko , v.n . , _
et al . _ , 1997 ,
78 , 4569 mohr , p. , vogt , k. , babilon , m. , enders , j. , hartmann , t. , hutter , c. , rauscher , t. , volz , s. , zilges , a. 2000 , phys .
488b , 127 mohr , p. , shizuma , t. , ueda , h. , goko , s. , makinaga , a. , hara , k. y. , hayakawa , t. , lui , y .- w .
, ohgaki , h. , utsunomiya , h. 2004 , phys .
c 69 , 032801(r ) mohr , p. kppeler , f. , gallino , r. , 2007 , phys .
c 75 , 012802(r ) nmeth zs , kppeler , f. , theis , c. , belgya , t. , yates , s. w. 1994 , apj , 426 , 357 nomoto , k. , et al .
2004 , in _ stellar collapse _ , ed .
fryer ( astrophysics and space science : kluwer ) , 277 ( astro - ph/0308136 ) n_tof collaboration , 2004 , phys .
93 , 161103 ohgaki , h. , _ et al .
_ , 1991 , ieee trans .
38 , 386 otsuki , k. , mathews , k.j . , & kajino , t. 2003 , new astrom . 8 , 767 zkan , n. , efe , g. , gray , r.t . , palumbo , a. , grres , j.m lee , h.y . ,
lamm , l.o . ,
rapp , w. , stech , e. , wiescher , m. , gyrky , gy.m flp , zs.m somorjai , e. 2007 , phys .
c 95 , 025801 zkan , n. , murphy , a.st.j . , boyd , r.n . , cole , a.l . ,
famiano , m. , guray , r.t . , howard , m. , sahin , l. , zach , j.j .
2002 , nucl .
phys . a 710 , 469 prantzos , n. , hashimoto , m. , rayet , m. , & arnould , m. 1990 , a&a , 238 , 455 pruet , j. , hoffman , r.d . , woosley , s.e . , janka , h .-
t . , buras , r. 2006 , apj , 644.1028 rauscher , t. , heger , a. , hoffman , r.d . ,
& woosley , s.e .
2002 , apj , 576 , 323 rauscher , t. 2006 , phys . rev .
c , 73 , 015804 rapp , w. , grres , j. , wiescher , m. , schatz , h. , & kppeler , f. 2006 , apj , 474 , 653 rayet , m. , prantzos , m. & arnould , m. 1990 , a&a , 227 , 271 rayet , m. , arnould , m. , hashimoto , m. , prantzos , n. , & nomoto , k. 1995 , a&a , 298 , 517 saitoh , t.r . , hashimoto , n. , sletten , g. , bark , r.a . , tormanen , s. , bergstrom , m. , furuno , k. , furutaka , k. , hagemann , g.b . , hayakawa , t. , komatsubara , t. , maj , a. , mitarai , s. , oshima , m. , sampson , j. , shizuma , t. , & varmette , p.g . 1999 ,
a. 660 , 121 spyrou , a .
becker , h .- w . ,
lagoyannis , a. , harissopulos , s. and rolfs , c. 2007 , phys . rev .
c. 76 , 015802 schatz , h. , et al .
, 1998 , phys .
294 , 167 schatz .
2001 , phys .
86 , 3471 seeger , p.a . ,
fowler , w.a . ,
clayton , d.d .
1965 , apj , 11 , 121 shen , j.j . &
lee , t. 2003 , apjl , 596 , l109 shizuma , t. _ et al .
c 72 , 025808 ( 2005 ) .
sneden , c. , cowan , j. j. , burris , d. l. , & truran , j. w. 1998 , apj , 496 , 235 sneden , c. , cowan , j. j. , ivans , i. i. , fuller , g. m. , burles , s. , beers , t. c. , & lawler , j. e .. 2000 , apj , 533 , l139 sneden , c. , cowan , j. j. , lawler , j. e. , burles , s. , beers , t. c. , & fuller , g. m. 2002 , apj , 566 , l25 takahashi , k. , & yokoi , k. , 1983 , nucl .
phys . a , 404 , 578 thielemann , f. , 2000 , http://nucastro.org/reaclib.html utsunomiya , h. , akimune , h. , goko , s. , ohta , m. , ueda , h. , yamagata , t. , yamasaki , k. , ohgaki , h. , toyokawa , h. , lui , y .- w . , hayakawa , t. , shizuma , t. , khan , e. & goriely , s. 2003 , phys . rev .
c 67 , 015807 utsunomiya , h. , mohr , p. , zilges , a. , rayet , m. 2006 , nucl .
a777 , 459 wanajo , s. 2006 , apj , 647 , 1323 woosley , s.e . & howard , w.m .
1978 , apjs , 36 , 285 woosley , s. e. , hartmann , d. h. , hoffman , r. d. & haxton , w .
c. 1990 , apj , 356 , 272 yin , q. , jacobsen , s.b .
, & yamashita , k. 2002 , nature , 415 , 881 yokoi , k. & takahashi , k. 1983 , nature , 305 , 198 * table.1 * list of the p - process isotopes .
the isotope abundances are taken from bivre and taylor ( 1993 ) .
( a ) the contribution of the r - process .
( b ) the contribution of the s - process .
( c ) the effect of the neutron magic number .
+ -stability line are mainly synthesized by the @xmath1- and @xmath12-processes . in contrast , the @xmath0-nuclei , for example @xmath115ba and @xmath116ce , are synthesized by the @xmath2-process .
there are pairs of a p - nucleus and an s - nucleus that is heavier than the p - nucleus by two neutrons . ]
-nucleus to @xmath0-nucleus with the same atomic number .
the circles and triangles are n@xmath24(s)/n@xmath24(1st p ) and n@xmath24(s)/n@xmath24(2nd p ) , respectively , where n@xmath24 is the solar abundance .
the first and second @xmath0-nuclei are lighter than the @xmath1-nucleus by two and four neutrons , respectively .
the dotted line is the constant value of n@xmath24(s)/n@xmath24(p ) = 23 .
the dot - dashed lines are 11.5 and 46 .
the observed ratios are almost constant within a factor of 2 . ] .
the filled circles mean the observed ratios in the solar system .
the open circles stand for the calculated ratios in the model with @xmath52 = 25 @xmath53 , @xmath54 = @xmath55 and @xmath56 = 10@xmath57 erg . ] - and @xmath1-nuclei , which are the members of the scalings .
the dot - dashed lines are the neutron magic numbers of n=50 and 82 .
the dot lines mean constant values around the centers of the @xmath0- and @xmath1-nucleus abundances in each region . ]
kr are pure @xmath1-nuclei since they are shielded against the @xmath8-decay after the @xmath12-process .
@xmath40kr is synthesized by a weak branch of the @xmath1-process , whereas @xmath43kr is located on the main flow of the @xmath1-process . ]
-nucleus to @xmath0-nucleus with the same atomic number , n@xmath24(s)/n@xmath24(p ) , where n@xmath24 is solar abundance .
the open circles and open triangles are modified n@xmath24(s)/n@xmath24(1st p ) and n@xmath24(s)/n@xmath24(2nd p ) , respectively .
the filled circles and filled triangles are original n@xmath24(s)/n@xmath24(1st p ) and n@xmath24(s)/n@xmath24(2nd p ) , respectively . in most cases ,
the modified and original ratios are same .
the dotted line is the average of these ratios n@xmath24(s)/n@xmath24(p ) = 23 .
the dot - dashed lines are 11.5 and 46 .
the observed ratios are almost constant within a factor of 2 .
, title="fig : " ] = 25 @xmath53 , @xmath54 = @xmath55 and @xmath56 = 10@xmath57 erg .
the filled triangles are those in the model with @xmath52 = 40 @xmath53 , @xmath54 = @xmath55 and @xmath56 = 10@xmath57 erg .
the open triangles are those in the model with @xmath52 = 25 @xmath58 , @xmath54 = @xmath55 and @xmath56 = 20 @xmath59 10@xmath57 erg .
the open squares are those in the model with @xmath52 = 25 @xmath58 , @xmath54 = 0.05 @xmath55 and @xmath56 = 10@xmath57 erg .
the dashed line is n(s)/n(p ) = 3 .
the dot - dashed lines are 1 and 9 .
the ratios in all the models are centered around 3 within a factor of 3 over a wide range of atomic number for @xmath54 @xmath3 40.,title="fig : " ] -nuclei are synthesized from the @xmath1-processed seed by the @xmath2-process at the supernova explosion .
the @xmath1-process depends strongly on the metallicity .
the n(s)/n(p ) ratio is proportional to the frequency of the s - process events and is time dependent . ]
decay of the neutron - deficient nuclei in the second path , respectively .
the dashed and dot - dashed line is 3 and 0.7 , respectively .
the lower panel presents the percentage of the first paths .
these are calculated in a model with @xmath52 = 25 @xmath53 , @xmath54 = @xmath55 and e = 10@xmath57 erg.,title="fig : " ] -nuclei are synthesized by two nuclear reactions paths .
the first is a direct photodisintegration reactions such as ( @xmath2,n ) reaction from heavier isotopes and the second is a flow located in neutron deficient side and subsequent @xmath8-decay .
arrows show a typical second path flow . ]
-nucleus @xmath89er is dominantly synthesized by the @xmath2-process . in contrast , the first @xmath0-nucleus @xmath72 is synthesized by the @xmath2- and @xmath1-processes . in high temperature environments of t @xmath5 10@xmath117 k , @xmath85dy becomes unstable by atomic effects and decays to @xmath85ho .
@xmath34er is synthesized by the @xmath1- and @xmath12-processes since it is not shielded against the @xmath8-decay after the @xmath12-process . ]
@xmath22ta is excited in a high temperature environment of @xmath118 @xmath3 10@xmath117 k. the ground state and excited states are populated from with each other via intermediate states lying higher excited states .
after a supernova explosion , the temperature of a nucleosynthesis site suddenly decreases .
the excited states in @xmath22ta decay to the ground state or the isomer .
the excitation energies are taken from ref . | analyzing the solar system abundances , we have found two empirical abundance scaling laws concerning the @xmath0- and @xmath1-nuclei with the same atomic number .
the first scaling is s / p ratios are almost constant over a wide range of the atomic number , where the @xmath0-nculei are lighter than the @xmath1-nuclei by two or four neutrons .
the second scaling is p / p ratios are almost constant , where the second @xmath0-nuclei are lighter than the first @xmath0-nucleus by two neutrons .
these scalings are a piece of evidence that most @xmath0-nuclei are dominantly synthesized by the @xmath2-process in supernova explosions .
the scalings lead to a novel concept of `` universality of @xmath2-process '' that the s / p and p / p ratios of nuclei produced by individual @xmath2-processes are almost constant , respectively .
we have calculated the ratios by @xmath2-process based on core - collapse supernova explosion models under various astrophysical conditions and found that the scalings hold for materials produced by individual @xmath2-processes independent of the astrophysical conditions assumed .
the universality originates from three mechanisms : the shifts of the @xmath2-process layers to keep their peak temperature , the weak @xmath1-process in pre - supernovae , and the independence of the s / p ratios of the nuclear reactions .
the results further suggest an extended universality that the s / p ratios in the @xmath2-process layers are not only constant but also centered on a specific value of 3 . with this specific value and the first scaling
, we estimate that the ratios of @xmath1-process abundance contributions from the agb stars to the massive stars are almost 6.7 for the @xmath1-nuclei of a @xmath3 90 .
we find that large enhancements of s / p ratios for ce , er , and w are a piece of evidence that the weak @xmath1-process actually occurred before sne . |
let @xmath0 denote the unit circle , @xmath1 denote the unit disk and @xmath2 denote the extended complex plane .
we also use @xmath3 for the exterior of the unit disk and @xmath4 for the sup norm over the set @xmath5 .
first , we recall a bernstein - type inequality proved by borwein and erdlyi in @xcite ( and in a special case , by li , mohapatra and rodriguez in @xcite ) .
we rephrase their inequality using potential theory ( namely , normal derivatives of green s functions ) and for the necessary concepts , we refer to @xcite and @xcite .
then we present one of our main tools , the `` open - up '' step in proposition [ prop : open - up ] , similar step was also discussed by widom , see @xcite , p. 205206 and
lemma 11.1 . this way we switch from polynomials and jordan arcs to rational functions and jordan curves .
then we use two conformal mappings , @xmath6 and @xmath7 to map the interior of the jordan domain onto the unit disk and to map the exterior of the domain onto the exterior of the unit disk respectively .
we transform our rational function with @xmath6 and `` construct '' a similar rational function ( approximate with another , suitable rational function ) so that the borwein - erdlyi inequality can be applied . our main theorem is the following .
[ thm : main ] let @xmath5 be an analytic jordan arc , @xmath8 not an endpoint .
denote the two normals to @xmath5 at @xmath9 by @xmath10 and @xmath11 .
then for any polynomial @xmath12 of degree @xmath13 we have @xmath14 where @xmath15 depends on @xmath9 and @xmath5 only and tends to @xmath16 as @xmath17 . remark .
this theorem was formulated as a conjecture in @xcite on page 225 .
theorem [ thm : main ] is asymptotically sharp as the following theorem shows .
[ thm : sharpness ] let @xmath5 be a finite union of disjoint , @xmath18 smooth jordan arcs and @xmath8 is a fixed point which is not an endpoint .
we denote the two normals to @xmath5 at @xmath9 by @xmath10 and @xmath11 .
then there exists a sequence of polynomials @xmath12 with @xmath19 such that @xmath20
the following theorem was proved in @xcite ( see also @xcite , p. 324
, theorem 7.1.7 ) , with slightly different notations .
if @xmath21 is a rational function , then @xmath22 denotes the maximum of the degrees of the numerator and denominator of @xmath21 ( where we assume that the numerator and the denominator have no common factors ) .
[ borwein - erdlyi ] let @xmath23 and let @xmath24 and @xmath25 .
if @xmath26 is a polynomial with @xmath27 and @xmath28 is a rational function , then @xmath29 if all the poles of @xmath21 are inside or outside of @xmath30 , then this result was improved in @xcite , theorem 2 and corollary 2 on page 525 using different approach .
we need to relax the condition on the degree of the numerator and the denominator .
if we could allow poles at infinity , then the degree of the numerator can be larger than that of the denominator .
more precisely , we can easily obtain the following using the notations from borwein - erdlyi theorem , if @xmath26 is a polynomial with @xmath31 and @xmath28 is a rational function , then @xmath32 let @xmath33 , and let @xmath34 , where @xmath35 , @xmath36 . then @xmath37 for @xmath38 , so @xmath39 since @xmath40 , therefore @xmath41 using borwein - erdlyi theorem for @xmath42 , @xmath38 , @xmath43 letting @xmath44 and combining the last three displayed estimates , we obtain the theorem . note that if we let all the poles tend to infinity , then we get back the original bernstein ( riesz ) inequality for polynomials on the unit disk .
let us also remark that the original proof of borwein and erdlyi also proves ( [ eq : rat_ineq_suff ] ) , with little modifications .
the relation with green s functions is as follows .
it is well known ( see e.g. @xcite , p.109 ) that green s function of the unit disk @xmath30 with pole at @xmath45 is @xmath46 and green s functions of the complement of the unit disk @xmath47 with pole at @xmath48 , @xmath49 and with pole at infinity are @xmath50 for the normal derivatives elementary calculations give ( @xmath38 , @xmath51 is the inner normal , @xmath52 is the outer normal ) @xmath53 they are also mentioned in @xcite , p.1739 . using this notation , we can reformulate these last two theorems as follows .
this is actually the result of borwein and erdlyi with slightly different wording .
[ thm : genborweinerdelyi]let @xmath54 be an arbitrary rational function with no poles on the unit circle where @xmath26 and @xmath55 are polynomials . denote the poles of @xmath21 on @xmath56 by @xmath57 where each pole is repeated as many times as its order .
then , for @xmath58 , @xmath59 note that if @xmath60 , then @xmath21 has a pole at @xmath61 , therefore it is repeated @xmath62 times and this pole at @xmath61 is taken into account in the second term of maximum .
inequality ( [ ineq : genborweinerdelyi ] ) is sharp , the factor on the right hand side can not be replaced for smaller constant , see , e.g. , @xcite , p. 324 .
let @xmath5 be a finite union of @xmath18 smooth , disjoint jordan arcs on the complex plane , that is , @xmath63 denote the endpoints of @xmath64 by @xmath65 , @xmath66 .
we need the following proposition to transfer our setting .
although we will use it for one analytic jordan arc , it can be useful for further researches .
after we worked out the proof , we learned that widom developed very similar open - up lemma in his work , see @xcite , p. 205 - 207 .
the difference is that he considers @xmath67 smooth arcs with hlder continuous @xmath68-th derivative ( see also p. 145 ) while we need this open - up technique for analytic arcs .
furthermore , there is a difference regarding the number of poles .
this is discussed after the proof .
[ prop : open - up]there exists a rational function @xmath69 and a domain @xmath70 such that @xmath71 is a compact set with @xmath72 components , @xmath73 is union of finitely many smooth jordan curves and @xmath69 is a conformal bijection from @xmath74 onto @xmath75 with @xmath76 . furthermore , if @xmath5 is analytic , then @xmath77 is analytic too .
first , we show that there are polynomials @xmath26 , @xmath55 such that @xmath78 , @xmath79 , @xmath80 and @xmath81 obviously , @xmath82 and the numerator is a polynomial of degree @xmath83 .
let @xmath84 . taking reciprocal , @xmath85 , that is
, the location of the poles are known .
our goal is to find @xmath86 such that @xmath87 or equivalently , @xmath88 must have @xmath16 residue everywhere , @xmath89 for all @xmath90 .
since @xmath91 s are pairwise different , @xmath92 , @xmath93 and @xmath94 are linearly independent , so we can choose @xmath95 s so that @xmath96 where @xmath97 will be specified later .
write @xmath98 in the form @xmath99 with suitable @xmath100 s .
it is easy to see that @xmath89 for all @xmath101 , furthermore @xmath102 . comparing the coefficients of @xmath103
, we obtain @xmath104 , @xmath105 . rearranging the expression for @xmath106
, @xmath97 must satisfy the following equation @xmath107 with this choice , there exists @xmath108 with the desired properties .
the domain @xmath74 is constructed as follows .
denote the unbounded component of @xmath109 $ ] by @xmath74 .
we prove that @xmath74 is a domain and its boundary consists of finitely many jordan curves and those curves are smooth .
locally , if @xmath110 for some @xmath64 and @xmath111 is not endpoint of @xmath64 , then , by the construction , @xmath111 is not a critical value . in other words , for any @xmath112 such that @xmath113 , we know @xmath114 ( @xmath112 is not a critical place ) .
if @xmath110 is an endpoint and @xmath115 is any of its inverse image , then @xmath116 by ( [ eq : fminimality ] ) and since the degree of @xmath26 and @xmath55 are minimal , @xmath117 .
therefore @xmath118 , and the inverse image @xmath119 $ ] of @xmath64 near @xmath115 is a smooth , simple arc .
so each bounded component of @xmath71 is such a compact set that it is a closure of a jordan domain .
using continuity and connectedness , @xmath120 $ ] has at least @xmath72 bounded components . if there were more than @xmath72 components , then we obtain contradiction as follows .
the boundary of each component is mapped into @xmath5 , so there should be more than @xmath83 critical points , but this contradicts the minimality of @xmath69 . denote the boundary of the components by @xmath121 , @xmath122 .
these @xmath121 s are smooth jordan curves and assume @xmath123 , @xmath124 $ ] .
it is clear that each component has nonempty interior and contains at least one pole of @xmath69 , otherwise @xmath69 maps that component onto some open , bounded , nonempty set and this set would intersect @xmath125 .
therefore each component contains exactly one pole which is simple by the minimality assumption .
now , @xmath126 is univalent on @xmath74 because of the followings .
take smooth jordan curves @xmath127 , @xmath124 $ ] satisfying the next properties : @xmath128 , @xmath129 as @xmath130 and @xmath131 as @xmath130 and @xmath132 .
since @xmath133 , @xmath134 as @xmath135 therefore @xmath136 as @xmath130 and , by continuity , @xmath137
. since @xmath69 has no critical values outside @xmath5 , the @xmath138 s are smooth jordan curves .
fix @xmath139 , then there is ( at least one ) @xmath140 with @xmath141 , because @xmath142 is open , @xmath143 and @xmath144 . if @xmath145 is small enough , then @xmath146 and @xmath147 ( @xmath122 ) , so @xmath148 .
therefore @xmath149 , so there is exactly one inverse image , this shows the univalence of @xmath69 .
we can give another proof for the univalence as follows .
there is a ( local ) branch of @xmath150 such that @xmath151=z / c_{1}+ .. $ ] as @xmath152 , in other words , @xmath61 is not a branch point of @xmath150 .
furthermore , the function @xmath69 has branch points only at @xmath153 s , @xmath154 and it behaves as a square root there .
therefore every analytic continuations along any curve in @xmath155 give the same function element .
now we use lemma 2 , p. 175
in @xcite with this ( local ) branch . therefore we can choose a ( global ) regular branch of @xmath150 such that @xmath156=\infty$ ] .
since this branch is regular and @xmath69 is a rational function , there is no other inverse image of @xmath61 by @xmath150 in @xmath74 . by the construction of @xmath74 and applying the maximum principle , we have @xmath157 , @xmath158 . using the majorization principle ( see @xcite , theorem 1 on p. 624 ) or theorem 4.4.1 on p. 112 from @xcite , we obtain that @xmath69 is conformal bijection from @xmath74 onto @xmath75 . as for the smoothness assertion ( @xmath77 analytic )
, this follows from standard considerations as follows . without loss of generality
, we may assume that @xmath159 , is a convergent power series for @xmath160 and @xmath161 is such that @xmath162 , @xmath163 and @xmath164 .
it is known , see e.g. @xcite , p. 286
, that the two branches of the inverse of @xmath69 near @xmath165 can be written as @xmath166 where @xmath167 are holomorphic functions .
denote them by @xmath168 and @xmath169 .
this way @xmath170=g_{0}\left(\kappa\left(t^{2}\right)\right)+t\sqrt{1+\kappa_{1}\left(t^{2}\right)}g_{1}\left(\kappa\left(t^{2}\right)\right)$ ] is a convergent power series in @xmath171 $ ] and similarly for @xmath172 $ ] and @xmath173 .
considering @xmath174 for @xmath171 $ ] , we see that @xmath175 , so @xmath176 is actually a convergent power series and it parametrizes the two joining arc . as for the number of poles , widom s open - up mapping
is constructed as iterating the joukowskii mapping ( composed with a suitable linear mapping in each step ) for each arc and that open - up mapping has @xmath177 different , simple poles and the location of poles also depends on the order of arcs .
in contrast , our open - up rational function has @xmath178 simple poles .
with this proposition , we switch from polynomials on jordan arcs to rational functions on jordan curves as follows .
we use the following notations , assumptions .
fix one , @xmath18 smooth jordan arc @xmath179 with endpoints @xmath180 and @xmath181 and let @xmath182 , @xmath183 , @xmath184 .
denote the two normal vectors of unit length at @xmath111 to @xmath179 by @xmath185 , @xmath186 , where @xmath187 .
we may assume that @xmath188 and @xmath189 depend continuously on @xmath111 .
we use the same letter for normals in different planes and from the context , it is always clear that which arc we refer to .
we use the rational mapping @xmath69 and the domain @xmath190 from the previous proposition for @xmath179 . denote the inward normal vector to @xmath77 at @xmath191 by @xmath192 and the outward normal vector to @xmath77 at @xmath112 by @xmath193 , @xmath194 .
it is easy to see that there are two inverse images of @xmath111 : @xmath195 ( such that @xmath196 ) and we can assume that @xmath197 are continuous functions of @xmath198 by reindexing @xmath115 and @xmath199 , we may assume that the normal vector @xmath200 is mapped by @xmath69 to the normal vector @xmath186 .
this immediately implies that @xmath201 , @xmath202 , @xmath203 are mapped by @xmath69 to @xmath185 , @xmath185 , @xmath186 respectively .
, @xmath111 , @xmath204 and @xmath205 with the normal vectors , scaledwidth=100.0% ] let us denote the domain @xmath206 by @xmath204 . since @xmath207 and @xmath69 is a conformal bijection from @xmath205 onto @xmath208
, @xmath69 is a conformal bijection from @xmath209@xmath204 onto @xmath208 . for simplicity ,
let us denote the inverse of @xmath69 onto @xmath204 by @xmath168 and onto @xmath205 by @xmath169 .
these geometrical objects are depicted in figure [ fig : onearconecurve ] where we indicated the normal vectors @xmath186 and @xmath200 with dashed arrows ( we fix the notations with their help ) and we indicated the other normal vectors with simple ( not dashed ) arrows ( their indexings are consequence of the earlier two vectors ) . [
prop : green_trf1]using the notations above , for the green s functions of @xmath210 and @xmath204 and for @xmath211 we have @xmath212 and , similarly for the other side , @xmath213 for arbitrary polynomial @xmath214 , let @xmath215
. then @xmath216 .
this immediately follows from the conformal invariance of green s functions @xmath217 and @xmath218 see e.g. @xcite , p. 107 , theorem 4.4.4 .
this proposition implies that it is enough to take into account the normal derivatives at , say , @xmath115 only , i.e. @xmath219 and @xmath220 only .
here @xmath204 is the bounded domain from the previous section and @xmath205 is the unbounded domain from the previous section .
actually , @xmath221 . as earlier , @xmath222 and @xmath223 . with these notations , @xmath224 .
using kellogg - warschawski theorem ( see e.g. @xcite p. 49 , theorem 3.6 ) , if the boundary is @xmath225 smooth , then the riemann mappings of @xmath226 onto @xmath227 respectively and their derivatives can be extended continuously to the boundary . under analyticity assumption , we can compare the riemann mappings as follows .
[ prop : green_deriv_g_one_g_two]let @xmath228 be fixed .
then there exist two riemann mappings @xmath229 , @xmath230 such that @xmath231 and @xmath232 , @xmath233 .
if @xmath224 is analytic , then there exist @xmath234 such that @xmath6 extends to @xmath235 , @xmath236 and @xmath237 is a conformal bijection , and similarly , @xmath7 extends to @xmath238 , @xmath239 and @xmath240 is a conformal bijection .
the existence of @xmath6 follows immediately from the riemann mapping theorem by considering arbitrary riemann mapping and composing this mapping with a suitable rotation and hyperbolic translation toward @xmath241 ( that is , @xmath242 with @xmath243 and @xmath244 , @xmath245 , and @xmath246 , @xmath247 ) .
the existence of @xmath7 follows the same way , using the same family of hyperbolic translations .
the extension follows from the reflection principle for analytic curves ( see e.g. @xcite pp .
16 - 21 ) .
from now on , we fix such two conformal mappings and let @xmath248\right]$ ] and @xmath249=\phi_{2}^{-1}\left[f_{2}^{-1}\left[\infty\right]\right]$ ] .
the domains of these analytic extensions are depicted on figure [ fig : tworiemann ] where @xmath250 is the grey region on the right and is mapped onto @xmath251 by @xmath6 which is the grey region on the left . using these mappings , we have the following relations between the normal derivatives of green s functions and blaschke factors .
[ prop : green_deriv_unitdisk]the followings hold @xmath252\right)=\frac{\partial}{\partial n_{1}\left(1\right)}g_{\bd}\left(1,a_{1}\right)=\frac{1-\left|a_{1}\right|^{2}}{\left|1-a_{1}\right|^{2}},\\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)=\frac{\partial}{\partial n_{2}\left(1\right)}g_{\bd^{*}}\left(1,a_{2}\right)=\frac{\left|a_{2}\right|^{2}-1}{\left|1-a_{2}\right|^{2}},\end{gathered}\ ] ] and if @xmath253 , then @xmath254\right)=\frac{\partial}{\partial n_{2}\left(1\right)}g_{\bd^{*}}\left(1,\infty\right)=1.\ ] ] the second equalities in all three lines follow from ( [ eq : ndgd ] ) , ( [ eq : ndgdsf ] ) and ( [ eq : ndgdsi ] ) . we know that @xmath255 and @xmath256 , moreover @xmath257 , @xmath258 imply that @xmath259 is mapped to @xmath260 by @xmath261 , @xmath233 and the mappings @xmath261 , @xmath233 also preserve the length at @xmath241 ( there is no magnifying factor @xmath262 unlike at proposition [ prop : green_trf1 ] ) . using the conformal mappings @xmath6 and @xmath7 , and the conformal invariance of green s functions , we obtain the first equalities in all three lines .
before we start the proof , let us recall three results .
the first one is gonchar - grigorjan estimate when we have one pole only .
see @xcite , theorem 2 on p. 572
( in the english translation ) .
let @xmath263 be a simply connected domain and its boundary is @xmath264 smooth .
let @xmath265 be a meromorphic function on @xmath266 such that it has only one pole .
assume that @xmath267 can be extended continuously to the boundary @xmath268 of @xmath266 .
denote @xmath269 the principal part of @xmath267 in @xmath266 ( with @xmath270 ) and let @xmath271 denote the holomorphic part of @xmath209@xmath267 in @xmath266 . denote the order of the pole of @xmath267 by @xmath272 .
then @xmath273 and there exists @xmath274 depending on @xmath266 only such that @xmath275 where @xmath276 denotes the sup norm over the boundary of @xmath266 . in the main result of this paper we are interested in asymptotics as @xmath17 . in particular , if @xmath277 , then @xmath278 , so we may write @xmath279 .
the second result is a special case of the bernstein - walsh estimate , see @xcite , p. 156
, theorem 5.5.7 a ) or @xcite , p. 153 .
let @xmath280 be a domain , @xmath281 and denote its green s function by @xmath282 with pole at infinity .
let @xmath283 be a meromorphic function which has only one pole at infinity and we denote the order of the pole by @xmath284 .
assume that @xmath285 can be extended continuously to the boundary @xmath286 of @xmath287 .
then @xmath288 where @xmath289 denotes the sup norm over @xmath286 .
the third result is a special case of a general construction of fast decreasing polynomials by totik , see @xcite , corollary 4.2 and theorem 4.1 too on p. 2065 .
let @xmath290 be a compact set , @xmath291 be a boundary point .
assume that @xmath292 satisfies the touching outer - disk - condition , that is , there exists a closed disk ( with positive radius ) such that its intersection with @xmath292 is @xmath293 .
then there exist @xmath294 such that for all @xmath284 there exists a polynomial @xmath295 with the following properties : @xmath296 , @xmath297 , @xmath298 and if @xmath299 , @xmath300 , then @xmath301 . to apply this third theorem
, we introduce several notations .
we need @xmath302 and its inverse @xmath303 .
note that @xmath304 , @xmath305 and let @xmath306 .
obviously , @xmath307 .
let @xmath308 and @xmath309 is chosen so that @xmath310 .
this @xmath311 depends on @xmath205 only .
let @xmath312 , this disk touches the unit disk at @xmath313 .
fix @xmath314 , @xmath209@xmath315 , such that @xmath316 . then for every
@xmath317 $ ] , @xmath318 consists of exactly two points , @xmath319 and @xmath320 .
it is easy to see that the length of the two arcs of @xmath321 lying in between @xmath322 and @xmath323 are different , therefore , by reindexing them , we can assume that the shorter arc is going from @xmath322 to @xmath323 counterclockwise .
elementary geometric considerations show that for all @xmath324 , @xmath325 with @xmath326 , we have ( since @xmath327 ) @xmath328 let @xmath329 obviously , this @xmath330 is a compact set and satisfies the touching - outer - disk condition at @xmath331 of totik s theorem .
see figure [ fig : gamma_two ] later .
consider @xmath332\cup\phi_{1}\circ\psi^{-1}\left[k_{w}^{*}\cap\bd^{*}\right]\cup g_{1}.\ ] ] this is a compact set and also satisfies the touching - outer - disk condition at @xmath333 of totik s theorem . obviously , @xmath334 , @xmath335 , @xmath336 and if @xmath337 , then @xmath338 and @xmath339 too .
now applying totik s theorem , there exists a fast decreasing polynomial for @xmath340 at @xmath341 of degree at most @xmath188 which we denote by @xmath342 .
more precisely , @xmath55 has the following properties : @xmath343 , @xmath344 on @xmath345 , @xmath346 and if @xmath347 , @xmath345 , then @xmath348 let @xmath349 , @xmath350 , @xmath351 and @xmath352 . in this subsection
, we let @xmath353 where @xmath12 is a fixed polynomial of degree @xmath13 and @xmath69 is the open - up rational function ( see proposition [ prop : open - up ] ) for @xmath5 ( from theorem [ thm : main ] ) .
actually , we use only the following facts .
@xmath21 is a rational function such that it has one pole in @xmath204 and one in @xmath205 .
we know that the poles of @xmath21 are @xmath354 $ ] and @xmath355 $ ] , and the order of the pole in @xmath204 is @xmath13 .
it is easy to decompose @xmath21 into sum of rational functions , that is , @xmath356 where @xmath42 is a rational function with pole in @xmath204 , @xmath357 and @xmath358 is a polynomial ( rational function with pole at @xmath61 ) .
this decomposition is unique .
we use the gonchar - grigorjan estimate ( [ est : gonchargrigorjan ] ) for @xmath358 on @xmath251 , so we have @xmath359 obviously , we have @xmath360 consider @xmath361 this is a meromorphic function in @xmath250 .
we may assume that @xmath362 has only one pole in @xmath250 otherwise we can decrease @xmath363 so that the pole in @xmath205 is not in @xmath364 .
we know that @xmath365 and @xmath366 .
we decompose `` the essential part of '' @xmath362 as follows @xmath367 where @xmath368 is a rational function , @xmath369 and @xmath370 is holomorphic in @xmath30 .
we use the gonchar - grigorjan estimate ( [ est : gonchargrigorjan ] ) again for @xmath362 on @xmath30 , this way the following sup norm estimate holds @xmath371 where @xmath372 is a constant independent of @xmath362 . as a remark ,
let us note that we may write @xmath373 for simplicity since we are interested in asymptotics as @xmath374 in the main theorem . otherwise , if @xmath375 or @xmath376 , then @xmath377 is a constant or linear polynomial and the error term @xmath378 in the main theorem ( theorem [ thm : main ] ) can be sufficiently large ( depending on @xmath5 and @xmath379 ) for these two particular values of @xmath13 . in this manner , we write @xmath380 in general , but we simplify it to @xmath381 frequently .
furthermore , we can estimate @xmath382 on @xmath383 as follows @xmath384 we also need to estimate @xmath55 outside @xmath30 ( and @xmath330 ) as follows .
using @xmath346 and bernstein - walsh estimate ( [ est : bernsteinwalsh ] ) , we can write for @xmath383 @xmath385 since the set @xmath386 is bounded , @xmath387 therefore , for all @xmath383 , @xmath388 this way we can continue ( [ est : phi_one_e_v ] ) and we use @xmath389 here and that @xmath368 is a rational function with no poles outside @xmath30 and the maximum principle for @xmath368 @xmath390 and here we used that @xmath42 has no pole in @xmath205 and the maximum principle .
we can estimate these three sup norms with the help of ( [ eq : f_one_norm_est ] ) and ( [ eq : phi_one_norm ] ) , ( [ eq : f_one_norm_est ] ) and ( [ eq : phi_one_e_norm_est ] ) , ( [ eq : phi_one_norm ] ) , ( [ eq : f_one_norm_est ] ) .
hence we have for @xmath383 @xmath391 approximate and interpolate @xmath370 as follows with rational function which has only one pole , namely at @xmath392 $ ] .
consider @xmath393 on @xmath394 .
using the properties of @xmath395 , we have @xmath396 and @xmath397 is a holomorphic function in @xmath394 .
we interpolate and use integral estimates for the error , see e.g. @xcite , p. 170
, proof of theorem 6.3.1 or @xcite , p. 11 .
therefore , let @xmath398 where @xmath399 .
we define the approximating polynomial @xmath400 it is well known that @xmath401 does not depend on @xmath402 . since @xmath313 is a double pole of @xmath403 , therefore @xmath401 and @xmath404 coincide there with @xmath397 and @xmath405 respectively .
the error of the approximating polynomial @xmath401 to @xmath397 is @xmath406 here @xmath407 can be arbitrary .
it is easy to see that for @xmath407 , @xmath408 and @xmath409 therefore , using ( [ eq : phi_one_e_v_est ] ) , we can estimate the error ( of approximation of @xmath401 to @xmath397 ) as follows @xmath410 which tends to @xmath16 as @xmath17 , because @xmath411 and @xmath412 considering @xmath413 , it is a rational function with pole at @xmath414 only , the order of its pole at @xmath414 is at most @xmath415 and we know that @xmath416 where @xmath15 is independent of @xmath12 and @xmath21 and depends only on @xmath205 and tends to @xmath16 as @xmath17 , furthermore @xmath417 now we interpolate and approximate @xmath418 . as earlier , we do not need the full information of this function , it is enough to deal with @xmath418 locally around @xmath241 and preserve the sup norm .
therefore we `` chop off '' `` the unnecessary parts of @xmath418 '' with the fast decreasing polynomial @xmath55 .
we have the following description about the growth of green s function .
[ lem : green_est]there exists @xmath419 depending on @xmath420 , that is , depending on @xmath205 only and is independent of @xmath421 and @xmath21 such that for all @xmath422 we have @xmath423 and @xmath424 furthermore , there exists @xmath425 which depends on @xmath205 and independent of @xmath421 and @xmath21 such that for all @xmath426 we have @xmath427 and @xmath428 for simplicity , let @xmath429 where @xmath430 , if @xmath431 and @xmath432 .
we can express green s function in the following ways for @xmath433 , @xmath434 and for @xmath435 @xmath436 the first displayed inequality in the lemma comes from continuity considerations and the conformal bijection properties . integrating this inequality along radial rays ,
we obtain ( [ eq : green_est_away ] ) .
if we are close to @xmath241 , then more is true : @xmath437 using continuity , we see that there exists @xmath425 such that for all @xmath438 , @xmath426 , we have @xmath439 in particular , for all @xmath440 from the segment @xmath441 $ ] , @xmath442 $ ] , @xmath443 and @xmath444 . therefore ,
integrating with respect to @xmath440 along @xmath441 $ ] , we obtain @xmath445 now we give the approximating polynomial as follows @xmath446 where @xmath447 can be arbitrary with @xmath448 and @xmath449 .
we remark that we use the same interpolating points , but we need a different @xmath447 for the error estimate .
now we construct @xmath450 for the estimate and investigate the error .
we use @xmath451 , @xmath452 and @xmath453 .
we give four jordan arcs that will make up @xmath454 .
let @xmath455 be the ( shorter , circular ) arc between @xmath456 and @xmath457 , @xmath458 be the longer circular arc between @xmath209@xmath459 and @xmath460 , @xmath461 and similarly @xmath462 be the two segments connecting @xmath458 and @xmath455 .
finally let @xmath454 be the union of @xmath458 , @xmath463 , @xmath455 and @xmath464 .
figure [ fig : gamma_two ] depicts these arcs and @xmath330 defined above . and
the arcs that make up @xmath454 ] we estimate the error of @xmath465 to @xmath466 on each integral separately : @xmath467 for the first term , we use bernstein - walsh estimate ( [ est : bernsteinwalsh ] ) for the polynomial @xmath358 on @xmath205 and the fast decreasing polynomial @xmath55 as follows .
if @xmath468 , then with ( [ eq : green_est_away ] ) , @xmath469 , therefore @xmath470 where we used ( [ eq : f_two_norm_est ] ) .
now we use the fast decreasing property of @xmath55 as follows .
we know that @xmath471 ( if @xmath472 ) and with the elementary geometric considerations ( [ eq : w_star_dist ] ) we have @xmath473 which is equivalent to @xmath474 ( this is true if @xmath13 is large ) .
it is also important that @xmath475 and @xmath476 therefore the growth order of the distances is preserved by @xmath477 .
hence the fast decreasing polynomial @xmath55 is small , see ( [ eq : q_fast_decreases ] ) , and we can write @xmath478 and integrating along @xmath458 , we can write for @xmath407 @xmath479 here we used @xmath451 .
we estimate the third term , the integral on @xmath455 , as follows for @xmath407 @xmath480 here , @xmath481 , @xmath482 , @xmath483 .
roughly speaking , @xmath358 grows and this time @xmath55 grows too ( the bad guys ) and only @xmath484 decreases ( the good guy ) .
we estimate their growth using bernstein - walsh estimate ( [ est : bernsteinwalsh ] ) for @xmath358 on @xmath205 and lemma [ lem : green_est ] ( and estimate ( [ eq : f_two_norm_est ] ) as well ) in the following way . here , as earlier , @xmath485 @xmath486 where in the last two steps we used @xmath487 from ( [ eq : w_star_dist ] ) and @xmath452 .
as for @xmath403 , @xmath488 where we used @xmath489 , @xmath453 and @xmath452 . as for @xmath55 ( this time it is a bad guy )
, we use bernstein - walsh estimate ( [ est : bernsteinwalsh ] ) for @xmath55 on @xmath490 and that @xmath491 . therefore , @xmath492 and we know that @xmath493 , hence @xmath494 here we used again ( [ eq : w_star_dist ] ) and the definition of @xmath495 .
we multiply together all these three last displayed estimates , this way we can continue our main estimate ( [ eq : third_term_main_est ] ) . note that @xmath496 cancels , and @xmath497 kills the factor @xmath498 , in more detail :
@xmath499 where we used several estimates : length of @xmath455 is at most @xmath500 , the definitions of @xmath501 and @xmath495 and that @xmath502 , therefore @xmath503 . for @xmath463 and @xmath464
, we apply the same estimate which we detail for @xmath463 only .
we again start with the integral for @xmath407 @xmath504 since @xmath505 , we can rewrite it in the form @xmath506 where @xmath507 ( with @xmath319 ) .
we use essentially the same steps to estimate @xmath358 ( the only one bad guy this time ) and @xmath403 and @xmath55 ( this time it is a good guy ) . in estimating @xmath358 ,
the only difference is that @xmath508 , so @xmath509 similarly for @xmath403 , we can write @xmath510 as for @xmath55 , we know that @xmath511 is far from @xmath313 so @xmath55 is small there .
more precisely , following the same argument as for @xmath458 , we know that @xmath473 , hence ( [ eq : q_fast_decreases ] ) holds for @xmath55 at @xmath511 , that is , we can write @xmath512 putting these all together , we see that @xmath513 cancels and actually @xmath55 make the integrand small .
so we can continue the estimate ( [ eq : second_term_main_est ] ) @xmath514 where we used that the length of @xmath463 is at most @xmath241 ( since @xmath315 ) and @xmath451 . summarizing these estimates on @xmath458 , @xmath455 and @xmath463 ( and also on @xmath464 )
, we have uniformly for @xmath515 , @xmath516 where @xmath15 tends to @xmath16 as @xmath17 but it is independent of @xmath12 and @xmath358 .
obviously , @xmath517 is a rational function with pole at @xmath518 only , the order of the pole at @xmath414 ( of @xmath517 ) is @xmath519 and using the properties of @xmath520 , we uniformly have for @xmath521 @xmath522 that is , @xmath523 since @xmath313 is double zero of @xmath524 , @xmath525 , and dividing both sides with @xmath526 , we obtain @xmath527 consider the `` constructed '' rational function @xmath528 this function @xmath529 has a pole at @xmath530 ( because of @xmath531 ) and the order of its pole at @xmath530 is at most @xmath13 , and @xmath529 has a pole at @xmath414 ( because of @xmath413 and @xmath517 ) and the order of its pole at @xmath414 is at most @xmath532 .
we use the identity @xmath533 to calculate the derivatives as follows @xmath534 where the second term is zero because of the fast decreasing polynomial ( @xmath535 ) and for the first term we can apply theorem 1.3 from @xcite in the following way ( @xmath536 ) : @xmath537 where @xmath15 depends on @xmath205 and @xmath115 only and tends to @xmath16 as @xmath538 ( note : @xmath539 )
. therefore @xmath540 this way we need to consider @xmath541 only .
the derivatives at @xmath241 of the original @xmath21 and @xmath529 coincide , because of ( [ eq : phi_decomp ] ) , ( [ eq : phi_one_e_deriv_at_one ] ) and ( [ eq : p_two_n_psi_deriv_at_one ] ) , so @xmath542 as for the sup norms , we use ( [ eq : phi_decomp ] ) , ( [ eq : phi_one_e_sup_err_est ] ) , ( [ eq : sup_norm_err_est_two ] ) , so we write @xmath543 now we apply the borwein - erdlyi inequality ( [ ineq : genborweinerdelyi ] ) for @xmath529 as follows : @xmath544 where the summation is taken over all poles in @xmath30 and in @xmath545 respectively , counting multiplicities .
we will continue this estimate later after simplifying these expressions . using propositions [ prop : green_deriv_unitdisk ] and [ prop : green_deriv_g_one_g_two ] ,
we can write @xmath546\right)\\ = n\frac{\partial}{\partial n_{2}\left(z_{0}\right)}g_{\bc_{\infty}\setminus k}\left(z_{0},\infty\right)\left|f'\left(u_{0}\right)\right|\end{gathered}\ ] ] where in the last step we used proposition [ prop : green_deriv_g_one_g_two ] with @xmath547 and identifying @xmath548 .
similarly , we can simplify the second term in the maximum in ( [ est : main_est_for_h ] ) @xmath549\right)\\ = \left(1+o\left(1\right)\right)n\frac{\partial}{\partial n_{1}\left(z_{0}\right)}g_{\bc_{\infty}\setminus k}\left(z_{0},\infty\right)\left|f'\left(u_{0}\right)\right|\end{gathered}\ ] ] where @xmath15 here does not depend on anything .
note that we `` used a slightly bit more the pole at @xmath414 '' , but it does not cause problem .
so we can continue the main estimate ( [ est : main_est_for_h ] ) @xmath550\right),\\ \left(1+o\left(1\right)\right)n\frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\\ \le\left\vert h\right\vert _
{ \partial\bd}\left(1+o\left(1\right)\right)n\\ \cdot\max\left(\frac{\partial}{\partial n_{1}\left(u_{0}\right)}g_{g_{1}}\left(u_{0},f_{1}^{-1}\left[\infty\right]\right),\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\right).\end{gathered}\ ] ] summarizing these estimates , we have for @xmath529 @xmath551\right),\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\right).\end{gathered}\ ] ] now we rewrite this inequality for @xmath552 using ( [ eq : h_qf_prime_at_one ] ) and ( [ est : qf_h_sup_norm ] ) , so @xmath553\right),\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\right)\\ + o\left(1\right)n\left\vert f\right\vert _ { \partial g_{2}}\cdot\max\left(\frac{\partial}{\partial n_{1}\left(u_{0}\right)}g_{g_{1}}\left(u_{0},f_{1}^{-1}\left[\infty\right]\right),\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\right).\end{gathered}\ ] ] now , we use the estimate @xmath554 and ( [ est : one_minus_q_f_prime ] ) , so @xmath555\right),\ \frac{\partial}{\partial n_{2}\left(u_{0}\right)}g_{g_{2}}\left(u_{0},f_{2}^{-1}\left[\infty\right]\right)\big)\right).\label{ineq : further_refs}\end{gathered}\ ] ] in the final step , we use @xmath556 and proposition [ prop : green_trf1 ] , so we get the main theorem .
in this section we show that the result is asymptotically sharp , that is , we prove theorem [ thm : sharpness ] . the idea is similar to that of @xcite .
note that we assume @xmath18 smoothness only .
we may assume that @xmath557 furthermore , we assume that @xmath558 and @xmath559 are defined on the component of @xmath5 containing @xmath9 and they are continuous there except for the endpoints . it is easy to see that for every @xmath560 there exists a compact set @xmath561 such that @xmath562 is finite union of disjoint , @xmath18 smooth jordan curves , @xmath563 , @xmath564 and the normal vector @xmath565 to @xmath566 ( pointing outward ) at @xmath9 is equal to @xmath11 and @xmath567 these conditions , roughly speaking , require that near @xmath9 , @xmath566 is on the @xmath10-side of @xmath5 and the whole @xmath566 shrinks to @xmath5 as @xmath568 .
figure [ fig : kstar ] depicts @xmath5 and the grey area is @xmath566 . and
@xmath566,scaledwidth=100.0% ] now we apply the sharpness result of @xcite ( theorem 1.4 , p. 194 ) .
this gives a sequence of polynomials for @xmath569 , say @xmath570 , with @xmath571 such that @xmath572 where @xmath573 depends on @xmath569 and @xmath9 and tends to @xmath16 as @xmath574 .
since @xmath575 was arbitrary , we see that @xmath576 , that is , choosing a suitable subsequence of @xmath577 we obtain the assertion .
the first author was supported by the european research council advanced grant no . 267055 , while he had a postdoctoral position at the bolyai institute , university of szeged , aradi v. tere 1 , szeged 6720 , hungary . the second author ,
bla nagy was supported by magyary scholarship : this research was realized in the frames of tmop 4.2.4 .
a/2 - 11 - 1 - 2012 - 0001 , , national excellence program - elaborating and operating an inland student and researcher personal support system .
the project was subsidized by the european union and co - financed by the european social fund .
the authors are deeply indebted to professor vilmos totik for several helpful suggestions and comments which enabled us to improve the presentation of our result and for drawing our attention to widom s open - up result .
akml98 s. i. kalmykov , _ majorization principles and some inequalities for polynomials and rational functions with prescribed poles _ , zap .
nauchn . sem . s .-
otdel . mat .
inst . steklov .
( pomi ) * 357 * ( 2008 ) , no .
analiticheskaya teoriya chisel i teoriya funktsii .
23 , 143157 , 227 .
mr 2765937 ( 2012b:30008 ) ch .
pommerenke , _ boundary behaviour of conformal maps _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , vol .
299 , springer - verlag , berlin , 1992 .
mr 1217706 ( 95b:30008 ) v. i. smirnov and n. a. lebedev , _ functions of a complex variable : constructive theory _
, translated from the russian by scripta technica ltd , the m.i.t .
press , cambridge , mass . ,
mr 0229803 ( 37 # 5369 ) edward b. saff and vilmos totik , _ logarithmic potentials with external fields _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , vol .
316 , springer - verlag , berlin , 1997 , appendix b by thomas bloom .
mr 1485778 ( 99h:31001 ) sergei kalmykov + institute of applied mathematics , febras , 7 radio street , vladivostok , 690041 , russia , + far eastern federal university , 8 sukhanova street , vladivostok , 690950 , russia and + bolyai institute , university of szeged , szeged , aradi v. tere 1 , 6720 , hungary + email address : [email protected] | in this paper we prove an asymptotically sharp bernstein - type inequality for polynomials on analytic jordan arcs .
also a general statement on mapping of a domain bounded by finitely many jordan curves onto a complement to a system of the same number of arcs with rational function is presented here .
this fact , as well as , borwein - erdlyi inequality for derivative of rational functions on the unit circle , gonchar - grigorjan estimate of the norm of holomorphic part of meromorphic functions and totik s construction of fast decreasing polynomials play key roles in the proof of the main result .. ] classification ( msc 2010 ) : 41a17 , 30c20 , 30e10 dedicated to professor vilmos totik on his sixtieth birthday |
symmetry has plaid an important role in particle physics@xcite .
all known basic forces of nature , i.e. , gravitational , electromagnetic , weak and strong forces , are governed by the symmetry principles .
three of them ( electromagnetic , weak and strong forces ) have turned out to be characterized by the gauge symmetry @xmath6 . thus the three fundamental gauge interactions among the building blocks or elementary particles ( quarks and leptons ) are mediated via the abelian and non - abelian yang - mills gauge fields@xcite .
the real world of particles has been found to be successfully described by quantum field theories ( qfts ) @xcite . in the meantime , the gauge invariance has been viewed as a basic principle@xcite in addition to the well - known basic principles of lorentz invariance and translational invariance .
qfts have also been applied to deal with effective theories for composite fields and particles at low energies and also critical phenomena ( or phase transitions ) in statistical mechanics and condensed matter physics@xcite .
nevertheless , qfts can not be defined by a straightforward perturbative expansion due to the presence of ultraviolet ( uv ) divergences .
namely , the definition of feynman diagrams in the perturbative expansion may be meaningless because of lack of convergence .
to avoid such difficulties , one may modify the behavior of field theory at very large momentum , or introduce so called regulators ( more general speaking , the regularization quanta ) , so that all feynman diagrams become well - defined finite quantities .
such a procedure is usually called regularization .
the most important features required for the regularization are that the regularization should maintain the basic symmetry principles of the theory , such as gauge invariance , lorentz invariance and translational invariance ( or poincare invariance ) , and also preserve the initial but well - defined divergent properties of the theory .
many regularization methods have been introduced in the literature , such as : cut - off regularization@xcite , pauli - villars regularization@xcite , schwinger s proper time regularization@xcite , bphz regularization@xcite , dimensional regularization@xcite , lattice regularization@xcite , differential regularization@xcite .
all the regularizations have their advantages and shortcomings .
the cut - off regularization is the simplest one by naively setting an upper bound to the integrating loop momentum .
this method is often used to treat qfts in statistical mechanics or in certain low energy dynamical systems , where the divergent behavior of the theory plays the crucial role , and the lorentz or poincare invariance beomes unimportant and the gauge symmetry is not involved at all . in contrast , the method fails in applying to the qfts for elementary particles in which lorentz or poincare invariance and yang - mills gauge invariance play an imporatnt role .
this is because the method destroys the principle of lorentz and translational ( or poincare ) invariance and also the principle of gauge invariance for gauge theories .
the spirit of pauli - villars s regularization is to modify the propagator .
its prescription is simple : replacing any propagator by a sum of propagators with large masses , and choosing appropriate coefficients so that the large momentum behavior becomes well controlled .
more general modifications on the propagators may be introduced in schwinger s proper time regularization .
whereas in pauli - villars s regularization , a set of regulator fields are usually introduced to modify the action of original theory . though pauli - villars s inspired regularizations may preserve a large number of symmetries of the theory , there are still a class of feynman diagrams which can not be regularized by this approach .
such field theories include the non - abelian yang - mills gauge theories@xcite and also the non - linear @xmath7-model . in ref.@xcite ,
a higher covariant derivative pauli - villars regularization was proposed for maintaining the gauge symmetry in non - abelian gauge theories .
nevertheless , it has been shown that such a regularization violates unitarity@xcite and also leads to an inconsistent quantum chromodynamics ( qcd)@xcite .
bphz prescription relies on the theorem proved by weinberg@xcite , which states that the requirement for the actual convergence of the amplitude corresponding to any graph is that the power - counting should give a negative superficial degree of divergence for the complete multiple integral for the whole amplitude and also for any subintegration defined by holding any one or more linear combinations of the loop momenta fixed .
the prescription does eliminate all superficial divergences and render the renormalized feynman integrals to be convergent .
the forest formula provides a powerful tool to construct proofs of renormalizability to all orders .
in fact , the method is a regularization - independent subtraction scheme up to the finite part . to proceed the practical calculation
so as to extract the finite part by means of the forest formula , one still has to adopt a regularization scheme .
as such a subtraction process is based on expanding around an external momentum , which modifies the structure of the feynman integral amplitudes , thus the gauge invariance is potentially destroyed when applying the bphz subtraction scheme to non - abelian gauge theories .
in addition , the unitarity , locality and causality may also not rigorously hold in such a subtraction scheme . in the lattice regularization
, both space and time are made to be the discrete variables .
the method can preserve gauge symmetries of the theory , but the principle of lorentz and translational ( or poincare ) invariance is lost in this method . though there is a great advantage that lattice regularization
may be extended to the nonperturbative calculations by a numerical method , it may also , at the same time , lead to a disadvantage due to a very complicated perturbative calculation .
the principle of dimensional regularization is to define all feynman diagrams by analytic continuation in the space - time dimension parameter , when the space - time dimension is required to be the initial one , the feynman diagrams will recover the initial logarithmically uv divergences but the quadratically uv divergences are suppressed .
this method provides the simplest symmetry principle preserving regularization in the perturbative calculations . whereas it seems to be meaningless beyond perturbative expansion as it involves continuation of feynman diagrams in space - time dimension parameter to arbitrary values .
the method also fails if quantities are specific to the initial space - time dimension , such as @xmath8 in four dimensions and the complete antisymmetric tensor @xmath9 as well as the case that scaling behavior becomes important .
the cases include the chiral , topological and supersymmetric theories .
for that a so - called dimensional reduction regularization was suggested@xcite as a variant of dimensional regularization , in which the continuation from dimension d=4 to d = n is made by compactification .
thus the number of field components remains unchanged and the momentum integrals are n - dimensional .
such a prescription may cause ambiguities in the finite parts of the amplitudes and also in the divergent parts of high order corrections .
the scheme has been applied particularly in supersymmetric models . in general , it seems to hold only at one loop level and is actually inconsistent with analytical continuation concerning @xmath8 @xcite . on the other hand
, the dimensional regularization may not be applicable to deal with the dynamics of effective qfts at a physically meaningful finite energy scale .
the well - known example is the derivation of gap equation in the gauged nambu - jona - lasinio model@xcite .
it was shown@xcite that the dimensional regularization can not lead to a correct gap equation .
this is because the dimensional regularization destroys the quadratic ` divergent ' ( or quadratic ` cut - off ' momentum ) term in the gap equation though it maintains gauge invariance . instead , when adopting the simple cut - off regularization to regulate loop integrals , one can arrive at a desired gap equation , but the gauge invariance is destroyed by a quadratically ` divergent ' ( or quadratic ` cut - off ' momentum ) term and also by ambiguities associated with arbitrary routing of loop momentum due to lack of translational invariance .
another example is for the evaluation of the hadronic matrix elements of four quark operators in the kaon decays @xmath10 .
if applying the dimensional regularization to the long - distance operator evaluation in the chiral perturbation theory , one arrives at a wrong sign for the leading order contributions when matching to the corresponding short - distance operator evaluation in perturbative qcd@xcite .
the reason is the same as the previous one , the dimensional regularization destroys the leading quadratic cut - off momentum terms which have the same sign as the leading order terms in perturbative qcd .
actually , the quadratic terms in the chiral perturbation theory were found to play a crucial role for understanding the @xmath11 rule and for providing a consistent prediction on the direct cp - violating parameter @xmath12 in kaon decays@xcite . from these two realistic and interesting examples , it is not difficult to further understand the shortcoming of the dimensional regularization . in general , the dimensional regularization suppresses the scaling behavior somehow via a cancellation between uv and ir divergences .
actually , in the dimensional regularization there is no ` divergent ' contribution that can appear in a power of cut - off energy scale .
the differential regularization was proposed as a method working directly on feynman graphs in coordinate space@xcite .
where one substitutes singular expressions by derivatives of well - behaved distributions .
however , in gauge theories , the unsatisfactory feature arises due to the arbitrariness of the different renormalization scales in the ward identities among renormalized green functions .
thus one has to adjust the scales in an appropriate way so as to preserve the gauge invariance .
lately , a so - called constrained differential regularization was introduced@xcite to fix the ambiguities of the local counterterms , so that the gauge invariance can be maintained in a convenient way . while from the calculational viewpoint , one loses the advantage of working directly in the momentum space . from the above points of view ,
one may arrive at the conclusion that up to now there exists no single satisfactory regularization that can be applied to all purposes in qfts .
it is then natural to ask whether one is able to find a regularization which can combine the advantages appearing in the above mentioned regularizations . to realize this purpose
, the new regularization should match at least four criteria : \(i ) the regularization is rigorous that it can maintain the basic symmetry principles in the original theory , such as : gauge invariance , lorentz invariance and translational invariance , except the anomaly induced symmetry breaking and the intrinsic mass scale caused conformal scaling symmetry breaking .
\(ii ) the regularization is general that it can be applied to both underlying renormalizable qfts ( such as qcd ) and effective qfts ( like the gauged nambu - jona - lasinio model and chiral perturbation theory ) .
\(iii ) the regularization is also essential in the sense that it can lead to the well - defined feynman diagrams with maintaining the initial but well - defined divergent properties of the theory , so that the regularized theory only needs to make an infinity free ( finite ) or infinity - controlled renormalization .
\(iv ) the regularization must be simple that it can provide the practical calculations .
+ the first and forth criteria are clearly understood .
the second and third ones are in principle related , for which we may recall the following point of view : it has generally been assumed that the fundamental laws of nature are governed by a quantum theory of fields since the time when the electromagnetic and also the weak and strong interactions were found to be well described by the qfts .
however , this idea has been challenged from the studies of quantum gravity . as it is known that the einstein s theory of general relativity is not a perturbatively renormalizable qft in the power - counting sense . also because of divergent problem in the perturbative expansion of qfts
, one has raised the issue that underlying theory might not be a quantum theory of fields , it could be something else , for example , string or superstring@xcite .
if this could turn out to be the case , even the quantum electrodynamics ( qed ) and qcd as well the quantum flavor dynamics ( qfd ) of electroweak interactions might be regarded as the effective qfts . in general , effective qfts
are thought to be resulted from low energy approximations of a more fundamental theory .
this may become more clear from the so - called folk theorem by weinberg@xcite .
the theorem states that any quantum theory that at sufficiently low energy and large distances looks lorentz invariant and satisfies the cluster decomposition principle will also at sufficiently low energy look like a quantum field theory .
according to such a folk theorem , there likely exists in any case a characteristic energy scale ( ces ) @xmath0 which can be either a fundamental - like energy scale ( such as the planck scale @xmath13 and/or the string scale @xmath14 in string theory ) or a dynamically generated energy scale ( for instance , the chiral symmetry breaking scale @xmath15 in chiral perturbation theory and the critical temperature for superconductivity ) , so that any effective qfts become meaningful only at a sufficiently low energy scale in comparison with the ces @xmath0 . in general , there is no restriction on the ces @xmath0 .
a particularly interesting case is to run the ces @xmath0 going to infinity . if in this case
, the qfts can remain to be well - defined by performing a renormalization , namely one can formally recover the initial divergent properties of qfts in the perturbative expansion at @xmath16 and consistently eliminate all infinities via an appropriate renormalization of coupling constants and fields in the original theory , we may then mention such kinds of qfts as the underlying renormalizable qfts .
it is not difficult to show that qfts in which there are only yang - mills gauge fields interacting with dirac spinor fields can be viewed as underlying renormalizable qfts . while the einstein s theory of general relativity may be regarded as an effective theory at the sufficiently low energy scale in comparison with the ces which is in the order of magnitude of the planck mass , i.e. , @xmath17 . on the other hand , basing on the idea of renormalization group developed by either wilson@xcite or gell - mann and low@xcite
, one should be able to deal with physical phenomena at any interesting energy scale by integrating out the physics at higher energy scales .
this implies that one can define the renormalized theory at any interesting renormalization scale or the so - called sliding energy scale ( ses ) @xmath1 which is not related to masses of particles or fields and can be chosen to be at any scale of interest , so that the physical effects above the ses @xmath1 are integrated in the renormalized couplings and fields .
with these considerations , it becomes reasonable to conjecture that there should exist an alternative new regularization scheme that can realize the above mentioned attractive properties .
of particular , the regularization for which we are looking must be governed by a physically meaningful ces @xmath0 and also a physically interesting ses @xmath1 , so that the laws of nature can well be described by a quantum theory of fields when the energy scale in the considered phenomena becomes sufficiently lower than the ces @xmath0 , and also the laws of nature at an interesting energy scale can well be dealt with by renormalizing at the ses @xmath1 which can be chosen to be the order of magnitude of the energy concerned in the considered process .
it is this motivation that comes to the purpose of the present paper which may be organized as follows : in section ii , by conceptually defining a set of so - called irreducible loop integrals ( ilis ) , we perform an explicit evaluation for the gauge field vacuum polarization graphs at one - loop order in the non - abelian gauge theory with a general @xmath18 gauge . as a consequence ,
we arrive at two necessary conditions for the regularized quadratically and logarithmically divergent ilis in order to maintain the gauge invariance of non - abelian gauge theories ( or to satisfy the so - called generalized ward identities ) . in analogous to the quantum electrodynamics , the gauge fixing term @xmath19 is found to be unaffected by the loop graphs . while it is unlike the usual case ( such as the case in the dimensional regularization )
, the tadpole graphs of gauge fields are found in the most general case to play an important role for maintaining manifest gauge invariance . at the end of section
, we present a set of consistency conditions which ensures the gauge invariance for all one loop graphs .
we then explore in section iii a new regularization method which satisfies the consistency conditions presented in section ii for the regularized divergent ilis . to be more clear , we begin with an explicit check on the two consistency conditions in the well - known cut - off regularization and dimensional regularization methods .
it is then easily seen why the cut - off regularization destroys the gauge invariance and the dimensional regularization does maintain the gauge invariance .
we then present an alternative new regularization method to achieve the desired four criteria mentioned above .
it is shown in section iv that the regulators ( or regularization quanta ) are indeed decouple from the regularized ilis . consequently ,
we arrive at , through a numerical check , three conjectures on mathematically interesting numbers and functional limits .
we then arrive at a consistent regularization scheme . in section
v , we study in details the overlapping feynmay integrals involved in the general two loop graphs and present an explicit evaluation for the corresponding ilis by adopting the usual feynman parameter method and a newly formulated uv - divergence preserving parameter method .
the latter method ensures that all the feynman parameter integrations contain no uv divergences .
the explicit forms and remarkable features of the ilis enable us to deduce a set of key theorems which contain the factorization theorem and subtraction theorem for overlapping divergences , and the reduction theorem for overlapping tensor type integrals as well as the relation theorem for the tensor and scalar type ilis . in section
vi , we present a general prescription for the new regularization and renormalization method . in particular , an explicit demonstration is provided for two loop diagrams .
the overlapping divergences are shown to be harmless after an appropriate subtraction .
the subtraction process can be made to be analogous to the one in the dimensional regularization .
all the feynman parameter integrations are convergent .
the scheme is seen to respects unitarity and causality .
we show in section vii how to evaluate the ilis of arbitrary loop graphs . after a detailed evaluation for the ilis of three loop graphs , a general procedure for the evaluation of any fold ilis
is presented , which should be practically useful to make a realistic computation for arbitrary loop diagrams . in section viii
, we pay our special attention to the issues on the conformal scaling symmetry breaking , the mass gap genesis in yang - mills gauge theories and the quark confinement and deconfinement .
our conclusions are presented in the last section .
some useful formulae and technical details concerned in the text are presented in appendices .
we start from the lagrangian of gauge theory with dirac spinor fields @xmath20 ( @xmath21 ) interacting with yang - mill gauge field @xmath22 ( @xmath23 ) @xmath24 with @xmath25 here @xmath26 are the generators of gauge group and @xmath27 the structure function of the gauge group with @xmath28 = i f_{abc } t^c $ ] . to quantize the gauge theory ,
it is necessary to fix the gauge by adding the gauge fixing term and introducing the corresponding faddeev - popov ghost term@xcite with the ghost fields @xmath29 . in the covariant gauge ,
the additional lagrangian @xmath30 has been found to take the following form @xmath31 where @xmath32 is an arbitrary parameter .
thus the modified lagrangian is given by @xmath33 based on this modified lagrangian , one can derive the feynman rules@xcite for propagators and vertex interactions@xcite . for simplicity , we begin with our considerations at one - loop level . to find out the gauge symmetry preserving conditions , it is of help to introduce a set of loop integrals which consist of the scalar type ones @xmath34 and tensor type ones @xmath35 where the number ( @xmath36 ) in the subscript labels the power counting dimension of energy momentum in the integrals . here
@xmath37 and @xmath38 are corresponding to the quadratically divergent integrals ( @xmath39 , @xmath40 ) and the logarithmically divergent integrals ( @xmath41 , @xmath42 ) .
we will explicitly see below that all one - loop feynman integrals of vacuum polarization feynman diagrams can be expressed in terms of the above set of integrals by adopting the usual feynman parameter method . in general , all feynamn integrals from the one - particle irreducible ( 1pi ) graphs can be evaluated into the above simple one - fold integrals .
note that the mass factor @xmath43 is regarded as an effective one which depends on the feynman parameters and the external momenta @xmath44 , i.e. , @xmath45 . to be conceptually helpful ,
we may mention the above set of loop integrals as the one - fold irreducible loop integrals ( ilis ) which are evaluated from one loop feynman diagrams . in general , n - fold ilis
are evaluated from n - loop overlapping feynman integrals of loop momenta @xmath46 ( @xmath47 ) and are generally defined as the loop integrals in which there are no longer the overlapping factors @xmath48 @xmath49 which appear in the original overlapping feynman integrals .
it will be shown that any loop integrals can be evaluated into the corresponding ilis by repeatedly using the feynman parameter method ( see appendix a ) and a newly formulated uv - divergence preserving parameter method ( see eq .
( 5.6 ) ) .
we now evaluate the vacuum polarization diagrams of gauge fields to one - loop order .
there are in general four non - vanishing one - loop diagrams ( see figures ( 1)-(4 ) ) .
their contributions to the vacuum polarization function are denoted by @xmath50 ( @xmath51 ) respectively .
the first three diagrams arise from the pure yang - mills gauge theory , their contributions to the vacuum polarization function may be labeled as @xmath52 .
the last diagram is from the fermionic loop and its contribution to the vacuum polarization function is denoted by @xmath53 .
the total contributions to the gauge field vacuum polarization function are then given by summing over all the four diagrams @xmath54 gauge invariance requires that @xmath55 which should be true for any non - abelian gauge group and valid for arbitrary fermion number @xmath56 .
this indicates that both parts @xmath57 and @xmath58 should satisfy the generalized ward identities @xmath59 in terms of the one - fold ilis , the vacuum polarization function @xmath57 from fermionic loop is simply given by ( for more details see appendix a ) @xmath60 \end{aligned}\ ] ] which shows that the gauge invariance is spoiled only by the quadratic divergent ilis .
it then implies that the regularization for which we are looking should lead to the following general condition @xmath61 so that the gauge non - invariant terms caused by the quadratically divergent ilis cancel each other and the vacuum polarization function @xmath57 satisfies the generalized ward identity and maintains the gauge invariance . here
the integrals with superscript @xmath62 means the regularized ilis .
obviously , under this condition the regularized vacuum polarization function @xmath63 from fermionic loop becomes gauge invariant and takes a simple form @xmath64 the gauge field vacuum polarization function @xmath58 in the pure yang - mills gauge theory is much complicated as it receives contributions from three diagrams ( for a more detailed evaluation see appendix a ) . summing over the contributions from the three diagrams and making some algebra , we then obtain @xmath65 \nonumber \\ & + & [ \ \left(5/2 - 2x(1-x ) -3(1-x)(1 - 2x ) \
\right ) p^2 g_{\mu\nu } - \left ( 1 + 4x(1-x)\ \right ) p_{\mu } p_{\nu } \ ] \
i_0 \nonumber \\ & + & \lambda \gamma(3 ) \ [ \ \left(\ ( 1-x ) ( 1 + 2x + 8x^2 ) p^2g_{\mu\nu}/2 - x(2+x-2x^2 ) p_{\mu}p_{\nu}\ \right ) \
i_0 - 4x^3 p^{\rho}p^{\sigma}g_{\mu\nu}i_{0\rho\sigma } \nonumber \\ & - & \left ( \ ( 1 - 2x - 2(3 - 2x)x^2 \ ) p^2 g_{\mu}^{\rho}g_{\nu}^{\sigma } - ( 3x -4x^2 + 4x^3 ) p^{\rho}\ ( g_{\mu}^{\sigma } p_{\nu } + g_{\nu}^{\sigma } p_{\mu } ) \
\right ) \ i_{0\rho\sigma}\ ] \nonumber \\ & + & \lambda \gamma(3 ) \ [ \ ( \ x+x^2 -6x^4 + 4x^5\ ) p_{\mu}p_{\nu } - ( \ x + 2x^2 -3x^3 -4x^4 + 4x^5 ) p^2 g_{\mu\nu } \ ] \ p^2\ i_{-2 } \nonumber \\ & + & \frac{1}{2}\lambda^2 \gamma(4)\ x(1-x)\ [ \ p^4 g_{\mu}^{\rho}g_{\nu}^{\sigma } + p_{\mu}p_{\nu}p^{\rho}p^{\sigma } - p^2 p^{\rho}\ ( g_{\mu}^{\sigma } p_{\nu } + g_{\nu}^{\sigma } p_{\mu } ) \ ] \ i_{-2\rho\sigma } \ \ } \end{aligned}\ ] ] to simplify the above expression , it is useful to notice the following property of the integrals with respect to the feynman parameter @xmath66
@xmath67 = \int_{0}^{1 } dx\ ( 1-x)\ i[x(1-x ) ] = 1/2 \int_{0}^{1 } dx\ i[x(1-x)]\ ] ] where the integrand @xmath68 $ ] is the function of the combination @xmath69 and is invariant under the change of the integral parameter @xmath70 . with the aid of this property
, one can derive more useful identities ( see appendix a ) . to be more useful
, we may formally express the tensor type ilis in terms of the scalar type ones @xmath71 here @xmath72 and @xmath73 are quantities relying on the regularization and expected to be determined from the gauge invariant conditions . adopting the above definitions and identities , the gauge field vacuum polarization function @xmath74 is simplified to be @xmath75\ i_0\ ( p^2g_{\mu\nu } - p_{\mu}p_{\nu } ) \nonumber \\ &
+ & \frac{1}{4}\lambda \gamma(3)\ [ \ 1 + 6x(1-x)(a_0 + 2 ) - 3a_0 \ ] i_{0}\ ( p^2g_{\mu\nu } - p_{\mu}p_{\nu } ) \nonumber \\ & + & \frac{1}{8}\lambda^2 \gamma(4)\ a_{-2}\ x(1-x)\ p^2\ i_{-2}\ ( p^2g_{\mu\nu } - p_{\mu}p_{\nu } ) \nonumber \\ & + & \frac{1}{4}\lambda \gamma(3)\ [ \ 2x(1-x)(a_0 + 2 ) -1\ ] \ i_0\ p_{\mu}p_{\nu } \\ & + & \lambda \gamma(3)\ x(1-x)[\ \left(\ 1 + 4x(1-x ) \right ) p_{\mu}p_{\nu } -\left(\ 1/2 + 6x(1-x ) \right ) p^2 g_{\mu\nu } \ ] p^2\ i_{-2 } \ \ } \nonumber \end{aligned}\ ] ] it is seen that not only the quadratically divergent integrals in the first and last terms , but also the logarimically divergent term ( the fifth term ) can in general destroy the gauge invariance .
to explicitly obtain the gauge invariant conditions , we may further adopt the identity @xmath76\
i_{0}\ = - \int_{0}^{1 } dx\ 2x(1-x)[\ 1 - 4x(1-x)\ ]
\ p^2\ i_{-2}\ ] ] which is obtained by performing partial integration with respect to @xmath66 and using the relevant integral identities presented in appendix a together with the following identity @xmath77 thus the integral in the fifth term of eq.(2.18 ) can be rewritten as @xmath78\ i_0 & = & - 2\ \int_{0}^{1 } dx\ x(1-x)[\ 1- 4x(1-x ) ] \ p^2 \
i_{-2 } \nonumber \\ & + & ( a_0 -1)\ \int_{0}^{1 } dx\ 2x(1-x)\ i_{0 } \end{aligned}\ ] ] substituting the above results into eq.(2.18 ) , the gauge field vacuum polarization function @xmath74 is further simplified to be @xmath79\ i_0\ \nonumber \\ & + & \frac{1}{4}\lambda \gamma(3)\ [ \ \left(\ 1 + 6x(1-x)(a_0 + 2 ) - 3a_0 \right ) i_{0}\ - 2x(1-x ) \left(\ 1 + 12x(1-x)\ \right ) p^2\ i_{-2 } \ ] \nonumber \\ & + & \frac{1}{8}\lambda^2 \gamma(4)\ a_{-2}\ x(1-x)\ p^2\ i_{-2}\ \ } \nonumber \\ & + & g^2 c_1 \delta_{ab}\ \int_{0}^{1 } dx\ \{\ 2(\ 2i_{2\mu\nu } - i_{2}g_{\mu\nu}\ ) + \lambda \gamma(3)\frac{a_0 -1}{2}\ p_{\mu}p_{\nu}\ x(1-x)\ p^2 \ i_{-2}\ \ } \end{aligned}\ ] ] here the gauge non - invariant term is associated with the factor @xmath80 .
it then implies that the new regularization for which we are looking should lead @xmath72 to be unit ( i.e. , @xmath81 ) .
based on the above simplified form @xmath58 , it becomes clear that in order to maintain the gauge invariance for the gauge field vacuum polarization function @xmath74 in the pure yang - mills gauge theories , the new regularization must satisfy the following two conditions @xmath82 where the superscript @xmath62 denotes the regularized ilis .
the first condition for the quadratically divergent ilis is the same as the one required from the gauge invariance of the gauge field vacuum polarization function @xmath83 due to fermionic loops .
the gauge invariance of the vacuum polarization function @xmath84 from non - abelian yang - mills gauge theories requires an additional condition concerning the logarithmically divergent ilis .
so far , we have provided at one - loop order a general proof for two necessary conditions that must be satisfied for any regularization to maintain the gauge symmetry . under these two conditions ,
the regularized whole gauge field vacuum polarization function @xmath85 can explicitly be expressed in terms of gauge invariant form @xmath86\ i_0^r\ \nonumber \\ & + & \frac{1}{2}\lambda \gamma(3)\ [ \ \left(\ 9x(1-x ) - 1 \right ) i_{0}^r\ - x(1-x ) \left(\ 1 + 12x(1-x)\ \right ) p^2\ i_{-2}^r \ ] \nonumber \\ & + & \frac{1}{8}\lambda^2 \gamma(4)\ a_{-2}\ x(1-x)\ p^2\ i_{-2}^r\ \ } \nonumber \\ & - & g^2 4n_f c_2 \delta_{ab } ( p^2g_{\mu\nu } - p_{\mu}p_{\nu})\ \int_{0}^{1 } dx\ 2x(1-x ) i_0^r(m ) \end{aligned}\ ] ] once applying again the identity eq.(2.19 ) , the gauge field vacuum polarization function @xmath87 can be re - expressed into a much simpler gauge invariant form @xmath88 \
i_0^r\ \nonumber \\ & - & n_f c_2\ 8x(1-x)\ i_0^r(m ) - 4c_1 \lambda\ [ \ 1 - 3\lambda a_{-2}/16\ ] \
x(1-x)\ p^2\ i_{-2}^r\ \ \ } \end{aligned}\ ] ] before proceeding , we would like to address the following observations implied from the above general evaluations : first , only uv divergent ilis destroy the generalized ward identities ( or gauge invariant conditions ) for the vacuum polarization function of gauge fields , which strongly suggests the necessity of regularizing the divergent ilis in order to make the divergent loop integrals to be meaningful . as expected , the uv convergent ilis are not constrained from gauge invariant conditions .
second , any consistent regularization must satisfy the necessary conditions resulted from the generalized ward identities ( or gauge invariant conditions ) @xmath89 .
it also becomes clear that under the two necessary conditions the gauge fixing term @xmath19 in non - abelian gauge theories is not affected by the loop graphs , which arrives at the same conclusion as the one in abelian gauge theory .
third , the quadratically divergent terms may not necessarily be a harmful source for the gauge invariance once a rigorous regularization can be found to satisfy the necessary conditions for the regularized quadratically divergent ilis , so that all the regularized quadratically divergent ilis cancel each other .
last but not least , in contrast to the usual case in the dimensional regularization , in the most general case , the tadpole graphs of yang - mills gauge fields actually play an essential role for maintaining the gauge invariance , which can explicitly be seen from the above general evaluation on the vacuum polarization function .
in fact , it is the tadpole graph that leads to the manifest gauge invariant form of the vacuum polarization function even without carrying out any explicit integrations over the feynman parameter @xmath66 and the loop momentum @xmath90 .
here we have only presented the explicit evaluation for the vacuum polarization function of gauge bosons . for all other two - point , three - point and four - point one loop feynman graphs of fermions , gauge and ghost bosons arising from the gauge theory described by the lagrangian eqs.(2.1 - 2.5 )
, it can be shown that they all maintain the gauge invariance and satisfy the generalized ward identities at one - loop level as long as the following conditions for the regularized ilis hold @xmath91 for quadratically divergent ilis , and @xmath92 for logarithmically divergent ilis , as well as @xmath93 for convergent ilis .
where we have used the notation @xmath94 we may mention the above conditions as the general regularization - independent _ consistency conditions _ for maintaining the gauge invariance of theories .
it is easily shown that the dimensional regularization does satisfy the above consistency conditions .
it is of interest to see that once the above consistency conditions for the regularized ilis hold , the divergent structure of the theories can be well characterized by only two regularized scalar type ilis @xmath95 and @xmath96 .
in general , one may not need to evaluate those ilis by any specific regularization scheme . as it has been turned out that in the underlying gauge theories the quadratic divergences arising from the vacuum polarization functions cancel each other , while the logarithmic divergences can be fully absorbed into the redefinitions of the coupling constants and the relevant fields . in this sense ,
one arrives at a regulator free scheme for the underlying theories .
mathematically , one only needs to prove that such a regularization exists and can result in the above consistency conditions among the ilis .
practically , we shall explicitly show in the following sections that there does exist , in addition to the dimensional regularization , an alternative regularization scheme that can lead to the above consistency conditions and in the meantime match to the four criteria stated in the introduction . before ending this section
, we would like to point out that the above consistency conditions hold exactly without any ambiguities , which differs from the so - called consistency relations imposed in the implicit regularization@xcite .
where some local arbitrary counterterms parameterized by ( finite ) differences of divergent integrals arise , they may ( or not ) be determined by the physical conditions .
the reason is that the consistency relations in the implicit regularization are imposed among the basic divergent integrals which are independent of the external momenta .
namely the mass factor in those basic divergent integrals remains the original mass of fields .
this is explicitly seen from the following identity and truncation which have been used in the implicit regularization @xmath97 } = \sum_{j=0}^{n } \frac{(-1)^j ( p_i^2 + 2p_i \cdot k ) ^j}{(k^2 -m^2)^{j+1 } } + \frac{(-1)^{n+1 } ( p_i^2 + 2p_i \cdot k ) ^{n+1}}{(k^2 -m^2)^{n+1 } [ ( k+p_i)^2 -m^2 ] } \ ] ] where @xmath44 are the external momenta and @xmath98 is chosen so that the last term is finite under integration over @xmath90 .
it must be this truncation that causes the local arbitrary counterterms .
we now proceed to explore a possible new regularization which can preserve the basic symmetry principles of the theory and lead to an infinity free renormalization . before doing so
, it is helpful to briefly look at the well - known two simple regaularizations , i.e. , cut - off regularization and dimensional regularization , through checking the consistency conditions resulted from the generalized ward identities ( or gauge invariant conditions ) . the regularized ilis in the cut - off regularization are presented in appendix b. from the explicit forms overthere , one easily sees that @xmath99 \neq \frac{1}{2 } g_{\mu\nu } i_2^r |_{cutoff}\ , \\ i_{0\ \mu\nu}^r |_{cutoff } & = & \frac{1}{4}g_{\mu\nu}\ i_0 |_{cutoff } + \frac{i}{64\pi^2}g_{\mu\nu } \left ( \frac{{\cal m}^2}{\lambda^2 + { \cal m}^2 } - \frac{{\cal m}^4}{2(\lambda^2 + { \cal m}^2)^2 } - \frac{1}{2 } \right ) \neq \frac{1}{4 } g_{\mu\nu } i_0^r |_{cutoff } \end{aligned}\ ] ] where @xmath100 is the cut - off momentum .
obviously , the regularized ilis by the simple cut - off regularization do not satisfy the consistency conditions .
in fact , not only the quadratically divergent terms can destroy the gauge invariance , even the finite terms may spoil the gauge conditions in the simple cut - off regularization .
whereas one may notice that the involved logarithmically divergent terms in the regularized ilis do satisfy the consistency conditions .
this is actually the case realized by the dimensional regularization . from the explicit results given in appendix b for the regularized ilis in the dimensional regularization , it is not difficult to yield , by using the identity @xmath101 , the desired relations @xmath102
it then becomes manifest why the dimensional regularization preserves gauge symmetry . to see the difference between the cut - off regularization and dimensional regularization
, one may make an expansion for @xmath103 .
using the properties of the @xmath104-function @xmath105 with the euler number @xmath106 , one has @xmath107 \\ & & i_0^r |_{dim . }
= \frac{i}{16\pi^2}\ [ \ \frac{2}{\varepsilon } - \gamma_e - \ln ( 4\pi { \cal m}^2 ) + o(\varepsilon ) \ ] \end{aligned}\ ] ] comparing them to the results in the cut - off regularization , the divergences given by @xmath108 must be logarithmic , i.e. , @xmath109 .
this may become more clear from the vanishing integral in the dimensional regularization @xmath110 it is this property that suppresses the quadratic divergence in the dimensional regularization .
actually , the dimensional regularization destroys all the divergences in a power of cut - off momentum because of the vanishing integrals @xmath111 which implies that divergences appearing in any loop integrals are expressed only in terms of the expansion by powers of @xmath108 in the dimensional regularization .
we now turn to investigate the possible new regularization .
it has been seen from the previous section that the divergences actually arise from loop integrals .
it is the uv divergent loop integrals which destroy the generalized ward identities of gauge invariance .
this observation naturally motivates us to propose a new regularization scheme which is supposed to regularize the ilis of loop graphs , i.e. , the whole feynman diagrams , instead of only the propagators carrying out in the pauli - villars inspired methods . unlike the pauli - villars regularization , we are not going to introduce any regulator fields to modify the action of original theory .
alternative to the dimensional regularization , we shall not change the space - time dimension of initial theories .
our motivation is only to modify the very high energy behavior of loop integrals and/or the very low energy behavior of loop integrals if the integrals concern ir divergent .
in fact , we do not well understand the physics at very short distances , especially the physics above the planck scale .
thus it must not be surprised that at infinitely large loop momentum , the feynman diagrams and the corresponding feynman integrals become ill - defined due to the lack of convergence .
they are actually meaningless because of the uv divergent integrals which destroy the basic gauge symmetry principle in gauge theories .
moreover , it is also a hard task in certain qfts to deal with the physics at a very low energy scale when the nonperturbative effects of strong interactions become important at that scale . in particular , the same serious problem may be faced when the feynman integrals also become ill - defined at zero momentum due to the existence of ir divergences .
all of these considerations strongly suggest the necessity of regularization for qfts . here
we are going to present a new regularization without introducing any additional lagrangian formulation to modify the original theory as usually done in the pauli - villars inspried regularization .
what we are doing is only to require the new regularization possessing the most important and necessary properties , namely : the regularization shall make the meaningless divergent ilis to be well - defined and physically meaningful , and also to be applicable to both underlying ( renormalizable ) qfts and effective qfts .
furthermore the new regularization shall preserve the initial divergent behavior and meanwhile maintain the important symmetry principles in the original theory .
more specifically we are going to work out the regularization which will ensure the regularized divergent ilis satisfying the consistency conditions of gauge invariance even when keeping the quadratic ` cut - off ' terms in the loop integrals .
in general , the quadratic ` cut - off ' terms will be found in the new regularization to be harmless for the underlying ( renormalizable ) qfts but very important for the effective qfts . of important
, we shall finally verify that the physical quantities are independent of any regulators ( or regularization quanta ) introduced in the regularization scheme , namely the regulators ( or regularization quanta ) will eventually be decouple from the well - defined regularized theory .
the prescription for the new regularization is simple : supposing that the ilis of loop graphs have been rotated into the four dimensional euclidean space of momentum , we then replace in the ilis the loop integrating variable @xmath112 and the loop integrating measure @xmath113 by the corresponding regularized ones @xmath114_l$ ] and @xmath115_l$ ] @xmath116_l \equiv k^2 + m^2_l\ , \qquad \int d^4 k \rightarrow \int [ d^4 k]_l \equiv \lim_{n , m_l^2 } \sum_{l=0}^{n } c_l^n \int d^4 k \end{aligned}\ ] ] in here @xmath117 ( @xmath118 ) may be regarded as the mass factors of regulators ( or regularization quanta ) . for @xmath119 , we shall have @xmath120 to maintain the original integrals .
if the original integrals are ir divergent , one can set @xmath121 to avoid ir divergent problem .
in general , the mass factors @xmath117 ( @xmath118 ) of the regulators ( or regularization quanta ) are introduced as unknown parameters and their values can be infinitely large .
it is seen that the initial integral is recovered by letting the regulator mass factors @xmath117 go to infinity , i.e. , @xmath122 ( @xmath118 ) . where @xmath123 ( @xmath124 ) are the corresponding coefficients and will be judiciously chosen to modify the short - distance behavior of loop integrals and remove the divergences .
specifically , in order to suppress the divergences at high orders , the coefficients @xmath123 with the initial condition @xmath125 may be chosen to satisfy the conditions @xmath126 so that the regularized integrals of @xmath127 vanish @xmath128_l \
( k^2 + m_l^2)^n \equiv \lim_{n , m_l^2 } \sum_{l=0}^{n } c_l^n \int d^4 k \ ( k^2 + m_l^2)^n = 0 \ \qquad ( n= 0 , 1 , \cdots ) \end{aligned}\ ] ] which coincide with the ones in the dimensional regularization . where the integer @xmath98 counts the corresponding numbers of regulators ( or regularization quanta ) and is in principle a free parameter
. its value can be infinitely large .
practically , the regulator number @xmath98 may also only need to reach a sufficiently large value around which the regularized ilis become insensitive to the choice for its specific values .
finally , we shall prove that the regularized ilis must not be sensitive to the choices for the various mass factors @xmath117 of regulators ( or regularization quanta ) and the corresponding coefficients @xmath123 as well as the regulator number @xmath98 .
we would like to point out that the order of operations for performing the integration over the loop momentum and making the limitation to the regulator mass factors is very important for divergent integrals .
one should first carry out the integration over the loop momentum , then take the infinite limit for the regulator mass factors , ( @xmath122 ( @xmath129 ) . physically
, this order of operations implies that @xmath130 namely , the loop momentum in the euclidean space goes to infinity in a much faster way than the mass factors of regulators ( or regularization quanta ) . before a detailed discussion
, one may notice that the present regularization scheme will result in a quite different property for the quadratically divergent integral in comparison with the dimensional regularization .
this is because in the dimensional regularization , the quadratically divergent integral vanishes for the massless case ( see eq.(3.7 ) ) . while in the present regularization scheme ,
only when the loop momentum @xmath90 becomes sufficiently large with @xmath131 , one then approximately has @xmath132_l \
\frac{1}{k^2 + m_l^2 } \sim \int_{k^2 \gg m_l^2 } [ d^4 k]_l \ \frac{1}{k^2 } \sum_{n=0}^{\infty } \left(\frac{m_l^2}{k^2 } \right)^n = 0 \end{aligned}\ ] ] which will lead to a distinguishable property in the present regularization scheme .
we shall come to a more detailed discussion below .
let us first apply the new regularization scheme to the scalar type divergent ilis @xmath133_l}{(2\pi)^4}\ \frac{1}{-k^2 - \hat{m}_l^2 } \equiv -i\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int \frac{d^4 k}{(2\pi)^4}\ \frac{1}{k^2 + \hat{m}_l^2 } \nonumber \\ & = & \frac{-i}{16\pi^2 } \
\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n\ \hat{m}_l^2 \ln \hat{m}_l^2 \\ i_0^r & = & i\int \frac{[d^4 k]_l}{(2\pi)^4}\ \frac{1}{(-k^2 - \hat{m}_l^2)^2 } \equiv i\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int \frac{d^4 k}{(2\pi)^4}\ \frac{1}{(k^2 + \hat{m}_l^2)^2 } \nonumber \\ & = & - \frac{i}{16\pi^2 } \ \lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n\ \ln \hat{m}_l^2 \end{aligned}\ ] ] with @xmath134 to check whether the above new method satisfies the consistency conditions for maintaining gauge invariance , we further evaluate the corresponding tensor type ilis by the same regularization scheme @xmath135_l}{(2\pi)^4}\ \frac{-k_{\mu}k_{\nu}}{(-k^2 - \hat{m}_l^2 ) ^2 } \equiv -i\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int \frac{d^4 k}{(2\pi)^4}\ \frac{k_{\mu}k_{\nu}}{(k^2 + \hat{m}_l^2 ) ^2 } \nonumber \\ & = & \frac{1}{4}g_{\mu\nu}\ \frac{-i}{16\pi^2 } \
\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int d^2 k \ [ \ 1 + \frac{-2\hat{m}_l^2 } { k^2 + \hat{m}_l^2 } + \frac{\hat{m}_l^4 } { ( k^2 + \hat{m}_l^2)^2 } \ ] \nonumber \\ & = & \frac{1}{2 } g_{\mu\nu}\ \frac{-i}{16\pi^2 } \
\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n\ \hat{m}_l^2 \ln \hat{m}_l^2 \nonumber \\ & = & \frac{1}{2 } g_{\mu\nu}\ i_2^r \\ i_{0\mu\nu}^r & = & i \int \frac{[d^4 k]_l}{(2\pi)^4}\ \frac{-k_{\mu}k_{\nu}}{(-k^2 - \hat{m}_l^2)^3 } \equiv i\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int \frac{d^4 k}{(2\pi)^4}\ \frac{k_{\mu}k_{\nu}}{(k^2 + \hat{m}_l^2)^3 } \nonumber \\ & = & \frac{1}{4}g_{\mu\nu}\ \frac{i}{16\pi^2 } \
\lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n \int d^2 k \ [ \ \frac{1}{k^2 + \hat{m}_l^2 } + \frac{-2\hat{m}_l^2 } { ( k^2 + \hat{m}_l^2)^2 } + \frac{\hat{m}_l^6 } { ( k^2 + \hat{m}_l^2)^3 } \ ] \nonumber \\ & = & \frac{1}{4}g_{\mu\nu}\ \frac{-i}{16\pi^2 } \ \lim_{n , m_l^2}\sum_{l=0}^{n } c_l^n\ \ln \hat{m}_l^2 \nonumber \\ & = & \frac{1}{4}g_{\mu\nu}\ i_0^r \end{aligned}\ ] ] which shows that the method does preserve gauge invariance .
note that in obtaining the above results , only two conditions have actually been used , i.e. , @xmath136 to be able to directly compare with the cut - off regularization and dimensional regularization , and also to explicitly check whether the present new regularization maintains the harmless and important quadratic ` cut - off ' terms , it is of help to find out the manifest expressions for the mass factors @xmath117 of regulators ( or regularization quanta ) and
the corresponding coefficients @xmath123 . for a self consistent reason ,
the explicit forms of @xmath117 and @xmath123 are assumed to be uniquely determined through the conditions given in eq.(3.11 ) .
this requires that for a given @xmath98 the number @xmath137 of conditions should be sufficient in comparison with the unknown variables @xmath138 which consist of n s mass parameters @xmath117 and n s coefficient functions @xmath123 . also for an economic reason , it is supposed that the regularization scheme only involves a minimal set of parameters . with these considerations
, we come to the following simple choice for the mass factors @xmath117 of regulators ( or regularization quanta ) @xmath139 here @xmath140 may be regarded as a basic mass scale for all regulators ( or regularization quanta ) .
the energy scale @xmath1 may be viewed as an intrinsic mass scale and will be seen to provide an ir cut - off energy scale . in the absence of ir divergences , the scale @xmath1 can in principle run to zero @xmath141 . in this case
, there is only one free parameter @xmath140 . in general , even without ir problem the scale @xmath1 can still be introduced to play the role of sliding energy scale ( ses ) .
a particularly interesting case is that a mass gap around the scale @xmath142 may be generated due to strong interactions , if this case happens , the ses @xmath1 is naturally set to be at the mass gap @xmath142 . substituting the above explicit form of @xmath117 to the mass - coefficient conditions given in eq.(3.11 ) for the regulators ( or regularization quanta ) , we arrive at the interesting constraints purely for the coefficients @xmath123 @xmath143 for a given @xmath98 , the coefficients @xmath123
are completely determined by solving @xmath98 s linear equations with the initial condition @xmath144 . the result is explicitly given by @xmath145 which may be regarded as the sign - associated combinations .
more precisely speaking , @xmath146 is the number of combinations of @xmath98 regulators ( or regularization quanta ) , taken @xmath147 at a time . the initial condition @xmath144 is required so that we shall recover the original ilis when no any regulator is introduced . up to now
, we have completely presented an explicit new regularization scheme in which there exists no free parameters when both the basic mass scale @xmath140 and the number @xmath98 of regulators ( or regularization quanta ) are taken to be infinitely large , i.e. , @xmath148 , and the ses @xmath1 is set to be zero @xmath141 .
what we shall do next step is to prove that the regularized ilis are insensitive to the regularization scheme .
more specifically the regularized ilis should eventually be independent of the basic mass scale @xmath140 and the number @xmath98 of regulators ( or regularization quanta ) when they are approaching to the infinitely large limits or practically when they are taken to be sufficiently large . to see that ,
let us apply the above explicit forms for the mass factors @xmath117 of regulators ( or regularization quanta ) and the coefficients @xmath123 to the regularized ilis .
after some algebra , we obtain the following explicit and simple forms for the regularized ilis @xmath149 \ \ } \\ i_0^r & = & \frac{i}{16\pi^2 } \ [ \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + y_0(\frac{\mu^2}{m_c^2 } ) \ ] \\ i_{-2}^r & = & - \frac{i}{16\pi^2 } \
\frac{1}{2\mu^2}[\ 1 - y_{-2}(\frac{\mu^2}{m_c^2 } ) \ ] \end{aligned}\ ] ] where we have introduced the following definitions and notations @xmath150 here @xmath151 and @xmath152 ( @xmath153 ) are the numbers of functional limits @xmath154 \ } \\ & & l_n = \lim_{n}l_n(n ) \equiv - \lim_{n}\sum_{l=1}^{n } c_l^n\frac{n!}{l^n}\ [ \ \sum_{l'=1}^{n } c_{l'}^n\ l ' \ln l ' \ ] ^n \end{aligned}\ ] ] note that the following notations are understood @xmath155 it is manifest that the energy scale @xmath156 sets an ir ` cut - off ' for the case @xmath157 which could occur in a massless theory with imposing on mass - shell conditions , and the energy scale @xmath0 provides an uv ` cut - off ' .
in fact , @xmath0 plays the role of characteristic energy scale ( ces ) and its magnitude is supposed to be known for the considered quantum theories , and @xmath1 plays the role of ses and it can in principle run from @xmath158 to @xmath4 if there is no ir divergent problem and no mass gap generated in the considered theory .
of particular , the method does maintain the quadratic ` cut - off ' term @xmath159 , which makes it different from the dimensional regularization .
though this feature is analogous to the naive cut - off regularization , while it is quite unlike the naive cut - off regularization as the new regularization ensures the consistency conditions for the regularized scalar and tensor type ilis . so that the new method preserves the basic symmetry principles of the theory , such as gauge invariance and lorentz invariance as well translational invariance , which is compatible to the dimensional regularization
it then becomes clear that the uv ` cut - off ' scale @xmath0 and the ir ` cut - off ' scale @xmath3 in the new regularization are completely distinguishable from the ones in the naive cut - off regularization . in the latter case ,
the cut - off momentum scales are introduced by imposing kinematically the upper and lower bound of the integrating loop momentum , which explicitly breaks the translational and lorentz invariance and also destroys the gauge invariance in gauge theories . in the former case , which is in contrast to the latter one ,
the energy scales @xmath0 and @xmath1 are entered intrinsically as two basic mass scales from the regulators ( or regularization quanta ) . for convenience and also to be distinguishable from the naive momentum cut - off , the energy scales @xmath0 and @xmath1 may be mentioned henceforth as intrinsic mass scales .
it is particularly noted that when taking the ces @xmath0 to be infinite , i.e. , @xmath160 , we then recover the initial but well - defined uv divergences of the ilis including the quadratically divergent terms . for non - abelian gauge theories of yang - mills fields interacting with dirac spinor fields ,
as all the quadratic divergent terms cancel each other in the new regularization scheme ( which has explicitly been shown in the previous section for the gauge field vacuum polarization function at one loop order ) , thus the resulting divergent behavior in the regularized gauge theories at @xmath160 is actually equivalent to the one by the dimensional regularization once one makes a simple replacement @xmath161 with taking @xmath162 and @xmath160 .
therefore , such non - abelian gauge theories may be viewed , according to our previous convention , as underlying renormalizable qfts . before ending this section
, we would like to emphasize that the new regularization described above differs in principle from the dimensional regularization .
the differences arise not only from the quadratic ` cut - off ' term , but actually from all the divergences and their origins .
this is because in the dimensional regularization the divergences are intimately correlated to the space - time dimensions .
therefore it is inevitable to meet divergences in the dimensional regularization when one is forced to take the space - time dimension @xmath163 to be four @xmath164 of the real world or to be the one in the original theories . unlike in the dimensional regularization ,
the space - time dimension in the new regularization is the same as the one in the original theory , the initial divergences of ilis are recovered and well - defined by taking the ces @xmath0 to be infinite , @xmath160 . in principle , @xmath0 is a free but well - controlled parameter and its value relies in general on the considered phenomena described by the quantum theories . in this sense , the ces @xmath0 is physically meaningful , so do the quadratic terms or other possible high order terms of @xmath0 . on the other hand , the finite terms in the new regularization
are given by the polynomial of @xmath165 rather than the one of @xmath166 in the dimensional regularization .
of particular , in the new regularization , one can always make on mass - shell renormalization even for a massless theory , like qcd with massless quarks .
this is because one can always choose the ses @xmath1 to be at an interesting finite energy scale .
nevertheless , the new regularization is formally compatible with the dimensional regularization as they satisfy the same consistency conditions .
finally , we want to point out that the prescription of the new regularization formally appears to be similar to the one of pauli - villars , but they are different in the following two aspects : first , the new regularization is applied to the ilis of feynman loop graphs rather than the propagators imposed in the pauli - villars regularization which manifestly spoils gauge invariance .
this is a conceptual extension to the pauli - villars regularization .
second , in the new regularization , the coefficients @xmath123 ( @xmath167 ) are well determined and independent of the regulator mass factors @xmath117 for @xmath168 ( @xmath169 ) . here the basic regulator mass scale @xmath170 and the number @xmath98 of regulators are taken to be infinitely large in the new regularization , which is basically a crucial generalization to the usual pauli - villars regularization in which the regulator mass scale @xmath170 is given @xmath171 and the number of regulators is normally chosen to be only a few limited one , say @xmath172 . as we will show below that for a finite @xmath98 and given @xmath170 , the integrals are regulator dependent ( one can explicitly see from table 1 in appendix c ) . only when taking the number @xmath98 and the mass scale @xmath170 of the regulators to be infinite ( or sufficiently large ) , then the resulting regularized theory becomes independent of ( or insensitive to ) the regulators
one may notice that the basic mass scale @xmath140 of the regulators ( or regularization quanta ) does not appear in the final expressions of the regularized ilis , while it is intrinsically correlated to the regulator number @xmath98 in such a way that : taking @xmath140 and @xmath98 to be infinite , @xmath173 and @xmath174 , but keeping the ces @xmath0 remain held fixed .
this implies that @xmath140 must approach to infinity in terms of an appropriate function of @xmath98 . to obtain its explicit form ,
we begin with the conjecture that @xmath175 here @xmath176 is supposed to be a constant mass scale . substituting this form into the definition of the ces @xmath0 ,
we arrive at the relation @xmath177 with @xmath178 being given by @xmath179 here the function @xmath180 is purely determined by the regulator number @xmath98 and supposed to have a finite limit at @xmath181 .
so far the regularized ilis are completely characterized by the three type of functional quantities @xmath180 , @xmath182 and @xmath183 ( @xmath184 ) , they are solely determined by the regulator number @xmath98 .
it then becomes clear that in order to prove the regularized ilis to be independent of the regulators ( or regularization quanta ) , one actually only needs to show that the regularized ilis are not sensitive to the regulator number @xmath98 at sufficiently large @xmath98 .
this is then equivalent to check whether the functional quantities @xmath180 , @xmath182 and @xmath183 approach to certain finite limits at @xmath174 , namely whether @xmath178 , @xmath151 and @xmath152 are eventually reached to the constant numbers . to be manifest ,
we perform , instead of an analytic proof , a numerical evaluation .
the numerical analysis is presented in appendix c .
the numerical results given in table 1 of appendix c truly show that for a sufficiently large @xmath98 the functions @xmath180 , @xmath182 and @xmath183 do approach to certain values which become less sensitive to the changes of the number @xmath98 once @xmath98 is going to be sufficiently large .
in general , we are led to the following three conjectures which have been checked by an appropriate functional fitting method up to the precision of @xmath185 ( for more details , see appendix c ) . + conjecture 1 .
@xmath186 which implies the following interesting functional limit @xmath187 conjecture 2 .
@xmath188 \right ) \nonumber \\ & = & 0.5772\cdots = \gamma_e \end{aligned}\ ] ] namely @xmath151 is equal to the euler number @xmath189 .
it then implies another interesting functional limit @xmath190 conjecture 3 .
@xmath191 which indicate the functional limits @xmath192 when the above conjectures hold , the functions @xmath193 defined in the previous section via the polynomials can now be expressed by the following simple functions @xmath194 it is then clear that the regulators ( or regularization quanta ) are indeed decouple from the regularized ilis .
furthermore , the consistency conditions eqs.(2.27 - 2.33 ) do hold in the new regularization ( one can easily check that by using the useful integrals given in appendix d ) . therefore it can be concluded that we do arrive at a consistent description for the new regularization scheme .
obviously , an analytical proof for the above conjectures must be very helpful and important .
it may also provide deep insights in mathematics .
the gauge symmetry preserving and infinity free regularization described above is well applied to regularize one loop graphs and the relevant ilis . in order to obtain a consistently regularized theory , such a regularization should be applicable to all loop graphs . furthermore , though the regularized loop graphs become well - defined and infinity free , the coupling constants and the fields in the quantum theories still need to be renormalized .
the only difference is that the subtraction terms become finite quantities instead of infinite ones .
the procedure of such subtractions ( or the so - called renormalization ) must be consistently carried out order by order in perturbative expansion .
in fact , whether the subtraction terms in the renormalization are finite or infinite , in a given order any new subtraction terms must be finite polynomials in external momenta in order to maintain unitarity of the theory .
such a statement should in general be independent of any regularization .
therefore , one shall be able to prove this statement in a more general way without involving any regularization . to be specific and also for an explicit demonstration , we will treat in this section the feynman integrals involved in two loop diagrams .
the generalization to more loop graphs is straightforward .
it is different from one loop graphs , two loop diagrams require us to make a careful treatment for overlapping divergences .
eventually , we must show that after appropriate subtractions of one loop divergences , the coefficient functions of all the divergent terms are finite polynomials in the external momenta .
namely one has to prove the theorem that all eventual divergences must be harmless divergences .
to arrive at this purpose , we shall first prove that overlapping divergences can completely be factorized .
more specifically we shall show that the overall divergences of two loop graphs and the sub - integral divergences corresponding to one loop graphs can be well separated .
it will be seen below that all of these properties will become manifest after performing the evaluation of two - fold ilis involved in two loop diagrams .
the explicit demonstration in the previous section shows that all divergences in one loop diagrams are harmless ones .
in fact , as all one loop integrals can be expressed in terms of the one - fold ilis by using the feynman parameter method , therefore the divergences are completely characterized by the one - fold ilis .
as the coefficient functions of all divergent one - fold ilis are finite polynomials of external momenta and masses , therefore all the divergences at one loop level are truly harmless ones .
we now come to study two loop graphs .
consider first the scalar type overlapping feynman integrals involved in the general two loop diagrams ( i.e. , the so - called general @xmath195 diagrams @xcite ) @xmath196 the above integral has been given in the euclidean space of momentum through an analytic wick rotation . here
@xmath197 since if one of them is zero , the integral is no longer an overlapping one and it actually becomes a factorizable one . in the gauge theories
, the usual power counting rule shows that the divergent overlapping integrals of two loop diagrams are at most quadratically divergent , i.e. , @xmath198 .
the above form of integrals can be regarded as the most general one for two loop graphs .
here we have supposed that one can always apply the feynman parameter method to all momentum factors @xmath199 , so that all propagators concerning @xmath199 and external momenta @xmath200 ( @xmath201 ) are combined into a single one .
similarly for the propagators concerning @xmath202 and @xmath203 .
therefore , the three mass factors @xmath204 , @xmath205 and @xmath206 are in general the functions of masses @xmath207 and external momenta @xmath200 ( @xmath201 ) @xmath208 based on the usual power counting rule , we may first recall some useful definitions following the ref.@xcite : ( i ) the sub - integral over @xmath199 is said to be convergent or divergent according to @xmath209 or @xmath210 , similarly , the sub - integral over @xmath203 is said to be convergent or divergent according to @xmath211 or @xmath212 ; ( ii ) the overall integral of @xmath213 diagram is said to be overall convergent or overall divergent according to @xmath214 or @xmath215 ; ( iii ) a harmless divergence is a divergence with its coefficient functions a polynomial of finite order in the external momenta . in the gauge theories , the overall divergence of a nontrivial overlapping integral is at most quadratic , thus @xmath216 . in two loop graphs
, we shall also involve the general tensor type feynman integrals @xmath217 similar to the one loop case , by adopting appropriate parameter methods , the tensor type feynman integrals can eventually be expressed in terms of the scalar type integrals through the aid of the metric tensor @xmath218 and the external momentum vector @xmath219 .
we shall first focus on the general scalar type feynman integrals @xmath220 and come back to the tensor type feynman integrals at the end of this section .
though the above general forms of feynman integrals may be more concise than the original ones due to the uses of feynman parameter method , they are not yet the desired ilis because of the overlapping momentum factor @xmath221 . to obtain the ilis of two loop graphs , according to the definition of ilis , we shall further adopt appropriate parameter methods to treat the factor @xmath221 . as the first step , we still apply the usual feynman parameter method to the sub - integral over @xmath199 ( one can also consider first the sub - integral over @xmath203 as the two sub - integrals are symmetric with the exchanges @xmath222 , @xmath223 , @xmath224 and @xmath225 , for convention , we will henceforth make the treatments according to the order @xmath199 , @xmath203 , @xmath226 ) . taking the feynman parameter variables @xmath227 and @xmath228 ,
the sub - integral over @xmath199 becomes a quadratic ( see appendix a ) @xmath229^{\alpha + \gamma } \left(k_2 ^ 2 + { \cal m}_2 ^ 2\right)^{\beta } } \end{aligned}\ ] ] with definition @xmath230 where we have made a momentum shift @xmath231 .
to further combine the denominators into a single quadratic , it is found to be of help to adopt an alternative identity @xmath232^{\alpha + \beta } } \ ] ] the advantage of this identity is that once @xmath233 approaches to infinite in a way similar to @xmath234 , i.e. , @xmath235 , the divergent structure of the integral coincides with the one of the term @xmath234 . for convenience
, we may mention this parameter method as an uv - divergence preserving parameter method , and the parameter @xmath233 as the corresponding uv - divergence preserving integrating variable .
its important consequence is that by appropriately applying this uv - preserving parameter method to the combined propagators obtained by carefully using the usual feynman parameter method , one can factor out all the uv divergences , so that the usual feynman parameter integrations contain no uv divergences .
only ir divergences could be hidden in the feynman parameter integrations .
this is crucial point for analyzing the overlapping divergences .
in fact , it will be shown next section that both uv and ir divergences can be simultaneously regularized in the new regularization . thus all the feynman parameter integrals are convergent .
taking @xmath236 and @xmath237 and making a momentum shift @xmath238 , we then obtain from eqs.(5.4 ) and ( 5.6 ) that @xmath239^{\alpha + \beta + \gamma } } \end{aligned}\ ] ] it is clear that once @xmath233 approaches to infinite in a way similar to @xmath240 , i.e. , @xmath241 , the usual power counting rule implies that the resulting divergent structure does coincide with the one in the original sub - integral over @xmath199 . in other word
, the uv divergent property of the sub - integral over @xmath199 will be characterized by the one of the integral over @xmath233 .
it will become more clear below from the explicit form of ilis . in general , @xmath242 ,
the sub - integral over @xmath199 becomes convergent and can be carried out safely .
it is straightforward to perform the integration over @xmath199 , which leads to the result @xmath243 where we have factored out the coefficient factor @xmath244 $ ] of the momentum square @xmath245 and introduced the definitions @xmath246 it is useful to change the dimensionless integrating variable @xmath233 into the dimensionful one @xmath247 . here @xmath248 with dimension of mass square is introduced as an universal energy scale . in general , it only requires @xmath249 in order to respect unitarity .
typically , one may take @xmath250 to reduce unnecessary parameters .
the variable @xmath251 may be regarded as a momentum - like one in connecting with the initial loop momentum @xmath199 . though @xmath251 can not wholly be compatible with the initial momentum @xmath240 , whereas the integral over @xmath251 does maintain the one - loop divergent behavior of the initial loop momentum @xmath199 . using the new integrating variable and noticing the formal relation of the integrating measure @xmath252 we can rewrite the above integral eq.(5.8 ) into the following form @xmath253 where @xmath254 is an overall one - fold ili involved in two loop graphs @xmath255 with @xmath256 according to the definition of ilis , the above resulting loop integrals may be called as the scalar type two - fold ilis . basing on such two - fold ilis and noticing the interesting property that @xmath257 vanishes in the infinite limit of @xmath251 , i.e. , @xmath258 we arrive at the following important observations : \(i ) for the simple case that @xmath259 and @xmath260 , it becomes manifest that the uv divergent behavior of the two - fold ilis at @xmath261 and @xmath262 coincides with the one of the original integral .
this explicitly shows that the one - loop uv - divergence preserving sub - integral over the momentum - like variable @xmath251 does characterize the high energy behavior of the sub - integral over the loop momentum @xmath199 in the initial integral .
\(ii ) the sub - integral over the loop momentum @xmath203 turns out to be an overall one - fold ili which is correlated to the sub - integral over the momentum - like variable only via the @xmath251-dependent mass factor @xmath263 .
\(iii ) when @xmath261 , the two sub - integrals over @xmath203 and @xmath264 become factorized ones due to the property @xmath265 at @xmath261 .
thus the divergent properties of the initial two loop integral can completely be described by the ones of the divergence factorizable two - fold ilis .
\(iv ) the most divergent behavior of the sub - integral over @xmath264 is governed by the power counting of @xmath266 which reflects the divergence of one - loop sub - diagram , and the divergent property of the sub - integral over @xmath203 in the ilis is solely characterized by the power counting of @xmath267 which actually describes the overall divergence of the general @xmath195 two loop diagrams .
\(v ) the external momentum dependence is contained only in the mass factor @xmath263 of the sub - integral over @xmath203 . at the divergent point of
the sub - integral over @xmath264 ( @xmath268 ) , the possible external momentum dependence is solely given by the mass function @xmath205 which appears only in the sub - integral over @xmath203 , therefore such a harmful divergence is expected to be eliminated by one subtraction term . from the above observations , it becomes clear that in the two - fold ilis the overall divergences of two loop graphs are completely factorized from the sub - integral divergences of one loop graphs .
consequently , we are able to deduce the following theorems without involving any regularization .
* theorem i * ( factorization theorem for overlapping divergences ) .
overlapping divergences which contain divergences of sub - integrals and overall divergences in the general feynman loop integrals become completely factorizable in the corresponding ilis .
this theorem is the crucial one to treat the problem of overlapping divergences involved in two and more loop diagrams . with the observations
( iv ) , ( iii ) and ( ii ) in the ilis ,
its proof becomes manifest .
it is actually a direct consequence of the ilis .
therefore , the whole demonstration of the theorem is equivalent to the evaluation of ilis from the general overlapping feynman integrals of loop graphs .
more explicit demonstrations will be presented in next section from the regularized ilis .
* theorem ii * ( subtraction theorem for overlapping divergences ) .
the difference of the general feynman loop integral with the subtraction term corresponding to the divergent sub - integral contains only harmless divergences .
this theorem is the central one to obtain a consist theory .
obviously , the factorization theorem ( theorem i ) becomes crucial to yield this theorem .
its proof is straightforward from the observations ( i)-(v ) in the corresponding ilis of the general feynman loop integrals .
let us present the simple demonstration based on the observations from the two - fold ilis .
note that as the external momentum dependence in the two - fold ilis only appears in the mass factor @xmath263 of the overall one - fold ili over the loop momentum @xmath203 , and also as the sub - integral over the momentum - like variable @xmath251 has the same one - loop uv divergent structure as the one over the loop momentum @xmath199 , one only needs to introduce one subtraction term of the sub - integral over the loop momentum @xmath199 @xmath269 where the superscript @xmath270 denotes the subtraction term . with particularly noticing the property @xmath271 at @xmath261 , it then becomes clear that the difference of the integrals , i.e. , ( @xmath272 ) , contains only harmless divergences .
its manifest demonstration will be given in next section for the regularized ilis .
* theorem iii * ( harmless divergence theorem ) .
if the general loop integral contains no divergent sub - integrals , then it contains only a harmless single divergence arising from the overall divergence .
this theorem may be deduced from the theorems i and ii .
its proof is also obvious with the observations ( iv ) , ( iii ) and ( ii ) in the ilis .
specifically , as @xmath273 and @xmath274 the integral over the momentum - like variable @xmath251 is convergent in the two - fold ilis . only the integral over the loop momentum @xmath203 may contain divergence .
as the integral over the loop momentum @xmath203 characterizes the overall divergence of the initial loop integral and it is actually an overall one - fold ili , also as the sub - integral over the momentum - like variable @xmath264 contains no external momentum - dependence , we then come to the statement in the theorem .
* theorem iv * ( trivial convergence theorem ) .
if the general loop integral contains no overall divergence and also no divergent sub - integrals , then it is convergent .
this theorem is really trivial and it is presented only for completeness .
we now turn to the tensor type feynman integrals .
it is not difficult to arrive at the following theorems : * theorem v * ( reduction theorem for overlapping tensor type integrals ) .
the general overlapping tensor type feynman integrals of arbitrary loop graphs are eventually characterized by the overall one - fold tensor type ilis of the corresponding loop graphs .
this theorem is the key theorem for the generalization of treatments and also for the prescriptions from one loop graphs to arbitrary loop graphs . from this theorem , it is not difficult to deduce the following theorem * theorem vi * ( relation theorem for tensor and scalar type ilis ) . for any fold tensor and scalar type ilis , as long as their power counting dimension of the integrating loop momentum are the same , then the relations between the tensor and scalar type ilis are also the same and independent of the fold number of ilis .
this theorem is crucial to extend the consistency conditions of gauge invariance from divergent one - fold ilis ( or one loop graphs ) to more fold ilis ( or more closed loop graphs ) .
in here we provide an explicit proof for the two loop case .
the extension of the proof to more closed loops is obvious . repeating the evaluation similar to the one for the scalar type two - fold ilis
, one can easily observe that all the tensor type feynman integrals with structures such as @xmath275 , @xmath276 and @xmath277 will be evaluated into the non - trivial tensor type ilis with tensor structures given only by @xmath277 plus the tensor type integrals which are constituted from the scalar type ilis and the metric tensor @xmath218 as well as the external momentum vector @xmath219 .
this can easily be seen from the fact that in evaluating the ilis all involving operations only concern the simple loop momentum shifts @xmath278 together with the integration over the loop momentum @xmath199 . here
the integration is well - defined as the sub - integral over @xmath199 becomes convergent after applying for the uv - divergence preserving parameter method .
we finally find that the non - trivial tensor type ilis in two loop graphs are in general given by @xmath279 where @xmath280 is a polynomial function of feynman parameter @xmath66 and its form only relies on the considered tensor structure . here
the tensor type ilis @xmath281 are defined as a non - trivial overall one - fold tensor type ilis with an effective mass factor @xmath263 @xmath282 the superscript @xmath283 means one - fold .
it is obvious that @xmath284\ \mu\nu \cdots}^{(2 ) } } { i _ { [ \delta ] } ^{(2 ) } } = \frac{i_{[\delta]\ \mu\nu \cdots}^{(1 ) } } { i_{[\delta]}^{(1 ) } } \end{aligned}\ ] ] where @xmath285 $ ] labels the power counting dimension of momentum .
it can be generalized to any fold ilis @xmath286\ \mu\nu \cdots}^{(n ) } } { i_{[\delta]}^{(n ) } } = \frac{i_{[\delta]\ \mu\nu \cdots}^{(1 ) } } { i_{[\delta]}^{(1 ) } } \end{aligned}\ ] ] here the superscript @xmath287 labels the fold number of ilis and its value is arbitrary .
the above explicit forms of the tensor type ilis provide the desired results and complete the proof for the theorem v and theorem vi .
it is clearly seen that we eventually only need to consider the non - trivial overall one - fold tensor type ilis .
all of the above observations and the deduced theorems imply the importance of evaluating the ilis of loop graphs .
it has been seen in the previous section that the problem of overlapping divergences is not relevant to any regularization . in the section
ii , it has been shown at one - loop level that the generalized ward identities ( or gauge invariant conditions ) are in general spoiled only by the divergent integrals .
this implies meaningless of divergent integrals and suggests the necessity of regularizing the divergent integrals .
consequently , consistency conditions between the regularized scalar and tensor type ilis have been resulted from the generalized ward identities at one - loop level . in section iii
, it has been demonstrated that the consistency conditions can provide stringent constraints on the regularization methods . in this section
, we shall apply , as a practical computation and also an explicit check , those theorems obtained in the previous section to the regularized ilis . in fact , by explicitly carrying out the relevant integrations in the regularized ilis , we shall arrive at an independent verification on those theorems .
we shall first generalize the regularization described in the section iii for the one - fold ilis to the two- and n - fold ilis with @xmath137 being arbitrary .
the general prescription for the new regularization method is simple : \(i ) analytically rotate the four dimensional minkowski space into the four dimensional euclidean space of momentum by using wick rotation .
\(ii ) appropriately evaluate feynman integrals of loop graphs into the corresponding ilis by adopting the usual feynman parameter method and the newly formulated uv - divergence preserving parameter method .
\(iii ) universally replace in the ilis the loop momentum square @xmath112 and the corresponding loop integrating measure @xmath113 as well as the uv - divergence preserving momentum - like variable @xmath288 ( or the uv - divergence preserving integrating parameter @xmath233 ) and the corresponding integrating measure @xmath289 ( or @xmath290 ) by the regularizing ones @xmath114_l$ ] and @xmath115_l$ ] as well as @xmath291_l$ ] ( or @xmath292_l$ ] ) and @xmath293_l$ ] ( or @xmath294_l$ ] ) , i.e. , @xmath295_l \equiv k^2 + m^2_l= k^2 + \mu_s^2 + l m_r^2\ , \qquad \int d^4 k \rightarrow \int [ d^4 k]_l \equiv \lim_{n , m_r } \sum_{l=0}^{n } c_l^n \int d^4 k \ , \\ & & \hat{k}^2 \rightarrow [ \hat{k}^2]_l \equiv \hat{k}^2 + m_l^2 = \hat{k}^2 + \mu_s^2 + l m_r^2\ , \qquad \int \frac{d^4 \hat{k}}{\hat{k}^2 } \rightarrow \int [ \frac{d^4 \hat{k}}{\hat{k}^2}]_l \equiv \lim_{n , m_r } \sum_{l=0}^{n } c_l^n \int \frac{d^4 \hat{k}}{\hat{k}^2 } \\ & & or \nonumber \\ & & \quad u \rightarrow [ u]_l \equiv u + m_l^2/q_o^2 = u + ( \mu_s^2 + l m_r^2)/q_o^2 \ , \qquad \int du \rightarrow \int [ du]_l \equiv \lim_{n , m_r } \sum_{l=0}^{n } c_l^n \int du \nonumber \end{aligned}\ ] ] where the coefficient function @xmath123 is given by eq .
( 3.21 ) and the mass scale @xmath170 is characterized via eqs.(4.1 - 4.4 ) .
the above prescription should be applicable to any fold ilis . here
one needs to distinguish the difference of the integrating measures between the loop momentum @xmath90 and the momentum - like variable @xmath296 . in general , the ses @xmath1 is taken to be a finite energy scale in order to avoid possible ir problem , so that one can always make on mass - shell renormalization even for a massless theory . with the above prescription , the two - fold ilis in eq .
( 5.18 ) are simply regularized into the following integrals @xmath297_l \ \sum_{i=0}^{\beta -1 } c_i^{\beta -1 } \frac{(x(1-x)q_o^2)^{i}}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) ^{\alpha + \gamma -1 + i } } \nonumber \\ & & \int [ d^4 k_2]_{l ' } \frac{1 } { ( \ k_2 ^ 2 + { \cal m}_2 ^ 2 + m_{l'}^2 + \mu^2_{\hat{k}_1 ^ 2 + m_{l}^2 } \ ) ^{\alpha + \beta + \gamma -2 } } \end{aligned}\ ] ] where the superscript @xmath62 means regularized ilis . based on the factorization theorem ( theorem i ) and the subtraction theorem ( theorem ii ) for overlapping divergences as well as the harmless divergence theorem ( theorem iii ) and the trivial convergence theorem ( theorem iv ) presented in the previous section , we shall be able to decompose the regularized ilis into the following general form @xmath298 where the superscripts ` @xmath299 ' and ` @xmath300 ' represent the regularized divergent and convergent ilis respectively , and the numbers ( 1 ) and ( 2 ) in the superscripts label the one - fold and two - fold ilis respectively .
the possible harmful divergences must appear only in the first term with double divergences in the two - fold ilis . to explicitly check the above decomposition
, we may first carry out the integration over the loop momentum @xmath203 as it is actually an overall one - fold ilis .
one can directly read off the results from appendix d for the regularized one - fold ilis .
we consider first the case in which the overall integral is logarithmically divergent , i.e. , @xmath301 .
the integration over @xmath203 results in the following explicit form @xmath302_l \
\sum_{i=0}^{\beta -1 } c_i^{\beta -1 } \frac{(x(1-x)q_o^2)^{i}}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) ^{\alpha + \gamma -1 + i } } \nonumber \\ & & \{\ [ \ \ln \frac{m_c^2}{\mu_m^2 } - \gamma_w + y_0(\frac{\mu_m^2}{m_c^2 } ) \ ] - [ \ \ln(\ 1 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}/\mu_m^2\ ) - z_0(\hat{k}_1 ^ 2 + m_l^2 ) \ ] \ \ } \ , \end{aligned}\ ] ] we consider next the case in which the overall integral is quadratically divergent , i.e. , @xmath303 .
the integration over @xmath203 leads to the result @xmath304_l \
\sum_{i=0}^{\beta -1 } c_i^{\beta -1 } \frac{(x(1-x)q_o^2)^{i}}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) ^{\alpha + \gamma -1 + i } } \nonumber \\ & & \{\ m_c^2 - \mu_m^2\ [ \
\ln \frac{m_c^2}{\mu_m^2 } - \gamma_w + 1 + y_2(\frac{\mu_m^2}{m_c^2 } ) \ ] \nonumber \\ & & - \mu^2_{\hat{k}_1 ^ 2 + m_l^2}\ [ \ \ln \frac{m_c^2}{\mu_m^2 } - \gamma_w + 1 + y_2(\frac{\mu_m^2}{m_c^2 } ) \ ] \nonumber \\ & & + [ \ \mu_m^2 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}\ ] [ \ \ln(\ 1 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}/\mu_m^2\ ) - z_2(\hat{k}_1 ^ 2 + m_l^2 ) \ ] \ \ } \end{aligned}\ ] ] here the functions @xmath305 and @xmath306 are defined as @xmath307 from the explicit forms of the functions @xmath308 and @xmath309 ( see eqs.(3.26 ) and ( 3.28 ) ) , it is easily seen that in the high energy limit the functions @xmath305 and @xmath306 are approaching to zero in terms of the inverse powers of @xmath251 , i.e. , @xmath310 it will be seen that the above properties are crucial for the treatment of overlapping divergences . to be more clear ,
consider first the simple case with @xmath259 and @xmath311 .
applying the usual power counting rule to the general overlapping loop integrals , one sees that the overall integral in this case is logarithmically divergent ( @xmath312 ) , the sub - integral over @xmath199 is also logarithmically divergent ( @xmath313 ) , and the sub - integral over @xmath203 is superficially convergent ( @xmath314 ) . turning to the corresponding ilis , the usual power counting rule shows that the most divergent part of the sub - integral over the momentum - like variable @xmath264 remains logarithmically one ( @xmath315 ) , whereas the sub - integral over @xmath203 in the corresponding ilis becomes logarithmically divergent one ( @xmath316 ) as it characterizes the overall divergence of the overlapping loop integrals .
one can easily carry out the integration for the regularized divergent part in the integral eq.(6.5 ) .
indeed , we find that the regularized ilis in this case can be written into the following form @xmath317 where the involving regularized one - fold ilis are given by @xmath318 \\ i_{0}^{(1)rd } & = & \pi^2 [ \ \ln \frac{m_c^2}{\mu_m^2 } - \gamma_w + y_0(\frac{\mu_m^2}{m_c^2 } ) \ ] \\
\hat{i}_{0}^{(1)rc } & = & -\pi^2\int_0 ^ 1 dx\ \frac{x(1-x)q_o^2 } { \mu_{q_o}^2 } [ \ 1 - y_{-2 } ( \frac{\mu_{q_o}^2}{m_c^2 } ) \ ] \end{aligned}\ ] ] with @xmath319 the regularized convergent part of the two - fold ilis is evaluated by @xmath320_l \
\frac{\hat{k}_1 ^ 2 + m_l^2}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) ^2 } \nonumber \\ & & [ \ \ln(\ 1 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}/\mu_m^2\ ) - z_0(\hat{k}_1 ^ 2 + m_l^2 ) \ ] \ \end{aligned}\ ] ] consider next the case with @xmath259 and @xmath321 . in this case , the general overlapping loop integral contains two logarithmically divergent sub - integrals over @xmath199 ( @xmath322 ) and @xmath203 ( @xmath323 ) .
the overall integral is quadratically divergent . in the corresponding ilis ,
the sub - integral over the momentum - like variable @xmath264 ( @xmath324 ) is logarithmically divergent , whereas the sub - integral over @xmath203 in ilis becomes quadratically divergent ( @xmath325 ) as it characterizes the overall divergences .
similar to the first case , it is not difficult to carry out the integration concerning the regularized divergent parts and yield the result with the following form @xmath326 where the regularized quadratically divergent one - fold ili and the regularized convergent one - fold ili are given by @xmath327 \ } \\
\hat{i}_{2}^{(1)rc } & = & -\pi^2\ \int_0 ^ 1 dx\ \int [ d \hat{k}_1 ^ 2 ] _
\frac { \mu^2_{\hat{k}_1 ^ 2 + m_l^2}}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) } \end{aligned}\ ] ] the regularized convergent two - fold ili is evaluated by @xmath328_l \ \frac{\mu_m^2 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}}{(\ \hat{k}_1 ^ 2 + x(1-x)q_o^2 + m_l^2 \ ) } \nonumber \\ & & [ \ \ln(\ 1 + \mu^2_{\hat{k}_1 ^ 2 + m_l^2}/\mu_m^2\ ) - z_2(\hat{k}_1 ^ 2 + m_l^2 ) \ ] \end{aligned}\ ] ] from the above explicit forms , it is easily seen that when taking the ces @xmath0 to be infinite , i.e. , @xmath329 , we then recover the initial divergent behaviors of the ilis ( or the ones of the corresponding general overlapping feynman integrals ) .
we shall be more interested in the possible harmful divergences contained in the above results at @xmath329 .
one notices that the divergences associated with possible non - polynomial external momentum dependent coefficient functions could occur only in the double divergent terms @xmath330 and @xmath331 .
this is because only the regularized one - fold ilis @xmath332 and @xmath333 contain the logarithmic term @xmath334 , in which the mass factor @xmath205 is in general the function of external momenta .
it is remarkable to note that the double divergent terms in the two cases contain a common one - fold ili , i.e. , @xmath335 .
this implies that the two harmful divergences at @xmath329 are actually the same .
therefore , we only need introduce one subtraction term to make them becoming harmless .
consider now the regularization to the subtraction term for the sub - integral over the loop momentum @xmath199 ( see eq.(5.17 ) ) @xmath336_l \frac{1}{(k_1 ^ 2 + m_o^2 + m_l^2 ) ^{\alpha + \gamma } } \int [ d^4 k_2 ] _ { l ' } \frac{1}{\left(k_2 ^ 2 + { \cal m}_2 ^ 2 + m_{l'}^2 \right)^{\beta } } = i_{\alpha\gamma}^{(1)r } i_{\beta}^{(1)r } \end{aligned}\ ] ] which is actually factorized and can easily be carried out .
for the case with @xmath259 and @xmath311 , we have @xmath337 and for the case with @xmath259 and @xmath321 , we yield @xmath338 where the explicit subtraction point @xmath339 is chosen .
it then becomes obvious that the differences @xmath340 contain only harmless divegencies at @xmath329 . here
@xmath341 is the additional convergent function arising from the subtraction @xmath342 \end{aligned}\ ] ] which is independent of the external momenta and vanishes if choosing @xmath343 , @xmath344 the above demonstrations for the two interesting cases have provided an explicit verification on the factorization theorem ( theorem i ) and the subtraction theorem ( theorem ii ) for overlapping divergences , as well as on the harmless divergence theorem ( theorem iii ) and the trivial convergence theorem ( theorem iv ) . practically speaking
, it provides an explicit demonstration on the regularization and renormalization prescriptions for two loop graphs .
obviously , such prescriptions can easily be generalized to more closed loops . to establish the consistency of the new regularization
, we shall further verify the generalized ward identities ( or gauge invariant conditions ) at two and more loop level .
as we have shown in section ii that only the divergent integrals could spoil the generalized ward identities .
on the other hand , the generalized ward identities only require the regularized tensor and scalar type divergent ilis to satisfy a set of consistency conditions .
therefore the verification is equivalent to check whether the consistency conditions between the regularized tensor and scalar type divergent ilis are still preserved at two and more loop level . for this purpose ,
one only needs to apply the theorems v and vi ( i.e. , the reduction theorem for overlapping tensor type integrals and the relation theorem for the tensor and scalar type ilis ) to the regularized tensor and scalar type divergent ilis .
obviously , the desired result can easily be achieved from those two theorems .
last but not least , it must be emphasized that the procedure respects unitarity and causality .
this is because : ( i ) the usual feynman parameter method and the newly formulated uv - divergence preserving parameter method do not violate unitarity and causality , this process is independent of regularization and can be made in any regularization schemes ; ( ii ) the evaluation of ilis involves just the shifts of integrating loop momenta and the convergent integrals over the loop momenta .
this procedure is well defined and ensured by the translational invariance and the safety of convergent integrals ; ( iii ) the regulators act on the whole feynman integrals of loop graphs rather than on the propagators , which distinguishes to some pauli - villars inspired regularization schemes .
the feynman integrals are well defined in the euclidean momentum space ; ( iv ) unlike the usual pauli - villars inspired regularization schemes , the mass factors and numbers of the regulators ( or regularization quanta ) in the new regularization are eventually taken to be infinitely large and decouple from the theory .
no additional singularities appear via the feynman propagators ; ( v ) also unlike the bphz subtraction scheme which is based on expanding around an external momentum , the new regularization and renormalization scheme introduces two intrinsic mass scales , i.e. , the ces @xmath0 and ses @xmath1 , to characterize the uv and ir behaviors of the feynman amplitudes , so that the structure of the corresponding amplitudes involving the external momenta is not changed .
this can be seen explicitly from the evaluation of the vacuum polarization function presented in section ii and appendix a ; ( vi ) it has been shown that in a given order any new subtraction terms are finite polynomials in external momenta , which is also necessary for maintaining unitarity of the theory .
in fact , the subtraction process in the new regularization scheme can be made in a similar way as the one in the dimensional regularization .
it has been seen that evaluating ilis for loop graphs is a crucial step in the new regularization method .
we shall present in this section a general description on the evaluation of ilis for arbitrary loop graphs . though the demonstration may concern some tedious formulae , while it must be very useful for a practical computation of more loop diagrams .
we begin with the scalar type overlapping loop integrals involved in the general n - loop diagrams with @xmath137 an arbitrary number @xmath345^{\alpha_{ij } } } \end{aligned}\ ] ] with @xmath346 and @xmath347 . for the loop diagrams in which the internal lines are topologically not crossed over each other
, the loop integrals are corresponding to the case with @xmath348 @xmath349^{\alpha_{i i+1 } } } \end{aligned}\ ] ] here @xmath346 and @xmath350 as if one of them vanishes , the corresponding sub - integral becomes no longer an overlapping one .
the most divergence of the above overlapping integrals is corresponding to the case with @xmath351 ( @xmath352 ) and @xmath353 ( @xmath354 ) , which is quadratically divergent .
this can easily be shown from the power counting rule as the momentum dimension of the overlapping integral in this case is two ( @xmath355 ) . the above overlapping integrals can be regarded as the most general ones for n - loop graphs . here
we have also supposed that one can always apply the usual feynman parameter method to all the momentum factors @xmath46 , so that all propagators concerning @xmath46 and the external momenta @xmath356 are combined into a single one .
similarly for the propagators concerning @xmath357 ( @xmath358 ) .
therefore , the mass factors @xmath359 and @xmath360 are in general the functions of masses @xmath361 and external momenta @xmath362 ( @xmath167 ) @xmath363 for simplicity , it is good enough to begin with the demonstration on the evaluation of ilis for three loop graphs .
its generalization to more loop graphs is obvious .
the procedure and prescription on the evaluation of ilis for arbitrary loop diagrams will be presented after an explicit demonstration . for three loop diagrams ,
the general overlapping loop integral can be expressed as @xmath364^{\alpha_{12 } } } \frac{1 } { [ ( k_1-k_3 + p_{13 } ) ^2 + { \cal m}_{13}^2 ] ^{\alpha_{13 } } } \frac{1 } { [ ( k_2-k_3 + p_{23 } ) ^2 + { \cal m}_{23}^2 ] ^{\alpha_{23 } } } \nonumber \end{aligned}\ ] ] as the first step , we apply the usual feynman parameter method ( see appendix a ) to the denominators containing the loop momentum @xmath199 . taking the feynman parameter variables @xmath365 ,
@xmath366 and @xmath367 , the sub - integral over @xmath199 becomes quadratic after making a momentum shift @xmath368 the resulting explicit form is given by @xmath369^{\alpha_1 + \alpha_{12 } + \alpha_{13 } } } \nonumber \\ & & \frac{1}{(\ k_2 ^ 2 + { \cal m}_2 ^ 2\ ) ^{\alpha_2 } } \
\frac{1 } { [ ( k_2-k_3 + p_{23 } ) ^2 + { \cal m}_{23}^2 ] ^{\alpha_{23 } } } \ \frac{1}{(\ k_3 ^ 2 + { \cal m}_3 ^ 2\ ) ^{\alpha_3 } } \end{aligned}\ ] ] with @xmath370 apply again the usual feynman parameter method to the remaining denominators containing the loop momentum @xmath203 . taking the feynman parameter variables @xmath371 and
@xmath372 , and making a momentum shift @xmath373 we yield @xmath374 ^{\alpha_1 + \alpha_{12 } + \alpha_{13 } } } \nonumber \\ & & \int_0 ^ 1 dy_1 \frac{(1-y_1)^{\alpha_2 -1 } y_1^{\alpha_{23 } - 1}}{[\ k_2 ^ 2 + y_1(1-y_1 ) ( k_3 - p_{13})^2 + { \cal m}_{y_1}^2 \ ] ^{\alpha_2 + \alpha_{23 } } } \frac{1}{(k_3 ^ 2 + { \cal m}_3 ^ 2 ) ^{\alpha_3 } } \end{aligned}\ ] ] with @xmath375 we now adopt as the second step the uv - divergence preserving parameter method .
taking @xmath376 and @xmath377 , the integral becomes @xmath378 k_2 ^ 2 + \mu_{k_3}^2 + u_1 { \cal m}_{y_1}^2 + { \cal m}_{x_1x_2}^2\ \}^{\alpha_1 + \alpha_{12 } + \alpha_{13 } + \alpha_2 + \alpha_{23 } } } \
\frac{1}{(k_3 ^ 2 + { \cal m}_3 ^ 2 ) ^{\alpha_3 } } \nonumber \end{aligned}\ ] ] with @xmath379 [ \ p_{12 } - y_1 ( k_3 - p_{23 } ) \ ] ^2 \nonumber \\ & + & [ x_2(1-x_2 ) + y_1(1-y_1 ) u_1 ] ( k_3 - p_{13})^2 \end{aligned}\ ] ] where we have made the momentum shift once more @xmath380\ ] ] in general , after performing the uv - divergence preserving parameter method , one can safely carry out the integration over the momentum @xmath199 as @xmath381 for a non - trivial overlapping integral .
after integration over @xmath199 , the integral gets the following form @xmath382^{(\alpha_1 + \alpha_{12 } + \alpha_{13}-2 ) + ( \alpha_2 + \alpha_{23 } ) } } \
\frac{1}{(k_3 ^ 2 + { \cal m}_3 ^ 2 ) ^{\alpha_3 } } \\ & & \frac{(1-x_1)^{\alpha_1 - 1 } ( x_1 - x_2 ) ^{\alpha_{12 } - 1 } x_2^{\alpha_{13 } -1 } ( 1-y_1)^{\alpha_2 -1 } y_1^{\alpha_{23 } - 1 } } { \ { \ k_2 ^ 2 + [ y_1 ( 1-y_1 ) + t_{1 xy } ] ( k_3 - p_{23})^2 + t_{u_1x } [ p_{12 } - y_1 ( k_3 - p_{23 } ) ] ^2 + \mu_{u_1}^2 \ \}^{\alpha_1 + \alpha_{12 } + \alpha_{13 } + \alpha_2 + \alpha_{23 } -2 } } \nonumber \end{aligned}\ ] ] with @xmath383 \\ & & t_{u_1xy } = \frac{x_2(1-x_2 ) - ( x_1-x_2)(1-x_1 + x_2)y_1(1-y_1 ) } { u_1 + ( x_1-x_2)(1-x_1 + x_2 ) } \\ & & \mu_{u_1}^2 = \frac{u_1 } { u_1 + ( x_1-x_2)(1-x_1 + x_2 ) } { \cal m}_{y_1}^2 + \frac{1 } { u_1 + ( x_1-x_2)(1-x_1 + x_2 ) } { \cal m}_{x_1x_2}^2 \end{aligned}\ ] ] where we have factored out the uv - divergence preserving integration over @xmath384 .
it is of interest to notice that the sub - integral over @xmath203 in eq.(7.10 ) becomes solely quadratic .
we can then repeatedly adopt the uv - divergence preserving parameter method .
taking @xmath385 ( k_3 - p_{23})^2 + t_{u_1x } [ p_{12 } - y_1 ( k_3 - p_{23 } ) ] ^2 + \mu_{u_1}^2 $ ] and @xmath386 , we have @xmath387^ { ( \alpha_1 + \alpha_{12 } + \alpha_{13 } - 2 ) + ( \alpha_2 + \alpha_{23 } ) } } \
\int_0^{\infty } du_2 \\ & & \frac{(1-x_1)^{\alpha_1 - 1 } ( x_1 - x_2 ) ^{\alpha_{12 } - 1 } x_2^{\alpha_{13 } -1 } ( 1-y_1)^{\alpha_2 -1 } y_1^{\alpha_{23 } - 1 } u_2^{\alpha_3 -1 } } { \ { \ k_2 ^ 2 + [ u_2 + y_1 ( 1-y_1 ) + \rho_{u_1 } ] \
k_3 ^ 2 + u_2 { \cal m}_3 ^ 2 + \mu_{u_1}^2 + p_{u_1}^2 \ \}^{\alpha_1 + \alpha_{12 } + \alpha_{13 } + \alpha_2 + \alpha_{23 } + \alpha_3 -2 } } \nonumber \end{aligned}\ ] ] with @xmath388\ p_{23}^2 + t_{u_1x}\ [ p_{12}^2 - 2y_1 p_{12}\cdot p_{23 } ] \nonumber \\ & & \rho_{u_1 } = t_{u_1x } y_1 ^ 2 + t_{u_1 xy } \end{aligned}\ ] ] where we have made the momentum shift @xmath389\ p_{23 \mu } + t_{u_1x } y_1 p_{12 \mu } } { u_2 + y_1 ( 1-y_1 ) + \rho_{u_1 } } \end{aligned}\ ] ] in general , as @xmath390 , we can safely perform the integration over @xmath203 and obtain the desired three - fold ilis @xmath391^ { ( \alpha_1 + \alpha_{12 } + \alpha_{13 } - 2 ) + ( \alpha_2 + \alpha_{23 } ) } } \nonumber \\ & & \int_0^{\infty } du_2 \frac{u_2^{\alpha_3 -1 } } { [ u_2 + y_1 ( 1-y_1 ) + \rho_{u_1 } ] ^ { ( \alpha_1 + \alpha_{12 } + \alpha_{13 } -2 ) + ( \alpha_2 + \alpha_{23 } -2 ) + \alpha_3 } } \ i^{(1)}_{\alpha_i \alpha_{ij } } ( { \cal m}_3 ^ 2 + \mu_{u_1 u_2 } ^2 ) \end{aligned}\ ] ] where @xmath392 defines the overall one - fold ilis @xmath393 with @xmath394\ { \cal m}_3 ^ 2 } { u_2 + y_1 ( 1-y_1 ) + \rho_{u_1 } } \nonumber \\ & = & \frac{\mu_{u_1}^2 - [ y_1 ( 1-y_1 ) + \rho_{u_1 } ] \
( { \cal m}_3 ^ 2 - p_{23}^2 ) + t_{u_1x}\ [ p_{12}^2 - 2y_1 p_{12}\cdot p_{23 } ] } { u_2 + y_1 ( 1-y_1 ) + \rho_{u_1 } } \end{aligned}\ ] ] similar to the two - fold ilis of two loop graphs discussed in section v , it is useful to introduce the momentum - like integrating variables and measures @xmath395 correspondingly , one needs to make the replacement for the relevant quantities @xmath396 it is noted that in the infinite limit of @xmath251 ( or @xmath384 ) , the quantities @xmath397 , @xmath398 , @xmath399 and @xmath400 approach to zero .
consequently , we have @xmath401 this ensures the factorization property for overlapping divergences in three loop graphs . here the sub - integrals over the momentum - like variables @xmath264 and @xmath402 characterize the uv divergent properties of one - loop and two - loop sub - diagrams respectively , and the integral over the loop momentum @xmath403 describes the overall divergent property of three loop diagrams . from the above demonstration on the evaluation of ilis for three loop graphs , we can
now straightforwardly generalize it to arbitrary loop graphs .
the general prescription on the evaluation of ilis may be summarized as follows : ( 1 ) . as the first step
, one shall repeatedly apply the usual feynman parameter method to the denominators containing the same loop momentum according to the given order say @xmath199 , @xmath203 , @xmath226 , so that one arrives at the integrals in which n - loop feynman integral contains correspondingly n s denominators . by making appropriate momentum shifts ,
each denominator is governed by the quadratic of momentum , say @xmath240 , @xmath245 , @xmath226 . with this procedure , it is not difficult to see that the momentum integral corresponding to the first denominator say @xmath199 becomes purely quadratic . in general , for n - loop overlapping integrals , one needs to repeatedly use the feynman parameter method by @xmath404 times .
( 2 ) . after the step ( 1 )
, one then adopts the uv - divergence preserving parameter method to the first two denominators , i.e. , the ones involving @xmath240 and @xmath245 .
after that , one can safely carry out the integration over @xmath199 as the resulting sub - integral in such a way becomes convergent .
note that the sub - integral over @xmath199 is quadratic and its integration is easily performed .
one first makes momentum shift again for the momentum involved in the second denominator , i.e. , @xmath203 , so that the integral over @xmath203 becomes purely quadratic . factoring out the coefficient of the momentum @xmath245 , one arrives at the integral which contains uv - divergence preserving momentum - like sub - integral over @xmath264 instead of the one over the loop momentum @xmath199 , and the remaining integral has the same form as the one resulting from the first step ( 1 ) but with the reduced integrals over the loop momenta @xmath203 , @xmath403 , @xmath226 .
( 4 ) . repeating the steps ( 2 ) and ( 3 ) to the remaining loop momentum integrals over @xmath203 , @xmath403 , @xmath226 , but keeping the integral over the last loop momentum ,
we then arrive at the desired ilis . in general , for n - loop overlapping integrals , the resulting ilis are the n - fold ones with ( n-1 ) sub - integrals over the momentum - like variables and one sub - integral over the loop momentum ( say @xmath405 ) . according to this procedure and prescription ,
the resulting ilis for any loop graphs possess all the properties observed in the ilis of two loop graphs ( see section v ) . in here for the ilis of more loop graphs , one only needs to notice the essential property that all the sub - integrals over the momentum - like variables @xmath264 , @xmath402 , @xmath226 characterize the divergences of one - loop , two - loop , @xmath226 sub - diagrams , respectively , and the single sub - integral over the initial loop momentum characterizes the overall divergences of the loop graph .
one can check that this procedure also ensures that all the uv divergences are characterized by the integrals over the momentum - like variables @xmath264 , @xmath402 , @xmath226 , the usual feynman parameter integrations contain no uv divergences and only ir divergences could be hidden in the feynman parameter integrations . for the regularized ilis in the new regularization scheme ,
all the feynman parameter integrations become convergent . it is then obvious that all the theorems deduced in section v and the prescriptions for the regularization and renormalization described in section vi hold for ilis of arbitrary loop graphs .
it is of interest to observe that feynman integrals of arbitrary loop graphs can always be evaluated into the corresponding ilis which contains only a single one - fold ili over the initial loop momentum which characterizes the overall divergences of the loop graph .
similarly , one can show that the subtracting the divergences from all the sub - integrals over the momentum - like variables @xmath264 ,
@xmath402 , @xmath226 , the overall divergence becomes harmless .
it is particularly noted that the new regularization method may cause the conformal scaling symmetry to be broken down due to the existence of two intrinsic mass scales @xmath0 and @xmath1 .
for instance , because of the quadratic ` cut - off ' terms of the ces @xmath0 , the conformal scaling symmetry can be broken down in a class of theories with scalar interactions .
in general , as long as taking the ses @xmath1 to be finite ( @xmath406 ) , the conformal scaling symmetry will be broken down even in gauge theories .
of particular , once a mass gap @xmath142 is generated via dynamical reasons of strong interactions , it becomes natural to set the ses @xmath1 to be at the mass gap @xmath142 .
an interesting example is for qcd in which an energy scale around @xmath407 gev has often be introduced to characterize the low energy dynamics of qcd .
this is because when the ses @xmath1 runs down to such a low energy scale , @xmath408 gev , the interactions become much stronger .
this feature may easily be understood from the behavior of asymptotic freedom of qcd@xcite . by evaluating the relevant one loop feynman diagrams , it is not difficult to check that the renormalized gauge coupling constant @xmath409 from the new regularization method is consistent with the one from the dimensional regularization at @xmath410 and @xmath411 @xmath412 \end{aligned}\ ] ] where only the polynomial of the finite terms @xmath413 differ from the one in the dimensional regularization . in obtaining the above result , the masses of quarks have been ignored . on the other hand , for a finite ses
@xmath1 , the conformal scaling symmetry is explicitly broken down due to higher dimensional effective interaction terms , i.e. , their momentum power counting dimensions are larger than four
. in general , the uv convergent but ir divergent feynman diagrams will result in such higher dimensional interaction terms . from the dimensional analysis
, these higher dimensional interactions can be written down in terms of the expansion of the inverse power of the ses @xmath1 @xmath414 which implies that the existence of a non - trivial solution for the ses @xmath1 is transmuted to the one of nonzero vacuum expectation values of @xmath415 and @xmath416 .
indeed , it has been turned out that at a low energy scale @xmath417 gev ( @xmath418 is the charm quark mass ) , the light quarks and gluons were found to have nonzero condensates @xcite due to strong interactions @xmath419 which implies that both conformal scaling symmetry and chiral symmetry are dynamically broken down at such a low energy scale .
this strongly indicates that a mass gap with @xmath420 must be generated and the quarks are going to be confined around such a low energy scale .
obviously , the conformal symmetry breaking scale or the mass gap @xmath421 must be related to the dynamically chiral symmetry breaking scale @xmath15 which has been known to be around @xmath422 gev .
the order of magnitude for the mass gap @xmath423 may be estimated from the quark and gluon condensates . from simple dimensional and large @xmath424 considerations , we have @xmath425 which holds in the large @xmath424 limit with @xmath426 being held fixed .
as expected , the mass gap @xmath423 and the chiral symmetry breaking scale @xmath15 are at the same order of magnitude , namely : @xmath427 correspondingly , the critical temperature for the quark deconfinement is estimated to be at the order of magnitude @xmath428 note that to obtain more definitive results , a detailed quantitative calculation is needed . only the perturbative evaluation of qcd is not enough to determine the numerical value for the mass gap .
it is necessary to develop low energy dynamics of qcd for a nonperturbation evaluation of qcd .
one may construct some effective perturbation theories to describe the low energy qcd .
the crucial issue concerns a consistent matching between the perturbative and nonperturbative qcd . in general , the ir cut - off in the perturbative qcd is related to the uv cut - off in the nonperturbative qcd . as a good example , it has been shown in ref.@xcite that the uv cut - off in the chiral perturbation theory of mesons does match to the ir cut - off in perturbative qcd via some appropriate matching conditions .
therefore , it is of interest to further investigate the ir and uv correspondence / matching of the short and long distance dynamics of qcd . as the new regularization is applicable to both underlying perturbative qcd and effective qfts characterizing nonperturbative qcd at low energy , it may provide us a practical tool to understand well the genesis of mass gap and the quark confinement via the ir and uv correspondence / matching of the short and long distance dynamics of qcd .
we have explicitly presented a proof for the existence of a symmetry principle preserving and infinity free regularization and renormalization method .
the main point for the new method is to regularize the whole feynman integrals of loop graphs rather than just the propagators of fields .
the other important point is to analyze an infinitely large number of regulators ( or regularization quanta ) rather than only a limited few regulators .
specifically , the concepts of irreducible loop integrals ( ilis ) and uv - divergence preserving parameter method as well as intrinsic mass scales ( i.e. , the characteristic energy scale ( ces ) @xmath0 and the sliding energy scale ( ses ) @xmath1 ) have been found to play an essential role .
consequently , the regularized qfts become well defined .
the method has been developed to be practically applicable to both underlying renormalizable qfts and effective qfts .
indeed , it provides an explicit demonstration that a quantum theory of fields can be applied to describe the laws of nature only when the considered energy scale is sufficiently lower than the ces @xmath0 .
thus the underlying renormalizable qfts , like yang - mills gauge theories , become particularly interesting as the ces @xmath0 in such theories is in principle allowed to be infinitely large @xmath411 through a suitable renormalization for the gauge coupling and fields , which is equivalent to the dimensional regularization with @xmath162 . in general , the scales @xmath0 and @xmath1 at a particularly interesting energy scale @xmath142 actually set the uv and ir cut - off energy scales respectively in the regularized qfts , but they distinguish from the naive cutoff momentum scales imposed kinematically to the upper and lower bounds of the internal loop momenta , the ces @xmath0 and the ses @xmath1 are more likely to behave as the dynamical mass scales , which remains to be understood from a deeper reason .
it is intriguing to see that the new regularization method provides a consistent way alternative to string theories for making qfts to be well defined finite theories .
though the explicit demonstration for it has been carried out only at one and two loop level , its proof to more loops has been shown to be a straightforward generalization . in this sense , the divergence problem of qfts may no longer be a strong reason for extending the underlying theory to go beyond the quantum theory of fields /elementary particles .
first of all , it must be very helpful to further study the possible interesting phenomena caused by the conformal scaling symmetry breaking .
in particular , the ir and uv correspondence between the short distance physics and long distance physics may play an important role to explore the nonperturbative effects and to understand well the matching between the short and long distance physics .
we believe that the conformal scaling symmetry and its breaking mechanism may eventually guide us to reveal some longstanding puzzles , such as : genesis of mass gap , unseen quarks , missing symmetries , origin of quark and lepton masses and mixing angles , and also small but nonzero cosmological constant . finally , we hope that the new method described in this paper can widen the applications of qfts
. + + * acknowledgement * the author is grateful to prof .
david j. gross for valuable discussions during the string conference at itp , beijing , especially for his important comments involving the intrinsic mass scales and conformal scaling symmetry breaking in the new regularization method .
he is also benefited from discussions with prof .
faddeev concerning the infrared cutoff energy scale and the mass gap in yang - mills gauge theories .
the author wants to express his thanks to profs .
chou , y.b .
dai , h.y .
guo , z.x .
liu , h.w .
peng , m. yu , z.x .
zhang , z.y .
zhu and many other colleagues for stimulating discussions and conversations , and c.j .
zhu for carefully reading the manuscript with useful comments .
he is also grateful to dr .
yan for the numerical check on the conjectures for three type of functional limits appearing in the new regularization method .
especially , he would like to thank profs .
li , w. palmer and l. wolfenstein for their kind hospitality to visit kentucky university , ohio - state university and carnegie - mellon university during his sabbatical on leave , where he was able to continue thinking the subject over .
this work was supported in part by the key projects of chinese academy of sciences and national science foundation of china .
in this appendix , we present a detailed evaluation of the gauge field vacuum polarization function from various non - vanishing diagrams . in the evaluation , we shall often use the feynman parameter formula @xmath429^{\alpha + \beta } } \ ] ] and its generalized form @xmath430^{\alpha_1 + \cdots + \alpha_n } } \end{aligned}\ ] ] to simplify the integrals .
there are in general four feynman diagrams ( see figures ( 1)-(4 ) ) .
consider first the diagram ( 1 ) which comes from the yang - mills trilinear interaction term . with the gauge boson propagator and
the yang - mills trilinear vertex of the feynman rules , one has @xmath431 with @xmath432 \nonumber \\ & + & \frac{\lambda}{(p+k)^2 } [ \ ( k^2 -2p^2)k_{\mu}k_{\nu } + ( p\cdot k ) ( p_{\mu}k_{\nu } + p_{\nu}k_{\mu } ) + ( p^2 - 2k^2 ) p_{\mu}p_{\nu } - ( k^2 - p^2)^2 g_{\mu\nu } ] \nonumber \\ & + & \frac{\lambda^2}{(p+k)^2 k^2 } [ \ p^4 k_{\mu}k_{\nu } + ( p\cdot k)^2 p_{\mu}p_{\nu } - ( p \cdot k ) p^2 ( p_{\mu}k_{\nu } + p_{\nu}k_{\mu } ) \end{aligned}\ ] ] after using the feynman parameter method with @xmath433 and @xmath434 , and replacing @xmath435 by @xmath436 , as well as performing some algebra , the gauge field vacuum polarization function @xmath437 can be expressed into the following general form in terms of the ilis @xmath438 \nonumber \\ & + & \lambda \gamma(3 ) \int_{0}^{1 } dx\ x\ [ \ - i_2 g_{\mu\nu } + i_{2 \mu\nu } + \left(\ 2(1-x)(1 + 2x ) p^2 g_{\mu\nu } + ( x^2 -2 ) p_{\mu}p_{\nu } \right)\ i_{0 } \nonumber \\ & - & \left(\ ( x + 2(1-x^2)\ ) p^2 g_{\mu}^{\rho}g_{\nu}^{\sigma } + 4x^2 p^{\rho}p^{\sigma } g_{\mu\nu } - ( 2x^2 + 1 ) p^{\rho}\ ( g_{\mu}^{\sigma } p_{\nu } + g_{\nu}^{\sigma } p_{\mu } ) \right ) i_{0 \rho\sigma } \nonumber \\ & - & \left(\ ( 1-x)^2(1 + 2x)^2 p^4 g_{\mu\nu } - ( 1 + x(1 - 2x)(2-x^2)\ ) p^2 p_{\mu}p_{\nu } \right)\ i_{-2}\ ] \nonumber \\ & + & \lambda^2\ \gamma(4)/2 \int_{0}^{1 } dx\ x(1-x)\ [ \ p^4 g_{\mu}^{\rho}g_{\nu}^{\sigma } + p_{\mu}p_{\nu}p^{\rho}p^{\sigma } - p^2 p^{\rho}\ ( g_{\mu}^{\sigma } p_{\nu } + g_{\nu}^{\sigma } p_{\mu } ) \
i_{-2\rho\sigma}\ ] \ \ } \end{aligned}\ ] ] where @xmath439 and @xmath440 are defined as @xmath441 and the mass term in all the ilis is given by @xmath442 the diagram ( 2 ) is the so - called tadpole diagram with no momentum flowing into the loop and it arises from the yang - mills quadrilinear interaction terms .
this diagram is quadratically divergent from power counting and vanishes in dimensional regularization , which has been thought to be the crucial point for preserving the gauge invariance in dimension regularization . as a general evaluation before applying for any regularization , it is better to treat it in the equal foot without worrying about quadratic divergences @xmath443 } { ( ( p+k)^2 + i\varepsilon)(k^2 + i\varepsilon ) } \end{aligned}\ ] ] in order to be able to express this loop integral by using the same ilis as those appearing in diagram ( 1 ) , we have inserted the momentum factor @xmath444 . adopting the same feynman parametrization and performing some algebra ,
we yield @xmath445 g_{\mu\nu } \nonumber \\ & + & \lambda \gamma(3 ) \int_{0}^{1 }
dx\ [ \ -\frac{1}{2 } i_2\ g_{\mu\nu } + ( 1-x ) i_{2\mu\nu } \ ] \nonumber \\ & + & \lambda \gamma(3 ) \int_{0}^{1 } dx\ [ \
\left(\ -\frac{1}{2}(1-x)(1 - 2x ) p^2 g_{\mu\nu } + ( 1-x)x^2 p_{\mu}p_{\nu } \ \right)\ i_0 \nonumber \\ & + & \left ( \ ( 1-x)^2(1 - 2x)p^2 g_{\mu}^{\rho}g_{\nu}^{\sigma } - 2(1-x)^2x p^{\rho}\ ( g_{\mu}^{\sigma } p_{\nu } + g_{\nu}^{\sigma } p_{\mu } ) \
\right)\ i_{0\rho\sigma}\ ] \nonumber \\ & + & \lambda \gamma(3 ) \int_{0}^{1 } dx\ ( 1-x)^2(1 - 2x)x^2 p_{\mu}p_{\nu } p^2 \ i_{-2 } \ \ } \end{aligned}\ ] ] which shows that without imposing any regularization , the tadpole graphs of yang - mills fields are in general no vanishing and has actually all divergent structures given by the ilis .
we now consider the diagram ( 3 ) which arises from the ghost - gauge field interaction term .
using the faddeev - popov ghost propagator and ghost - gauge field vertex with including a minus sign for the closed loop of grassmann fields , one has @xmath446
\end{aligned}\ ] ] in obtaining the second equality , we have used the same feynman parametrization as the previous ones . consider finally the diagram ( 4 ) due to fermion - gauge field interactions .
using the fermion propagator and fermion - gauge field vertex with including a minus sign for the closed fermion loop , one reads @xmath447
\end{aligned}\ ] ] where we have used the definition for the group theory factor @xmath448 note also that the mass factor in the ilis is modified by the fermion mass @xmath449 as indicated in the notation @xmath450
@xmath451 to simplify the integrals , the following identities are found to be helpful @xmath452 = \int_{0}^{1 } dx\ x(1-x)\ i[x(1-x ) ] + \frac{1}{2}\int_{0}^{1 } dx\ i[x(1-x ) ] \nonumber \\ & & \int_{0}^{1 } dx\ x^3\ i[x(1-x ) ] = \frac{3}{2}\int_{0}^{1 } dx\ x(1-x)\ i[x(1-x ) ] + \frac{1}{2}\int_{0}^{1 } dx\ i[x(1-x ) ] \nonumber \\ & & \int_{0}^{1 } dx\ x^4\ i[x(1-x ) ] = \int_{0}^{1 } dx\ x^2(1-x)^2\ i[x(1-x ) ] \nonumber \\ & & \qquad + 2\int_{0}^{1 } dx\ x(1-x)\ i[x(1-x ) ] + \frac{1}{2 } \int_{0}^{1 } dx\ i[x(1-x ) ] \\ & & \int_{0}^{1 } dx\ x^5\ i[x(1-x ) ] = \frac{5}{2}\int_{0}^{1 } dx\ x^2(1-x)^2\ i[x(1-x ) ] \nonumber \\ & & \qquad + \frac{5}{2}\int_{0}^{1 } dx\ x(1-x)\ i[x(1-x ) ] + \frac{1}{2 } \int_{0}^{1 } dx\ i[x(1-x ) ] \nonumber \end{aligned}\ ] ] +
we present here some regularized one - fold ilis in the cut - off and dimensional regualrizations .
the relevant one - fold ilis in the text are @xmath453 consider first the cut - off regularization , the ilis can be easily evaluated in four dimensional euclidean space of momentum by performing a wick rotation @xmath454 \\ & & i_0^r = i\int^{\lambda^2 } \frac{d^4 k}{(2\pi)^4}\ \frac{1}{(-k^2 - { \cal m}^2)^2 } = \frac{i}{16\pi^2 } \ [ \
\ln \frac{\lambda^2 + { \cal m}^2}{{\cal m}^2 } + \frac{{\cal m}^2}{\lambda^2 + { \cal m}^2 } -1 \ ] \end{aligned}\ ] ] and @xmath455 \\ i_{0\ \mu\nu}^r & = & i\int^{\lambda^2 } \frac{d^4 k}{(2\pi)^4}\ \frac{-k_{\mu}k_{\nu}}{(-k^2 - { \cal m}^2)^3 } = \frac{1}{4}g_{\mu\nu}\ \frac{i}{16\pi^2 } \ [ \ \ln \frac{\lambda^2 + { \cal m}^2}{{\cal m}^2 } + \frac{2{\cal m}^2}{\lambda^2
+ { \cal m}^2 } - \frac{{\cal m}^4}{2(\lambda^2 + { \cal m}^2)^2 } - \frac{3}{2 } \ ] \end{aligned}\ ] ] where @xmath100 is the cut - off momentum .
note that the above integrals are carried out in the euclidean space and the final results are given by rotating back to the minkowski space , all integrations over momentum will be performed in this way except with a specific mention .
consider next the dimensional regularization in which the space - time dimension @xmath163 is taken @xmath456 @xmath457 and @xmath458
we first present in the following table some numerical results for the functions @xmath180 , @xmath182 and @xmath183 when the regulator number @xmath98 is taken to be typically large .
+ + table 1 .
the quantities @xmath180 , @xmath182 and @xmath183 ( @xmath459 ) as functions of the regulator number @xmath98 . + [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] + + as the functions behave well .
they are expected to approach some finite values at @xmath181 , one may use a functional fitting method to extract their values at @xmath181 with the required precision . here
we choose the following fitting function @xmath460 here @xmath461 are the fitting coefficients .
in general , one can choose , instead of @xmath462 , other functions with vanishing limit at @xmath181 .
taking @xmath463 , @xmath464 and @xmath465 ( @xmath153 ) , one then has @xmath466 , @xmath467 and @xmath468 ( @xmath153 ) .
practically , by computing a set of numerical values of the functions @xmath180 , @xmath182 and @xmath183 ( @xmath153 ) for some sufficiently large values of @xmath98 , and solving the corresponding linear equations , one is then able to obtain @xmath469 with the required precision .
here we only present the solutions at the precision @xmath185 @xmath470 they are resulted from a set of values for the functions @xmath180 , @xmath182 and @xmath183 ( @xmath471 ) obtained for @xmath98 from @xmath472 to @xmath473 . for a consistent check , such solutions are found to provide , with the same precision , a prediction for the numerical values of the functions @xmath180 , @xmath182 and @xmath183 ( @xmath153 ) at @xmath474 ( see table ) .
the above numerical analysis provides a reasonable check on the three conjectures made in section iv .
using the definitions and prescriptions of the new regularization , we present the following useful regularized ilis @xmath475_l \frac{1}{k^2 + m_l^2 + { \cal m}^2 } = \pi^2 \ { m_c^2 - \mu^2 [ \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + 1 + y_2(\frac{\mu^2}{m_c^2 } ) \ ] \ } \\ & & \int [ d^4 k]_l \frac{1 } { ( k^2 + m_l^2 + { \cal m}^2 ) ^2 } = \pi^2 \ { \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + y_0(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{1 } { ( k^2 + m_l^2 + { \cal m}^2)^3 } = \frac{1}{2\mu^2 } \pi^2 \ { \ 1- y_{-2}(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{1 } { ( k^2 + m_l^2 + { \cal m}^2 ) ^{\alpha } } = \pi^2 \frac{\gamma(\alpha -2)}{\gamma(\alpha ) } \frac{1 } { ( \mu^2 ) ^{\alpha -2 } } \ { \ 1 - y_{-2(\alpha -2 ) } ( \frac{\mu^2}{m_c^2 } ) \ \ } \qquad \alpha > 3 \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^2 } = \frac{1}{2 } \delta_{\mu\nu } \pi^2 \ { m_c^2 - \mu^2 [ \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + 1 + y_2(\frac{\mu^2}{m_c^2 } ) \ ] \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^3 } = \frac{1}{4 } \delta_{\mu\nu } \pi^2 \ { \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + y_0(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^4 } = \frac{1}{4}\delta_{\mu\nu } \frac{1}{3\mu^2 } \pi^2 \ { \ 1- y_{-2}(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^{\alpha + 1 } } = \frac{1}{2 } \delta_{\mu\nu } \pi^2 \frac{\gamma(\alpha -2)}{\gamma(\alpha + 1 ) } \frac{1 } { ( \mu^2 ) ^{\alpha -2 } } \ { \ 1 - y_{-2(\alpha -2 ) } ( \frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu}k_{\rho}k_{\sigma } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^3 } = \frac{1}{8 } \delta_{\ { \mu\nu\rho\sigma\ } } \pi^2 \ { m_c^2 - \mu^2 [ \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + 1 + y_2(\frac{\mu^2}{m_c^2 } ) \ ] \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu}k_{\rho}k_{\sigma } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^4 } = \frac{1}{24 } \delta_{\ { \mu\nu\rho\sigma\ } } \pi^2 \ { \ \ln \frac{m_c^2}{\mu^2 } - \gamma_w + y_0(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } k_{\rho}k_{\sigma } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^5 } = \frac{1}{24 } \delta_{\ { \mu\nu\rho\sigma\ } } \frac{1}{4\mu^2 } \pi^2 \ { \ 1- y_{-2}(\frac{\mu^2}{m_c^2 } ) \ \ } \\ & & \int [ d^4 k]_l \frac{k{\mu}k_{\nu } k_{\rho}k_{\sigma } } { ( k^2 + m_l^2 + { \cal m}^2 ) ^{\alpha + 2 } } = \frac{1}{4 } \delta_{\ { \mu\nu\rho\sigma\ } } \pi^2 \frac{\gamma(\alpha -2)}{\gamma(\alpha + 2 ) } \frac{1 } { ( \mu^2 ) ^{\alpha -2 } } \ { \ 1 - y_{-2(\alpha -2 ) } ( \frac{\mu^2}{m_c^2 } ) \ \ } \\ \end{aligned}\ ] ] with @xmath476_l \equiv \lim_{n\to \infty } \sum_{l=0}^{n } c_l^n \int d^4 k = \lim_{n\to \infty } \sum_{l=0}^{n } ( -1)^l \frac{n!}{(n - l)!\ l ! } \int d^4 k \end{aligned}\ ] ] where @xmath477 the explicit forms for @xmath478 , @xmath308 and @xmath479 are given in eqs.(4.10 - 4.13 ) .
99 see e.g. : t.d .
lee , _ particle physics and introduction to field theory _ , ( harwood academic publishers gmbh , 1981 ) ; _ symmetries , asymmetries , and the world of particles _ , ( university of washington press , 1988 )
yang and r.l .
mills , phys . rev . * 96 * , 191 ( 1954 ) . see e.g. : s. weinberg , _ the quantum theory of fields_-i and ii , ( cambridge university press , cambridge , 1995 ) . see also : ta - pei cheng and ling - fong li , _ gauge theory of elementary particle physics _ , ( oxford university press , walton street , oxford ox2 6dp , 1984 ; reprinted 1985 - 1996 ) ;
+ yuan - ben dai , _ gauge theory of interactions _ , scientific publishing , beijing , 1987 ; + m. e. peskin and d.v .
schroeder , _ an introduction to quantum field theory _ , addison - wesley publishing co. , 1995 .
see e.g. : g. t hooft , _ under the spell of the gauge principle _ , ( world scientific publishing co. , 1994 ) .
see e.g. : jean zinn - justin , _ quantum field theory and critical phenomena _ , ( oxford university press , walton street , oxford ox2 6dp , 1996 ) .
w. heisenberg , ann .
* 32 * , 20 ( 1938 ) .
w. pauli and f. villars , rev .
* 21 * , 434 ( 1949 ) .
j. schwinger , phys . rev . * 82 * , 664 ( 1951 ) .
n.n . bogoliubov and o.s .
parasiuk , acta math .
, * 97 * , 227 ( 1957 ) ; n.n . bogoliubov and d.v .
shirkov , _ introduction to the theory of quantized fields _ , interscience , new york , 1959 ; + k. hepp , comm .
* 2 * , 301 ( 1966 ) ; + w. zimmermann , cmmmun .
phys . * 15 * , 208 ( 1969 ) ; `` lectures on elementary particles and quantum field theory '' , brandeis summer institute 1970 , edited by s. deser , m. grisaru and h.pendleton , mit press . g.t
hooft and m. veltman , nucl . phys . *
b44 * , 189 ( 1972 ) ; + j. ashmore , lett .
nuovo cimento * 4 * , 289 ( 1972 ) ; + c.g .
bollini and j.j giambiaggi , phys .
40b * ( 1972 ) 566 ; nuovo cimento * 12b * ( 1972 ) .
wilson , _ new phenomena in subnuclear physics _ , erice 1975 , edited by a. zichichi ( plenum , new york 1977 ) ; + t. reisz , commun .
phys . * 117 * , 79 , 639 ( 1988 ) ; + h.b .
nielsen and m. ninomiya , nucl . phys . *
b185 * , 20 ( 1981 ) ; + t. banks , l. susskind and j. kogut , phys . rev . * d13 * , 1043 ( 1976 ) .
freedman , k. johnson and j.i .
latorre , nucl .
b371 * , 353 ( 1992 ) ; d.z .
freedman , g. grignani , k. johnson and n. rius , ann .
phys . * 218 * , 75 ( 1992 ) ; + p.e .
haagense and j.i .
latorre , phys .
b283 * , 293 ( 1992 ) ; p.e .
haagense and j.i .
latorre , ann .
b221 * , 77 ( 1993 ) .
y. nambu and g. jona - lasinio , phys . rev . *
122 * , 345 ( 1961 ) .
t. gherghetta , phys . rev . *
d50 * , 5985 ( 1994 ) .
a.a . slavnov , theor .
33 * , 977 ( 1977 ) .
leon , c.p .
martin and f.r .
ruiz , phys .
b355 * , 531 ( 1995 ) .
martin and f.r .
ruiz , nucl .
b436 * , 545 ( 1995 ) .
s. weinberg , phys .
118 * , 838 ( 1959 ) .
w. siegel , phys .
b84 * , 193 ( 1979 ) .
w. siegel , phys .
b94 * , 37 ( 1980 ) .
wu , phys .
rev . * d64 * 016001 ( 2001 )
. f. del aguila , a. culatti , r. munoz tapia and m. perez - victoria , phys . lett . *
b419 * , 263 ( 1998 ) ; m. perez - victoria , phys . lett . *
b442 * , 315 ( 1998 ) .
see e.g. : m.b .
green , j. h. schwarz and e. witten , _ superstring theory _ , ( cambridge university press , cambridge , 1987 ) ; + d.j .
gross , j.a .
harvey , e. martinec and r. rohm , nucl .
b256 * , 235 ( 1985 ) ; * b267 * , 75 ( 1986 ) ; + j. polchinski , _ string theory _
( cambridge university press , cambridge , 1998 ) . s. weinberg , hep - th/9702027 , 1997 .
wilson , phys .
rev . * b4 * , 3174 , 3184 ( 1971 ) ; rev
phys . * 47 * , 773 ( 1975 ) .
m. gell - mann and f.e .
low , phys .
rev . * 95 * , 1300 ( 1954 ) .
faddeev and v.n .
popov , phys .
* 25b * , 29 ( 1967 ) .
feynman , acta phys .
polonica * 24 * , 697 ( 1963 ) .
de witt , phys . rev . * 162 * , 1195 , 1239 ( 1967 ) .
batistel , a.l . mota and m.c .
nemes , mod .
a13 , 1597 ( 1998 ) ; hep - th/0203261 and references therein .
d.j . gross and f. wilczek , phys .
rev . lett . * 30 * , 1343 ( 1973 ) . h. d. politzer , phys .
lett . * 30 * , 1346 ( 1973 ) .
shifman , a.i .
vainstein and v.i .
zakharov , nucl .
* b147 * , 385 ( 1979 ) .
vacuum polarization diagrams of gauge field : ( 1 ) from trilinear gauge boson interactions ; + ( 2 ) from quartic gauge boson interactions ; ( 3 ) from gauge - ghost boson interactions ; + ( 4 ) from fermion - gauge boson interactions . | through defining irreducible loop integrals ( ilis ) , a set of consistency conditions for the regularized ( quadratically and logarithmically ) divergent ilis are obtained to maintain the generalized ward identities of gauge invariance in non - abelian gauge theories .
the ilis of arbitrary loop graphs can be evaluated from the corresponding feynman loop integrals by adopting an ultraviolet ( uv ) divergence preserving parameter method .
overlapping uv divergences are explicitly shown to be factorizable in the ilis and be harmless via suitable subtractions . a new regularization and renormalization method
is presented in the initial space - time dimension of the theory .
the procedure respects unitarity and causality . of interest
, the method leads to an infinity free renormalization and meanwhile maintains the symmetry principles of the original theory except the intrinsic mass scale caused conformal scaling symmetry breaking and the anomaly induced symmetry breaking .
tadpole graphs of yang - mills gauge fields are found to play an essential role for maintaining manifest gauge invariance via cancellations of quadratically divergent ilis . quantum field theories ( qfts )
regularized through the new method are well defined and governed by a physically meaningful characteristic energy scale ( ces ) @xmath0 and a physically interesting sliding energy scale ( ses ) @xmath1 which can run from @xmath2 to a dynamically generated mass gap @xmath3 or to @xmath4 in the absence of mass gap and infrared ( ir ) problem . for @xmath5 ,
the initial uv divergent properties of qfts are recovered and well - defined .
in fact , the ces @xmath0 and ses at @xmath3 play the role of uv and ir cutoff energy scales respectively .
it is strongly indicated that the conformal scaling symmetry and its breaking mechanism play an important role for understanding the mass gap and quark confinement .
the new method is developed to be applicable for both underlying renormalizable qfts and effective qfts .
it also leads to a set of conjectures on mathematically interesting numbers and functional limits which may provide deep insights in mathematics . |
our mean field study follows the approach of ref .
@xcite ; it is based on the following hamiltonian for rare earth ( r ) moments @xmath23 at site @xmath24 of the pyrochlore lattice : @xmath81 in this expression , @xmath27 is written in terms of @xmath28 stevens operators @xcite with @xmath75 k , @xmath76 k , @xmath77 k , @xmath78 k,@xmath79 k , @xmath80 k , in weybourne conventions .
these parameters lead to cef excitations at 6.2 , 7.5 , 13.8 , 44.9 , 47.7 , 52.6 and 69.6 mev which compare well with the first levels determined experimentally through neutron scattering at 6.4 , 7.3 and 15.4 mev @xcite .
@xmath82 is the sum of the anisotropic exchange tensor @xmath83 and of the @xmath84 dipolar interaction , limited to the contribution of the nearest - neighbors .
various conventions have been used to define @xmath29 @xcite . here , @xmath29 is defined in the @xmath30 frame linked with a r - r bond @xcite : @xmath85 considering for instance the pair of er@xmath3 ions at @xmath86 and @xmath87 , where @xmath88 is the cubic lattice constant , we define the local bond frame as : @xmath89 , @xmath90 and @xmath91 .
this hamiltonian , written in terms of bond - exchange constants , has the advantage to provide a direct physical interpretation of the different parameters .
note that @xmath32 is an anti - symmetric exchange constant ( dzyaloshinskii - moriya like ) , while @xmath31 are symmetric terms .
owing to the form of the dipolar interaction , we have : @xmath92 with @xmath93 and where @xmath94 is the nearest neighbour distance in the pyrochlore lattice .
if we combine them with the anisotropic exchange constants @xmath95 , we obtain : @xmath96 as is usual in mean - field approximations , a self - consistent treatment is carried out to solve the problem : starting from a random configuration for the @xmath97 , the contribution to @xmath98 at site @xmath24 is diagonalized in the hilbert space of the er@xmath3 magnetic moment defined by the @xmath99 basis vectors .
this yields the energies @xmath100 and the wave functions @xmath101 .
the updated magnetic moments , @xmath102 , at each step of the iteration procedure , are given by : @xmath103 with @xmath104 these are then used to proceed to site @xmath105 , and this is repeated until convergence .
this anisotropic exchange hamiltonian can be rewritten in terms of couplings between the spin components of a pseudospin @xmath64 defined in subspace spanned by the ground cef doublet states @xcite : @xmath106 \end{aligned}\ ] ] @xmath107 is the set of effective exchange parameters .
note that the sanserif " notation refers to local bases .
the states of this pseudospin @xmath64 span the ground cef wavefunctions doublet , using the relation : @xmath108 for the pyrochlores , of fd@xmath109 m space group , @xmath110 matrix is diagonal and takes the form : @xmath111 we finally obtain the following relations : @xmath112 and , conversely : @xmath113
the description of the possible magnetic structures in er@xmath0ti@xmath0o@xmath1 is based on the symmetry analysis performed in the fd@xmath109 m space group for a @xmath5 propagation vector .
different structures are possible , belonging to different irreducible representations . as explained in the main text , er@xmath0ti@xmath0o@xmath1 undergoes a transition towards an antiferromagnetic nel phase below @xmath114 k @xcite .
this ordered phase corresponds to the @xmath7 basis vector of the @xmath46 irreducible representation .
the magnetic moment at site i is defined in a local frame @xmath115 given in table [ table1 ] . owing to symmetry ,
the moments in the @xmath7 configuration are of the form @xmath43 .
the experiment favors @xmath116 , hence a non - collinear structure , in which the magnetic moments are perpendicular to the local @xmath4 axes . in the general case ,
the projection of the moment along the cef axis is : @xmath117 while the angle formed with the xy plane is : @xmath118 assuming classical magnetic moments of equal amplitude @xmath119 on the different sites , and calculating the coupling matrices @xmath120 , it is possible to calculate analytically the classical energy per spin in the @xmath7 state : @xmath121 in the xy case , @xmath116 : @xmath122 .@xmath115 frame for the different sites of a tetrahedron . [ cols="^,^,^,^,^ " , ]
the spin dynamics is calculated using the random phase approximation ( rpa ) @xcite .
spin excitations are constructed on the basis of the transitions between the mean - field states @xmath123 of energy @xmath100 .
these transitions occur between the set of cef modes , perturbed or split by the molecular field . because of the interactions between the magnetic moments , these transitions acquire a dispersion .
two quantities are essential , namely the matrix element of @xmath124 that connects the two states @xmath125 and @xmath126 : @xmath127 and the energy of the transition @xmath128 . + following the generalized susceptibility approach @xcite , it is useful to introduce labels associated with a transition @xmath129 as well as matrices @xmath130 and energies @xmath131 so that : @xmath132 and : @xmath133 with @xmath134 . the rpa generalized spin - spin susceptibility tensor @xmath135 is a solution of @xcite : @xmath136 { \tilde \chi}_{s'',s ' } \label{rpa}\ ] ] here , @xmath137 is the 3 @xmath138 3 single - ion spin - spin susceptibility : @xmath139 the spin - spin correlation function @xmath140 is then given by : @xmath141 \end{aligned}\ ] ] where @xmath142 denotes the position of the ion at site @xmath24 and @xmath143 is the magnetic form factor of the er@xmath3magnetic ions .
+ we proceed by rewriting the rpa eq .
( [ rpa ] ) using eq .
( [ eq16 ] ) : @xmath144 owing to the definition of @xmath130 , it is possible to show that this matrix has just one non zero eigenvalue @xmath145 : @xmath146 while the corresponding ( normalized ) eigenvector is nothing but @xmath147 .
multiplying eq .
( [ eq17 ] ) on the left and on the right by @xmath148 and @xmath149 respectively , we get : @xmath150 we are then left with an inversion problem to determine @xmath151 and then @xmath152 . from this equation , it is also worth noting that the pole of the susceptibility are given by the eigenvalues of the matrix @xmath153 : @xmath154
the blue ( resp .
green ) points , obtained from the @xmath155 t ( resp 2.5 t ) inelastic neutron scattering data , lie around the blue ( resp .
green ) plane , which is better evidenced in the inset .
the red line is the intersection of the two planes , and the red point the best fit ( see text).,width=302 ] data is taken from refs .
@xmath11@xmath12 $ ] .
, width=226 ] a fitting procedure in a four parameters space may be hazardous . reducing the parameter space
often requires a combination of different experimental techniques giving complementary information as well as a detailed characterization of the system in function of external parameters , such as the temperature or the magnetic field . ref .
@xcite claims that er@xmath0ti@xmath0o@xmath1 has a small dzyaloshinskii - moriya interaction @xmath156 .
although it may slightly improve the agreement between simulations and experimental data , we checked that such a contribution has a weak effect on the excitation spectrum . in
what follows , we therefore consider @xmath157 .
exhaustive calculations of the scattering function @xmath158 have been performed as a function of the remaining parameters @xmath159 , assuming equipopulated @xmath7 magnetic domains at zero magnetic field and only one domain at @xmath160 t. experimental data were fitted by matching the location of the maximum neutron intensity in several directions .
the spectral weight repartition has also been taken into account , checking that numerical intensities roughly matche the experimental ones .
all the considered parameters giving a good agreement are reported in blue ( resp .
green ) for @xmath155 t ( resp .
2.5 t ) in figure [ fig - fit ] .
interestingly , while many sets of parameters capture the zero field data , the ones obtained at @xmath160 t appear to be much restrictive : only a very small portion of parameter space accounts for the corresponding experimental dispersions .
this also points out very simple linear correlations between parameters ( which is better evidenced in the inset ) : the relevant coupling constants are contained into a plane , thus reducing the parameter space to a surface .
different relations between the coupling constants ( or surfaces ) are obtained considering the zero ( @xmath161 ) or the finite magnetic field data ( @xmath162 ) .
combining these two latter relations reduces the relevant parameter space to a line represented in red in the fig .
[ fig - fit ] , whose parametric equations is : @xmath163 along this line , the best set of parameters , fitting all the experimental data , is finally obtained for @xmath164 k ( red point in fig . [ fig - fit ] ) .
we finally checked that this set of parameters allows one to capture the magnetization curve measured at @xmath165 mk ( see fig .
[ fig - mh ] ) @xcite , except near the quantum critical field driven transition at 1.6 t @xcite , where the mean - field theory is not expected to be accurate .
data were collected on the in5 time - of - flight instrument ( ill ) in its single - crystal set up .
measurements were carried out with an incident neutron wavelength of 6 and 8.5 , in zero field and under an applied magnetic field of 1.5 and 2.5 t along @xmath53 $ ] at 50 mk .
note that ref .
[ ] shows data along @xmath166 $ ] while the main body of ref .
[ ] shows in field data only .
figure [ fig - gap ] shows selected maps from the 8.5 data , allowing one to observe the spin gap at the zone center @xmath167 .
those data were obtained by integrating the full 4d data - set in narrow q - range by steps of 0.05 rlu " , and then reassembled to produce the maps . the data by ross _ et al .
_ is shown for comparison .
note that the latter was obtained with a resolution of 12 @xmath2ev @xcite , while it was 20 @xmath2ev in the present experiment .
the two acoustic - like modes coming from different magnetic domains are clearly visible along @xmath168 , along with the depletion of the intensity below the spin gap . | examples of materials where an order by disorder " mechanism is at play to select a particular ground state are scarce .
it has recently been proposed , however , that the antiferromagnetic xy pyrochlore er@xmath0ti@xmath0o@xmath1 reveals a most convincing case of this mechanism .
observation of a spin gap at zone centers has recently been interpreted as a corroboration of this physics . in this paper , we argue , however , that the anisotropy generated by the interaction - induced admixing between the crystal - field ground and excited levels provides for an alternative mechanism .
it especially predicts the opening of a spin gap of about 15 @xmath2ev , which is of the same order of magnitude as the one observed experimentally .
we report new high resolution inelastic neutron scattering data which can be well understood within this scenario .
geometrically frustrated magnetism is a forefront research topic within condensed matter physics , as testified by the wealth of exotic phenomena discovered over the past years @xcite .
for instance , the problem of an xy antiferromagnet on the pyrochlore lattice ( the celebrated lattice of corner sharing tetrahedra ) has been considered with much interest since this model displays an extensive classical degeneracy @xcite along with classical and quantum order by disorder ( obd ) effects @xcite . the elegant concept of obd @xcite is a cornerstone of ordering in frustrated condensed matter systems .
obd comes into play by selecting a ground state , either because fluctuations away from this particular configuration allow for a relative gain of entropy compared to other classically degenerate states , or because quantum mechanical zero point fluctuations define a minimum in the total energy . until now
, the number of confirmed examples for obd in real materials have remained scarce @xcite .
for obd to be an efficient selection mechanism , the classical ground state degeneracy must be extremely robust and the minimal theoretical model not openly subject to additional terms that would spoil the accidental emerging symmetry and lift the degeneracy .
recently , the xy pyrochlore antiferromagnet er@xmath0ti@xmath0o@xmath1 has been proposed as a candidate that satisfies these conditions in a rather compelling way @xcite . given the unique position of er@xmath0ti@xmath0o@xmath1 among frustrated quantum magnets , it is of foremost importance to scrutinize the soundness of this proposal .
the crystal electric field ( cef ) acting on the kramers er@xmath3 ion is responsible for a strong xy - like anisotropy , with easy magnetic planes perpendicular to the local @xmath4 ternary axes @xcite .
combined with antiferromagnetic interactions , an extensive classical degeneracy is expected @xcite . despite this degeneracy , er@xmath0ti@xmath0o@xmath1 undergoes a second order phase transition towards an antiferromagnetic non - collinear @xmath5 nel phase at @xmath6=1.2 k @xcite . in this configuration , denoted @xmath7 and depicted in fig .
[ fig1](a ) , the magnetic moments are perpendicular to the @xmath4 axes @xcite and make a zero net magnetic moment per tetrahedron . a theory based on a hamiltonian written in terms of interacting pseudospins 1/2 , each describing the
single - ion cef ground doublet , along with four anisotropic nearest - neighbor exchange parameters @xmath8 , has been proposed for er@xmath0ti@xmath0o@xmath1 @xcite .
for the set of parameters determined by inelastic neutron scattering ( ins ) experiments in a large applied magnetic field @xcite , the theory @xcite predicts a quantum obd selection of @xmath7 , on the basis of a linear spin wave calculation @xcite , as well as thermal obd at @xmath9 , also selecting @xmath7 @xcite .
another consequence of obd in er@xmath0ti@xmath0o@xmath1 is the opening of a spin gap , previously inferred from epr experiments @xcite as well as from deviation of the @xmath10 law in specific heat measurements @xcite , and very recently confirmed from ins measurements @xcite .
however , while the spin gap is a necessary consequence , it is not a definitive proof of this scenario : whatever the mechanism , a spin gap is expected since the ordered @xmath7 ground state breaks a global discrete symmetry @xcite . in this work ,
we follow a different route and consider an anisotropic bilinear exchange hamiltonian written for the er@xmath3 moments along with the cef contribution ( henceforth referred to as model a ) . as shown in ref .
@xmath11@xmath12 $ ] ,
an energetic selection of the @xmath7 state is possible for a specific range of anisotropic exchange parameters , owing to magneto - crystalline effects described by the cef ( see also ref .
@xmath11@xmath12 $ ] ) .
we model the spin excitations spectra within this cef - induced energetic selection scenario .
the comparison with new ins data allows one to determine a new set of anisotropic exchange parameters .
these are compatible with those determined from in - field ins experiments @xcite and based on the pseudospin 1/2 hamiltonian ( referred to as model b ) , from which quantum @xcite and thermal ( at @xmath6 ) obd @xcite is predicted .
both approaches lead to a spin gap of the correct order of magnitude . however , in model a , the spin gap results strictly from the admixing of the cef levels via the mean - field @xcite .
our results revive the debate regarding the ordering mechanism in pyrochlore antiferromagnets and illustrate that the argument of quantum obd being the chief governing mechanism causing @xmath7 ordering in er@xmath0ti@xmath0o@xmath1 is not definitive .
more generally , they emphasize the limitations of the projection onto the pseudospin 1/2 subspace ( shift from model a to model b ) with solely bilinear anisotropic spin - spin coupling in describing even qualitatively the physics of highly frustrated rare - earth pyrochlores .
magnetic configuration .
( b ) illustration of the energetic selection by molecular field induced magneto - crystalline anisotropy : the ground state energy of hamiltonian @xmath13 computed for @xmath14 t shows minima along the axes @xmath15 degrees .. , however slightly tilted out of the xy plane ( @xmath16 90 ) .
( c ) shows the corresponding tilt angle as a function of @xmath17 .
( d , e ) show respectively the curvature along @xmath18 and @xmath19 of the potential wells as a function of the magnetic field @xmath17 .
( f ) shows the evolution of the spin gap as a function of the renormalization @xmath20 of the stevens coefficients @xmath21 ( see text ) .
, width=302 ] _ cef energetic selection mechanism ( model a ) _
@xmath22 this approach considers a mean - field anisotropic bilinear exchange hamiltonian written for the er@xmath3 moments @xmath23 at sites @xmath24 of the pyrochlore lattice ( see refs .
$ ] and @xmath11@xmath12 $ ] ) .
it also contains explicitly the cef contribution , @xmath25 : @xmath26 where @xmath27 is written in terms of @xmath28 stevens operators @xcite .
the @xmath21 have been determined to fit a number of experiments , including the intensities and positions of the crystal field levels , as well as the susceptibility @xcite . in the following ,
those values are considered as fixed parameters .
@xmath29 denotes an anisotropic coupling tensor , defined in the @xmath30 frame attached to the er@xmath3@xmath22er@xmath3 bonds @xcite .
it is described by 3 symmetric parameters , @xmath31 , and an anti - symmetric exchange constant ( dzyaloshinskii - moriya like ) , @xmath32 @xcite . in this model
, the molecular field induces an admixture between the ground and excited cef levels , leading to an effective magnetic anisotropy .
this point is best evinced by considering the problem of an er@xmath3 ion in a local magnetic field @xmath33 : @xmath34 .
figure [ fig1](b ) shows the ground state energy of @xmath35 computed as a function of @xmath18 and @xmath19 ( in the local basis ) for a field @xmath14 t , which is the actual order of magnitude of the molecular field in er@xmath0ti@xmath0o@xmath1 ( see below ) .
minima along the 6-fold directions of the cef , _ slightly tilted _
away from the xy plane perpendicular to the local @xmath36 $ ] direction , are clearly observed . as shown in fig .
[ fig1](c ) , the tilt grows as @xmath37 but remains less that one degree for realistic values of @xmath17 .
the potential well in the vicinity of the minima can be approximated by a highly anisotropic harmonic potential whose curvature along @xmath18 ( denoted by @xmath38 ) is far steeper than along @xmath19 ( denoted by @xmath39 ) .
the average curvature , given by @xmath40 @xcite , is approximately @xmath41 k at @xmath14 t , a value about the same order of magnitude as the one emerging from zero point fluctuations ( quantum obd ) @xcite .
returning to the model a eq .
( [ eqmodela ] ) , this effective anisotropy combined with appropriate exchange parameters stabilizes the @xmath7 state ( note that other phases , namely a canted ferromagnet as well as the antiferromagnetic palmer and chalker state can also be stabilized depending on the values of @xmath42 @xcite ) . here , the moment direction at each of the 4 sites of a primitive tetrahedron basis is given by @xmath43 with @xmath44 ( in the cubic frame ) .
note that this is allowed by symmetry for the @xmath7 state , but not for the other component ( @xmath45 ) of the @xmath46 two - dimensional representation @xcite .
the dynamical structure factor @xmath47 that exposes the spin dynamics is modeled by a random phase approximation ( rpa ) calculation ( see refs @xmath11@xmath12 $ ] and supplemental material @xcite ) . for the relevant set of @xmath42 parameters
( see below ) , numerical calculations show the opening of a spin gap @xmath48 @xmath2ev at brillouin zone centers . to emphasize explicitly the influence of the cef levels in causing this gap ,
calculations have been performed for @xmath21 parameters multiplied by a renormalization coefficient @xmath20 .
this has the effect of rescaling all cef energy gaps by @xmath20 . in these calculations ,
the exchange coupling remains fixed , so that the molecular field and the nel temperature are essentially unchanged , but the effective magnetic anisotropy inherited by admixing of the ground cef doublet with excited cef states decreases with increasing @xmath20 .
@xmath47 is also globally unchanged , yet the spin gap gradually decreases and tends to zero for large @xmath20 , numerical calculations show that @xmath49 while @xmath50 . ultimately , as @xmath51 , the @xmath52 classical ground state degeneracy within the hilbert space strictly composed of of a direct product of single - ion cef ground doublets is recovered ( see fig .
[ fig1]f ) .
in that limit , quantum @xcite and thermal @xcite obd would become the sole mechanism able to lift the accidental degeneracy .
_ comparison with experiments _
@xmath22 to determine the @xmath42 couplings , the ins data were fitted to the calculated @xmath47 within the rpa .
the neutron measurements were performed on a large er@xmath0ti@xmath0o@xmath1 single crystal grown with the floating zone technique .
the crystal was inserted in a copper sample holder and attached on the cold finger of a dilution fridge , allowing one to cool the sample down to 50 mk .
data were collected on the in5 time - of - flight instrument ( ill ) which combines high flux with position sensitive detectors allowing for single crystal spectroscopy .
measurements were carried out with an incident neutron wavelength of 6 in zero field and under an applied magnetic field of 1.5 and 2.5 t along @xmath53 $ ] .
the spin excitation spectra measured along the high symmetry directions of the cubic unit cell at 50 mk are shown in fig .
[ fig - tof ] .
these results compare well with prior measurements ( see ref . [ ] and the supplemental material of ref .
[ ] ) .
because of the magnetic @xmath7 domains , the identification of the expected four different spin wave branches is not straightforward .
this means that the inelastic peaks in fig .
[ fig - tof ] contain several modes within the experimental resolution .
this is evidenced in the high resolution set - up , using a wavelength of 8.5 .
the highly gaussian ( nearly triangular ) profile of the resolution line inherent to the counter - rotating disk choppers instrument , which yields an energy resolution of about 20 @xmath2ev , reveals two acoustic - like modes coming from different magnetic domains ( see supplemental material ) .
using the same high resolution set - up , the zero field data confirm the opening of a spin gap at zone centers : as shown in fig .
[ fig - tof](c ) , the energy resolution permits to discriminate between the inelastic scattering and the strong bragg intensity at the @xmath54 ( 111 ) position . above the elastic line ,
the neutron intensity first shows a dip and then a peak , a behavior that is typical of a spin gap .
fitting the data through a lorentzian profile convoluted with the resolution function ( see blue line in fig .
[ fig - tof]c ) yields @xmath55 @xmath2ev .
this value compares very well with previous estimates @xcite . to determine the exchange parameters , we calculate @xmath47 assuming an equal population of the six @xmath7 magnetic domains . on the basis of exhaustive calculations as a function of the parameters @xmath42 in zero and applied magnetic field ,
the ins data were fitted by matching the location of the maximum ins intensity in several directions .
a good agreement is found for the following values : @xmath56 many others sets that capture independently the magnetization or the excitation spectrum can be found but the present determination provides values that capture all these experimental data @xcite . with these exchange parameters ,
the spin gap is evaluated at @xmath57 15 @xmath2ev , a value smaller than @xmath58 , but of the correct order of magnitude . *
4c coupling & model a & model b & ref . @xmath11@xmath12 $ ] + @xmath59 & 4.54 ( @xmath60 0.1 ) & 4.3 ( @xmath60 0.1 ) & 4.2 ( @xmath60 0.5 ) + @xmath61 & 5.84 ( @xmath60 0.1 ) & 6.0 ( @xmath60 0.1 ) & 6.5 ( @xmath60 0.75 ) + @xmath62 & 0.92 ( @xmath60 0.1 ) & -1.5 ( @xmath60 0.1 ) & -0.88 ( @xmath60 1.5 ) + @xmath63 & -0.87 ( @xmath60 0.1 ) & -2.2 ( @xmath60 0.1 ) & -2.5 ( @xmath60 1.8 ) + _ discussion _ @xmath22 it is instructive to compare our results ( eq . ( [ kresults ] ) ) with those found using the pseudospin @xmath64 approach ( model b ) @xcite .
the corresponding anisotropic exchange hamiltonian is described in detail in ref .
$ ] and is based on the anisotropic couplings @xmath65 acting between pseudospin @xmath64 components written in their local basis .
the calculation of the dynamical structure factor for this model was performed within the holstein - primakov approximation , using the _ spinwave _ software developed at the llb @xcite .
following the same fitting procedure as above , a set of parameters is obtained which largely confirm the results of ref .
@xmath11@xmath12 $ ] ( see table [ table1 ] ) .
the most striking point is that models a and b lead to very similar @xmath47 .
this is due to the fact that both models adopt a predominant effective hamiltonian with bilinear couplings in terms of pseudospin @xmath64 operators when projected in the cef ground doublet @xcite .
specifically , there is a relationship @xcite between the two sets of anisotropic exchange couplings based on the @xmath66 and @xmath67 land factors deduced from the ground state doublet wavefunctions @xcite .
table [ table1 ] , which allows to compare @xmath42 transformed in the @xmath68 language with the values determined from model b and from ref .
$ ] , shows that the sets of values are similar whether determined from either model .
further , owing to the obd mechanism , model b leads to a spin gap @xmath69= 21 @xmath2ev @xcite , a value of the same order of magnitude as the one ( @xmath70 = 15 @xmath2ev ) obtained in model a. while the maps in fig .
[ fig - tof ] demonstrate an overall agreement with experiment , some discrepancies are observed nonetheless , which equally affect models a and b. the most important difference concerns the acoustic - like mode stemming from ( 0,0,2 ) , which seems to disperse continuously up to 0.45 mev in the neutron data . within the experimental uncertainty , there is no gap opening when this branch crosses the optical one ( see the arrows in fig .
[ fig - tof](a ) , left and right columns of the @xmath71=0 panel ) .
such a gap opening occurs in the calculations , separating the acoustic branch from a higher energy optical branch .
furthermore , both models predict two well - separated modes at the zone centers @xmath72 = ( 1,1,1 ) , ( 2,2,0 ) and ( 0,0,2 ) at about 0.45 and 0.5 mev , whereas a single one is observed in experiment ( middle column of the @xmath71=0 panel in fig .
[ fig - tof](a ) ) .
more elaborate models are probably necessary to explain these features , taking into account the long - range part of the dipolar interaction or more complex coupling terms than bilinear ones . reference @xmath11@xmath12 $ ] reports the evolution of the gap as a function of temperature , and ascertains that it varies as the square of the @xmath7 order parameter .
the gap calculated in the framework of model a shows instead a linear evolution with the order parameter ( see fig .
[ fig - tof](d ) ) .
the difference between the linear and squared order variations is most pronounced in a narrow temperature range spanning @xmath73 to @xmath74 .
unfortunately , in this temperature range , we believe that the experimental uncertainty in ref . @xmath11@xmath12
$ ] is too large to allow one to discriminate between the two behaviors .
further experiments are planned to shed light on this issue .
to conclude , the present study shows that the molecular field induced admixture between cef levels generates an effective magnetic anisotropy as a plausible mechanism for an energetic selection of the @xmath7 state .
the proposed model captures a number of key features of the inelastic neutron scattering data , including the opening of a spin gap .
its order of magnitude shows that the proposed mechanism appears as efficient as the obd scenario , questioning the completeness of the projected pseudospin @xmath64 model @xcite as a minimal model of er@xmath0ti@xmath0o@xmath1 .
our study raises the question whether cooperating quantum @xcite and thermal @xcite order by disorder is the sole or even the principal mechanism for the selection of @xmath7 in this material , and whether its advocacy as a rare example of obd @xcite will stand the test of time . on a more positive note
, it seems plausible that quantum fluctuations _ and _ anisotropy induced by cef admixing cooperate to select @xmath7 in er@xmath0ti@xmath0o@xmath1 .
conversely , one might ask whether their _ competition _ might be responsible in part for some of the perplexing properties of er@xmath0sn@xmath0o@xmath1@xcite .
we acknowledge e. lhotel , l. jaubert and p. mcclarty for fruitful discussions , as well as f. damay for a careful reading of the manuscript .
we also thank s. turc ( cryogeny group at ill ) for his technical help while setting up the magnet and the dilution fridge .
research at the perimeter institute for theoretical physics is supported by the government of canada through industry canada and by the province of ontario through the ministry of economic development & innovation .
99 , edited by c. lacroix , p. mendels , and f. mila ( springer - verlag , berlin , 2011 ) .
j. s. gardner , m. j. p. gingras , and j. e. greedan , rev .
mod .
phys . * 82 * , 53 ( 2010 ) .
s. t. bramwell and m. j. p. gingras , science * 294 * , 1495 ( 2000 ) .
s. t. bramwell , m. j. p. gingras , and j. n. reimers , j. appl .
phys .
* 75 * , 5523 ( 1994 ) .
j. m. d. champion and p. c. w. holdsworth , j. phys . : condens .
matter * 16 * , s665 ( 2004 ) .
p. a. mcclarty , p. stasiak , and m. j. p. gingras , phys .
rev .
b * 89 * , 024425 ( 2014 ) .
m. e. zhitomirsky , m. v. gvozdikova , p. c. w. holdsworth , and r. moessner , phys .
rev .
lett . * 109 * , 077204 ( 2012 ) .
l. savary , k. a. ross , b. d. gaulin , j. p. c. ruff , and l. balents , phys . rev .
lett .
* 109 * , 167201 ( 2012 ) .
a. w. c. wong , z. hao , and m. j. p. gingras , phys .
rev .
b * 88 * , 144402 ( 2013 ) .
j. oitmaa , r. r. p. singh , b. javanparast , a. g. r. day , b. v. bagheri , and m. j. p. gingras , phys .
rev .
b * 88 * , 220404 ( 2013 ) .
h. yan , o. benton , l. d. c. jaubert , and n. shannon , f arxiv:1311.3501 ( 2013 ) .
m. e. zhitomirsky , p. c. w. holdsworth , and r. moessner phys .
rev .
b * 89 * , 140403(r ) ( 2014 ) . j. d. m. champion , m. j. harris , p. c. w. holdsworth , a. s. wills , g. balakrishnan , s. t. bramwell , e. cizmar , t. fennell , j. s. gardner , j. lago , d. f. mcmorrow , m. orendac , a. orendacova , d. mck .
paul , r. i. smith , m. t. f. telling , and a. wildes , phys .
rev .
b. * 68 * , 020401(r ) , ( 2003 ) .
j. villain , r. bidaux , j .-
p .
carton , and r. conte , j. phys . * 41 * , 1263 ( 1980 ) .
e. f. shender , sov .
phys .
jetp * 56 * , 178 ( 1982 ) .
t. yildirim , turk .
j. phys .
* 23 * , 47 ( 1999 ) .
w. j. blte , r.f .
wielinga and w. j. huiskamp , physica * 43 * , 549 ( 1969 ) .
m. j. harris , s. t. bramwell , t. zeiske , d. f. mcmorrow , and p. j. c. king , j. magn .
magn . mater . *
177 * , 757 ( 1998 ) .
r. siddharthan , b. s. shastry , a. p. ramirez , a. hayashi , r. j. cava , and s. rosenkranz , phys .
rev .
lett . * 83 * , 1854 ( 1999 ) .
a. poole , a. s. wills , and e. lelivre - berna , j. phys .
: condens .
matter * 19 * , 452201 ( 2007 ) .
j. p. c. ruff , j. p. clancy , a. bourque , m. a. white , m. ramazanoglu , j. s. gardner , y. qiu , j. r. d. copley , m. b. johnson , h. a. dabkowska , and b. d. gaulin , phys .
rev .
lett .
* 101 * , 147205 ( 2008 ) .
s. s. sosin , l. a. prozorova , m. r. lees , g. balakrishnan , and o. a. petrenko , phys .
rev .
b * 82 * , 094428 ( 2010 ) . p. dalmas de rotier , a. yaouanc , y. chapuis , s. h. curnoe , b. grenier , e. ressouche , c. marin , j. lago , c. baines , and s. r. giblin , phys . rev .
b * 86 * , 104424 ( 2012 ) .
k. a. ross , y. qiu , j. r. d. copley , h. a. dabkowska , and b. d. gaulin , phys .
rev .
lett . * 112 * 057201 ( 2014 ) .
p. a. mcclarty , s. h. curnoe , and m. j. p. gingras , journal of physics : conference series * 145 * , 012032 ( 2009 ) .
s. guitteny , s. petit , e. lhotel , j. robert , p. bonville , a. forget , and i. mirebeau , phys .
rev .
b * 88 * , 134408 ( 2013 ) .
b. g. wybourne , _ spectroscopic properties of rare earths _ , ( interscience , new york , 1965 ) .
j. jensen and a. r. mackintosh , _ rare earth magnetism _ , clarendon press , oxford , 1991 .
h.b .
cao , i. mirebeau , a. gukasov , p. bonville , and c. decorse , phys .
rev .
b * 82 * , 104431 ( 2010 ) . the crystal field is modeled by the following coefficients : @xmath75 k , @xmath76 k , @xmath77 k , @xmath78 k,@xmath79 k , @xmath80 k , in weybourne conventions .
see supplemental information .
b. z. malkin , t. t. a. lummen , p. h. m. van loosdrecht , g. dhalenne , and a.r .
zakirov , j. phys . : condens .
matter * 22 * , 276003 ( 2010 ) .
b. javanparast , a. day , z. hao , and m.j.p . gingras ; unpublished .
y. j. kao , m. enjalran , a. del maestro , h. r. molavian , and m. j. p. gingras , phys .
rev .
b * 68 * , 172407 ( 2003 ) . s. petit , in _ collection sfn _ 12 ( 2011 ) 105121 , published by edp sciences .
http://dx.doi.org/10.1051/sfn/201112006 .
j. d. thompson , p. a. mcclarty , h. m. ronnow , l. p. regnault , a. sorge , and m. j. p. gingras , phys .
rev .
lett . * 106 * 187202 ( 2011 ) .
s. petit , j. robert , s. guitteny , p. bonville , c. decorse , j. ollivier , h. mutka , m.j.p .
gingras , i. mirebeau , to appear in phys .
rev .
b(r ) . |
let @xmath2 be a graph with vertex set @xmath7 and edge set @xmath8 . if two vertices @xmath9 and @xmath10 share an edge , then they are called _ adjacent _ , otherwise they are called _ non - adjacent_. the open neighborhood of a vertex @xmath9 is @xmath11 @xmath10 is adjacent to @xmath9 in @xmath12 , and the closed neighborhood of @xmath9 is @xmath13 = n(u)\cup \{u\}$ ] . for a subset @xmath14 of @xmath7 ,
the set @xmath15 is adjacent to some @xmath16 is the open neighborhood of @xmath14 in @xmath2 .
two distinct vertices @xmath17 are adjacent twins if @xmath13=n[v]$ ] and non - adjacent twins if @xmath18 .
a set @xmath19 is called a twin - set of @xmath2 if @xmath20 are twins in @xmath2 for every pair of distinct vertices @xmath21 .
the distance @xmath22 between two vertices @xmath23 is the length of a shortest path between them and the diameter @xmath24 of @xmath2 is @xmath25 .
we refer to the book @xcite for the general graph theoretic notations and terminology not described in this paper . for a graph @xmath2 , an _ automorphism _ of @xmath2 is a bijective mapping @xmath0 on @xmath7 such that @xmath26 if and only if @xmath27 . the set of all automorphisms of a graph @xmath2 forms a group , denoted by @xmath28 , under the operation of composition of functions . for a vertex @xmath10 of @xmath2 ,
the set @xmath29 is the _ orbit _ of @xmath10 under @xmath28 , denoted by @xmath30 , and two vertices in the same orbit are _
similar_. if @xmath9 and @xmath10 are similar , then we write @xmath31 , if they are not similar , then we write @xmath32
. we define @xmath33 for some @xmath34 ( set of all vertices of @xmath2 which are more than one in their orbits ) .
also consider @xmath35 @xmath36 and @xmath37 .
note that if @xmath2 is a rigid graph ( a graph with @xmath38 ) , then @xmath39 .
a well - established fact is that every automorphism is also an _ isometry _ , that is , for @xmath40 and @xmath41 , @xmath42 .
+ an automorphism @xmath41 is said to _ fix _ a vertex @xmath43 if @xmath44 .
the set of automorphisms that fix a vertex @xmath43 is a subgroup of @xmath28 .
it is called the _ stabilizer _ of @xmath10 and is denoted by @xmath45 . for @xmath46 and @xmath41 ,
an automorphism @xmath47 is said to _ fix _ the set @xmath1 if for every @xmath48 , we have @xmath44 .
the set of automorphisms that fix @xmath1 is a subgroup @xmath49 of @xmath28 and @xmath50 . if @xmath1 is a set of vertices for which @xmath51 , then @xmath1 _ fixes _ the graph @xmath2 and we say that @xmath1 is a _ fixing set _ of @xmath2 .
erwin and harary introduced the _ fixing number _ , @xmath52 , of a graph @xmath2 in @xcite and it is defined as the minimum cardinality of a set of vertices that fixes @xmath2 . a fixing set containing @xmath52 number of vertices
is called a _ minimum fixing set _ of @xmath2 . the notion of fixing set
has equivalence with another notion , determining set , introduced by boutin in @xcite .
a set @xmath53 is said to be a _ determining set _ for a graph @xmath2 if whenever @xmath54 so that @xmath55 for all @xmath56 , then @xmath57 for all @xmath58 , i.e. , if two automorphisms @xmath47 and @xmath59 agree on @xmath60 , then they must agree on @xmath7 .
the equivalence between the definitions of a fixing set and a determining set was given in @xcite .
the minimum cardinality of a determining set of a graph @xmath2 , denoted by @xmath6 , is called the _ determining number _ of @xmath2 . in this paper
, we will use both the terms , fixing and determining .
term ` fixing ' is used to fix vertices and ` determining ' is used to determine automorphisms .
however , the term ` fixing set ' and ` determining set ' can be used interchangeably .
a vertex @xmath61 is called a _ fixed _
vertex if @xmath62 for all @xmath41 , i.e. , @xmath63 .
a vertex @xmath64 is said to _ fix a pair _ @xmath65 , if @xmath66 or @xmath67 whenever @xmath68 . if @xmath69 , then @xmath70 and hence question of fixing pair @xmath71 by any vertex of @xmath2 does not arise .
let @xmath65 and the set @xmath72 and @xmath73 for all @xmath74 is called the fixing set ( or an @xmath0-set ) relative to the pair @xmath71 .
it is also further assumed that if @xmath75 , then @xmath76 .
hence , @xmath77 . let @xmath61 and the set @xmath78 @xmath79 and @xmath80 , for all @xmath81 is called the _ fixed neighborhood _ of @xmath82 . also , if @xmath61 is a fixed vertex , then @xmath83 . the _ fixing graph _
, @xmath84 , of a graph @xmath2 is a bipartite graph with bipartition @xmath85 and a vertex @xmath86 is adjacent to a pair @xmath65 if @xmath82 fixes @xmath65 .
let a set @xmath87 , then @xmath88 @xmath82 fixes @xmath71 for some @xmath89 . in the fixing graph , @xmath84 , the minimum cardinality of a subset @xmath1 of @xmath7 such that @xmath90 is the fixing number of @xmath2 .
an upper bound on @xmath52 was given by erwin and harary by using another well - studied invariant , metric dimension , defined in the following way : let @xmath91 be a @xmath92-subset of @xmath7 and , for each vertex @xmath58 , define @xmath93 .
a @xmath92-set @xmath94 is called a resolving set for @xmath2 if for every pair @xmath20 of distinct vertices of @xmath2 , @xmath95 . the _ metric dimension _ ,
@xmath96 , is the smallest cardinality of a resolving subset @xmath94 .
a resolving set of minimum cardinality is a metric basis for @xmath2 . following lemma and theorem
were given in @xcite : if @xmath94 is a metric basis for @xmath2 , then @xmath97 is trivial . for every connected graph @xmath2 , @xmath98 .
considering the fact that the metric dimension is greater than or equal to fixing number and automorphisms preserve distances , metric dimension and fixing number are closely related notions @xcite .
cceres et al . studied this relation and answered the following question which appeared first in @xcite : can the difference between both parameters of a graph of order @xmath99 be arbitrarily large ?
the resolving number , @xmath100 , of a graph @xmath2 is the minimum @xmath92 such that every @xmath92-set of vertices is a resolving set of @xmath2 .
resolving number of a graph @xmath2 gives natural upper bound to the metric dimension of @xmath2 , i.e. , @xmath101 .
for every pair @xmath102 of integers with @xmath103 , existence of a connected graph @xmath2 with @xmath104 and @xmath105 , was given in @xcite .
motivated by definition of resolving number and resemblance between parameters metric dimension and determining number , we defined determined number of a graph @xmath2 in @xcite .
the _ determined number _ of a graph @xmath2 , @xmath5 , is the minimum @xmath92 such that every @xmath92-@xmath106 of vertices is a determining set for a graph @xmath2 .
it may be noted that @xmath107 . in @xcite , we found determined number of some well known graphs and discussed some of its properties . in the next section ,
we study some properties of @xmath0-sets following the study of @xmath108-sets by tomescu and imran @xcite .
to see the contribution of each vertex in resolving a graph , the concepts of resolving share and resolving percentage were introduced in @xcite . in the second section ,
we introduce the concept of the _ fixing share _ which tells the participation of each vertex of a fixing set in fixing a graph @xmath2 .
we also define the _ fixing percentage _ in @xmath2 , by using the concept of fixing share of each element of a fixing set of @xmath2 , which is the measure of the amount of fixing done by a fixing set in @xmath2 .
then we compute the fixing share and the fixing percentage in paths and cycles . in the third section
, we will prove existence of a graph @xmath2 for which @xmath4 for a given positive integer @xmath3 .
[ l1 ] if there exists an automorphism @xmath41 such that @xmath109 and if @xmath110 for some @xmath61 , then @xmath111 . since automorphism @xmath47 is an isometry ,
so @xmath112 and by hypothesis @xmath113 .
thus @xmath62 and @xmath114 .
hence , @xmath115 .
[ l2 ] let @xmath2 be a cycle of order @xmath99 and let @xmath116 be a pair of distinct vertices in @xmath2 , + @xmath117 @xmath118 , if @xmath99 is even and @xmath22 is odd .
+ @xmath119 @xmath120 , if @xmath99 is even and @xmath22 is even , where @xmath121 are the antipodal vertices with @xmath122 , @xmath123 .
+ @xmath124 @xmath125 , if @xmath99 is odd and @xmath64 is the vertex with @xmath126 .
[ p1 ] let @xmath2 be a path of order @xmath99 and @xmath127 where @xmath128 is adjacent to @xmath129 with @xmath130 , then @xmath131 and + ( i ) @xmath132 , if @xmath99 is even . + ( ii ) @xmath133 , if @xmath99 is odd .
two vertices @xmath9 and @xmath10 in a graph @xmath2 are said to be distance similar if @xmath134 for all @xmath135 .
[ l1 ] @xmath136 if and only if @xmath17 are distance similar .
if @xmath136 , then this implies that @xmath9 can be fixed either by fixing @xmath9 or by fixing @xmath10 .
now if @xmath9 and @xmath10 are not distance similar , then there exists a vertex @xmath137 such that @xmath138 .
if we fix @xmath139 by an automorphism @xmath140 , then @xmath141 implies @xmath142 .
now @xmath143 implies @xmath144 and fact that @xmath0 is an isometry implies @xmath145 , a contradiction .
thus @xmath17 are distance similar .
conversely , let @xmath17 are distance similar and @xmath146 , then there exists at least one vertex @xmath147 such that @xmath148 and @xmath149 for all @xmath150 . since @xmath0 is an isometry and @xmath148 , so @xmath151 , a contradiction that @xmath17 are distance similar . in a complete graph ,
every pair is distance similar .
therefore , we have the following corollary : [ l2 ] let @xmath2 be a complete graph of order @xmath99 and @xmath17 be a pair of distinct vertices , then @xmath65 and @xmath136 . [ l2 ]
let @xmath152 be a complete bipartite graph , + ( i ) @xmath136 , if both @xmath17 are in same partite sets .
+ ( ii ) @xmath118 , if both @xmath17 are not in same partite sets and @xmath153 .
+ ( iii ) @xmath76 , if both @xmath17 are not in same partite set and @xmath154 [ prop1 ] let @xmath2 be a graph of order @xmath155 and fixing number @xmath92 , then @xmath156 .
+ let @xmath157 and @xmath158 , then @xmath159 and @xmath160 .
let @xmath161 .
we will prove that @xmath162 .
suppose @xmath163 , then there are at most @xmath164 pairs in @xmath165 which are not adjacent to @xmath10 .
let @xmath166 , where @xmath167 .
note that , @xmath168 fixes @xmath169 for each @xmath170 , @xmath171 .
hence , @xmath168 is adjacent to pair @xmath169 in @xmath84 for each @xmath170 , @xmath171
. therefore , @xmath172 .
hence , @xmath173 , which is a contradiction .
thus , @xmath174 and consequently , @xmath175 .
if @xmath17 are twins in a connected graph @xmath2 , then @xmath176 for every vertex @xmath177 . from this we have the following remarks : [ c01 ] if @xmath17 are twins in a connected graph @xmath2 and @xmath1 is a fixing set for @xmath2 , then @xmath9 or @xmath10 is in @xmath1 . moreover , if @xmath178 and @xmath179 , then @xmath180 is also a fixing set for @xmath2 . [ r01 ]
if @xmath14 is a twin - set in a connected graph @xmath2 of order @xmath99 with @xmath181 , then every fixing set for @xmath2 contains at least @xmath182 vertices from @xmath14 .
[ d31 ] ( sole fixer ) let @xmath2 be a connected graph and @xmath1 be a minimum fixing set of @xmath2 .
let @xmath65 . if @xmath183 , then @xmath82 is called sole fixer for the pair @xmath71 . ,
title="fig:",width=302 ] consider the graph @xmath184 in figure 1 with vertex set @xmath185 and @xmath186 .
a fixing set of @xmath184 with minimum cardinality is @xmath187 .
since @xmath188 and @xmath189 , so @xmath190 is the sole fixer of the pair @xmath191 and @xmath192 is the sole fixer of the pair @xmath193 with respect to @xmath187 .
[ r02 ] let @xmath2 be a connected graph and @xmath1 be a minimum fixing set of @xmath2 .
let @xmath65 be a pair and @xmath194 be the sole fixer of @xmath71 with respect to @xmath1 .
then , @xmath82 is considered to has 1-share in fixing pair @xmath71 .
further , if @xmath71 is also fixed by , say @xmath195 other vertices of @xmath196 , then @xmath82 is considered to have @xmath197 share in fixing the pair @xmath71 .
for example in @xmath184 , the vertex @xmath190 is considered to have 1-share in fixing the pair @xmath191 with respect to @xmath198 . however , the vertex @xmath190 is considered to have @xmath199 -share in fixing the pair @xmath200 , as @xmath201
. if @xmath1 is a fixing set with minimum cardinality for a connected graph @xmath2 and @xmath194 , then the fixing share of @xmath82 in @xmath1 is defined as a measure of the amount of fixing done by @xmath82 in @xmath2 .
formally , we have the following definition : [ d32 ] ( fixing share ) let @xmath2 be a connected graph and @xmath1 be a minimum fixing set of @xmath2 .
let @xmath194 and @xmath202 .
we define a set @xmath203 and @xmath204 .
the fixing share , @xmath205 of an @xmath194 in @xmath1 is defined as @xmath206 in example of @xmath184 we have + @xmath207 and @xmath208 .
also @xmath209 , @xmath210 , @xmath211 , @xmath212 . thus @xmath213 and similarly @xmath214 .
[ d33 ] ( fixing sum and percentage ) let @xmath1 be a minimum fixing set for a connected graph @xmath2 and let @xmath215 , called the fixing sum in @xmath2 .
then the quantity @xmath216 , denoted by @xmath217 , is the measure of the amount of fixing done by @xmath1 in @xmath2 , and we call it the fixing percentage of @xmath1 in @xmath2 . for graph @xmath184 , @xmath218 and @xmath219 .
[ t32 ] let @xmath2 be a complete graph of order @xmath220 and @xmath1 be a minimum fixing set of @xmath2 .
let @xmath48 , then @xmath221 .
therefore @xmath222 and @xmath223 .
let @xmath224 . since @xmath2 is a complete graph ,
so @xmath225 and @xmath226 .
also cardinality of a minimum fixing set of @xmath2 is @xmath227 .
let @xmath228 be a minimum fixing set of @xmath2 , also @xmath1 is a twin - set in @xmath2 .
then for each @xmath170 , @xmath229 , @xmath230 , where @xmath231 and @xmath232 .
it can be seen that for each @xmath233 where @xmath231 , @xmath234 , @xmath235 ( because @xmath236 appears in exactly two @xmath237 , where @xmath238 ) .
also @xmath239 ( because @xmath240 appears in exactly one @xmath237 , where @xmath241 ) .
so @xmath242 .
hence , @xmath222 and @xmath223 .
[ t33 ] let @xmath2 be a path of order @xmath243 and @xmath244 be a minimum fixing set of @xmath2 . then fixing share @xmath245 and @xmath246 . as @xmath247 and one of the end vertices of @xmath2 forms a fixing set @xmath1 .
also @xmath248 and @xmath10 is the sole fixer of @xmath249 pairs in @xmath165 .
thus @xmath250 and @xmath251 for all @xmath252 and hence , @xmath245 .
thus @xmath246 for all @xmath243 .
since , a path @xmath253 is a graph with fixing number @xmath254 and a complete graph @xmath255 is the graph with fixing number @xmath227 , hence we can deduce that for a connected graph @xmath2 of order @xmath243 , @xmath256 and @xmath257 .
two vertices @xmath17 in a connected graph @xmath2 of order @xmath99 are said to be antipodal vertices if @xmath258 .
[ t34 ] let @xmath2 be a cycle graph of order @xmath259 and @xmath1 be a minimum fixing set of @xmath2 .
let @xmath48 , then the fixing share @xmath260 .
therefore @xmath261 and @xmath262 .
for all @xmath259 , let @xmath263 be a minimum fixing set of @xmath2 consisting of two non - antipodal vertices @xmath17 .
then @xmath264 .
two cases arise : + case i ( when @xmath99 is even ) + we notice that @xmath265 .
there are two types of pairs @xmath266 in @xmath267 .
@xmath117 there are @xmath268 pairs @xmath266 in @xmath267 such that @xmath269 ( as @xmath10 is the sole fixer for @xmath268 pairs in @xmath267 .
@xmath119 there are remaining @xmath270 pairs @xmath266 in @xmath267 such that @xmath271 ( as @xmath9 and @xmath10 equally participate to fix the remaining @xmath270 pairs in @xmath267 ) .
thus , @xmath272=\frac{1}{2}{n\choose 2}$ ] for each @xmath48 .
+ case ii ( when @xmath99 is odd ) + we notice that @xmath273 .
there are two types of pairs @xmath266 in @xmath267 .
@xmath117 there are @xmath274 pairs @xmath266 in @xmath267 such that @xmath269 ( as @xmath10 is the sole fixer for @xmath274 pairs in @xmath267 .
@xmath119 there are remaining @xmath275 pairs @xmath266 in @xmath267 such that @xmath271 ( as @xmath9 and @xmath10 equally participate to fix the remaining @xmath275 pairs in @xmath267 ) . thus , @xmath276=\frac{1}{2}{n\choose 2}$ ] for each @xmath48 .
+ so @xmath277 and hence @xmath262 .
a vertex of degree at least 3 in a tree is called a _
major vertex _ of tree .
an end vertex @xmath9 of a tree @xmath278 is said to be a terminal vertex of a major vertex @xmath10 , if @xmath279 for every other major vertex @xmath139 of @xmath278 .
the terminal degree , @xmath280 , of a major vertex @xmath9 of @xmath278 is the number of terminal vertices of @xmath9 .
let @xmath278 be a tree which is not a path and every major vertex of @xmath278 has different terminal degree and all terminal vertices rooted at a major vertex are at the same distance from that major vertex , @xmath1 is a fixing set of @xmath278 , @xmath48 be a terminal vertex of a major vertex of @xmath278 , then @xmath281 .
let @xmath10 be a terminal vertex of a major vertex @xmath9 of @xmath278 and @xmath280 is different from terminal degrees of remaining major vertices of @xmath278 , so @xmath30 contains remaining terminal vertices of major vertex @xmath9 of @xmath278 .
let @xmath282 and @xmath283 .
now , @xmath1 being a fixing set of @xmath278 contains @xmath284 vertices of @xmath30 , say @xmath285 .
the only pairs of vertices fixed by vertex @xmath10 are pairs of vertices in @xmath30 . thus fixing share of @xmath10
has equal contribution from @xmath286 .
@xmath287 + @xmath288 + .
+ . + @xmath289 + as no vertex in @xmath290 can fix pairs @xmath291
, @xmath292 .
thus each @xmath293 where @xmath294 , does not contain any pair @xmath295 , @xmath292 . thus @xmath296
in the next theorem , we will prove the existence of a graph @xmath2 for a given positive integer @xmath3 such that @xmath4 .
for every positive integer @xmath3 , there exists a graph @xmath2 such that @xmath4 .
we choose @xmath297 .
let @xmath298 be a bipartite graph , where @xmath299 and ordered set @xmath300 and both @xmath14 and @xmath94 are disjoint . before defining adjacencies
, we assign coordinates to each vertex of @xmath14 by expressing each integer @xmath233 @xmath301 in its base 2 ( binary ) representation .
we assign each @xmath302 @xmath303 the coordinates @xmath304 where @xmath305 @xmath306 is the value in the @xmath307 position of binary representation of @xmath233 . for integers
@xmath170 @xmath308 and @xmath233 @xmath301 , we join @xmath309 and @xmath310 if and only if @xmath311 .
this completes the construction of graph @xmath2 .
next we will prove that @xmath312 .
we denote @xmath313 is adjacent to @xmath314 and it is obvious to see that @xmath315 as if @xmath316 , then @xmath317 .
number of vertices in each @xmath318 is the permutation of @xmath92 digits in which digit 1 is appears @xmath170 times and digit 0 appears @xmath319 times , hence @xmath320 .
as @xmath315 , so minimum determining set @xmath60 must have @xmath321 vertices from each @xmath318 , @xmath322 , for otherwise if @xmath323 and @xmath324 for some @xmath170 , then there exists an automorphism @xmath41 such that @xmath325 because @xmath9 and @xmath10 have only one common neighbor @xmath309 , which is a contradiction that @xmath60 is a determining set .
moreover @xmath326 as each @xmath309 , @xmath327 , is fixed while fixing at least @xmath321 vertices in each @xmath318 .
hence , @xmath328 next we will find @xmath5 .
as order of @xmath2 is @xmath329 and set of all vertices of @xmath2 except one vertex forms a determining set of @xmath2 .
it can be seen that @xmath330 , for otherwise if @xmath331 and @xmath332 for some @xmath170 , then the set @xmath333 consisting of @xmath334 is not a determining set , which implies that @xmath330 .
hence for the graph @xmath2 , we have @xmath335 as required .
in this paper , we have described some properties of @xmath0-sets in graphs .
we have found bound on cardinality of edge set of the fixing graph of a graph @xmath2 .
we have defined fixing share of a vertex in a fixing set @xmath1 of a graph @xmath2 and studied it for vertices in fixing sets of some common classes of graphs .
finally we have proved existence of a graph @xmath2 for a given positive integer @xmath3 for which @xmath4 .
j. cceres , d. garijo , l. puertas , c. seara , on the determining number and the metric dimension of graphs , _ electron . j. combin . , _ 17 ( 2010 )
. g. chartrand , c. poisson , p. zhang , resolvability and the upper dimension of graphs , _ computers and maths . with appl .
39 ( 2000 ) 19 - 28 .
_ g. chartrand and l. lesniak , graphs and digraphs , @xmath336rd ed . , _ chapman and hall , london , _ 1996 .
d. erwin , f. harary , destroying automorphism by fixing nodes , _ disc .
, _ 306 ( 2006 ) ,
3244 - 3252 . m. fehr , s. gosselin , o. r. oellermann , the metric dimension of cayley digraphs , _ disc .
306(2006 ) , 31 - 41 . c. r. gibbons , j. d. laison , fixing numbers of graphs and groups , _ electron .
j. combin . , _ 16 research paper 39 , ( 2009 ) .
f. haray , methods of destorying the symmetries of a graph , _ bull .
malaysian math .
24(2 ) , 2001 .
f. harary , r. a. melter , on the metric dimension of a graph , _ ars combin . , _
2(1976 ) , 191 - 195 .
i. javaid , m. murtaza , m. asif , f. iftikhar , on @xmath92-sets related to automorphism determination of graphs , _ submitted for publication_. | in this paper , we deduce some properties of @xmath0-sets of connected graphs .
we introduce the concept of fixing share of each vertex of a fixing set @xmath1 to see the participation of each vertex in fixing a connected graph @xmath2 .
we also define a parameter , called the fixing percentage , by using the concept of fixing share , which is helpful in determining the measure of the amount of fixing done by the elements of a fixing set @xmath1 in @xmath2 .
it is shown that for every positive integer @xmath3 , there exists a graph @xmath2 with @xmath4 , where @xmath5 is the determined number and @xmath6 is the determining number of @xmath2 . |
after stone @xcite and zhegalkin @xcite , boole s main contribution to science @xcite can be understood as the realization that the mathematics of logical phenomena is controlled to a large extend by the field @xmath0 with two elements ; in contrast the mathematics of classical physical phenomena is controlled to a large extend by the field @xmath1 of real numbers .
the switch from @xmath2 to @xmath1 corresponds with a deep ontological jump from logical to physical phenomena .
the switch from @xmath1 to @xmath3 corresponds to the jump from classical to quantum physics . +
what makes the logic / physics jump possible is the fact that @xmath2 may be regarded as an object of two different categories . on the one hand , it is a field @xmath4 with sum and product defined by making @xmath5 the neutral element and @xmath6 the product unit . on the other hand ,
it is a set of truth values with @xmath5 and @xmath6 representing falsity and truth , respectively .
indeed , @xmath7 is a boolean algebra : a complemented distributive lattice with minimum @xmath5 and maximum @xmath6 .
the operations @xmath8 and @xmath9 correspond with the logical connectives @xmath10 , @xmath11 , and @xmath12 the two viewpoints are related by the identities : @xmath13 these identities , together with the inverse relation @xmath14 , allow us to switch back and forth from the algebraic to the logical viewpoint .
+ by and large , the logical and algebraic viewpoints have remained separated . in this work , in order to explore quantum - like phenomena in characteristic @xmath15 , we place ourselves at the jump .
our algebraic viewpoint is , in a sense , complementary to the quantum logic approach initiated by birkhoff and von neumann @xcite .
for example , while the meet in quantum logic is a commutative connective , our quantum analogue for the meet turns out to be non - commutative .
the appearance of non - commutative operations is a common tread in quantum physics @xcite . + we take as our guide the well - known fact that the quantization of canonical phase space may be identified with the algebra of differential operators on configuration space . in analogy with the real / complex case , we introduce @xmath16 the algebra of boolean differential operators on @xmath17 we provide a couple of presentations by generators and relations of @xmath16 , giving rise to the boole - weyl algebras @xmath18 and the shifted boole - weyl algebras @xmath19 .
we call these algebras the quantum boolean algebras .
we study the structural coefficients of @xmath18 and @xmath19 in various bases .
+ having introduced the quantum boolean algebras , we proceed to study them from the logical and set theoretical viewpoints . for us ,
the main difference between classical and quantum logic is that classical observations , propositions , can be measured without , in principle , modifying the state of the system ; quantum observations , in contrast , are quantum operators : the measuring process changes the state of the system .
indeed , regardless of the actual state of the system , after measurement the system will be in an eigenstate of the observable .
quantum observables are operators acting on the states of the system , and thus quite different to classical observables which are descriptions of the state of the system . + this work is organized as follows . in section [ s1 ]
we review some elementary facts on regular functions on affine spaces over @xmath2 . in section [ s2 ]
we introduce @xmath16 the algebra of boolean differential operators on @xmath20 . in section [ bwa ]
we introduce the boole - weyl algebra @xmath18 which is a presentation by generators and relations of @xmath16 .
we describe the structural coefficients of @xmath18 in several bases . in section [ s3 ]
we introduce the shifted boole - weyl algebra @xmath19 which is another presentation by generators and relations of @xmath16 .
we describe the structural coefficients of @xmath19 in several bases . in section [ s5 ]
we discuss the logical aspects of our constructions .
we propose a quantum operational calculus for which boolean differential operators play the same role that truth functions play in the classical propositional calculus . in section [ s5 ]
we adopt a set theoretical viewpoint ; we show that just as the classical propositional calculus is intimately related with @xmath21 , the algebra of sets of subsets of @xmath22 , the quantum operational calculus is intimately related with @xmath23 the algebra of sets of subsets of two disjoint copies of @xmath22 . in section [ s7 ] we make some closing remarks and mention a few topics for future research .
our main goal in this work is to study the boolean analogue for the weyl algebras , and to describe those algebras from a logical viewpoint .
fixing a field @xmath24 the weyl algebra @xmath25 can be identified with the @xmath24-algebra of algebraic differential operators on the affine space @xmath26 by definition @xcite the ring @xmath27 $ ] of regular functions on @xmath28 is the ring of functions @xmath29 such that there exists a polynomial @xmath30 $ ] with @xmath31 for all @xmath32 if @xmath24 is a field of characteristic zero , then the ring of regular functions on @xmath28 can be identified with @xmath33 $ ] the ring of polynomials over @xmath24 .
let @xmath34 be the derivations of @xmath33 $ ] given by @xmath35 for @xmath36=\{1, ... ,n \}.$ ] the algebra @xmath37 of differential operators on @xmath28 is the subalgebra of @xmath38)$ ] generated by @xmath39 and the operators of multiplication by @xmath40 for @xmath41 $ ] .
+ by definition , the weyl algebra @xmath42 is the @xmath24-algebra defined via generators and relations as @xmath43 where @xmath44 is the free associative @xmath24-algebra generated by @xmath45 and @xmath46 is the ideal generated by the relations @xmath47 and @xmath48 for @xmath36 $ ] , @xmath49 for @xmath50 $ ] , and @xmath51 for @xmath41 $ ] .
+ the weyl algebra @xmath42 comes with a natural representation @xmath52)$ ] which sends @xmath53 to @xmath39 and @xmath40 to the operator of multiplication by @xmath40 .
this representation induces an isomorphism of algebras @xmath54 + we proceed to study the analogue of the weyl algebras for the boolean affine spaces @xmath55 first , we review some basic facts on regular functions on @xmath56 let @xmath57 be the ring of maps from @xmath20 to @xmath2 with pointwise addition and multiplication . the ring @xmath58 $ ] of regular functions on @xmath20 is the sub - ring of @xmath57 containing the maps @xmath59 for which there exists a polynomial @xmath60 $ ] such that @xmath31 for all @xmath61 in this case @xmath58 $ ] is not a polynomial ring ; instead we have the following result .
_ there is an exact sequence of @xmath2-algebras @xmath62 \longrightarrow \mathbb{z}_2[\mathbb{a}^n ] \longrightarrow 0\ ] ] where @xmath63 is the ideal generated by the relations @xmath64 for @xmath65.$ ] _ the map @xmath66 \rightarrow \mathbb{z}_2[\mathbb{a}^n]$ ] sends a polynomial @xmath67 to the map @xmath68 given by @xmath69 clearly @xmath70 and @xmath40 define the same map @xmath71 .
thus @xmath63 is in the kernel of the map @xmath66 \rightarrow \mathbb{z}_2[\mathbb{a}^n]$ ] .
let @xmath72 $ ] be such that @xmath73 for all @xmath74 .
we can write @xmath75 where @xmath76 $ ] is a degree @xmath6 polynomial in @xmath77 .
therefore @xmath78 where @xmath79.$ ] since @xmath80 for all @xmath74 , then @xmath81 for all @xmath82 , and thus we obtain that @xmath83 we have shown that the polynomials @xmath84 and @xmath85 define the @xmath5 function thus , by induction , we conclude that @xmath86 , which implies that @xmath87 and therefore @xmath67 belong to @xmath63 .
therefore the ring @xmath58 $ ] of regular functions on @xmath20 can be identified with the quotient ring @xmath88 / < x_1 ^ 2 + x_1 , .... , x_n^2 + x_n > .\ ] ] often we think of @xmath20 as a ring , with coordinate - wise sum and product .
we identify @xmath20 with @xmath89 $ ] , the set of subsets of @xmath90 $ ] , as follows : @xmath74 is identified with the subset @xmath91 $ ] such that @xmath92 iff @xmath93 . with this identification
the product @xmath94 of elements in @xmath20 agrees with the intersection @xmath95 of the sets @xmath96 and @xmath97 ; the sum @xmath98 corresponds with the symmetric difference @xmath99 ; the element @xmath100 is identified with the complement @xmath101 of @xmath96 .
note that @xmath102 .
note also that @xmath103 $ ] denotes the set of all families of subsets of @xmath90 $ ] .
+ for @xmath104 $ ] let @xmath105 be the map such that @xmath106 for @xmath107 $ ] non - empty let @xmath108 $ ] be the monomial @xmath109 . also set @xmath110 the monomial @xmath111 defines the map @xmath112 given by @xmath113 for @xmath107 $ ] non - empty let @xmath114 $ ] be the monomial @xmath115 .
also set @xmath116 the monomial @xmath117 defines the map @xmath118 given by @xmath119 lemma [ bases ] below follows from the definitions and the m@xmath120bius inversion formula @xcite which can be stated as follows : given maps
@xmath121 \rightarrow r$ ] , with @xmath87 a ring of characteristic @xmath15 , then @xmath122 [ bases]_the following identities hold in @xmath58 $ ] : @xmath123 @xmath124 @xmath125 _
notice that @xmath126 = \mathrm{m}(\mathbb{z}_2^n , \mathbb{z}_2)$ ] , indeed a map @xmath127 can be written as @xmath128 @xmath129 from lemma [ bases ] we see that there are several natural bases for the @xmath2-vector space @xmath130 = \mathbb{z}_2[x_1, ...
,x_n ] / < x_1 ^ 2 + x_1 , .... , x_n^2 + x_n > , \ ] ] namely we can pick @xmath131 \}$ ] , @xmath132 \}$ ] , or @xmath133
\}.$ ] we use the following notation to write the coordinates of @xmath134 $ ] in each one of these bases @xmath135}f(a)m^a = \sum_{a \in \mathrm{p } [ n]}f_x(a)x^a = \sum_{a \in \mathrm{p}[n]}f_w(a)w^a.\ ] ] we obtain three linear maps @xmath136 from @xmath58 $ ] to @xmath57 . the coordinates @xmath137 ,
@xmath138 and @xmath139 are connected , via the m@xmath120bius inversion formula , by the relations : @xmath140 @xmath141 the maps @xmath142 fail to be ring morphisms .
instead we have the identities : @xmath143 we define a predicate @xmath144 on finite sets as follows : given a finite set @xmath96 , then @xmath145 holds iff the cardinality of @xmath96 is an odd number . in other words
, @xmath144 is the map from finite sets to @xmath2 such that @xmath146 iff the cardinality of @xmath96 is odd .
_ let @xmath147 $ ] .
a @xmath24-covering of @xmath148 $ ] is a sequence @xmath149 such that @xmath150 let @xmath24-@xmath151 be the set of @xmath24-coverings of @xmath96 by elements of @xmath152 .
then @xmath96 belongs to @xmath152 iff @xmath153-@xmath154 is odd for every @xmath155 .
indeed , let @xmath134 $ ] be given by @xmath156}1_c(a)x^a,\ ] ] where @xmath157 \rightarrow \mathbb{z}_2 $ ] is the characteristic function of @xmath152 .
since @xmath158 for every @xmath134 $ ] , we have that @xmath159}1_c(a)x^a = f = f^k= \sum_{a \in \mathrm{p } [ n]}\left ( \sum_{a_1 \cup ... \cup a_k = a}\prod_{i=1}^k 1_c(a_i)\right)x^a= \sum_{a \in \mathrm{p } [ n ] } o(k\mbox{-}cov_c(a ) ) x^a .\ ] ] we conclude that @xmath160 , and thus @xmath161 iff @xmath162 is odd . _
next we consider the algebra of differential operators on affine boolean spaces .
note that the derivations @xmath39 , although well - defined on @xmath66 $ ] , do not define operators on @xmath58 $ ] ; indeed if we had such an operator , then @xmath163 it is thus necessary to introduce an alternative definition for @xmath39 .
the boolean partial derivative @xmath164 of a map @xmath165 is given @xcite by @xmath166 where @xmath167 has vanishing entries except at position @xmath168 .
with this definition @xmath169 , the contradiction above does not arise , and we obtained well - defined operators @xmath170 \longrightarrow \mathbb{z}_2[\mathbb{a}^n].\ ] ] the operators @xmath39 fail to be derivations ; instead they satisfy the twisted leibnitz identity @xmath171 where the shift operators @xmath172 \longrightarrow \mathbb{z}_2[\mathbb{a}^n]$ ] are given by @xmath173 indeed : @xmath174g(x ) + f(x+e_i ) [ g(x+ e_i)+g(x)]\\ & = & \partial_if(x ) g(x ) + s_if(x)\partial_ig(x).\end{aligned}\ ] ] the operators @xmath39 are nilpotent , @xmath175 since : @xmath176 _ the algebra @xmath16 of boolean differential operators on @xmath20 is the subalgebra of @xmath177)$ ] generated by @xmath39 and the operators of multiplication by @xmath40 for @xmath41.$ ] _ [ rel]_the following identities hold for @xmath65 $ ] and @xmath178 : @xmath179 @xmath180 _ we have already shown that @xmath64 and @xmath181 . for the other identities we have that : @xmath182 @xmath183 @xmath184 @xmath185 @xmath186 @xmath187 @xmath188 @xmath189 the operator @xmath39 acts on the bases @xmath190 as follows : @xmath191 from these expressions we obtain that : + @xmath192 if and only if @xmath193 , i.e. @xmath194 + @xmath195 if @xmath196 and @xmath197 if @xmath198 , i.e. @xmath199 } } { \sum}f_x(a)x^{a - i } .$ ] + @xmath200 if @xmath196 and @xmath201 if @xmath198 , i.e. @xmath202 } } { \sum}f_w(a)w^{a - i}.$ ] + more generally one can show by induction , for @xmath203,$ ] that : @xmath204 by definition @xmath205)$ ] acts naturally on @xmath58 $ ] , so we have a map @xmath206 \rightarrow \mathbb{z}_2[\mathbb{a}^n].\ ] ] [ do ] _ consider maps @xmath207\times \mathrm{p}[n ] \rightarrow \mathbb{z}_2 $ ] and @xmath208 \rightarrow \mathbb{z}_2 $ ]
. + 1 . let @xmath209 } } { \sum}d(a , b)m^a\partial^b \in \mathrm{bdo}_n$ ] , @xmath210 } } { \sum}f(c)m^c \in \mathbb{z}_2[\mathbb{a}^n]$ ] , and @xmath211 } } { \sum}df(a)m^a $ ]
. then we have that @xmath212 2 .
let @xmath213 } } { \sum}d_x(a , b)x^a\partial^b \in \mathrm{bdo}_n$ ] , @xmath210 } } { \sum}f_x(c)x^c \in \mathbb{z}_2[\mathbb{a}^n]$ ] , and @xmath214 } } { \sum}df_x(e)x^e$ ] .
then @xmath215 _ @xmath216 } } { \sum}d(a , b)f(c)m^a\partial^bm^c= \underset{{a , e \subseteq b , c } } { \sum}d(a , b)f(c)m^am^{c+e}\ ] ] @xmath217 } \left ( \sum_{e \subseteq b } d(a , b)f(a+e ) \right)m^a.\ ] ] @xmath218 } } { \sum}d_x(a , b)f_x(c)x^a\partial^bx^c= \underset{{a , b \subseteq c } } { \sum}d_x(a , b)f_x(c)x^{a \cup c \setminus b}\ ] ] @xmath219 } } \left ( \underset{{\underset { a \cup ( c\setminus b)=e}{a,\ b \subseteq c } } } { \sum } d_x(a , b)f_x(c ) \right)x^e.\ ] ] [ egre ] _ for @xmath220 we have that @xmath221).$ ] _ note that @xmath222 ) \right ) = \mathrm{dim}(\mathbb{z}_2[\mathbb{a}^n])\mathrm{dim}(\mathbb{z}_2[\mathbb{a}^n])=2^n2^n=2^{2n}.$ ] the set @xmath223 \}$ ] has @xmath224 elements and generates @xmath16 as a @xmath2 vector space , thus it is enough to show that it is a linearly independent set .
suppose that @xmath225 } } { \sum}f(a , b)x^a\partial^b = \sum_{b \in \mathrm{p } [ n]}\left ( \sum_{a \in \mathrm{p } [ n]}f(a , b)x^a \right ) \partial^b = 0.\ ] ] pick a minimal set @xmath226 $ ] such that @xmath227 } } { \sum}f(a , c)x^a \neq 0.$ ] we have that : @xmath228}f(a , b)x^a \partial^b \right)(x^c)= \sum_{b \in \mathrm{p } [ n ] } \left ( \sum_{a \in \mathrm{p } [ n]}f(a , b)x^a \right ) \partial^b ( x^c ) = \sum_{a \in \mathrm{p } [ n]}f(a , c)x^a = 0.\ ] ] therefore , since @xmath132 \}$ ] is a basis for the regular functions we have that @xmath229 in contradiction with the fact @xmath227 } } { \sum}f(a , c)x^a \neq 0.$ ] we conclude that @xmath230 yielding the desired result
. put together proposition [ do ] and theorem [ egre ] provide a couple of explicit ways of identifying @xmath16 with @xmath231 the algebra of square matrices of size @xmath232 with coefficients in @xmath233 note that @xmath231 may be identified with @xmath234\times \mathrm{p}[n ] , \mathbb{z}_2).$ ] moreover , we can identify @xmath234\times \mathrm{p}[n ] , \mathbb{z}_2)$ ] with the set of directed graphs with vertex set @xmath89 $ ] and without multiple edges as follows : given a matrix @xmath235 its associated graph has an edge from @xmath97 to @xmath96 iff @xmath236 + let @xmath237 be the @xmath2-linear map constructed as follows .
consider the bases @xmath238 \}$ ] for @xmath16 and @xmath239 \}$ ] for @xmath58 $ ] .
for @xmath203 $ ] , let @xmath240 be the matrix of @xmath241 on the basis @xmath242 .
the action of @xmath241 on @xmath243 is given by @xmath244 therefore , the matrix @xmath240 is given for @xmath245 $ ] by the rule @xmath246 _ the graph of @xmath247 is show in figure [ g1 ] .
[ g1],height=113 ] for a second representation consider the @xmath2-linear map @xmath248 constructed as follows . consider the bases @xmath249 \}$ ] for @xmath16 and @xmath250 \}$ ] for @xmath58 $ ] . for @xmath203 $ ] let @xmath251 be the matrix of @xmath252 on the basis @xmath111 .
the action of @xmath252 on @xmath253 is given by @xmath254 therefore , the matrix @xmath251 is given for @xmath245 $ ] by the rule @xmath255 _ the graph associated to the matrix @xmath256 is shown in figure 2 . _ .
[ bra1],height=124 ]
let us motivate from the viewpoint of canonical quantization our definition of boole - weyl algebras .
canonical phase space , for a field @xmath24 of characteristic zero , can be identified with the affine space @xmath257 .
the poisson bracket on @xmath258 $ ] in canonical coordinates @xmath259 on @xmath257 is given by @xmath260 equivalently , the poisson bracket is given for @xmath261 $ ] by @xmath262 canonical quantization may be formulated as the problem of promoting the commutative variables @xmath40 and @xmath263 into non - commutative operators @xmath264 and @xmath265 satisfying the commutation relations : @xmath266= 0 , \ \ \ [ \widehat{y}_i , \widehat{y}_j ] = 0 , \
\ \ [ \widehat{y}_i , \widehat{x}_j ] = \delta_{i , j}.\ ] ] note that the free algebra generated by @xmath264 and @xmath265 subject to the above relations is precisely what is called the weyl algebra . +
now let @xmath267 and consider the affine phase spaces @xmath268 .
let @xmath259 be canonical coordinates on @xmath268 .
the analogue of the poisson bracket @xmath269 \otimes \mathbb{z}_2[\mathbb{a}^{2n } ] \rightarrow \mathbb{z}_2[\mathbb{a}^{2n}]\ ] ] can be expressed for @xmath270 $ ] as @xmath271 where @xmath272 and @xmath273 are the boolean derivatives along the coordinates @xmath274 and @xmath275 clearly , the full set of axioms for a poisson bracket will not longer hold , e.g. boolean derivatives are not derivations .
nevertheless , the bracket is still determined by its values on the canonical coordinates : @xmath276 canonical quantization consists in promoting the commutative variables @xmath40 and @xmath263 into non - commutative operators @xmath264 and @xmath265 satisfying the commutation relations : @xmath266= 0 , \ \ \
[ \widehat{y}_i , \widehat{y}_j ] = 0 , \ \ \ [ \widehat{y}_i , \widehat{x}_j ] = 0 , \
\mbox{for } \
i\neq j , \ \mbox{and } \ \ \ [ \widehat{y}_i , \widehat{x}_i ] _ { s_i}= 1.\ ] ] note that in the last relation we did not use the commutator but the shifted commutator @xmath277_{s_i}=fg + s_ifg ; \ ] ] this choice is expected since the operators @xmath278 are twisted derivations instead of usual derivations .
the relation @xmath279_{s_i}= 1 $ ] can be equivalently written using the commutator as @xmath280 = \widehat{y}_i + 1.\ ] ] we are ready to introduce the boole - weyl algebras @xmath18 , which we also call quantum boolean algebras .
the boole - weyl algebra @xmath18 is the free algebra generated by @xmath264 and @xmath265 subject to the relations above .
the algebras @xmath18 are the analogue of the weyl algebras in the boolean context .
_ the algebra @xmath18 is the quotient of @xmath281 , the free associative @xmath2-algebra generated by @xmath259 , by the ideal @xmath282 generated by the relations @xmath283 and @xmath284 for @xmath41 $ ] , @xmath285 and @xmath48 for @xmath36,$ ] and @xmath286 for @xmath50.$ ] _
[ qbas]_the map @xmath287)$ ] sending @xmath40 to the operator of multiplication by @xmath40 , and @xmath53 to @xmath39 , descends to an isomorphism @xmath288)$ ] of @xmath2-algebras . _ by theorem [ rel ] the given map descends . by definition
it is a surjective map @xmath289).\ ] ] moreover , this map is an isomorphisms since @xmath290)\right)$ ] . indeed using the commutation relations it is easy to check that the natural map @xmath88/<x_i^2 + x_i>\otimes \ \mathbb{z}_2[y_1, ... ,y_n]/<y_i^2 > \longrightarrow \mathrm{bw}_n\ ] ] is surjective .
if @xmath291 } } { \sum}f(a , b)x^a \otimes y^b$ ] is in the kernel of the latter map , then the boolean differential operator @xmath291 } } { \sum}f(a , b)x^a \partial^b$ ] would vanish , and therefore the coefficients @xmath292 must vanish as well .
thus @xmath293/<x_i^2 + x_i>)\mathrm{dim}(\mathbb{z}_2[y_1, ... ,y_n]/<y_i^2>)= 2^n2^n\ ] ] @xmath294)\mathrm{dim}(\mathbb{z}_2[\mathbb{a}^n])= \mathrm{dim}(\mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^n])).\ ] ] _ the map @xmath287)$ ] sending @xmath40 to the operator of multiplication by @xmath295 , and @xmath53 to the operator @xmath39 , descends to an isomorphism @xmath288)$ ] of @xmath2-algebras .
_ follows from the fact that @xmath296 and @xmath297 satisfy exactly the same relation as @xmath40 and @xmath298 [ coro]_any identity in @xmath18 involving @xmath40 and @xmath297 has an associated identity involving @xmath296 and @xmath297 obtained by replacing @xmath40 by @xmath299 _ _ for @xmath300 $ ] the following identities hold in @xmath18 : @xmath301 @xmath302 where @xmath303 _ \1 . by theorem [ qbas ] it is enough to show that the differential operators associated with both sides of the equation are equal .
consider the operator of multiplication by @xmath165 and let @xmath304 be any other map .
the twisted leibnitz rule @xmath305 can be extended , since @xmath306 and @xmath53 commute , to the identity : @xmath307 in particular we obtain that @xmath308 @xmath309 \2 .
we have that : @xmath310 @xmath311 where the last identity follows from the fact that @xmath312 , @xmath313 .
+ 3 . from the relations
@xmath314 for @xmath315 and @xmath316 we can argue as follows . if a letter @xmath53 is placed just to the left of a @xmath317 we can move it to the right , since these letters commute .
if instead we have a product @xmath318 , then three options arises : a ) @xmath53 moves to the right of @xmath40 ; b ) @xmath53 absorbs @xmath40 ; c ) @xmath40 and @xmath53 annihilate each other leaving an @xmath6 .
call @xmath319 the set of indices for which c ) occurs , and @xmath320 the set of indices for which either b ) or c ) occur .
then @xmath321 and the set for which option a ) occurs is @xmath322 .
thus the desired identity is obtained .
we have that : @xmath323 where @xmath324 _
@xmath325 @xmath326 @xmath327 @xmath328 _ _ for @xmath329 $ ] let @xmath330 $ ] and @xmath331
. then @xmath332 } o\{(a_1, ... ,a_k)\in a_1\times ... \times a_k \ | \ a_1\sqcup ...
\sqcup a_k = b\ \}y^{b}.\ ] ] in particular , for @xmath333 $ ] and @xmath334 , we get that @xmath335}o\{a_1, ... ,a_k\in a \ | \ a_1\sqcup ...
\sqcup a_k = b\ \ } y^{b}.\ ] ] for example , if @xmath336 $ ] then for @xmath337 we have that : @xmath335}o\{a_1, ... ,a_k\in \mathrm{p}[n ] \ | \
a_1\sqcup .... \sqcup a_{k } = b\ \ } y^{b}=
\sum_{b \in \mathrm{p}[n]}(k^{|b| } \ \mbox{mod } \ 2 ) \
y^{b},\ ] ] thus @xmath158 if @xmath24 is odd , and @xmath338 if @xmath24 is even .
_ from lemma [ bases ] we see that there are several natural basis for @xmath18 , namely : @xmath339 \ \ } , \ \ \ \ { x^ay^b \ | \ a , b \in \mathrm{p}[n ] \ \ } , \ \ \ \ { w^ay^b \ | \ a , b \in \mathrm{p}[n ] \ \ } .\ ] ] we write the coordinates of @xmath340 in these bases as : @xmath341 } f_m(a , b ) m^ay^b = \sum_{a , b \in \mathrm{p}[n ] } f_x(a , b ) x^ay^b = \sum_{a , b \in \mathrm{p}[n ] } f_w(a , b ) w^ay^b.\ ] ] these coordinates systems are connected by the relations : @xmath342 @xmath343 [ tt1 ] _ for @xmath344 the following identities hold for @xmath345:$ ] _ 1 .
@xmath346 2 .
@xmath347 @xmath348 \1 .
let @xmath349}f_{m}(a , b)m^ay^b,$ ] @xmath350}g_{m}(c , d)m^cy^d,$ ] then we have that : @xmath351}f_{m}(a , b)g_{m}(c , d ) m^ay^bm^cy^d\ ] ] @xmath352}f_{m}(a , b)g_{m}(c , d)\underset{{\underset { e\setminus d \subseteq a+c \subseteq b } { d\subseteq e } } } { \sum}m^ay^{e}\ ] ] @xmath353 \2 .
let @xmath349}f_{x}(a , b)x^ay^b,$ ] @xmath350}g_{x}(c , d)x^cy^d,$ ] then we have that : @xmath354}f_{x}(a , b)g_{x}(c , d)x^ay^bx^cy^d\ ] ] @xmath355 } \underset{{a \subseteq e,\ d \subseteq h } } { \sum}f_{x}(a , b)g_{x}(c , d)c(a , b , c , d , e , h)x^ey^h,\ ] ] where @xmath356 _ let @xmath357}f_m(a , b)m^ay^b$ ] and @xmath358}g_m(a , b)m^ay^b$ ] .
then @xmath359 for a non - vanishing summand we must have that @xmath360 , @xmath361 , and @xmath362 .
the conditions @xmath363 implies that @xmath364 and @xmath365 , thus @xmath366 .
we conclude that @xmath367 iff @xmath364 , @xmath368 and @xmath369 thus @xmath370 if @xmath371 .
for @xmath372 we get @xmath373 in particular we get that @xmath374 .
_ _ let @xmath375}m^ay^b= \sum_{a , b \in \mathrm{p}[n]}f_m(a , b)m^ay^b$ ] .
we have that @xmath376 note that if @xmath377 is not equal to @xmath90 $ ] , then there are an even number of choices for @xmath97 , thus we can assume that @xmath378 and @xmath379.$ ] the condition @xmath380 $ ] becomes trivial , and therefore @xmath381=0 $ ] if @xmath382 and @xmath383 if @xmath384 .
therefore we have that @xmath385}m^a.\ ] ] _ _ let @xmath386}x^{\{i\}}y^{\{i\ } } = \sum_{a , b \in \mathrm{p}[n ] } r_x(a , b ) x^ay^b\in \mathrm { bw}_n$ ] then we have that : @xmath387 @xmath388 clearly @xmath389 @xmath390 and @xmath391 moreover , we have that @xmath392 .
if @xmath393 , then @xmath394 for some @xmath41.$ ] if @xmath395 , then there are two options for @xmath320 leading to a vanishing coefficient .
thus we may assume that @xmath396 and then necessarily @xmath397 .
thus we conclude that @xmath398 if instead @xmath399 , then @xmath400 , @xmath401 and @xmath402 let @xmath403 and suppose that @xmath404 and @xmath405 .
then @xmath406 and @xmath407 .
all together we conclude that @xmath408}x^{\{i\}}y^{\{i\ } } + \sum_{i\neq j \in[n]}x^{\{i , j\}}y^{\{i , j\}}.\ ] ] _ _ from corollary [ coro ] we see that if @xmath409}w^{\{i\}}y^{\{i\}}$ ] then @xmath410}w^{\{i\}}y^{\{i\ } } + \sum_{i\neq j \in[n]}w^{\{i , j\}}y^{\{i , j\}}.\ ] ] equivalently , if @xmath409}y^{\{i\ } } + \sum_{i \in[n]}x^{\{i\}}y^{\{i\ } } $ ] then @xmath410}y^{\{i\ } } + \sum_{i \in[n]}x^{\{i\}}y^{\{i\}}+
\sum_{i\neq j \in[n]}y^{\{i , j\ } } + \sum_{i\neq j \in[n]}x^{\{i , j\}}y^{\{i , j\ } } + + \sum_{i\neq j \in[n]}\left(x^{\{i\ } } + x^{\{j\}}\right)y^{\{i , j\}}.\ ] ] _
so far , the operators @xmath39 have played the main role . in this section
we take an alternative viewpoint and let the operators @xmath306 be the main characters .
recall that the boolean derivatives and the boolean shift operators are related by the identities @xmath411 and @xmath412 . for @xmath413
$ ] let @xmath414 and @xmath415 we get that : @xmath416 [ dos ] _ consider maps @xmath207\times \mathrm{p}[n ] \rightarrow \mathbb{z}_2 $ ] and @xmath208 \rightarrow \mathbb{z}_2 $ ] .
let @xmath209 } } { \sum}d(a , b)m^a s^b \in \mathrm{bdo}_n$ ] , @xmath210 } } { \sum}f(c)m^c \in \mathbb{z}_2[\mathbb{a}^n]$ ] , and @xmath211 } } { \sum}df(a)m^a $ ] .
then we have that @xmath417}{\sum}}d(a , b)f(a+b).\ ] ] 2 .
let @xmath213 } } { \sum}d_x(a , b)x^as^b \in \mathrm{bdo}_n$ ] , @xmath210 } } { \sum}f_x(c)x^c \in \mathbb{z}_2[\mathbb{a}^n]$ ] , and @xmath418 } } { \sum}df_x(d)x^d$ ] .
then @xmath419 _ + + 1 .
@xmath420 } } { \sum}d(a , b)f(c)m^as^bm^c= \underset{{a , b , c } } { \sum}d(a , b)f(c)m^am^{b+c}$]@xmath421@xmath422 + @xmath423 } \left ( \sum_{b \in \mathrm{p}[n ] } d(a , b)f(a+b ) \right)m^a.$ ] + 2 .
@xmath420 } } { \sum}d_x(a , b)f_x(c)x^as^bx^c= \underset{{a , e \subseteq b \cap c } } { \sum}d_x(a , b)f_x(c)x^{a \cup c \setminus e}$ ] + @xmath424 } } \left ( \underset{{\underset { a \cup ( c\setminus e)=d}{a , \ e \subseteq b \cap c } } } { \sum } d_x(a , b)f_x(c ) \right)x^d.$ ] proposition [ dos ] and theorem [ egre ]
provide a couple of explicit ways of identifying @xmath16 with @xmath231 the algebra of square matrices of size @xmath232 with coefficients in @xmath233 consider the @xmath2-linear map @xmath237 sending @xmath425 to @xmath426 the matrix of @xmath425 on the basis @xmath242 .
the action of @xmath425 on @xmath243 is given by @xmath427 if @xmath428 and @xmath5 otherwise .
therefore , the matrix @xmath426 is given for @xmath245 $ ] by the rule @xmath429 _ the graph of the matrix @xmath430 is shown in figure [ g3 ] . _ .
[ g3],height=37 ] for a second representation consider @xmath2-linear the map @xmath248 sending @xmath431 to @xmath432 the matrix of @xmath431 on the basis @xmath111 .
the action of @xmath431 on @xmath253 is given by @xmath433 therefore , the matrix @xmath434 is given for @xmath245 $ ] by the rule @xmath435 _ the graph of the matrix @xmath436 is shown in figure [ g4 ] . _ .
[ g4],height=86 ] next we introduce the shifted boole - weyl algebra @xmath19 , a quite useful and easy to handle presentation for the algebra of boolean differential operators @xmath16 . _ the algebra @xmath19 is the quotient of @xmath437 , the free associative @xmath2-algebra generated by @xmath438 , by the ideal @xmath439 generated by the relations @xmath440 and @xmath441 for @xmath41 $ ] , @xmath285 and @xmath48 for @xmath36,$ ] and @xmath442 for @xmath50.$ ] _ _
the map @xmath443)$ ] sending @xmath40 to the operator of multiplication by @xmath40 , and @xmath306 to the shift operator in the @xmath168-direction , descends to an isomorphism @xmath444)$ ] of @xmath2-algebras . _
the map @xmath445 sending @xmath40 to @xmath40 and @xmath306 to @xmath446 descends to an algebra isomorphisms @xmath447 .
the result then follows from theorem [ qbas ] . _
the map @xmath443)$ ] sending @xmath40 to the operator of multiplication by @xmath295 , and @xmath306 to the shift operator in the @xmath168-direction , descends to an isomorphism @xmath444)$ ] of @xmath2-algebras .
_ follows from the fact that @xmath296 and @xmath448 satisfy exactly the same relation as @xmath40 and @xmath449 _ any identity in @xmath19 involving @xmath40 and @xmath448 has an associated identity involving @xmath296 and @xmath448 obtained by replacing @xmath40 by @xmath299 _ _ for @xmath300 $ ] the following identities hold in @xmath19 : @xmath450 @xmath451 _ \1 . for any @xmath134
$ ] we have that : @xmath452 \2 .
@xmath453 + 3 . from the identity
@xmath454 we see that as @xmath306 pass to the right of @xmath40 , it may or may not absorb @xmath40 .
the set @xmath455 is the set of indices for which @xmath40 is absorbed by @xmath306 .
@xmath456 _ _ @xmath457}m^c= m^{\overline{c}}s^{[n ] } , \ \
\overline{c}=[n]\setminus c. \ \ \ \ 2 . \ m^{\overline{c}}s^{[n]}m^cs^d = m^{\overline{c}}s^{\overline{d}}.\ ] ] @xmath458}x^c=\sum_{k \subseteq c } x^{k}s^{[n]}. \ \ \ \ \ \ \ \ \ \ \ 4 .
\ x^as^{[n]}x^cs^d = \sum_{k \subseteq c } x^{a \cup k}s^{\overline{d}}.\ ] ] from lemma [ bases ] we see that there are several natural basis for @xmath18 , namely : @xmath459 \ \ } , \ \ \ \ { x^as^b \ | \ a , b \in \mathrm{p}[n ] \ \ } , \ \ \ \ { w^as^b \ | \ a , b \in \mathrm{p}[n]\ \ } .\ ] ] we write the coordinates of @xmath460 in these bases as : @xmath341 } f_{m , s}(a , b ) m^as^b = \sum_{a , b \in
\mathrm{p}[n ] } f_{x , s}(a , b ) x^as^b = \sum_{a , b \in \mathrm{p}[n ] } f_{w , s}(a , b ) w^as^b.\ ] ] these coordinates systems are connected by the relations : @xmath461 @xmath462 [ tt2 ] _ for @xmath463 the following identities hold for @xmath464 $ ] : _ 1 .
@xmath465 } } { \sum}f_{m , s}(a , c)g_{m , s}(a+c , b+c).$ ] 2 .
@xmath466 } o\{k \subseteq b\cap
a \cup c\setminus k = e \ \}f_{x , s}(a , b)g_{x , s}(c , b+h).$ ] \1 .
let @xmath349}f_{m , s}(a , b)m^as^b,$ ] @xmath350}g_{m , s}(c , d)m^cs^d,$ ] then we have that : @xmath467}f_{m , s}(a , b)g_{m , s}(c , d ) m^as^bm^cs^d=\ ] ] @xmath468 } f_{m , s}(b+c , b)g_{m , s}(c , d ) m^{b+c}s^{b+d}\ ] ] @xmath469 } \left ( \underset{{\underset { b+d = f } { b+c = e } } } { \sum}f_{m , s}(b+c , b)g_{m , s}(c , d)\right ) m^{e}s^{f}\ ] ] @xmath469 } \left ( \underset{{b \in \mathrm{p}[n ] } } { \sum}f_{m , s}(e , b)g_{m , s}(e+b , f+b)\right ) m^{e}s^{f}.\ ] ] \2 .
let @xmath349}f_{x , s}(a , b)x^as^b$ ] , @xmath350}g_{x , s}(c , d)x^cs^d,$ ] then we have that : @xmath467 } f_{x , s}(a , b)g_{x , s}(c , d ) x^as^bx^cs^d\ ] ] @xmath470 , \ k
\subseteq b\cap c } f_{x , s}(a , b)g_{x , s}(c , d ) x^{a \cup c\setminus k}s^{b+d}\ ] ] @xmath471}\left(\underset { { \underset { \underset{a \cup c\setminus k = e , \
b+d = h } { k \subseteq b\cap c } } { a , b , c , d \in \mathrm{p}[n ] } } } { \sum } f_{x , s}(a , b)g_{x , s}(c , d ) \right ) x^{e}s^{h}\ ] ] @xmath471}\left(\underset { { \underset { \underset{a \cup c\setminus k = e } { k \subseteq b\cap c } } { a \subseteq e , \
b , c \in \mathrm{p}[n ] } } } { \sum } f_{x , s}(a , b)g_{x , s}(c , b+h ) \right ) x^{e}s^{h}=\ ] ] @xmath471}\left(\sum_{a \subseteq e , \ b , c \in \mathrm{p}[n ] } o\{k \subseteq b\cap c \ | \ a \cup c\setminus k = e \ \}f_{x , s}(a , b)g_{x , s}(c , b+h ) \right ) x^{e}s^{h}.\ ] ] _ suppose that @xmath349}f_{m , s}(a , b)m^as^b,$ ] and @xmath350}g_{m , s}(c , d)m^cs^d,$ ] are actually regular functions on @xmath20 , i.e. @xmath472 if @xmath473 , and @xmath474 if @xmath475
. a non - vanishing term in the formula @xmath476 } } { \sum}f_{m , s}(a , c)g_{m , s}(a+c , b+c)\ ] ] must have @xmath477 , and then we must also have that @xmath478 , and @xmath479 thus in this case the product @xmath480 is , as expected , just the pointwise product of functions on @xmath20 . _ _ let @xmath349}f_{m , s}(a , b)m^as^b,$ ] and suppose that @xmath481}g_{m , s}(c , d)m^cs^d$ ] is such that @xmath474 if @xmath482.$ ] then a non - vanishing summand in the formula @xmath476 } } { \sum}f_{m , s}(a , c)g_{m , s}(a+c , b+c)\ ] ] can only arise for @xmath378 . therefore @xmath483,b+\overline{a}).$ ] for example , we have that @xmath484}m^as^{\overline{a } } \right)\left ( \sum_{d \in \mathrm{p}[n]}m^{[n]}s^d \right ) = \sum_{a , b \in \mathrm{p}[n]}m^as^b.\ ] ] as another example consider @xmath349}m^as^b$ ] and @xmath485}s^{[n]}$ ] . in this case
we get that : @xmath486}m^as^b\right)\left ( m^{[n]}s^{[n ] } \right)= \sum_{a\in \mathrm{p}[n]}m^as^a.\ ] ] _
in this section we study quantum boolean algebras from a logical viewpoint .
we assume the reader to be familiar with the language of operads and props @xcite .
first we review the basic principles of classical propositional logic @xcite which may be summarized as : * propositions are words in a certain language .
propositions are either simple or composite .
let @xmath22 be the finite set of simple propositions and @xmath487 the set of all propositions .
composite propositions are obtained from the simple propositions using the logical connectives .
there are several options for the choice of connectives , the most common ones being @xmath488 * a truth function @xmath489 is associated to each proposition @xmath490 , where @xmath491 is the set of maps from @xmath22 to @xmath2 . the map @xmath492 $ ] sending @xmath493 to @xmath494 is such that @xmath495 is the evaluation at @xmath96 for each @xmath496 , and @xmath497 @xmath498 @xmath499 @xmath500 where the action of the connectives on truth functions are induced by the corresponding operations on @xmath233 the map @xmath501 $ ] is surjective , and there is a systematic procedure to tell when two propositions have the same associated truth function . for our purposes , it is convenient to describe @xmath487 using the binary connectives @xmath502 and @xmath503 , and the constants @xmath504 in logical terms the product @xmath502 is the logical conjunction , @xmath503 the exclusive or , and @xmath5 and @xmath6 represent falsity and truth , respectively .
+ @xmath487 is defined recursively as the set of words in the symbols @xmath505 such that : * @xmath506 , @xmath507 and @xmath508 * if @xmath509 then @xmath510 and @xmath511 are also in @xmath487 .
we defined recursively the notion of sub - words on @xmath487 as follows .
for all @xmath512 we set : @xmath493 is a sub - word of @xmath493 ; @xmath493 is a sub - word of @xmath510 and @xmath511 ; if @xmath493 is a sub - word of @xmath513 and @xmath513 is a sub - word @xmath514 , then @xmath493 is a sub - word of @xmath514 .
+ next we define an equivalence relation @xmath515 , also denoted by @xmath516 , on @xmath487 .
given @xmath517 we set : @xmath518 the relation @xmath519 can be defined in purely syntactic terms as follows : @xmath493 and @xmath513 are related iff @xmath520 or there exists a sequence @xmath521 , for some @xmath155 , such that @xmath522 , and @xmath523 , and @xmath524 is obtained from @xmath525 by replacing a sub - word of @xmath525 by an equivalent word according to the following relations : * associativity and commutativity for @xmath502 and @xmath503 : @xmath526 @xmath527 , + @xmath528 @xmath529 .
* distributivity : @xmath530 * additive and multiplicative units : @xmath531 and @xmath532 . * additive nilpotency : @xmath533 . * multiplicative idempotency : @xmath534 .
let @xmath535 be the category of sets and mappings , and @xmath536 the category of finite sets .
let @xmath537 the functor sending @xmath22 to the free boolean ring generated by @xmath22 , i.e. @xmath538 for a map
@xmath539 the map @xmath540 is such that @xmath541 for any @xmath542 . let @xmath543 be the functor given by @xmath544 thus we have that : @xmath545,\ ] ] i.e. @xmath546 is the ring of boolean functions on @xmath547 + note that the sequence @xmath548 \ | \
n\geq 0 $ ] } can be identified with the endomorphism operad of @xmath2 in @xmath535 , i.e. @xmath549=\mathrm{m}(\mathbb{z}_2^n,\mathbb{z}_2)$ ] as sets with @xmath550-actions .
+ let @xmath551 be the free operad in @xmath535 generated by @xmath552 and @xmath553 for given @xmath554 , the set of all propositions @xmath487 is actually equal to the free @xmath551-algebra generated by @xmath22 .
+ we also denote by @xmath551 the functor @xmath555 sending @xmath22 to the @xmath487 , and @xmath539 to its natural extension @xmath556 sending @xmath22 to @xmath557 via @xmath137 , and respecting the logical connectives . + let @xmath558 be the category of equivalence relations .
objects in @xmath558 are pairs @xmath559 where @xmath87 is an equivalence relation on @xmath560 .
a morphism @xmath561 in @xmath558 is a map @xmath562 such that @xmath563 we have a functor @xmath564 sending @xmath565 to the quotient set @xmath566 .
+ the pair @xmath567 yields a functor @xmath568 sending @xmath22 to @xmath569 and the corresponding functor @xmath570 moreover , the results of section [ s1 ] imply the identification @xmath571 \ \
\mbox{as boolean rings}.\ ] ] note also that we obtain a purely syntactic description of the operad @xmath548 \ | \
n\geq 0 \}$ ] as @xmath572/\mathbb{r}[n ] \ | \
n\geq 0 \}.\ ] ] classical logic main concern is the pre - order @xmath573 of entailment on @xmath487 .
recall that @xmath574 $ ] is a poset with @xmath575 if @xmath576 for all @xmath577 .
+ for @xmath517 , we set @xmath578 iff @xmath579 or equivalently , iff there is @xmath580 such that @xmath581 the entailment relation @xmath573 can be defined syntactically terms as follows : @xmath582 next we move from the classical to the quantum situation .
as mentioned in the introduction , quantum observables are operators instead of propositions . in analogy with the classical case
we identify operators with words in a certain language .
truth functions are replaced by boolean differential operators .
we think of quantum logic as arising from the following principles : * we find again the simple / composite dichotomy . for a set @xmath22
we let @xmath583 be a set disjoint from @xmath22 whose elements are of the form @xmath584 for @xmath585 given @xmath22 we let @xmath586 be the set of all quantum observables ; @xmath586 is a set of words in a certainly language to be specified below .
elements of @xmath586 are called operators .
the set of operators @xmath586 is obtained from the set of simple operators @xmath587 using the binary connectives product @xmath502 and sum @xmath503 , and the constants @xmath588 ; i.e. @xmath586 is defined recursively as the set of words in the symbols @xmath589 such that : * * @xmath590 @xmath591 , and @xmath592 * * if @xmath593 then @xmath510 and @xmath511 are also in @xmath586 . * the logical interpretation of the connectives @xmath502 and @xmath503 are as follows .
the product @xmath594 correspond with the logical @xmath11 , but there is also a temporal dimension to it : @xmath594 may be interpreted as `` act with the operator @xmath513 , and then act with the operator @xmath595 '' the connective @xmath503 correspond to @xmath596 , the exclusive or .
the constants @xmath5 and @xmath6 may be interpreted as : `` reset to @xmath5 '' and `` leave it as it is '' .
* let @xmath597)=\mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[a \
x])$ ] be the algebra of boolean differential operators on @xmath547 we think of @xmath598 as the quantum analogue of the boolean algebra @xmath599 $ ] of truth functions .
just as we have a map from propositions to truth functions , we have a map @xmath600)$ ] from operators to boolean differential operators given by : * * @xmath495 is the operator of multiplication by @xmath96 , for @xmath496 . *
* @xmath601 is @xmath602 the boolean derivative along @xmath96 , for @xmath496 . * * @xmath603 and @xmath604 for @xmath605 * * @xmath606 , the operator identically equal to @xmath5 , and @xmath607 is the identity operator .
we think of the composition @xmath608 of operators in @xmath609)$ ] as the quantum analogue of the meet @xmath610 , or equivalently the product , on @xmath599 = \mathrm{m}(\mathbb{z}_2^x , \mathbb{z}_2)$ ] .
indeed @xmath608 is an extension of the classical meet .
consider the inclusion map @xmath599 \rightarrow \mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^x])$ ] sending @xmath611 $ ] into the operator of multiplication by @xmath137 .
this map is additive and multiplicative , thus showing that the quantum structures are , as they should , an extension of the classical ones .
+ the map @xmath600)$ ] turns out to be surjective , and there is a well - defined procedure to tell when two operators are assigned the same boolean differential operator .
sub - words are defined in @xmath586 as in the classical .
we define an equivalence relation @xmath515 , also denoted by @xmath516 , on @xmath586 .
for @xmath612 we set : @xmath518 @xmath519 is defined syntactically as follows : @xmath493 and @xmath513 are related iff @xmath520 or there exists a sequence @xmath521 , for some @xmath155 , such that @xmath522 , and @xmath523 , and @xmath524 is obtained from @xmath525 by replacing a sub - word of @xmath525 by an equivalent word according to the following relations : * associativity for the product : @xmath613 * associativity and commutativity for @xmath503 : @xmath528 @xmath614 .
* distributivity : @xmath530 * additive and multiplicative units : @xmath531 and @xmath532 * additive nilpotency : @xmath533 . *
multiplicative idempotency and nilpotency : @xmath615 and @xmath616 , for @xmath585 * commutation relations : @xmath617 and @xmath618 , for @xmath619 @xmath620 for @xmath621 and @xmath622 for @xmath585 let @xmath623 be the category of finite sets and bijections @xcite .
we regard @xmath624 , and @xmath625)$ ] as functors @xmath626 .
the pair @xmath627 defines a functor @xmath628 .
the results of section [ bwa ] imply the following identification @xmath629 ) \ \ \ \mbox{in particular } \ \ \ \mathbb{o}(x)/\mathbb{r}(x ) = \mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^x]).\ ] ] next we define the quantum entailment pre - order @xmath573 on @xmath586 .
first we define a pre - order on @xmath609)$ ] ; for @xmath630)$ ] we set @xmath631 for example , if @xmath84 and @xmath85 are projections onto the subspaces @xmath632 and @xmath633 of @xmath599 $ ] , respectively , then @xmath634 iff @xmath635 + for @xmath612 , we set @xmath578 iff there is @xmath636 such that @xmath637 equivalently , the entailment relation @xmath638 can be defined syntactically as : @xmath639 as expected , quantum entailment is an extension of classical entailment .
indeed , we have functors @xmath640 @xmath641 , \leq ) , ( \mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^ { ( \ ) } ] ) , \leq ) $ ] from @xmath623 to the category of pre - ordered sets .
these functors fit into the following commutative diagram of natural transformations : @xmath642 \ar[d ] & ( \mathbb{o } , \vdash ) \ar[d ] \\ ( \mathbb{z}_2[\mathbb{a}^ { ( \ ) } ] , \leq ) \ar[r ] & ( \mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^{(\ ) } ] ) , \leq ) } \ ] ] where the top horizontal arrow is the natural inclusion , the bottom horizontal arrow sends @xmath137 into the operator of multiplication by @xmath137 , and the vertical arrows are the valuation maps from propositions and operators to truth functions and differential operators , respectively .
+ our previous considerations yield a syntactic presentation of the diagonal of the endomorphism prop of @xmath643 $ ] in the category @xmath2-@xmath644 .
indeed we have that : @xmath645/\mathbb{r}[n]=\mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^{n}])= \mathrm{end}_{\mathbb{z}_2}(\mathbb{z}_2[\mathbb{a}^1]^{\otimes n}).\ ] ]
the link between classical propositional logic and the algebra of sets arises as follows .
recall that there is a map @xmath646 sending each proposition to its truth function .
since @xmath647 can be identified with @xmath648 we obtain a map @xmath649 assigning to each proposition @xmath493 a set @xmath494 of subsets of @xmath22 .
moreover , the logical connectives intertwine nicely with the set theoretical operations on subsets , namely : @xmath650 we stress the , often overlooked , fact that classical propositional logic describes the set theoretical operations present in @xmath648 that are common to all sets of the form @xmath651 i.e. the extra algebraic structures present in @xmath652 when @xmath653 play no significative role in the logic / set theory relation outlined above .
this is why whereas @xmath654 have been massively studied , the algebraic structures on @xmath655 , for @xmath656 , have seldom attracted any attention .
+ we proceed to consider the analogue statements in the quantum scenario .
we present our results for sets of the form @xmath90 $ ] .
it should be clear , however , that the same constructions apply for arbitrary finite sets . + as in the classical case we have a map @xmath657 ) \ $ ] sending operators ( words in a certain language ) to boolean differential operators .
as shown in section [ bwa ] it is possible to identify @xmath177)$ ] with the boolean - weyl algebra @xmath18 , and with the symmetric boolean - weyl algebra @xmath19 .
moreover , we described several explicit bases for these algebras .
for example , each @xmath658 can be written in an unique way as : @xmath659}f(a , b)x^ay^b.\ ] ] thus boolean differential operators can be identified with maps from @xmath89 \times \mathrm{p}[n]$ ] to @xmath2 , and we get the identifications : @xmath660 ) \simeq \mathrm{bdo}_n \simeq \mathrm{bw}_n \simeq \mathrm{m}(\mathrm{p}[n ] \times \mathrm{p}[n ] , \mathbb{z}_2 ) \simeq \mathrm{p}(\mathrm{p}[n ] \times \mathrm{p}[n ] ) \simeq \mathrm{p}\mathrm{p}([n ] \sqcup [ n]).\ ] ] we adopt the following conventions .
we identify @xmath90 \sqcup [ n]$ ] with the set @xmath661=\{1,2, ... ,n , \widetilde{1 } , \widetilde{2 } , ... , \widetilde{n } \}.\ ] ] given @xmath662 $ ] we let @xmath663 be the corresponding subset of @xmath664=\{\widetilde{1 } , \widetilde{2 } , ... , \widetilde{n } \}.$ ] an element @xmath665 $ ] will be written as @xmath666 with @xmath667 $ ] .
note that we have a natural map @xmath668 \rightarrow \mathrm{p}[n ] \times \mathrm{p}[n]$ ] given by @xmath669 .
we use indices without tilde to denote monomials of regular functions , and indices with tilde to denote the boolean derivatives or shift operators .
the identification @xmath177 ) = \mathrm{p}\mathrm{p}[n,\widetilde{n}]$ ] allow us to give set theoretical interpretations to the algebraic structures on boolean differential operators .
unlike the classical case , the quantum structures are not defined for an arbitrary sets of the form @xmath652 , quite the contrary , they very much depend on the fact that @xmath670.$ ] + below we consider pairs @xmath671 where @xmath632 is a @xmath2-algebra and @xmath672 is an @xmath632-module .
a morphism @xmath673 between such pairs , is given by @xmath2-linear maps @xmath674 and @xmath675 such that @xmath137 is an algebra morphism , and @xmath676 for all @xmath677 + the additive structure @xmath678 \times \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \rightarrow \mathrm{p}\mathrm{p}[n,\widetilde{n}]$ ] on @xmath679 $ ] is given by @xmath680 we consider several isomorphic products @xmath681 and @xmath682 on @xmath679 $ ] displaying different combinatorial properties .
the various products correspond with the various bases for @xmath18 and @xmath19 .
_ there are maps @xmath683 \times \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \rightarrow \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \ \mbox{and } \
\circ:\mathrm{p}\mathrm{p}[n,\widetilde{n } ] \times \mathrm{p}\mathrm{p}[n ] \rightarrow \mathrm{p}\mathrm{p}[n]$ ] , turning @xmath679 $ ] into a @xmath2-algebra and @xmath684 $ ] into a module over @xmath679,$ ] such that the pair @xmath685 , \mathrm{p}\mathrm{p}[n])$ ] is isomorphic to @xmath686 via the maps @xmath687 _ from theorem [ tt1 ] and proposition [ do ] we see that the desired products @xmath608 are constructed as follows . for @xmath688,$ ]
the product @xmath689 $ ] is given by @xmath690 \
o\ { b \in \mathrm{p}[n],c \in b
\ | \ c_2 \subseteq a_2 , \ a_1 \sqcup \widetilde{b } \in a , \
a_2 \setminus c_2
\subseteq a_1 + c_1 \subseteq b \ } \right\}.\ ] ] let @xmath691 $ ] and @xmath692 $ ] , then @xmath693 $ ] is given by @xmath694 \ \
o\{b \subseteq c \in \mathrm{p}[n ] \
| \ a \sqcup \widetilde{c } \in a , \ a + b \in f \ } \ \right\ } .\ ] ] _ in @xmath695 $ ] we have that @xmath696 indeed @xmath697 if there is a odd number of pairs @xmath698 with certain properties .
note that @xmath699 , @xmath700 , @xmath701 and thus necessarily @xmath702 and @xmath703 .
moreover , we must have that @xmath704 , that is @xmath705 .
thus either @xmath706 or @xmath707 yielding the desired result .
_ _ for @xmath333 $ ] set @xmath708 .
let @xmath692 $ ] , the we have that : @xmath709 \ \ | \ \ o\{b \subseteq
c \in \mathrm{p}[n ] \ | \
\widetilde{c } \in a , \ a + b \in f \ } \ \right\ } .\ ] ] note that @xmath710}m^a=1 $ ] and thus : @xmath711 _ _ for @xmath712 $ ] let @xmath713 \ |
\ a_1 = a_2 \in a \
let @xmath714 $ ] , then @xmath715 \
o\{e \subseteq a \
\in f \ } \ \right\ } \
\ \mbox{and therefore}\ ] ] @xmath716 _ _ there are maps @xmath717 \times \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \rightarrow \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \ \mbox{and } \ \bullet : \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \times \mathrm{p}\mathrm{p}[n ] \rightarrow \mathrm{p}\mathrm{p}[n]$ ] such that the pair @xmath685 , \mathrm{p}\mathrm{p}[n])$ ] is isomorphic to @xmath686 via the maps @xmath718 _ from theorem [ tt1 ] and proposition [ do ] we see that the desired products @xmath719 are constructed as follows . for @xmath688,$ ]
the product @xmath720 $ ] is such that @xmath721 iff @xmath722 let @xmath691 $ ] and @xmath692 $ ] , then @xmath723 $ ] is given by @xmath724 \ \
\ o\ { b \in a , c \in f \ | \ b_2 \subseteq c , \ b_1 \cup ( c \setminus b_2)=a\ } \
\right\ } .\ ] ] _ in @xmath695 $ ] we have that @xmath725 indeed , we must have @xmath726 and @xmath727 , and thus there are three options for @xmath728 $ ] , namely , @xmath729 @xmath730 and @xmath731 giving rise to the sets @xmath732 respectively . _ _ for @xmath733 $ ] set @xmath734 .
let @xmath692 $ ] then we have that : @xmath735 \
o\{b \in a , c \in f \ | \ b \subseteq c , \
c \setminus b = a\ } \
\right\}\ ] ] therefore we get that @xmath736 _ _ for @xmath712 $ ] let @xmath713 \
| \ a_1 = a_2 \in a \ \}$ ] .
let @xmath714 $ ] , then @xmath737 @xmath738 in particular we have that : @xmath739}\bullet \mathrm{p}[n]= \left\ { \emptyset \right\}$ ] and thus @xmath740 } } { \sum}x^a \partial^a \right)\circ \left(\underset{{b \in \mathrm{p}[n ] } } { \sum}x^b\right)=1.\ ] ] _ _ there are maps @xmath741
\times \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \rightarrow \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \ \mbox{and } \
\star:\mathrm{p}\mathrm{p}[n,\widetilde{n } ] \times \mathrm{p}\mathrm{p}[n ] \rightarrow \mathrm{p}\mathrm{p}[n]$ ] , turning @xmath679 $ ] into a @xmath2-algebra and @xmath684 $ ] into a module over @xmath679,$ ] such that the pair @xmath685 , \mathrm{p}\mathrm{p}[n])$ ] is isomorphic to @xmath686 via the maps @xmath742 _ from theorem [ tt2 ] and proposition [ dos ] we see that the desired products @xmath743 are constructed as follows . for @xmath688,$ ]
the product @xmath744 $ ] is given by @xmath745 \
o\ { b \in \mathrm{p}[n ] \ | \
a_1 \sqcup \widetilde{b } \in a , \
( a_1 + b ) \sqcup \widetilde{(a_2 + b ) } \in b \ } \right\}.\ ] ] let @xmath691 $ ] and @xmath692 $ ] , then @xmath746 $ ] is given by @xmath747 \ \ | \ \
o\{b \in \mathrm{p}[n ] \
| \ a \sqcup \widetilde{b } \in a , \ a + b \in f \ } \right\ } .\ ] ] _ in @xmath695 $ ] we have that @xmath748 from the equation @xmath749 we see that @xmath750 and @xmath751 .
also we must have @xmath752 , which holds , and @xmath753 which implies that @xmath754 _ _ for @xmath333 $ ] set @xmath708 .
let @xmath692 $ ] , the we have that : @xmath755 \ \ | \ \
o\{b \in a \
\ a + b \in f \ } \ \right\}.\ ] ] therefore @xmath756 } m^b \right)=oa\sum_{a \in \mathrm{p}[n ] } m^a .\ ] ] _ _ for @xmath712 $ ] set @xmath713 \
| \ a_1 = a_2 \in a \
let @xmath714 $ ] , then @xmath757 therefore @xmath758 where @xmath759 if @xmath760 and @xmath761 if @xmath762 . _ _ let @xmath763 be as in the previous example , then @xmath764 if @xmath765 and @xmath766 otherwise .
indeed , @xmath665 $ ] belongs to @xmath767 if @xmath768 , i.e. @xmath769 and @xmath770 , i.e. @xmath765 and @xmath771 _ _ for @xmath712 $ ] set @xmath772 \ | \ \overline{a_1 } = a_2 \in a \ \}$ ] .
then @xmath773 if @xmath90 \in a$ ] and @xmath766 if @xmath90 \not \in a$ ]
. indeed , @xmath665 $ ] belongs to @xmath774 if @xmath775 , i.e. @xmath776 and @xmath777 , i.e. @xmath90 \in a$ ] and @xmath778 _ _ there are maps @xmath779 \times \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \rightarrow \mathrm{p}\mathrm{p}[n,\widetilde{n } ] \ \mbox{and } \
\ast:\mathrm{p}\mathrm{p}[n,\widetilde{n } ] \times \mathrm{p}\mathrm{p}[n ] \rightarrow \mathrm{p}\mathrm{p}[n]$ ] , turning @xmath679 $ ] into a @xmath2-algebra and @xmath684 $ ] into a module over @xmath679,$ ] such that the pair @xmath685 , \mathrm{p}\mathrm{p}[n])$ ] is isomorphic to @xmath686 via the maps @xmath780 _ from theorem [ tt2 ] and proposition [ dos ] we see that the desired products @xmath743 are constructed as follows . for @xmath688,$ ]
the product @xmath781 $ ] is given by @xmath782 \
| \ o\ { b , c , d , e \in \mathrm{p}[n ] \ |
\ e \subseteq c\cap d,\ b\cup d\setminus e= a_1,\ b \sqcup
\widetilde{c } \in a , \ d
\sqcup \widetilde{(c+a_2 ) } \in b \ }
\right\}.\ ] ] let @xmath691 $ ] and @xmath692 $ ] , then @xmath783 $ ] is given by @xmath784 \ \ | \ \
o\{b , c , d \in \mathrm{p}[n ] , e \in f \ | \ c\subseteq d\cap e , \ b\cup e\setminus c= a , \ b
\sqcup \widetilde{d } \in a \ \ } \ \right\ } .\ ] ] _ in @xmath695 $ ] we have that @xmath785 indeed we must have @xmath786 , @xmath787 , @xmath788 , and @xmath789 . since @xmath790 , there are two options , either @xmath791 and then @xmath792 and @xmath793 , or @xmath794 and then @xmath795 and @xmath796 _ _ for @xmath733 $ ] set @xmath734 .
let @xmath692 $ ] then we have that : @xmath797 \
o\{c \in \mathrm{p}[n ] , d \in a , e \in f \ | \ c \subseteq d \cap e ,
\ e\setminus c = a\ } \ \right\}.\ ] ] therefore @xmath798 _ _ for @xmath712 $ ] let @xmath713 \
| \ a_1 = a_2 \in a \ \}$ ] .
let @xmath714 $ ] , then @xmath799 \ \
\ o\{b\in a , c \in \mathrm{p}[n ] , e \in f \ | \ c\subseteq b\cap e , \
b\cup e\setminus c= a \ \ } \ \right\}.\ ] ] @xmath800 _
we introduced an approach for the study of quantum - like phenomena in characteristic @xmath15 .
our approach is developed as follows : + 1 ) quantization of the canonical phase space @xmath801 over a field @xmath24 of characteristic zero may be identified with the @xmath24-algebra of algebraic differential operators on @xmath802 the weyl algebra provides an explicit description by generators and relations of the latter algebra .
+ 2 ) classical propositional logic may be identified , to a good extend , with the study of regular functions on @xmath20 .
we introduced the algebra @xmath16 as a suitable analogue for algebraic differential operators on @xmath20 .
we showed that @xmath16 coincides with the algebra of linear endomorphisms of regular functions on @xmath20 .
we introduced a couple of presentations by generators and relations for @xmath16 , namely , the quantum boolean algebras @xmath18 and @xmath803 + 3 ) we shift back from the algebro - geometric viewpoint , and study the quantum boolean algebras from the logical and set theoretical viewpoints .
+ our work leaves several open questions and problems for future research : + 1 ) we considered the structural aspects of quantization in characteristic @xmath15 .
the dynamical aspects will be considered elsewhere .
+ 2 ) we studied the analogue for the weyl algebra in characteristic @xmath804 recently , @xcite , there have been a remarkable interest in characteristic @xmath6 .
it would be interesting to study the analogue for the weyl algebra in characteristic @xmath6 .
+ 3 ) categorification of the weyl algebra has been considered in @xcite ; categorification of the boole - weyl and shifted boole - weyl algebras remain to be addressed .
+ 4 ) the symmetric powers of weyl algebras and linear boolean algebras in characteristic zero were studied in @xcite and @xcite , respectively .
the analogue problems for the quantum boolean algebras and the linear quantum boolean algebras are open .
+ 5 ) our logical interpretation of quantum boolean algebras was based on a specific choice of connectives .
it remains to study other connectives , perhaps with a more direct logical meaning .
+ a. deitmar , schemes over f1 , in g. van der geer , b. moonen , r. schoof ( eds . ) , number fields and function fields - two parallel worlds , progress in mathematics 239 , birkh@xmath806user , basel 2005 , pp . | we introduce quantum boolean algebras which are the analogue of the weyl algebras for boolean affine spaces .
we study quantum boolean algebras from the logical and set theoretical viewpoints . |
often , when a mathematical object is drawn in some `` random '' manner , the probability that it is isomorphic to a given object is inversely proportional to the size of the automorphism group of the latter .
the cohen
lenstra heuristics @xcite , which make predictions on the distribution of class groups of `` random '' algebraic number fields , are , in fact , a special case of this rule , provided that one passes to arakelov class groups . however , arakelov class groups may have infinitely many automorphisms , so that a difficulty arises in comparing the sizes of their automorphism groups .
this difficulty is addressed in the present paper ; the number theoretic implications will appear in a later one .
our main results , formulated as theorems [ thm : qgintro ] and [ thm : mainintro ] below , express that , for certain pairs of modules @xmath0 and @xmath1 over certain types of ring , one can meaningfully define the ratio of the size of the automorphism group @xmath2 of @xmath1 to the size of @xmath3 , even when their orders @xmath4 and @xmath5 are infinite .
if @xmath3 can be naturally embedded in @xmath2 as a subgroup of finite index , then the ratio mentioned may be defined to be that index .
our approach consists of giving a canonical definition of such an index in a more general situation .
we shall write @xmath6 for the `` index of automorphism groups '' that we define .
denote by @xmath7 the ring of integers , by @xmath8 the field of rational numbers , by @xmath9 the multiplicative group of positive rational numbers , by @xmath10 $ ] the group ring of a group @xmath11 over a ring @xmath12 , and by @xmath13 the index of a subgroup @xmath14 of a group @xmath11 . by `` module ''
we shall always mean `` left module '' .
[ thm : qgintro ] let @xmath11 be a finite group , let @xmath15 be a finitely generated @xmath16$]-module , and put @xmath17 is a finitely generated @xmath18$]-module with @xmath19 as @xmath16$]-modules@xmath20 .
then there exists a unique function @xmath21 such that 1 . if @xmath0 , @xmath22 , @xmath1 , @xmath23 and @xmath24 , @xmath25 , then @xmath26 ; 2 . if @xmath0 , @xmath1 , @xmath27 , then @xmath28 ; 3 .
if @xmath29 , and @xmath30 is a submodule of finite index , then with @xmath31 and @xmath32 mapping @xmath33 to @xmath34 , one has @xmath35 to explain part ( c ) , we remark that it is not hard to show that one has @xmath36 , and that the three cardinal numbers @xmath37 , @xmath38 , @xmath39 are finite ( see section [ sec : autcomm ] ) . since these three numbers may be thought of as the ratio of the sizes of @xmath2 and @xmath14 , of @xmath14 and @xmath40 , and of @xmath3 and @xmath40 , respectively , one may think of the expression in ( c ) as the ratio of the sizes of @xmath2 and @xmath3 .
the same argument shows that one has indeed @xmath41 if @xmath2 and @xmath3 are finite . as an example , let @xmath11 be the trivial group , and put @xmath42 . then each @xmath36 is isomorphic to the direct sum of @xmath43 with a finite abelian group @xmath44 , and @xmath3 is isomorphic to a semidirect product @xmath45 , where both @xmath46 and @xmath47 are finite . writing @xmath29 similarly , and `` cancelling '' @xmath48 , one is led to believe that @xmath49 making this informal argument rigorous ( see proposition [ prop : fingengps ] ) , one discovers that if a function as in theorem [ thm : qgintro ] exists , it must be given by the formula just stated . however , that this formula does define a function meeting all conditions , in particular ( c ) , is not obvious . likewise , for general @xmath11 the uniqueness statement of theorem
[ thm : qgintro ] is easy by comparison to the existence statement .
our proof of theorem [ thm : qgintro ] is given in section [ sec : proofs ] .
there is little doubt that one can prove theorem [ thm : qgintro ] using a suitable theory of covolumes of arithmetic groups .
instead , we will give an entirely algebraic proof , obtaining the theorem as a special case of a much more general result , of which the formulation requires some terminological preparation .
_ isogenies .
_ a _ group isogeny _ is a group homomorphism @xmath50 with @xmath51 and @xmath52 , and its _ index _
@xmath53 is defined to be @xmath54 . for a ring @xmath12 ,
an @xmath12_-module isogeny _ is an @xmath12-module homomorphism that is an isogeny as a map of additive groups .
a _ ring isogeny _ is a ring homomorphism that is an isogeny as a map of additive groups .
the index of an isogeny of one of the latter two types is the index of the induced group isogeny on the additive groups . _
commensurabilities . _ if @xmath55 , @xmath56 are objects of a category @xmath57 , then a _ correspondence _ from @xmath55 to @xmath56 in @xmath57
is a triple @xmath58 , where @xmath59 is an object of @xmath57 and @xmath60 and @xmath61 are morphisms in @xmath57 ; we will often write @xmath62 to indicate a correspondence . a
_ group commensurability _ is a correspondence @xmath58 in the category of groups for which both @xmath63 and @xmath64 are isogenies , and the _ index _ @xmath65 of such an isogeny is defined to be @xmath66 . for a ring @xmath12
, one defines @xmath12-module commensurabilities and their indices analogously , replacing the category of groups by the category of @xmath12-modules .
likewise , one defines ring commensurabilities and their indices . _ endomorphisms and automorphisms .
_ let @xmath12 be a ring , and let @xmath67 @xmath68 be a correspondence of @xmath12-modules .
we define the _ endomorphism ring _
@xmath69 of @xmath70 to be the subring @xmath71 , @xmath72 of the product ring @xmath73 .
there are natural ring homomorphisms @xmath74 and @xmath75 sending @xmath76 to @xmath77 and @xmath78 , respectively ; we shall write @xmath79 for the ring correspondence consisting of @xmath69 and those two ring homomorphisms . similarly , writing @xmath80 for the multiplicative group of invertible elements of a ring @xmath81 , we define the _ automorphism group _ @xmath82 of @xmath70 to be the group @xmath83 , and we write @xmath84 for the group correspondence consisting of @xmath82 and the natural maps @xmath85 , @xmath86 .
a _ domain _ is a commutative ring without zero divisors and with @xmath87 .
a ring is _ semisimple _ if all short exact sequences of modules over the ring split .
we can now formulate the general result that we announced .
[ thm : mainintro ] let @xmath88 be an infinite domain such that for all non - zero @xmath89 the ring @xmath90 is finite , let @xmath91 be the field of fractions of @xmath88 , let @xmath92 be a semisimple @xmath91-algebra of finite vector space dimension over @xmath91 , let @xmath93 be a sub-@xmath88-algebra with @xmath94 , and let @xmath0 , @xmath1 be finitely generated @xmath12-modules .
then : 1 . there is an @xmath12-module commensurability @xmath68 if and only if the @xmath92-modules @xmath95 and @xmath96 are isomorphic ; 2 .
if @xmath97 is an @xmath12-module commensurability , then @xmath79 is a ring commensurability , and @xmath98 is a group commensurability ; 3 .
if @xmath70 , @xmath99 are @xmath12-module commensurabilities , then one has @xmath100 part ( c ) is the essential statement of theorem [ thm : mainintro ] .
it shows that one can define @xmath101 , independently of @xmath70 , if one has @xmath102 and @xmath103 is an @xmath12-module commensurability .
one deduces the existence part of theorem [ thm : qgintro ] from theorem [ thm : mainintro ] by putting @xmath104 , @xmath105 , @xmath106 $ ] , and @xmath107 $ ] .
we prove theorem [ thm : mainintro](a ) and ( b ) as theorems [ thm : parta ] and [ thm : partb ] , respectively , but with the semisimplicity assumption on @xmath92 replaced by the condition that @xmath12 be left - noetherian ; the following result shows that this is a weaker condition .
[ thm : noethintro ] let @xmath12 be a ring as in the statement of theorem [ thm : mainintro ]
. then @xmath12 is left - noetherian and right - noetherian .
for a proof , see theorem [ thm : q ] .
the point of theorem [ thm : noethintro ] is that @xmath12 is not required to be finitely generated as a @xmath88-module .
as an aside , we will characterise , in theorem [ thm : r ] , the rings @xmath88 satisfying the hypotheses of theorem [ thm : mainintro ] . the full statement of theorem [ thm : mainintro ] will be proven in section [ sec : proofs ] .
there we also give an example ( example [ ex : notss ] ) showing that theorem [ thm : mainintro](c ) may fail if @xmath12 is left - noetherian , but @xmath92 is not semisimple .
isogenies , commensurabilities , and their indices have many formal properties , and it is to these that section 2 is devoted . among other things , we shall , under certain conditions , define the _ composition _ @xmath108 of two correspondences @xmath109 and @xmath70 .
composition plays an important role in the proof of theorem [ thm : mainintro ] .
in fact , the index of a commensurability is multiplicative in compositions of commensurabilities . more importantly , and less formally , if in the situation of theorem [ thm : mainintro ] one has commensurabilities @xmath97 and @xmath110 of finitely generated @xmath12-modules , then one has @xmath111 ( see theorem [ thm : compos ] ) .
this property is the first key to the proof of theorem [ thm : mainintro](c ) .
the following elegant result , which will be proven as theorem [ thm : o ] , will be useful in passing from endomorphism rings to automorphism groups .
[ thm : unitsintro ] let @xmath112 be a ring isogeny . then the induced group homomorphism @xmath113 is a group isogeny . if in addition the map @xmath112 is surjective ,
then so is the induced map @xmath113 . the second key result for the proof of theorem [ thm : mainintro](c ) reads as follows .
[ thm : finexpintro ] let @xmath114 be a semisimple ring that is finitely generated as a module over its centre @xmath115 . then @xmath116)$ ] is an abelian group of finite exponent .
the proof is given in section [ sec : unitssemisimple ] ( theorem [ thm : x ] ) .
in fact , we prove an explicit version of theorem [ thm : finexpintro ] .
a _ central simple algebra _ over a field @xmath117 is a ring @xmath114 that is simple in the sense that it has precisely two two - sided ideals ; that has centre equal to @xmath117 ; and that has finite dimension as a vector space over @xmath117 ; it is a well - known result ( * ? ? ?
* ( 7.22 ) ) that , under these conditions , that dimension is a square .
[ thm : expintro ]
let @xmath117 be a field , and let @xmath114 be a central simple algebra over @xmath117 .
let the dimension of @xmath114 as a vector space over @xmath117 be @xmath118 , where @xmath109 is a positive integer .
then the group @xmath119)$ ] is abelian of exponent dividing @xmath109 .
our proof of theorem [ thm : expintro ] ( see theorem [ thm : w ] ) makes use of wedderburn s factorisation theorem for polynomials over division rings .
theorem [ thm : finexpintro ] is an immediate consequence of theorem [ thm : expintro ] . in section [ sec : skewcorres ] , we place ourselves in the situation of theorem [ thm : mainintro ] , but replacing the semisimplicity assumption on @xmath92 by the condition that @xmath12 be left - noetherian .
we show that the category whose objects are finitely generated @xmath12-modules , and whose morphisms are certain equivalence classes of @xmath12-module correspondences , is equivalent to the category of finitely generated @xmath92-modules ; theorem [ thm : equivcat ] gives a precise statement .
we deduce that the set of equivalence classes of commensurabilities @xmath120 forms a group that may be identified with @xmath121 ( corollary [ cor : g_l ] ) . in section [ sec : autcomm ] we show that @xmath122 and @xmath123 give rise to homomorphisms from this group to @xmath9 .
if @xmath92 is semisimple , then theorem [ thm : finexpintro ] will imply that these homomorphisms are trivial , and theorem [ thm : mainintro](c ) will easily follow .
this section is devoted to formal properties of isogenies and commensurabilities , and of their indices .
we begin by recalling a basic notion from category theory , for which we refer to ( * ? ? ?
i , 11 ) .
let @xmath57 be a category , and let @xmath124 be a diagram in @xmath57 .
we say that @xmath125 is a _ fibre product _ of @xmath92 and @xmath126 over @xmath114 if @xmath127 is a diagram in @xmath57 with the property that @xmath128 , and with the universal property that for any diagram @xmath129 that satisfies @xmath130 , there exists a unique morphism @xmath131 such that @xmath132 and @xmath133 . when a fibre product exists , it is unique up to a unique isomorphism , so in that case we may speak of _ the _ fibre product of @xmath92 and @xmath126 over @xmath114 . in the category @xmath134 of groups , the fibre product of @xmath124 exists , and it is given by @xmath135 with @xmath136 and @xmath137 being the projection maps to @xmath92 and @xmath126 , respectively . throughout this section @xmath57
will denote a category in which for every diagram @xmath124 the fibre product of @xmath92 and @xmath126 over @xmath114 exists , equipped with a functor @xmath138 that preserves fibre products .
the main examples we have in mind are the category of groups with the identity functor , the category of rings with the functor that sends a ring to its underlying additive group , and the category of finitely generated left @xmath12-modules for a left - noetherian ring @xmath12 , with the functor that sends an @xmath12-module to its underlying abelian group .
an _ isogeny _ in @xmath57 is a morphism that becomes an isogeny in @xmath134 .
a _ commensurability _ in @xmath57 is a correspondence in @xmath57 that becomes a commensurability in @xmath134 .
we will often think of an isogeny @xmath139 as a special case of a commensurability , which we will denote by @xmath140 , namely @xmath141 .
the _ index _ @xmath53 of an isogeny @xmath63 in @xmath57 is defined to be the index of the image of @xmath63 in @xmath134 , and the index of a commensurability is defined analogously , as in the introduction . for each of the results [ prop : isog ] [ prop : composition ] below
, it will be clear that it holds for @xmath57 if it holds for @xmath134 .
we will therefore tacitly assume that @xmath142 in the proofs of those results .
[ prop : isog ] let @xmath92 , @xmath114 , @xmath126 be objects in @xmath57 and let @xmath143 be the composition of two morphisms @xmath144 . if two of @xmath63 , @xmath64 , @xmath143 are isogenies , then so is the third .
moreover , we then have @xmath145 .
we have an exact sequence of pointed sets @xmath146 in which each map has the property that all its non - empty fibres have equal cardinality .
hence , any term that sits between two finite sets in the above sequence is itself finite
. the first assertion of the proposition easily follows .
moreover , if all terms in the sequence are finite , then the alternating product of their cardinalities is 1 , which proves the second assertion .
[ def : compos ] let @xmath147 and @xmath148 be correspondences in @xmath57 .
we define the _ composition _ of @xmath70 with @xmath109 by @xmath149 where @xmath136 , @xmath137 are the canonical morphisms from @xmath150 to @xmath55 , respectively @xmath56 .
[ rmrk : assoc ] it follows from the universal property of fibre products , and a routine diagram chase , that composition of correspondences is associative up to canonical isomorphism .
[ prop : fibre ] let @xmath151 be morphisms in @xmath57 , and suppose that @xmath143 is an isogeny .
let @xmath152 be the fibre product of @xmath55 and @xmath56 over @xmath114 .
then : 1 .
the morphism @xmath136 is an isogeny ; 2 .
if the image of @xmath64 in @xmath134 has finite kernel , then so does the image of @xmath137 ; 3 .
if @xmath64 is an isogeny , then so is @xmath137 .
we first prove part ( a ) .
assume that @xmath143 is an isogeny .
we have @xmath153 which is finite by assumption .
further , the kernel of @xmath154 is equal to @xmath155 , so @xmath156 , which is also finite .
so @xmath136 is an isogeny .
similarly , @xmath157 , which proves part ( b ) .
finally , part ( c ) is symmetric in @xmath55 and @xmath56 , and so follows from part ( a ) .
[ def : skew ] a _ skew correspondence _
is a correspondence @xmath158 in which @xmath63 is an isogeny .
[ prop : composition ] if @xmath159 and @xmath160 are skew correspondences , respectively commensurabilities , then @xmath161 is a skew correspondence , respectively a commensurability . moreover , if @xmath70 and @xmath109 are commensurabilities , then we have @xmath162 the first two assertions follow immediately from propositions [ prop : fibre ] and [ prop : isog ] .
the third one follows from proposition [ prop : isog ] and a routine diagram chase , which we leave to the reader .
[ def : equiv ] let @xmath147 and @xmath163 be two correspondences .
we say that @xmath70 and @xmath109 are _ equivalent _ if there exists a commensurability @xmath164 such that @xmath165 and @xmath166 .
we will call such a commensurability an _ equivalence _ between @xmath70 and @xmath109 . [
prop : equivrel ] being equivalent in the sense of definition [ def : equiv ] is an equivalence relation .
the relation is clearly symmetric .
reflexivity is also clear , since an equivalence between @xmath167 and itself is given by @xmath168 .
transitivity follows from proposition [ prop : composition ] .
[ def : inverse ] the _ inverse _ of a correspondence @xmath147 is defined to be @xmath169 .
[ lem : equivinverse ] let @xmath70 , @xmath170 and @xmath160 be correspondences .
then : 1 .
the correspondence @xmath171 is equivalent to the composition @xmath172 ; 2 . if @xmath70 is equivalent to @xmath173 , then @xmath174 is equivalent to @xmath175 .
the proof is easy , and is left to the reader .
[ prop : equivcompos ] let @xmath70 , @xmath170 and @xmath109 , @xmath176 be correspondences .
suppose that @xmath70 is equivalent to @xmath173 , and @xmath109 is equivalent to @xmath177 .
then @xmath108 is equivalent to @xmath178 .
let @xmath158 , @xmath179 .
first , we prove the proposition in the special case that @xmath180 , and @xmath181 , where @xmath182 is an isogeny .
let @xmath183 be the fibre product of the diagram @xmath151 , and let @xmath184 be the fibre product of the diagram @xmath185
. thus @xmath186 , and @xmath187 .
since @xmath188 , the universal property of fibre products guarantees the existence of a unique map @xmath189 with the property that @xmath190 and @xmath191 : @xmath192_{p_0'}\ar[d]^{i}\ar[ddr]^{p_1'}&\\ & w\ar[d]^p & x\times_b y \ar[dl]^{p_0}\ar[dr]_{p_1 } & & \\ &
x\ar[dl]^f\ar[dr]_g & & y\ar[dl]^h\ar[dr]_j & \\ \;\;\;\;a\;\ ; & & b & & \;\;c.\;\;\;\ ; } \ ] ] moreover , it is easy to see that @xmath193 is the fibre product of the diagram @xmath194 .
it follows from proposition [ prop : fibre ] that @xmath195 is an isogeny , which proves that @xmath108 is equivalent to @xmath196 .
now , we prove the proposition in the special case that @xmath197 , and @xmath173 is arbitrary .
write @xmath198 for the equivalence relation between correspondences .
let @xmath199 , and let @xmath200 be an equivalence between @xmath70 and @xmath173 . since @xmath201 is an isogeny , we have @xmath202 , and since @xmath203 is an isogeny , we have @xmath204 .
we deduce from the special case of the proposition that we just proved that @xmath205 .
now , we prove the general case . by lemma [ lem : equivinverse ] and by the special case we just proved , we have @xmath206 .
it therefore follows from lemma [ lem : equivinverse](b ) , that @xmath207 . by the special case of the proposition that we have already proven , we also have @xmath208 , and the proposition follows .
[ prop : equiv ] if @xmath70 and @xmath109 are two equivalent commensurabilities , then @xmath209 .
this is an immediate consequence of proposition [ prop : isog ] .
the term `` inverse '' is justified by the following result .
[ prop : inverse ] given a commensurability @xmath147 , the composition @xmath210 is equivalent to the commensurability @xmath211 , and the composition @xmath212 is equivalent to the commensurability @xmath213 . first , we prove the assertion on @xmath214 . by definition , @xmath215 , where @xmath216 is the fibre product of the diagram @xmath217 . by the universal property of the fibre product
, we have a unique map @xmath218 with the property that @xmath219 .
@xmath220_(0.3)i\ar@/^5ex/[dd]^(0.3)f & \\ & x\times_b x\ar[dl]_(.6){p_0}\ar[dr]^(.6){p_1 } & \\ x \ar[d]_{f}\ar[dr]^(.3){g } & a\ar[dl]_(.3){\id}\ar[dr]^(.3){\id } & x\ar[d]^f\ar[dl]_(.3){g}\\ a & b & a } \ ] ] since @xmath64 is an isogeny , it follows from proposition [ prop : fibre ] that @xmath136 is an isogeny . by proposition [ prop : isog ] , the morphism @xmath195 is also an isogeny , so @xmath221 defines an equivalence between @xmath214 and @xmath211 .
the claim for @xmath222 follows by applying the above argument to @xmath223 in place of @xmath70 .
[ def : skewcom ] we define @xmath224 to be the category with the same objects as in @xmath57 , and where , for objects @xmath225 , the morphisms from @xmath92 to @xmath114 are the equivalence classes of skew correspondences @xmath226
. we also define @xmath227 to be the category with the same objects , and where the morphisms from @xmath92 to @xmath114 are the equivalence classes of commensurabilities @xmath226 .
it follows from remark [ rmrk : assoc ] and propositions [ prop : equivrel ] and [ prop : equivcompos ] , that these are indeed categories , i.e. that composition of morphisms is well - defined and associative .
proposition [ prop : inverse ] implies that @xmath227 is a ( generally large ) groupoid , i.e. every morphism in @xmath227 is an isomorphism .
in fact , we have the following sharper result .
[ prop : maxgroupoid ] the category @xmath227 is the maximal subgroupoid of @xmath224 . let @xmath147 be a skew correspondence , and let @xmath228 be a two - sided inverse in @xmath224 .
so @xmath108 is equivalent to the commensurability @xmath229 , and @xmath230 is equivalent to @xmath231 , and in particular both compositions are commensurabilities .
we wish to prove that @xmath64 is then necessarily an isogeny , and for this it is enough to assume that @xmath142 .
let @xmath232 be the fibre product of the diagram @xmath233 , and let @xmath234 be the fibre product of the diagram @xmath235 . since @xmath230 is a commensurability , the morphism @xmath236 is an isogeny , so @xmath237 is finite .
also , since @xmath108 is a commensurability , the morphism @xmath238 is an isogeny , so @xmath239 is finite .
but @xmath240 . so @xmath64 is an isogeny , as claimed .
in the present section we prove that an isogeny of rings induces an isogeny of multiplicative groups .
we begin by recalling some standard ring theoretic facts , which can be found in @xcite .
the _ jacobson radical _ of a ring @xmath81 , denoted by @xmath241 , is the intersection of the maximal left ideals of @xmath12 .
let @xmath81 be a ring , and @xmath242
. then the following are equivalent : 1 .
@xmath243 ; 2 .
@xmath244 is contained in every maximal right ideal of @xmath245 ; 3 .
@xmath244 annihilates every simple left @xmath81-module ; 4 .
@xmath244 annihilates every simple right @xmath81-module ; 5 .
@xmath246 for all @xmath247 , @xmath248 .
see @xcite .
[ lem : j ] let @xmath249 be a two - sided ideal of @xmath81 with @xmath250
. then the map @xmath251 is surjective .
moreover , @xmath252 is a unit if and only if @xmath253 is a unit in @xmath254 .
let @xmath253 be a unit in @xmath254 , and let @xmath255 be its inverse .
then we have @xmath256 , @xmath257 , so @xmath258 has both a right inverse , namely @xmath259 , and a left inverse , namely @xmath260 .
it follows that @xmath258 is a unit in @xmath81 .
a ring is called _ simple _ if it has exactly two two - sided ideals .
a ring @xmath81 is called _ semisimple _ if all short exact sequences of left @xmath81-modules split .
any semisimple ring is a finite direct product of simple rings . if @xmath81 is a semisimple ring , then the opposite ring @xmath261 is also semisimple .
a left - artinian ring is semisimple if and only if its jacobson radical is 0 . if @xmath81 is an arbitrary ring , then @xmath262 . in particular , if @xmath81 is left - artinian , then @xmath263 is semisimple .
all these facts can be found in ( * ? ? ? * and 4 ) .
[ lem : k ] let @xmath81 and @xmath264 be rings , let @xmath112 be a surjective ring homomorphism , and suppose that @xmath81 is left - artinian .
then the induced group homomorphism @xmath113 is surjective .
first , suppose that @xmath81 is semisimple .
then @xmath81 can be written as the product of finitely many simple rings .
since the kernel of @xmath112 is a two - sided ideal of @xmath81 , it must be a subproduct , and @xmath264 may then be identified with the complementary subproduct .
surjectivity of @xmath113 is now obvious .
we pass to the general case .
denote by @xmath249 the image of @xmath241 in @xmath264 , which is a two - sided ideal of @xmath264 . the map @xmath112 induces a surjective ring homomorphism @xmath265 , where the ring @xmath263 is semisimple , so by the first part of the proof the induced map @xmath266 is surjective . by lemma [ lem :
j ] , the map @xmath267 is also surjective , so the map @xmath268 induced by the composed ring homomorphism @xmath269 is surjective as well .
now let @xmath270 , and choose @xmath271 that maps to @xmath272 .
then the image @xmath273 of @xmath258 in @xmath264 satisfies @xmath274 , so @xmath275 belongs to the image @xmath276 of @xmath277 in @xmath264 .
let @xmath278 map to @xmath275 .
then @xmath279 maps to @xmath280 , and we have @xmath281 because @xmath271 and @xmath282 .
this proves surjectivity of @xmath113 , as required . for a different proof of lemma [ lem : k ] , see @xcite .
[ lem : l ] let @xmath81 be a ring , and let @xmath249 , @xmath283 be two - sided ideals .
then the kernel of the natural ring homomorphism @xmath284 equals @xmath285 , and its image is the fibre product @xmath286 .
the proof is straightforward , and is left to the reader .
[ lem : m ] let @xmath112 be a surjective ring isogeny
. then the induced group homomorphism @xmath113 is surjective .
let @xmath249 be the kernel of the isogeny @xmath112 .
then @xmath249 is finite , and we may identify @xmath264 with @xmath254 .
write @xmath287 for the endomorphism ring of the additive group of @xmath249 .
let @xmath245 and @xmath12 , respectively , be the kernel and the image of the ring homomorphism @xmath288 sending @xmath289 to the map @xmath290
. then @xmath12 , being a subring of @xmath287 , is a finite ring , @xmath245 is a two - sided ideal of @xmath81 , and we have a ring isomorphism @xmath291 . by lemma [ lem : l ] , the combined map @xmath292 induces a ring isomorphism @xmath293 .
now we first prove that the map @xmath294 is surjective .
let @xmath295 .
write @xmath296 for the image of @xmath258 in @xmath297 . since @xmath12 is finite and hence left - artinian , by lemma [
lem : k ] we can choose @xmath298 mapping to @xmath299
. then @xmath300 belongs to @xmath301 , so @xmath302 is a unit of @xmath303 that maps to @xmath295 .
this proves that @xmath294 is surjective . from @xmath304
it follows that for each @xmath305 the element @xmath306 has inverse @xmath307 and therefore belongs to @xmath80 ; this implies @xmath308 , so that by lemma [ lem : j ] the map @xmath309 is surjective .
the composed map @xmath113 is then surjective as well .
[ lem : n ] let @xmath81 be a subring of a ring @xmath264 such that the index @xmath310 of additive groups is finite .
then : 1 .
the ring @xmath264 has a two - sided ideal @xmath249 with @xmath311 for which the ring @xmath312 is finite ; 2 .
the index @xmath313 is finite .
\(a ) put @xmath314 .
then @xmath249 is a two - sided ideal of @xmath264 that is contained in @xmath81 , and we proceed to show that @xmath249 has finite index in @xmath264 . put @xmath315 and @xmath316 .
then we have @xmath317 .
denote by @xmath318 the finite abelian group @xmath319 , by @xmath320 its endomorphism ring , and by @xmath321 the ring opposite to @xmath320 .
both of these rings are finite .
the natural left and right @xmath81-module structures on @xmath318 induce a ring homomorphism @xmath322 of which @xmath245 is the kernel .
it follows that @xmath245 is a two - sided ideal of @xmath81 of finite index in @xmath81 .
there is a well - defined group homomorphism j & & ( d_d , d ) + x & & ( ( y+e)(z+e)yxz+e ) for @xmath323 , @xmath324 .
its kernel is @xmath249 , and since @xmath325 is finite , the group @xmath326 is finite . because each of @xmath319 , @xmath327 , @xmath326 is finite
, the ring @xmath312 is finite .
this proves ( a ) .
\(b ) let @xmath249 be as in ( a ) .
then @xmath312 and @xmath328 are finite , so the kernel @xmath329 ( say ) of the natural group homomorphism @xmath330 has finite index in @xmath331 . if @xmath332 , then @xmath333 , so both @xmath247 and @xmath334 are in @xmath276 , which is contained in @xmath81 .
this proves @xmath335 , so @xmath80 has finite index in @xmath331 as well .
this proves ( b ) .
lemma [ lem : n ] also appears in @xcite
. we can now prove theorem [ thm : unitsintro ] of the introduction , which reads as follows .
[ thm : o ] let @xmath112 be a ring isogeny . then the induced group homomorphism @xmath113 is a group isogeny .
if in addition the map @xmath112 is surjective , then so is the induced map @xmath113 .
the last assertion is lemma [ lem : m ] .
for the first assertion , let @xmath249 and @xmath318 be the kernel , respectively the image of the map @xmath112 .
then the kernel of @xmath336 is contained in @xmath276 and therefore finite , and by lemma [ lem : m ] the image is all of @xmath337 .
hence @xmath336 is a group isogeny .
further , the inclusion map @xmath338 is obviously injective , while by lemma [ lem : n](b ) the index of @xmath337 in @xmath331 is finite .
hence @xmath338 is a group isogeny . by proposition [ prop : isog ] ,
the composed map @xmath113 is also a group isogeny .
this section is devoted to some properties of infinite domains all of whose proper quotients are finite . [
lem : p ] let @xmath88 be a domain such that for all non - zero @xmath89 the ring @xmath90 is finite , let @xmath91 be the field of fractions of @xmath88 , let @xmath15 be a finite dimensional @xmath91-vector space , and let @xmath0 be a sub-@xmath88-module of @xmath15 .
then for all non - zero @xmath89 the @xmath88-module @xmath339 is finite of order dividing @xmath340 , with equality if @xmath0 is finitely generated and @xmath341 .
let @xmath89 be non - zero .
first suppose that @xmath0 is finitely generated .
let @xmath342 be a finite subset that generates it as a @xmath88-module , let @xmath343 be a maximal subset that is linearly independent over @xmath91 , and let @xmath344 be the @xmath88-module generated by @xmath345 .
then @xmath1 is @xmath88-free of rank @xmath346 , so @xmath347 is finite of order @xmath348 , with equality if @xmath345 is a @xmath91-basis of @xmath15 or , equivalently , if @xmath341 . by maximality of @xmath345 , we can , for each @xmath349 , choose a non - zero element @xmath350 such that @xmath351 , and @xmath352 is then a non - zero element of @xmath88 satisfying @xmath353 . because @xmath354 is finite , its subgroup
@xmath355 is finite as well , and since the latter group is isomorphic to @xmath356 and to @xmath357 , we find that @xmath356 and @xmath357 are finite of the same order .
the group @xmath358 is finite of order @xmath359 , so @xmath339 is also finite , of order @xmath360 with equality if @xmath341 . passing to the general case ,
let @xmath361 be the set of finitely generated sub-@xmath88-modules @xmath22 of @xmath0 , which is a directed partially ordered set by inclusion .
then @xmath0 is the injective limit of all @xmath362 , and @xmath339 is the injective limit of the modules @xmath363 , all of which have order dividing @xmath340 .
the injective limit has then also order dividing the same number .
this completes the proof of lemma [ lem : p ] . we now prove theorem [ thm : noethintro ] .
[ thm : q ] let @xmath88 be a domain such that for all non - zero @xmath89 the ring @xmath90 is finite , let @xmath91 be the field of fractions of @xmath88 , let @xmath92 be a semisimple @xmath91-algebra of finite vector space dimension over @xmath91 , and let @xmath93 be a sub-@xmath88-algebra with @xmath364
. then @xmath12 is left - noetherian and right - noetherian .
let @xmath249 be a left ideal of @xmath12 .
then @xmath365 is a left @xmath92-ideal , so by semisimplicity of @xmath92 it is a direct summand of the left @xmath92-module @xmath92 .
thus the endomorphism ring of the latter module contains an idempotent with image @xmath365 .
since the endomorphisms of the left @xmath92-module @xmath92 are the right multiplications by elements of @xmath92 , it is equivalent to say that we can choose an idempotent @xmath366 with @xmath367 .
we have @xmath368 , so we can choose a non - zero element @xmath89 with @xmath369 . multiplying the chain of inclusions @xmath370 by @xmath371 on the right , which when restricted to @xmath249 is just the identity map , we obtain @xmath372 , where @xmath373 because @xmath374 is central . by lemma [ lem : p ] , the group @xmath375 is finite , so @xmath376 is finite as well .
hence @xmath249 is , as a left @xmath12-module , generated by @xmath377 together with a finite set , and is therefore finitely generated .
this proves that @xmath12 is left - noetherian . applying this result to @xmath378 and @xmath379 , we find that @xmath12 is right - noetherian as well .
[ ex : notnoeth ] if we assume @xmath380 , then the semisimplicity condition on @xmath92 is actually necessary for the conclusion of theorem [ thm : q ] to be valid for all @xmath12 . to see this ,
assume that @xmath92 is not semisimple , or equivalently that @xmath381 , and choose a sub-@xmath88-algebra @xmath382 that is finitely generated as a @xmath88-module and satisfies @xmath383 .
then the ring @xmath384 is not left - noetherian because @xmath385 is not finitely generated as a left @xmath12-ideal .
if it were , then the non - zero @xmath91-vector space @xmath386 would be finitely generated as a @xmath345-module and hence as a @xmath88-module , which for @xmath380 is impossible .
[ lem : finquodomain ] let @xmath88 be an infinite commutative ring .
suppose that there exists a faithful @xmath88-module @xmath1 with the property that for all non - zero @xmath89 , @xmath347 is finite .
then @xmath88 is a domain .
let @xmath387 be non - zero .
we have an exact sequence of @xmath88-modules @xmath388 the left and right terms are finite by assumption , so @xmath389 is finite .
but since @xmath88 is infinite , and @xmath1 is a faithful module , @xmath1 is also infinite , and so @xmath390 .
the following result gives a description of the rings @xmath88 that occur in theorem [ thm : mainintro ] .
[ thm : r ] let @xmath391 be an infinite commutative ring .
then the following assertions are equivalent : 1 . for each non - zero @xmath392 ,
the ring @xmath393 is finite ; 2 .
the ring @xmath391 is a domain , and each non - zero prime ideal @xmath394 of @xmath391 is finitely generated as an ideal and has finite index in @xmath391 ; 3 .
either @xmath391 is a field , or it is a one - dimensional noetherian domain with the property that for every maximal ideal @xmath395 of @xmath391 the field @xmath396 is finite . first we prove that ( a ) implies ( b ) . from ( a )
it follows , by lemma [ lem : finquodomain ] applied to @xmath397 , that @xmath88 is a domain .
now let @xmath394 be a non - zero prime ideal , and let @xmath398 be non - zero .
then we have @xmath399 , and since @xmath90 is finite , the index of @xmath394 in @xmath88 is finite and @xmath400 is finite . hence @xmath394 is generated by @xmath374 together with a finite set , and is therefore finitely generated .
now we prove that ( b ) implies ( c ) . by (
* theorem 2 ) , each commutative ring of which every prime ideal is finitely generated is noetherian .
hence ( b ) implies that @xmath88 is noetherian .
if @xmath394 is a non - zero prime ideal , then @xmath401 is a finite domain , and therefore a field .
hence each non - zero prime ideal is maximal , so @xmath88 has krull dimension @xmath402 or @xmath403 ; in the former case it must be a field .
finally , suppose that ( c ) holds .
then , we will deduce ( a ) by showing that for any non - zero ideal @xmath249 of @xmath88 , the ring @xmath404 is finite .
suppose that there exists a non - zero ideal @xmath249 in @xmath88 such that @xmath404 is infinite .
since @xmath88 is noetherian , we may , without loss of generality , assume that @xmath249 is maximal among ideals with this property .
so @xmath404 is infinite , but its quotient by any non - zero ideal is finite .
it follows from lemma [ lem : finquodomain ] , applied to @xmath405 , that @xmath404 is a domain , so @xmath249 is a prime ideal of @xmath88 .
it is also non - zero , so ( c ) implies that @xmath249 is maximal , and therefore that @xmath404 is finite , which is a contradiction .
by a _ division ring _ we mean a ring @xmath318 with the property @xmath406 . if @xmath318 is a division ring and @xmath407 is a positive integer , then @xmath408 denotes the ring of @xmath407 by @xmath407 matrices over @xmath318 .
if @xmath11 is a group , then @xmath409 denotes the maximal abelian quotient of @xmath11 .
[ lem : s ] let @xmath407 be a positive integer , let @xmath318 be a division ring , and for @xmath410 and @xmath411 , let @xmath412 be the diagonal matrix with @xmath413th entry equal to @xmath247 and all other entries equal to @xmath403 .
then each map @xmath414 is a group homomorphism @xmath415 , they all induce the same group homomorphism @xmath416 , and if @xmath417 or @xmath418 then @xmath419 is surjective .
it is clear that each @xmath414 is a group homomorphism , and that for each @xmath247 all @xmath420 are conjugate to each other , so that all @xmath414 induce the same map @xmath421 . for @xmath195 , @xmath411 , @xmath422 , and @xmath423 , let @xmath424 be the matrix obtained from the unit matrix by replacing the @xmath425-entry by @xmath247
; then one has @xmath426 .
the subgroup of @xmath427 generated by all @xmath428 is denoted by @xmath429 . by (
* chapter iv , theorem 4.1 ) , we have @xmath430 .
each @xmath428 is a transvection of the right @xmath318-vector space @xmath431 in the sense of ( * ? ? ?
* chapter iv , definition 4.1 ) , and if @xmath432 or @xmath418 , then by ( * ? ? ? * chapter iv , section 2 ) each transvection belongs to @xmath433 $ ] , so that @xmath434 $ ] , and therefore @xmath435{\cdot}\delta_n(d^\times).\ ] ] this implies that @xmath419 is surjective .
it is clearly also surjective if @xmath436 .
[ lem : t ] let @xmath407 be a positive integer , let @xmath318 be a division ring , and for each @xmath410 let @xmath437 be @xmath247 times the identity matrix .
then @xmath438 is a group homomorphism @xmath415 , and the group @xmath439)\ ] ] is abelian of exponent dividing @xmath407 .
it is clear that @xmath438 is a group homomorphism .
if we have @xmath440 , then @xmath427 is a non - abelian group of order @xmath441 , so that @xmath442 has order @xmath443 and the conclusion of the lemma is valid .
assume now that @xmath417 or @xmath418 , so that the map @xmath419 from lemma [ lem : s ] is surjective .
denote by @xmath444 the map induced by @xmath438 .
for each @xmath410 one has @xmath445 and therefore @xmath446 , so the surjectivity of @xmath419 yields @xmath447 and the lemma is proved .
the following result is due to wedderburn @xcite .
[ thm : u ] let @xmath318 be a division ring with centre @xmath448 , let @xmath449 , and let @xmath450 $ ] be an irreducible polynomial with leading coefficient @xmath403 such that @xmath451 . put @xmath452 .
then there exist @xmath453 , @xmath454 ,
, @xmath455 such that in @xmath456 $ ] one has @xmath457 see ( * ? ? ?
* theorem ( 16.9 ) ) .
[ lem : v ] let @xmath318 be a division ring that has finite vector space dimension @xmath458 over its centre @xmath448 , where @xmath374 is a positive integer
. then the group @xmath459)$ ] is abelian of exponent dividing @xmath374 . since @xmath459)$ ] is a quotient of @xmath460 , it is an abelian group .
. it will suffice to show that the image @xmath462 of @xmath463 in the quotient @xmath459)$ ] has order dividing @xmath374 .
the subfield @xmath464 of @xmath318 is contained in a maximal subfield of @xmath318 , and each maximal subfield of @xmath318 is an extension field of @xmath448 of degree @xmath374 , by ( * ? ? ?
* ( 7.22 ) ) .
hence we have @xmath465=l$ ] for some divisor @xmath466 of @xmath374 , and @xmath463 is a zero of an irreducible polynomial @xmath450 $ ] of degree @xmath466 with leading coefficient @xmath403 . using theorem [ thm : u ] we find @xmath453 ,
, @xmath455 such that @xmath467 .
mapping this identity to the abelian group @xmath459)$ ] we obtain @xmath468 , so @xmath469 , as required .
this proves lemma [ lem : v ]
. we can now prove theorem [ thm : expintro ] and deduce theorem [ thm : finexpintro ] .
we recall the statements .
[ thm : w ] let @xmath117 be a field , and let @xmath114 be a central simple algebra over @xmath117 .
let the dimension of @xmath114 as a vector space over @xmath117 be @xmath118 , where @xmath109 is a positive integer .
then the group @xmath119)$ ] is abelian of exponent dividing @xmath109 . by ( * ? ? ?
* , theorem 1 ) , there are a positive integer @xmath407 and a division ring @xmath318 with @xmath470 such that @xmath114 is , as an algebra over @xmath117 , isomorphic to @xmath408 .
then @xmath318 has finite degree @xmath458 over @xmath117 , and @xmath471 . by lemma [ lem :
v ] , the cokernel of the natural group homomorphism @xmath472 has exponent dividing @xmath374 , and by lemma [ lem : t ] the cokernel of the natural group homomorphism @xmath473 has exponent dividing @xmath407 .
it follows that the cokernel of the natural group homomorphism @xmath474 has exponent dividing @xmath471 .
[ thm : x ] let @xmath114 be a semisimple ring that is finitely generated as a module over its centre @xmath115 .
then @xmath116)$ ] is an abelian group of finite exponent . in the case
the semisimple ring @xmath114 is simple , our hypothesis that it be finite over its centre implies that it is a central simple algebra over @xmath115 , and the assertion follows from theorem [ thm : w ] . generally , by ( * ? ? ?
* chapter 1 , theorem ( 3.5 ) ) the ring @xmath114 is a product of finitely many semisimple rings that are simple , and the result follows from the case we just did .
in this section we formulate and prove the equivalence between categories of certain modules announced at the end of the introduction . we will need the notion of a skew correspondence ( definition [ def : skew ] ) , and the constructions of the categories @xmath224 and @xmath227 ( definition [ def : skewcom ] ) .
[ not : z ] the following assumptions will be in force throughout the present section : @xmath88 is an infinite commutative ring that satisfies the equivalent conditions of theorem [ thm : r ] , with field of fractions @xmath91 ; further , @xmath92 is a @xmath91-algebra of finite vector space dimension over @xmath91 , and @xmath12 is a left - noetherian sub-@xmath88-algebra of @xmath92 with the property that @xmath475 .
by an @xmath12-module we shall always mean a left @xmath12-module .
if @xmath0 is a finitely generated @xmath12-module , let @xmath476 denote the set of all elements of @xmath0 that have a non - zero annihilator in @xmath88 . since the image of @xmath88 in @xmath12 is central
, @xmath476 is an sub-@xmath12-module .
[ lem : fintors ] let @xmath0 be a finitely generated @xmath12-module , and let @xmath361 be a sub-@xmath12-module .
then @xmath361 is finite if and only if it is contained in @xmath476 .
first , we show that @xmath476 is finite .
since @xmath12 is left - noetherian , @xmath476 is finitely generated as an @xmath12-module .
so there exists a non - zero @xmath392 that annihilates @xmath476 , and @xmath476 is then a finitely generated module over the ring @xmath477 , which is finite by lemma [ lem : p ] .
this proves one implication . for the converse ,
let @xmath478 be a finite sub-@xmath12-module .
then for each @xmath479 , the set @xmath480 is finite , so that the annihilator of @xmath247 in @xmath88 has finite index in @xmath88 ; in particular it is non - zero , since @xmath88 is assumed to be infinite , so @xmath481 .
[ thm : parta ] let @xmath0 , @xmath1 be two finitely generated @xmath12-modules .
then there exists a commensurability of @xmath12-modules @xmath68 if and only if there exists an isomorphism of @xmath92-modules @xmath482 . first
, suppose that we have an isomorphism @xmath483 of @xmath92-modules .
it follows from lemma [ lem : fintors ] that @xmath0 is commensurable with @xmath484 , and similarly for @xmath1 , so by proposition [ prop : composition ] we may assume without loss of generality that @xmath0 and @xmath1 are @xmath391-torsion free . they then embed into @xmath485 , respectively @xmath486 , and by `` clearing denominators '' , we can find @xmath487 such that @xmath488 is contained in @xmath1 , and @xmath489 contains @xmath490 .
since @xmath491 and @xmath492 are finite by lemma [ lem : p ] , it follows that the correspondence @xmath493 is a commensurability .
conversely , suppose that we have a commensurability @xmath494 .
then the kernels and cokernels of @xmath495 are finite @xmath12-modules , and so are @xmath88-torsion modules by lemma [ lem : fintors ] .
they are therefore annihilated by the functor @xmath496 , so the maps @xmath497 and @xmath498 are isomorphisms . [
lem : welldefined ] let @xmath0 , @xmath1 be finitely generated @xmath12-modules , and let @xmath167 and @xmath499 be equivalent skew correspondences .
let @xmath500 denote the map of @xmath92-modules @xmath501 induced by @xmath63 , and similarly for @xmath502 .
then @xmath503 .
let @xmath164 be an equivalence between @xmath167 and @xmath504 .
since @xmath201 and @xmath203 are isogenies , lemma [ lem : fintors ] implies that @xmath505 and @xmath506 are both invertible .
moreover , we have q_z f & = & ( q_z h)(q_z q)(q_z p)^-1 , + q_z g & = & ( q_z j)(q_z q)(q_z p)^-1 , + so @xmath507 . let @xmath508 , respectively @xmath509 denote the category of finitely generated @xmath12-modules , respectively finitely generated @xmath92-modules .
by lemma [ lem : welldefined ] , we may define a functor @xmath510 from @xmath511 to @xmath509 by sending an @xmath12-module @xmath0 to the @xmath92-module @xmath512 , and an equivalence class of skew correspondences represented by @xmath494 to the map of @xmath92-modules @xmath513 .
the verification that @xmath510 respects composition of morphisms , and thus does define a functor , is easy and is left to the reader .
the next three lemmas will show that the functor @xmath510 has dense image , is full , and is faithful .
[ lem : dense ] any element of @xmath509 is isomorphic to @xmath514 for some @xmath12-module @xmath0 .
let @xmath15 be an @xmath92-module with finite generating set @xmath114 .
let @xmath0 be the sub-@xmath12-module of @xmath15 generated by @xmath114 over @xmath12 .
then the @xmath92-module @xmath514 is isomorphic to @xmath15 .
[ lem : full ] let @xmath0 , @xmath1 be finitely generated @xmath12-modules , and let @xmath515 be a morphism of @xmath92-modules . then there exists a skew correspondence @xmath97 such that @xmath516 .
let @xmath517 be the image of @xmath0 in @xmath512 , and let @xmath518 be the image of @xmath1 in @xmath519 . by lemma [ lem : fintors ] ,
the natural map @xmath520 gives rise to a commensurability @xmath521 , and similarly we have a commensurability @xmath522 .
since @xmath517 and @xmath518 are finitely generated as @xmath12-modules , and since @xmath518 generates @xmath519 over @xmath91 , we may choose a non - zero @xmath89 such that @xmath523 is contained in @xmath518 .
let @xmath64 be the inclusion @xmath524 , and define the correspondence @xmath525 .
it follows from lemma [ lem : p ] that @xmath526 is a skew correspondence . by proposition [ prop :
composition ] , the composition @xmath527 is also a skew correspondence , and it is easy to see that @xmath516 . [
lem : faithful ] let @xmath0 , @xmath1 be finitely generated @xmath12-modules , and let @xmath70 , @xmath528 be two skew correspondences such that @xmath529 .
then @xmath70 and @xmath109 are equivalent .
let @xmath158 , and @xmath179 .
we will show that @xmath70 and @xmath109 are equivalent by showing that the fibre product @xmath530 is a commensurability .
first , assume that the images of @xmath63 , @xmath64 , @xmath143 , and @xmath413 are @xmath88-torsion free .
then @xmath63 and @xmath64 factor through @xmath531 , and similarly for @xmath143 and @xmath413 . by lemma [ lem : fintors ] ,
the quotient maps @xmath532 and @xmath533 are isogenies , so after replacing @xmath70 and @xmath109 by equivalent commensurabilities , we may assume that @xmath55 and @xmath56 are @xmath88-torsion free .
it then follows from lemma [ lem : fintors ] that @xmath63 , @xmath64 , @xmath143 , and @xmath413 are injective . since @xmath529 , we have @xmath534 and it follows that the canonical injection @xmath535 is an isomorphism . by proposition [ prop :
composition ] , the fibre product @xmath536 of the diagram @xmath537 is a commensurability , which proves this special case of the lemma . we now prove the general case . by applying lemma [ lem : fintors ] with @xmath538 , and similarly for @xmath64 , @xmath143 , and @xmath413 , we may choose a non - zero @xmath89 such that the images of @xmath539 , @xmath540 , @xmath541 , and @xmath542 are @xmath88-torsion free .
it is easy to see that @xmath70 is equivalent to @xmath543 , and @xmath109 is equivalent to @xmath544 .
so the general case follows from the special case above .
[ thm : equivcat ] the functor @xmath545 is an equivalence of categories .
the result follows by combining lemmas [ lem : dense ] , [ lem : full ] , and [ lem : faithful ] .
if @xmath0 is a finitely generated @xmath12-module , we let @xmath546 denote the group of equivalence classes of commensurabilities @xmath120 under composition . it may be viewed as a full sub - groupoid of @xmath547 whose only object is @xmath0 .
[ cor : g_l ] let @xmath0 be a finitely generated @xmath12-module .
then the map @xmath548 , @xmath549 is a group isomorphism . by proposition [ prop : maxgroupoid ] , the category @xmath547 is the maximal sub - groupoid of @xmath511 .
so theorem [ thm : equivcat ] implies that the functor @xmath510 induces an equivalence of categories from @xmath547 to the category whose objects are the finitely generated @xmath92-modules , and whose morphisms are the @xmath92-module isomorphisms .
the corollary follows by restricting @xmath510 to the full sub - groupoid @xmath546 of @xmath511 .
it is in the present section that we construct ring and group commensurabilities out of module commensurabilities .
here we retain the assumptions of notation [ not : z ] .
let @xmath550 @xmath68 be a correspondence of @xmath12-modules . in the introduction we defined the endomorphism ring of @xmath70 to be @xmath551 ,
we also recall the correspondence @xmath553 , given by sending @xmath554 to @xmath77 and @xmath78 , respectively , and the induced correspondence of automorphism groups @xmath555 . if @xmath556 is an isogeny , we let @xmath140 be the commensurability @xmath557 , as in section [ sec:2 ] . [
lem : endisog ] let @xmath556 be an isogeny of finitely generated @xmath12-modules . then the correspondence @xmath558 is a commensurability of rings . we first show that @xmath137 has finite kernel .
we have @xmath559 which is finite since @xmath0 is finitely generated and @xmath560 is finite by assumption .
next , we show that the image of @xmath137 has finite additive index in @xmath561 . the modules @xmath476 and @xmath562 are finite , so by lemma [ lem : fintors ] there exist non - zero @xmath563 such that @xmath564 annihilates @xmath476 , and @xmath565 annihilates @xmath562 .
thus , @xmath566 is injective , and the image contains @xmath567 , so @xmath568 defines a homomorphism @xmath569 .
given @xmath570 , we may therefore define @xmath571 , which has the property that @xmath572 .
so the image of @xmath137 contains @xmath573 , which has finite additive index in @xmath561 by lemma [ lem : p ] .
this proves that @xmath137 is an isogeny .
we now show that the image of @xmath136 has finite additive index in @xmath574 .
given any @xmath575 , we may define @xmath576 , @xmath577 , where @xmath578 are as before .
we then have @xmath579 .
so the image of @xmath136 contains @xmath580 , which has finite additive index in @xmath574 by lemma [ lem : p ] . finally , we show that @xmath136 has finite kernel .
we have @xmath581 where the last isomorphism follows from lemma [ lem : fintors ] and the assumption that @xmath562 is finite . invoking lemma [ lem : fintors ] again
, it follows that @xmath582 is finite , so @xmath136 is an isogeny .
[ thm : partb ] let @xmath0 , @xmath1 be finitely generated @xmath12-modules .
then for any commensurability @xmath583 , the correspondence @xmath584 is a ring commensurability , and the induced correspondence @xmath585 @xmath586 is a group commensurability .
the correspondence @xmath587 is canonically isomorphic to the composition of @xmath588 with @xmath589 .
the correspondences @xmath590 and @xmath591 are commensurabilities by lemma [ lem : endisog ] , so @xmath587 is a commensurability by proposition [ prop : composition ] .
the assertion on @xmath592 follows from theorem [ thm : o ] by passing to the unit groups .
[ thm : compos ] let @xmath97 , @xmath593 be commensurabilities of @xmath12-modules .
then : 1 .
the ring commensurability @xmath594 is equivalent ( see definition [ def : equiv ] ) to the composition of ring commensurabilities @xmath595 , and the group commensurability @xmath596 is equivalent to the composition @xmath597 ; 2 .
we have @xmath598 \(a ) write @xmath583 , @xmath599 @xmath600 .
we claim that there is an isogeny @xmath601 that makes the following diagram of endomorphism rings commute : @xmath602^{i } \ar[dl]\ar[dr ] & \\
\end c \ar[d]\ar[dr ] & \end ( d\circ c ) \ar[dl]\ar[dr ] & \end d \ar[dl]\ar[d ] & \\
\end l & \end m & \end n , } \ ] ] where all unlabelled morphisms are the ones defined in the introduction .
an element of @xmath603 is a pair of triples ( ( , , ) , ( , , ) ) , + l , x,,m , y , n , satisfying @xmath604 , @xmath605 , @xmath606 , @xmath607 , and the fibre product condition in fact demands that @xmath608 .
an element of @xmath609 is a triple @xmath610 satisfying @xmath611 , @xmath612 , where @xmath136 , @xmath137 are the canonical projection maps from @xmath613 to @xmath55 , respectively @xmath56 .
define i & c_md & & ( dc ) + & ( ( , , ) , ( , , ) ) & & ( , ( , ) , )
. a routine diagram chase , which we leave to the reader , shows that the image of @xmath195 is indeed contained in @xmath609 . to see that this definition of @xmath195 makes
the above diagram of endomorphism rings commute is also routine , and will also be omitted .
it remains to check that @xmath195 is an isogeny .
the correspondence @xmath614 consists of @xmath603 , together with the maps to @xmath574 and @xmath615 . by theorem [ thm : partb ] , the correspondences @xmath587 and @xmath616 are commensurabilities , so by proposition [ prop : composition ] the correspondence @xmath595 is a commensurability . in particular
, the morphism @xmath617 is an isogeny .
also , @xmath618 is an isogeny by theorem [ thm : partb ] .
the fact that @xmath195 is an isogeny therefore follows from proposition [ prop : isog ] .
this proves our claim .
the isogeny @xmath195 defines an equivalence between @xmath619 and @xmath595 .
this proves part ( a ) for endomorphism rings . by passing to the unit groups and applying theorem [ thm : o ] to the isogeny @xmath195
, we also obtain part ( a ) for automorphism groups .
part ( b ) immediately follows from part ( a ) by propositions [ prop : equiv ] and [ prop : composition ] . [
prop : equivequiv ] let @xmath0 , @xmath1 be finitely generated @xmath12-modules , and let @xmath70 , @xmath620 @xmath68 be two commensurabilities .
if @xmath70 is equivalent to @xmath109 , then @xmath587 is equivalent to @xmath616 , and @xmath592 is equivalent to @xmath621 .
let @xmath158 and @xmath179 .
first , assume that an equivalence between @xmath70 and @xmath109 is given by an isogeny @xmath622 , so that we have the following commutative diagram : @xmath623_{p}\ar[dl]_{h}\ar[dr]^j & \\ l & x\ar[r]^{g}\ar[l]_{f } & m. \\ } \ ] ] as before , write @xmath624 , and define @xmath625 , @xmath626 similarly .
then @xmath109 is canonically isomorphic to @xmath627 . by theorem [ thm : compos ] , the commensurability @xmath616 is equivalent to @xmath628 . by proposition [ prop : inverse ] ,
the composition @xmath629 is equivalent to @xmath630 .
so by proposition [ prop : equivcompos ] the commensurability @xmath628 is equivalent to @xmath631 , which is canonically isomorphic to @xmath632 .
applying theorem [ thm : compos ] again , we find that @xmath632 is equivalent to @xmath633 . finally , @xmath634 is canonically isomorphic to @xmath70 , and the special case of the proposition follows .
passing to the general case , let @xmath164 be an equivalence between @xmath70 and @xmath109 . since @xmath201 is an isogeny , @xmath70 is equivalent to @xmath635 , and since @xmath203 is an isogeny , @xmath109 is equivalent to @xmath636 .
the result therefore follows from the special case we just did .
let @xmath637 denote the category of rings , and @xmath134 the category of groups .
theorem [ thm : compos ] and proposition [ prop : equivequiv ] imply that there is a functor from @xmath547 to @xmath638 that takes an @xmath12-module @xmath0 to the ring @xmath574 , and an equivalence class of @xmath12-module commensurabilities , represented by a commensurability @xmath70 , to the equivalence class of ring commensurabilities represented by @xmath587 .
further , theorem [ thm : o ] shows that we have the functors @xmath639 and @xmath640 from @xmath638 to @xmath641 which take a ring to the additive , respectively multiplicative group of the ring .
finally , propositions [ prop : equiv ] and [ prop : composition ] imply that we have the functor @xmath642 from @xmath641 to the group @xmath9 , thought of as a groupoid with one object .
to summarise , we have the functors of groupoids @xmath643 let @xmath0 be a finitely generated @xmath12-module , and let @xmath15 denote the @xmath92-module @xmath512 .
the isomorphism of corollary [ cor : g_l ] and the functors ( [ eq : groupoids ] ) then induce group homomorphisms [ eq : homs ] _ a v & _ l & & _ > 0 , + & c & & ( ( c ) ) + & c & & ( (c ) ) .
[ lem : throughtorsion ] let @xmath0 be a finitely generated @xmath12-module , write @xmath484 , and let @xmath644 denote the quotient map .
then the isomorphism in @xmath547 given by the commensurability @xmath645 induces an isomorphism @xmath646 that commutes with the maps @xmath647 and @xmath648 defined in ( [ eq : homs ] ) .
let @xmath649 denote the commensurability @xmath650 .
then the isomorphism @xmath646 is given by composition on the right with @xmath649 and on the left with @xmath651 .
it easily follows from theorem [ thm : compos ] , proposition [ prop : composition ] and proposition [ prop : equiv ] that this isomorphism commutes with the maps ( [ eq : homs ] ) .
[ prop : throughcentre ] let @xmath0 be a finitely generated @xmath12-module , and denote the @xmath92-module @xmath512 by @xmath15 .
let @xmath652 be an element of @xmath653 .
then its image in @xmath638 under the first functor of ( [ eq : groupoids ] ) is the identity morphism on @xmath574 . by lemma [ lem : throughtorsion ]
, we may assume that @xmath0 is @xmath88-torsion free .
thus , @xmath0 injects into @xmath654 .
for any sub-@xmath12-module @xmath361 of @xmath15 , write @xmath655
. then the injection @xmath656 induces a map @xmath657 , which is injective and whose image is exactly @xmath658 .
let @xmath659 be arbitrary
. then the isomorphism @xmath660 identifies @xmath652 with the equivalence class of commensurabilities represented by @xmath661 , where @xmath662 is the inclusion map .
we have @xmath663 the commensurability @xmath587 is then of the form @xmath664 , where @xmath665 , and @xmath666 .
it follows that if @xmath652 is an element of @xmath667 , then @xmath136 and @xmath137 are equal . in this case , the commensurability @xmath668 defines an equivalence between @xmath587 and @xmath669 , the identity morphism on @xmath574 in @xmath638 .
the following result is an immediate consequence of proposition [ prop : throughcentre ] .
[ cor : throughcentre ] the two group homomorphisms @xmath670 of ( [ eq : homs ] ) factor through @xmath671 .
[ rmrk : ie ] the computation in the proof of proposition [ prop : throughcentre ] shows that the group homomorphism @xmath672 is given by @xmath673 and analogously for @xmath674 .
in this section , we will prove our main results .
we begin with theorem [ thm : mainintro ] .
we recall the statement .
[ thm : maingeneral ] let @xmath88 be an infinite domain such that for all non - zero @xmath89 the ring @xmath90 is finite , let @xmath91 be the field of fractions of @xmath88 , let @xmath92 be a semisimple @xmath91-algebra of finite vector space dimension over @xmath91 , let @xmath93 be a sub-@xmath88-algebra with @xmath94 , and let @xmath0 , @xmath1 be finitely generated @xmath12-modules .
then : 1 . there is an @xmath12-module commensurability @xmath68 if and only if the @xmath92-modules @xmath95 and @xmath96 are isomorphic ; 2 .
if @xmath97 is an @xmath12-module commensurability , then @xmath79 is a ring commensurability , and @xmath84 is a group commensurability ; 3 .
if @xmath70 , @xmath99 are @xmath12-module commensurabilities , then one has @xmath100 by theorem [ thm : q ] , the ring @xmath12 is left - noetherian , so the assumptions of notation [ not : z ] are satisfied . parts
( a ) and ( b ) of the theorem therefore follow from theorems [ thm : parta ] and [ thm : partb ] , respectively .
we now prove part ( c ) .
let @xmath675 be @xmath12-module commensurabilities . by theorem [ thm : compos ] , the assertion of part ( c ) is equivalent to the statement that @xmath676 so we may , without loss of generality , assume that @xmath677 , and it will suffice to show that the homomorphisms @xmath678 defined in ( [ eq : homs ] ) are trivial . here
@xmath15 denotes the @xmath92-module @xmath512 .
let @xmath114 denote the @xmath91-algebra @xmath679 , so that @xmath680 . since @xmath9 is abelian , both homomorphisms @xmath122 and @xmath674 factor through @xmath681 $ ] . by corollary [ cor :
throughcentre ] , they also factor through @xmath682 .
since @xmath92 is a semisimple ring , and since @xmath15 is a finitely generated @xmath92-module , it follows that @xmath15 is a finite direct sum of simple modules , so by schur s lemma @xmath114 is a direct product of matrix rings over division rings , and in particular a semisimple ring . by theorem
[ thm : x ] , the quotient @xmath116)$ ] is an abelian group of finite exponent .
since @xmath9 is torsion - free , any homomorphism @xmath116)\rar \q_{>0}$ ] must be trivial .
[ ex : notss ] the following example demonstrates that if we replace the semisimplicity assumption on @xmath92 by the condition that @xmath12 be left - noetherian , then the conclusion of theorem [ thm : mainintro](c ) need no longer hold .
recall from equation ( [ eq : homs ] ) , that @xmath689 defines a group homomorphism from @xmath690 to @xmath9 , which factors through @xmath691)$ ] .
the map @xmath692 defines an isomorphism of this quotient with @xmath693 . for @xmath694 , one
easily computes , using remark [ rmrk : ie ] , that @xmath695 .
it follows that both @xmath689 and @xmath123 map @xmath696 to @xmath697 , and are therefore far from trivial .
[ thm : qg ] let @xmath11 be a finite group , let @xmath15 be a finitely generated @xmath16$]-module , and put @xmath17 is a finitely generated @xmath18$]-module with @xmath19 as @xmath16$]-modules@xmath20 .
then there exists a unique function @xmath21 such that 1 . if @xmath0 , @xmath22 , @xmath1 , @xmath23 and @xmath24 , @xmath25 , then @xmath26 ; 2 . if @xmath0 , @xmath1 , @xmath27 , then @xmath28 ; 3 .
if @xmath29 , and @xmath30 is a submodule of finite index , then with @xmath31 and @xmath32 mapping @xmath33 to @xmath34 , one has @xmath35 existence immediately follows from theorem [ thm : maingeneral ] . in particular , property ( c ) follows by taking the commensurability @xmath698 , where @xmath699 is the inclusion map , and noticing that @xmath14 is precisely @xmath82 . to show uniqueness , note that the conditions of the theorem imply that the function @xmath700 , if it exists , is uniquely determined by its values on @xmath7-free modules . indeed
, if @xmath564 and @xmath565 are the exponents of the @xmath7-torsion submodule of @xmath0 , respectively of @xmath1 , then condition ( b ) requires that @xmath701 condition ( c ) determines the values of @xmath702 and @xmath703 , so @xmath6 is determined by @xmath704 . clearly , the modules @xmath705 and @xmath490 are both @xmath7-free . 1 .
there exists a commensurability @xmath68 if and only if @xmath0 and @xmath1 have the same rank ; 2 . if @xmath708 and @xmath709 , where @xmath710 and @xmath711 are finite abelian groups , then @xmath712 we now prove part ( b ) .
first we compute @xmath713 .
the split exact sequence @xmath714 induces a surjective map @xmath715 whose kernel is easily seen to be canonically isomorphic to @xmath716 .
it follows that if @xmath70 is the commensurability @xmath717 , then the map @xmath718 is an isomorphism , while the map @xmath719 is onto , with kernel of cardinality @xmath720
. hence @xmath721 . | we develop a theory of commensurability of groups , of rings , and of modules .
it allows us , in certain cases , to compare sizes of automorphism groups of modules , even when those are infinite .
this work is motivated by the cohen
lenstra heuristics on class groups .
the number theoretic implications will be addressed in a separate paper . |
supernova remnants ( snrs ) provide a laboratory to study various aspects of interstellar medium ( ism ) evolution across the whole electromagnetic spectrum . with the advent of high spatial resolution telescopes in both the x - ray and infrared ( ir ) regimes , it is possible to probe the interaction of the rapidly moving shock wave with the dust and gas of the surrounding ambient medium and to study the ejecta products of the sn itself .
the expanding shockwave sweeps up and heats gas to @xmath0 k , causing it to shine brightly in x - rays .
dust grains embedded in the hot , shocked plasma are collisionally heated , causing them to radiate at ir wavelengths , and slowly destroying them in the process via sputtering .
because the physical processes behind x - ray and ir emission are related , a combined approach to studying snrs using both energy ranges can reveal more information than either could on its own .
to better characterize ir emission from snrs , we conducted an imaging survey with the _
spitzer space telescope _ of @xmath1 known snrs in the large and small magellanic clouds .
the clouds were chosen because of their known distance and relatively low galactic ir background .
subsequently , we obtained ir spectra of several of the snrs in the lmc using the infrared spectrograph ( irs ) on _ spitzer_. we report here on irs observations of two of these , snrs b0509 - 67.5 ( hereafter 0509 ) and b0519 - 69.0 ( hereafter 0519 ) .
both are remnants of thermonuclear sne @xcite and have fast , non - radiative shocks ( several thousand km s@xmath2 ) @xcite
. there is no evidence in either for slower , radiative shocks .
they are both located in the lmc , at the known distance of @xmath3 kpc .
in addition , both have ages determined from light echoes @xcite , with 0509 being 400 @xmath4 and 0519 being 600 @xmath5 years old .
they are nearly identical in size , having an angular diameter of @xmath6 , which corresponds to a physical diameter of @xmath7 pc . in @xcite
hereafter paper i , we used 24 and 70 @xmath8 m imaging detections of both snrs to put limits on the post - shock density and the amount of dust destruction that has taken place behind the shock front , as well as put a limit on the dust - to - gas mass ratio in the ambient medium , which we found to be a factor of several times lower than the standard value for the lmc of @xmath9% ( weingartner & draine 2001 , hereafter wd01 ) .
these limits , however , were based on only one ir detection , at 24 @xmath8 m , with upper limits placed on the 70 @xmath8 m detection .
both remnants were detected with _
@xcite at 15 and 24 @xmath8 m ( 0519 was also detected at 11 @xmath8 m ) .
seok et al .
applied single - temperature dust models to _ akari _ data , deriving warm dust masses of 8.7 @xmath10 @xmath11 and @xmath12 @xmath11 in 0509 and 0519 , respectively . with full spectroscopic data , we can place much more stringent constraints on the dust destruction and dust - to - gas mass ratio in the ism .
we also explore alternative dust models , such as porous and composite grains .
additionally , we examine archival data from the reflection grating spectrometer ( rgs ) onboard the _ xmm - newton _ x - ray observatory . the high spectral resolution of rgs
allows us to measure the strength of lines in x - ray spectra .
we can use post - shock densities derived from ir fits to predict the strength of lines arising from shocked ambient medium , and can thus derive relative strengths of ejecta contributions to oxygen lines . doing this
requires knowledge of the shock compression ratio , @xmath13 , defined as @xmath14 ( where @xmath15 and @xmath16 are postshock and preshock hydrogen number densities , respectively ) , which for a standard strong shock ( mach number @xmath17 1 ) is 4
. however , shock dynamics will be modified @xcite if the shock is efficient at accelerating cosmic rays , and @xmath13 can be increased by a factor of several .
we use several representative values of @xmath13 in modeling x - ray line strengths .
we mapped both objects using the long wavelength ( 14 - 38 @xmath8 m ) , low - resolution ( @xmath18 64 - 128 ) ( ll ) spectrometer on _
spitzer s _ irs ( program i d 40604 ) . for 0509 , we stepped across the remnant in seven ll slit pointings , stepping perpendicularly 5.1@xmath19 each time .
this step size is half the slit width , and is also the size of a pixel on the ll spectrograph .
we then shifted the slit positions by 56@xmath19 in the parallel direction and stepped across the remnant again , ensuring redundancy for the entire object .
this process was repeated for each of the two orders of the spectrograph .
each pointing consisted of two 120-second cycles , for a total observing time of 6720 s. for 0519 , the process was identical in terms of number of positions and step sizes in between mappings , but each pointing consisted of four 30-second cycles , for a total observing time of 3360 s. the difference was due to the fact that 0519 is several times brighter in the wavelength range of interest , and we did not want to saturate the detectors . to estimate the uncertainties in the data , we fit a second - order polynomial to the line - free wavelength regions ( @xmath20 @xmath8 m ) in the spectra where the signal - to - noise was adequate for modeling .
the standard deviations of the data points in these regions were used for @xmath21 fitting of dust emission models ( see section [ errors ] ) .
we obtained @xmath22 values of 0.830 and 2.53 mjy for 0509 and 0519 , respectively .
the spectra were processed at the _ spitzer science center _ using version 17.1 of the irs pipeline .
we ran a clipping algorithm on the spectra which removes both hot and cold pixels that are more than 3-@xmath22 away from the average of the surrounding pixels . to extract the spectra , we used spice , the _ spitzer _ custom extraction tool .
once all spectra were extracted , we stacked spectra from the same spatial location , improving the signal - to - noise ratio of the sources . for background subtraction , we use the off source slit positions that come when one of the two slit orders is on the source . in the end
, we have seven ( overlapping ) background subtracted spectra for each remnant . as we show in section [ results ] , this allows us to do spatially resolved spectroscopy , despite the fact that the remnants are only @xmath23 in diameter .
we processed archival _ xmm - newton _ rgs data from the xmm science archive with version 8.0 of the science analysis subsystem ( sas ) software for xmm .
rgs provides high - resolution ( @xmath24 @xmath25 ) x - ray spectroscopy in the wavelength range from 5 - 38 .
b0509 - 67.5 was observed on 4 july 2000 ( obs .
i d 0111130201 , pi m. watson ) for 36 ksec . of these 36 ksec ,
only the last @xmath26 s were affected by flaring .
we discarded these and use the remaining 33.5 ksec for our analysis .
we use the spectra from both rgs 1 and 2 , and bin the data using the ftool _ grppha _ to a minimum of 25 counts per bin .
both rgs orders contained @xmath27 counts after time - filtering .
b0519 - 69.0 was observed on 17 september 2001 ( obs .
i d 0113000501 , pi j. kaastra ) for 48.4 ksec .
after time filtering , we obtained @xmath28 ksec of useful data , with @xmath29 counts in both rgs 1 and 2 .
we also grouped these by a minimum of 25 counts per bin .
since rgs is a slitless spectrometer , spatial information is degraded for extended sources , and the spectrum is smeared by the image of the source . in order to model rgs spectra
, the response files generated by sas must be convolved with a high - resolution x - ray image . for that purpose
, we used archival broadband chandra images from 2001 ( obs .
i d 776 , pi j. hughes ) , along with the ftool _ rgsrmfsmooth _ , to produce new response matrices .
in the top left panel of figure [ 24um ] , we show the mips 24 @xmath8 m image of 0509 , with overlays as described in the caption .
immediately obvious is the large asymmetry in 0509 , where the remnant brightens from the faint ne hemisphere to the brighter sw . dividing the remnant in half and measuring the 24 @xmath8 m fluxes from each half
, we find a flux ratio of five between these two regions . at x - ray wavelengths ,
we measure a much more modest ratio of 1.5 from archival broadband _
chandra _ images . using irs , we separated the remnant into 7 overlapping regions from which spectra were extracted . for our analysis , we chose two regions ( shown on figure [ 24um ] ) that do not overlap spatially and provide adequate signal to noise spectra .
these regions roughly correspond to the bright and faint halves of the remnant , and the spectra are shown in figure [ faintvsbright ] . in figure [ 0509faint ] and
figure [ 0509bright ] , we show both of these spectra separately , with models overlaid ( models are discussed in section [ modeling ] ) .
the ratio of integrated fluxes from 14 - 35 @xmath8 m for the two regions is @xmath30 .
this differs from the factor of 5 measured from the photometric images because the slits are not exactly aligned with the bright and faint halves of the remnant , and because of the differences in the mips and irs bandpasses .
a few obvious features immediately stand out about both spectra .
first , there is no line emission at all seen in either spectrum .
this is not unexpected , since the shocks in 0509 are some of the fastest snr shocks known , at @xmath31 km s@xmath2 @xcite , and there is no optical evidence for radiative shocks .
second , though both spectra show continuum from warm dust , there are obvious differences in the spectra .
an inflection around 18 @xmath8 m can be seen in the spectrum from the bright half , while this feature is not as clear in the faint spectrum .
we attribute this feature to the si - o - si bending mode in amorphous silicate dust . in section [ disc ] , we will explore reasons for the different spectral shapes and the differences in brightness between the two halves of the remnant . in the bottom left panel of figure [ 24um ]
, we show the 24 @xmath8 m image of 0519 , which does not show the large scale asymmetries that 0509 does , although three bright knots can be seen in the image .
these same three knots can be identified in both h@xmath32 @xcite and x - ray images , where _ chandra _ broadband data shows that the three knots are prominent in the 0.3 - 0.7 kev band @xcite . in paper
i , we measured the total flux from the three knots added together and found that collectively they represent only about 20% of the flux .
nevertheless , spectra extracted from slits that contain a bright knot do show differences in continuum slope from those extracted where no knots are present .
we discuss our interpretation of these knots in section [ disc ] . as with 0509
, there are no lines seen in the irs data ( see figure [ 0519irs ] ) , as shock speeds are also quite high in this object .
we now turn our attention to modeling the emission seen in irs with numerical models of collisionally heated dust grains .
we follow a procedure identical to that followed in our previous work on these objects ; see paper i. the heating rate of a dust grain immersed in a hot plasma depends on both the gas density ( collision rate ) and gas temperature ( energy per collision ) .
in general , the electron temperature of a shocked plasma can be determined from fits to x - ray spectra . in the case of these objects , plane - shock models without collisionless heating of electrons at the shock front yield the temperature .
dust heating models do depend on this temperature , but the dependency is not large @xcite .
gas density is a much more sensitive parameter for dust heating models , and is determined by the temperature of the grains .
we regard postshock gas density as a free parameter in our model , and fine tune the density to match the observed ir spectra .
we take into account sputtering by ions , which destroys small grains and sputters material off large grains .
we use a plane - shock model which superimposes regions of increasing shock age ( or sputtering timescale ) @xmath33 , where @xmath34 is post - shock proton density , while keeping temperature behind the shock constant .
inputs to the model are an assumed pre - shock grain - size distribution , grain type and abundance , proton and electron density , ion and electron temperature . the code calculates the heating and sputtering for grains from 1 nm to 1 @xmath8 m , producing a grain temperature and spectrum for each grain size .
the spectra are then added in proper proportions according to the post - sputtering size distribution to produce a model spectrum that can be compared with observations .
we use sputtering rates from @xcite with enhancements for small grains from @xcite .
we assume that sputtering yields are proportional to the amount of energy deposited by incoming particles , accounting for partial transparency of grains to protons and alpha particles at high energies @xcite , particularly relevant for 0509 and 0519 .
we assume that grains are compact spheres , but we also report results and implications if grains are porous ( i.e. contain a nonzero volume filling fraction of vacuum ) . in general , our method is to fix as many of the input parameters as possible , based on what is known from other observations . for instance , @xcite used optical light echoes to constrain the ages of both remnants , yielding ages of @xmath35 yrs . for 0509 and @xmath36 yrs . for 0519 .
shock velocities are known approximately for both objects from measurements of ion temperatures @xcite . in 0509 ,
the standard shock jump conditions for a 7000 km s@xmath2 shock , the assumed speed for the ne half of the remnant @xcite with no ion - electron equilibration would produce proton temperatures of @xmath37 kev .
recently , @xcite used vlt spectroscopy to detect a broad h@xmath32 component from the rim of 0509 , determining the shock speed to be @xmath38 km s@xmath2 in the ne .
they give several possibilities for the proton temperature in the ne , and note that energy from escaping cosmic - rays limits the post - shock proton temperature to a value of @xmath39 0.85 of that expected from the strong shock jump conditions .
adopting this value of 0.85 , along with a shock speed of 6000 km s@xmath2 and no electron heating gives a post - shock proton temperature of @xmath3970 kev .
we use this value as an input to dust heating models , except where noted otherwise ( see section [ iontemp ] )
. this upper limit on proton temperature makes any fit to the density for 0509 a lower limit , and we discuss the effects of changing the proton temperature in section [ iontemp ] . the shock speed in 0519 is a bit more uncertain .
@xcite quote a range of 2600 - 4500 km s@xmath2 between the limits of minimal and full ion - electron equilibration .
@xcite measured line widths of x - ray iron lines seen in _ xmm - newton _ rgs spectra to obtain a velocity of 2927 km s@xmath2 .
we adopt a value of 3000 km s@xmath2 for this work and use the standard shock jump proton temperature assuming no electron - ion equilibration , which is 21 kev .
the actual proton temperature may differ from this estimate , but the fitting of dust spectra in the ir is only weakly dependent on the proton temperature ( see section [ iontemp ] ) .
the overall normalization of the model to the data ( combined with the known distance to the lmc ) provides the mass in dust .
since our models calculate the amount of sputtering that takes place in the shock , we can then determine the amount of dust present in the pre - shock undisturbed ism that has been encountered by the blast wave .
while we are sensitive only to `` warm '' dust , and could in principle be underestimating the mass in dust if a large percentage of it is too cold to radiate at irs wavelengths , it is difficult to imagine a scenario where any significant amount of dust goes unheated by such hot gas behind the high - velocity shocks present in both remnants , particularly given the upper limits at 70 @xmath8 m reported in @xcite ( and the fact that there are no radiative shocks observed anywhere in these remnants ) .
the only fitted parameters in the model are the post - shock proton number density , which affects the shape of the model spectrum , and the radiating dust mass , which affects the overall normalization .
all other variable parameters ( e.g. electron density , sputtering timescale ) depend directly on the density .
we use an optimization algorithm to minimize the value of @xmath21 in the regions of the spectra with high signal - to - noise ( for both remnants , @xmath20 @xmath8 m ) to find the best - fitting value of the density .
we then calculate upper and lower limits on the fitted density as the 90% confidence intervals in @xmath21 space ( where confidence intervals correspond to @xmath40 + 2.71 ) , using the standard deviations listed in section 2 .
the errors on both the density and dust mass present are of order 25% for 0509 and 10% for 0519 .
larger errors in 0509 are due to a poorer signal - to - noise ratio in the spectrum .
we assume that the plasma temperature is fixed at the values listed above , but again note that the dependency on proton and electron temperature is small ( see section [ iontemp ] ) .
there are also several sources of uncertainty in modeling emission from warm dust grains that are hard to quantify statistically .
first and foremost , we are limited by both the spatial resolution of _ spitzer _ for such small snrs and the low signal - to - noise ratio of the spectra .
we use a plane - shock model with constant plasma temperature to model dust emission , which is only an approximation to a spherical blast wave .
estimates of swept gas mass assume constant ambient density ahead of the shock .
sputtering rates for such high temperature protons and alpha particles are probably only accurate to within a factor of 2 ( but see appendix a ) , and we use the simple approximation that the sputtering rate varies according to the amount of energy deposited into a grain of given size and composition .
lastly , but possibly most importantly , we do not know the degree of porosity and mixing of various grain types .
although we have full spectral mapping of both objects , the 10.5@xmath19 width of the irs slit severely limits the amount of spatially resolved spectroscopy we can do on remnants that are only 30@xmath19 in diameter , as both 0509 and 0519 are . despite this , we do have several disjoint slit positions which isolate spectra from the most prominent features in the images . because dust radiates as a modified blackbody spectrum , emission from a small amount of warmer dust can overwhelm that from a larger amount of colder dust , particularly at the wavelengths of interest here .
we show in figure [ 0509faint ] and figure [ 0509bright ] spectra extracted from two regions of 0509 , which we label the `` faint '' and `` bright '' portions of the remnant .
fits to these regions require a density contrast of @xmath41 in the post - shock gas , with the higher density required in the `` bright '' region ( higher densities means hotter dust , hence more short wavelength emission ) .
although it might be possible that a density gradient of this order actually exists in the ambient medium surrounding 0509 ( the angular size of 30@xmath19 in the lmc corresponds to a linear diameter of about 7 pc ) , the implications from this model lead to several scenarios which are unlikely and require an appeal to special circumstances . in comparing the mips image with images of the remnant at other wavelengths , one can clearly see an enhancement in the _ chandra _ broadband image @xcite on the southwest side of the remnant .
this enhancement is mostly present in the energy range containing fe l - shell emission lines .
the h@xmath32 image , shown in figure [ 24um ] , shows a uniform periphery around the remnant , with the exception of a brightness enhancement in the sw , relatively well confined to a small region . to compare the images , we convolved the h@xmath32 image to the resolution of the 24 @xmath8 m image using the mips 24 @xmath8 m psf .
the result is shown in figure [ halpha_conv ] .
the images are morphologically similar , and from these comparisons we conclude that there is not an overall ne - sw density gradient in the ism .
rather , the ism is mostly uniform except for the sw , where the remnant is running into a localized region of higher density . to obtain the conditions for the whole of the remnant
, we fit a model to the `` faint '' region , freezing all parameters to those reported in table 1 and allowing post - shock density to vary ( which also causes the sputtering timescale to vary ) .
if we assume a standard lmc dust model @xcite , we get a post - shock density of @xmath15 = 0.59 @xmath42 .
the reduced @xmath21 value for this model was 1.04 .
values in table [ resultstable0509 ] correspond to this `` faint '' region .
we can use irs to estimate the spatial extent of the bright region .
first , by looking at the 2-d spectra from the ll slit located directly on top of the brightest region in the mips 24 @xmath8 m image , we fit a gaussian to the spatial profile ( indicated by an arrow in figure [ 24um_7points ] ) of the emission between 16 and 17 @xmath8 m , and found a fwhm of the brightness profile of 12.7 .
this wavelength is the shortest at which we could achieve a good signal - to - noise ratio for this region .
the _ spitzer _
resolution at 16 @xmath8 m is about 4.75 . in the direction perpendicular to the slit positions
, we had 7 slit pointings , each separated by 5.1 .
extracting spectra from each of these ( overlapping ) positions , subtracting the background , and integrating over the irs bandpass gave us 7 measurements of the brightness as a function of position .
the locations from which these spectra were extracted are shown as diamonds in figure [ 24um_7points ] , and the resulting normalized intensity function is shown in figure [ intensitymap ] .
the brightness peaks at position 3 , the position we used to examine 2-d spectra , as described above , and falls off relatively quickly on both sides away from that position .
our assumption that the bright region of 0509 is the result of a more localized enhancement in density implies that the spectrum extracted from that region is then a composite of emission from the uniform parts of the remnant and from the small , denser region . _
spitzer _ does not have the resolution required to isolate this dense region , but we can make a crude isolation by assuming that the two slit positions in figure [ 24um ] cover an equal surface area on the remnant , and subtracting the faint spectrum from the bright .
the residual spectrum is of less than ideal signal - to - noise .
nevertheless , a fit to this spectrum implies a gas density of @xmath43 @xmath42 , or about an order of magnitude higher than the rest of the remnant .
a small region of hot dust outshines a more massive region of cooler dust at 24 @xmath8 m , which would explain the factor of 5 ratio in the flux between the two sides .
x - rays from this object are dominated by ejecta , not swept - up ism , and there are several factors to consider beyond density when considering the flux of h@xmath32 coming from a region .
higher resolution observations at all wavelengths will shine more light on this issue . for 0519
we have a similar scenario , except that instead of one bright region , we have three bright knots . as noted previously , these knots correspond spatially with knots seen in both x - rays and h@xmath32 ( see figure [ 24um ] ) .
we adopt an identical strategy here , isolating spectra from slit positions that do not overlap with one of the bright knots .
as it turns out , there is only one slit position in this object that is nearly completely free of emission from a knot , a slit position that goes directly across the middle of the remnant . the spectrum from this region is shown in figure [ 0519irs ] .
we assume that the conditions within this slit position are indicative of the remnant as a whole . using parameters found in table 1 ,
we obtain a post - shock density of @xmath15 = 6.2 @xmath42 ( reduced @xmath21 = 1.13 ) .
we do not have adequate signal - to - noise to isolate the bright knots individually .
however , we approximate the density required by noting that a fit to the spectrum extracted from the entire slit position that contains the brightest of the three knots ( the northernmost knot ) requires a density that is higher by a factor of @xmath44 . given that this spectrum contains emission from both the uniform parts of the remnant and the bright knot , it is likely that the density in the knot itself is perhaps a factor of @xmath45 higher that the more uniform parts of the remnant , which is roughly consistent with our results from paper i.
while it is typically assumed in dust grain models that grains are solid bodies composed of various materials , this is likely not a physically valid model ( see @xcite , and references therein ) .
it is possible to relax the assumption that grains are compact spheres with a filling fraction of unity .
we have updated dust models to include the porosity of dust grains ( where porosity , @xmath46 , is the fraction of the grain volume that is composed of vacuum , instead of grain material ) , as well as allowing for grains to be composite in nature , i.e. a grain made up of silicates , graphite , amorphous carbon , and vacuum .
the details of this model are discussed in @xcite .
we summarize here the important results .
perhaps the most significant effect of introducing porosity to grains is that the grain volume per unit mass increases , since the overall density of the grain is lowered . as a result , grains are more efficient radiators of energy , and less dust mass in total is required to reproduce an observed ir luminosity . for a grain of the same mass ,
the grain is heated to a higher temperature if it is solid than if it is porous . as a result of the equilibrium temperature being lower for porous grains ,
more heating is necessary to fit the same spectrum with a porous grain model than a solid grain model . for a fixed plasma temperature , this
can only be achieved by increasing the post - shock density . in summary , increasing the porosity of grains has the dual effects of raising the necessary postshock density required to heat grains to an observed temperature , while lowering the total amount of dust required to reproduce an observed luminosity .
what remains is to choose a grain `` recipe '' to use to fit the data .
since there are few constraints from the ism at this time , we have adopted here two values of @xmath47 , 25 and 50% , to explore the effects of grain porosity of fitting ir data .
our `` _ compact _ '' model , which we used to derive results given above , assumes separate populations of solid silicate and graphite grains with appropriate size distributions for the lmc ( wd01 ) .
our `` _ porous _ '' grain models are based on those of @xcite . for the 25% porous model
, we assume that grains are composed of 25% vacuum , 50% silicate , and 25% amorphous carbon .
the 50% porous grains contain 50% vacuum , 33.5% silicate , and 16.5% amorphous carbon , distributed from 0.0025 to 1.5 @xmath8 m .
the narrow bandwidth of the irs data does not allow us to determine which model fits the data better , as we obtained equally good fits from all models ( albeit with a different value of density ) , and report results from all .
a caveat to using these porous grain distributions is that both of them were optimized for the galaxy , and not the lmc .
however , to the best of our knowledge , a size distribution appropriate for porous grains in the lmc has not yet been developed .
the results are as follows : for 0509 , the best - fit @xmath47 = 25% model had a post - shock density , @xmath15 , of 1.1 @xmath42 ( @xmath21 = 1.05 ) , while the @xmath47 = 50% model required an @xmath15 of 2.2 @xmath42 ( @xmath21 = 1.07 ) . for 0519 ,
the best - fit densities were 9.5 ( @xmath21 = 1.08 ) and 17 ( @xmath21 = 1.07 ) @xmath42 for @xmath46 = 25 and 50% , respectively . in all cases ,
we assume lmc abundances , such that @xmath48 = 1.2@xmath15 .
although dust models in the ir are a powerful diagnostic of the post - shock gas , they are insensitive to pre - shock gas conditions , which determine the total swept mass . unlike many snrs ,
both of these objects have relatively well - defined ages and distances .
the post - shock densities are related to the pre - shock values by @xmath13 , which is 4 in the case of an unmodified , strong shock , but may be greater than 4 in the case of cosmic - ray modification of the blast wave . for a given @xmath13 , we can calculate the pre - shock density , and thus the total amount of swept up material .
the product of post - shock electron density and total gas mass overrun by the blast wave is proportional to the emission measure ( em ) of the x - ray emitting plasma . when combined with the ionization timescale ( @xmath49 ) and electron temperature of the gas , we have all the information necessary to generate a model x - ray spectrum for the shocked ambient medium , which we can compare directly to observed x - ray spectra .
the difficulty in this approach lies in disentangling x - ray emission from the shocked ambient medium from that arising from reverse - shocked ejecta . for large remnants ,
this could be done spatially using ccd spectra , but both 0509 and 0519 are @xmath23 in diameter with the ejecta not well separated from the shocked ism , rendering such an approach impossible .
for 0509 , previous studies @xcite have used chandra ccd spectra to model emission from the remnant , but these studies are subject to the inherent difficulty in using poor spectral resolution ccd spectra to disentangle ism and ejecta emission .
an alternative approach is to take high - resolution spectra from grating spectrometers , such as rgs on _ xmm - newton_. if lines can be identified as arising only from shocked ambient medium , then model fits to these lines could be directly compared with predicted line strengths from the model described above .
this can be most easily done with n lines , as virtually no n is produced in type ia explosions .
@xcite fit the 24.8 n ly-@xmath32 line in the spectrum of 0509 , but we find that this line is very weak and unconstraining to the fitting procedure used here . here
we generate model o ly@xmath32 and k@xmath32 lines visible in rgs spectra and compare them directly to observed line fluxes . in figure [ rgsspectra ]
, we show rgs spectra from both remnants .
as seen in figure [ rgsspectra ] , fe l - shell lines dominate the spectrum of 0509 , but strong lines from both h and he - like o are clearly visible in both cases at 18.97 and 21.7 . in order to measure the strength of observed o lines
, we fit the data with a two - component model .
one model , a _
vpshock _ model containing only iron , was fit to the data between 14 - 18 , which are dominated by iron lines .
we then added a second _ vpshock _ model at fixed normal lmc abundances ( _ wilms _ abundance model in xspec @xcite , with abundances of heavy elements set to 0.4 ; except that n = 0.1 ) with ionization timescale and normalization free to fit the data between 9 - 14 and 18 - 23 , regions where the signal - to - noise was good enough to constrain the fit .
we use two absorption components , one resulting from galactic absorption with column density , @xmath50 , equal to 5 @xmath51 10@xmath52 @xmath53 @xcite , and one from the lmc ( at fixed lmc abundances ; same as above ) , equal to 2 @xmath51 10@xmath52 @xmath53 @xcite .
these models were used only to determine the strength and width of oxygen lines .
we freeze the electron temperature for the second _ vpshock _ model to 2 kev .
this is roughly the number one gets from models of coulomb heating of electrons behind a shock of 7000 km s@xmath2 with a preshock density of 0.25 @xmath42 .
although we do not know the preshock density _ a priori _ , we use this as a representative value and note that the dependence of the emission measure of the gas on the electron temperature is small .
we introduce an additional line smoothing parameter to both models to allow for the fact that lines are broadened by doppler motions . for this model , we find that oxygen line fluxes measured from rgs spectra in 0509 are as follows : o ly@xmath32 = 4.41 @xmath54 ergs @xmath53 s@xmath2 and o k@xmath32 = 5.05 @xmath54 ergs @xmath53 s@xmath2 .
the oxygen line widths obtained from fitting are @xmath55 = 12350 ( 11220 , 13630 ) km s@xmath2 . @xcite
measured the widths of fe lines in the rgs spectrum of this object , obtaining line widths of 11500 ( 10500 , 12500 ) km s@xmath2 .
our technique for fitting rgs data from 0519 was identical , except that we extended the fitting range out to 27 to account for the fact that we had better signal - to - noise at long wavelengths in this case .
we again used two absorption components ; one for the galaxy with @xmath50 = 6 @xmath51 10@xmath52 @xmath53 @xcite and one for the lmc with @xmath50 = 1.6 @xmath51 10@xmath56 @xmath53 @xcite .
measured line strengths for this remnant are 5.81 @xmath54 ergs @xmath53 s@xmath2 and 3.91 @xmath57 ergs @xmath53 s@xmath2 for o ly@xmath32 and o k@xmath32 , respectively .
we obtain a line width for oxygen lines of @xmath55 = 3470 ( 3170 , 4000 ) km s@xmath2 . using a three - component nei model , @xcite measure fe line widths in this object to be 4400 ( 4280 , 4520 ) km s@xmath2 .
the also report that neither ne nor mg is required to fit the _ xmm _ spectra .
we find that normal abundances of ne and mg from the shocked ism are allowed by the data , and it is not necessary to exclude them from the fitting procedure .
oxygen line widths in 0519 are comparable to the assumed shock velocity , while those in 0509 are significantly higher . in section [ hydro ] , we discuss hydrodynamic modeling of 0519 , and find that given the age of 0519 , a lower shock speed of 2450 km s@xmath2 is preferred to match the observed radius of the forward shock .
line widths from a shock of this speed would fall within errors of the observed line widths above .
ir spectral fitting gives us a postshock density , which we can then divide by @xmath13 of the shock to get the preshock density . for a standard , `` strong '' shock
, @xmath13 is @xmath58 , but modification of shocks by cosmic ray acceleration will increase this .
an additional complication for the cases of these two objects is that they are both only 30@xmath19 in diameter , which is quite small for the telescope resolution at _ spitzer _ wavelengths , meaning that the spectra shown in figures 2 - 4 are the average spectra over the entire postshock region . to account for this , we computed hydrodynamic models ( see section [ hydro ] ) and calculated the mean , em - weighted @xmath13 over the post - shock region from the leading edge of the blast wave to the contact discontinuity . for the case of a standard shock ,
this mean @xmath13 is 2.65 .
this is lower than the immediate postshock value of 4 due to the sphericity of the blast wave .
the postshock density drops behind a spherical shock .
taking the example of 0509 , dust model fits to ir spectra give a postshock density of @xmath59 @xmath42 .
dividing this by 2.65 , we get an ambient pre - shock density , @xmath16 , of 0.22 @xmath42 .
the known distance to the lmc of 50 kpc , combined with the measured angular size of the remnants , gives a shocked volume of 5.7 @xmath60 @xmath61 .
multiplying the pre - shock density by the volume gives the value reported in table 2 for the swept - gas mass , 1.47 @xmath11 . dividing the dust mass inferred from fits to ir data , 1.9 @xmath62 @xmath11 , by this number
, we arrive at an ambient dust - to - gas mass ratio of 1.3 @xmath62 , the number we report in table 1 .
this dust - to - gas ratio is not folded into the models at any point , nor is it ever assumed .
we repeat these steps for various values of @xmath13 , determining different swept gas masses and dust - to - gas mass ratios .
however , note that the post - shock density and total dust mass remain constant for each case , since those were determined from fits to ir data .
we include effects of sputtering of dust grains by ions in deriving the dust masses above . in figures [ 0509_2dplot ] and [ 0519_2dplot ]
, we show the best fit values of post - shock density and dust mass for various grain models .
x - rays in 0509 and 0519 are coming mostly from reverse - shocked ejecta .
the shocked ambient medium may contribute substantially , particularly to continua and o lines , though it is difficult to evaluate this contribution based solely on x - ray data .
however , because fits to ir data provide us with the post - shock density , and the ages and sizes of these remnants are well - known , we can use x - ray emission codes to predict line strengths coming from the shocked ambient medium , assuming typical lmc abundances . in tables 2
& 3 , we show predicted oxygen line strengths as a function of @xmath13 and grain porosity for both remnants .
these line strengths were calculated in _
xspec _ by taking the plane - shock models as described in the section above and varying the ionization timescale and emission measure appropriately for each case .
real ages of 400 and 600 years for 0509 and 0519 , respectively , were divided by 3 in calculating @xmath63 as a correction factor for applying a plane - shock model to a spherical blast wave @xcite .
em was calculated as @xmath64 , where @xmath48 is post - shock electron density , @xmath16 is pre - shock density ( @xmath65 @xmath66 , where v is total volume enclosed by the blast wave ( approx .
@xmath67 @xmath61 for both remnants ) .
we chose values for @xmath13 of 4 , 8 , and 12 , which correspond to mean em - weighted @xmath13 values of 2.65 , 4.8 , and 6.8 . in the tables
, we quote our predicted o line strengths from the blast wave as a fraction of the observed strengths from rgs spectra .
values @xmath39 1 imply a contribution to o lines from ejecta , while values @xmath68 1 are inconsistent with observations and are ruled out . for 0519 , this rules out all models of porous grains .
however , this requirement also rules out compact grain models that are unmodified from `` standard '' shock jump conditions .
only those shocks with significantly higher @xmath13 are possible based on x - ray line strengths observed .
for 0509 , only one case is ruled out , that of highly porous ( @xmath47 = 50% ) grains heated behind a shock with standard jump conditions .
if grains are compact , then virtually all o line emission seen in rgs spectra would have to come from ejecta .
recent near - ir observations of type ia sne @xcite suggest that the entire progenitor is burned in the explosion and that o and ne are byproducts of carbon burning , meaning that strong x - ray lines from oxygen in the ejecta may be unlikely .
the favored explosion model for 0509 from @xcite contains 0.04 @xmath11 of unburned oxygen , the least of any of the explosion models considered for this remnant .
nevertheless , even this relatively small amount of ejecta oxygen could be enough to produce observed o lines .
model line ratios require at least some contribution from reverse - shocked ejecta to account for observed o ly@xmath32 strength .
@xcite find that at least some of the energy of the shock in 0509 must be deposited into cosmic rays .
of course , if cosmic - ray pressure is indeed non - negligible , then the standard case no longer applies , and cosmic - ray modified shocks would be required .
our assumed proton temperature of 70 kev already implies modification from the standard shock jump conditions , which would predict @xmath69 @xmath70 kev .
however , the dependency of the dust heating models on this number is small ( see section [ iontemp ] ) .
the relationship between cosmic - ray pressure and shock compression ratio is model dependent @xcite and beyond the scope of this paper , but models in table 1 with @xmath71 would be preferred .
the data do not allow us to firmly distinguish between compact and porous grain fits , as both can be consistent with the data under the right circumstances .
instead , we summarize implications for each model .
if dust grains in the ism are compact , the oxygen lines visible in x - ray spectra of 0509 are dominated by contributions from the ejecta , for all values of @xmath13 .
the inferred preshock density is inversely proportional to @xmath13 ( see table [ resultstable0509 ] )
. a `` standard '' @xmath13 value of 4 yields a preshock density of @xmath16 = 0.22 @xmath42 , significantly higher than either the value of 0.05 @xmath42 quoted by @xcite or that of 0.02 @xmath42 from @xcite , but below the 0.43 @xmath42 of @xcite . in 0519 , it is possible for nearly all of o lines to be coming from shocked ambient medium ; however , the minimum @xmath13 value required to match the o k@xmath32 line strength observed is 12 . for this value of @xmath13 ,
a small ejecta contribution to the o ly@xmath32 line is still necessary for line ratios to match .
if grains are porous , a lower ejecta contribution to oxygen lines observed in 0509 would be permitted , even for values of @xmath13 that are closer to the standard shock value .
however , in 0519 , much higher @xmath13 values would be required to bring model oxygen line fluxes down to observed levels , even with no contribution from the ejecta .
we require significantly less oxygen in the ejecta than @xcite , who report an o ejecta mass of 0.2 - 0.3 @xmath11 .
minimum values for @xmath13 would be 20 and 34 for @xmath47 = 25% and 50% , respectively .
we can set an upper limit to @xmath13 using the observed diffuse synchrotron brightness of the lmc near 0509 and 0519 , and requiring that the brightening by compression of the ambient synchrotron emissivity not exceed the observed radio brightnesses of 0509 and 0519 .
compression will boost the energy of electrons and also increase the mean magnetic field strength in the emissivity @xmath72 .
if the electron energy distribution is @xmath73 between energies @xmath74 and @xmath75 , then the emissivity is given by @xmath76 where @xmath77 and @xmath78 cgs .
( in general , @xmath79 in the notation of pacholczyk [ 1970 ] , with @xmath80 and @xmath81 . )
taking @xmath82 @xmath83 for simplicity , we find that the electron energy density @xmath84 obeys @xmath85 in the absence of turbulent effects and magnetic - field amplification , the tangential component of @xmath86 will be increased by @xmath13 while the radial component will remain the same , raising the magnitude of @xmath86 by a factor @xmath87 . for a spherical remnant encountering a uniform magnetic field ,
it is straightforward to show that the mean amplification factor @xmath88 is @xmath89 for @xmath90 .
thus , after compression , the synchrotron emissivity interior to the remnant is given by @xmath91 for a mean diffuse surface brightness upstream of @xmath92 w m@xmath93 sr@xmath2 , the mean emissivity in the lmc just depends on the line - of - sight depth of synchrotron - emitting material in the lmc .
calling that quantity @xmath94 , we have @xmath95 .
compressing this as above predicts the post - shock surface brightness for a snr of radius @xmath96 , @xmath97 , to be about @xmath98 the diffuse radio brightness of the lmc has been mapped with atca and parkes ( hughes et al .
2007 ) . from the image at 20 cm ( resolution @xmath99 ) , we estimate surface brightnesses near 0509 and 0519 of about 1 and 2 mjy / beam respectively , or about @xmath100 and @xmath101 w m@xmath93 sr@xmath2 at 1 ghz , respectively ( using a mean spectral index of the diffuse emission of 0.3 ; hughes et al . 2007 ) .
we take @xmath102 kpc . for 0519
, a high - resolution image is available @xcite , which shows a shell peak brightness of about 6 mjy / beam at 20 cm , giving ( for @xmath82 and a fwhm of @xmath103 ) @xmath104 w m@xmath93 sr@xmath2 , and @xmath105 .
there is no high - resolution image of 0509 available ; using the mean remnant surface brightness ( @xmath106 w m@xmath93 sr@xmath2 from data in mills et al . 1984 )
slightly overestimates @xmath107 , but gives a value of 25 .
larger values of @xmath13 would give higher radio brightnesses than we observe .
if shocks are indeed modified by cosmic - ray accleration , both @xmath13 and the plasma temperature behind the shock would be altered from the standard strong shock jump conditions @xcite .
@xcite report evidence in 0509 for decreased proton temperatures due to cosmic - ray modification of the shock . in this work
, we find that higher than normal values of @xmath13 are also necessary to avoid overpredicting oxygen lines in the x - ray spectra of 0519 ; such a modification would likewise lower proton temperatures from the value assumed .
dust model fits to ir data are dependent only on these post - shock conditions , and are insensitive to the degree of modification at the shock .
a detailed examination of this relationship is beyond the scope of this paper , but we can make some general comments about the effect that lowering the proton temperature behind the shock would have on fitting the data . in the case of most snrs , heating of grains by protons can be neglected , as heating is dominated by collisions with electrons .
however , for remnants with very fast shocks , proton heating becomes non - negligible @xcite . in table
[ resultstable0509 ] , we use a proton temperature of 70 kev for modeling 0509 .
we take the case of 25% porous grains as an example ( where @xmath15 was fit to be 1.1 @xmath42 ) . raising the proton temperature by a factor of @xmath108 to 115 kev
, the temperature expected from a 7000 km s@xmath2 shock with no electron - ion equilibration lowers the fitted density to 1.0 @xmath42 , an effect of only about 10% .
lowering by a factor of @xmath109 to @xmath110 kev raises @xmath15 to 1.3 @xmath42 .
finally , setting the proton temperature to @xmath111 = @xmath112 = 2 kev ( a lower limit ; although unphysical in this object given that @xcite report detection of broad ly@xmath113 uv line with width 3710 km s@xmath2 ) requires @xmath15 = 2.0 @xmath42 .
these effects are non - negligible , but are not enough to affect the overall conclusions stated above .
we can derive a conservative lower limit on the pre - shock density in 0519 using one of two methods .
the h@xmath32 fluxes from @xcite give a firm lower limit on @xmath16 of 0.2 @xmath42 at our assumed shock speed into a fully neutral medium , and assuming 0.2 h@xmath32 photons per atom entering the shock . the radio upper limit to @xmath13 ( see section [ radiolimits ] ) of 25 , which translates to a mean @xmath13 of @xmath114 , combined with the minimum post - shock density of 6.2 @xmath42 yields a lower limit on @xmath16 of 0.5 @xmath42 .
an upper limit on the pre - shock density in 0519 is given in table 3 as 0.91 @xmath42 , which is a factor of @xmath30 lower than the pre - shock density of @xmath16 = 2.4 @xmath42 quoted by @xcite , who derive their ambient density by fitting a six - component nei model to the x - ray spectra from _ chandra _ and _ xmm - newton_. in their model , most of the oxygen comes from the ejecta , which requires a very high ionization timescale of @xmath115 @xmath42 s ( a factor of @xmath116 higher than our value reported in table 3 ) and thus a very high em ( @xmath117 @xmath61 ) for the shocked csm component , implying a higher pre - shock density .
there is roughly a factor of 2 - 3 between our lower and upper limits to the pre - shock density . to explore both of these limits
, we use the virginia hydrodynamics ( vh-1 ) time - dependent hydrodynamic numerical code in one dimension .
vh-1 is a hydrodynamics code which solves the eulerian equations of fluid mechanics using the piecewise parabolic method , described in @xcite .
we assume spherical symmetry on a numerical grid with 512 zones .
we modify @xmath13 for the forward shock by varying the adiabatic index , @xmath118 (= @xmath119 for a normal shock ) , of the shocked ambient medium @xcite .
softening the equation of state effectively makes the shocked plasma more compressible , yielding @xmath13 @xmath68 4 .
we use the exponential ejecta density profiles of @xcite and assume an explosion energy of @xmath120 ergs with an ejected mass of 1.4 @xmath11 . in figure [ fsrs ] , we plot the evolution of the forward and reverse shocks with time for varying values of @xmath118 , corresponding to immediate postshock @xmath13 values of 4 , 8 , and 12 , using identical scalings to that of @xcite ( compare with figure 2a of their paper ) . with the mean post - shock density fixed at 6.2 @xmath42 ( the gas density required to fit the spectrum assuming compact grains ) , we examined @xmath13 values of 12 and 25 , corresponding to @xmath118 values of 1.18 and 1.085 , mean @xmath13 in the post - shock region of 6.8 and 12.5 , and pre - shock densities of 0.91 and 0.50 , respectively .
the lower density case , with @xmath121 , reaches a radius of 3.6 pc in 550 years with a shock velocity at that time of 2950 km s@xmath2 .
while this age is roughly consistent with the known age of this object from light echoes , the shock velocity is too high to be consistent with the measured width of oxygen lines .
this shock speed would imply a width , @xmath55 , of 5660 km s@xmath2 .
the measured value was @xmath122 km s@xmath2 .
the higher density case requires 630 years to reach the observed radius , with a current shock speed of 2450 km s@xmath2 .
the line width implied from this shock speed is 4490 km s@xmath2 .
while this is still higher than our observed line width , it is much closer to being within the quoted errors , and is clearly favored over the lower density case .
errors in determining the post - shock density from ir fits are about 20% , but this has almost no effect on the inferred pre - shock density allowed by not overpredicting oxygen lines in rgs spectra .
however , this number is sensitive to the soft x - ray absorbing column density used in the models .
if the lmc absorption component , detailed in section [ 0519xray ] , is increased , a higher pre - shock density will be allowed before oxygen lines are overproduced . to be within the errors of the line widths measured for oxygen lines ,
the lmc @xmath50 value would only need to be modestly increased from 1.6 to 1.7 @xmath123 @xmath53 .
@xcite report an @xmath50 of 2.6 @xmath123 @xmath53 , which would easily accommodate a higher density .
uncertainties in the exact distance to the remnant are also non - negligible here .
we assume a distance to the lmc of 50 kpc , but recent measurements of the distance to the lmc range from 47.8 kpc @xcite to 51.1 kpc @xcite .
the lower of these two would increase the observed radius of the remnant from 3.6 to 3.77 pc , which would bring the observed line width within errors of the shock velocity from hydro modeling .
the unknown position of 0519 within the depth of the lmc may also be important . of particular note in all results presented in tables 2 and 3
is that given the amount of gas that these remnants have swept up at this stage of their evolution , the amount of dust observed by _
spitzer _ is lower than what is expected , even accounting for the effects of sputtering ( although there are a few values in table 2 that are acceptable ; namely , those with a high @xmath13 ) .
the amount of dust sputtered is @xmath109% for compact grains in 0509 , and 34% for compact grains in 0519 .
uv studies of grain absorption in the ism @xcite yield a dust - to - gas mass ratio in the lmc of 2.5 @xmath62 .
recent wide - field far - ir observations of the lmc give a value of @xmath124 for this ratio @xcite .
this shortfall is consistent with our previous studies of both lmc and galactic snrs @xcite .
similar results have been found for sn 1987a @xcite in the lmc , although the shocks there probe circumstellar material , and have not yet reached the ambient ism ( but see dwek et al .
2008,2010 ) , and for g292.0 + 1.8 @xcite and puppis a @xcite in the galaxy . of all parameters in the model , dust mass is the _
least _ model - dependent . as a check of this
, we examine the case of 0519 . assuming compact grains , we derive a total radiating dust mass of 1.5 @xmath125 based on model fits to irs data .
we then made fits to the data using a greatly simplified model , that of a single grain at a single temperature , ignoring heating , sputtering , and grain - size distribution .
this model yielded a total dust mass of about 1.3 @xmath125 .
as a further check , we used an approximate analytic expression from @xcite for the mass of dust , dependent only on the temperature of the grain and the observed ir luminosity , obtaining a value of 1.7 @xmath126 .
these three different approaches of decreasing complexity only differ by at most a factor of 30% . taking our _ minimum _ value of swept - up gas mass ( 5.3 @xmath11 )
, we derive dust - to - gas mass ratios of 2.4 - 3.2 @xmath127 , an order of magnitude lower than the expected value .
dust - to - gas mass ratios in 0509 are closer to expected values , but are , in most cases in table [ resultstable0509 ] , still too low .
it may be that snrs that are in regions of low background sufficient to allow a high enough signal - to - noise ratio for a detailed study in the mid- and far - ir have such a low background because they are in atypical regions of the ism , creating an observational bias .
in addition , dust - to - gas mass ratio determinations from snrs are _ spot _ measurements , and are not averaged over long lines of sight or large fields of view .
porous dust grains require a higher density relative to compact grains to reproduce the same spectrum , because they are more efficient radiators of energy than compact grains .
for this reason , the inferred dust mass from the porous grain models , seen in table 2 , is lower by a factor of @xmath128 .
however , for a given @xmath13 , the inferred gas mass is higher for porous grains , so that the dust - to - gas mass ratios for porous models are lower by a factor of @xmath41 for @xmath47 = 25% and @xmath129 for @xmath47 = 50% , when compared to the compact grain case .
it should be noted though , that typical dust - to - gas mass ratios calculated for the ism in general ( such as the 0.25% given above from wd01 ) inherently assume compact grains .
one possibility that has been suggested for snrs of both type ia and cc origin @xcite is that large amounts of dust are indeed present , but are too cold ( t@xmath130 @xmath131 k ) to be detectable at _ spitzer _ wavelengths and are only visible in the sub - millimeter regime .
if this is true , then the total dust - to - gas mass ratio for the ism could be quite different from what we report here , which is based only on `` warm '' dust .
this claim has been disputed for kepler s snr @xcite based on a detection at 70 @xmath8 m and an upper limit at 160 @xmath8 m , but we lack detection at either wavelength here , and upper limits are unconstraining . _ herschel _ observations would provide the key wavelength coverage between the mid - ir and sub - mm necessary to settle this issue . a recent _ herschel _ observation of cas
a at wavelengths up to 500 @xmath8 m revealed only 0.075 @xmath11 of dust there , much of which may be newly - formed ejecta dust @xcite .
how cold would dust have to be to escape detection at 70 @xmath8 m , assuming a standard dust - to - gas mass ratio for the lmc ? obviously , the answer depends on the choice of models from tables 2 & 3 . as an example
, we consider the @xmath47 = 0 model for 0519 with the highest value for @xmath13 ( the only model for this remnant that seems to not be ruled out by x - ray lines ) .
the swept gas mass for this case is 6.1 @xmath11 , and assuming a dust - to - gas ratio of 0.35% gives a dust mass of 2.1 @xmath132 @xmath11 . in order to avoid
violating the 70 @xmath8 m upper limit of 121 mjy @xcite , post - shock dust in this remnant would need to be @xmath133 k. if we choose the standard @xmath13 case , the implied dust mass would be 5.5 @xmath134 @xmath11 , and the required post - shock dust temperature would be @xmath135 k. both of these numbers are significantly lower than those fit to irs data or those predicted by grain heating models .
such low temperatures could only be acheived by having the dust reside in very dense clumps .
shocks driven into such clumps should quickly become radiative , but no [ o iii ] or [ s ii ] emission is seen from either remnant .
we present mid - ir spectral observations of two young snrs in the lmc , 0509 - 67.5 and 0519 - 69.0 , as well as analysis of archival high - resolution x - ray spectra . by fitting dust heating and sputtering models to ir spectra , we can determine post - shock gas density .
we derive post - shock densities of 0.59 and 6.2 @xmath42 in 0509 and 0519 , respectively , using a compact grain model .
porous grains require a density higher by a factor of a few . by assuming a value for the shock compression ratio , @xmath13
, we can infer the pre - shock density , swept gas mass , strength of oxygen ly@xmath32 and k@xmath32 x - ray lines , and dust - to - gas mass ratio for the ambient ism . derived values for the pre - shock density of the ism vary greatly depending on the model used and compression ratio assumed , but for the sake of comparison , values listed in tables 2 & 3 are comparable to densities inferred from various line - of - sight hi column densities through the lmc .
these column densities vary from 10@xmath52 @xmath53 to a few @xmath123 @xmath53 , which , assuming an lmc depth of 1 kpc , yields densities of @xmath136 . for standard strong shocks , @xmath13 is 4 , but modification by cosmic rays would increase this by an unknown amount .
we report values of the quantities above for several plausible values of @xmath13 . in principle
, this method could be used in reverse to determine @xmath13 .
this would require x - ray emission from shocked ambient medium that is well - separated from ejecta emission ; thus , x - ray lines could be modeled as solely arising from material at cosmic abundances .
we believe this method can be used for older , larger remnants in the lmc , such as dem l71 . based on our analysis , a significant fraction of the o line emission seen in the x - ray spectrum of 0509 arises from the ejecta .
if grains are highly porous , then the ejecta contribution is less , but a contribution is still required because the line flux ratios observed still do not match the model ratios .
both standard shocks and cosmic - ray modified shocks with higher @xmath13 values can provide acceptable fits for 0509 .
the data for 0509 are inconclusive in favoring either compact or porous grains .
compact grains require nearly all of the oxygen seen in x - ray spectra to be coming from shocked ejecta , while porous grains would significantly lower the ejecta contribution , implying a higher contribution from the forward - shocked material . in 0519
, there is significant evidence for higher than standard compression , and little contribution from ejecta to o x - ray lines .
a shock speed of @xmath39 2300 km s@xmath2 is favored for this object from hydrodynamical simulations , in order to reproduce the observed size , age , and x - ray line widths .
compact grains are favored for this remnant .
the standard value of @xmath137 is possible only if absorption to the remnant is significantly higher than has been reported .
the column density from the lmc would need to be at least @xmath138 @xmath53 to bring oxygen lines in the model down to below observed values .
@xcite derive an upper limit for 0519 of @xmath139 , implying an upper limit to the hi column density of @xmath140 @xmath141 @xmath53 @xcite .
we derive dust - to - gas mass ratios that are lower by a factor of several than what is generally expected in the ism .
since the general properties of dust in the ism come from optical / uv absorption line studies averaged over long lines of sight , it is possible that there exist large local variations on smaller scales .
the ratios presented here probe parsec scales .
another potential explanation is that the sputtering rate for dust grains is significantly understated in the literature .
however , since the deficit of dust is about an order of magnitude , sputtering rates would also have to be increased by this amount to account for the discrepancy . also , because sputtering rates typically assume a compact grain , further work in this field will be needed to determine if these rates are appropriate for porous grains .
extremely high values for @xmath13 ( @xmath68 50 ) could bring dust - to - gas mass ratios up to more typical values , but such values are ruled out by radio observations .
a possible observational bias could exist , in that snrs that are easy to study in the ir ( i.e. , well separated from sources of ir confusion ) could be found preferentially in low - dust regions of the ism .
studies such as this will benefit greatly from the increases in both spatial resolution and sensitivity of future generations of telescopes , such as the _ james webb space telescope_. being able to spatially separate the dust spectra right behind the shock from that further inside the shell will be crucial to reducing some of the uncertainties listed above .
longer wavelength observations with the _ herschel space observatory _ would be useful for detecting or placing upper limits on any cold dust that may be present in the remnant .
we thank john blondin for providing the vh-1 hydrodynamics code and for useful discussions on interpretation , and the anonymous referee for many useful comments which improved the paper .
we acknowledge support from _ spitzer _
guest observer grant rsa 1328682 .
_ facilities : _ , lccccc b0509 - 67.5 & 6000 & 2.0 & 70 & 400 & 1 , 2 + b0519 - 69.0 & 3000 & 1.5 & 21 & 600 & 1 , 2 + [ inputtable ] lccccccccc 0 & 0.59@xmath142 & 0.22@xmath143 & 2.97@xmath144 @xmath145 & 0.089@xmath146 & 0.026@xmath147 & 0.1@xmath148 & 1.47@xmath149 & 1.90@xmath150 @xmath62 & 1.3@xmath151 @xmath62 + 0 & 0.59@xmath142 & 0.12@xmath152 & 2.97@xmath144 @xmath145 & 0.048@xmath153 & 0.017@xmath154 & 0.054@xmath155 & 0.8@xmath156 & 1.90@xmath150 @xmath62 & 2.4@xmath157 @xmath62 + 0 & 0.59@xmath142 & 0.09@xmath158 & 2.97@xmath144 @xmath145 & 0.036@xmath159 & 0.014@xmath160 & 0.041@xmath161 & 0.6@xmath162 & 1.90@xmath150 @xmath62 & 3.2@xmath163 @xmath62 + + + 25% & 1.1@xmath164 & 0.42@xmath165 & 5.53@xmath166 @xmath145 & 0.32@xmath167 & 0.12@xmath168 & 0.39@xmath169 & 2.80@xmath170 & 9.37@xmath171 @xmath127 & 3.3@xmath172 @xmath127 + + + 50% & 2.2@xmath173 & 0.83@xmath174 & 1.11@xmath175 @xmath176 & 1.25@xmath177 & 0.62@xmath178 & 1.30@xmath179 & 5.53@xmath180 & 4.1@xmath181 @xmath127 & 7.4@xmath182 @xmath10 + [ resultstable0509 ] lccccccccc 0 & 6.2@xmath183 & 2.34@xmath184 & 4.69@xmath185 @xmath176 & 9.92@xmath186 & 2.10@xmath187 & 2.63@xmath188 & 15.6@xmath189 & 1.95@xmath190 @xmath62 & 1.3@xmath191 @xmath127 + 0 & 6.2@xmath183 & 1.29@xmath192 & 4.69@xmath185 @xmath176 & 5.47@xmath193 & 1.16@xmath194 & 1.46@xmath195 & 8.60@xmath196 & 1.95@xmath190 @xmath62 & 2.3@xmath197 @xmath127 + 0 & 6.2@xmath183 & 0.91@xmath198 & 4.69@xmath185 @xmath176 & 3.86@xmath199 & 0.82@xmath200 & 1.02@xmath201 & 6.06@xmath202 & 1.95@xmath190 @xmath62 & 3.2@xmath203 @xmath127 + + + 25% & 9.5@xmath204 & 3.58@xmath205 & 7.18@xmath206 @xmath176 & 23.2@xmath207 & 3.83@xmath208 & 4.19@xmath209 & 23.9@xmath210 & 9.34@xmath211 @xmath127 & 3.9@xmath212 @xmath10 + + + 50% & 17@xmath213 & 6.42@xmath214 & 1.29@xmath215 @xmath216 & 74.6@xmath217 & 7.80@xmath218 & 7.95@xmath219 & 42.8@xmath220 & 4.80@xmath221 @xmath127 & 1.1@xmath222 @xmath10 + [ resultstable0519 ] m image of snr b0509 - 67.5 , overlaid with regions of spectral extraction as described in text , where green region is `` faint '' region and yellow marks `` bright '' region .
top right : star - subtracted h@xmath32 image of 0509 .
bottom left : mips 24 @xmath8 m image of snr b0519 - 69.0 , with extraction slit shown in yellow .
bottom right : star - subtracted h@xmath32 image of 0519 .
spectral extraction regions in both remnants are approximately 10.5@xmath19 @xmath51 45@xmath19 .
fwhm of mips 24 @xmath8 m psf is approximately 7@xmath19 .
h@xmath32 images obtained with the 4-m blanco telescope at noao / ctio . in all images , north is up and east is to the left .
[ 24um ] , width=453 ] , width=642 ] m irs spectrum of the `` faint '' region of 0509 , overlaid with model fit .
[ 0509faint ] , width=604 ] m irs spectrum of the `` bright '' region of 0509 , overlaid with model fit .
[ 0509bright ] , width=604 ] m irs spectrum of 0519 , extracted from a slit placed across the middle of the remnant , free of emission from the bright knots seen in the 24 @xmath8 m image ; with model overlaid .
[ 0519irs ] , width=604 ] image of b0509 - 67.5 from figure [ 24um ] , convolved with the mips 24 @xmath8 m psf .
right : mips 24 @xmath8 m image , at native resolution .
[ halpha_conv ] , width=566 ] m image of 0509 .
arrow shows both direction of slit placement and length of inferred brightness enhancement in that spatial direction .
diamonds are extraction centers for the parallel slit positions extracted in the text ( note that arrow coincides with position 3 ) .
intensity function plotted in figure [ intensitymap ] corresponds to diamonds from left - to - right in this image .
green circle is mips 24@xmath8 m psf .
[ 24um_7points ] , width=566 ] .
[ intensitymap ] , width=566 ] , width=453 ] , width=548 ] , but for 0519 [ 0519_2dplot ] , width=548 ] , of the shocked medium .
top curves correspond to the forward shock ; bottom curves correspond to the reverse shock . @xmath223 and @xmath224 are normalized to values given in @xcite .
solid line here corresponds to the solid line in figure 2a of their paper . note that decreasing @xmath118 has the effect of shrinking the post - shock volume .
effective @xmath118 values of 5/3 , 1.285 , and 1.18 correspond to effective @xmath13 values of 2.65 , 4.8 , and 6.8 , respectively .
[ fsrs ] , width=529 ] ) as a function of proton energy . taken from the nist _
pstar _ database ( see text ) .
[ si02stopping ] , width=566 ] , width=604 ] = 2 kev ) is included in this model .
projected range is defined as the average length traveled by a particle into a material .
[ spectracompare ] , width=604 ] . [ spectraratio ] , width=604 ]
andersen , h.h .
, & ziegler , j.f .
1977 , _ hydrogen : stopping powers and ranges in all elements _
( new york : pergamon ) arendt .
r.g . , dwek , e. , blair , w.p . ,
ghavamian , p. , hwang , u. , long , k.s . , petre , r. , rho , j. , & winkler , p.f .
2010 , apj , 725 , 585 badenes , c. , hughes , j.p . , cassam - chenai , g. , & bravo , e. 2008 , apj , 680 , 1149 badenes , c. , bravo , e. , borkowski , k.j . , & dominguez , i. 2003 , apj , 593 , 358 barlow , m.j . ,
2010 , a&a , 518 , 138 berezhko , e.g. & ellison , d.c .
1999 , apj , 526 , 385 blair , w.p . ,
ghavamian , p. , long , k.s . ,
williams , b.j . , borkowski , k.j .
, reynolds , s.p . , & sankrit , r. 2007 , apj , 662 , 998 blondin , j.m . , & ellison , d.c .
2001 , apj , 560 , 244 borkowski , k.j . , lyerly , w.j . ,
& reynolds , s.p . 2001 , apj , 548 , 820 borkowski , k.j . ,
et al . 2006 , apj , 642 , 141 , paper i bouchet , p. , et al .
2006 , apj , 650 , 212 bragg , w.h . , & kleeman , l. 1905 , phil . mag .
10 , 318 clayton , g.c . , wolff , m.j . ,
sofia , u.j .
, gordon , k.d . , & misselt , k.a .
2003 , apj , 588 , 871 cox , n.l.j . , cordiner , m.a . , cami , j. , foing , b.h . ,
sarre , p.j . , kaper , l. , & ehrenfreund , p. 2006 ,
a&a , 447 , 991 colella , p. , & woodward , p.r .
1984 , j. of comp .
54 , 174 dickel , j.r .
, & milne , d.k .
1995 , aj , 109 , 200 draine , b. t. , & salpeter , e. e. 1979 , apj , 231 , 77 dunne , l. et al .
2003 , nature , 424 , 285 dwarkadas , v. v. , & chevalier , r.a .
1993 , apj , 497 , 807 dwek , e. , 1987 , apj , 322 , 812 dwek , e. , & werner , m.w . 1981 , apj , 248 , 138 dwek , e. at al .
2008 , apj , 676 , 1029 fischer , p. , eppacher , c. , hofler , g. , & semrad , d. 1996 , nucl .
methods phys .
, 218 , 817 ghavamian , p. , blair , w.p . ,
sankrit , r. , raymond , j.c . , &
hughes , j.p .
2007 , apj , 664 , 304 gomez , h.l . , dunne , l. , ivison , r.j . ,
reynoso , e.m . ,
thompson , m.a . ,
sibthorpe , b. , eales , s.a . , delaney , t.m . ,
maddox , s. , isaak , k. 2009 , mnras , 397 , 1621 grocholski , a.j . , sarajedini , a. , olsen , k.a.g . ,
tiede , g.p . , & mancone , c.l .
2007 , aj , 134 , 680 helder , e.a . ,
kosenko , d. , & vink , j. , 2010 , arxiv:1007.3138 hughes , j.p . , et al .
1995 , apj , 444 , 81 hughes , a. , staveley - smith , l. , kim , s. , wolleben , m. , filipovi , m. 2007 , mnras.382 , 543 jones , f.c . & ellison , d.c .
1991 , spa .
rev . , 58 , 259 jurac , s. , johnson , r.e .
, & donn , b. 1998 , apj , 503 , 247 koerwer , j.f . 2009 , aj , 138 , 1 kosenko , d. , vink , j. , blinnikov , s. , rasmussen , a. 2008 , a&a , 490 , 223 kosenko , d. , helder , e. , vink , j. 2010 , arxiv:1001.0983 lee , h. et al . 2009 , apj , 706 , 441 marion , g.h . ,
hoeflich , p. , gerardy , c.l . ,
vacca , w.d . ,
wheeler , j.c . , robinson , e.l .
2009 , aj , 138 , 727 mathis , j.s .
, 1996 , apj , 472 , 643 meixner , m. , et al .
2010 , a & a , 518 , l71 mills , b.y . ,
turtle , a.j .
, little , a.g .
, & durdin , j.m .
1984 , austr.j.phys . , 37 , 321 .
nozawa , t. , kozasa , t. , & habe , a. 2006 , apj , 648 , 435 pacholczyk , a.g .
1970 , _ radio astrophysics _
( san francisco : freeman ) .
rest , a. , et al .
2005 , nature , 438 , 1132 seok , j. , et al .
2008 , pasj , 60 , 453 serra daz - cano , l. & jones , a.p .
2008 , a&a , 492 , 127 shen , y. , draine , b.t . , & johnson , e.t .
2008 , apj , 689 , 260 smith , r.c . ,
kirshner , r.p . ,
blair , w.p . , & winkler , p.f .
1991 , apj , 375 , 652 staveley - smith , l. , kim , s. , calabretta , m.r .
, haynes , r.f . , & kesteven , m.j .
2003 , mnras , 339 , 87 tuohy , i. r. , dopita , m. a. , mathewson , d. s. , long , k.s . , & helfand , d. j. 1982 , apj , 261 , 473 vink , j. , yamakazi , r. , helder , e.a .
, & schure , k.m .
2010 , apj , 722 , 1727 warren , j.s . , & hughes , j.p . 2004 , apj , 608 , 261 weingartner , j.c . , & draine , b.t .
2001 , apj , 548 , 296 williams , b.j .
2006 , apj , 652 , 33 williams , b.j .
2010 , ph.d .
thesis , arxiv:1005:1296 wilms , j. , allen , a. & mccray , r. 2000 , apj , 542 , 914 ziegler , j.f .
1977 , _ helium : stopping powers and ranges in all elements _
( new york : pergamon )
in young snrs , and in 0509 in particular , proton temperatures are much higher than electron temperatures , and heating by protons can not be neglected as it can in slower shocks where temperatures have had time to equilibrate . here
we investigate whether simple analytic approximations to proton and alpha particle energy deposition rates are valid at ion energies approaching or exceeding 100 kev .
the stopping power ( energy loss per unit path length ) and projected range ( average length traveled by a particle into a material ) for both electrons and protons into a given grain are generally given by analytic expressions approximating experimental data .
for electrons , these expressions fit the experimental data to within 15% for the energy range of 20 ev to 1 mev @xcite . for protons and alpha particles ,
the approximate stopping power , based on experimental data from @xcite and @xcite , is given by @xcite as where @xmath226 is the mass density of the grain .
the authors caution that this expression is valid only for @xmath227 kev .
we raise here two questions : 1 ) are these approximations appropriate for energies at or exceeding the 100 kev threshold ( as may well be the case in very young snrs with shock speeds exceeding 6000 km s@xmath2 ) ?
2 ) these approximations are based on data over 30 years old ; are they still valid ? to answer both of these questions , we turn to more recent laboratory data on the stopping power of protons and alpha particles in various materials , obtained from the online _ pstar _ and _ astar _ databases published and maintained by the national institute of standards and technology . in figure [ si02stopping ] , we show the stopping power of a proton in silicon dioxide as a function of energy .
the stopping power turns over above @xmath228 kev .
since , for an extremely fast shock , there will be a significant population of protons with energies in this range , this turnover should be properly accounted for in calculating the projected range of the particle .
interstellar grains are more complicated than silicon dioxide ( although it is possible that sio@xmath229 is a minor component of ism dust )
. since the nist databases only have information available for actual materials that can be measured in the lab , materials like `` astronomical silicate ( mgfesio@xmath230 ) '' are not available .
the individual stopping powers for the constituent elements are available , however , and can be combined to approximate the stopping power for a grain of arbitrary composition . to do this ,
we use `` bragg s rule , '' @xcite , given by where @xmath232 is the total stopping power of a molecule @xmath233 , and @xmath234 and @xmath235 are the stopping powers of the individual constituents .
since mg is not included in the nist database , we use the results of @xcite .
thus , for mgfesio@xmath230 , one has the calculated projected range for protons based on this stopping power ( similarly for graphite ) are given by the following polynomial expressions in logarithmic - space , where @xmath237 is the energy of the impinging particle in kev : ultimately , it is necessary to determine how much energy is deposited into the grain by an impinging particle , in order to calculate grain heating rates and determine whether these improved energy deposition rates matter . in figure
[ heatingrateratio ] , we show the ratio of the heating rate , @xmath242 , calculated from the nist energy deposition rates and that calculated from the analytical approximations of @xcite and @xcite , for a proton temperature of 100 kev , which is relevant for the fast shocks , such as those seen in 0509 .
figure [ spectracompare ] shows a comparison of the spectra of grains heated behind such a shock , for both the modified energy deposition rates described above and the analytical approximations .
figure [ spectraratio ] shows the ratio of the spectra .
the main difference is in the 18 - 40 @xmath8 m range , but the maximum difference is only @xmath243% .
the use of analytical approximations to proton heating rates should be valid for most cases .
nonetheless , we use the modified rates in reporting results in this work .
because heating by protons is non - negligible in the case of such fast shocks and high ion temperatures , it is necessary to account for relative gas - grain motions between dust grains and ions in the plasma .
grain heating rates for protons and alpha particles are increased by 35 and 112% , respectively , over the case where grains are at rest with respect to the plasma .
grain heating is still dominated by electrons , though , so the overall heating rate is increased by only @xmath244 25% for the moving grain case .
when folded through the model , this results in a decrease of about 10% in the plasma densities necessary to fit irs data .
these numbers are reported throughout the text .
we assume that the charge - to - mass ratio of the grains is sufficiently low that grains are not immediately affected by the passage of the shock . | we present results from mid - ir spectroscopic observations of two young supernova remnants ( snrs ) in the large magellanic cloud ( lmc ) done with the _ spitzer space telescope_. we imaged snrs b0509 - 67.5 and b0519 - 69.0 with _ spitzer _ in 2005 , and follow - up spectroscopy presented here confirms the presence of warm , shock heated dust , with no lines present in the spectrum .
we use model fits to _ spitzer _ irs data to estimate the density of the postshock gas .
both remnants show asymmetries in the infrared images , and we interpret bright spots as places where the forward shock is running into material that is several times denser than elsewhere .
the densities we infer for these objects depend on the grain composition assumed , and we explore the effects of differing grain porosity on the model fits .
we also analyze archival _ xmm - newton _ rgs spectroscopic data , where both snrs show strong lines of both fe and si , coming from ejecta , as well as strong o lines , which may come from ejecta or shocked ambient medium .
we use model fits to irs spectra to predict x - ray o line strengths for various grain models and values of the shock compression ratio . for 0509 - 67.5
, we find that compact ( solid ) grain models require nearly all o lines in x - ray spectra to originate in reverse - shocked ejecta .
porous dust grains would lower the strength of ejecta lines relative to those arising in the shocked ambient medium . in 0519 - 69.0
, we find significant evidence for a higher than standard compression ratio of 12 , implying efficient cosmic - ray acceleration by the blast wave . a compact grain model is favored over porous grain models .
we find that the dust - to - gas mass ratio of the ambient medium is significantly lower than what is expected in the ism . |
green functions of yang - mills theories are gauge - dependent quantities .
they can , however , be used as a starting point for the evaluation of hadronic observables ( see for example @xcite ) .
thus , the study of the infrared ( ir ) behavior of propagators and vertices is an important step in our understanding of qcd . in particular ,
the confinement mechanism for color charges @xcite could reveal itself in the ir behavior of ( some of ) these green functions .
this ir behavior should also be relevant for the description of the deconfinement transition and of the deconfined phase of qcd .
indeed , at high temperature color charges are expected to be debye - screened and the ( electric and magnetic ) screening masses should be related to the ir behavior of the gluon propagator ( see for example @xcite ) . among the gauge - fixing conditions employed in studies of yang - mills green functions ,
a very popular choice is the landau gauge , which in momentum space reads @xmath20 . from the continuum perspective
this gauge has various important properties , including its renormalizability , various associated nonrenormalization theorems @xcite and a ghost - antighost symmetry @xcite . in the past few years
many analytic studies of green functions in landau gauge have focused on the solution of the yang - mills dyson - schwinger equations ( dses ) , which are the exact quantum equations of motion of the theory ( see for example @xcite ) . since the dses are an infinite set of coupled equations , any attempt of solving them requires a truncation scheme .
then , some green functions ( usually the gluon and the ghost propagators ) are obtained self - consistently from the considered equations , while all the other green functions entering the equations are given as an input . for the coupled dses of gluon and ghost propagators two solutions
have been extensively analyzed ( see for example chapter 10 in ref .
@xcite and ref .
@xcite for recent short reviews ) .
the scaling solution @xcite finds in @xmath21 and 4 an ir - enhanced ghost propagator @xmath14 and a vanishing gluon propagator @xmath5 at zero momentum .
in particular , the ir behavior of the two propagators should be given respectively by @xmath22 and by @xmath23 with @xmath24 @xcite .
on the other hand , the massive solution @xcite gives ( for @xmath12 and 4 ) a free - like ghost propagator in the ir limit , i.e. @xmath25 , and a massive behavior for the gluon propagator , that is , @xmath13 and @xmath26 .
the existence of two different types of solution for the coupled gluon and ghost dses is now understood as due to the use of different auxiliary _
boundary conditions_. these conditions can be given in terms of the value of the ghost dressing function @xmath27 at a given momentum scale @xmath28 @xcite .
in particular , if one considers @xmath29 , it is clear that @xmath30 gives an ir - enhanced ghost propagator @xmath14 while @xmath31 yields a free - like behavior for @xmath14 at small momenta . as stressed in ref .
@xcite ( see also the discussion in section 4.2.2 of ref .
@xcite ) , the scaling condition @xmath30 relies on a particular cancellation in the ghost dse which , in turn , implies a specific `` critical '' value @xmath18 for the coupling constant @xmath15 @xcite .
thus , at least from the mathematical point of view , there is a one - parameter family of solutions for the gluon and ghost dses , labelled by @xmath15 or , equivalently , by @xmath32 : in the case @xmath19 one recovers the scaling solution while , for all cases @xmath33 , the solution is a massive one .
one gets a negative ghost propagator at small momenta . ] in @xmath34 , the su(3 ) physical value of the coupling seems to select the massive solution @xcite . at this point
we should recall that , when considering gauge - dependent quantities in non - abelian gauge theories , one has to deal with the existence of gribov copies @xcite ( see also ref .
@xcite for a recent review ) . indeed , for compact non - abelian lie groups defined on the 4-sphere @xcite or on the 4-torus @xcite , it is impossible to find a continuous choice of one ( and only one ) connection @xmath35 on each gauge orbit .
the effect of gribov copies is not seen in perturbation theory @xcite , i.e. the usual faddeev - popov - quantization procedure is correct at the perturbative level .
however , these copies could be relevant at the non - perturbative level , i.e. in studies of the ir properties of yang - mills theories .
different approaches have been proposed in order to quantize a yang - mills theory while taking into account the existence of gribov copies ( see for example @xcite ) .
the one usually considered , both in the continuum and on the lattice , is based on restricting the functional integration to a subspace of the hyperplane of transverse configurations .
the original proposal , made by gribov @xcite , was based on the observation that the landau gauge condition @xmath36 allows for ( infinitesimally ) gauge - equivalent configurations if the landau - gauge faddeev - popov operator @xmath37 has zero modes .
( here @xmath38 is the usual covariant derivative . ) indeed , since an infinitesimal gauge transformation @xmath39 gives @xmath40 , it is clear that the exclusion ( in the path - integral measure ) of the zero modes of @xmath41 implies that gauge copies connected by such infinitesimal gauge transformations are ignored in the computation of expectation values . in order to exclude these zero modes ,
gribov considered a stronger condition by requiring that the functional integration be restricted to the region @xmath42 of gauge configurations @xmath43 defined as @xmath44 this set , known as the ( first ) gribov region , clearly includes the vacuum configuration @xmath45 , for which the faddeev - popov operator is given by @xmath46 .
the region @xmath42 can also be defined ( see for example @xcite ) as the whole set of local minima of the functional @xmath47 \ , = \ , \int \ , { \ensuremath{\mathrm{d}}}^d x \ , a_{\mu}^b(x ) \ , a_{\mu}^b(x)$
since usually each orbit allows for more than one local minimum of @xmath47 $ ] , it is clear that the region @xmath42 is not free of gribov copies . on the contrary , in the interior of the so - called fundamental modular region @xmath48 , given by the set of the absolute minima of the functional @xmath47 $ ] ,
no gribov copies occur @xcite .
the characterization of the fundamental modular region @xmath48 , i.e. finding the absolute minima of the _ energy _ functional @xmath47 $ ] , is a problem similar to the determination of the ground state of a spin glass system @xcite .
thus , a local formulation of a yang - mills theory , with the functional measure delimited to @xmath48 , is not available , whereas a practical way of restricting the physical configuration space to the region @xmath42 was introduced by gribov @xcite . to this end
, he required that the ghost dressing function @xmath49 can not have a pole at finite nonzero momenta . after setting @xmath50
this condition can be written as @xmath51 where @xmath0 is the so - called gribov ghost form - factor @xcite . indeed ,
since the ghost propagator is given by @xmath52 i.e. it is related to the inverse of the faddeev - popov matrix @xmath41 , the above inequality known as the no - pole condition should be equivalent to the restriction of the functional integration to the gribov region @xmath42 , i.e. to the condition @xmath53 . from the discussion above ,
it is clear that both scaling and massive solutions of dses satisfy the no - pole condition , i.e. @xmath54 for @xmath10 .
indeed , in the scaling case @xcite , this condition [ together with the condition @xmath30 ] is imposed from the beginning to the solution of the dses .
on the contrary , for the massive solution , the no - pole condition is either verified a posteriori , as in ref .
@xcite , or used [ together with the condition @xmath31 ] as an input for the solution of the dses , as in refs .
in particular , in ref .
@xcite , the value of @xmath57 is fixed using lattice data .
the restriction to the first gribov region @xmath42 is also always implemented in lattice numerical simulations of green functions in landau gauge by ( numerically ) finding local minima of the functional @xmath47 $ ] .
results obtained using very large lattice volumes @xcite ( see also chapter 10 in ref .
@xcite , section 3 in ref.@xcite and ref .
@xcite for recent short reviews ) have shown that in @xmath12 and 4 the gluon propagator @xmath5 is finite and nonzero in the limit @xmath7 while the ghost propagator @xmath14 behaves as @xmath58 .
on the contrary , for @xmath4 the lattice data @xcite are in quantitative agreement with the scaling solution and one finds @xmath59 .
since the region @xmath42 is not free of gribov copies , their ( possible ) influence on the numerical evaluation of gluon and ghost propagators has been studied by various groups @xcite .
it has been found that these effects are usually observable only the ir limit and that any attempt to restrict the functional integration to the fundamental modular region @xmath48 gives a stronger suppression at small momenta for both propagators , i.e. reducing the value of @xmath60 and increasing that of @xmath32 .
more recently , it has been suggested @xcite that the one - parameter family of solutions obtained for the gluon and ghost dses should be related to gribov - copy effects and that the value of @xmath32 could be used as a gauge - fixing parameter .
this analysis finds indeed ir - enhanced ghost propagators ( and sometimes a disconcerting over - scaling ) . on the other hand
, the gluon propagator still shows a finite nonzero value at zero momentum , that is , @xmath13 .
moreover , this approach does not explain why the numerical results found in @xmath4 are different from those obtained in @xmath12 and 4 , even though gribov copies inside the first gribov region @xmath42 are clearly present in any space - time dimension @xmath61 . from the analytical point of view , following gribov s approach , zwanziger modified the usual yang - mills action in order to restrict the path integral to the first gribov region @xmath42 @xcite although this restriction is obtained using a non - local term , the gribov - zwanziger ( gz ) action can be written as a local action and it is proven @xcite to be renormalizable . at tree level
the gz gluon propagator is given by @xmath62 , where @xmath63 is a parameter with mass - dimension 1 .
at the same time , the ghost propagator is given by @xmath64 .
thus , as in the scaling solution of the gluon and ghost dses , the gluon propagator is null at zero momentum within the gz approach by considering a non - analytic behavior for the free energy of the system @xcite . ] and the ghost propagator is ir - enhanced @xcite .
these tree - level results , also in agreement with the original work by gribov @xcite , have been confirmed by one - loop calculations in the three- and four - dimensional cases @xcite .
more recently , the gz action has been modified by considering ( for @xmath12 and 4 ) dimension - two condensates @xcite .
the corresponding action , called the refined gribov - zwanziger ( rgz ) action , still imposes the restriction of the functional integration to the region @xmath42 and it is renormalizable .
however , the rgz action allows for a finite nonzero value of @xmath60 and for a free - like ghost propagator @xmath65 in the ir limit .
thus , nonzero values for these dimension - two condensates yield for the gluon and ghost propagators an ir behavior in agreement with the massive solution of the gluon and ghost dses .
indeed , the rgz tree - level gluon propagator describes well the numerical data in the su(2 ) case @xcite , for @xmath12 and 4 , and in the su(3 ) case with @xmath66 .
it is also interesting to note that the fitting values for the dimension - two condensates are very similar for the su(2 ) and su(3 ) gauge groups in the four - dimensional case .
as stressed above , the restriction of the functional integration to the first gribov region @xmath42 and the no - pole condition ( [ eq : nopole ] ) are key ingredients in the study of the ir sector of yang - mills theories in landau gauge .
however , to our knowledge , a detailed investigation of the properties of the gribov form - factor @xmath0 as well as of the possible implications of the no - pole condition was missing up to now , although some interesting one - loop results were already presented in refs.@xcite . in particular , in appendix b.2 of @xcite it was shown that , if the gluon propagator @xmath5 is positive , then in the @xmath6 case the derivative @xmath67 is negative for all values of @xmath68 , i.e. @xmath0 is largest at @xmath69 . also , in ref . @xcite it was proven that in the rgz framework the form - factor @xmath0 presents a logarithmic ir singularity @xmath70 for @xmath4 .
this result precluded the use of the rgz action in the two - dimensional case , leading to a first interpretation of the different behavior found in lattice numerical simulations for the @xmath6 case , compared to the @xmath12 and 4 cases .
similar findings have been ( more recently ) presented in refs .
@xcite . in this work
we collect some general properties of the gribov form - factor @xmath0 and we study the consequences of imposing the no - pole condition .
in particular , in section [ sec : ghost - prop ] , using the expression for @xmath0 obtained from the evaluation of the ghost propagator at one loop , we prove that @xmath0 attains its maximum value at @xmath69 for any dimension @xmath71 . since this expression for @xmath0 depends on the gluon propagator @xmath5 , in the same section we also investigate ( for a general @xmath2-dimensional space - time ) which ir behavior of the gluon propagator is necessary in order to satisfy the no - pole condition @xmath9 . by considering a generic ( and sufficiently regular ) gluon propagator @xmath5 , we find in the @xmath4 case that @xmath0 is unbounded unless the gluon propagator is null at zero momentum .
more exactly , we find @xmath72 in the @xmath7 limit , in agreement with @xcite .
this result does not apply to the @xmath12 and 4 cases .
indeed , in these cases one can introduce , for all values of @xmath68 , simple finite upper bounds for the gribov form - factor . in section [ sec : sigma234 ]
we present explicit one - loop calculations for @xmath0 using for the gluon propagator @xmath5 linear combinations of yukawa - like propagators ( with real and/or complex - conjugate poles ) , which have been recently used to model lattice data of the gluon propagator in the su(2 ) case @xcite .
besides confirming the results obtained in section [ sec : ghost - prop ] , we also find that the ghost propagator admits a one - parameter family of behaviors @xcite labelled by the coupling constant @xmath15 , considered as a free parameter .
moreover , the massive solution @xmath73 , corresponding to @xmath74 , is obtained for all values of @xmath15 smaller than a `` critical '' value @xmath18 . at the `` critical '' value @xmath18 , implying @xmath75 , one finds an ir - enhanced ghost propagator .
( as already stressed above , the case @xmath76 corresponds to @xmath77 and one obtains a negative ghost propagator at small momenta . ) finally , in section [ sec : dse ] , we analyze the dse for @xmath0 .
we stress that in this case we do not try to solve the dse but we focus only on general properties of this equation .
as we will see , considering ir - finite ghost - gluon vertices , we confirm and extend the one - loop analysis of the no - pole condition presented in section [ sec : ghost - prop ] . in particular , after introducing bounds for the gribov form - factor , we show again for @xmath4 that the gluon propagator @xmath5 must vanish at zero momentum in order to keep @xmath0 finite . on the contrary ,
such a constraint does not apply in the three- and four - dimensional cases .
we also present alternative evidence for the @xmath4 result using a spectral representation for the ghost propagator in the dse .
it is important to note that all our results in sections [ sec : ghost - prop ] and [ sec : dse ] apply irrespective of which set of gribov copies ( inside the region @xmath42 ) is considered , i.e.they are not affected by the so - called gribov noise .
we end with our conclusion in section [ sec : concl ] .
some technical details have been collected in four appendices . in particular , in appendix [ sec : hyper ] we present properties of the gauss hypergeometric function @xmath78 that are relevant to prove some of our results .
, width=302 ]
in this section , as well as in section [ sec : sigma234 ] below , we consider the one - loop - corrected landau - gauge ghost propagator , diagrammatically represented in figure [ figghost ] .
this propagator can be written [ for the su(@xmath1 ) gauge group in the @xmath2-dimensional case ] as @xmath79 where @xmath80 is the tree - level gluon propagator [ not necessarily given by @xmath81 and @xmath82 is the usual projector onto the transverse sub - space , i.e. @xmath83 .
we have also considered the tree - level ghost - gluon vertex @xmath84 , where @xmath28 is the outgoing ghost momentum .
the indices @xmath85 refer , respectively , to the incoming ghost , to the gluon and to the outcoming ghost . after using @xmath86 , valid for the adjoint representation ,
we obtain @xmath87 \ ; , \ ] ] where @xmath0 is the momentum - dependent function @xmath88 finally , we can write [ as in eq .
( [ eq : gsigma ] ) ] @xmath89 which corresponds to the usual resummation of an infinite set of diagrams into the self - energy . note that this resummation makes sense only when @xmath90 , i.e. when the no - pole condition ( [ eq : nopole ] ) is satisfied .
clearly the function @xmath0 is dimensionless and it should go to zero for @xmath91 , modulo possible logarithmic corrections . also , this function coincides with the so - called gribov ghost form - factor @xcite , even though the latter is obtained in a slightly different way .
@xcite compared to our eq.([sigma1 ] ) . however , this is due to the fact that in eq .
( [ eq : gini ] ) above we considered for the landau - gauge gluon propagator the usual expression @xmath92 while , in the derivation of the gribov ghost form - factor , one usually writes @xmath93 , as in eq .
( 255 ) of ref .
] as discussed in the introduction , the no - pole condition @xmath9 for @xmath10 should be equivalent to the restriction of the path integral to the first gribov region @xmath42 [ defined in eq .
( [ defgribovregion ] ) ] . in this section
we will derive general properties of @xmath0 in @xmath71 space - time dimensions .
in particular , as we will see below , the weaker condition @xmath94 is already sufficient to obtain a strong constraint on the ir behavior of the gluon propagator @xmath5 in the @xmath4 case .
following appendix b.2 of ref .
@xcite one can show that , if the gluon propagator @xmath95 is positive , then in the @xmath6 case and for all values of @xmath68 we have @xmath96 with @xmath0 defined in eq .
( [ sigma1 ] ) .
this implies that @xmath0 attains its maximum value at @xmath69 . to this end
, we choose the positive @xmath97 direction parallel to the external momentum @xmath28 and write ( using polar coordinates ) @xmath98 the integral in @xmath99 can be evaluated using contour integration on the unit circle and the residue theorem .
this yields @xmath100 where @xmath101 represents the integral on the unit circle @xmath102 , we indicated with @xmath103 the complex - conjugate of @xmath104 and @xmath105 is the step function .
this integral is also evaluated in eqs .
( [ eq:2f1-uno ] ) and ( [ eq:2f1-due ] ) in the appendix [ app : hsc ] ( for the general @xmath2-dimensional case ) . considering also
( [ eq:2f1-d2 ] ) and eq .
( [ eq : omegad ] ) , with @xmath4 , we find @xmath106 \label{eq : sigma2d } \\[2 mm ] & = & \frac{1}{8 \pi } \left [ \ , \int_0^{p^2 } \frac{{\ensuremath{\mathrm{d}}}q^2}{p^2 } \ , \mathcal{d}(q^2 ) + \ , \int_{p^2}^{\infty } \frac{{\ensuremath{\mathrm{d}}}q^2}{q^2 } \ , \mathcal{d}(q^2 ) \ , \right ] \\[2 mm ] & = & \frac{1}{8 \pi } \int_0^{\infty } \ , { \ensuremath{\mathrm{d}}}q^2 \ , \mathcal{d}(q^2 ) \ , \left [ \ , \frac{\theta(p^2 - q^2)}{p^2 } \ , + \ , \frac{\theta(q^2 - p^2)}{q^2 } \ , \right ] \ ; .
\label{eq : d2res}\end{aligned}\ ] ] then , by using @xmath107 , where @xmath108 is the dirac delta function , the derivative of @xmath0 with respect to @xmath68 yields @xmath109 which is clearly negative , for any value of @xmath68 , if @xmath95 is positive
. we can evaluate the limit @xmath110 of this derivative using , for example , the trapezoidal rule \ , + \ , { \cal o}(b - a)^3 \ ; , \ ] ] which can be obtained by integrating @xmath111 / ( b - a)$ ] .
thus , the trapezoidal rule is equivalent to using a linear taylor expansion @xmath112 for @xmath113 with the first derivative @xmath114 approximated by a ( first forward ) finite difference @xmath115 / ( b - a)$ ] . ]
@xmath116 \ , = \ , - \ , \lim_{p^2 \to 0 } \ , \frac{g^2 n_c \mathcal{d}(0)}{8 \pi p^2 } \ ; .
\label{eq : lim - deriv}\ ] ] one arrives at the same result after writing eq .
( [ eq : sigmader2 ] ) as @xmath117 where @xmath118 is a primitive of @xmath5 , that is , @xmath119 where we indicate with @xmath120 the first derivative with respect to the variable @xmath68 .
the limit @xmath110 then yields again eq .
( [ eq : lim - deriv ] ) . clearly , one finds an ir singularity at @xmath121 , unless @xmath122 . if this condition is satisfied , using again the trapezoidal rule , we have from eq .
( [ eq : lim - deriv ] ) that @xmath123 for a gluon propagator @xmath5 that is regular at small momenta , i.e. that can be expanded as @xmath124 at small @xmath68 , the above limit is finite . on the other hand ,
if the leading ir behavior of @xmath5 is proportional to @xmath125 with @xmath126 , as found for example in the @xmath6 case in refs.@xcite , then the above limit gives a singular value , due to the non - integer - power ( and non - analytic ) behavior of @xmath5 .
here we prove that for @xmath4 and for any gluon propagator @xmath5 that goes to zero sufficiently fast at large momenta , e.g. as @xmath58 , and that is reasonably regular at small momenta , e.g. that can be expanded at @xmath127 as @xmath128 ( with @xmath129 and @xmath130 and @xmath131 finite ) ) below , the taylor expansion of @xmath5 at @xmath121 is of the type considered here with @xmath132 . ] the ghost form - factor ( [ sigma1 ] ) displays a logarithmic divergence for @xmath7 proportional to @xmath60 . indeed , by considering eq .
( [ eq : sigma2d ] ) , one obtains @xmath133 \\[2 mm ] & = & \frac{1}{8 \pi } \lim_{\lambda \to \infty } \left\ { \ , \int_0^{p^2 } \frac{{\ensuremath{\mathrm{d}}}x}{p^2 } \ , \mathcal{d}(x ) \ , + \ , 2 \ , \int_p^{\lambda } \frac{{\ensuremath{\mathrm{d}}}q}{q } \ , \mathcal{d}(0 ) \ , + \ , 2 \ , \int_p^{\lambda } \frac{{\ensuremath{\mathrm{d}}}q}{q } \ , \left [ \mathcal{d}(q^2 ) \ , - \ , \mathcal{d}(0 ) \right ] \ , \right\ } \\[2 mm ] & = & \frac{1}{8 \pi } \lim_{\lambda \to \infty } \left\ { \ , \int_0^{p^2 } \frac{{\ensuremath{\mathrm{d}}}x}{p^2 } \ , \mathcal{d}(x ) \ ,
+ \ , \mathcal{d}(0 ) \
, \ln\left ( \frac{\lambda^2}{p^2 } \right ) + \ , 2 \ ,
\int_p^{\lambda } \frac{{\ensuremath{\mathrm{d}}}q}{q } \ ,
\frac{- q^{2 \eta}}{q^{2 \eta } + m } \ , \left [ \frac{\mathcal{d}(q^2 ) \ , - \ ,
\mathcal{d}(0)}{- q^{2 \eta } } \ , \left(q^{2 \eta } + m\right ) \right ] \
, \right\ } \\[2 mm ] & = & \frac{1}{8 \pi } \lim_{\lambda \to \infty } \left\ { \ , \frac{\hat{d}(p^2 ) - \hat{d}(0)}{p^2 } \ , + \ , \mathcal{d}(0 ) \
, \ln\left ( \frac{\lambda^2}{p^2 } \right ) \ , - \ , \int_{p^2}^{\lambda^2 } { \ensuremath{\mathrm{d}}}x \ , \frac{x^{\eta-1}}{x^{\eta } + m } \ , \left [ \frac{\mathcal{d}(x ) \ , - \ , \mathcal{d}(0)}{-x^{\eta } } \ , \left(x^{\eta } + m\right ) \right ] \ , \right\ } \ ; , \end{aligned}\ ] ] where @xmath134 , @xmath135 is again a primitive of @xmath136 and @xmath137 is a ( finite ) constant .
if we indicate with @xmath138 the quantity in square brackets in the last line , then we have @xmath139 & & \qquad \qquad \qquad \qquad \ ; + \ , \frac{1}{\eta } \ , \int_{p^2}^{\lambda^2 } \ , { \ensuremath{\mathrm{d}}}x \ , \ln\left ( x^{\eta } + m \right ) \ , h'(x ) \
, \biggr\ } \ ; . \label{eq : sigmaint}\end{aligned}\ ] ] note that for @xmath140 we find @xmath141 and the last term in eq .
( [ eq : sigmaint ] ) is zero . since @xmath142
we also have that @xmath143 and the two logarithmic singularities for infinite @xmath48 cancel each other .
thus , we get @xmath144 & & \qquad \qquad \ ; + \ , \frac{1}{\eta } \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln\left ( x^{\eta } + m \right ) \ , \frac{\eta m \left [ \mathcal{d}(x ) \ , - \ , \mathcal{d}(0 ) \right ] \ , - \ , x \ , \left ( x^{\eta } + m \right ) \
, \mathcal{d}'(x)}{x^{1+\eta } } \ , \biggr\ } \ ; .
\label{eq : sigmafin}\end{aligned}\ ] ] if @xmath145 at large @xmath97 , it is easy to check . ]
that @xmath0 is null for @xmath146 , as expected . at the same time , in the limit @xmath110 we obtain @xmath147 & & \quad + \ , \int_{0}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln\left ( x^{\eta } + m \right ) \ , \frac{\eta m \left [ \mathcal{d}(x ) \ , - \ , \mathcal{d}(0 ) \right ] \ , - \ , x \
, \left ( x^{\eta } + m \right ) \ , \mathcal{d}'(x)}{x^{1+\eta } } \ , \biggr\ } \ ; , \label{eq : sigmafin2}\end{aligned}\ ] ] where we used @xmath148 and @xmath149 is a finite constant .
for the gluon propagator .
] finally , in appendix [ app : relax ] we show that , under the assumptions made for the gluon propagator , can be relaxed .
] the last term on the r.h.s . of eq .
( [ eq : sigmafin2 ] ) is finite .
thus , the only ir singularity in the ghost form - factor @xmath0 is proportional to @xmath150 .
this result is in qualitative agreement with @xcite .
an ir singularity plaguing the @xmath6 calculation has also been recently obtained in ref .
@xcite . an alternative ( equivalent ) proof by essentially doing a taylor expansion of @xmath5 at @xmath121 .
] can be done by performing an integration by parts .
then , eq . ( [ eq : sigma2d ] ) becomes @xmath151 \label{eq : sigma-1l - ini } \\[2 mm ] & = & \frac{1}{8 \pi } \left [ \ , \frac{\hat{d}(p^2 ) - \hat{d}(0)}{p^2 } \
, + \ , \ln\left ( x \right ) \ , \mathcal{d}(x ) \ , \bigr|_{p^2}^{\infty } \ , - \ , \int_{p^2}^{\infty } { \ensuremath{\mathrm{d}}}x \ , \ln\left ( x \right ) \ , \mathcal{d}'(x ) \ , \right ] \\[2 mm ] & = & \frac{1}{8 \pi } \left [ \ , \frac{\hat{d}(p^2 ) - \hat{d}(0)}{p^2 } \ , - \ , \ln\left ( p^2 \right ) \ , \mathcal{d}(p^2 ) \ , - \ , \int_{p^2}^{\infty } { \ensuremath{\mathrm{d}}}x \ , \ln\left ( x \right ) \ , \mathcal{d}'(x ) \ , \right ] \ ; , \label{form2}\end{aligned}\ ] ] where we used the assumption @xmath145 at large @xmath97 .
note that the above result coincides with eq .
( [ eq : sigmafin ] ) when @xmath152 , which implies @xmath153 . a second integration by parts yields @xmath154 \ , \mathcal{d}'(x ) \ , \bigr|_{p^2}^{\infty } \ , + \ , \int_{p^2}^{\infty } { \ensuremath{\mathrm{d}}}x \ , \left [ \ , x \ , \ln\left ( x \right ) \,-\ , x\ , \right ] \ , \mathcal{d}''(x )
\ , \biggr\ } \\[2 mm ] & = & \ ! \frac{1}{8 \pi } \biggl\ { \ , \frac{\hat{d}(p^2 ) - \hat{d}(0)}{p^2 } \ , - \ , \ln\left ( p^2 \right ) \ , \mathcal{d}(p^2 ) \ , + \ , \left [ \ , p^2 \ , \ln\left ( p^2 \right ) \,-\ , p^2\ , \right ] \ , \mathcal{d}'(p^2 ) \ , + \ , \int_{p^2}^{\infty } { \ensuremath{\mathrm{d}}}x \ , \left [ \ , x \ , \ln\left ( x \right ) \,-\ , x\ , \right ] \ , \mathcal{d}''(x ) \ , \biggr\ } \ ; .\end{aligned}\ ] ] here we used the hypothesis that @xmath155 goes to zero sufficiently fast at large momenta , e.g. as @xmath156 . as before
, one easily sees that @xmath0 is null for @xmath146 ( see appendix [ app : relax ] ) . at the same time , under the assumptions made for the gluon propagator @xmath5 , in the limit @xmath110 we obtain @xmath157 \ , \mathcal{d}''(x ) \
, \biggr\ } \label{eq : sigma0due}\ ] ] and we again find we will prove that the integral on the r.h.s.of eq .
( [ eq : sigma0due ] ) is finite under the hypotheses made for the gluon propagator @xmath5 . in the same appendix
we will also show how these hypotheses can be relaxed in this case . ]
an ir singularity proportional to @xmath8 , unless one has @xmath122 .
thus , in the @xmath6 case and using a generic ( sufficiently regular ) gluon propagator , a null value for @xmath60 is a necessary condition to obtain a finite value for @xmath158 at one loop . as a consequence ,
the condition @xmath122 must be imposed if one wants to satisfy the no - pole condition ( [ eq : nopole ] ) and keep the functional integration inside the first gribov region @xmath42 .
it is important to stress again that our proofs apply to any gribov copy inside the first gribov horizon , i.e.the result @xmath122 is not affected by the gribov noise .
we can easily extend the result @xmath159 to the @xmath2-dimensional case by using for the integral in @xmath160 the so - called @xmath161-max approximation or angular approximation ( see for example @xcite ) .
the same approach allows us to show that the ir singularity @xmath8 is present only in the two - dimensional case . indeed , by using hyperspherical coordinates ( see appendix [ app : hsc ] ) and by considering the positive @xmath162 direction parallel to the external momentum @xmath28
, we can write the @xmath2-dimensional ghost form - factor ( [ sigma1 ] ) as @xmath163 in the @xmath161-max approximation one substitutes @xmath164 with @xmath58 , for @xmath165 , and with @xmath166 , for @xmath167 . then
, we obtain @xmath168 \ , \int \ , \left [ 1 - \cos^2(\phi_1 ) \right ] \ , { \ensuremath{\mathrm{d}}}\omega_d \\[2 mm ] & = & \frac{\omega_d}{(2\pi)^d } \ , \frac{d-1}{2 \ , d } \ , \left [ \ , \int_0^{p^2 } \ , { \ensuremath{\mathrm{d}}}q^2 \ , \frac{q^{d-2}}{p^2 } \ , \mathcal{d}(q^2 ) \ , + \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}q^2 \ , \frac{q^{d-2}}{q^2 } \ , \mathcal{d}(q^2 ) \ , \right ] \label{eq : sigmadd } \\[2 mm ] & = & \frac{\omega_d}{(2\pi)^d } \ , \frac{d-1}{2 \ , d } \ , \int_0^{\infty } \ , { \ensuremath{\mathrm{d}}}q^2 \ , q^{d-2 } \ , \mathcal{d}(q^2 ) \ , \left [ \ , \frac{\theta(p^2 - q^2)}{p^2 } \ , + \ , \frac{\theta(q^2 - p^2)}{q^2 } \ , \right ] \ ; , \label{eq : sigmaderiv}\end{aligned}\ ] ] where we used eq .
( [ eq : firstint ] ) .
note that , for @xmath4 and using eq .
( [ eq : omegad ] ) one gets the exact result ( [ eq : d2res ] ) . by repeating
the argument shown in the section [ sec:2dderivative ] , the proof of the inequality ( [ eq : ineq ] ) follows directly from eq .
( [ eq : sigmaderiv ] ) . at the same time , we can write eq .
( [ eq : sigmadd ] ) as @xmath169 \ , = \ , i_d(p^2,\infty ) \ ; , \label{eq : sigmaddbis}\ ] ] where the integral @xmath170 is defined in eq .
( [ eq : id - ell ] ) . in appendix
[ sec : hyper ] we have also shown that , for @xmath171 , this integral is finite when the gluon propagator @xmath5 is finite and nonzero at @xmath121 .
thus , using the @xmath161-max approximation , we find that only in the @xmath6 case the condition @xmath122 is necessary in order to obtain a finite value for the gribov form - factor @xmath0 for all values of @xmath68 . of course
, in case of ultraviolet ( uv ) divergences we should regularize the integral defining @xmath0 , as done for example in section [ sec:4d ] below for the @xmath34 case using the modified minimal subtraction ( @xmath172 ) scheme and dimensional regularization .
one can also consider a fixed momentum @xmath173 and subtract . ]
the value @xmath174 from the gribov form - factor @xmath0 . due to the use of the @xmath161-max approximation
the result of the subtraction is very simple .
indeed , instead of eq .
( [ eq : sigmaddbis ] ) we have the relation @xmath175 \ ; , \label{eq : sigmamom}\ ] ] which is valid for @xmath176 as well as for @xmath177
. then , we find again @xmath178 if @xmath136 is positive .
we can also easily check that , for @xmath13 and @xmath171 , the gribov form - factor @xmath0 in eq .
( [ eq : sigmamom ] ) does not display an ir singularity .
one can improve the results obtained in the previous section by considering the formulae reported in appendices [ app : hsc ] and [ sec : hyper ] , which allow us to perform the angular integration in eq.([eq : sigmadangular ] ) without approximations .
indeed , we have we explicitly show how to extend the proof to a mom scheme , i.e. by subtracting the value @xmath174 from @xmath0 where @xmath173 is a fixed momentum . ]
@xmath179 with @xmath180 defined in eq .
( [ eq : igeneral - dell ] ) . since @xmath181 in this case ,
for @xmath71 we can also make use of the inequalities ( [ eq : i - upper - nu1 ] ) and write @xmath182 note that @xmath183 is the same integral obtained on the right - hand side of eq .
( [ eq : sigmaddbis ] ) .
thus , the @xmath161-max approximation of the previous section provides , for @xmath12 and 4 , an upper bound for the gribov ghost factor . on the contrary , for @xmath4 ,
the above inequalities become equalities . at the same time , as one can see in appendix [ sec : hyper ] , the integral @xmath183 is finite ( for @xmath184 ) also if @xmath60 is nonzero , i.e. we do not need to impose the condition @xmath122 in order to attain a finite value for @xmath0 in the ir limit . by evaluating the derivative with respect to @xmath68 of the result ( [ eq : igeneral - dell2 ] ) we also obtain @xmath185 & & \qquad \qquad \qquad \qquad \qquad \ ; - \ , \frac{q^2 \ , \theta(p^2 - q^2)}{p^6 } \ , _ 2f'_1\left(1 , 1-d/2 ; 1+d/2 ; q^2/p^2\right ) \nonumber \\[2 mm ] & & \left .
\qquad \qquad \qquad \qquad \qquad \ ; + \ , \frac{\theta(q^2 - p^2)}{q^4 } \ , _ 2f'_1\left(1 , 1-d/2 ; 1+d/2 ; p^2/q^2\right ) \ , \right ] \ ; , \label{eq : sigmaderd}\end{aligned}\ ] ] where @xmath186 indicates the derivative with respect to the variable @xmath187 of the gauss hypergeometric function @xmath78 ( see appendix [ sec : hyper ] ) .
here we used again the properties of the theta and of the dirac delta functions and eq .
( [ eq : gauss ] ) . for @xmath4 , the last two terms in eq .
( [ eq : sigmaderd ] ) are null [ see eq .
( [ eq : derivativef ] ) ] and , using the result ( [ eq:2f1-d2 ] ) , we find again eq .
( [ eq : sigmader2 ] ) . in the @xmath34 case
one can use the expression ( [ eq:2f1-d4 ] ) for the gauss hypergeometric function @xmath188 . then , from eq .
( [ eq : sigmaderd ] ) or , equivalently , by evaluating the derivative with respect to @xmath68 of eq .
( [ eq:4dmom ] ) in appendix [ sec : mom ] we find that @xmath189 \\[2 mm ] & = & \frac{g^2 n_c}{32 \ , \pi^2 } \ , \left [ \ , \int_0 ^ 1 \ , { \ensuremath{\mathrm{d}}}y \ ; \mathcal{d}(y^2 p^2 ) \ , \left ( 2 y^5 \,-\ , 3 y^3 \right ) \ , - \ , \int_{p}^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , \frac{\mathcal{d}(q^2)}{q } \ , \right ] \ ; , \end{aligned}\ ] ] where @xmath190 and we have used eq .
( [ eq : omegad ] ) . for @xmath191
both terms in square brackets are negative , i.e. the derivative @xmath192 is negative for all values of the momentum @xmath28 .
let us note that in the original work by gribov @xcite the same result was proven [ see comment after eq .
( 37 ) in the same reference ] under the much stronger hypothesis of a gluon propagator @xmath95 decreasing monotonically with @xmath193 over the main range of integration .
a similar analysis can be done in the @xmath194 case using eq .
( [ eq:2f1-d3 ] ) . in order to simplify the notation we define @xmath195 \ ; .\ ] ] this gives @xmath196 \ ; .\ ] ] then , after setting @xmath12 in eq . and using eq . , we obtain @xmath197 & & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad + \ , \frac{\theta(q^2-p^2)}{q^4 } \ , \psi'\left(\frac{p^2}{q^2}\right ) \ , \biggr ] \ , \biggr\ } \\[2 mm ] & = & \frac{g^2 n_c}{6\pi^2 } \ , \biggl\ { \ , \int_0^{p^2 } \ , { \ensuremath{\mathrm{d}}}x \ , \sqrt{x } \,\mathcal{d}(x ) \ , \biggl [ \ , -\frac{1}{p^4 } \ , \psi\left(\frac{x}{p^2}\right ) \ , - \ , \frac{x}{p^6}\ , \psi'\left(\frac{x}{p^2}\right ) \ , \biggr ] \ , + \ , \int_{p^2}^{\infty } \ , \frac{{\ensuremath{\mathrm{d}}}x}{x^{3/2 } } \ , \mathcal{d}(x ) \ ,
\psi'\left(\frac{p^2}{x}\right ) \ , \biggr\ } \ ; , \end{aligned}\ ] ] where we also made the substitution @xmath134 . next , the change of variable @xmath198 in the first integral and @xmath199 in the second integral yield @xmath200 \ , + \ , \int_{0}^{1 } \ , \frac{{\ensuremath{\mathrm{d}}}y}{p } \ , \frac{1}{\sqrt{y } } \ , \mathcal{d}(p^2/y ) \ , \psi'(y ) \ , \biggr\ } \ ; .\end{aligned}\ ] ] as one can see in figure [ figdiff ] , the factor @xmath201 $ ] is negative for @xmath202 $ ] . at the same time , from eq .
( [ eq : d3derivative2f1 ] ) we know that @xmath203 is negative for @xmath204 ( see also the corresponding plot in figure [ figdiff ] ) .
thus , for a positive gluon propagator @xmath5 , the @xmath194 derivative @xmath205 is negative for @xmath10 .
$ ] ( full line ) and @xmath206 ( dashed line ) for @xmath202 $ ] .
[ figdiff],width=302 ] finally , we can consider a general @xmath171 and , after suitable changes of variables ( for @xmath10 ) , we write @xmath207 & & \left .
\;\;\ ; - \ , \int_0^{1 } \ , { \ensuremath{\mathrm{d}}}y \ , y^{d/2 - 1 } \
; \mathcal{d}(y p^2 ) \ , \bigl [ \ ; _ 2f_1\left(1 , 1-d/2 ; 1+d/2 ; y\right ) \ , + \ , y \ ; _ 2f'_1\left(1 , 1-d/2 ; 1+d/2 ; y\right ) \ , \bigr ] \right\ } \ ; . \label{eq : sigmaderd2}\end{aligned}\ ] ] note that the dependence on @xmath68 is only in the global factor @xmath208 and in the argument of the gluon propagator . from appendix
[ sec : hyper ] we know that the derivative @xmath209 is negative for @xmath210 $ ] and @xmath171 and that , under the same hypotheses , the expression in square brackets is positive . thus , for a positive gluon propagator , both term in the r.h.s . of the above expression are negative and we have proven that , for any dimension @xmath71 , the gribov form - factor @xmath0 ( at one loop ) is monotonically decreasing with @xmath68 , i.e. it gets its maximum value at @xmath69 .
in the previous section we have proven that , at one - loop level and for a sufficiently regular gluon propagator @xmath5 , the gribov ghost form - factor @xmath0 is always finite in three and four space - time dimensions while , in @xmath4 , one needs to impose @xmath11 in order to avoid an ir singularity of the type @xmath150 . in this section
we present an explicit calculation of @xmath0 at one loop for @xmath21 and 4 using , for the gluon propagator , results recently presented in ref .
@xcite from fits to lattice data of @xmath5 in the su(2 ) case .
the expressions obtained below for the ghost propagator @xmath14 will be used in a subsequent work @xcite to model lattice data of su(2 ) ghost propagators . in this section , besides recovering the same results reported in section [ sec : ghost - prop ] , we also find that the ghost propagator @xmath14 admits a one - parameter family of behaviors @xcite labelled by the coupling constant @xmath15 , considered as a free parameter . the no - pole condition
@xmath16 implies @xmath17 , where @xmath18 is a `` critical '' value .
moreover , for @xmath15 smaller than @xmath18 one has @xmath74 and the ghost propagator shows a free - like behavior in the ir limit , in agreement with the so - called massive solution of gluon and ghost dses @xcite . on the contrary , for @xmath19
one finds @xmath75 and the ghost propagator is ir enhanced @xcite . in ref .
@xcite the su(2 ) gluon propagator was fitted in 2 , 3 and 4 space - time dimensions using , respectively , the functions is @xmath211 and not @xmath212 as in the rest of the manuscript . ]
@xmath213 @xmath214 and @xmath215 the last two propagators are tree - level gluon propagators that arise in the study of the rgz action @xcite .
the first one is a simple generalization of the form ( [ eq : f4dgluon ] ) that fits well the @xmath6 data .
note that these three functions can be written as a linear combination of propagators of the type @xmath216 , where @xmath217 is in general a complex number .
[ in the @xmath6 case we need to consider the more general form @xmath218 with @xmath219 . ]
thus , in order to evaluate @xmath0 in eq .
( [ sigma1 ] ) using the above gluon propagators @xmath5 , we first consider the integral @xmath220 the evaluation of @xmath221 can be done in three and four space - time dimensions by introducing feynman parameters and applying the usual shift in the momentum @xmath222 .
the integration then yields @xmath223 \nonumber \\[2 mm ] & & \ ; - \frac{1}{(4\pi)^{d/2 } } \int_0 ^ 1 { \ensuremath{\mathrm{d}}}x \int_0^{1-x } { \ensuremath{\mathrm{d}}}y \left [ \frac{1}{2 } \theta^{d/2 - 2 } \ , \gamma(2-d/2 ) \ , + \ , x^2p^2 \ , \theta^{d/2 - 3 } \ , \gamma(3-d/2 ) \right ] \label{integr}\end{aligned}\ ] ] with @xmath224 since the gamma function has the behavior @xmath225 for small @xmath97 , it is clear that the first two integrals are uv finite for @xmath226 while the third one is uv finite for @xmath227 . below we will calculate the integral for @xmath228 3 and 4 .
we start from the case @xmath12 , where all terms are finite , and then we evaluate the integral for the case @xmath228 4 , using the @xmath172 scheme . on the contrary ,
as stressed above , in the @xmath6 case one needs to evaluate the more general function @xmath229 with @xmath219 .
this case will be treated ( in a slightly different way ) in section [ sec:2dghost ] .
most of the analytic results reported in this section have been checked using mathematica and/or maple . in the @xmath194 case the residual @xmath97- and @xmath161-integrations in eq .
are straightforward and give @xmath230 \ , + \ , \left [ \ , - \frac{(p^2 - \omega^2 ) \sqrt{\omega^2}}{32
\pi p^2 \omega^2 } + \frac{\pi p}{64 \pi \omega^2 } - \frac{(p^2+\omega^2)^2}{32 \pi p^3 \omega^2 } \arctan\left ( \frac{p}{\sqrt{\omega^2 } } \right ) \
, \right ] \nonumber \\[2 mm ] & & \quad + \ , \left [ \ , \frac{3 \ , ( p^2 - \omega^2 ) \sqrt{\omega^2}}{32
\pi p^2 \omega^2 } + \frac{3 p^4 - 2 p^2 \omega^2 + 3 ( \omega^2)^2}{32 \pi p^3 \omega^2 } \arctan\left ( \frac{p}{\sqrt{\omega^2 } } \right ) - \frac{3 \pi p}{64 \pi \omega^2 } \ , \right ] \ ; , \end{aligned}\ ] ] where the three square brackets highlight the contribution from the three terms in eq . .
here we have only made the assumption @xmath10 . by simplifying the above result
, we find @xmath231 & = & \frac{1}{32 \pi
p^3 \omega^2 } \ , g(p^2,\omega^2 ) \label{integr2}\end{aligned}\ ] ] where @xmath232 in order to use the result we need to write the gluon propagator as @xmath233 here @xmath234 , @xmath235 and @xmath236 are the three roots of the cubic equation , with respect to the variable @xmath68 , obtained by setting equal to zero the denominator of eq . .
thus , by combining eqs . , and we can write for the function @xmath0 in the @xmath194 case the expression @xmath237\ ] ] or @xmath238 \label{eq : sigma3d}\ ] ] with @xmath239 given in eq . . in general ,
the roots @xmath234 , @xmath240 and @xmath236 are all real or there is one real root , for example @xmath234 , and two complex - conjugate roots , i.e. @xmath241 , implying also @xmath242 . since the fits in ref .
@xcite support the latter case we write @xmath243 and @xmath244 then , following for example @xcite , we find for @xmath240 the relations @xmath245 and similar results for @xmath236 .
we also use the expression ( see for example @xcite ) @xmath246 this allows us to write the function @xmath0 only in term of real quantities , i.e.@xmath247 \label{eq : sigma3d2}\ ] ] where @xmath239 is given in eq .
( [ eq : f3d ] ) above .
also , we have @xmath248 with @xmath249 f_2(p^2 ) & = & \frac{\left ( a v \ , + \ , b w \right ) \sqrt{r+v } \,-\ , \left ( b v \ , - \ , a w \right ) \sqrt{r - v}}{\sqrt{2 } \ , \pi \ ,
r^2 } \ ; , \\[2 mm ] f_3(p^2 ) & = & - \frac{1}{p^2 } \ , \frac{a \sqrt{r+v } \,-\ , b \sqrt{r - v}}{\sqrt{2 } \
, \pi } \ ; , \\[2 mm ] f_4(p^2 ) & = & a(p^2 ) \ ; \frac{p^4 \left ( a v \ , + \ , b w \right ) \,+\ , 2\,a\,p^2\,r^2 \,+\,r^2 \left ( a v \ , - \ , b w \right)}{2 \ , \pi \ ,
r^2\ , p^3 } \ ; , \\[2 mm ] f_5(p^2 ) & = & -l(p^2 ) \ ; \frac{p^4 \left ( b v \ , - \ , a w \right ) \,+\ , 2\,b\,p^2\,r^2 \,+\,r^2 \left ( b v \ , + \ , a w \right)}{2 \ , \pi \ ,
r^2\ , p^3}\end{aligned}\ ] ] and @xmath250 \pi \,+\ , \arctan\left ( \frac{\sqrt{2 } \ , p \ , \sqrt{r+v}}{r\,-\,p^2 }
\right ) \qquad & \mbox{if } \;\;\ ; r\,-\,p^2 < 0 \end{array } \right . \ ; , \\[3 mm ] l(p^2 ) & = & \ln\left [ \frac{\sqrt{p^4 \,+\ , 2 \ , p^2\,v \,+\ , r^2 } } { r \,+\ , p \ , \left ( p \,+\ , \sqrt{2}\,\sqrt{r - v } \right ) } \right ] \ ; , \\[3 mm ] r & = & \sqrt{v^2+w^2 } \label{eq : r3d } \label{eq : f33d } \ ; .\end{aligned}\ ] ] one can check that @xmath0 is null in the limit @xmath91 . finally , by expanding @xmath0 around @xmath69 in eqs .
( [ eq : sigma3d2])([eq : r3d ] ) we find @xmath251 & & \quad + \sqrt{r - v}\;\frac{9\,b\,r^2-\left(b\,v+a\,w\right)\,\left(2\,v+r\right ) } { 24\,\sqrt{2}\,\pi\,r^3 } \ , - \ , \frac{\alpha\,r^2 + 2 \ , ( a\,v+b\,w ) \omega_1
^ 2}{32 \ , \omega_1 ^ 2 \ , r^2 } \ , p \ , + \ , o\left(p^2\right )
\ ; , \label{eq : sigmair3d}\end{aligned}\ ] ] which implies @xmath252 at very small momenta .
however , if the constant term in the above expression is equal to @xmath253 , yielding @xmath254 , then one gets in the ir limit @xmath255 or @xmath256 , depending on whether the term @xmath257 vanishes or not . in particular , in the original gz case , i.e. when the terms containing @xmath234 and @xmath258 are absent , we do recover the usual @xmath259 behavior . also note that for purely imaginary poles , i.e. when @xmath260 ( and @xmath261 ) , the condition @xmath254 simplifies to @xmath262 clearly , for a given value of @xmath1 and with a suitable choice of @xmath15 , one can always set @xmath75 in eq .
( [ eq : sigmair3d ] ) .
for example , using the numerical data in the second column of table xi of ref.@xcite and @xmath263 , we find from eq .
( [ eq : sigmair3d ] ) the result @xmath264 and we have @xmath254 if @xmath265 .
thus , if one considers @xmath15 as a free parameter , then eq .
( [ eq : sigmair3d ] ) gives a one - parameter family of behaviors , labeled by @xmath15 .
for a specific value of @xmath266 we have @xmath254 and one finds an ir - enhanced ghost propagator at one loop . on the contrary , for @xmath267
we obtain @xmath74 and @xmath268 in the ir limit .
finally , for @xmath269 the no - pole condition @xmath16 is not satisfied , i.e. the ghost propagator is negative in the ir limit .
these findings are in qualitative agreement with the dse results obtained in refs .
finally , note that at small momenta the function @xmath0 in the above formula ( [ eq : sigma3dnum ] ) is decreasing as @xmath68 increases , as expected from section [ sec : d - exact ] .
we want now to evaluate @xmath270 in eq . for @xmath66 .
as stressed above , in this case we have to deal with uv divergences .
we do the calculation in the @xmath172 renormalization scheme using dimensional regularization with @xmath271 . for the first term in eq .
we have @xmath272 \ , = \ , \frac{1}{16 \pi^2 } \int_0 ^ 1 { \ensuremath{\mathrm{d}}}x \left [ \frac{2}{\varepsilon } \ , - \ , \gamma_e \ , + \ , \ln\left(4 \pi\right ) \ , - \ , \ln\left(\delta\right ) \right ] \ ; , \ ] ] where @xmath273 is the euler constant .
then , using the usual @xmath172 prescription , we find @xmath274 \
, = \ , \frac{-1}{16 \pi^2 } \int_0 ^ 1 { \ensuremath{\mathrm{d}}}x \left [ \ln\left(\frac{p^2}{{\ensuremath{\overline \mu}}^2}\right ) \ , + \ , \ln\left(1-x\right ) \ , + \ , \ln\left(x + \frac{\omega^2}{p^2}\right ) \right]\ ] ] and the @xmath275 integration yields @xmath276 \nonumber \\[2 mm ] & = & \frac{-1}{16 \ , p^2 \pi^2 } \left
[ - 2 p^2 \ , + \ , p^2 \ , \ln\left(\frac{p^2 + \omega^2}{{\ensuremath{\overline \mu}}^2}\right ) \ , + \ , \omega^2 \ , \ln\left(\frac{p^2 + \omega^2}{\omega^2}\right ) \right ] \ ; , \label{4d6}\end{aligned}\ ] ] where @xmath277 is the renormalization scale . for the second term in eq . , which is also divergent , we first perform the @xmath161 integration exactly , obtaining @xmath278 \ ; . \label{eq:4dsecondterm}\ ] ] the @xmath279 expansion then gives @xmath280 \
; , \ ] ] where we have already applied the @xmath172 prescription , and after integrating in @xmath275 we find @xmath281 \right\ } \ ; .
\label{4d7}\ ] ] finally , the third term , which is finite , yields @xmath282 \ , = \ , - \frac{1}{16 \pi^2 \omega^2 } \int_0 ^ 1 { \ensuremath{\mathrm{d}}}x \ , x^2 p^2 \ , \left [ \ln\left ( x + \frac{\omega^2}{p^2 } \right ) \ , - \ , \ln\left(x\right ) \right ] \\[2 mm ] & = & - \frac{1}{96 p^4 \omega^2 \pi ^2 } \ ,
\left [ \ , p^2 \omega^2 ( p^2 - 2 \omega^2 ) \ , + \ , 2 \ , p^6 \ln\left ( \frac{\omega^2}{p^2 } \right ) \ , + \ , 2 \left(p^6 + \omega^6\right ) \ln\left ( \frac{p^2+\omega^2}{\omega^2 } \right ) \ , \right ] \ ; . \label{4d5}\end{aligned}\ ] ] by summing the three results above we ultimately find ( in the @xmath172 scheme ) @xmath283 \label{4d8}\ ] ] with @xmath284 f_2(p^2,\omega ) & = & - \ , \left(p^6 - p^4 \omega^2 + 3 \ , p^2 \omega^4 + \omega^6 \right ) \ln\left(\frac{p^2 + \omega^2}{\omega^2}\right ) \ ; , \\[2 mm ] f_3(p^2,\omega ) & = & p^2 \omega^2 \left [ 5 \ , p^2 + \omega^2 + p^2 \ ,
\ln\left(\frac{p^2}{{\ensuremath{\overline \mu}}^2}\right ) - 4 \
\ln\left(\frac{p^2 + \omega^2}{{\ensuremath{\overline \mu}}^2}\right ) \right ] \ ; . \label{eq : f3}\end{aligned}\ ] ] as shown in ref .
@xcite , in the @xmath34 case the fit of the gluon - propagator data is done using the expression ( [ eq : f4dgluon ] ) .
thus , in order to use the above result , we need to write the gluon propagator as @xmath285 where @xmath286 are the roots of the quadratic equation , with respect to the variable @xmath68 , obtained by setting equal to zero the denominator of eq .
( [ eq : f4dgluon ] ) .
then , the ghost form - factor in the @xmath172 scheme is given by @xmath287 \label{eq : sigma4d}\ ] ] and we have @xmath288^{-1 } \ ; .\ ] ] note that the function @xmath289 is real .
from @xcite we know that @xmath290 are complex - conjugate roots , i.e.@xmath291 and @xmath292 . by writing @xmath293 and @xmath294
we find @xmath295 \label{eq : sigma4d2}\ ] ] with @xmath296 \,-\ , ( b v - a w ) [ a_1(p^2)-a_2(p^2 ) ] \ ; , \\[2 mm ] t_2(p^2 ) & = & a [ 5+\ell_1(p^2)+\ell_2(p^2)+\ell_3(p^2)-4 \ell_4(p^2 ) ] \ , - \ , b[a_1(p^2)-a_2(p^2 ) - 4 a_3(p^2 ) ] \ ; , \\[2 mm ] t_3(p^2 ) & = & [ 1 - 3 \ell_3(p^2 ) ] ( a v^3-b w v^2+v a w^2-b w^3 ) \,- \ , 3 a_2(p^2 ) ( b v^3+a w v^2+v b w^2+a w^3 ) \ ; , \\[2 mm ] t_4(p^2 ) & = & \ell_3(p^2 ) ( a v^4 - 2 w b v^3 - 2 v b w^3-a w^4 ) \ , + \ , a_2(p^2 ) ( b v^4 + 2 a w v^3 + 2 v a w^3-b w^4)\end{aligned}\ ] ] and @xmath297 also note that , at large momenta , one gets @xmath298 finally , by expanding @xmath289 around @xmath69 in eqs .
( [ eq : sigma4d2])([eq : r ] ) we obtain @xmath299 & & \quad + \frac{\left [ -11 + 6 \ln\left ( \frac{p^2}{r } \right ) \right ] \,\left(a\,v+w\,b \right ) + 6 ( b\,v - a\,w ) \ , \arctan\left ( { \frac{w}{v } } \right ) } { 192\ , \pi^2 \ , r^2 } \ , p^2 \ , + \ , o \left ( { p}^{4 } \right ) \ ; . \label{eq : sigmair}\end{aligned}\ ] ] thus , if @xmath300 we have that @xmath301 at small momenta ( plus logarithmic corrections ) .
clearly , also in @xmath34 , we obtain a one - parameter family of behaviors , labelled by the value of @xmath15 , and the ir - enhanced ghost propagator corresponds to the upper value of @xmath15 allowed by the no - pole condition ( [ eq : nopole ] ) . with the numerical values reported in the second column of table iv of ref .
@xcite and @xmath263 we find @xmath302 \ , p^2 \label{eq : sigmq4d}\ ] ] and the condition @xmath300 corresponds ) and the value of @xmath18 . ] to @xmath303 .
note again the negative sign of the leading order corrections at small momenta ( see section [ sec : d - exact ] ) . as stressed in section [ sec : yukawa ] above , ref .
@xcite has shown that the fit of the gluon - propagator data in the @xmath6 case can be done using the expression @xmath304 where @xmath305 is real , @xmath292 , @xmath291 and @xmath286 are the roots of the quadratic equation , with respect to the variable @xmath68 , obtained by setting equal to zero the denominator of eq .
( [ eq : f2dgluon ] ) .
thus , in order to evaluate the ghost form - factor @xmath0 we need to consider the function @xmath306 , defined in eq .
( [ eq : deffofkwithe ] ) above . to this end
, we can choose again the positive @xmath97 direction parallel to the external momentum @xmath28 and consider polar coordinates .
then , after evaluating the angular integral we find @xmath307 \ ; , \label{eq : deffofkwithesimpl}\ ] ] valid both for @xmath308 and for @xmath309 . in the case
@xmath308 the momentum integration is straightforward giving @xmath310 \\[2 mm ] & = & \lim_{\lambda \to \infty } \ , \frac{1}{4 \pi } \left\ { \ , \frac{1}{2 p^2 } \ , \ln\left(1 + \frac{p^2}{\omega^2}\right ) \ , + \ , \frac{1}{\omega^2 } \ , \left [ \ln\left(\lambda\right ) \ , - \ , \ln\left(p\right ) \ , - \ , \frac{1}{2 } \ln\left(\lambda^2 + \omega^2\right ) \,+ \ , \frac{1}{2 } \ln\left(p^2 + \omega^2\right ) \right ] \ , \right\ } \\[2 mm ] & = & \frac{1}{8 \pi } \left [ \ , \frac{1}{p^2 } \ , \ln\left(1 + \frac{p^2}{\omega^2}\right ) \ , + \ , \frac{1}{\omega^2 } \ , \ln\left(1 + \frac{\omega^2}{p^2}\right ) \ , \right ] \ ; .
\label{eq : fwithe0}\end{aligned}\ ] ] note that the second term above blows up logarithmically in the ir limit @xmath7 , in agreement with the result obtained in section [ sec : d=2sing ] . for @xmath309
the second integral in eq .
( [ eq : deffofkwithesimpl ] ) can be written , after the change of variable @xmath311 , as @xmath312 where @xmath313 is the incomplete beta function , which is defined for @xmath314 @xcite , implying @xmath315 in our case .
for the first integral in eq .
( [ eq : deffofkwithesimpl ] ) we can not use directly the changes of variable @xmath316 and @xmath317 because we get an incomplete beta function ( [ eq : betainc ] ) with @xmath318 . in this case
it is convenient to introduce a feynman parameter ( using non - integer exponents ) and write @xmath319^{2-\eta/2 } } \\[2 mm ] & = & \frac{1}{2 p^2 } \int_0 ^ 1 { \ensuremath{\mathrm{d}}}x \ , x^{-\eta/2 } \ , \biggl\ { \ , - \frac{p^2}{\left[p^2 + ( 1-x ) \omega^2\right]^{1-\eta/2 } } \nonumber \\[2 mm ] & & \qquad \qquad + \ , \frac{2}{\eta } \left[p^2 + ( 1-x ) \omega^2\right]^{\eta/2 } \ , - \ , \frac{2}{\eta } \left[(1-x ) \omega^2\right]^{\eta/2 } \ ,
\biggr\ } \ ; , \end{aligned}\ ] ] where we have also done the integration in @xmath320 .
after suitable changes of variables , the last formula can be written as @xmath321 & & \quad \ , + \ , \frac{p^2+\omega^2}{\eta \ , p^2 \ , ( \omega^2)^{1-\eta/2}}\ , b\left(\frac{\omega^2}{p^2+\omega^2 } ; 1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ , - \ , \frac{1}{\eta \ , p^2 \ , ( \omega^2)^{-\eta/2}}\,b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ ; , \end{aligned}\ ] ] where @xmath322 is the beta function .
thus , by summing the two results above , we find @xmath323 \\[2 mm ] & = & \frac{(\omega^2)^{\eta/2 - 1}}{4 \pi \ , \eta } \ , b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \nonumber \\[2 mm ] & & \quad + \ , \frac{(\omega^2)^{\eta/2}}{4 \pi \ , \eta \
, p^2 } \ , \frac{p^2+\omega^2}{\omega^2 } \ , \left [ b\left(\frac{\omega^2}{p^2+\omega^2 } ; 1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ , - \ , b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \right ] \ ; . \label{eq : fwithe}\end{aligned}\ ] ] note that for @xmath127 the incomplete beta function @xmath324 becomes the beta function @xmath325 .
then , by taylor expanding @xmath326 for small momenta @xmath28 , we obtain @xmath327 \ ; , \label{eq : fat0}\ ] ] yielding a constant contribution at @xmath127 . using the expression ( [ eq : d2d ] ) , the ghost form - factor in the @xmath6 case is given by @xmath328 \label{eq : sigma2d } \ ; , \ ] ] with @xmath270 and @xmath329 defined , respectively , in eqs .
( [ eq : fwithe0 ] ) and ( [ eq : fwithe ] ) .
of course , the function @xmath0 is real . by writing @xmath293 and @xmath294
we get for the first two terms above @xmath330 \right .
\nonumber \\[2 mm ] & & \left .
\qquad \ , + \ ,
\frac{1}{r^2 } \ , \left [ \left(a v + b w\right ) \ell_5(p^2 ) \ , - \ , \left(b v - a w\right ) a_3(p^2 ) \right ] \ ,
\right\ } \ ; , \label{eq : sigmareal2d}\end{aligned}\ ] ] where @xmath331 , @xmath332 , @xmath333 and @xmath334 have already been defined in eqs .
( [ eq : l2 ] ) , ( [ eq : a2 ] ) , ( [ eq : a3 ] ) and ( [ eq : r ] ) and @xmath335 as shown in section [ sec : d=2sing ] , there is a logarithmic singularity @xmath336 at small momenta proportional to the gluon propagator at zero momentum , that is , @xmath337 .
we also have @xmath338 \ ; , \label{eq : imaginaryw}\ ] ] where we have indicated with @xmath339 the imaginary part of the expression in square brackets .
one can easily check that @xmath0 is null at large momenta .
finally , the results ( [ eq : sigmareal2d ] ) and ( [ eq : imaginaryw ] ) , together with the expressions ( [ eq : fat0 ] ) and ( [ eq : sigma2d ] ) , allow us to evaluate the behavior of the ghost propagator at small momenta .
we obtain @xmath340 \ , - \ , \frac{b v - a w}{r^2 } \left [ \ , \arctan\left ( \frac{w}{v } \right ) \ , - \ , \frac{w p^2}{r^2 } \ , \right ] \ , + \ , o \left ( { p}^{4 } \right ) \right\ } \nonumber \\[2 mm ] & & \qquad - 2 \ , c \ , \im \left [ \frac{(\omega_+^2)^{\eta/2 - 1}}{4 \pi \ , \eta } \ , b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ , - \ , \frac{p^{\eta}}{4 \pi \ , \eta \ , ( 1 + \eta/2 ) \ , \omega^2_+ } \ , + \ , o \left ( { p}^{2+\eta } \right ) \right ] \\[2 mm ] & = & \frac{1}{8 \pi } \left\ { \ , \frac{a p^2}{2 r^2 } \ , + \ , \frac{a v + b w}{r^2 } \left [ \ , 1 \ , + \ , \ln\left(\frac{r}{p^2}\right ) \ , \right ] \ , - \ , \frac{b v - a w}{r^2 } \left [ \ , \arctan\left ( \frac{w}{v } \right ) \ , - \ , \frac{w p^2}{r^2 } \ , \right ]
\ , \right\ } \nonumber \\[2 mm ] & & \quad - 2 \ , c \ , \sin\left [ \left(\frac{\eta}{2}-1\right ) \arctan\left(\frac{w}{v}\right ) \right ] \frac{r^{\eta/2 - 1}}{4 \pi \ ,
\eta } \ , b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ , - \ , \frac{2 \ , c \ , w\ , p^{\eta}}{4 \pi \ , \eta \ , ( 1 + \eta/2 ) \ , r^2 } \ , + \ , o \left ( { p}^{2+\eta } \right ) \ ; . \label{eq : sigma2dexp}\end{aligned}\ ] ] note that , if @xmath75 , one finds a ghost propagator with a behavior @xmath341 in the ir limit . as in @xmath194 and in @xmath34
we have a one parameter family of solutions labelled by the value of @xmath15 . as explained in ref .
@xcite , the @xmath6 data for the gluon propagator suggest the relations @xmath342 and @xmath343 , implying @xmath344 and @xmath345 .
then , we find @xmath346 \frac{(2 v^2)^{\eta/2 - 1}}{4 \pi \ , \eta } \
, b\left(1-\frac{\eta}{2 } , 1+\frac{\eta}{2 } \right ) \ , - \ , \frac{c \ , p^{\eta}}{4 \pi \ , \eta \ , ( 1 + \eta/2 ) \ ,
v } \ , - \ , \frac{a p^2}{32 \pi v^2 } \ , + \ , o \left ( { p}^{2+\eta } \right ) \ ; .\ ] ] using the approximate result @xmath347 ( see again ref .
@xcite ) this formula simplifies to @xmath348 on the contrary , for @xmath263 and with the numerical values reported in @xcite see the second column of table xiv and , for the exponent @xmath211 , the last line of table xiii we find for eq.([eq : sigma2dexp ] ) the numerical results @xmath349 the coefficient @xmath350 , multiplying the logarithmic ir singularity , is zero within error and we have omitted the corresponding term . note that @xmath0 decreases for increasing momenta @xmath68 , as proven in section [ sec:2dderivative ] above . also note that we have @xmath254 for @xmath351 and in this case the ghost propagator behaves as @xmath352 in the ir limit .
the one - loop analysis above has shown that , in the @xmath6 case , an ir singularity @xmath150 appears in the gribov form - factor @xmath0 when @xmath110 .
thus , one needs a null gluon propagator at zero momentum in order to satisfy the no - pole condition @xmath16 . on the contrary , for @xmath12 and 4
, we found that @xmath0 is finite also for @xmath191 . in this section
we improve our analysis by considering the dse for the ghost propagator @xmath14 ( see for example @xcite ) .
as stressed in the introduction , here we do not try to solve the ghost propagator dse , but instead we concentrate on general properties of this equation for different space - time dimensions .
in particular , the results obtained in section [ sec : ghost - prop ] are confirmed by considering a generic ( sufficiently regular ) gluon propagator @xmath5 and an ir - finite ghost - gluon vertex @xmath353 . in the @xmath6 landau gauge the dse for the ghost propagator
is written as @xmath354 where @xmath355 , the gluon and the ghost propagators respectively @xmath5 and @xmath14 are full propagators and we indicated with @xmath353 the full ghost - gluon vertex .
the above result implies @xmath356 if one uses eq .
( [ eq : finalg ] ) . for a tree - level ghost - gluon vertex @xmath357 and using the transversality of the gluon propagator we finally find @xmath358 which should be compared to the one - loop result ( [ sigma1 ] ) . as in section [ sec:2dderivative ]
above , we can choose the @xmath97 direction along the external momentum @xmath28 obtaining ( using polar coordinates ) @xmath359 } \ ; , \label{eq : sigmads}\ ] ] with @xmath360 .
this equation will be analyzed below using two different approaches .
a first result can , however , be easily obtained using again the @xmath161-max approximation , as in section [ sec : dapprox ] above .
this gives us @xmath361 } \
, + \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \frac{\mathcal{d}(x)}{x \ , \left [ 1 - \sigma(x ) \right ] } \ , \right\ } \ ; , \ ] ] where we have done the angular integration and set @xmath362 . in the limit of small momenta @xmath68
we then obtain @xmath363 } \
, + \ , \int_{0}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \frac{\mathcal{d}(x)}{x \ , \left [ 1 - \sigma(x ) \right ] } \ , \right\ } \ ; .\ ] ] in order to avoid ir singularities in the above equation we have to impose @xmath364 , i.e. the gluon propagator should be null at zero momentum . in particular ,
if @xmath74 , i.e. for a free - like ghost propagator at small momenta , it is sufficient to have @xmath309 . on the contrary , if the ghost propagator is ir enhanced and @xmath365 for small @xmath97 with @xmath366 , then the condition @xmath367 should be satisfied .
note that the predictions of the scaling solution @xcite , i.e. @xmath368 and @xmath369 , are consistent with the above inequality .
the same results can also be obtained by setting @xmath69 directly in eq .
( [ eq : sigmads ] ) .
this makes the @xmath370 integral trivial and gives @xmath371 } \ ; .
\label{eq : sigma0wrong}\ ] ] note , however , that in both cases we essentially miss the logarithmic ir singularity @xmath70 which is found below . in the first case this is probably related to the very crude @xmath161-max approximation . on the contrary , in eq
( [ eq : sigma0wrong ] ) , this is due to the ( improper ) exchange of the @xmath222 integration with the @xmath110 limit @xcite .
since the gribov form - factor is non - negative , we can easily construct a lower bound for the l.h.s . of eq .
( [ eq : sigmads ] ) by writing @xmath372 & = & i_2(p^2 , \infty ) \ , = \ , \frac{1}{4 \pi } \left [ \ , \int_0^p \frac{{\ensuremath{\mathrm{d}}}q}{p^2 } \ , q \ , \mathcal{d}(q^2 ) \ , + \ , \int_p^{\infty } \frac{{\ensuremath{\mathrm{d}}}q}{q } \ , \mathcal{d}(q^2 ) \ , \right ] \ ; , \label{eq:2dbound2}\end{aligned}\ ] ] where we use the definitions ( [ eq : igeneral - dell ] ) , ( [ eq : id - ell ] ) and the relations ( [ eq : i - upper - nu1 ] ) . the last integral in the above equation has already been analyzed in section [ sec : d=2sing ] , where it was shown that @xmath373 develops an ir singularity proportional to @xmath374 if @xmath375 .
thus , @xmath0 also is ir singular , unless @xmath122 .
one can also find an upper bound for @xmath0 and check that the ir singularity is indeed only logarithmic . to this end
we can notice that , if @xmath74 , one can write case and in the one - loop approximation , the function @xmath0 is decreasing as @xmath68 increases , i.e. the maximum value of @xmath0 is obtained for @xmath121 ( see section [ sec:2dderivative ] ) .
however , the proof presented here can be easily modified for the case when @xmath9 for all momenta @xmath28 and the maximum value of @xmath0 is not attained at @xmath127 .
finally , one should note that in the dse ( [ eq : sigma2 ] ) one uses explicitly eq .
( [ eq : finalg ] ) . thus , when estimating the integral in eq .
( [ eq : sigmads ] ) , we can not simply impose @xmath376 but we have to consider the stronger condition @xmath377 . ]
@xmath378 } \ , = \
, \frac{i_2(p^2 , \infty)}{1 - \sigma(0 ) } \ ; , \ ] ] where we have also used eqs .
( [ eq:2dbound1 ] ) and ( [ eq:2dbound2 ] ) above .
therefore , the upper bound also blows up as @xmath374 in the ir limit . at the same time , if @xmath254 , with @xmath379 at small momenta we find @xmath380 & & \qquad \quad + \ , \int_0^{\infty } \frac{q \ , { \ensuremath{\mathrm{d}}}q}{4 \pi^2 } \ , \mathcal{d}(q^2 ) \ , \int_0^{2 \pi } { \ensuremath{\mathrm{d}}}\theta \ , \frac{1 - \cos^2(\theta)}{s^2 } \ , \left [ \ , \frac{1}{1 - \sigma(s^2 ) } \ , - \ , \frac{1}{c \ ,
s^{2\kappa } } \ , \right ] \ ; .\end{aligned}\ ] ] note that the quantity in square brackets in the last integral is finite at @xmath381 if the behavior of @xmath0 is given by @xmath382 with @xmath383 .
moreover , this quantity goes to 1 at large momenta and its absolute value is clearly bounded from above by some positive constant @xmath384 if @xmath385 $ ] .
hence , we have @xmath386 for @xmath387 we can also use the upper bound in eq .
( [ eq : igeneral - ell - upper ] ) and write @xmath388 where @xmath389 is a positive constant .
thus , we have again an ir singularity proportional to @xmath374 if @xmath60 is not zero .
we conclude that @xmath0 can be finite solely if @xmath122 .
let us remark that the only hypothesis considered in this case is the ir expansion @xmath390 with @xmath391 and @xmath383 .
also note that the @xmath6 lattice data @xcite show for the ghost propagator an ir behavior in good agreement with the so - called scaling solution @xcite that predicts @xmath369 .
thus , the condition @xmath391 is verified in both cases .
one can also note that , by considering in eq.([eq : sigmafull ] ) the full ghost - gluon vertex @xmath392 , instead of the tree - level one @xmath393 , the above results still applies for an ir - finite vertex .
this hypothesis is usually adopted in dse studies of gluon and ghost propagators @xcite and it is confirmed by lattice data @xcite . in this section
we analyze the dse in an alternative way , also avoiding the @xmath161-max approximation .
to this end , let us first consider the @xmath370-integral using contour integration . after setting @xmath394 we find @xmath395 } \
, = \ , \frac{i}{4 } \oint { \ensuremath{\mathrm{d}}}z \ , \frac{(z^2 - 1)^2}{z^2 \ , ( -q+kz ) \ , ( k - qz ) } \ , \frac{1}{1 \,- \ , \sigma\left [ \ , ( -q+kz ) \ , ( k - qz)z^{-1 } \ , \right ] } \ ; , \ ] ] where the integral @xmath101 is again taken on the unit circle @xmath102 . clearly , besides the poles at @xmath396 and at @xmath397 in the first denominator on the r.h.s . of the above equation
, one has to consider possible divergences in the function @xmath398 } \ ; .\ ] ] in particular , if we assume ghost enhancement , i.e. @xmath254 , then @xmath399 is divergent at @xmath400 and at @xmath401 .
note that these divergences are not necessarily poles of the function @xmath399 .
indeed , @xmath399 could display a branch cut in the unit disc or one passing through it .
for example , the usual @xmath4 dse scaling solution has @xmath402 in the limit @xmath403 , where @xmath404 is a fractional number .
this behavior signals a non - analyticity for @xmath405 at the origin and implies a non - analyticity for the function @xmath399 at @xmath401 or at @xmath400 .
also , since the ghost is `` massless '' we should expect that the ghost propagator develops a branch cut along the real axis for @xmath406 .
then , @xmath400 or @xmath401 would correspond to branch points of the function @xmath399 , making quite difficult the evaluation of the contour integral in the above expression . in order to overcome this problem
, we make the hypothesis that a spectral representation for the ghost propagator can be introduced , i.e. we write case , the theory should be uv finite and we do not need to consider renormalization factors here . ]
@xmath407 which reproduces the branch cut in @xmath405 for @xmath408 ( see for example @xcite ) .
if we assume @xmath409 and write @xmath410 it is clear that the spectral density @xmath411 must satisfy the normalization condition @xmath412 also note that the tree - level ghost propagator @xmath413 corresponds to the spectral density @xmath414 , where @xmath415 is the dirac delta function .
this case will be used below to recover results obtained in the one - loop analysis carried on in sections [ sec:2dderivative ] and [ sec : d=2sing ] . in the general case ,
the spectral density @xmath411 is proportional to the discontinuity of the ghost propagator along the cut .
has a branch cut along a curve @xmath416 in the complex plane and if it goes to zero sufficiently fast at infinity , using cauchy s theorem we could write down an integral relation similar to eq . with the variable @xmath417 running over the curve @xmath418 , with @xmath419 .
also , possible poles can be included by adding @xmath420-functions to the spectral density @xmath411 or , equivalently , by pulling the pole terms out of the spectral integral . ]
considering eqs .
( [ eq : sigmads ] ) and ( [ contextra10 ] ) we can write @xmath421 } & = & \frac{i}{4 } \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \oint { \ensuremath{\mathrm{d}}}z \ , \frac{z^2 \ , + \ , \overline{z}^2 \ , - \ , 2}{-p q z^2 \ , + \ , \left(p^2 + q^2 + t\right ) z - p q } \\[2 mm ] & = & - \frac{i}{4 p q } \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \oint \frac{{\ensuremath{\mathrm{d}}}z}{z^2 } \ , \frac{(z^2 - 1)^2}{z^2 \ , - \ , \left(p^2 + q^2 + t\right )
z / \left(p q\right ) + 1 } \ ; , \label{contextra11}\end{aligned}\ ] ] where we indicated with @xmath103 the complex - conjugate of @xmath104 . thus ,
using the representation we can avoid dealing directly with the integral of an unknown function along the branch cut . in exchange
, we have in our formulae an extra integration of the ( also unknown ) spectral density @xmath411 .
nevertheless , as we will see below , the above equation will allow us to control the @xmath110 limit [ at least in the case @xmath422 .
to this end , let us first note that in the contour integral ( [ contextra11 ] ) there is a double pole at @xmath423 and there are single poles at @xmath424 since @xmath425 we have that @xmath426 and one can check that the pole @xmath427 lies within the unit disc while @xmath428 lies outside of it .
moreover , for @xmath429 ( which implies @xmath430 and @xmath431 ) the two poles coincide and we have @xmath432 .
it is also easy to check that the residues , inside the unit circle , for the @xmath187-integrand are @xmath433 \mathcal{r}\text{es}_{z = z_{- } } & = & \frac{\sqrt{\left[(p+q)^2+t\right ] \
, \left[(p - q)^2+t\right ] } } { p^2 q^2 } \ ; .\end{aligned}\ ] ] then , using the residue theorem , we find @xmath434 } \ , = \
, \frac{\pi}{2 } \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \frac{p^2+q^2+t \ , - \ , \sqrt{\left[(p+q)^2+t\right ] \ , \left[(p - q)^2+t\right]}}{p^2 q^2}\ ] ] and we can write the ghost dse ( [ eq : sigmads ] ) as @xmath435 \ , \left[(p - q)^2+t\right]}}{p^2 q^2 } \\[2 mm ] & = & \int_0^{\infty } \frac{{\ensuremath{\mathrm{d}}}x}{16 \pi } \ , \mathcal{d}(x ) \ , \int_0^\infty
\ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \frac{p^2+x+t \ , - \ , \sqrt { t^2 \ , + \ , 2 t \left(p^2 + x\right ) \ , + \ , \left(p^2 - x\right)^2 } } { p^2 x } \label{contextra13 } \ ; .\end{aligned}\ ] ] note that , for @xmath414 and using @xmath436 \ , \left[(p - q)^2\right ] } = \left\ { \begin{array}{ll } p^2 \ , - \ , q^2 \qquad & \mbox{if } \quad p^2 > q^2 \\[3 mm ] q^2 \ , - \ , p^2 \qquad & \mbox{if } \quad q^2 >
p^2 \end{array } \right.\ ] ] we find from eq .
( [ contextra12 ] ) the one - loop result ( [ 2 ] ) . also note that , by taylor expanding the integrand at @xmath69 , one finds @xmath437 } \ ; , \ ] ] where we used the definition ( [ contextra10 ] ) . as shown above
[ see eq .
( [ eq : sigma0wrong ] ) ] , this result can also be obtained immediately by setting @xmath69 in eq . .
however , as already pointed out below eq .
( [ eq : sigma0wrong ] ) and in ref .
@xcite , one should not exchange the @xmath222 integration and the @xmath110 limit . therefore , in order to properly evaluate @xmath0 for small momenta @xmath68 , we write eq .
( [ contextra13 ] ) as @xmath438 & & \quad \ , + \ , \lim_{p^2 \to 0 } \ , \int_{p^2}^{\infty } \frac{{\ensuremath{\mathrm{d}}}x}{16 \pi } \ , \mathcal{d}(x ) \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \frac{p^2+x+t \ , - \ , \sqrt { t^2 \ , + \ , 2 t \left(p^2 + x\right ) \ , + \ , \left(p^2 - x\right)^2 } } { p^2 x } \ ; .
\label{nieuw1}\end{aligned}\ ] ] the first integral can be estimated using the the trapezoidal rule .
we then obtain @xmath439 & = & \lim_{p^2 \to 0 } \ , \frac{p^2}{32 \pi } \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \left [ \ , \mathcal{d}(p^2 ) \
, \frac{2 p^2+t \ , - \ , \sqrt { t^2 \ , + \ , 4 t p^2 } } { p^4 } \ , + \ , \frac{2 \
, \mathcal{d}(0)}{p^2+t } \ , \right ] \\[2 mm ] & = & \lim_{p^2 \to 0 } \ , \frac{1}{32 \pi } \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \left [ \ , \mathcal{d}(p^2 ) \ , \frac{2 p^2+t \ , - \ , \sqrt { t^2 \ , + \ , 4 t p^2 } } { p^2 } \ , \right ] \ , + \ , \lim_{p^2 \to 0 } \frac{\mathcal{d}(0)}{16 \pi \left [ \ , 1-\sigma(p^2 ) \right ] } \ ; , \label{nieuw2}\end{aligned}\ ] ] where we used again eq .
( [ contextra10 ] ) . for the second integral
we define @xmath440 and find @xmath441 & = & - \ln(p^2 ) \ , \frac{\mathcal{d}(p^2)}{16 \pi } \ , \int_0^\infty { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \frac{2 p^2+t \ , - \ , \sqrt { t^2 \ , + \ , 4 t p^2 } } { p^2 } \ , - \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln(x ) \ , \mathcal{g}'(x , p^2 ) \ ; , \label{nieuw3}\end{aligned}\ ] ] where @xmath120 refers to the derivative w.r.t .
the @xmath97 variable and we used the fact that @xmath136 goes to zero at large @xmath97 . note that , in the one - loop case @xmath414 , we have @xmath442 and eq .
( [ nieuw3 ] ) becomes @xmath443 in agreement with eqs .
( [ eq : sigma-1l - ini ] ) and ( [ form2 ] ) . by collecting the above results we can therefore write @xmath444 \ , \int_0^\infty \ , { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \frac{2 p^2+t \ , - \ , \sqrt { t^2 \ , + \ , 4 t p^2 } } { p^2 } \nonumber \\[2 mm ] & & \qquad \qquad \qquad \qquad - \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln(x ) \ , \mathcal{g}'(x , p^2 ) \ , \biggr\ } \ ; . \label{nieuw5b}\end{aligned}\ ] ] we can now verify that the last integral in the above expression is finite . indeed , we have @xmath445 & + & \frac{\mathcal{d}(x)}{16 \pi } \ , \int_0^\infty { \ensuremath{\mathrm{d}}}t \ , \rho(t ) \ , \left [ \ , \frac{1}{p^2 } \ , - \ , \frac{x + t - p^2 } { p^2 \ , \sqrt { t^2 \ , + \ , 2 t \left(p^2 + x\right ) \ , + \ , \left(p^2 - x\right)^2 } } \ , \right ] \ ; .\end{aligned}\ ] ] then , for large @xmath97 we find @xmath446 \ ; , \ ] ] where we used the normalization condition ( [ eq : normalization ] ) , while for small @xmath97 we have @xmath447 \ ; , \ ] ] where we used the definition ( [ eq : spectral ] ) .
thus , the integral @xmath448 has no ir and uv singularities if @xmath136 and @xmath155 goes to zero sufficiently fast when @xmath97 goes to zero and to infinity . at the same time we need the integral @xmath449 to be finite .
we can also check that the integral @xmath450 is finite and nonzero if @xmath411 is non - negative .
that is negative for some values of @xmath417 .
however , they apply if one can explicitly verify that the integral in eq .
( [ eq : integral ] ) is indeed finite and nonzero .
also note that we can not simply consider the limit @xmath7 of the integral ( [ eq : integral ] ) , since the factor multiplying @xmath411 vanishes in this limit and we might erroneously conclude that the above expression is equal to zero .
indeed , already in the tree - level case , i.e. for @xmath451 , one finds that the integral ( [ eq : integral ] ) is non - zero and equal to 2 .
finally , one could also formally expand the integrand in powers of @xmath68 leading to @xmath452 where @xmath453 is the usual binomial coefficient .
however , this series is not converging for all values of @xmath454 .
moreover , for @xmath455 we have a term proportional to @xmath456 and this integral is divergent . indeed , by ( formally ) setting @xmath121 in eq .
we find @xmath457 ] to this end let us first note that the numerator @xmath458 is non - negative since @xmath459 when @xmath460 . moreover , if we define @xmath461 it is clear that @xmath462 , since the quantity @xmath463 is negative for @xmath464 .
this implies @xmath465 where we used again the normalization condition ( [ eq : normalization ] ) . at the same time we can write @xmath466 where we use the fact that the function @xmath467 is positive and gets its minimum value , equal to @xmath468 , for @xmath469 .
then , using the definition ( [ eq : spectral ] ) and the condition @xmath470 , we can write @xmath471 } \ , \geq \ , \frac{3}{2 } \ ; .\ ] ] from the above results we conclude that in eq .
( [ nieuw5b ] ) we have two possible ir singularities , i.e. the term @xmath472 $ ] , if @xmath75 , and the term proportional to @xmath473 . in both cases we need to impose the condition @xmath122 in order to avoid the singularity .
thus , we find again that a massive gluon propagator in the @xmath4 case is not compatible with the restriction of the functional integration to the first gribov region . in the @xmath194 case we expect no uv divergences when using dimensional regularization above at one - loop level , the evaluation of the ghost propagator in @xmath194 usually involves gamma functions with half - integer arguments , which do not generate infinities . indeed , for nonnegative values of @xmath474 with @xmath474 integer , one has @xcite @xmath475 and @xmath476 , where @xmath477 denotes the double factorial . ] and the dse for the gribov form - factor is simply @xmath478 & = & \int_0^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , \frac{q^2}{(2\pi)^3 } \ , \mathcal{d}(q^2 ) \ , \int \ , { \ensuremath{\mathrm{d}}}\omega_3 \ , \frac{1 - \cos^2(\phi_1)}{s^2 \ , \left [ \ , 1 \ , - \ , \sigma(s^2 ) \ , \right ] } \ ; , \label{eq : dse3d}\end{aligned}\ ] ] where we used the tree - level ghost - gluon vertex @xmath479 and @xmath480 .
we can now work as in section [ sec:2dbounds ] and use the results of appendix [ sec : hyper ] . in this way we obtain the upper bounds
@xmath481 if @xmath482 , and @xmath483 if @xmath484 with @xmath485 for small @xmath68 . in the latter case we also need the conditions @xmath486 and @xmath383 . as we saw in eq .
( [ eq : limiti ] ) , under simple assumptions for the gluon propagator @xmath95 , the integral @xmath487 is finite in the ir limit @xmath7 .
thus , in both cases the upper bound of @xmath0 is also finite is nonnegative and eqs.([eq : igeneral - dell ] ) and ( [ eq : i - upper - nu1 ] ) , the lower bound @xmath488 clearly applies .
however , since @xmath0 is finite , this bound does not add any relevant information to our analysis .
note that this observation can be made also for the @xmath34 case described in section [ sec:4ddse ] . ] and , in order to have a finite value for @xmath158 in the @xmath194 case we do not need to set @xmath122 .
this result also applies when an ir - finite ghost - gluon vertex is included in the ghost dse ( [ eq : dse3dfull ] ) .
let us also note that the scaling solution predicts in the 3d case @xcite a value @xmath489 for which the condition @xmath490 is satisfied . in @xmath34 the dse for @xmath0
is given by ( see for example @xcite ) @xmath491 } \ ; , \ ] ] where @xmath492 and @xmath493 are the renormalization constants for the ghost propagator and the ghost - gluon vertex respectively .
] and @xmath480 . in order to eliminate these constants from the expression for
@xmath0 we can subtract . ] the same equation for some fixed value @xmath494 and set @xmath495 , using the non - renormalization of the ghost - gluon vertex in landau gauge @xcite .
this gives @xmath496 } \nonumber \\[2 mm ] & & \qquad \qquad \ ; - \ , \int_0^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , \frac{q^3}{(2\pi)^4 } \ , \mathcal{d}(q^2 ) \ , \int \ , { \ensuremath{\mathrm{d}}}\omega_4 \ , \frac{1 - \cos^2(\phi_1 ) } { t^2 \ , \left [ \ , 1 \ , - \ , \sigma(t^2 ) \ , \right ] } \ ; , \label{eq : dse4dbis}\end{aligned}\ ] ] with @xmath497 .
clearly , considering @xmath498 at large momenta , the uv ( logarithmic ) divergence of the two integrals cancels out . in order to derive upper bounds for the above expression without spoiling the cancellation of uv divergences and
since we are interested in the ir limit @xmath7 , we can consider a momentum scale @xmath499 sufficiently large and write @xmath500 } \nonumber \\[2 mm ] & & \;\ ; + \ , \int_{\ell}^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , \frac{q^3}{(2\pi)^4 } \ , \mathcal{d}(q^2 ) \ , \int \ , { \ensuremath{\mathrm{d}}}\omega_4 \ , \frac{1 - \cos^2(\phi_1 ) } { s^2 \ , \left [ \ , 1 \ , - \ , \sigma(s^2 ) \ , \right ] } \ , - \ , \int_0^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , \frac{q^3}{(2\pi)^4 } \ , \mathcal{d}(q^2 ) \ , \int \ , { \ensuremath{\mathrm{d}}}\omega_4 \ , \frac{1 - \cos^2(\phi_1 ) } { t^2 \ , \left [ \ , 1 \ , - \ , \sigma(t^2 ) \ , \right ] } \ ; . \label{eq : dse4dter}\end{aligned}\ ] ] for small momenta @xmath28 only the first integral on the r.h.s . of
the above equation can produce an ir singularity .
following the analysis presented in the 3d case above we can then write @xmath501 } \
, \leq \ , \frac{i_4(p^2 , \ell)}{1 \ , - \ , \sigma(0 ) } \ ; , \ ] ] if @xmath482 , and @xmath501 } \ , \leq \ , \left ( \ , \frac { m '' } { c } \ , + \ , m \ , \right ) \ , i_4(p^2 , \ell ) \ ; , \ ] ] if @xmath502 with @xmath383 and @xmath503 .
again , thanks to the result ( [ eq : limiti ] ) , both upper bounds are finite in the ir limit @xmath7 also for @xmath13 .
an alternative proof can be given by working directly with eq .
( [ eq : dse4d ] ) and using dimensional regularization , i.e. considering a dimension @xmath271 . in this case
we can write @xmath504 with and @xmath505 , the integral in eq.([eq : sigmad ] ) is no longer dimensionless . to keep the dimensionality correct we should , as always , scale out a dimensional factor @xmath506 where @xmath507 is a mass scale , which could then be combined with the coupling constant @xmath15 , making @xmath0 dimensionless also for @xmath505 .
this is important when evaluating the @xmath279 expansion in order to single out uv divergencies .
since here we are mainly interested in the ir behavior of @xmath0 , we do not keep track explicitly of all the terms depending on @xmath279 and we simply consider the coupling @xmath15 dimensionful . ]
@xmath508 } \ ; . \label{eq : sigmad}\ ] ] then , if we can show that no ir singularities occur for @xmath509 , the uv infinity that appears for @xmath510 is taken care of by the renormalization factors . in order to show that @xmath511 is ir finite we can work as done above and write @xmath512 or @xmath513 depending on the value of @xmath514 . in the latter case we considered again the ir expansion @xmath515 and the conditions @xmath383 and @xmath503 .
we conclude that also in the @xmath34 case , @xmath158 is finite if @xmath60 is also finite ( but not necessarily null ) .
let us remark that the ir exponent usually obtained in the scaling solution @xcite is @xmath516 in the @xmath34 case , i.e. the condition @xmath517 is satisfied .
also note that when @xmath518 the hypergeometric function @xmath519 is not convergent at @xmath520 and we can not use the above proofs in order to derive properties of the gribov form - factor .
however , these large values of @xmath521 imply for the ghost propagator @xmath14 a very strong ir enhancement with a behavior at least as singular as @xmath522 in @xmath34 and at least as singular as @xmath523 for @xmath12 .
summarizing , in this manuscript we have considered general properties of the landau - gauge gribov ghost form - factor @xmath0 for su(@xmath1 ) euclidean yang - mills theories in @xmath71 space - time dimensions .
this form - factor is in a one - to - one correspondence with the ghost propagator @xmath14 via eq .
( [ eq : gsigma ] ) . also , @xmath0 is bounded by 1 if the no - pole condition ( [ eq : nopole ] ) is imposed , i.e. if one restricts the functional integration to the first gribov region @xmath42 . the main result of this work is an exact proof of qualitatively different behavior of @xmath0 for @xmath3 with respect to @xmath4 . in particular , for @xmath4 ,
the gluon propagator @xmath5 needs to vanish at zero momentum in order to avoid in @xmath0 an ir singularity proportional to @xmath150 . on the contrary , for @xmath12 and 4 ,
an ir - finite ghost form - factor @xmath0 is obtained also when @xmath13 .
these results were proven , in section [ sec : ghost - prop ] , using perturbation theory at one loop and , in section [ sec : dse ] , by considering the dse for the ghost propagator .
let us stress again that in dse studies of correlation functions in minimal landau gauge , besides using the no - pole condition , a specific boundary condition is usually imposed on the gribov ghost form - factor at zero momentum . here , instead
, we have tried to prove general properties of the gribov ghost form - factor @xmath0 when the restriction to the first gribov horizon is considered . at the same time , in section [ sec : sigma234 ] , we have presented closed analytic expressions for the gribov form - factor @xmath0 at one loop , considering for the gluon propagator linear combinations of yukawa - like propagators ( with real and/or complex - conjugate poles ) .
these functional forms , briefly described in eqs .
( [ eq : f2dgluon])([eq : f4dgluon ] ) , were recently used to fit lattice data of the gluon propagator in the su(2 ) case @xcite .
the expressions obtained for @xmath0 confirm the results presented in section [ sec : ghost - prop ] .
these expressions also show that , for the ghost propagator @xmath14 , there is a one - parameter family of behaviors @xcite labelled by the coupling constant @xmath15 , when it is considered as a free parameter .
the no - pole condition @xmath16 then implies @xmath17 , where @xmath18 is a `` critical '' value . for @xmath15
smaller than @xmath18 one has @xmath74 and the ghost propagator is a massive one . on the contrary , at the `` critical '' value @xmath18 , i.e. for @xmath75 , one finds an ir - enhanced ghost propagator .
as stressed in the introduction , the physical value of the coupling is expected to select the actual value of @xmath158 .
present results @xcite give @xmath74 in the four - dimensional su(3 ) case .
our findings imply that a massive gluon propagator can not be obtained in the two - dimensional case , in disagreement with some of the results presented in ref .
@xcite ( see their table 2 ) .
case was also obtained in ref .
@xcite . on the other hand
, the author of @xcite stressed that a full understanding of the @xmath6 case would require a more detailed investigation . ] a possible massive behavior for the gluon propagator in the @xmath6 case was also explicitly conjectured in ref .
@xcite as a gribov - copy effect .
however , since our @xmath6 result is valid for any gribov copy inside the first gribov region , we have shown that , at least for the @xmath6 gluon propagator @xmath5 in the minimal landau gauge , gribov - copy effects do not alter our conclusion for the value of the gluon propagator at zero momentum , i.e. @xmath60 must vanish .
this observation also represents an explicit counterexample to the identification of the one - parameter family of solutions for the gluon and ghost dses @xcite with different gribov copies , as suggested in @xcite .
d. dudal and n. vandersickel are supported by the research - foundation flanders ( fwo ) .
a. cucchieri acknowledges financial support from the special research fund of ghent university ( bof ugent ) and thanks cnpq and fapesp for partial support .
we thank t. mendes , j. rodriguez - quintero and s. p. sorella for fruitful comments .
following appendix b in ref .
@xcite one can easily perform , using hyperspherical coordinates , the angular integrations necessary for our calculations . to this end
let us recall that , in @xmath2 dimensions , one has the following relations between cartesian coordinates @xmath524 ( with @xmath525 ) and hyperspherical coordinates @xmath526 ( with @xmath527 ) : @xmath528 the hyperspherical coordinates take values , respectively , @xmath529 , @xmath530 $ ] for @xmath531 and @xmath532 . at the same time , the volume measure is given by @xmath533 with @xmath534 note that , in the usual three - dimensional case , this notation correspond to @xmath535 , @xmath536 and @xmath537 .
here we want to evaluate the integral @xmath538 if we indicate with @xmath539 the well - known result @xmath540 where @xmath541 is the gamma function , then we have @xmath542 where all other angular integrations have already been evaluated .
the integral in the denominator can be written as @xmath543 where @xmath544 is the beta function . in our calculations
we consider two different functions @xmath545 , i.e.@xmath546 and @xmath547^{\nu } } \ ; .\ ] ] in the first case the integration gives @xmath548 \ , { \ensuremath{\mathrm{d}}}\omega_d \ , = \ , \frac{\omega_d}{b\left(\frac{d-1}{2 } , \frac{1}{2}\right ) } \ , \int_0^{\pi } \ , \sin^{d-2}(\phi_1 ) \ , \left [ 1 - \cos^2(\phi_1 ) \right ] \ , { \ensuremath{\mathrm{d}}}\phi_1\ ] ] and the integral in the numerator is @xmath549 collecting these results we find @xmath548 \ , { \ensuremath{\mathrm{d}}}\omega_d \ , = \ , \omega_d \ , \frac{b\left(\frac{d+1}{2 } , \frac{1}{2}\right)}{b\left(\frac{d-1}{2 } , \frac{1}{2}\right ) } \ , = \ , \omega_d \ , \frac{d-1}{d } \ ; , \label{eq : firstint}\ ] ] where we used @xmath550 . in the second case ,
i.e. when considering the integral @xmath551^{\nu } } \ , { \ensuremath{\mathrm{d}}}\omega_d \ ; , \ ] ] we have @xmath552^{\nu } } \ , { \ensuremath{\mathrm{d}}}\phi_1 \ ; .\ ] ] we can now use the result ( see for example formula @xmath553 in @xcite ) @xmath554^{\nu } } \ , { \ensuremath{\mathrm{d}}}\theta \ , = \ , b(\mu , 1/2 ) \ ; _ 2f_1\left(\nu , \nu -\mu+1/2 ; \mu+1/2 ;
a^2\right ) \ ; , \ ] ] which is valid for @xmath555 and @xmath556 . here
@xmath557 is the gauss hypergeometric function ( see appendix [ sec : hyper ] ) .
then , we find @xmath558^{\nu } } \ , { \ensuremath{\mathrm{d}}}\omega_d & = & \frac{\omega_d}{b\left(\frac{d-1}{2 } , \frac{1}{2}\right ) } \ , \frac{1}{q^2 } \ , b\left(\frac{d+1}{2 } , \frac{1}{2}\right ) \ ; _ 2f_1\left(\nu , \nu - d/2 ; 1+d/2 ; p^2/q^2\right ) \\[2 mm ] & = & \frac{\omega_d}{q^2 } \ ,
\frac{d-1}{d } \ ; _ 2f_1\left(\nu , \nu - d/2 ; 1+d/2 ; p^2/q^2\right ) \label{eq:2f1-uno}\end{aligned}\ ] ] if @xmath167 and @xmath559^{\nu } } \ , { \ensuremath{\mathrm{d}}}\omega_d \ , = \ , \frac{\omega_d}{p^2 } \ , \frac{d-1}{d } \ ; _ 2f_1\left(\nu , \nu - d/2 ; 1+d/2 ; q^2/p^2\right ) \label{eq:2f1-due}\ ] ] if @xmath165 .
let us recall that the gauss hypergeometric function @xmath557 is defined @xcite for @xmath560 by the series @xmath561 where @xmath562 is a so - called pochhammer symbol .
this series is converging for @xmath563 .
it is also converging for @xmath102 if @xmath564 , where @xmath565 indicates the real part . when this condition is satisfied one can use , for @xmath520 , the result ( see formula 9.122.1 in ref .
@xcite ) @xmath566 in eqs .
( [ eq:2f1-uno ] ) and ( [ eq:2f1-due ] ) the hypergeometric function appears with @xmath567 .
therefore , the corresponding series is converging inside the unit circle for any dimension @xmath568 . at the same time we have @xmath569 and
@xmath570 is finite on the unit circle for @xmath571 .
also , using the relation ( see equation 9.131.1 in ref .
@xcite ) @xmath572 we can write @xmath573 and for @xmath574 we have that @xmath575 , i.e. the hypergeometric function @xmath570 is clearly positive for @xmath576 $ ] . in the case @xmath181 the above results simplify .
in particular , we have convergence of the series on the unit circle for any dimension @xmath61 .
then eq .
( [ eq : gausstheorem ] ) yields @xmath577 which is positive . note that for @xmath4 the above result gives @xmath578 .
one can actually check that , when @xmath4 , the function @xmath579 is simply equal to 1 for any value @xmath580 .
indeed , from the series representation ( [ eq : serie ] ) it is obvious that @xmath581 for @xmath582 ( and/or for @xmath583 ) .
the same result can be obtained by considering eq .
( [ eq:2f1sing ] ) and the relation ( see equation 9.121.1 in ref .
@xcite ) @xmath584 yielding @xmath585 the hypergeometric function @xmath586 is actually very simple for any even dimension @xmath587 , with @xmath588 .
indeed , in this case we are considering the function @xmath589 and the coefficient @xmath590 is either zero or negative .
this implies that the series ( [ eq : serie ] ) is actually a polynomial in @xmath187 .
for example , for @xmath66 we have the very simple expression @xmath591 note that it is also possible to find a closed form for the hypergeometric function @xmath586 in the @xmath194 case .
indeed , by considering the relations ( see equations 9.137.8 and 9.137.1 in ref .
@xcite ) @xmath592 0 \ , & = & \ , _ 2f_1(a , b;c;z ) \ , c \ , \left [ \ , c - 1 - \left(2 c - a - b - 1\right ) \ , z \ ,
\right ] \ , + \ , _ 2f_1(a , b;c+1;z ) \ , \left(c - a\right ) \ , \left(c - b\right ) \ , z \nonumber \\[2 mm ] & & \qquad \qquad \qquad \qquad \qquad \qquad \qquad + \ , _ 2f_1(a , b;c-1;z ) \ , c \ , \left(c - 1\right ) \ , \left(z - 1\right)\end{aligned}\ ] ] and ( see equation 9.131.1 in @xcite ) @xmath593 we can write @xmath594 & = & \frac{3}{4 } \ , + \ , \frac{1 - z}{4 } \ , \frac{3 \ , \left(1 - z\right)}{2 z } \ , \left [ \ , _ 2f_1\left(1,1/2;1/2;z\right ) \ , - \ , _ 2f_1\left(1,1/2;3/2;z\right ) \
, \right ] \\[2 mm ] & = & \frac{3}{4 } \ , + \ , \frac{3 \ , \left(1 - z\right)}{8 z } \ , \left [ \ , 1 \ , - \ , \left(1 - z\right ) \ , _ 2f_1\left(1,1/2;3/2;z\right ) \ , \right ] \\[2 mm ] & = & \frac{3}{4 } \ , + \ , \frac{3 \ , \left(1 - z\right)}{8 z } \ , \left [ \ , 1 \ , - \ , \sqrt{1 - z } \ ; _ 2f_1\left(1/2,1/2;3/2;\frac{z}{z-1}\right ) \ , \right ] \\[2 mm ] & = & \frac{3}{4 } \ , + \ , \frac{3 \ , \left(1 - z\right)}{8 z } \ , \left [ \ , 1 \,-\ , \frac{1 - z}{\sqrt{z } } \ , \text{arcsinh}\left(\sqrt{\frac{z}{1 - z } } \ , \right ) \ , \right ] \ ; , \label{eq:2f1-d3}\end{aligned}\ ] ] where we have also used eq .
( [ eq:2f1power ] ) , the relations @xmath595 and ( see 9.121.27 in @xcite ) @xmath596 from the expression ( [ eq:2f1-d3 ] ) it is easy to check that @xmath597 and that @xmath598 , as expected . using the above series ( [ eq : serie ] )
one can verify that the derivative of @xmath557 , with respect to the variable @xmath187 , is given by @xmath599 thus , in the case of interest for us , we have @xmath600 these results can be written as @xmath601 where we made use of eq .
( [ eq:2f1sing ] ) .
when the hypergeometric function @xmath602 is finite in the unit circle , i.e. for @xmath571 , it is clear that this derivative is positive for @xmath603 and @xmath602 attains its maximum value at @xmath520 , equal to @xmath604 and its minimum value , equal to 1 , at @xmath423 . on the contrary , the same derivative is negative when @xmath605 . in this case
@xmath602 is largest at @xmath423 , with @xmath606 , and smallest at @xmath520 with a value given by eq .
( [ eq:2f1max ] ) . finally , for @xmath607 the derivative is null and the hypergeometric function @xmath602 is equal to 1 for all values of @xmath576 $ ] .
thus , for @xmath571 we can always write @xmath608 , for some positive constants @xmath609 and @xmath610 and with @xmath187 taking values in the interval @xmath611 $ ] . for @xmath612 these results again simplify , yielding @xmath613 as expected , for @xmath4 this derivative is zero since @xmath614 .
the result above also simplifies for @xmath66 , for which we find on the right - hand side of eq .
( [ eq : derivativef ] ) the value @xmath615 , as already known from eq .
( [ eq:2f1-d4 ] ) . at the same time , for @xmath12 we can write @xmath616 and the derivative is clearly negative for any value @xmath617 . the same result can actually be proven for any dimension @xmath2 larger than 2 .
indeed , the hypergeometric function @xmath618 is finite in the unit circle for @xmath171 and using eq .
( [ eq:2f1sing ] ) we can easily verify that it is also positive for @xmath576 $ ] .
thus , the derivative in eq .
( [ eq : derivativef ] ) is negative for @xmath171 ( and @xmath576 $ ] ) . as a consequence , under the same hypotheses
, we have that the hypergeometric function @xmath619 has its maximum value , equal to 1 , at @xmath423 , and its minimum value , equal to @xmath620 , at @xmath520 [ see eq .
( [ eq : gauss ] ) ] . using the definition ( [ eq : a - n ] ) of the pochhammer symbol @xmath621
it is also easy to verify that @xmath622 which implies @xmath623 \ ; .\ ] ] then , using eq .
( [ eq : serie ] ) , one can prove the relation @xcite @xmath624 & = & a \ , \left [ \ , \sum_{n=0}^{\infty } \ , \frac{(a+1)_n \ , ( b)_n}{(c)_n } \ , \frac{z^n}{n ! } \ , - \ , \sum_{n=0}^{\infty } \ , \frac{(a)_n \ , ( b)_n}{(c)_n } \
, \frac{z^n}{n ! } \ , \right ] \\[2 mm ] & = & a \ , \left [ \ , _
2f_1(a+1,b;c;z ) \ , - \ , _ 2f_1(a , b;c;z)\ , \right ] \ ; .\end{aligned}\ ] ] in particular , we can write @xmath625 note that the r.h.s . in
the above relation is finite in the unit circle for @xmath171 and , using again eq .
( [ eq:2f1sing ] ) , we can verify that it is also positive for @xmath576 $ ] .
the above results , together with the eqs .
( [ eq:2f1-uno ] ) and ( [ eq:2f1-due ] ) , allow to write a lower and an upper bound for the integral @xmath626^{\nu } } \ ; .
\label{eq : igeneral - dell}\ ] ] indeed , after considering the angular integration , we have ( for @xmath627 ) @xmath628 & & \qquad \qquad \qquad \ ; \left . + \ ,
\int_p^{\ell } \ , { \ensuremath{\mathrm{d}}}q \ , q^{d-3 } \ , \mathcal{d}(q^2 ) \ , _ 2f_1(\nu , \nu - d/2 ; 1+d/2 ; p^2/q^2 ) \ , \right ] \ ; .
\label{eq : igeneral - dell2}\end{aligned}\ ] ] then , for @xmath629 we obtain @xmath630 with with a behavior @xmath166 at large momenta , the second integral in eq .
( [ eq : id - ell ] ) is uv divergent if @xmath631 and @xmath632 . ]
@xmath633 \ ; .
\label{eq : id - ell}\ ] ] also , in the limit @xmath7 , we find ( for @xmath61 ) @xmath634 \ ; , \label{eq : limiti}\ ] ] where we used the trapezoidal rule .
clearly , for @xmath13 , the first term is ir - finite if @xmath71 while the second term is finite for @xmath171 .
finally , note that for @xmath181 the condition @xmath629 simplifies to @xmath635 . at the same time , for @xmath71 the inequalities ( [ eq : igeneral - ell - upper ] ) become @xmath636 and for @xmath4 we have @xmath637 .
in section [ sec : d=2sing ] we have proven , using two different approaches , that in the @xmath6 case one needs to set @xmath11 in order to have @xmath94 .
the assumptions made for the gluon propagator were rather general .
indeed , for the first proof one needs , for small momenta @xmath68 , an expansion of the gluon propagator of the type @xmath638 , with @xmath129 and @xmath130 and @xmath131 finite . at the same time , for large momenta @xmath68 , we required @xmath639 let us recall that we are indicating with @xmath118 a primitive of @xmath5 and that @xmath640 is the first derivative with respect to the variable @xmath68 . in section [ sec : d=2sing ] above we considered for @xmath5 a large @xmath68 behavior of the type @xmath58 .
however , it is clear that a weaker condition can also be used .
indeed , the behavior @xmath641 with @xmath642 also allows to satisfy the above conditions . in order to check this
, one should recall that @xmath641 implies implies @xmath641 , is not always true ( see for example ref .
@xcite and section 7.3 in ref .
this is why the so - called de lhpital s rule does not always apply ( see also footnote [ foot : tauberian ] ) . ]
@xmath643 and these two asymptotic behaviors yield the limits in eq .
( [ eq : dlim1 ] ) . under the same hypothesis
we can also verify that the integral [ see eq .
( [ eq : sigmafin2 ] ) ] @xmath644 \ , - \ , x \ , \left ( x^{\eta } + m \right ) \ , \mathcal{d}'(x)}{x^{1+\eta}}\ ] ] is finite . to this end , let us first consider the integral @xmath645 \ , - \ , x \
, \left ( x^{\eta } + m \right ) \ , \mathcal{d}'(x)}{x^{1+\eta}}\ ] ] with @xmath646 and @xmath28 small . in the limit @xmath647 the integrand behaves as @xmath648\ ] ] and
no singularity arises in the integration from @xmath649 to @xmath650 ( if @xmath129 ) .
at the same time , for large @xmath97 the same integrand behaves as @xmath651 \ ; .\ ] ] thus , for sufficiently large @xmath652 , the integral @xmath653 \ , - \ , x \ , \left ( x^{\eta } + m \right ) \ , \mathcal{d}'(x)}{x^{1+\eta}}\ ] ] can be approximated by @xmath654 \ ; . \label{eq : ia}\ ] ] after integrating by parts and avoid the partial integration .
] we then find @xmath655 \
, \right|_{\ell^2}^{\infty } \ , + \ , \int_{\ell^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \left [ \ , \frac{m \mathcal{d}(0)}{x^{1+\eta } } \,+\ , \frac{\mathcal{d}(x)}{x } \ , \right ] \ ; , \label{eq : iaparts}\ ] ] which is clearly finite under the assumptions made for the gluon propagator @xmath5 . finally , the remaining term @xmath656 \ , - \ , x \ , \left ( x^{\eta } + m \right ) \
, \mathcal{d}'(x)}{x^{1+\eta } } \ ; , \ ] ] with @xmath657 , can be easily bounded by making the assumption that neither @xmath136 nor @xmath155 display a singularity for @xmath658 $ ] .
thus , we can conclude that the integral is indeed finite .
we further notice that @xmath659 is null in the limit @xmath660 .
going back to the first proof in section [ sec : d=2sing ] , one can verify that the above conditions allow to show that the gribov form - factor @xmath0 goes to zero for @xmath68 going to infinity and that the term @xmath150 is the only singularity of @xmath0 in the ir limit @xmath110 .
the situation is , of course , very similar in the second proof . in this case
we considered a finite value for @xmath60 and the limit @xmath661 \ , \mathcal{d}'(p^2 ) \ ,
= \ , 0 \ ; .
\label{eq : dlim3}\ ] ] we also imposed , for large momenta @xmath68 , the limits @xmath662 \ , \mathcal{d}'(p^2 ) = \lim_{p^2 \to \infty } \ , \ln(p^2 ) \ , \mathcal{d}(p^2 ) = \lim_{p^2 \to \infty } \ , \frac{\hat{d}(p^2)}{p^2 } \ , = \ , 0 \ ; .
\label{eq : dlim4}\ ] ] clearly , any gluon propagator with an ir behavior of the type @xmath663 , with @xmath309 and with @xmath60 and @xmath664 finite satisfies the limit ( [ eq : dlim3 ] ) .
also , if we make the hypothesis @xmath665 , with @xmath642 for large values of @xmath68 we have , in the same limit , as before , one needs to be careful and make hypotheses on @xmath640 and not on @xmath5 .
for example , after making assumptions on the uv behavior of @xmath5 one could employ de lhpital s rule in order to obtain @xmath666 and conclude that @xmath667 \ , \mathcal{d}'(p^2 ) = 0 $ ] follows from the condition @xmath668 . however , it is easy to find counterexamples to this result .
a classical one is @xmath669 for which @xmath667 \ , \mathcal{d}'(p^2)$ ] is undetermined even though @xmath668 clearly holds .
thus , in order to apply the result in eq .
( [ 6 ] ) , additional auxiliary ( also called tauberian ) conditions have to be imposed to the function @xmath5 , for example on @xmath155 or on @xmath670 ( see again ref .
in particular , imposing @xmath191 is not sufficient since @xmath671/p^2 $ ] is also a counterexample .
here we find it simpler to make hypotheses directly on the asymptotic behavior of @xmath640 .
however , one should also notice that the large-@xmath97 behavior @xmath665 , with @xmath642 , could also imply @xmath672 ( see for example ref .
thus , we also have to impose explicitly the condition @xmath673 for @xmath146 . ]
@xmath674 and @xmath675 and one can easily prove the results in eq.([eq : dlim4 ] ) . as a consequence
, we can also verify that the integral [ see eq .
( [ eq : sigma0due ] ) ] @xmath676 \ , \mathcal{d}''(x ) \label{eq : dlim5}\ ] ] is finite for any @xmath677 .
indeed , we can integrate by parts . ] obtaining @xmath678 \ , \mathcal{d}'(x ) \bigr|_{p^2}^{\infty } \ , - \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln(x ) \,\mathcal{d}'(x ) \\[2 mm ] & = & \left [ \ , p^2 \ , \ln(p^2 ) \ , - \ , p^2 \ , \right ] \ , \mathcal{d}'(p^2 ) \ , - \ , \int_{p^2}^{\infty } \ , { \ensuremath{\mathrm{d}}}x \ , \ln(x ) \,\mathcal{d}'(x ) \ ; . \label{eq : ib}\end{aligned}\ ] ] note that the integral on the r.h.s . of eq .
( [ eq : ib ] ) also appears in the second term of eq.([eq : ia ] ) .
thus , using eqs .
( [ eq : iaparts ] ) and ( [ eq : dlim4 ] ) we have that , for large @xmath68 , the integral @xmath679 is finite and @xmath680 . at the same time , for @xmath68 going to zero we have @xmath681 and the integrand in eq .
( [ eq : dlim5 ] ) behaves as @xmath682 x^{\eta -1}$ ] .
thus , with @xmath309 , no singularity arises from the integration at @xmath649 .
this result completes the conditions necessary ( in the second proof ) to show that @xmath0 goes to zero at large momenta and that the ir singularity @xmath150 appears in the limit @xmath110 .
finally , due to the well - known results @xmath683 for @xmath684 , it is clear that the above proofs can also be generalized to asymptotic behaviors that include logarithmic functions . at large momenta
these logarithmic corrections could be present , for example , if one uses , as an input in the evaluation of @xmath0 , a gluon propagator obtained in perturbation theory beyond the tree - level term .
let us start from eq .
( [ eq : igeneral - dell2 ] ) , with @xmath181 and @xmath685 , and subtract @xmath174 from @xmath0 , where @xmath173 is a fixed momentum
. then we can write @xmath686 & & \qquad \qquad \quad \ , - \ , \int_0^{\mu } \ , { \ensuremath{\mathrm{d}}}q \ , q^{d-1 } \ , \mathcal{d}(q^2 ) \ , \frac { _ 2f_1\left(1 , 1-d/2 ; 1+d/2 ; q^2/\mu^2\right ) } { \mu^2 } \nonumber \\[2 mm ] & & \qquad \qquad \quad \ , + \ , \int_{p}^{\mu } \ , { \ensuremath{\mathrm{d}}}q \ , q^{d-3 } \ , \mathcal{d}(q^2 ) \ , _ 2f_1\left(1 , 1-d/2 ; 1+d/2 ; p^2/q^2\right ) \nonumber \\[2 mm ] & & \qquad \qquad \quad \ , + \ , \int_{\mu}^{\infty } \ , { \ensuremath{\mathrm{d}}}q \ , q^{d-3 } \ , \mathcal{d}(q^2 ) \ , \left [ \ , _ 2f_1\left(1 , 1-d/2 ; 1+d/2 ; p^2/q^2\right ) \right .
\nonumber \\[2 mm ] & & \qquad \qquad \qquad \qquad \qquad \ , - \ , \left . _ 2f_1\left(1 , 1-d/2 ; 1+d/2
; \mu^2/q^2\right ) \ , \right ] \ , \biggr\ } \ ; . \label{eq:4dmom}\end{aligned}\ ] ] in the case @xmath66 we can use the result ( [ eq:2f1-d4 ] ) in appendix [ sec : hyper ] .
thus , the last integral in the above expression becomes @xmath687 which is uv - finite for any gluon propagator that goes to zero at large momenta .
the apparent linear divergence for large @xmath68 in the above integral is of course canceled by the third integral in eq.([eq:4dmom ] ) above , i.e. by @xmath688 then , for @xmath146 and for @xmath498 at large momenta one only gets a logarithmic contribution @xmath689 , as expected . using the above formula ( [ eq:4dmom ] ) , the proof that @xmath0 is ir - finite for @xmath191 can be obtained as in section [ sec : d - exact ] .
indeed , for @xmath68 going to zero we have to consider only the first and the third integrals and the last one is regular at @xmath69 [ see eq .
( [ eq : uv4d ] ) ] . ] in eq .
( [ eq:4dmom ] ) . then , using again the result ( [ eq:2f1-d4 ] ) we can write @xmath690 and no singularity arises in the limit @xmath110 if @xmath13 . at the same result one arrives by setting @xmath691 and by integrating explicitly the l.h.s . in the above equation .
d. zwanziger , phys .
* d65 * , 094039 ( 2002 ) . c. lerche , l. von smekal , phys .
* d65 * , 125006 ( 2002 ) .
m. q. huber , r. alkofer , c. s. fischer , k. schwenzer , phys .
* b659 * , 434 ( 2008 ) .
r. alkofer , j. greensite , j. phys .
* g34 * , s3 ( 2007 ) .
a. c. aguilar , a. a. natale , jhep * 0408 * , 057 ( 2004 ) .
p. boucaud et al .
, jhep * 0606 * , 001 ( 2006 ) .
a. c. aguilar , d. binosi , j. papavassiliou , phys .
* d78 * , 025010 ( 2008 ) .
d. zwanziger , aip conf . proc .
* 1343 * , 176 ( 2011 ) .
d. zwanziger , pos * facesqcd * , 023 ( 2010 ) .
j. a. gracey , phys .
b * 632 * , 282 ( 2006 ) [ erratum - ibid .
* 686 * , 319 ( 2010 ) ] .
j. a. gracey , jhep * 0605 * , 052 ( 2006 ) .
a. cucchieri , t. mendes , a. mihara , jhep * 0412 * , 012 ( 2004 ) .
a. cucchieri , a. maas , t. mendes , phys .
* d74 * , 014503 ( 2006 ) . e. -m .
ilgenfritz , m. muller - preussker , a. sternbeck , a. schiller , i. l. bogolubsky , braz .
j. phys .
* 37 * , 193 ( 2007 ) . | we study general properties of the landau - gauge gribov ghost form - factor @xmath0 for su(@xmath1 ) yang - mills theories in the @xmath2-dimensional case .
we find a qualitatively different behavior for @xmath3 with respect to the @xmath4 case .
in particular , considering any ( sufficiently regular ) gluon propagator @xmath5 and the one - loop - corrected ghost propagator , we prove in the @xmath6 case that the function @xmath0 blows up in the infrared limit @xmath7 as @xmath8 . thus , for @xmath4 , the no - pole condition @xmath9 ( for @xmath10 ) can be satisfied only if the gluon propagator vanishes at zero momentum , that is , @xmath11 . on the contrary , in @xmath12 and 4 , @xmath0 is finite also if @xmath13 .
the same results are obtained by evaluating the ghost propagator @xmath14 explicitly at one loop , using fitting forms for @xmath5 that describe well the numerical data of the gluon propagator in two , three and four space - time dimensions in the su(2 ) case .
these evaluations also show that , if one considers the coupling constant @xmath15 as a free parameter , the ghost propagator admits a one - parameter family of behaviors ( labelled by @xmath15 ) , in agreement with previous works by boucaud et al . in this case the condition @xmath16 implies @xmath17 , where @xmath18 is a `` critical '' value .
moreover , a free - like ghost propagator in the infrared limit is obtained for any value of @xmath15 smaller than @xmath18 , while for @xmath19 one finds an infrared - enhanced ghost propagator .
finally , we analyze the dyson - schwinger equation for @xmath0 and show that , for infrared - finite ghost - gluon vertices , one can bound the ghost form - factor @xmath0 . using these bounds
we find again that only in the @xmath4 case does one need to impose @xmath11 in order to satisfy the no - pole condition .
the @xmath4 result is also supported by an analysis of the dyson - schwinger equation using a spectral representation for the ghost propagator .
thus , if the no - pole condition is imposed , solving the @xmath4 dyson - schwinger equations can not lead to a massive behavior for the gluon propagator .
these results apply to any gribov copy inside the so - called first gribov horizon , i.e. the @xmath6 result @xmath11 is not affected by gribov noise .
these findings are also in agreement with lattice data . |
in a recent work @xcite , maddah - ali and niesen considered the caching problem which deals with improving the content delivery efficiency , _
i.e. _ , the total traffic rate @xmath2 in the delivery phase , through the utilization of local cache memory of capacity @xmath3 each where the coded contents are strategically prepared in the placement phase , in a system with @xmath4 files and @xmath5 users .
it was shown coded caching can be rather beneficial , and in fact orderly optimal , while uncoded caching solution suffers a significant loss .
subsequent works extended it to decentralized caching placements @xcite , caching with nonuniform demands @xcite , and online caching placements @xcite , etc .. despite these advances , the fundamental tradeoff between the delivery traffic rate @xmath2 and cache memory capacity @xmath3 is not fully known except the two - user two - file case .
this is partly due to the fact that the main focus of these existing investigations @xcite is on the regime when the number of files and the number of users are both large , where coded caching can provide the largest gain over the uncoded counterpart .
however , within the problem setting of@xcite , the number of users @xmath5 accounted for in the placement phase is also the total number of simultaneous content requests in the delivery phase , and in some applications the number of simultaneous data requests can be quite small . in such scenarios , better understanding of the fundamental limits of the caching problem , when either @xmath4 or @xmath5 is small , in fact becomes rather important .
the outer bound provided in @xcite was obtained through a cut - set argument , and it is generally suspected to be not tight .
however , stronger outer bounds appear difficult to find analytically . in this short note
, we provide an improved outer bound for the case with @xmath0 files and @xmath1 users , which is obtained using the computational approach developed in @xcite .
this outer bound is included in the online collection of `` solutions of computed information theoretic limits ( scitl ) '' at @xcite , where the data files for the proof can be downloaded for further processing .
we use the problem definition and notation given in @xcite : @xmath3 is the capacity of the local memory cache size , and @xmath2 is the multicast rate in the delivery transmission ; each file is assumed to have unit size , _
@xmath7 , since normalizing by @xmath8 does not cause any essential loss here .
the main result of this note is the following theorem . the memory - delivery - rate tradeoff for the @xmath6 coded caching problem must satisfy : @xmath9 .
[ fig : region],width=377 ] this region is illustrated in fig .
[ fig : region ] .
we note that the point @xmath10 , which was shown to be achievable in @xcite , is optimal since it is on the boundary of the outer bound .
the first two and the last two inequalities in ( [ eqn : bounds ] ) are already known in @xcite .
the remaining ones are new , and we discuss them in some details in the next section . for reference ,
the inner bound given in @xcite through a centralized placement algorithm is also plotted .
we shall not provide the full details of the proof in this note , since some of them can be rather large ( in a table form , or long when written down as chains of inequalities ) .
however , one of the inequalities is discussed in some depth in section [ sec : translation ] to illustrate the manner this bound is obtained , and interested readers are referred to @xcite for more details .
in this section we discuss the symmetry we utilized to reduce the complexity of the computation , which is important to understand the tabulation based proof . in the sequel we use @xmath11 to denote the @xmath12-th ( random )
file , @xmath13 to denote the content stored at the @xmath12-th user .
the multicasted message in the delivery phase is written as @xmath14 , meaning that it is the message when the first user demands the @xmath12-th file , the second user demands the @xmath15-th file , and the third user demands the @xmath16-th file .
the files are mutually independent .
similar as in @xcite , it can be shown that without loss of generality we can consider only symmetric codes . however , here the symmetry structure is different from that in @xcite . in the computational approach ,
we utilize the following type of symmetry to reduce the computation .
let three sets of random variables be given as @xmath17 let a permutation function be defined as @xmath18 on the set of @xmath19 .
the permutation operates on the set of @xmath20 as follows @xmath21 and on the set @xmath22 as follows @xmath23 for example the permutation function @xmath24 maps @xmath25 to @xmath26 , but maps any @xmath27 to @xmath28 , and thus the set @xmath29 to @xmath30 .
we call a given caching code symmetric , if for any @xmath31 , and any permutation @xmath32 , we have @xmath33 as a reality check , consider @xmath34 under the aforementioned permutation : it should be equal to @xmath35 ; notice that @xmath36 is a function of @xmath37 and under the permutation , @xmath36 is still a function of @xmath38 .
intuitively , we prove that symmetric codes are without loss of optimality based on the idea of time ( space ) sharing : encode @xmath39 of the three files and place the coded cache in a permuted order at the three users , which will remove any asymmetry in the code .
we plan to explore the symmetry structure further in a subsequent work .
.the entropy terms used in the proof . [ cols="^,^ " , ] to illustrate the proof obtained using the computational approach , consider the inequality @xmath40 , which is particularly simple to prove .
this proof is a direct translation of the solution obtained using the computational approach , which are given in table [ tab : correspondence ] and table [ table : cancellation ] , where we use @xmath7 to denote the unit of the file size .
we can alternatively write the following chain of inequalities @xmath41 where ( s ) is for reason of symmetry , and ( a ) is because the message @xmath42 together with the coded content cached at user-@xmath43 , _
i.e. , @xmath44 _ , can recover @xmath45 , ( b ) is because of the assumption that each file has unit size as well as the chain rule , and ( c ) is because the coded content cached at user-@xmath46 , _
i.e. , @xmath47 _ , together with @xmath42 can recover @xmath36 , and @xmath47 together with @xmath48 can also recover @xmath49 . note the five inequalities in ( [ eqn : chain ] )
correspond to the five rows in table [ table : cancellation ] , though not in the same order , and not the exact same form because of the symmetry structure and application of certain chain rule simplifications .
an improved outer bound is given for the caching problem when @xmath6 in this note .
this result is part of the online collection of `` solutions of computed information theoretic limits ( scitl ) '' hosted at @xcite , which hopefully in the future can serve as a data depot for information theoretic limits obtained through computational approaches .
some results in this collection requires non - trivial variation of the approach outlined in @xcite , the details of which will be presented elsewhere . | the fundamental limit of coded caching is investigated for the case with @xmath0 files and @xmath1 users . an improved outer bound
is obtained through the computational approach developed by the author in an earlier work .
this result is part of the online collection of `` solutions of computed information theoretic limits ( scitl ) '' . |
the present work concerns the geometric structure of the flow of a vector field in four - dimensional spacetime .
we work from the perspective that the generating vector field satisfying some set of governing evolution equations is the primary quantity , and the flow is the secondary ( or derived ) quantity . assuming _ a - priori _ a smooth generating vector field , we introduce a generally covariant measure of the flow geometry called the _ referential gradient of the flow_. the main result of this work is the explicit relation between the referential gradient of the flow and the generating vector field , and is provided for from two equivalent perspectives : a lagrangian specification with respect to a generalized parameter , and an eulerian specification making explicit the evolution dynamics .
furthermore , we provide explicit non - trivial conditions which govern the transformation properties of the referential gradient object .
the layout of this paper is as follows .
section ( [ s : flowandrefgrad ] ) provides the standard differential geometry context ( [ ss : preliminaries ] ) and formalism of mathematical flow representations ( [ ss : generalconnectivitymap ] ) in order to provide a rigorous framework from which to define the referential gradient .
the main results of the paper are put forth in section ( [ s : refgraddefined ] ) where we define the referential gradient of the flow . in section ( [ ss : lagrangespecrefgrad ] ) , we prove the lagrangian specification theorem ( [ thm : refgradlagrangian ] ) from which is given a non - local , closed - form functional solution with respect the generating vector field .
furthermore , due to the non - local nature of the lagrangian specification of the referential gradient , in section ( [ ss : refgradtransforms ] ) we prove three lemmas that identify the non - trivial referential gradient transformation conditions : lemma ( [ lem : refgrad ] ) provides the condition for manifest covariance ; lemma ( [ lem : refgradgroup ] ) provides for the group property with respect to the connectivity parameter ; and lemma ( [ lem : representationrelations ] ) provides the proper relations between the corresponding referential gradient representations associated with a change of integration variable . in section ( [ ss : refgraddynamics ] ) , we prove the eulerian specification theorem ( [ thm : refgraddynamics ] ) which in a coordinate chart makes explicit the referential gradient dynamics at each point of the manifold . throughout the paper , we work in both coordinate - free language , show coordinate - dependent expressions , and give explicit illustrative examples .
greek indices @xmath0 denote spacetime components , and latin indices @xmath1 denote only spatial components ; in addition we employ the standard einstein summation convention .
let @xmath2 be a simply connected region of spacetime with ( semi - riemannian ) metric @xmath3 .
furthermore , assume the spacetime topology may be foliated as @xmath4 , where @xmath5 is a three - dimensional spatial hypersurface , and @xmath6 is the time axis .
let @xmath7 be an atlas of @xmath8 , such that an arbitrary point @xmath9 has coordinate component functions @xmath10 , where @xmath11 and @xmath12 are identified with coordinate time and 3-space , respectively
. given the atlas @xmath7 over @xmath8 , we have on each chart @xmath13 the local coordinate basis @xmath14 and local dual basis @xmath15 , defined , respectively , via @xmath16 and @xmath17 , where @xmath18 is the kronecker delta .
the invariant interval is given in coordinates by , @xmath19 where in general , in the coordinate chart @xmath13 the metric components are given functions of position , @xmath20 .
we assume an affine connection @xmath21 , that satisfies the standard axioms ( see e.g. , ref .
@xcite @xmath22 3.1 ) . in a coordinate chart @xmath13 ,
the connection components are defined via the action on basis vectors , @xmath23 and @xmath24 , such that the covariant derivative of a mixed tensor field @xmath25 is given by , @xmath26 in addition , the covariant derivative with respect to @xmath27 in the direction of the vector field @xmath28 is given by , @xmath29 , where @xmath30 is the contraction map ( with some abuse of notation where the meaning is clear , we write the contraction map @xmath31 , interpreted as the standard scalar product ) .
beyond what is physically reasonable , we make no assumptions regarding the metric or the connection . without loss of generality
, one may perform particular calculations using the levi - civita connection with components given by , @xmath32 let @xmath33 be a smooth vector field everywhere on @xmath8 ( unless explicitly stated otherwise , by @xmath34 we mean @xmath35 ) . in a chart @xmath13 , the vector field
may be written , @xmath36 where the component functions @xmath37 .
identify the vector field @xmath38 with a four - velocity field in minkowski spacetime , curvilinear coordinates ( see e.g. , ref .
@xcite @xmath22 6.2 ) , the contravariant components of which are given by , @xmath39 where @xmath40 is the velocity field in a co - moving coordinate frame .
identify the vector field @xmath38 with the four - magnetic field ( see e.g. , ref .
@xcite @xmath22 13.10.2 ) , the contravariant components of which are given by , @xmath41 where @xmath42 is the levi - civita tensor density , and @xmath43 . in the frame of the observer , @xmath44 is the electromagnetic tensor , and @xmath45 are the covariant components of the four - velocity field .
[ d : mcflow ] _ ( the flow of a vector field ) _ let @xmath46 be an open set , and @xmath47 and open interval containing @xmath48 .
the _ flow _ of @xmath49 is a map @xmath50 , such that for any point @xmath51 , @xmath52 we refer to @xmath49 as the _ generating vector field _ , @xmath53 as the _ connectivity parameter _ ( associated with the generating vector field ) , @xmath54 as the _ reference point _ , and @xmath55 as the _ reference condition_. if @xmath9 for some coordinate chart @xmath13 , then so also lies a segment of @xmath56 in @xmath13 . for that segment
the coordinate components of equation ( [ e : mcflow ] ) with respect to the basis @xmath57 are , @xmath58 equations ( [ e : mcflowcoords ] ) represent an initial value problem in ( four ) first - order differential equations in the parameter @xmath53 , covariant under a change of coordinates . depending on the nature of the generating vector field , equations ( [ e : mcflowcoords ] ) may be non - autonomous or autonomous in the parameter @xmath53 .
hence , for fixed reference condition , by standard theorems of existence , uniqueness , and extension for ordinary differential equations ( see e.g. , refs .
@xcite ) , the solution @xmath59 exists , is unique , smooth , and maximal for any @xmath9 and all @xmath60 . for a generating vector field identified with the four - velocity field ( [ ex : fourvelocity ] ) , the associated connectivity parameter is identified with time , and equations ( [ e : mcflowcoords ] ) are non - autonomous . for a generating vector field identified with the four - magnetic field ( [ ex : fourmagneticfield ] ) , the associated connectivity parameter is identified with a distance per magnetic field strength , and equations ( [ e : mcflowcoords ] ) are autonomous . by equation ( [ e : mcflowcoords ] )
the units of the flow @xmath56 are identified with the position coordinates @xmath61 .
hence , the units of the connectivity parameter @xmath53 are coordinate - dependent ; that is , in a given coordinate chart , @xmath53 has coordinate units per generating vector field units . in @xmath22 [ ss : refgraddynamics ] , we return to a full discussion of the coordinate representation of the connectivity parameter . from the flow
may be defined two collections of maps : [ d : mcflow ] _ ( the orbit of @xmath54 ) _ for fixed reference point @xmath62 , the _ orbit of @xmath54 _ is the @xmath63 map @xmath64 from an interval in @xmath6 into @xmath8 ; e.g. , @xmath65 such that , @xmath66 if @xmath9 , then the coordinate maps , @xmath67 are smooth curves with tangent vector everywhere defined by , and equal to , the ( smooth ) generating vector field . for a generating vector field identified with the four - velocity field ( [ ex : fourvelocity ] ) in the non - relativistic limit ,
the orbit of @xmath54 is a _
streamline_. for a generating vector field identified with the four - magnetic field ( [ ex : fourmagneticfield ] ) in the non - relativistic limit , the orbit of @xmath54 is a _
magnetic line of force _ ( or _ magnetic field line _ ) .
[ d : mcmap ] _ ( the connectivity map ) _ for fixed connectivity parameter @xmath68 , the _ connectivity map _ is the one - parameter group of ( active ) diffeomorphisms @xmath69 , @xmath70 if @xmath71 , then for all possible values of connectivity parameter @xmath72 , the flow @xmath73 solutions form a lie group @xmath74 .
that is to say , for each value of @xmath72 is associated a smooth transformation of the space to itself @xmath75 , such that the connectivity map satisfies the following group properties : @xmath76 for any @xmath77 ; @xmath78 is the identity element ; and @xmath79 is the inverse element , such that @xmath80 . for divergence - free generating vector fields ,
the flow is identified with the group of volume - preserving diffeomorphisms @xmath81 ( see e.g. , ref .
@xcite @xmath22 i.1 ) .
the flow represents an equivalence class under a affine transformations of the connectivity parameter , @xmath82 with @xmath83 and @xmath84 ( see e.g. , ref .
@xcite @xmath22 7.4 ) .
in particular , the connectivity parameter may be identified with an arc length measure @xmath85 along the orbit @xmath86 from the reference point @xmath62 , where @xmath87 by a change of variable @xmath88 with , @xmath89 for a smooth generating vector field @xmath49 , provided @xmath90 for all @xmath62 , it can be shown the transformation ( [ e : arclengthcondition ] ) exists and is unique .
furthermore , under transformation ( [ e : arclengthcondition ] ) , the generating vector field for the re - parametrized flow @xmath91 , is the unit direction of the generating vector field @xmath92 , such that , @xmath93 additionally , if @xmath9 , then equation ( [ e : arclengthmcflow ] ) finds a coordinate expression similar to that of ( [ e : mcflow ] ) , @xmath94 we call the re - parametrized flow @xmath95 of equations ( [ e : arclengthmcflow ] ) , the _ arc length representation _ , reflecting the fact that the connectivity parameter itself is identified with an arc length measure along the orbit issuing from the reference point @xmath62 . in the arc length representation of the flow for a generating vector field identified with the four - velocity field ( [ ex : fourvelocity ] ) , the connectivity parameter is a measure of the total time along the streamline . in the arc length representation of the flow for a generating vector field identified with the four - magnetic field ( [ ex : fourmagneticfield ] ) in the non - relativistic limit , the connectivity parameter is a measure of the total spatial distance along the magnetic line of force . unless explicitly noted we will work with the flow representation @xmath56 associated with the full generating vector field @xmath49 . for completeness
though , the flow representations @xmath56 and @xmath95 are implicitly related everywhere via the unit generating field @xmath96 , @xmath97 in a coordinate chart @xmath13 , equation ( [ e : flowarclengthflowrelation ] ) is , @xmath98 where @xmath99 is the metric given in the coordinate chart @xmath100 .
given a generating vector field @xmath49 , the determination of the flow @xmath56 with a particular reference condition @xmath55 , represents an initial value problem , via equations ( [ e : mcflow ] ) .
the geometric flow structure may be discerned by examining the dependence on the reference condition ( see e.g. , ref .
@xcite @xmath22 32 ) .
let @xmath101 be an open ball of radius @xmath102 about the reference point @xmath62 , and @xmath103 be a bounded open interval containing @xmath48 and @xmath53 with @xmath104 , such that @xmath105 for all points @xmath106 and any @xmath107 .
denote the non - zero _ reference shift vector _ @xmath108 such that @xmath109 , where the unit direction vector @xmath110 .
[ d : refgrad ] _ ( the referential gradient of the flow ) _
the _ referential gradient of the flow _ is the matrix - valued function @xmath111 defined via the vector , @xmath112 provided the limit exists for arbitrary reference shift vector @xmath110 . the vector @xmath113 is the referential gradient of the flow corresponding to the direction of the reference shift vector @xmath114 .
it follows immediately for @xmath115 , definition ( [ e : formalrefgrad ] ) reduces to @xmath116 , thus @xmath117 .
a somewhat weaker form of the definition of the referential gradient for any @xmath118 is such that for any reference shift vector @xmath119 and @xmath120 , @xmath121 geometrically , for every @xmath118 , the referential gradient @xmath122 is a measure of the relative change of the flow @xmath56 with respect to a shift in the reference point @xmath123 . in other words , for a given orbit @xmath124 , the referential gradient @xmath122 contains the deformation information of all `` neighboring '' orbits @xmath125 of similar length @xmath126 .
let @xmath127 be in a coordinate chart .
the reference shift direction vector @xmath128 is given by @xmath129 , and the coordinate expression for @xmath113 is , @xmath130 provided the limit exists . when @xmath115 , the referential gradient corresponding to the reference shift direction vector @xmath114 is @xmath131 , hence @xmath132 .
one may always construct a coordinate representation of the referential gradient from the coordinate representation of the flow via definition ( [ e : formalrefgradcoords ] ) .
however , in typical physical systems of interest , the governing equations describe the evolution of the generating vector field , and the associated flow is derived therefrom .
hence , the following theorem gives the explicit functional dependence of the referential gradient of the flow on the generating vector field . _
( generalized lagrangian specification of the referential gradient ) _ [ thm : refgradlagrangian ] let @xmath133 be a smooth , complete generating vector field with associated flow @xmath134 .
the referential gradient @xmath111 , for @xmath135 and @xmath136 , is the unique solution to the differential equation , @xmath137 where @xmath138 is the identity .
furthermore , in @xmath135 , the solution is given explicitly by the absolutely and uniformly convergent , path - ordered exponential , @xmath139 f \left ( \ 0 , p \
\right ) \\ % f \left ( \ \lambda , p \
\right ) = \mathcal{p } \text{exp } \biggl [ \ \int_{0}^{\lambda } dt \ \nabla b \bigl ( \ \phi_{p } \left ( t \right ) \
\bigr ) \ \biggr ] f \left ( \ 0 , p \
\right ) \\ \\\ ] ] the system of differential equations ( [ e : mcflow ] ) and ( [ e : refgradholonomy ] ) together with their reference conditions are often referred to as the _ system of equations of variations _ for the flow ( see e.g. , ref .
@xcite @xmath22 32 ) .
moreover , equations ( [ e : refgradholonomy ] ) and ( [ a : pexpreferentialgradient ] ) constitute a generalized _ lagrangian specification of the referential gradient _ ( with respect to the connectivity parameter ) . in @xmath22
[ ss : refgraddynamics ] , we construct an _ eulerian specification of the referential gradient _ equivalent to equations ( [ e : refgradholonomy ] ) .
we begin by deriving equation ( [ e : refgradholonomy ] ) . using the fundamental relation between the connection flow and generating field , equations ( [ e : mcflow ] ) , for any @xmath107
, we construct the flow difference , @xmath140 \\ % \\ % \end{split } % \displaystyle % \phi \left ( 0 , p + h \right ) - \phi \left ( 0 , p \right ) = h \\
% \end{array}\ ] ] equation ( [ e : refgradholonomy ] ) follows upon dividing equation ( [ e : mcflowdifference ] ) by @xmath141 and in the limit @xmath142 .
consider the lhs of equation ( [ e : mcflowdifference ] ) .
the vector @xmath114 does not depend on the connectivity parameter @xmath53 . using definition ( [ e : formalrefgrad ] )
, it follows immediately,@xmath143 \\ % \\ % \displaystyle % \lim_{\vert h \vert \rightarrow 0 } \
\frac{1}{\vert h \vert } \ \frac{\partial}{\partial \lambda } \biggl ( \ \phi \bigl ( \lambda , p + \vert h \vert \ \hat{h } \bigr ) - \phi \bigl ( \lambda , p \bigr ) \
\biggr ) = \frac{\partial}{\partial \lambda } \
\biggl [ \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \ \biggr ] \\ % \\ \displaystyle \lim_{\vert h \vert \rightarrow 0 } \ \frac{1}{\vert h \vert } \ \frac{\partial}{\partial \lambda } \biggl ( \ \phi \bigl ( \lambda , p + \vert h \vert \ \hat{h } \bigr ) - \phi \bigl ( \lambda , p \bigr ) \
\biggr ) = \frac{\partial f \left ( \ \lambda , p \ \right)}{\partial \lambda } \cdot \hat{h } \\ \\ % \end{array}\ ] ] consider the rhs of equation ( [ e : mcflowdifference ] ) .
let @xmath144 and denote the non - zero shift vector @xmath145 such that @xmath146 , where the shift direction vector @xmath147 . by assumption ,
the generating vector field @xmath38 is smooth for all @xmath62 , thus we may taylor expand the rhs of equation ( [ e : mcflowdifference ] ) ( see appendix [ a : generalizedfirstordertaylorexpansion ] ) . noting the identity @xmath148 , the rhs of equation ( [ e : mcflowdifference ] ) may be written , @xmath149 hence , @xmath150 \\
% \displaystyle % & = \lim_{\vert h \vert \rightarrow 0 } \
\frac{1}{\vert h \vert } \
\biggl [ \ b \bigl ( \ y + \vert w \vert \ \hat{w } \
\bigr ) - b \bigl ( y \bigr ) \ \biggr ] \\ % \\ % \displaystyle % & = \lim_{\vert h \vert \rightarrow 0 } \
\frac{1}{\vert h \vert } \
\biggl [ \ w \cdot \nabla b \bigl ( y \bigr ) + w \cdot \int^{1}_{0 } ds \
\biggl ( \ \nabla b \bigl ( \ y + s \ \vert w
\vert \ \hat{w } \
\bigr ) - \nabla b \bigl ( y \bigr ) \
\biggr ) \ \biggr ] \\ % \\
\displaystyle \lim_{\vert h \vert \rightarrow 0 } \
\frac{1}{\vert h \vert } & \ \biggl [ \ b \bigl ( \ \phi \bigl ( \lambda , p + \vert h \vert \ \hat{h } \bigr ) \
\bigr ) - b \bigl ( \ \phi \bigl ( \lambda , p \bigr ) \
\bigr ) \ \biggr ] \\
\displaystyle & = \lim_{\vert h \vert \rightarrow 0 } \
\frac{w}{\vert h \vert } \cdot \nabla b \bigl ( y \bigr ) \\
\displaystyle & + \lim_{\vert h \vert \rightarrow 0 } \
\frac{w}{\vert h \vert } \cdot \int^{1}_{0 } ds \
\biggl ( \ \nabla b \bigl ( \ y + s \ \vert w \vert \ \hat{w } \
\bigr ) - \nabla b \bigl ( y \bigr ) \
\biggr ) \\ \end{split } \end{array}\ ] ] recall , @xmath151 ; by definition ( [ e : formalrefgrad ] ) , @xmath152 it remains to show the integral remainder goes to zero in the limit @xmath142 .
it suffices to show @xmath153 in the limit @xmath142 . by assumption ,
the generating vector field is smooth , then by standard theorems of the smoothness of ode solutions , so also is the associated flow smooth ( e.g. , ref .
@xcite , theorem 17.19 ) ; thus , @xmath154 the integral remainder term in equation ( [ e : mcflowdifferencerhs1 ] ) is zero since @xmath155 , and , @xmath156 \\ \displaystyle & = \bigl ( \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \
\bigr ) \cdot \int^{1}_{0 } ds \
\biggl ( \ \nabla b \bigl ( y \bigr ) - \nabla b \bigl ( y \bigr ) \
\biggr ) = 0 \\ \end{split } % \\
% \displaystyle % \lim_{\vert h \vert \rightarrow 0 } \
\frac{w}{\vert h \vert } \cdot \int^{1}_{0 } ds \
\biggl ( \ \nabla b \bigl ( \ y + s \ \vert w \vert \ \hat{w } \
\bigr ) - \nabla b \bigl ( y \bigr ) \
\biggr ) \ \biggr ] = 0 \\ % \end{array}\ ] ] hence , the rhs of equation ( [ e : mcflowdifference ] ) is , @xmath157 \\ \displaystyle & = \bigl ( \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \lambda , p \bigr ) \
\bigr ) \\ \end{split}\ ] ] equating the lhs with the rhs , respectively , equations ( [ e : mcflowdifferencelhs ] ) and ( [ e : mcflowdifferencerhs ] ) , @xmath158 \\ % \\ % \displaystyle % \frac{\partial f \left ( \ \lambda , p \ \right)}{\partial \lambda } \cdot \hat{h } = \bigl ( \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \lambda , p \bigr ) \
\bigr ) \\ \displaystyle \frac{\partial}{\partial \lambda } \
\bigl ( \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \
\bigr ) = \bigl ( \ f \left ( \ \lambda , p \
\right ) \cdot \hat{h } \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \lambda , p \bigr ) \
\bigr ) \\ % \end{array}\ ] ] finally , since the reference shift vector @xmath114 is arbitrary , and noting the contraction map @xmath148 remains unambiguous for @xmath159 fields , the differential equation ( [ e : refgradholonomy ] ) follows immediately , @xmath160 the reference condition @xmath117 is given by definition ( [ e : formalrefgrad ] ) at @xmath115 .
we proceed to construct the solution ( [ a : pexpreferentialgradient ] ) . by assumption ,
the generating vector field @xmath38 is smooth and complete for all @xmath62 , and so therefore the gradient @xmath161 is smooth and everywhere non - singular for all @xmath62 .
thus , the system of equations ( [ e : refgradholonomy ] ) is a well - posed initial value problem and standard theorems for existence , uniqueness , and extension for ordinary differential equations apply ( see e.g. , refs .
@xcite ) . to construct solution ( [ a : pexpreferentialgradient ] )
, we consider the integral formulation of the initial value problem ( [ e : refgradholonomy ] ) ; by the fundamental theorem of calculus ( see e.g. , ref .
@xcite @xmath22 1.4 ) , equation ( [ e : refgradholonomy ] ) is equivalent to the integral equation , @xmath162 where @xmath117 is the identity . for fixed
@xmath126 , equation ( [ e : integralnearbyfllinearization ] ) is a linear fredholm equation of the second kind .
relaxing the fixed @xmath53 condition , allowing the upper - limit of integration to vary over all @xmath72 , equation ( [ e : integralnearbyfllinearization ] ) is a linear volterra equation of the second kind .
these integral equations are well known , and the convergence , existence , and uniqueness properties of the solutions are well defined ( see e.g. , ref .
@xcite ) . we proceed to show the existence of a solution to equations ( [ e : integralnearbyfllinearization ] ) via the resolvent kernel method . for a non - singular kernel @xmath163 ,
equation ( [ e : integralnearbyfllinearization ] ) may be solved directly by a method of iteration , such that if there exists a set of functions @xmath122 that satisfy equations ( [ e : integralnearbyfllinearization ] ) , then these functions also satisfy the iterated set of equations . substituting the rhs of equation ( [ e : integralnearbyfllinearization ] ) into the integrand and expanding , we get the two - fold iteration equation , @xmath164 \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \ \bigr)\\ % \\ % \end{split } % \\ \begin{split } \displaystyle & f \left ( \ \lambda , p \
\right ) = f \left ( \ 0 , p \
\right ) + \int_{0}^{\lambda } d\sigma_{0 } \ f \left ( \ 0 , p \ \right ) \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \\ \displaystyle & + \int_{0}^{\lambda } d\sigma_{0 } \
\int_{0}^{\sigma_{0 } } d\sigma_{1 }
\ f \left ( \ \sigma_{1 } , p \ \right ) \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{1 } , p \bigr ) \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \\ \end{split } \end{array}\ ] ] for ease of notation , considering @xmath54 as a fixed parameter , we define , @xmath165 hence , we may write the 2-fold iteration equation ( [ a : implicitdefgrad1 ] ) as , @xmath166 repeating this procedure @xmath167-times , and collecting terms , we obtain the @xmath167-fold iterated equation , @xmath168 \\ % \begin{split } \displaystyle + \int_{0}^{\lambda } d\sigma_{0 } \int_{0}^{\sigma_{0 } } d\sigma_{1 } \ \dotsi \int_{0}^{\sigma_{n } } d\sigma_{n+1 } \ f & \left ( \ \sigma_{n+1 } , p \
\right ) \cdot n^{\left ( 0 \right ) } \left ( \sigma_{n+1 } \right ) \\ \displaystyle & \dotsi n^{\left ( 0 \right ) } \left ( \sigma_{1 } \right ) \cdot n^{\left ( 0 \right ) } \left ( \sigma_{0 }
\right ) \\ \end{split } \end{array}\ ] ] where the @xmath169 term ( @xmath170 ) is defined by the recursive formula , @xmath171 \cdot n^{\left ( 0 \right ) } \left ( \sigma_{0 } \right ) \\ % \\ % \\ % \end{array}\ ] ] thus , as @xmath172 , there exist a set of functions , @xmath122 , that satisfy equations ( [ e : integralnearbyfllinearization ] ) , and are given by the infinite series , @xmath173 \\\ ] ] where each term @xmath174 is given by the recursion relations ( [ a : nthtermrecurrsionrelation ] ) and ( [ a : order2iterationkernels ] ) .
we proceed to show the @xmath167-fold iterated equation ( [ a : nfolditeration ] ) converges absolutely and uniformly for every @xmath167 , and therefore so does formula ( [ a : defgradgeneralsolution ] ) as @xmath172 . to show convergence of the first @xmath167-terms in equation ( [ a : nfolditeration ] ) , let @xmath175 for all @xmath176 ; @xmath177 exists since , by assumption , @xmath49 is smooth
. then , by the recursive formula ( [ a : nthtermrecurrsionrelation ] ) , the absolute value of each successive iterated kernel is also bounded , @xmath178 where @xmath179 is the ( non - negative ) total length of the integration path .
formula ( [ a : boundedkernels ] ) is the @xmath169 term of the absolutely and uniformly convergent exponential series for any ( finite ) @xmath179 , @xmath180 = m \ \sum_{n = 0}^{\infty } \
\frac{1}{n ! } \
\bigl [ \ m \
\vert \lambda \vert \ \bigr]^{n}\ ] ] therefore , the first @xmath167-terms in equation ( [ a : nfolditeration ] ) are bounded by the absolutely and uniformly convergent series ( [ a : nthkernelconvergence ] ) for every @xmath118 . to show the final iteration term is bounded , let @xmath181 for all @xmath176 be the absolute upper bound of the referential gradient function @xmath182 everywhere along the integration path .
@xmath183 exists as long as the gradient of the generating vector field remains smooth and non - singular for any @xmath176 .
then , by the relations ( [ a : boundedkernels ] ) , the final iteration term of the @xmath167-fold iterated equations ( [ a : nfolditeration ] ) satisfies the inequality , @xmath184^{n+1 } \\
\end{array}\ ] ] formula ( [ a : nthtermiterationconvergence ] ) is the @xmath185 term of the same absolutely and uniformly convergent exponential series ( [ a : nthkernelconvergence ] ) with the coefficient @xmath183 , @xmath186 = m \
\sum_{n = 0}^{\infty } \ \frac{1}{n ! } \
\bigl [ \ m \vert \lambda \vert \ \bigr]^{n}\ ] ] thus , the final term in the @xmath167-fold iterated equation ( [ a : nfolditeration ] ) tends to zero as @xmath172 , for any finite @xmath118 .
consequently , the @xmath167-fold iterated equation ( [ a : nfolditeration ] ) converges absolutely and uniformly to the functions @xmath122 , given by equations ( [ a : defgradgeneralsolution ] ) .
we proceed to show that the functions @xmath122 , given by the absolutely and uniformly convergent series of equations ( [ a : defgradgeneralsolution ] ) , indeed satisfy equations ( [ e : integralnearbyfllinearization ] ) . substituting equations ( [ a : defgradgeneralsolution ] ) and definition ( [ a : order2iterationkernels ] ) into equations ( [ e : integralnearbyfllinearization ] ) , @xmath187 \cdot n^{\left ( 0 \right ) } \left ( \sigma \right ) \\ % \\ \begin{split } \displaystyle f \left ( \ \lambda , p \
\right ) & = f \left ( \ 0 , p \
\right ) + f \left ( \ 0 , p \ \right ) \cdot \int_{0}^{\lambda } d\sigma \ n^{\left ( 0 \right ) } \left ( \sigma \right ) \\ \displaystyle & + f \left ( \ 0 , p \ \right ) \cdot \int_{0}^{\lambda } d\sigma \int_{0}^{\sigma } d\chi \ \sum_{n = 0}^{\infty } \ n^{\left ( n \right ) } \left ( \chi \right ) \cdot n^{\left ( 0 \right ) } \left ( \sigma \right ) \\ \end{split } % \end{array}\ ] ] expanding the third term on the rhs of equation ( [ a : satisfysolution ] ) , @xmath188 \cdot n^{\left ( 0 \right ) } \left ( \sigma \right ) \\
\end{split } \end{array}\ ] ] noting the recursion relation ( [ a : nthtermrecurrsionrelation ] ) for successive terms , ( [ a : expandedthirdterm1 ] ) becomes , @xmath189 \\ \end{split}\ ] ] substituting ( [ a : expandedthirdterm2 ] ) into ( [ a : satisfysolution ] ) , @xmath190 \\ \end{split}\ ] ] collecting terms , we recover equations ( [ a : defgradgeneralsolution ] ) , @xmath191 \\\ ] ] thus , equations ( [ a : defgradgeneralsolution ] ) , with recursion relations ( [ a : nthtermrecurrsionrelation ] ) and ( [ a : order2iterationkernels ] ) , satisfy equations ( [ e : integralnearbyfllinearization ] ) .
finally , it remains to be shown the iterated solution ( [ a : defgradgeneralsolution ] ) may be cast in closed form , given by the path - ordered exponential @xmath192 solution ( [ a : pexpreferentialgradient ] ) .
noting recursion relations ( [ a : nthtermrecurrsionrelation ] ) and ( [ a : order2iterationkernels ] ) , equation ( [ a : defgradgeneralsolution ] ) may be written , @xmath193 \\ % \\ % \displaystyle % f \left ( \ \lambda , p \
\right ) = f \left ( \ 0 , p \ \right ) \ \cdot \ \biggl [ \ i \ + \ \sum_{n = 1}^{\infty } \
\idotsint \limits_{\lambda \geq \sigma_{0 } \geq \dotsi \ge \sigma_{n-1 } \geq 0 } d\sigma_{1 } \dotsi d\sigma_{n } \
\nabla b \bigl ( \ \phi \bigl ( \sigma_{n } , p \bigr ) \
\bigr ) \ \dotsi \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \ \biggr ] \begin{split } \displaystyle & f \left ( \ \lambda , p \ \right ) = f \left ( \ 0 , p \ \right ) \cdot \biggl [ \ i + \int_{0}^{\lambda } d\sigma_{0 } \ \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \\ \displaystyle & + \sum_{n = 1}^{\infty } \ \int_{0}^{\lambda } d\sigma_{0 } \ \dotsi \int_{0}^{\sigma_{n-1 } } d\sigma_{n } \
\nabla b \bigl ( \ \phi \bigl ( \sigma_{n } , p \bigr ) \
\bigr ) \ \dotsi \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \ \biggr ] \\
\end{split } % \\
% \end{array}\ ] ] we introduce the _ product ordering operator _ @xmath194 ( also known as the _ path - ordered product operator _ ; see e.g. , ref .
@xcite @xmath22 4.2 ) , the action of which is to permute the factors @xmath195 composing the kernel of each term in equation ( [ a : oinfiteration ] ) such that the integration connectivity parameter values @xmath196 appear in order from smallest to largest .
explicitly , for any @xmath197 , the path - ordered product operator is , @xmath198 \\ % % \displaystyle % & = \bigg \lbrace \begin{array}{c c } \nabla b \bigl ( \ \phi \bigl ( \sigma_{i } , p \bigr ) \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{j } , p \bigr ) \
\bigr ) \ & \ \ \sigma_{i } < \sigma_{j } \\
\nabla b \bigl ( \ \phi \bigl ( \sigma_{j } , p \bigr ) \
\bigr ) \cdot \nabla b \bigl ( \ \phi \bigl ( \sigma_{i } , p \bigr ) \
\bigr ) \ & \ \ \sigma_{j } < \sigma_{i } \end{array } \\ % \end{split } \end{array}\ ] ] where @xmath199 is the heavyside step function , with @xmath200 . in general ,
the ( free ) lie algebra @xmath201 consisting of @xmath202 evaluated at different connectivity parameter values is non - abelian ; e.g. , for any fixed @xmath197 and @xmath203 , @xmath204 \ne 0 \\\ ] ] however , in the special cases that the ( [ e : pocommutator ] ) equals zero , the ordered product operator @xmath194 reduces to the standard product . transforming integration limits from @xmath205 , to a uniform @xmath167-cube @xmath206 for every @xmath207 , and using the product ordering operator , the @xmath169 term ( @xmath170 ) integration in equation ( [ a : oinfiteration ] ) may be rewritten , @xmath208 % \\ \displaystyle & \int_{0}^{\lambda } d\sigma_{0 } \ \dotsi \int_{0}^{\sigma_{n-1 } } d\sigma_{n } \
\nabla b \bigl ( \ \phi \bigl ( \sigma_{n } , p \bigr ) \
\bigr ) \ \dotsi \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \\ \displaystyle & = \frac{1}{n ! } \ \int_{0}^{\lambda } d\sigma_{0 } \
\dotsi \int_{0}^{\lambda } d\sigma_{n } \
\mathcal{p } \biggl ( \ \nabla b \bigl ( \ \phi \bigl ( \sigma_{n } , p \bigr ) \
\bigr ) \ \dotsi \nabla b \bigl ( \ \phi \bigl ( \sigma_{0 } , p \bigr ) \
\bigr ) \ \biggr ) \\ \end{split}\ ] ] furthermore , the nested integral on rhs of equation ( [ a : poinfiteration ] ) is simply @xmath167 independent factor integrations . hence ,
@xmath209 from equations ( [ a : oinfiteration ] ) and ( [ a : poinfiteration2 ] ) , we may formally define the _ path ordered exponential function _
@xmath192 , via its series representation , @xmath210 cases in which the ( free ) lie algebra is abelian , i.e. , ( [ e : pocommutator ] ) is equal to zero for every @xmath197 , the path ordered exponential ( [ a : poexponential ] ) reduces to the standard exponential .
equation ( [ a : poexponential ] ) is simply a formally compact form of the bracketed terms in equation ( [ a : oinfiteration ] ) .
noting @xmath117 is the identity ( given by definition ( [ e : formalrefgrad ] ) at @xmath115 ) , we get the final form of the referential gradient solution ( [ a : pexpreferentialgradient ] ) , @xmath211 for consistency , we note the formal path ordered exponential solution ( [ a : pexpreferentialgradient ] ) trivially reduces to the identity at @xmath115 .
it is a nearly trivial matter to write the coordinate representation of equations ( [ e : refgradholonomy ] ) and solution ( [ a : pexpreferentialgradient ] ) .
let @xmath127 be in a coordinate chart .
the reference shift vector may be written @xmath212 .
thus , the vector @xmath213 representing the referential gradient of the flow vector corresponding to the reference shift vector @xmath114 , evolves with the connectivity parameter @xmath118 according to , @xmath214 since the reference shift vector @xmath212 is arbitrary , we may write the first - order differential equations ( [ e : refgradholonomy ] ) for the coordinate - dependent component representation @xmath215 , @xmath216 where the reference condition @xmath217 is given by definition ( [ e : formalrefgrad ] ) at @xmath115 . furthermore , in the coordinate chart @xmath13 , solution ( [ a : pexpreferentialgradient ] ) is written , @xmath218 the kernel is the covariant derivative of the generating vector field is given by , @xmath219 where the position @xmath220 is evaluated along the flow .
the integration operation of the referential gradient solution ( [ a : pexpreferentialgradient ] ) places powerful non - trivial restrictions on this object s transformation properties . in this section ,
we explore the various conditions demanded in order that : 1 ) the components of the referential gradient transform as a tensor , 2 ) the referential gradient forms a group under translations of the connectivity parameter , and 3 ) the relationship between representations of the referential gradient under a change in integration variable .
it is important to note , due to the integration the coordinate representation of the lagrangian specification of the referential gradient solution ( [ e : refgradproofcoords ] ) is valid only to the extent that it domain of definition @xmath127 .
the first lemma makes explicit the condition under which the referential gradient components in any coordinate representation transform as a tensor under smooth changes of coordinates ; namely , so long as the set @xmath221 . _
( coordinate transformation of the referential gradient ) _ [ lem : refgrad ] let @xmath7 and @xmath222 be coordinate charts on @xmath8 . under smooth coordinate transformations
@xmath223 , the coordinate representation of the referential gradient transforms as a tensor , @xmath224 if and only if @xmath225 . to prove necessity , we assume @xmath225 , and show the transformation ( [ e : refgradcoordtransform ] ) follows from the series solution ( [ a : oinfiteration ] ) for any @xmath118 . in the chart @xmath13 , the coordinate representation of the referential gradient series solution ( [ a : oinfiteration ] )
is given by , @xmath226 where we have made use of the reference condition @xmath132 . by assumption @xmath225 , and each kernel factor covariant derivative of the generating vector field transforms under smooth coordinate transformations as , @xmath227 for every @xmath228 . making use of the identity @xmath229
, the @xmath230 product kernel transforms as , @xmath231 by induction , the @xmath169 product kernel transforms as , @xmath232 upon substitution of ( [ e : kerneltransform ] ) and ( [ e : nthorderproductkerneltransform ] ) into ( [ e : seriessolncoords ] ) , and using the fact that the connectivity parameter is independent of coordinates , equation ( [ e : refgradcoordtransform ] ) follows immediately , @xmath233 \\
% \end{split } % \\ f^{\mu}_{\nu } \bigl ( \ \lambda , x_{i } \left ( p \right ) \
\bigr ) = \lambda^{\beta}_{\nu } \ \lambda^{\mu}_{\alpha } \
f^{\alpha}_{\beta } \bigl ( \ \lambda , x_{j } \left ( p \right ) \
\bigr ) % \end{array}\ ] ] to prove sufficiency , let the chart @xmath13 cover the set @xmath234 , and let @xmath235 .
for any reference point @xmath236 , the coordinate representation of the referential gradient @xmath215 in the chart @xmath13 is well - defined for all @xmath118 .
furthermore , there exists a @xmath237 such that @xmath238 for any @xmath239 , thus equation ( [ e : refgradcoordtransform ] ) is not defined .
hence , upon shrinking @xmath103 to the extent that @xmath221 guarantees equation ( [ e : refgradcoordtransform ] ) for smooth coordinate transformations @xmath223 . whereas the previous lemma concerned the tensorial nature of the referential gradient components in any coordinate representation , the second lemma concerns the group properties of the referential gradient with respect to the connectivity parameter @xmath118 .
namely , due to the integration , the group property of the referential gradient with respect to translations of connectivity parameter is a direct consequence of the group property of the flow along which it is evaluated . _
( group structure of the referential gradient ) _ [ lem : refgradgroup ] let @xmath240 and @xmath241 .
the referential gradient forms a ( lie ) group in the connectivity parameter .
if and only if the flow satisfies @xmath243 for every @xmath244 and @xmath245 , or the representation of @xmath246 in any coordinate chart is a constant . to prove the lemma , it suffices to work in coordinates ; let @xmath127 be in a coordinate chart . to show necessity , we begin with the defining equation ( [ e : refgradholonomycoords ] ) . without loss of generality , fix @xmath247
such that @xmath248 .
noting , @xmath249 , equation ( [ e : refgradholonomycoords ] ) becomes , @xmath250 suppose the referential gradient satisfies the group condition , equation ( [ e : fgroup ] ) , then we may write , @xmath251 substituting equation ( [ e : refgradholonomycoords ] ) , @xmath252 fixing @xmath253 , we find the necessary condition that the referential gradient is a group in the connectivity parameter follows , @xmath254 equation ( [ e : necessarycondition ] ) is true for the representation of @xmath246 in any coordinate chart if @xmath243 for every @xmath244 and @xmath245 , or is constant ( independent of both the connectivity parameter @xmath118 and the reference point @xmath62 ) . to show sufficiency , we make use of the path - ordered exponential functional solution , equation ( [ e : refgradproofcoords ] ) . for fixed @xmath255 such that @xmath256 , @xmath257 the integral argument in the path - ordered exponential may be written as a sum , @xmath258 where , in the first integral on the rhs we have made the substitution @xmath259 .
since the flow @xmath56 forms a group in the connectivity parameter , the second integral on the rhs is equivalent to pivoting off of a different reference point ; namely , @xmath260 .
hence , we may write , @xmath261 by the baker - campbell - hausdorff theorem ( see appendix [ a : bchtheorem ] ) , sufficiency follows from the condition that the ( free ) lie algebra is abelian ; e.g. , the following commutator is zero , @xmath262 \\ \\
% \begin{split } % \displaystyle % = \int^{\lambda_{2}}_{0 } & d\tau \
\nabla_{\eta } \
b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \tau , q \bigr ) \
\bigr ) \ \bigr ) \ \int_{0}^{\lambda_{1 } } d\sigma \ \nabla_{\nu } \ b^{\eta}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \sigma , p \bigr ) \
\bigr ) \ \bigr ) \\ % \displaystyle % & - \int_{0}^{\lambda_{1 } } d\sigma \ \nabla_{\eta } \
b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \sigma , p \bigr ) \
\bigr ) \ \bigr ) \ \int^{\lambda_{2}}_{0 } d\tau \ \nabla_{\nu } \ b^{\eta}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \tau , q \bigr ) \
\bigr ) \\ % \\ % \end{split } % \\
% \begin{split } % \displaystyle % = \int_{0}^{\lambda_{2 } } & d\tau \
\int^{\lambda_{1}}_{0 } d\sigma \ \biggl ( \ \nabla_{\eta } \
b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \tau , q \bigr ) \
\bigr ) \ \bigr ) \ \nabla_{\nu } \
b^{\eta}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \sigma , p \bigr ) \
\bigr ) \ \bigr ) \\ % \displaystyle % & - \nabla_{\eta } \ b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \sigma , p \bigr ) \
\bigr ) \ \bigr ) \ \nabla_{\nu } \
b^{\eta}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \tau , q \bigr ) \
\bigr ) \ \bigr ) \
\biggr ) \\ % \\ % \end{split } %
\\ \displaystyle = \int_{0}^{\lambda_{2 } } d\tau \ \int^{\lambda_{1}}_{0 } d\sigma \ \biggl [ \ \nabla_{\nu } \ b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \tau , q \bigr ) \
\bigr ) , \nabla_{\nu } \ b^{\mu}_{i } \bigl ( \ x_{i } \bigl ( \ \phi \bigl ( \sigma , p \bigr ) \
\bigr ) \ \bigr ) \ \biggr ] \\ \end{array}\ ] ] the rhs of equation ( [ e : groupcommutator ] ) is zero if the kernel is zero , to wit , @xmath263 = 0 \\\ ] ] in general , equation ( [ e : groupcommutatorabelian ] ) is non - zero unless the representation of @xmath246 in any coordinate chart is : 1 ) evaluated at the same point , or 2 ) independent of the connectivity parameter and reference point .
the second case is trivial .
the first case requires the condition @xmath264 . substituting for @xmath265 the change of integration variable used in equation ( [ e : pexpdecomp ] )
, we recover the condition , @xmath266 .
lemma ( [ lem : refgradgroup ] ) simply states that the group property of the referential gradient with respect to translations of the connectivity parameter is a direct consequence of the similar group property of the flow ; the constant @xmath246 representation is the trivial case . in a follow - up paper , we explore the geometric and topological implications of this lemma .
the third lemma relates the referential gradient of the respective representations of the flow .
recall from section ( [ ss : generalconnectivitymap ] ) , the flow @xmath56 of a vector field is an equivalence class under affine transformations of the connectivity parameter . as such
, the flow may always be represented in a form @xmath267 in which the connectivity parameter measures an arc - length from the reference point via relations ( [ e : arclengthcondition ] ) , ( [ e : arclengthmcflow ] ) , and ( [ e : flowarclengthflowrelation ] ) . for ease of notation , by @xmath268 $ ] and @xmath269 $ ] we mean , respectively , the lagrangian specification of the referential gradient solution ( [ a : pexpreferentialgradient ] ) constructed from of the full generating vector field @xmath49 with flow representation @xmath56 , and that constructed from the unit generating vector field @xmath270 with arc - length flow representation @xmath267 . _
( relations between referential gradient representations ) _ [ lem : representationrelations ] let @xmath133 be a non - null generating vector field .
the representations @xmath268 $ ] and @xmath269 $ ] are related by , @xmath271 = \mathcal{p } \text{exp } \int^{l}_{0 } ds \
\biggl ( \ \nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) + \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \\ \displaystyle f \left [ \ b \ \right ] = \mathcal{p } \text{exp } \int^{\lambda}_{0 } d\sigma \ \biggl ( \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) - \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \end{array}\ ] ] furthermore , if the ( free ) lie algebras are abelian , @xmath272 = 0 \\
\int^{\lambda}_{0 } d\sigma \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ , \ \int^{l}_{0 } ds \
\frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ] = 0 \\ \end{array}\ ] ] then relations ( [ e : representationsrelatations ] ) may be written , @xmath273 = f \left [ \ b \ \right ] \
\mathcal{p } \text{exp } \biggl ( \ \int^{l}_{0 } ds \ \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \\ \displaystyle f \left [ \ b \ \right ] = f \left [ \ b \ \right ] \ \mathcal{p } \text{exp } \biggl ( \ - \int^{\lambda}_{0 } d\sigma \ \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \end{array}\ ] ] we note , in the case of a null generating vector field , @xmath274 . hence ,
the unit generating direction field @xmath270 does not exist , and therefore neither does the representation @xmath269 $ ] .
we begin by deriving relations ( [ e : representationsrelatations ] ) . by theorem ( [ thm : refgradlagrangian ] ) , each representation independently satisfies the corresponding system of equations , @xmath275 = f \left [ \ b \ \right ] \cdot \nabla b \bigl ( \ \phi \left ( \lambda , p \right ) \
\bigr ) \\ \\
\displaystyle \frac{\partial}{\partial l } \
f \left [ \ b \ \right ] = f \left [ \ b \ \right ] \cdot \nabla b \bigl ( \ \psi \left ( l , p \right ) \
\bigr ) \\ \end{array}\ ] ] with respective reference conditions , @xmath268 = i$ ] at @xmath115 , and @xmath269 = i$ ] at @xmath276 .
the full generating vector field and its unit generating vector field are related by @xmath277 .
therefore , the covariant derivative may be written , respectively , @xmath278 where act expansion is evaluated along their respective flow representations @xmath279 and @xmath91 , and in both cases the magnitude of the vector field is given by @xmath280 .
substituting relations ( [ e : gradbdecomp ] ) into the differential equations ( [ e : systemrepresentation ] ) , @xmath281 \\ \displaystyle & = \vert b \vert \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ f \left [ \ b \ \right ] \cdot \biggl ( \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) + \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \\ \begin{split } \displaystyle & \frac{\partial}{\partial l } \ f \left [ \ b \ \right ] \\
\displaystyle & = \frac{1}{\vert b \vert \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) } \ f \left [ \ b \ \right ] \cdot \biggl ( \ \nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) - \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \end{split } \end{array}\ ] ] hence , by theorem ( [ thm : refgradlagrangian ] ) , the path - ordered exponential solutions to the respective equations ( [ e : systemrepresentation ] ) and ( [ e : fullgvfrepresentation ] ) must be equivalent , @xmath282 = \mathcal{p } \text{exp } \int^{\lambda}_{0 } d\sigma \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \\ \displaystyle & = \mathcal{p } \text{exp } \
\int^{\lambda}_{0 } d\sigma \ \vert b \vert \bigl ( \ \phi \left ( \sigma , p \right ) \
\biggl ( \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) + \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \\
\begin{split } \displaystyle & f \left [ \ b \
\right]= \mathcal{p } \text{exp } \int^{l}_{0 } ds \
\nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \\ \displaystyle & = \mathcal{p } \text{exp } \
\int^{l}_{0 } \
\frac{ds}{\vert b \vert \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) } \ \biggl ( \ \nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) - \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \end{array}\ ] ] the affine transformation ( [ e : arclengthcondition ] ) remunerating the connectivity parameter is a diffeomorphism between the full generating vector field representation and the the arc length representation , @xmath283 .
hence , equations ( [ e : componentrepresentationssolutions ] ) become ( [ e : representationsrelatations ] ) , @xmath284 & = \mathcal{p } \text{exp } \int^{\lambda}_{0 } d\sigma \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \\ \displaystyle & = \mathcal{p } \text{exp } \
\int^{l}_{0 } ds \
\biggl ( \ \nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) + \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \\
\begin{split } \displaystyle f \left [ \ b \ \right ] & = \mathcal{p } \text{exp } \int^{l}_{0 } ds \
\nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \\ \displaystyle & = \mathcal{p } \text{exp } \ \int^{\lambda}_{0 } d\sigma \ \biggl ( \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) - \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \end{array}\ ] ] we proceed to prove the remaining portion of the lemma by noting , in general , the ( free ) lie algebras generated by the integral terms on the respective lhs of equations ( [ e : representationsrelatations ] ) are not abelian .
however , in the special case that the commutators , @xmath272 = 0 \\ \\ \displaystyle \biggl [ \ \int^{\lambda}_{0 } d\sigma \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ , \ -\int^{\lambda}_{0 }
d\sigma \ \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ] = 0 \\ \\ \end{array}\ ] ] the relations ( [ e : representationsrelatationdecomposition ] ) follow immediately from an application of the bch theorem ( see appendix [ a : bchtheorem ] ) to equations ( [ e : componentrepresentationssolutions ] ) , @xmath285 \\ \displaystyle & = \mathcal{p } \text{exp } \biggl ( \ \int^{l}_{0 } ds \
\nabla b \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \ \mathcal{p } \text{exp } \biggl ( \ \int^{l}_{0 } ds \ \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \psi \left ( s , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \\
\begin{split } \displaystyle & f \left [ \ b \ \right ] \\
\displaystyle & = \mathcal{p } \text{exp } \biggl ( \ \int^{\lambda}_{0 } d\sigma \ \nabla b \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \ \mathcal{p } \text{exp } \biggl ( \ -\int^{\lambda}_{0 } d\sigma \ \frac{b}{2 } \ \frac{\nabla b^{2}}{b^{2 } } \bigl ( \ \phi \left ( \sigma , p \right ) \
\bigr ) \ \biggr ) \\ \\ \end{split } \end{array}\ ] ] given a general affine transformation remunerating the connectivity parameter , lemma ( [ lem : representationrelations ] ) amounts to a change of integration variable prescription ; see equations ( [ e : componentrepresentationssolutions ] ) and equations ( [ e : componentrepresentationssolutions2 ] ) . as such
, this lemma applies to any field of scalar multiples of the generating vector field and associated flow representation . in this section
, we derive an equivalent eulerian specification of the referential gradient ; that is the sixteen , coupled , non - linear partial differential equations which govern the referential gradient field dynamics in a coordinate chart .
furthermore , there is no identified closed form general solution in the eulerian specification akin to equation ( [ a : pexpreferentialgradient ] ) .
we establish the coordinate representation of the rate of change with respect to the connectivity parameter , via the covariant derivative along the direction of the generating vector field given in coordinates .
the following proposition follows as a consequence of equation ( [ e : mcflowcoords ] ) , the rate of change with respect to connectivity parameter is given by the intrinsic directional derivative operator in the direction of the generating vector field , @xmath286 applying operator ( [ e : rocconnparam ] ) to @xmath56 yields the identity , @xmath287 recall the remark of section ( [ ss : generalconnectivitymap ] ) following equation ( [ e : mcflowcoords ] ) , in a coordinate chart @xmath13 the units of the flow @xmath56 are identified with positions @xmath288 .
hence , in the chart @xmath13 , @xmath289 in a four - dimensional spacetime coordinate chart @xmath13 , identify the generating field with the four - velocity field @xmath290 ; the associated connectivity parameter is then identified with proper time @xmath291 .
hence , equation ( [ e : rocconnparam ] ) is the intrinsic derivative operator with respect to proper time ( see e.g. , ref .
@xcite @xmath22 4.2 ) @xmath292 for simplicity , assume minkowski spacetime and cartesian coordinates ; @xmath293 and @xmath294 .
thus , the four - velocity is given by @xmath295 .
then , in the non - relativistic limit @xmath296 , equation ( [ ex:4velintrinsicderiv ] ) reduces to the standard material derivative , @xmath297 identify the generating field with the four - magnetic field given by equation ( [ ex : fourmagneticfield ] ) . for simplicity ,
assume minkowski spacetime and cartesian coordinates ; @xmath293 and @xmath294 .
hence , @xmath298 where @xmath299 and @xmath300 respectively denote the three - space electric and magnetic vector fields and @xmath301 is the three - velocity of the observer , and @xmath302 is the lorentz factor .
hence , the intrinsic derivative operator with respect to the connectivity parameter is , @xmath303 which , in the non - relativistic limit @xmath304 , reduces to the ( stationary ) directional derivative along the field , @xmath305 having identified the coordinate representation of the rate of change with respect to connectivity parameter , the following theorem is immediately obvious . _
( eulerian specification of the referential gradient ) _ [ thm : refgraddynamics ] let @xmath133 be a smooth , complete generating vector field with associated flow @xmath134 .
the eulerian specification of the referential gradient @xmath306 , for @xmath135 and @xmath136 , is the coordinate representation of the lie derivative with respect to the generating vector field , @xmath307 \\
% \displaystyle % \biggl [ \ b \bigl ( \ \phi \left ( \lambda , p \right ) \
\bigr ) \ , \ f \left ( \ \lambda , p \ \right ) \cdot \hat{h } \ \biggr ] = 0 \\\ ] ] for arbitrary reference shift direction vector @xmath308 , and all @xmath118 . moreover ,
if the metric connection is torsion - free , the eulerian specification of the referential gradient reduces to , @xmath309 the eulerian specification of the referential gradient follows upon application of proposition ( [ e : rocconnparam ] ) to equation ( [ e : directionalrefgradholonomyproof ] ) .
@xmath310 to construct the coordinate representation of equation ( [ e : dirrefgradcommutatorproof ] ) , let @xmath311 . by equation ( [ e : refgradholonomy ] ) of theorem ( [ thm : refgradlagrangian ] ) , in the coordinate chart @xmath13 the lineal element @xmath312 characterizing the lagrangian specification of the referential gradient at a distance @xmath118 from the reference point @xmath313 , is equivalent to the point @xmath314 ; hence , @xmath315 for ease of notation , in what follows we represent the general position @xmath316 at variable @xmath118 in the coordinate chart @xmath13 .
hence , the position at @xmath115 may be written as @xmath317 . in the coordinate chart @xmath13 , therefore , the @xmath57 component of equation ( [ e : dirrefgradcommutatorproof ] ) is @xmath318 expanding the covariant derivatives in equation ( [ e : dirrefgradcoorddynamics ] ) , @xmath319 where the torsion @xmath320 } \left ( \ x_{i } \
\right)$ ] .
the lhs of equation ( [ e : covderdirrefgradcoord ] ) is simply the lie bracket of the vector fields @xmath321 and @xmath322 .
hence , the @xmath323-component of the lie derivative of the vector @xmath322 with respect to the generating vector field @xmath321 is given by , @xmath324 moreover , equation ( [ thm : dirrefgradcommutator2 ] ) follows immediately for a torsion - free metric connection @xmath325 . in the practical application of theorem ( [ thm : refgraddynamics ] ) , it is useful to use the levi - civita connection
. furthermore , recall the generating vector field @xmath321 is assumed known _ a - priori _ everywhere . noting the reference shift direction vector @xmath326 is an arbitrary albeit fixed ( i.e. , coordinate - independent ) vector , the eulerian specification of the referential gradient is given by the following set of ( sixteen ) coupled first - order , partial differential equations , @xmath327 with reference condition @xmath328 .
the eulerian specification of the referential gradient in minkowski space , @xmath329 , satisfies the following set of coupled first - order partial differential equations , @xmath330^{i } \\
\displaystyle & = \frac{f^{0}_{\nu } \left ( \ x_{i } \
\right)}{c } \
\frac{\partial b^{i}_{i } \left ( \ x_{i } \
\right)}{\partial t } \\
\end{split } \end{array}\ ] ] with reference conditions at the point @xmath331 given by , @xmath332
this paper introduces the _ referential gradient of the flow _ of a vector field , a generally covariant measure of the geometric structure of the flow of a vector field in four - dimensional spacetime .
we assume _ a - priori _ the generating vector field exists , is everywhere smooth , and satisfies some set of governing evolution equations .
the mathematical formalism of flows is provided as background from which the referential gradient object is defined .
we provided the explicit relation between the referential gradient of the flow and the generating vector field from two equivalent perspectives : a lagrangian specification with respect to a generalized connectivity parameter , and an eulerian specification making explicit the evolution dynamics at each point of the manifold .
the lagrangian specification theorem ( [ thm : refgradlagrangian ] ) yields a general closed - form functional solution with respect the generating vector field in terms of a generalized connectivity parameter . while the eulerian specification theorem ( [ thm : refgraddynamics ] ) makes explicit the referential gradient dynamics at each point of the manifold . due to the integration , we prove three transformation lemmas that identify the conditions under which the referential gradient transforms as a 1 - 1 tensor , forms a group with respect to the connectivity parameter , and the proper change of variable relations between the corresponding referential gradient representations . lemma ( [ lem : refgrad ] ) proves manifest covariance of the referential gradient provided the closure of its domain of definition is contained within coordinate chart overlap .
lemma ( [ lem : refgradgroup ] ) identifies the necessary and sufficient conditions that the referential gradient forms a group with respect to translations of connectivity parameter ; that is , in general , 1 ) the group property is a direct consequence of the similar group property of the flow , or 2 ) the representation of @xmath246 in any coordinate chart is independent of the connectivity parameter and reference point .
finally , since the flow of a vector field represents an equivalence class under affine transformations of the connectivity parameter , lemma ( [ lem : representationrelations ] ) provides the proper relations associated with a change of integration variable ; in particular , between the two most natural representations of the flow . in a follow - up paper , we develop a geometric mechanics and thermodynamics using the lagrangian specification of the referential gradient . in this context
we explore the importance of lemma ( [ lem : refgradgroup ] ) with respect to topological invariants such as the linking number , as well as the consequences of relaxing the smoothness assumption of the _ a - priori _ generating vector field which lead to some powerful topological constraints on the storage , transport , and release of field energy in a system .
furthermore , in this context , lemma ( [ lem : representationrelations ] ) will prove important in application to systems in which only partial information of the generating vector field may known .
the author would like to acknowledge l. a. fisk at the university of michigan , and b. j. lynch at the university of california at berkeley for insightful discussions in the development of this research .
this research was supported , in part , by nsf grant ags-1043012 , and nasa lws grant nnx10aq61 g .
by standard taylor expansion theorem ( see e.g. , ref .
@xcite , theorem a.58 ) , any smooth vector field @xmath333 for any fixed @xmath334 , may be written , @xmath335 where @xmath336 is the covariant derivative along the vector @xmath337 . in a chart @xmath13 , the vector field @xmath338 , and the covariant derivative along the vector @xmath339
is given by , @xmath340 hence the @xmath341 component of equation ( [ ae : generaltaylorexpansion ] ) may be written , @xmath342 ds \\ \end{split } \end{array}\ ] ] it follows immediately the integral remainder is equal to zero when the vector @xmath343 .
the exponential of a matrix satisfies the identity ( dinkin s formula , see ref .
@xcite ) , @xmath344 + \sum_{n } \ a_{n } \ c_{n } \bigl ( \ x , y \ \bigr ) \
\biggr ) \\\ ] ] where the @xmath345 are constant coefficients , and the @xmath346 are homogeneous ( lie ) polynomials in @xmath347 and @xmath348 of degree @xmath167 ( i.e. , nested commutators ) . the first few @xmath349 terms are well - known , and given by ( see ref .
@xcite ) , @xmath350
\ \bigr ] + \bigl [ \ \left [ \ x , y \
\right ] , y \ \bigr ] \\ \\
\displaystyle a_{2 } = 1/24 \ \ \ & \ \ \ c_{2 } = \bigl [ \ \bigl [ \ x , \left [ \ x , y \
\right ] \ \bigr ] , y \ \bigr ] \\ \\
\displaystyle a_{3 } = 1/720 \ \ \ & \ \ \ \begin{split } c_{3 } = \bigl [ \ x , \bigl [ \ x , \bigl [ & \ x , \left [ \ x , y
\ \right ] \ \bigr ] \ \bigr ] \ \bigr ] \\
\displaystyle & - \bigl [ \ \bigl [ \ \bigl [ \ \left [ \ x , y
\ \right ] , y \ \bigr ] , y \ \bigr ] , y \ \bigr ] \\ \end{split } \\
\end{array}\ ] ] hirsch , m. w. , & smale , s. : _ differential equations , dynamical systems , and linear algebra_. pure and applied mathematics , ed .
eilenberg , s. , & bass , h. , academic press inc .
, harcourt brace jovanovich , 1974 | assuming _ a - priori _ a smooth generating vector field , we introduce a generally covariant measure of the flow geometry called the referential gradient of the flow .
the main result is the explicit relation between the referential gradient and the generating vector field , and is provided for from two equivalent perspectives : a lagrangian specification with respect to a generalized parameter , and an eulerian specification making explicit the evolution dynamics .
furthermore , we provide explicit non - trivial conditions which govern the transformation properties of the referential gradient object . |
[ [ section ] ] in this article , we use results about hodge modules on abelian varieties @xcite to prove two conjectures about zeros of holomorphic one - forms on smooth complex projective varieties .
one conjecture was formulated by hacon and kovcs @xcite and by luo and zhang @xcite ; the former partly attribute the question to carrell @xcite , and also explain why it is natural to consider varieties of general type .
[ conj : main ] if @xmath0 is a smooth complex projective variety of general type , then the zero locus @xmath1 of every global holomorphic one - form @xmath2 is nonempty .
the statement is a tautology in the case of curves ; for surfaces , it was proved by carrell @xcite , and for threefolds by luo and zhang @xcite .
it was also known to be true if the canonical bundle of @xmath0 is ample @xcite , and more generally when @xmath0 is minimal @xcite . for varieties that are not
necessarily of general type , luo and zhang proposed the following more general version of the conjecture , in terms of the kodaira dimension @xmath3 , and proved it in the case of threefolds @xcite .
[ conj : strong ] let @xmath0 be a smooth complex projective variety , and let @xmath4 be a linear subspace such that @xmath5 is empty for every nonzero one - form @xmath6
. then the dimension of @xmath7 can be at most @xmath8 .
[ [ section-1 ] ] since every holomorphic one - form on @xmath0 is the pullback of a holomorphic one - form from the albanese variety @xmath9 , it is natural to consider an arbitrary morphism from @xmath0 to an abelian variety , and to ask under what conditions the pullback of a holomorphic one - form must have a zero at some point of @xmath0 .
our main result is the following theorem , which implies both of the above conjectures .
[ thm : main ] let @xmath0 be a smooth complex projective variety , and let @xmath10 be a morphism to an abelian variety . if @xmath11 for some integer @xmath12 and some ample line bundle @xmath13 on @xmath14 , then @xmath5 is nonempty for every @xmath15 in the image of @xmath16 .
the condition in the theorem is equivalent to asking that , up to birational equivalence , the morphism @xmath17 should factor through the iitaka fibration of @xmath18 .
[ [ section-2 ] ] one consequence of is the following .
[ cor : smooth ] if @xmath10 is a smooth morphism from a smooth complex projective variety onto an abelian variety , then @xmath19 .
since @xmath17 is smooth , @xmath20 is empty for every nonzero @xmath21 .
the desired inequality then follows from because @xmath22 is injective . in particular , there are no nontrivial smooth morphisms from a variety of general type to an abelian variety ; this was proved by viehweg and zuo @xcite when the base is an elliptic curve , and then by hacon and kovcs @xcite in general .
together with subadditivity , has the following application . if @xmath10 is a smooth morphism onto an abelian variety , and if all fibers of @xmath17 are of general type , then @xmath17 is birationally isotrivial .
we have to show that @xmath0 becomes birational to a product after a generically finite base - change ; it suffices to prove that @xmath23 , in viehweg s terminology .
since the fibers of @xmath17 are of general type , we know from the main result of @xcite that @xmath24 on the other hand , @xmath25 by , and so we are done .
successive versions of this result go back to migliorini @xcite ( for smooth families of minimal surfaces of general type over an elliptic curve ) , kovcs @xcite ( for smooth families of minimal varieties of general type over an elliptic curve ) , and viehweg - zuo @xcite ( for smooth families of varieties of general type over an elliptic curve ) . over a base of arbitrary dimension ,
kovcs @xcite and zhang @xcite have shown that if the canonical bundles of the fibers is assumed to be ample , the morphism @xmath17 must actually be isotrivial .
[ [ section-3 ] ] before we get into the details of the proof , we explain how implies and hence also , which is a special case .
the statement of the conjecture is vacuous when @xmath26 , and so we shall assume from now on that @xmath27 . let @xmath28 be a birational modification of @xmath0 , such that @xmath29 is a smooth model for the iitaka fibration .
the general fiber of @xmath30 is then a smooth projective variety of dimension @xmath31 and kodaira dimension zero , and therefore maps surjectively to its own albanese variety @xcite*theorem 1 . by a standard argument @xcite*proof of theorem 13
, the image in @xmath9 of every fiber of @xmath30 must be a translate of a single abelian variety .
letting @xmath14 denote the quotient of @xmath9 by this abelian variety , we obtain @xmath32 , and the following commutative diagram : @xmath33 by construction , @xmath34 dominates @xmath35 , and so we conclude from that all holomorphic one - forms in the image of @xmath36 have a nonempty zero locus .
since this subspace has codimension at most @xmath37 in the space of all holomorphic one - forms , we see that must be true .
[ [ section-4 ] ] let @xmath10 be a morphism from a smooth projective variety to an abelian variety .
we define @xmath38 , and introduce the set @xmath39 to prove , we have to show that the projection from @xmath40 to @xmath41 is surjective ; what we will actually show is that @xmath42 maps onto @xmath41 .
[ [ section-5 ] ] in our proof , we use techniques from generic vanishing theory and from saito s theory of hodge modules . to motivate the use of generic vanishing theory ,
suppose for a moment that one could take @xmath43 in ; in other words , suppose that there was a nontrivial morphism @xmath44 by one of the generic vanishing results due to green and lazarsfeld @xcite*theorem 3.1 , the existence of a holomorphic one - form @xmath45 without zeros implies that @xmath46 for general @xmath47 .
this leads to the conclusion that @xmath48 which contradicts the fact that @xmath13 is ample , because a general translate of an ample divisor does not contain @xmath35 . in reality
, only some large power of @xmath49 will have a section after an tale cover that makes @xmath13 sufficiently divisible and so we are forced to work on a resolution of singularities of a @xmath50-fold branched covering of @xmath0 . now the additional singularities in the morphism to @xmath14 prevent the simple argument from above from going through . to fix this problem
, we use an extension of generic vanishing theory to hodge modules , developed in @xcite . with the help of this tool ,
we can still obtain when there is a holomorphic one - form @xmath45 without zeros .
[ [ section-6 ] ] [ par : vz ] our method owes a lot to the paper @xcite , in which viehweg and zuo show , among other things , that there are no smooth morphisms from a complex projective variety @xmath0 of general type to an elliptic curve @xmath51 . to orient the reader
, we shall briefly recall a key step in their proof @xcite*lemma 3.1 ; a similar technique also appears in the work of kovcs @xcite . using a carefully chosen resolution of singularities of a certain branched covering of @xmath0 , viehweg and zuo
produce a logarithmic higgs bundle @xmath52 on the elliptic curve @xmath51 , whose higgs field @xmath53 has logarithmic poles along a divisor @xmath54 ; by construction , @xmath54 contains the set of points @xmath55 where the morphism from @xmath0 to @xmath51 is singular .
they also construct a higgs subbundle @xmath56 with two properties : the restriction of the higgs field satisfies @xmath57 and @xmath58 is an ample line bundle ( where @xmath59 ) .
since any subbundle of @xmath60 has degree @xmath61 , it follows that @xmath55 can not be empty : otherwise , the image of @xmath58 under some iterate of the higgs field would be an ample subbundle of @xmath60 .
[ [ section-7 ] ] the associated graded of a hodge module may be thought of as a generalization of a higgs bundle .
recall that a hodge module on a smooth complex algebraic variety @xmath0 is a very special kind of filtered @xmath62-module @xmath63 ; in particular , @xmath64 is a regular holonomic left @xmath65-module , and @xmath66 is a good increasing filtration by @xmath67-coherent subsheaves .
the associated graded @xmath68 is coherent over the symmetric algebra @xmath69 .
it therefore determines a coherent sheaf @xmath70 on the cotangent bundle @xmath71 , whose support is the so - called characteristic variety @xmath72 of the @xmath62-module .
a longer summary of saito s theory can be found in @xcite*sections 2.12 , and of course in the introduction to @xcite .
[ [ section-8 ] ] for later use , we point out a connection between and hodge modules , having to do with the properties of the set @xmath73 . the structure sheaf @xmath67 is naturally a left @xmath65-module ; the direct image functor for @xmath62-modules takes it to a complex @xmath74 of regular holonomic @xmath75-modules . according to kashiwara s estimate for the behavior of the characteristic variety , @xmath76 where the notation is as in the following diagram : @xmath77 one consequence of saito s theory
is that @xmath67 , equipped with the obvious filtration ( @xmath78 ) , is actually a hodge module . because @xmath17 is projective , @xmath79 ( with the induced filtration ) is then a complex of hodge modules on @xmath14 , and saito s decomposition theorem
@xcite*thorme 5.3.1 gives us a non - canonical splitting @xmath80 in the derived category of filtered @xmath62-modules .
this means that the set @xmath73 contains the characteristic varieties of the hodge modules @xmath81 .
[ [ section-9 ] ] [ par : idea ] the above considerations suggest a way to generalize the construction in @xcite to the setting of .
namely , suppose that we manage to find a hodge module @xmath63 on the abelian variety @xmath14 , and a graded @xmath82-submodule @xmath83 denote by @xmath84 and @xmath70 the associated coherent sheaves on @xmath85 , and suppose that the following three conditions are satisfied : 1 .
[ en : condition-1 ] there is a morphism @xmath86 from a smooth projective variety , such that @xmath63 is a direct summand of some @xmath87 .
[ en : condition-2 ] the support of @xmath84 is contained in the set @xmath88 .
[ en : condition-3 ] for some @xmath89 , the sheaf @xmath90 is isomorphic to @xmath91 , where @xmath13 is an ample line bundle on @xmath14 .
if that is the case , we can use the generic vanishing theory for hodge modules that we developed in @xcite to show that @xmath40 projects onto @xmath41 .
[ prop : surjective ] if a hodge module @xmath63 and a graded @xmath82-module @xmath92 with the above properties exist , then the projection from @xmath40 to @xmath41 must be surjective .
let @xmath93 be the normalized poincar bundle on @xmath94 .
using its pull - back to @xmath95 as a kernel , we define the fourier - mukai transform of @xmath96 to be the complex of coherent sheaves @xmath97 on @xmath98 .
when the hodge module @xmath63 comes from a morphism to an abelian variety as in [ en : condition-1 ] , the complex @xmath51 has special properties @xcite*section 4.5 : [ en : gvt - a ] @xmath51 is a _ perverse coherent sheaf _ , meaning that its cohomology sheaves @xmath99 are zero for @xmath100 , and are supported in codimension at least @xmath101 otherwise .
gvt - b ] the union of the supports of all the higher cohomology sheaves of @xmath51 is a finite union of translates of triple tori in @xmath98 .
gvt - c ] the dual complex @xmath102 has the same properties . here
a triple torus , in simpson s terminology , means a subset of the form @xmath103 where @xmath104 is a morphism to another abelian variety . roughly speaking , [ en : gvt -
a ] is derived using a vanishing theorem by saito @xcite*lemma 2.5 , whereas [ en : gvt - b ] is derived using a result by arapura and simpson about cohomology support loci of higgs bundles @xcite , with the help of the decomposition in .
it follows that the @xmath105-th cohomology sheaf of the complex @xmath51 is locally free outside a finite union of translates of triple tori of codimension at least two .
indeed , the locus where @xmath106 is not locally free is @xmath107 which is of the asserted kind because @xmath102 also satisfies [ en : gvt - a ] and [ en : gvt - b ] . since the higher cohomology sheaves of @xmath51 are supported in a finite union of translates of triple tori of codimension at least two , the restriction of the complex @xmath51 to the subspace @xmath108 is a single locally free sheaf for general @xmath109 . translating this back into a statement on @xmath110 involves a few simple manipulations with the formula for the fourier - mukai transform ; the result is that @xmath111 is a locally free sheaf on @xmath41 . here @xmath112 and @xmath113 are the projections to the two factors , and @xmath114 is the line bundle corresponding to @xmath109 .
now suppose that the projection from @xmath40 to @xmath41 was _ not _ surjective . because the support of @xmath84 is contained in @xmath115 by [ en : condition-2 ] , the subsheaf @xmath116 would then be torsion , and therefore zero .
as @xmath41 is a vector space , we are forced to the conclusion that the corresponding graded @xmath117-module @xmath118 is also zero . but
this contradicts our assumption in [ en : condition-3 ] that @xmath119 for some @xmath89 .
in fact , if @xmath120 has no global sections , then every global section of @xmath121 has to vanish set - theoretically along @xmath35 ; but then @xmath35 is contained in a general translate of an ample divisor , which is absurd .
[ [ section-10 ] ] the remainder of the paper is devoted to constructing the two objects in under the assumptions of .
from now on , we let @xmath0 be a smooth complex projective variety of dimension @xmath122 and kodaira dimension @xmath27 .
we also assume that we have a morphism @xmath10 to an abelian variety , such that @xmath123 has a section for some @xmath12 and some ample line bundle @xmath13 on @xmath14 .
using the geometry of abelian varieties , this can be improved as follows .
after a finite tale base change on @xmath14 , we can find an ample line bundle @xmath13 such that the @xmath50-th power of @xmath124 has a section for some @xmath125 .
fix an ample line bundle @xmath126 on @xmath14 , and choose @xmath12 such that @xmath127 has a section .
let @xmath128 denote multiplication by @xmath129 ; then @xmath130 for some ample line bundle @xmath13 on @xmath14 , which clearly does the job .
since the conclusion of is unaffected by finite tale morphisms , we may assume for the remainder of the argument that the @xmath50-th power of the line bundle @xmath131 has a nontrivial section ; for the sake of convenience , we shall take @xmath12 to be the smallest integer with this property .
[ [ section-11 ] ] a nontrivial section @xmath132 defines a branched covering @xmath133 of degree @xmath50 , unramified outside the divisor @xmath134 ; see @xcite*3 for details .
since @xmath50 is minimal , @xmath135 is irreducible ; let @xmath136 be a resolution of singularities that is an isomorphism over the complement of @xmath134 , and define @xmath137 and @xmath138 .
the following commutative diagram shows all the relevant morphisms : @xmath139{drr}{h } \arrow[bend left=40]{rr}{\varphi } & x_d \rar{\pi } & x \dar{f } \\
& & a \end{tikzcd}\ ] ] by construction , @xmath135 is embedded in the total space of the line bundle @xmath140 , and so the pullback @xmath141 has a tautological section ; the induced morphism @xmath142 is an isomorphism over the complement of @xmath134 . after composing it with @xmath143 , we obtain for every @xmath144 an injective morphism @xmath145 which is actually an isomorphism over the complement of @xmath134 . pushing forward to @xmath0 , and using the fact that @xmath146 is injective , we find that the morphisms @xmath147 are also injective .
[ [ section-12 ] ] let @xmath148 be the symmetric algebra on the dual of @xmath149 , and consider the complex of graded @xmath150-modules @xmath151 placed in cohomological degrees @xmath152 , where @xmath153 and @xmath154 .
the differential in the complex is induced by the evaluation morphism @xmath155 . concretely , let @xmath156 be a basis , and denote by @xmath157 the dual basis ; then the formula for the differential is @xmath158 we use similar notation on @xmath159 as well .
[ lem : morphism ] there is a morphism of complexes of graded @xmath160-modules @xmath161 induced by the individual morphisms in .
the morphisms in commute with the differentials in the two complexes because @xmath162 for every @xmath163 .
[ [ section-13 ] ] we denote by @xmath164 the complex of coherent sheaves on @xmath165 associated with the complex of graded @xmath150-modules @xmath166 . [ lem : support ] the support of @xmath164 is equal to @xmath167 .
let @xmath168 denote the first projection ; then @xmath169 with differential induced by the tautological section of @xmath170 .
this shows that @xmath164 is equal to the pullback of the koszul resolution for the structure sheaf of the zero section in @xmath71 via the morphism @xmath171 .
in particular , @xmath172 is equal to @xmath173 , in the notation of .
[ [ section-14 ] ] the reason for introducing @xmath166 is that it is closely related to the direct image @xmath79 .
in fact , we have the following refinement of kashiwara s estimate . [
lem : laumon ] the associated graded of the hodge module @xmath81 is @xmath174 this is proved in @xcite*proposition 2.11 . in a nutshell ,
a result by laumon gives the isomorphism @xmath175 in the derived category of graded @xmath160-modules ; to deduce the assertion about individual cohomology sheaves , one has to use the fact that the complex @xmath79 is strict @xcite*thorme 5.3.1 .
this ensures that every @xmath81 is again a filtered @xmath62-module ; a priori , it is only an object of a larger abelian category , due to the fact that the derived category of filtered @xmath62-modules is the derived category of an exact category . loosely speaking
, this means that , for hodge modules , taking the associated graded commutes with direct images ; arbitrary filtered @xmath62-modules do not have this property .
[ [ section-15 ] ] we are now in a position to carry out the construction suggested in .
for the hodge module @xmath63 , we take @xmath176 ; it is really a hodge module because the morphism @xmath177 is projective @xcite*thorme 5.3.1 .
as explained in , one has an isomorphism @xmath178 as graded modules over @xmath179 .
now define @xmath92 to be the image of @xmath180 in @xmath181 , using the the morphism provided by . in this way , we get a graded @xmath82-submodule of @xmath182 as in .
[ [ section-16 ] ] it remains to check that @xmath63 and @xmath92 have all the required properties .
[ prop : construction ] @xmath63 and @xmath92 satisfy the conditions in [ en : condition-1][en : condition-3 ] .
it is obvious from the construction that [ en : condition-1 ] holds . to prove [ en : condition-2 ] , note that @xmath84 is by definition a quotient of the coherent sheaf @xmath183 the complex in parentheses
is supported in the set @xmath40 by ; consequently , the support of @xmath84 is contained in @xmath115 , as required .
to finish up , we shall argue that [ en : condition-3 ] is true when @xmath184 .
we clearly have @xmath185 and @xmath186 , and the morphism @xmath187 in is injective . after pushing forward to the abelian variety @xmath14 , we find that the resulting morphism @xmath188 is still injective .
but @xmath189 , and so @xmath90 is isomorphic to @xmath91 .
now shows that @xmath40 projects onto @xmath41 ; this means that every holomorphic one - form in the image of @xmath190 has a nonempty zero locus .
we have proved , , and .
[ [ section-17 ] ] to illustrate how our proof works , we shall briefly compare it with the argument by viehweg and zuo ( sketched in ) ; either method shows that there are no smooth morphisms @xmath191 from a variety of general type to an elliptic curve .
it is easy to see that @xmath115 dominates @xmath192 if and only if @xmath17 has at least one singular fiber , so let us quickly see how either method leads to a contradiction . in the case of viehweg and zuo ,
@xmath17 smooth implies that @xmath193 , and hence that the iterates under the higgs field @xmath194 of the ample line bundle @xmath195 are all nonzero ( because subbundles of @xmath60 have degree @xmath61 ) ; but this is not possible because @xmath196 .
in other words , smoothness of @xmath17 and positivity of @xmath195 would prevent the higgs field from being nilpotent , contradicting the fact that it is always nilpotent .
in our case , let @xmath197 be a nonzero global vector field .
the method used in produces a nonzero graded @xmath198-submodule @xmath199 which is @xmath200-torsion if @xmath201 ; but this is not possible because the ambient module is free over @xmath198 .
in other words , smoothness of @xmath17 and positivity of @xmath202 would create a nontrivial torsion submodule , contradicting the fact that the larger module is always free .
more generally , our branched covering construction is nearly identical to the one in @xcite , except that we do not need to be as careful in choosing the initial section of @xmath203 or the resolution of singularities ; this is due to the power of saito s theory .
the idea of having two objects one with good properties coming from hodge theory , the other encoding the singularities of the morphism @xmath17 also comes from viehweg and zuo .
but the actual mechanism behind the proof of based on generic vanishing theory is different , because higgs bundles and hodge modules have different properties .
[ [ section-18 ] ] we thank lawrence ein , sndor kovcs , and mircea musta for several useful discussions , and especially ein for discovering a serious mistake in an earlier attempt to prove the conjectures .
we also thank an anonymous referee for many comments that have greatly improved the presentation . [
[ section-19 ] ] during the preparation of this paper , mihnea popa was partially supported by grant dms-1101323 from the national science foundation , and christian schnell by grant dms-1331641 . | we show that every holomorphic one - form on a smooth complex projective variety of general type must vanish at some point .
the proof uses generic vanishing theory for hodge modules on abelian varieties . |
in this paper we derive exact solutions for a diffusive predator - prey system @xcite , @xmath1 where @xmath2 , @xmath3 , @xmath4 and @xmath5 are positive parameters .
subscripts @xmath6 and @xmath7 denote partial derivatives .
the equations are expressed in dimensionless variables , where scaling of space and time have been introduced so to have the equation in a simple form .
the biological meaning of the each term has been discussed in @xcite , which we briefly review .
the model is of the predator - prey kind , with @xmath8 and @xmath9 being the densities of prey and predator .
spatial redistribution of the population is governed by diffusive dynamics , with both species having the same diffusivities , @xcite . in absence of the predator ,
the temporal dynamics of the prey density is of the allee type @xcite , small populations being not viable .
presence of the predator population negatively affects the prey population . in turn , the predator population is totally dependent on prey availability , as the only term in the predator equation that represents the growth of the population is the @xmath10 term , where @xmath2 quantifies the gain in natality due to prey consumption . the parameters @xmath5 and @xmath4 represent the per capita mortality rate of prey and predator respectively , in the linear , small populations , limit .
finally , the @xmath11 is a closure relation taking into account the effects of higher trophic levels @xcite . to investigate the dynamics of the above diffusive predator - prey system the authors of ref . @xcite
have assumed the following relations between the parameters , namely @xmath12 and @xmath13 . in other words
it has been assumed that the per capita mortality rate of prey and predator are equal and the rate of biomass production at the lower level must be consistent with the rate of biomass assimilation at the upper level of the web @xcite . under this assumption eq .
( [ qqq1 ] ) reads @xmath14 in our work also we consider the predator - prey system ( [ qqq6 ] ) only .
we construct exact analytic solutions to the equation ( [ qqq6 ] ) in order to understand the properties of this model for different parametric values .
to do so we employ the @xmath0 expansion method @xcite .
this method has been applied to several nonlinear evolutionary equations . here
we demonstrate that the utility of this method in exploring dynamics of a diffusive predator - prey system .
while implementing @xmath0 expansion method to equation ( [ qqq6 ] ) we obtain exact solutions for two different wave speeds ( vide eqs .
( [ qqq17 ] ) and ( [ qqq18 ] ) ) . in both the cases we give three different types of exact solutions .
the plan of the paper is as follows . to begin with , in sec .
2 , we describe the @xmath0 expansion method . in sec . 3 , we consider eq .
( [ qqq6 ] ) and derive exact solutions of it . finally , we present our conclusion in sec .
in this section we discuss briefly the method of finding exact solutions for a system of nonlinear partial differential equations ( pdes ) using @xmath15-expansion method .
suppose that the system of nonlinear pdes is of the form @xmath16 where @xmath17 and @xmath18 are two unknown functions and @xmath19 and @xmath20 are polynomials in @xmath17 and @xmath18 and their partial derivatives .
the @xmath15 expansion method involves the following four steps .
@xmath21 let us introduce the travelling wave reduction @xmath22 in the pde ( [ aaa1 ] ) so that the latter becomes @xmath23 where prime denotes differentiation with respect to the new variable @xmath24 .
@xmath25 suppose that the solution of @xmath26 can be expressed by a polynomial in @xmath27 , that is @xmath28 with @xmath29 is the solution of the second order linear damped harmonic oscillator equation @xmath30 in the above @xmath31 and @xmath32 are constants and @xmath33 , @xmath34 .
the positive integers @xmath4 and @xmath35 can be determined by substituting ( [ aaa4 ] ) in ( [ aaa3 ] ) and considering the homogeneous balance between the highest order derivative and nonlinear terms appearing in ( [ aaa3 ] ) .
@xmath36 substituting ( [ aaa4 ] ) in ( [ aaa3 ] ) and eliminating the variable @xmath37 in the resultant equations by using ( [ aaa5 ] ) one gets two polynomials equations in @xmath27 . now equating each coefficients of @xmath38 and @xmath39 to zero
one obtains a set of algebraic equations for the parameters @xmath31 and @xmath32 . solving these algebraic equations one can get exact values for these coefficients .
@xmath40 substituting the values of @xmath31 and @xmath32 and @xmath41 and the general solution of ( [ aaa5 ] ) in ( [ aaa4 ] ) one can obtain three different types of travelling wave solutions for the given system of nonlinear pdes .
in this section , we apply the method described in the previous section to the nonlinear pdes ( [ qqq6 ] ) and construct exact solutions . substituting ( [ aaa2 ] ) into ( [ qqq6 ] )
we get the following system of ordinary differential equations ( odes ) , namely @xmath42 suppose that the solution of odes ( [ qqq9 ] ) can be expressed by a polynomial in @xmath27 which is of the form ( [ aaa4 ] ) . substituting ( [ aaa4 ] ) and their derivatives in ( [ qqq9 ] ) and performing the homogeneous balance between @xmath43 and @xmath44 and @xmath45 with @xmath46 in resultant equation we find @xmath47 and @xmath48 .
so we fix the polynomials ( [ aaa4 ] ) be of the form @xmath49 substituting the expressions ( [ qqq13 ] ) and their derivatives in ( [ qqq9 ] ) and rearranging the resultant equation in the descending powers of @xmath50 we arrive at @xmath51\big(\frac{g'}{g}\big)^3+[3\alpha_1\lambda - c\alpha_1+k\alpha_1 ^ 2+\frac{\alpha_1 ^ 2}{\sqrt{\delta}}-3\alpha_1 ^ 2\alpha_0-\alpha_1\beta_1]\big(\frac{g'}{g}\big)^2 & & \nonumber \\ + [ ( 2\mu+\lambda^2)\alpha_1-c\lambda \alpha_1-\beta\alpha_1 + 2k\alpha_0\alpha_1+\frac{2\alpha_1\alpha_0}{\sqrt{\delta}}-3\alpha_0 ^ 2\alpha_1-\alpha_1\beta_0-\alpha_0\beta_1]\big(\frac{g'}{g}\big ) & & \nonumber \\ + ( \mu \alpha_1\lambda - c\mu \alpha_1-\beta \alpha_0+\alpha_0 ^ 2+\frac{\alpha_0 ^ 2}{\sqrt{\delta}}-\alpha_0 ^ 3-\alpha_0\beta_0 ) = 0 , \qquad \qquad \qquad\end{aligned}\ ] ] @xmath52\big(\frac{g'}{g}\big)^3+[3\beta_1\lambda - c\beta_1+k\alpha_1\beta_1 - 3\delta\beta_1 ^ 2\beta_0]\big(\frac{g'}{g}\big)^2 & & \nonumber \\ + [ ( 2\mu+\lambda^2)\beta_1-c\lambda \beta_1-\beta\beta_1+k\alpha_0\beta_1+k\alpha_1\beta_0 - 3\delta\beta_0 ^ 2\beta_1]\big(\frac{g'}{g}\big ) & & \nonumber \\ + ( \mu \beta_1\lambda - c\mu \beta_1-\beta \beta_0+k\alpha_0\beta_0-\delta\beta_0 ^ 3 ) = 0 .
\qquad \qquad \end{aligned}\ ] ] equating the coefficients of @xmath38 , @xmath53 to zero in equations ( [ qqq15 ] ) and ( [ qqq16 ] ) we get the following set of algebraic equations , namely @xmath54 solving the above system of algebraic equations ( [ qqa17 ] ) and ( [ qqa18 ] ) we obtain two sets of values for the constants @xmath55 , @xmath56 , @xmath57 , @xmath58 and @xmath41 : @xmath59 @xmath60 since both the sets separately satisfy the algebraic equations in @xmath61 and @xmath62 they individually form a compatible solution . from each
set we derive an exact solution for the nonlinear pdes ( [ qqq6 ] ) . to begin with
let us take the values given in @xmath63 with the values given in @xmath64 , the solution @xmath65 reads @xmath66 it is known that the linear damped harmonic oscillator equation ( [ aaa5 ] ) admits three different types of solutions depending on the values of @xmath67 and @xmath32 , namely case 1 : @xmath68 @xmath69 case 2:@xmath70 @xmath71 case 3 : @xmath72 @xmath73 substituting ( [ qa17])-([qa19 ] ) into ( [ qqq19 ] ) we arrive at the following form of solutions , namely case 1 : @xmath74 @xmath75 ) shown for time @xmath76 ; when @xmath77 ; parameters are @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 . ] ) shown for time @xmath84 ; when @xmath85 ; parameters are @xmath86 , @xmath87 , @xmath88 , @xmath89 , @xmath82 , @xmath90 . ] ) shown for time @xmath91 ; when @xmath92 ; parameters are @xmath80 , @xmath93 , @xmath94 , @xmath82 , @xmath83 . ]
case 2 : @xmath95 @xmath96 case 3 : @xmath97 @xmath98 where @xmath99 , @xmath100 , @xmath101 and @xmath102 are arbitrary constants . for the second set of values we end up with the same form of solution with the only difference in the values in the parameters @xmath41 , @xmath5 and @xmath67 : case 1 : @xmath103 @xmath104 case 2 : @xmath105 @xmath106 case 3 : @xmath97 @xmath107 where @xmath108 , @xmath109 , @xmath101 and @xmath102 are arbitrary constants .
in this paper , we have constructed exact solutions for a diffusive predator - prey system which is modeled by a system of two coupled nonlinear pdes . using the @xmath0 expansion method ,
we have derived exact solutions for two different wave speeds .
the solutions that we have obtained are singular and can not be taken at face value describing actual situations in ecology .
notwithstanding they present a very interesting property : depending on the sign of @xmath110 , the nature of the solution changes from a single structure to a periodic one , which is akin to pattern forming systems .
we express @xmath110 in terms of the original parameters in the equation , we get @xmath111 . therefore , if @xmath112 we have a single structure like in fig.(1 ) . and if @xmath113 we have a periodic structure .
the meaning of @xmath2 is that it measures the gain in natality obtained by the predator , and @xmath5 is its mortality .
it follows that periodic structure formation comes from the strength of mortality . in a more intuitive way ,
@xmath114 is the inverse of the typical time for a predator population to decay in absence of preys .
if this time is short , we have a periodic pattern , the population continuously decaying and recovering . if this time is long , a smoother dynamics shows up .
the equations for which we could find new solutions , besides the previously known , @xcite , have the obvious drawback that matching of coefficients is necessary .
this is the same situation in most cases in the subject of exact solutions for reaction - diffusion equations , beyond the specific cases of interest in biology , @xcite .
however , the broad view insight gained remains of interest as the systems considered contain many elements of more realistic , non solvable ones . * *
rak and ms wish to thank cnpq ( brazil ) and dst ( india ) for the financial support through major research projects . | we construct exact solutions for a system of two nonlinear partial differential equations describing the spatio - temporal dynamics of a predator - prey system where the prey per capita growth rate is subject to the allee effect . using the @xmath0 expansion method ,
we derive exact solutions to this model for two different wave speeds . for each wave velocity
we report three different forms of solutions .
we also discuss the biological relevance of the solutions obtained .
reaction - diffusion equations , predator - prey system , @xmath0-expansion method . |
blazars are radio sources that have prominent synchrotron cores
. their broad - band spectra are smooth and flat from the radio up to the infrared / optical bands ( or beyond that ) , the emission is variable , sometimes on time - scales as short as minutes to hours , and is often polarised ( see e.g. urry & padovani ( 1995 ) and kollgaard ( 1994 ) for reviews ) .
the blazar population comprises bl lacs and flat spectrum radio quasars , which are , respectively , featureless optical spectrum objects and strong broad emission line objects .
it is believed that the continuum radiation of blazars is predominantly non - thermal emission from a relativistic jet pointing to the observer at a small angle to the line of sight , as proposed by blandford & rees ( 1978 ) . traditionally , apart from the radio - loudness , core - jet morphology and flatness of the broad - band spectrum , an object is only classified as a bl lac if it has a strong non - thermal optical component relative to the starlight and weak emission lines .
the emission lines are commonly quantified by the equivalent width ( ew ) of the brightest line and the strength of the non - thermal optical component is measured through the relative depression of the continuum at 4000 , the 4000 break contrast , c. the usual limits adopted are ew @xmath2 5 and c @xmath2 0.25 ( e.g. stocke et al . , 1991 ) .
these are empirical limits , based on the properties of a relatively small number of objects .
results from newer samples have been showing that the 4000 break contrast distribution of bl lacs and other `` non - bl lac '' objects does not reveal a clear bi - modality ( e.g. march et al . , 1996 ; hereinafter m96 , laurent - muehleisen et al . , 1998 ;
caccianiga et al .
, 1999 ; rector et al . , 2000 ) .
as first noted by m96 the traditional classification of a bl lac is almost certainly too restrictive , and an `` expanded '' definition that includes objects with c @xmath2 0.4 and objects with larger emission line equivalent widths was proposed .
more recently , landt et al . (
2004 ) have suggested that there is a natural separation between weak- and strong - lined sources in the [ oiii ] 5007 /[oii ] 3727 plane which clarifies blazar classification schemes .
their weak - lined class contains the majority of classical bl lacs while the strong - lined class has the majority of the broad - line blazars .
their approach is interesting but might be complicated by the fact that they include a wide range of luminosities in their sample . in this paper
we look at continuum spectral energy distributions ( sed ) of objects in a sample with a restricted range of luminosities .
+ the seds of blazars are distinctive by having a 2-hump shape , in units of @xmath3 f@xmath4 vs. @xmath3 ( f@xmath4 the flux density and @xmath3 the frequency ) .
in general , the lower frequency hump is attributed to synchrotron radiation and the second hump to inverse compton scattering by the synchrotron electrons .
the first hump is characterised by the frequency @xmath5 at which the synchrotron component begins to decline .
this frequency is directly related to a break in the synchrotron electron energy spectrum in the jet .
some bl lacs ( and flat spectrum radio quasars ) show the first energy peak in the infrared / optical wavelength range and these are often labelled low frequency peaked bl lacs ( lbls ) .
on the other hand there are those that show the first energy peak in the euv region , the high frequency peaked bl lacs ( hbls ) .
lbls are mainly found in the radio - selected samples , whereas the hbls are mainly found in the x - ray selected samples . although some authors disagreed ( e.g. polatidis , 1989 ) , for some time it was thought that there was a bi - modal lbl - hbl population .
however , the discovery of more bl lacs , mainly in the x - ray / radio surveys , has been showing that there is a continuous distribution of the peak frequencies ( e.g. perlman et al . 1998 , 2001 ) and that the apparent dichotomy was due to selection effects ( e.g. padovani et al .
( 2003 ) and references therein ) .
it is now widely accepted that bl lacs have a broad @xmath5 distribution , spanning @xmath6 5 orders of magnitude in frequency ( 10@xmath7 @xmath8 @xmath2 10@xmath9 hz ) , the extremes being populated by lbls and hbls . +
a potentially very important result has been pointed out by fossati et al .
( 1998 ) and ghisellini et al .
( 1998 , 2002 ) . they found a sequence involving the frequency
at which the peak of the non - thermal emission occurs and the radio luminosity ; the synchrotron peak position and the radio luminosity appear anti - correlated in the sense that lower peak positions are associated with the more luminous radio emitters .
donato et al . ( 2001 ) , based on x - ray spectra of blazars , also found the same anti - correlation .
this recognition of a blazar spectral sequence appears to be an important breakthrough in our understanding and it is therefore vital to check if the result holds when a wider range of parameter space is explored .
+ much of the research on blazars has been based on very bright and relatively small samples , either selected on the radio band , e.g. `` 1 jy sample '' ( stickel et al . , 1991 ) or in the x - ray band , e.g. emss ( gioia et al . , 1990 ; stocke et al . , 1991 ; maccacaro et al . ,
1994 ) . as a consequence ,
the knowledge on these objects has been built on the properties of the most powerful ones , which may not be representative of the population as a whole .
it would be , therefore , very useful to have a new radio - selected sample of relatively low radio - luminosity objects in order to investigate further the luminosity - peak frequency sequence . in particular ,
if the trend in the data is telling us about the physical properties of the blazars , and not about selection effects , the low - luminosity objects should be found to have relatively high synchrotron peak frequencies .
such an investigation is particularly timely as recent results point to a more complicated scenario , in which objects from deep surveys do not follow the anti - correlation trend ( padovani et al . , 2003 ) .
+ in order to investigate this issue and others we have been analysing a low - luminosity radio - selected sample , the 200 mjy sample ( antn et al , 2004 , hereinafter a04 ; m96 ) .
producing statistically well - defined samples of blazars is difficult since traditionally it has relied on the detection of an optical non - thermal component in the sed .
this introduces strong luminosity - dependent selection effects ( browne & march , 1993 ; march & browne , 1995 , landt et al .
2002 , 2004 ; rector & stocke , 2001 ) and many analyses of blazar population statistics can be suspect for this reason .
therefore we have been exploring an approach to sample selection focusing on the low frequency ( radio , mm and sub - mm ) properties , particularly the possession of milliarcsecond vlbi core - jet structure of the type commonly found in superluminal radio sources . with our 200 mjy core - jet sample
we have a set of objects well suited to extend our knowledge of the seds of blazar - like objects . here
we address the following questions : * what is the range of synchrotron peak frequencies amongst the low - luminosity radio selected objects ? do these objects follow the luminosity peak frequency sequence ? * how does the peak frequency influence the classification of the objects ? * are there any selection effects influencing the luminosity peak frequency sequence ?
the paper is organised as follows . in section [ sec - seds ]
we present a summary of the properties of the sources discussed here , as well as the analysis of their seds . in section [ sec - peak ]
we discuss the relation between the sed type and the optical classification of the objects .
the properties of the 200 mjy objects in the framework of the spectral sequence found by fossati et al .
( 1998 ) and ghisellini et al .
( 1998 , 2002 ) are analysed in section [ sec - specseq ] .
specifically , we fit a model consisting of a low frequency power law plus a higher frequency parabola to both the 200 mjy seds and those from other samples in order to look for systematic trends .
the main points of this work are summarised in section [ sec - conclusion ] . throughout the paper
we assume h@xmath10 = 65 kms@xmath11 mpc@xmath11 and @xmath12=0 .
1st column 1950.0 iau name of the object , 2nd column is the radio morphology , 3rd column is the optical classification , 4th column is the sed type , 5th column is the brightness temperature . _
radio morphology classification key _ : ( * c+j * ) core+jet vlba structure according to : 1 .
bondi et al .
( 2001 ) 2 .
bondi et al .
( 2004 ) 3 .
gabuzda et al .
( 1998 ) 4 .
gabuzda et al .
( 1996 ) 5 .
gabuzda & cawthorne ( 1996 ) 6 .
gabuzda et al .
( 1989 ) 7 .
henstock et al . ( 1995 ) 8 .
kollgaard et al .
( 1996 ) 9 .
nrao vlba calibrators list 10 .
pearson et al . ( 1993 ) 11 .
piner et al . ( 1999 ) 12 .
taylor et al .
( 1994 ) 13 .
usno astrometric vlbi database 14 .
venturi et al.(1993 ) 15 .
the object 1744 + 260 shows core - jet morphology at merlin resolution ( augusto et al . , 1998 ) .
taylor et al .
_ optical classification keys _ : * bll*=bl lac , * blc*=bl lac candidate , * syf1*=seyfert-1 type , * syf2*=seyfert-2 type , * peg*=passive elliptical galaxies
. _ sed classification keys _ : * bpl+ib * broadband flat spectra with one or more bumps , * bpl * broken power - law broadband spectra .
the brightness temperatures where computed from vlbi maps , at @xmath13 15 ghz or @xmath14 5 ghz presented in : 1 .
kellermann et al .
( 1998 ) 2 .
bondi et al .
( 2004 ) 3 .
henstock et al . (
taylor et al .
( 1994 ) [ cols="<,^,^,^,^,^ " , ] the objects analysed in this paper are a subsample of the 200 mjy sample ( a04 , m96 ) .
this is a nearby radio - selected sample ( most of the objects have z@xmath2 0.2 ) , that comprises objects with core - dominated morphology at 8.4 ghz , with s@xmath15 @xmath16 200 mjy , flat radio spectra between 1.4 and 5 ghz ( @xmath17 @xmath18 0.5 ) and r @xmath19 17 mag .
+ the objects under consideration in the present work are presented in table [ tabela ] .
they were selected on the basis of the results from the multiwavelength study of the 200 mjy sample , presented in a04 .
that study has revealed a subset of objects with uniform radio / mm properties , flat broadband spectra , some being classified as bl lac objects but some having different optical classifications , as follows : 1 . from the analysis of the seds of the 200 mjy sample as a whole ( a04 ) , it became clear that there was a subset of objects which had , at least at wavelengths down to the sub - mm , flat broadband spectra ( in units of flux density vs frequency ) before steepening , and they have been classified as broken - power - law type objects ( * bpl * ) in table 1 .
there are a few objects showing an infrared bump that we interpreted as dust emission , and they are listed in table 1 as * bpl+ib * ( broken - power - law + infrared bump ) .
2 . all the objects have a * core - jet * radio structure when observed with vlbi techniques ( bondi et al 2004 , 2001 and references in table 1 ) .
this ensures that we are dealing with a set of objects in which the radio emission is mainly from the active nucleus and not from compact lobes , as would be the case with sources with cso - like radio structures .
in addition we note that the possession of core - jet vlbi radio structures is in general a good circumstantial evidence that we are dealing with emission from relativistic jets .
for example , britzen et al .
( britzen , personal communication ) have made extensive vlbi observations of a complete sample of 297 flat spectrum radio sources ( cjf : taylor et al .
1996 ) and they find that @xmath690% of those cjf sources with core jet structures show superluminal motion .
+ for most of the objects the brightness temperatures could be estimated , and they are presented in table 1 .
the temperatures were computed according to : + + t@xmath20= 1.8@xmath2110@xmath22 @xmath23 } } } { \nu^2 \ { \mbox { \small [ ghz ] } } \ \ \theta^2}$ ] + with @xmath24 being the source size in milliarcsec ( caccianiga et al .
note that t@xmath20 is a lower limit , as the size of the emitting region is most probably smaller than the beamsize .
all the objects have brightness temperatures larger than @xmath25 k , showing that the emission is certainly non - thermal in origin .
more interesting is the fact that the t@xmath20 of bl lacs and of the other objects are very similar .
3 . based on optical spectral properties of the objects , in particular the 4000 break contrast and emission line strength , they are classified as ( see a04 and m96 ) : + @xmath26 bl lac objects ( * bll * ) , objects that obey the `` classical '' bl lac definition : c @xmath19 0.25 and ew @xmath19 5 .
+ @xmath26 bl lac candidates ( * blc * ) , objects with relatively small break contrast and ew : 0.25 @xmath2 c @xmath2 0.4 and ew @xmath2 40 .
+ @xmath26 seyfert - like objects , objects with relatively strong broad ( * sy1 * ) or narrow ( * sy2 * ) emission lines : ew @xmath27 40 .
+ @xmath26 passive elliptical galaxies ( * pegs * ) , objects with high contrast and weak emission lines : c @xmath16 0.40 and ew @xmath2 40 . the multiwavelength study of the 200 mjy sample has revealed an homogeneous subset of objects which have many of the properties expected of bl lacs but not all of which met the traditional bl lac selection criteria . as it can be inferred from table [ tabela ] all the objects that are optically classified as bl lacs
have seds which fall into the category bpl .
more interesting is the fact that objects other than bl lacs look similar to bl lacs up to relatively high frequencies . to investigate this further , the seds of the objects in table [ tabela ] were model fitted to estimate the peak frequency , and look for any statistical difference between bl lacs and non - bl lac objects .
we assumed the fossati et al .
( 1997 ) model , that consists of a power - law , to represent the low frequencies , plus a parabola for the higher frequencies .
the model differentiates two frequencies : up to a certain frequency the radiation is taken to be optically thick and it is represented by the power law , then the emission becomes optically thin and it is represented by the parabolic function with the maximum occurring at a frequency @xmath5 .
more complex and physically based models have been considered by others ( e.g. padovani et al 2003 ) .
but , as in our case the aim is to compare the peak frequencies amongst objects , it is sufficient to use a convenient analytical function for the modelling .
the same protocol was used for all the objects .
the flux densities were k - corrected , assuming the spectral indices from sambruna et al .
( 1996 ) and in the computation of the luminosities a redshift of 0.3 was assumed if there was no redshift available for the object . when it was clear that the near infrared / optical emission was predominantly starlight ( not agn emission ) , the data points were ignored during the fitting process .
we note that the optical data point is our estimate of the non - thermal emission , based on the 4000 break contrast as described in a04 .
figure [ figuras ] shows the seds of four objects , each representing a different optical class : 1055 + 567 is a bl lac , 0651 + 428 is a bl lac candidate , 0125 + 487 is a seyfert-1 like object with blazar characteristics and 1241 + 735 is a peg .
black area represent the @xmath5 distribution of known bl lac objects , width=302 ] the seds were analysed in the manner described in the previous section and the distribution of the synchrotron peak frequencies ( @xmath5 ) is shown in figure [ peaks ] . we draw attention to the fact that the @xmath5 are widely and fairly _ continuously _ distributed from @xmath6 10@xmath28 to 10@xmath29 hz with the well - known bl lacs ( c @xmath2 0.25 & ew @xmath2 5 ) and other objects ( c @xmath27 0.25 & ew @xmath27 5 ) occupying overlapping regions .
the @xmath5 distribution strongly points to the conclusion that we are dealing with a single group of objects drawn from the same population ( otherwise one would not expect a continuous distribution ) , containing a wide spread of non - thermal cut - offs , ranging over 4 orders of magnitude in frequency .
+ in the lower frequency tail of the distribution , we suggest that we are seeing the red end of the blazar sequence and , if one wished to maintain the tradition of labelling objects by where their spectral peaks lie , one might call the objects with @xmath5 lying at frequencies smaller than the near - infrared as very low - frequency - peaked bl lacs , or vlbls .
we do not , however , believe it helpful to make this distinction and wish , rather , to emphasise the similarities . these objects : + @xmath26
all have flat broad - band spectra up to frequencies of @xmath6100 ghz or higher .
+ @xmath26 all have radio structures that consist of compact vlbi cores plus jets .
+ how do the values of @xmath5 relate to the traditional optical classifications of bl lacs , seyfert - like objects and pegs ?
it is of course of interest that not all the objects listed in table [ tabela ] , and that we would like to label as blazars , meet the usual `` bl lac '' criteria . in figure [ cpeak ] the values of @xmath5 are plotted against the 4000 break contrasts .
as expected , the @xmath5 distributions of bl lacs are different from those of seyfert - like , bl lac candidates and peg objects ; bl lacs have @xmath5 in the near - infrared / optical / x - ray bands and the rest of objects have their @xmath5 located mainly between the long and near infrared bands . of course it should be no surprise that the strength of the synchrotron component in the optical band ( which is the basis of normal bl lac classification ) is dependent on the frequency at which the synchrotron emission begins to decline .
note that the importance of this effect on traditional classifications decreases with increasing non - thermal luminosity . in the case of powerful objects ,
their non - thermal optical component is strong enough to compete with the starlight ( from the host galaxy ) and to fill the 4000 break , even in the case of a relatively low @xmath5 .
that is , when there is a wide range of peak frequencies , as we believe is the case , it can lead to a selection - induced anti - correlation between @xmath5 and luminosity ( see below ) .
this implies that a very low luminosity bl lac is only recognised as such if it has a very high frequency peak , whereas for a high luminosity object recognition is possible for a wide range of peak frequencies .
a crucial question is whether or not the systematic shift of the average synchrotron peak frequency with luminosity , pointed out by fossati et al . ( 1998 ) and donato et al .
( 2001 ) is confirmed for our 200 mjy sample of low - luminosity radio - selected objects .
this spectral sequence is important because , if true , it represents a great empirical simplification which is backed by an elegant physical explanation in terms of different inverse compton losses occurring in the jet due to ambient photon densities which vary with object luminosity ( ghisellini et al .
1998 ; georganopoulos et al . , 2001 ) .
+ unfortunately there are also selection effects which might give rise to a spurious correlation of the form that is observed ( see also perlman et al .
this is because the result is based on all available blazar seds taken from a whole range of samples with different selection criteria .
one of the possible selection effects
i.e. that recognition of bl lacs with low cut of frequencies is more difficult at low luminosities than high luminosities was already pointed out in section [ sec - peak ] .
another possible selection effect is that in the samples analysed by fossati et al .
, the lower luminosity objects are mainly represented by the x - ray selected objects ( e.g. the slew sample ) and therefore likely to be mainly hbl - type . on the other hand ,
the higher luminosity objects represented by the 1 jy and 2 jy samples , are mainly radio - selected objects , and therefore the majority are of lbl - type .
thus a correlation of the form observed could arise from the type of objects available for analysis .
+ the 200 mjy being a radio - selected sample , but of relatively low average radio - luminosity , occupies a critical range of parameter space and is thus well suited to enable us to check the reality of the radio - luminosity / spectral sequence : if they follow the expected luminosity trend , they should have high peak frequencies . to ensure that we are comparing like with like
, the seds of the slew and 1 jy objects listed in fossati et al ( 1998 ) were re - fitted in an identical manner to the 200 mjy objects ( section [ sec - sedfitting ] ) . figure [ lum5nupeak ] presents the peak frequencies against radio luminosities at 5 ghz for the 1 jy , slew and 200 mjy objects ( the latter plotted in different symbols depending on the optical classification ) .
the 200 mjy objects fill a region of intermediate radio luminosity relative to the 1 jy and slew objects , showing objects with low radio luminosities and low peak frequencies .
this is best shown in figure [ lumpeak ] that presents the peak frequency distributions of the objects and their respective radio luminosities at 5 ghz .
it is clear that the distributions of _ radio luminosities _ in the slew survey and the 200 mjy samples are very similar .
thus , on the basis of the proposed spectral sequence , we would expect the 200 mjy objects and slew objects to have the same distribution of _ peak frequencies_. it is , however , evident from figure [ lumpeak ] that they do not ( k - s test rejects this hypothesis at a significance level of @xmath30 ) .
the 200 mjy peak frequency distribution is more similar to that of the 1 jy sample , suggesting that the method of selection ( i.e. radio or x - ray ) has more influence on the peak frequency distribution than does the intrinsic luminosity .
+ we suggest that when there is a population with a broad distribution of spectral shapes , a low frequency band of selection biases the samples to objects that are stronger at low frequencies and a high frequency band of selection biases the samples to objects that are strong at high frequencies .
sources strong at low frequencies tend to have low peak frequencies and visa versa . in radio astronomical parlance ,
selecting at high frequencies is the way to select flat spectrum sources and selecting at low frequencies produces lots of steep spectrum sources ( kellermann et al . , 1968 ) . also , as argued in section [ sec - peak ] , there is a bias against recognising as traditional bl lacs , low luminosity objects with synchrotron peak frequencies in the infrared and lower frequencies , which also acts to reinforce the fossati et al . correlation .
how valid is our approach to defining blazar samples ?
the advantage of ignoring the optical emission is that it is much less subject than other methods to luminosity - dependent selection effects , something that is quite important when comparing the statistical properties of samples with a range of luminosities .
+ perhaps the most controversial thing we have done is to include objects with seyfert - like spectra in our 200 mjy sub - sample .
we point out , however , that three of the eight seyfert - like objects , namely 0125 + 487 , 1646 + 466 and 2116 + 81 , are well - established blazars ( jackson & march , 1999 ; perlman et al . , 1998 ) .
if we exclude the remaining five objects ( see figure [ lumpeaknosy ] ) it is still clear that the peak frequency distribution for the 200 mjy sample objects is still much more like that of the 1jy sample of high luminosity objects than that of the lower luminosity slew survey objects ( k - s test rejects the hypothesis that 200 mjy and slew samples have similar peak frequency distributions at a significance level of @xmath31 ) .
+ we thus believe that we have shown that amongst low - luminosity radio sources there are nuclear synchrotron emitters with peaks in their seds falling at lower frequencies than have been found before .
these objects may not meet the criteria usually required for classification as a blazar but we see no good physical reason why they should be denied that status and argue that our `` non - optical '' method of sample selection has significant advantages over the tradition methods .
based on the assumption that these are really blazars , we have looked at where these objects fit on the sequence of blazar seds with luminosity investigated by fossati et al .
( 1998 ) and ghisellini et al .
we find that they occupy part of the luminosity / peak frequency plane hitherto almost empty and thus much weaken the significance of the correlation found by fossati et al .
( 1998 ) and ghisellini et al .
our results are consistent with the results from padovani et al .
we compare peak frequency distributions of samples selected in diverse ways and argue that the distribution of peak frequencies for each sample has much more to do with the initial sample selection frequency than it has to do with intrinsic agn luminosity .
we thank alessandro caccianiga and maria march for useful discussions , and eric perlman for helpful comments .
snia antn acknowledges the financial support from the portuguese fundao para a cincia e tecnologia through the grant sfrh / bpd/5692/2001 , the european commission , tmr programme , research network contract erbfmrxct96 - 0034 ceres " and the financial support from jodrell bank observatory visitor grant .
ian browne acknowledges the financial support from the portuguese fundao para a cincia e tecnologia through the project eso / fnu/43803/2001 .
perlman , e. s. , padovani , p. , landt , h. , stocke , j. t. , costamante , l. , rector , t. , giommi , p. & schachter , j. f. 2001 , asp conf .
ser . 227 : blazar demographics and physics , 200 . edited by paolo padovani and c. megan urry | we analyse a group of radio sources , a subset of the 200 mjy sample , all of which have core - jet radio structures measured with vlbi and have flat spectra stretching from the radio to the millimetre / sub - millimetre band .
thus the objects have most of the properties expected of blazars .
however , they display varied optical properties ranging from `` seyfert - like '' objects , through bl lac objects , to `` normal '' elliptical galaxies .
we investigate the distribution of synchrotron peak frequencies in their spectral energy distributions ( seds ) and find a broad distribution between @xmath0 and @xmath1 hz .
our conclusion is that we should consider virtually all objects in the sample as blazars since much of the diversity in their classification based on traditional optical criteria arises from differences in the frequency at which the non - thermal emission begins to decline . specifically , an object is only classified as bl lac when its peak frequency falls in the near ir / optical range .
we determine peak frequencies using the same method for objects from other blazar samples .
an important result is that our objects do not follow the blazar spectral sequence proposed by fossati et al . and ghisellini et al .
in which , on average , peak frequencies increase as the radio luminosity decreases .
most of our low radio - luminosity sources have peaks in their seds at low frequencies , not at the expected high frequencies .
we suggest that at least part of the systematic trend seen by fossati et al . and ghisellini et al .
results from selection effects .
[ firstpage ] galaxies : active - bl lacertae objects : general . |
coalescing neutron star ( ns ) binaries are likely to be one of the most important sources of gravitational radiation for the ground - based laser - interferometer detectors in ligo @xcite , virgo @xcite , geo600 @xcite , and tama @xcite
. these interferometers are most sensitive to gw signals in the frequency range from about @xmath5 to @xmath6 , which corresponds to the last several thousand orbits of the inspiral . during this period ,
the binary orbit is decaying very slowly , with the separation @xmath7 and phase @xmath8 following the standard theoretical treatment for the inspiral of two point masses ( see , e.g. , @xcite ) .
theoretical templates for the corresponding quasi - periodic gw signals covering an appropriate range of values for the ns masses , as well as orbital phases and inclination angles , can be calculated to great precision ( see , e.g. , @xcite and references therein ) , and template matching techniques can therefore be used to extract signals from noisy interferometer data . when the binary separation @xmath7 has decreased all the way down to a few ns radii ,
the system becomes dynamically unstable @xcite and the two stars merge hydrodynamically in @xmath9ms .
the characteristic gw frequency of the final burst - like signal is @xmath10 , outside the range accessible by current broadband detectors .
these gw signals may become detectable , however , by the use of signal recycling techniques , which provide increased sensitivity in a narrow frequency band @xcite .
these techniques are now being tested at geo600 , and will be used by the next generation of ground - based interferometers .
point - mass inspiral templates break down during the final few orbits . instead , 3d numerical hydrodynamic calculations are required to describe the binary merger phase and predict theoretically the gw signals that will carry information about the ns equation of state ( eos ) .
the first hydrodynamic calculations of binary ns mergers in newtonian gravity were performed by nakamura , oohara and collaborators using a grid - based , eulerian finite - difference code @xcite .
rasio and shapiro ( @xcite , hereafter rs ) later used lagrangian sph calculations to study both the stability properties of close ns binaries and the evolution of dynamically unstable systems to complete coalescence .
since then , several groups have performed increasingly sophisticated calculations in newtonian gravity , exploring the full parameter space of the problem with either sph @xcite or the eulerian , grid - based piecewise parabolic method ( ppm ) @xcite , and focusing on topics as diverse as the gw energy spectrum @xcite , the production of r - process elements @xcite , and the neutrino emission as a possible trigger for gamma - ray bursts @xcite .
some of these newtonian calculations have included terms to model approximately the effects of the gravitational radiation reaction @xcite .
the first calculations to include the lowest - order ( 1pn ) corrections to newtonian gravity , as well as the lowest - order dissipative effects of the gravitational radiation reaction ( 2.5pn ) were performed by shibata , oohara , and nakamura , using an eulerian grid - based method @xcite .
more recently , the authors ( @xcite , hereafter paper 1 and paper 2 , respectively ) , as well as ayal et al.@xcite , have performed pn sph calculations , using the pn hydrodynamics formalism developed by blanchet , damour , and schfer ( @xcite , hereafter bds ) .
these calculations have revealed that the addition of 1pn terms can have a significant effect on the results of hydrodynamic merger calculations , and on the theoretical predictions for gw signals .
for example , paper 1 showed a comparison between two calculations for initially corotating , equal - mass binary ns systems . in the first calculation ,
radiation reaction effects were included , but no 1pn terms were used , whereas a complete set of both 1pn and 2.5pn terms were included in the second calculation . it was found that the inclusion of 1pn terms affected the evolution of the system both prior to merger and during the merger itself .
the final inspiral rate of the pn binary just prior to merger was much more rapid , indicating that the orbit became dynamically unstable at a greater separation .
additionally , the gw luminosity produced by the pn system showed a series of several peaks , absent from the newtonian calculation .
most previous hydrodynamic calculations of binary ns mergers have assumed corotating initial conditions , and many modeled the stars as initially spherical .
however , real binary ns are unlikely to be described well by such initial conditions . just prior to contact
, tidal deformations can be quite large and the stars can have very nonspherical shapes @xcite .
nevertheless , because of the very low viscosity of the ns fluid , the tidal synchronization timescale for coalescing ns binaries is expected to always be longer than the orbital decay timescale @xcite .
therefore , a corotating state is unphysical .
in addition , at large separation , the ns in these systems are expected to be spinning slowly ( see , e.g. , @xcite ; observed spin periods for radio pulsars in double ns systems are @xmath11ms , much longer than their final orbital periods ) .
thus , the fluid in close ns binaries should remain nearly _ irrotational _ ( in the inertial frame ) and the stars in these systems can be described approximately by irrotational riemann ellipsoids @xcite . in paper 2 , we showed that nonsynchronized initial conditions can lead to significant differences in the hydrodynamic evolution of coalescing binaries , especially in the amount of mass ejected as a result of the rotational instability that develops during the merger .
there are many difficulties associated with numerical calculations of binary mergers with irrotational initial conditions , especially in pn gravity .
foremost of these is the difficulty in preparing the initial , quasi - equilibrium state of the binary system . in synchronized binaries ,
the two stars are at rest in a reference frame which corotates with the system , and thus relaxation techniques can be used to construct accurately the hydrostatic equilibrium initial state in this corotating frame ( see @xcite and paper 1 ) . for irrotational systems , no such frame exists in which the entire fluid would appear to be in hydrostatic equilibrium .
instead , one must determine self - consistently the initial velocity field of the fluid in the inertial frame .
otherwise ( e.g. , when simple spherical models are used ) , spurious oscillations caused by initial deviations from equilibrium can lead to numerical errors .
this is especially of concern in pn gravity , where the strength of the gravitational force at any point in the ns contains terms proportional to the gravitational potential and pressure at 1pn order .
another serious problem for hydrodynamic calculations with irrotational initial conditions is the issue of spatial resolution and numerical convergence . as was first pointed out in rs2 , in a frame corotating with the binary orbital motion ,
irrotational stars appear counterspinning so that , when they first make contact during the coalescence , a vortex sheet is formed along the interface .
this tangential discontinuity is kelvin - helmholtz unstable at all wavelengths@xcite .
the sheet is expected to break into a turbulent boundary layer which propagates into the fluid and generates vorticity through dissipation on small scales .
how well this can be handled by 3d numerical calculations with limited spatial resolution is unclear .
the outline of our paper is as follows .
section ii presents a summary of our numerical methods , including a brief description of our pn sph code , and an explanation of the method used to construct irrotational initial conditions in pn gravity .
additionally , we give the parameters and assumptions for all new calculations discussed in this paper .
section iii presents our numerical results based on sph calculations for several representative binary systems , all with irrotational initial configurations , but varying mass ratios and ns eos . to test our numerical methods
, we also study the effects of changing the initial binary separation and the numerical resolution .
section iv presents gw energy spectra calculated from the runs in this and previous papers , as well as a discussion of how the measurement of spectral features could constrain the ns eos . a summary of our pn results and possible directions for further research , including the possibility of fully relativistic sph calculations ,
are presented in section v.
all our calculations were performed using the post - newtonian ( pn ) sph code described in detail in papers 1 and 2 .
it is a lagrangian , particle - based code , with a treatment of relativistic hydrodynamics and self - gravity adapted from the pn formalism of bds . as in our previous work
, we use a hybrid 1pn/2.5pn formalism , in which radiation reaction effects are treated at full strength ( i.e. , corresponding to realistic ns parameters ) , but 1pn corrections are scaled down by a factor of about 3 to make them numerically tractable ( see below ) .
we did not perform any new newtonian calculations for this paper .
all the poisson - type field equations of the bds formalism are solved on grids of size @xmath12 , including the space for zero - padding , which yields the proper boundary conditions .
shock heating , which is normally treated via an sph artificial viscosity , was ignored , since it plays a negligible role in binary coalescence , especially for fluids with a very stiff eos .
all runs , except those used to study the effects of numerical resolution on our results , use @xmath13 sph particles per ns ( i.e. , the total number of particles @xmath3 ) , independent of the binary mass ratio @xmath14 . the number of sph particle neighbors is set to @xmath15 for all runs with @xmath3 . unless indicated otherwise
, we use units such that @xmath16 , where @xmath17 and @xmath18 are the mass and radius of one ns . in calculations for unequal - mass binaries
, we use the mass and radius of the primary . as in our previous papers , we compute the gravitational radiation reaction assuming that the speed of light @xmath19 in our units , which corresponds to a neutron star compactness @xmath20 . for a standard ns mass of @xmath21
this also corresponds to a radius @xmath22 km .
the unit of frequency ( used throughout sec .
iv ) is then @xmath23 in order to keep all 1pn terms sufficiently small with respect to newtonian quantities , we calculate them assuming a somewhat larger value of the speed of light , @xmath24 , which would correspond to ns with @xmath25 . as in our previous work , all calculations in this paper use a simple polytropic eos , i.e. , the pressure is given in terms of the mass - energy density by @xmath26 .
we use @xmath1 to represent a typical stiff ns eos , and @xmath0 to represent a somewhat softer eos .
the values of the polytropic constant @xmath27 for all our ns models ( determined by the mass - radius relation ) are the same as those used in paper 2 ( see table 1 therein ) .
models for single ns are taken from papers 1 and 2 .
it should be noted that models of ns with identical masses and polytropic constants but different numbers of sph particles may have slightly different effective radii , due to numerical resolution effects near the stellar surface .
it is not easy to define a meaningful and accurate origin of time for merger calculations .
the moment of first contact between the two stars is difficult to determine accurately , since it involves the smoothing lengths of particles located near the surface of each ns , where the method is least accurate .
therefore , following the convention in our earlier papers , we set the absolute timescale for each run by defining the moment of peak gw luminosity to be at @xmath28 in our units
. many runs therefore start at @xmath29 , but this is merely a matter of convention .
to construct irrotational binary configurations in quasi - equilibrium , we start from our relaxed , equilibrium models for single ns ( paper 2 , sec .
these models are transformed linearly into irrotational triaxial ellipsoids , with principal axes taken from the pn models of lombardi , rasio , and shapiro for equilibrium ns binaries @xcite .
for example , for an equal - mass system containing two @xmath1 polytropes with an initial separation @xmath30 , we find from their table iii that the principal axes are given by @xmath31 , @xmath32 , and @xmath33 , where @xmath34 , @xmath35 , and @xmath36 are measured along the binary axis ( x - direction ) , the orbital motion ( y - direction ) , and the rotation axis ( z - direction ) , respectively .
for an initial separation @xmath37 , we find @xmath38 , @xmath39 , and @xmath40 . for the initial velocity field of the fluid we adopt the simple form assumed for the internal fluid motion in irrotational riemann ellipsoids@xcite , @xmath41 where @xmath42 is the orbital angular velocity .
it is easy to verify that this initial velocity field has zero vorticity in the inertial frame . calculating self - consistently the correct value of @xmath42 corresponding to a quasi - equilibrium circular orbit proves to be much more difficult in pn gravity than in newtonian gravity , especially for irrotational configurations . in newtonian gravity ,
the gravitational force between the two stars is independent of the magnitude of the tidal deformation of the bodies to lowest order .
thus , even if the initial configuration of matter in the binary is slightly out of equilibrium , the orbit calculated for the two stars will still be almost perfectly circular .
in pn gravity , the situation is quite different . from appendix
a of paper 1 , we see that the gravitational acceleration , in our pn formalism , denoted there as @xmath43 , is calculated as @xmath44)\nabla u_*+\ldots,\ ] ] where @xmath45 is the the 1pn - adjusted velocity , and @xmath46 , @xmath47 , and @xmath48 are the pressure , rest - mass density , and gravitational potential , respectively , as defined in the bds formalism ( note that @xmath48 as defined here is a positive quantity ) .
the quantity in brackets is the 1pn correction to the gravitational force , and depends explicitly on the initial thermodynamic state of the ns .
small oscillations of each star about equilibrium result in errors when calculating the gravitational force felt by each component of the binary . in addition , the gravitational force is affected by the orbital velocity of the ns .
thus , when we try to calculate the orbital velocity from the centripetal accelerations of the respective ns , as in eq . ( 1 ) of paper 2 , @xmath49 where @xmath50 and @xmath51 are the center - of - mass velocities of the primary ( located initially on the positive x - axis ) and the secondary ( located on the negative x - axis ) , respectively , we face the problem that the rhs of the equation is itself a function of @xmath42 . for an initially synchronized system
, we solve this problem by relaxing the matter in a frame corotating with the binary , and in the process find the equilibrium value of @xmath42 self - consistently ( see paper 1 , appendix a ) .
for an irrotational binary , this is not possible , since we can not describe a priori the proper velocity field toward which the matter must relax .
we solve the problem instead by a more direct method . realizing that any simple approximation of the orbital velocity should be near the correct pn value
, we alter the expression for @xmath42 to include a correction factor which , we hope , will remove the effect of any small deviations away from equilibrium on the gravitational accelerations felt by each ns . therefore , we introduce a parameter @xmath52 such that @xmath53 for each value of @xmath52 , we iterate eqs .
[ eq : vx ] , [ eq : vy ] , and [ eq : kap ] over @xmath42 to determine a self - consistent velocity field at @xmath54 .
we then perform trials for each value of @xmath52 , i.e. , we perform purely dynamical integrations ( with radiation reaction turned off ) for about one full orbit , until we find one which produces a nearly circular orbit ( such that @xmath7 changes by no more than @xmath55 ) .
the results of one such set of calculations , for equal - mass ns with a @xmath1 eos and an initial separation @xmath30 , is shown in the top panel of fig .
[ fig : rvstkap ] .
we see that for @xmath56 , which works extremely well for all runs performed in newtonian gravity , the orbital angular velocity is too small , and the pericenter of the orbit lies within the dynamical stability limit , causing the system to merge ( artificially ) within the first orbital period . for @xmath57
, we see that the orbit is elliptical , with our initial state corresponding to pericenter . for a value of @xmath58
we obtain a very nearly circular orbit .
a residual small - amplitude oscillation is shown in greater detail in the middle panel of fig .
[ fig : rvstkap ] .
we see that it occurs not on the orbital period timescale , but rather on the internal dynamical timescale of the stars .
it is the direct result of small - amplitude pulsations of each ns about equilibrium .
this is illustrated more clearly in the bottom panel of the figure , which shows the clear correlation between the radial acceleration of each ns ( shown as a fraction of the total inward gravitational acceleration ) and the central density of each ns .
the numerical noise is an artifact of the small number of particles located near the very center of each star for the density curve , and of precision limits in the calculation for the acceleration curve . when averaged over time , we see nearly perfect correlation between the two quantities . also shown in the middle panel of fig .
[ fig : rvstkap ] is a thick solid line depicting the dynamical evolution of a run with @xmath58 and radiation reaction included , to give a sense of the proper inspiral velocity in relation to the spurious radial velocities resulting from oscillations .
we have performed several large - scale sph calculations of ns binary coalescence , assuming an irrotational initial condition for the binary system .
table [ table : runs ] summarizes the relevant parameters of all runs performed , listing the adiabatic exponent @xmath59 , the mass ratio @xmath14 , the initial separation @xmath60 , and the number of sph particles .
we continue to use the same nomenclature for our sph runs introduced in paper 2 , although we present here a new run e1 , using an improved initial configuration calculated by the method described in the previous section .
run e1 is for a system with a @xmath1 eos , equal - mass ns , and an initial separation @xmath30 , and is similar in all respects to run b1 of paper 2 ( called the pn run in paper 1 ) , except that the initial condition is irrotational .
it was continued until a quasi - stationary remnant configuration was reached .
run e2 is for a system with the same @xmath1 eos , but a mass ratio of @xmath61 , and was started from a smaller initial separation @xmath37 , since binaries with smaller masses take longer to coalesce . in runs f1 and f2 ,
the eos has @xmath0 , but we use the same mass ratios and initial separations as runs e1 and e2 , respectively .
additionally , to assess the numerical accuracy and convergence of calculations for irrotational binaries , we performed three runs which were primarily designed to test the dependence of the physical results on numerical parameters .
first , we performed one run , labeled t2 , identical to e1 except for a smaller initial separation @xmath37 .
this run was used to study how accurately the irrotational flow is maintained during the early stages of inspiral , and the effects of a small amount of spurious tidal synchronization on the gw signal .
we then repeated run t2 using @xmath62 and @xmath63 sph particles per ns ( for a total of @xmath64 and @xmath4 particles , respectively ) , to study the effect of numerical resolution on calculations where we know small - scale instabilities will develop .
the number of neighbors was adjusted in the two runs to be @xmath65 and @xmath66 , respectively .
this choice is dictated by the convergence and consistency properties of our basic sph scheme : convergence toward a physically accurate solution is expected when _ both _
@xmath67 _ and _ @xmath68 , but with @xmath69 @xcite .
the primary consideration behind the choice of the initial separation at @xmath37 ( rather than @xmath30 ) was the high computational cost of a run with @xmath4 .
the coalescence process for binary ns systems is essentially the same qualitatively whether they are initially synchronized or irrotational . prior to merger , both
ns show tidal elongation as well as the development of a tidal lag angle @xmath70 , as noted in papers 1 and 2 , created as the ns continuously try to maintain equilibrium while the coalescence timescale gets shorter and shorter .
the inner edge of each ns rotates forward relative to the binary axis , and the outer edge of each ns rotates backward .
we define @xmath70 to be the angle in the horizontal plane between the axis of the primary moment of inertia of each ns and the axis connecting the centers of mass of the respective ns . for equal - mass systems ,
the angle is the same for both ns . for binaries with @xmath71 we always find a larger lag angle for the secondary than the primary .
mass shedding through the outer lagrange points is suppressed in irrotational binaries , but the formation of a differentially rotating remnant is quite similar .
that said , it is important to understand the different computational challenges presented by irrotational systems . to demonstrate this , in fig .
[ fig : dvirr ] we show the evolution of run e1 , with a @xmath1 eos , a mass ratio @xmath72 , and an initial separation @xmath30 .
it is in all ways similar to run b1 , except that the ns start from an irrotational configuration .
rather than plot sph particle positions , we instead show the density contours of the matter in the orbital plane , overlaying the velocity of the material in the corotating frame , defined by taking a particle - averaged tangential velocity , such that @xmath73_i}{\sum_im_i(r_{cyl})_i } , \label{eq : omegac}\ ] ] where the cylindrical radius is defined as @xmath74 . in synchronized binaries , the material maintains only a small velocity in the corotating frame prior to first contact .
in contrast , for irrotational binaries material on the inner edge of each ns is counterspinning in the corotating frame , and thus we see a large discontinuity in the tangential velocity when first contact is made . this surface layer , initially at low density , is kelvin - helmholtz unstable to the formation of turbulent vortices on all length scales .
meanwhile , material on the outer edge of each ns has less angular momentum in an irrotational binary configuration than in a synchronized one .
thus , there is less total angular momentum in the system , and mass shedding is greatly suppressed , as we will discuss further in sec .
[ sec : irrotq1 ] . at late times , though , the remnant rotates differentially , with the same profile seen in the synchronized case , namely @xmath75 attaining its maximum value at the center of the remnant , and decreasing as a function of radius . in table
[ table : gw ] , we show some basic quantities pertaining our gw results , as well as the initial time for all runs , and the tidal lag angle @xmath70 which existed at the moment of first contact .
the negative values of @xmath76 merely reflect the fact that our runs require more than 20 dynamical times before reaching peak gw luminosity .
lag angles @xmath70 are listed for the primary and secondary , respectively , in systems with @xmath77 .
each gw signal we compute typically shows an increasing gw luminosity as the stars approach contact , followed by a peak and then a decline as the ns merge together .
most runs then show a second gw luminosity peak of smaller amplitude .
for all of our runs we list the maximum gw luminosity @xmath78 and maximum gw amplitude @xmath79 , where @xmath80 , @xmath81 , and @xmath82 are defined by eqs .
( 23)-(25 ) of paper 1 ( see also eqs .
[ eq : hpl ] and [ eq : hcr ] below ) .
quantities referring to the first and second luminosity peaks are denoted `` 1 '' and `` 2 '' , respectively .
in addition , we show the time @xmath83 at which the second peak occurs .
we continued several of our runs to late times to study the full gw signal produced during the coalescence , as well as to study the properties of the merger remnants that may form in these situations .
for each of these runs , we list several of the basic parameters of the merger remnant in table [ table : final ] , using the values computed for the remnant at @xmath84 .
we identify the remnant mass @xmath85 , defining the edge of the remnant by a density cut @xmath86 , as well as the gravitational mass of the remnant in the pn runs , where the gravitational mass , which differs from the rest mass , is given by @xmath87 ( see paper 1 for more details ) .
additionally , we list the kerr parameter @xmath88 , central and equatorial values of the angular velocity , @xmath89 and @xmath90 , the semi - major axis @xmath34 and ratios of the equatorial and vertical radii @xmath91 and @xmath92 , and the ratio of the principal moments of inertia @xmath93 .
to study the effect of the eos on the evolution of irrotational ns binaries , we calculated mergers of both @xmath1 and @xmath0 polytropes ( runs e1 and f1 , respectively ) .
a comparison of the binary separations as a function of time , along with the dimensionless gw luminosities and amplitudes are shown in fig .
[ fig : gweos ] . immediately apparent is a difference in the location of the dynamical stability limit found in the two calculations .
the orbit of the binary system containing ns with a softer eos ( run f1 ) remains stable at separations where the binary system with the stiffer eos has already begun to plunge inward dynamically .
we see that , as in the synchronized case presented in paper 2 , the peak gw luminosity in dimensionless form is larger for the softer eos , but after a secondary luminosity peak the remnant relaxes toward a spheroidal , non - radiating configuration ( with essentially no emission whatsoever after @xmath94 ) .
we also note that the secondary peak occurs sooner after the primary peak , by a factor of @xmath95 .
even though they are presumed to be unphysical , calculations started from a synchronized initial condition make up much of the body of work performed to date on the binary ns coalescence problem . noting this
, we compare our irrotational run e1 , with parameters detailed above , to one similar in every respect but started with a synchronized initial condition , our run b1 . in the top panel of fig .
[ fig : gwspin ] , we see that the inspiral tracks do not align particularly well .
synchronized binaries contain more total energy , and are thus less dynamically stable than irrotational ones , leading to a more rapid inspiral , even before the stability limit is reached .
additionally , the binary separation `` hangs up '' earlier , at a separation @xmath96 , indicating the onset of mass shedding , as matter begins to expand radially outward from the system .
the middle and bottom panels of the figure show the gw signals and luminosity , respectively , for the two runs .
while the initial luminosity peaks are similar for both runs , both in amplitude and morphology , the secondary peaks are vastly different .
the secondary peak for the synchronized system is of considerably greater magnitude than in the irrotational case , and delayed relative to it .
we conclude that , while the adiabatic index seems to be the dominant factor in determining the gw signal during the merger itself , the initial velocity profiles of the ns play a key role in the evolution of the remnant , as well as affecting the orbital dynamics during inspiral . to better understand the features found in the gw signals of these calculations , particle plots for runs e1 , b1 , and f1 are shown in fig .
[ fig : xyeos ] . comparing the leftmost panels
, we see that at @xmath28 , when the gw luminosity peaks , the mass configurations are qualitatively similar , although more low - density material is seen at the edges of the newly forming remnant in the run with the @xmath0 eos , conforming to the general density profile expected for a softer eos .
more subtle is the greater extension of the matter found in the synchronized run .
since material on the outer edge of each ns has greater angular momentum when a synchronized initial condition is chosen , the calculation shows that such an initial condition allows for greater efficiency at channeling material outward during the final moments of inspiral .
this difference is made abundantly clear by a comparison of the calculations at @xmath97 , shown in the center panels .
we see extensive mass shedding from the synchronized run , much less from the irrotational runs .
there is significantly more mass shedding from the run with the softer eos , but most of the material remains extremely close to the remnant .
finally , by @xmath98 , we see that the softer eos produces a nearly spherical remnant , whereas the calculations with a stiffer choice of eos produce remnants which are clearly ellipsoidal , and will continue to radiate gws for some time , albeit at a much lower amplitude than at the peak .
the strong influence of the choice of both eos and initial velocity profile on the final state of the remnant is shown in fig .
[ fig : finaleos ] for the same three runs . in the top panel ,
we see the angular velocity profiles of the remnants at @xmath84 .
we see that the choice of eos plays an important role near the center of the remnant , but at @xmath99 , the velocity profiles are essentially identical .
the pattern holds as well for the mass profiles , which are shown in the bottom panel .
the softer eos leads to a more centrally condensed remnant , as we would expect , but the remnants formed in both irrotational calculations contain virtually all the system mass within @xmath96 , with no more than @xmath55 escaping to larger radii .
there is slightly more mass shedding past @xmath100 for the softer choice of eos , since more of the low - density material originally found at the edges of the ns is shed through the outer lagrange points of the system . by comparison
, the synchronized run sheds almost @xmath101 of the total system mass past @xmath100 , even though the angular velocity profile at small radii is nearly the same as for the irrotational run with the same choice of eos .
even though all well - measured ns masses in relativistic binary pulsars appear roughly consistent with a single ns mass @xmath102 @xcite , it is important to consider cases where the two ns have somewhat different masses .
the roles played by the primary and secondary in unequal - mass binary mergers are remarkably different compared to the picture developed above for equal - mass systems . in paper 2 , we found that the primary ns in the system remains virtually undisturbed , simply settling at the center of the newly formed merger remnant .
the secondary is tidally disrupted prior to merger , forming a single thick spiral arm .
most of the material originally located in the secondary eventually forms the outer region of the merger remnant , but a significant amount of material is shed to form a thick torus around the central core . in fig .
[ fig : xyq8 ] , we show particle plots for runs e2 and f2 , with a @xmath1 and @xmath0 eos , respectively , both with mass ratio @xmath61 . in the leftmost panels , at @xmath28 , we see that the secondary , located on the left , is tidally disrupted as it falls onto the primary . for the softer , @xmath0 eos ( run f2 ) , we see a greater extension of the secondary immediately prior to merger , as well as early mass shedding from the surface of the primary , as material from the secondary essentially blows it off the surface of the newly forming remnant .
this process continues , so that by @xmath97 ( center panels ) , the secondary has begun to shed a considerable amount of mass in a single spiral arm which wraps around the system . much like in the equal - mass case ,
the spiral arm is much broader for the softer eos .
mass loss from the primary is greatly reduced in the system with the stiffer eos , with only a scattering of particles originally located in the primary lifted off the surface .
finally , by @xmath98 ( right panels ) , we see that the spiral arm has in both cases begun to dissipate , leaving a torus around the merger remnant containing approximately @xmath103 of the total system mass .
although the merger process is significantly different for equal - mass and unequal - mass binaries , the details of the inspiral phase are reasonably similar . in particular , the evolution of the binary separation for runs with @xmath61 , shown in the top panel of fig .
[ fig : gwq8 ] is roughly similar to what was found in fig .
[ fig : gweos ] for binaries with @xmath72 . for both choices of the eos
, the dynamical stability limit is located at approximately the same separation for both @xmath61 and @xmath72 binaries , although prior to the onset of instability the more massive equal - mass binaries show a more rapid stable inspiral . as we found before
, the dynamical stability limit occurs further inward for the softer choice of the ns eos . in paper 2
, we found that the scaling of the maximum gw amplitude and luminosity as a function of the system mass ratio followed a steeper power law in synchronized binaries than would be predicted by newtonian point - mass estimates .
newtonian physics predicts that @xmath104 and @xmath105 for merging binary systems .
rs found the scaling in their numerical calculations to roughly follow empirical power law relationships given by @xmath106 and @xmath107 .
the discrepancy results from the unequal role played by the two components during the final moments before plunge .
the primary , which remains relatively undisturbed , contributes rather little to the gw signal , especially during the final moments before coalescence .
thus , the gw power is reduced as the mass ratio is decreased .
similar results were found in paper 2 for pn calculations of synchronized binaries , especially a steeper decrease in the gw luminosity as a function of the mass ratio for binaries with a soft eos . in the middle and bottom panels of fig .
[ fig : gwq8 ] we show the gw signals and luminosity , respectively , for runs e2 and f2 .
we find the same strong decrease in the gw power , especially for the softer @xmath0 eos .
we expect that there should be a strong observational bias towards detecting mergers of equal - mass components , especially if the eos is softer .
all our results presented above for equal - mass binaries used an initial binary separation @xmath30 , in part for the sake of comparison with previous calculations for synchronized binaries .
this choice agrees with the standard approach of utilizing the largest possible separation for which the calculation can be performed using a reasonable amount of computational resources .
this also has the advantage that any small deviations from equilibrium will generally be damped away before the ns actually make contact .
it also allows for the best determination of the dynamical stability limit , however , a large initial separation can create problems in the case of irrotational binaries , since numerical shear viscosity inherently present in sph codes can lead to some degree of tidal synchronization of the ns during the inspiral phase .
thus , by the time the merger takes place , the ns will no longer be completely irrotational . to study this effect
, we calculated mergers for equal - mass ns with a @xmath1 eos starting at initial separations of both @xmath30 ( the aforementioned run e1 ) and @xmath37 ( run t2 ) . in the top panel of fig .
[ fig : rvstsep0 ] , we show the binary separation as a function of time for both runs , noting the good agreement throughout . at the very end , during the merger itself , we do see the beginning of a slight discrepancy , attributable in large part to greater mass shedding in the calculation started at greater separation .
this is similar to what was seen in sec .
[ sec : irrotq1 ] , where run b1 , which had greater spin angular momentum , showed greater mass shedding , but the effect is greatly reduced in magnitude here since the ns in run t2 are nowhere near complete synchronization at the moment of first contact . in the bottom panel of the figure , we plot the ratio of the net spin angular momentum of the ns about their own centers of mass to the total angular momentum of the binary system , as a function of time .
we see that the ns do gradually acquire a rotation pattern which corresponds to the direction of corotation , although there is nowhere near enough time to synchronize the binary .
the effect is greatly enhanced immediately prior to merger in both cases , as the ns develop tidal lag angles and become distorted . by the time the binary initially started from @xmath30
reaches a separation of @xmath108 , the net angular momentum around each ns around its center of mass is equal to approximately @xmath109 the value we would expect should the binary be synchronized .
this difference persists throughout the inspiral phase when the two calculations are compared . in fig .
[ fig : gwsep0 ] , we compare the gw signals and luminosities for the two runs .
we find excellent agreement between the two waveforms , both in amplitude and in phase .
both runs show the modulated , damped gw luminosity which is characteristic of all runs we have computed using pn gravity .
there is a slight difference in the amplitude of the signal during the second gw luminosity peak , but we expect such differences to be minor in light of such issues as the uncertainty in the equation of state and the larger problem of a proper relativistic treatment of gravitation . to focus on the effect of varying @xmath60 on the final results , we show the final mass and angular velocity profiles of the remnants for the two calculations in fig .
[ fig : finsep0 ] .
the results are in good agreement , although we see that the greater spin angular momentum of the run started at greater initial separation leads to approximately three times as much mass being deposited in a halo which surrounds the remnant while remaining gravitationally bound to it . in both cases , however , the total mass in the halo is less than @xmath55 of the total system mass .
the inner region of the remnant in the run started from @xmath30 actually spins slightly slower than in the run started further inward , even though the ns have a greater spin angular momentum at the moment of first contact , but only because angular momentum transport outward was marginally more efficient in this case .
we conclude that the choice of initial separation plays very little role in determining the results of our calculations , so long as we start from an initial separation @xmath110 . as discussed in sec .
[ sec : summary ] , 3d calculations with limited spatial resolution could lead to gw signals which are dependent upon the number of particles used . to test this , gw signals computed from ns merger calculations , we performed runs t1 , t2 , and t3 , which all have equal - mass ns , start from the same initial separation @xmath37 , and use a @xmath1 eos , but vary by two orders of magnitude in the number of sph particles used , from run t1 with @xmath111 to run t3 with @xmath112 sph particles .
although we could have used @xmath30 as an initial separation , we felt that the smaller initial separation was justified given the large computational overhead required to do a calculation with a million sph particles .
to the best of our knowledge , run t3 is the largest and highest resolution sph calculation of irrotational binary ns coalescence to date .
a comparison of the gw signals in both polarizations , as well as the gw luminosities , is shown in fig .
[ fig : gwnp ] .
we see that the lowest resolution run t1 produces a gw signal clearly different than higher resolution runs t2 and t3 .
the difference is due in part to initial oscillations of the ns about quasi - equilibrium .
such oscillations are greatly reduced by increasing the number of sph particles . since all three calculations started with the same approximate ellipsoidal models ( see sec .
[ sec : initconf ] ) , we conclude that the amplitudes of the initial fluctuations result primarily from numerical noise . the two runs with higher resolution show good agreement from beginning to end , producing nearly identical gw luminosities .
reassuringly , the gw signals also remain in phase throughout , indicating that calculations with @xmath2 can indeed model well these irrotational binary mergers . to estimate the precision which can be achieved for these numerical resolutions , we show in table [ table : gravnp ] a comparison of gw quantities computed at various times for each of the three runs . at @xmath113 , @xmath28 , and @xmath97
, we show the total gw strain @xmath114 , and the instantaneous frequency of the gws @xmath115 , defined by the relations @xmath116 we see that for the two high resolution runs , no quantity shown varies by more than about @xmath117 , whereas the difference is more than @xmath118 in the computed gw strains at late times between our highest and lowest resolution runs .
the vortices forming at the surface of contact are shown in detail in figs .
[ fig : dvnp1 ] and [ fig : dvnp2 ] .
density contours in the orbital plane are overlaid with velocity vectors , which are plotted in the corotating frame of the binary , as defined by eq .
[ eq : omegac ] .
the upper left panels show the evolution of run t1 , with the bottom left and right panels representing runs t2 and t3 , respectively . in fig .
[ fig : dvnp1 ] , we show the state of the three runs at @xmath28 .
immediately apparent is that vortices have formed to the largest extent in the lowest resolution run , whereas in the higher resolution run there is little sign of particles mixing , except at a large distance along the vortex sheet from the center of the newly forming remnant . by @xmath119 , shown in fig .
[ fig : dvnp2 ] , we see that there is a slight difference between the high resolution calculations with regard to the direction of the material flowing along the vortex sheet . in the highest resolution run , the streams of material flow nearly in a straight line from one vortex to the other , whereas in the middle and lowest resolution runs , there is a larger region of material which is accelerated toward the very center of the remnant .
overall , though , there is excellent agreement between the two highest resolution calculations .
the agreement of the gw signals calculated from runs t2 and t3 lead us to believe that the small - scale differences seen in the matter near the vortex sheet does not carry over into the bulk of the mass , responsible for the gw emission .
essentially , the quadrupole moment at any given instant is most sensitively dependent upon the orientation of the densest regions at the cores of the respective ns , which are unaffected by the small - scale motion in the vortex sheet .
the infall of the cores is driven by dynamical instability , leading them to plunge inward and merge , disrupting the vortex sheet and leading to the formation of the merger remnant .
zhuge , centrella , and mcmillan first pointed out the importance of gw energy spectra for the interpretation of merger signals @xcite .
in particular , on the basis of newtonian sph calculations , they showed how the observation of particular features in the spectra could directly constrain the ns radii and eos . since the detection of ns merger events
will likely be made in narrow - band interferometers , it is especially important to understand the frequency dependence of the gw signals , and not just their time behavior ( as represented by waveforms ) .
following the approach in @xcite , we calculate the gw energy spectrum from each of our calculations as follows .
we first take the fourier transforms of both polarizations of the gw signal , @xmath120 and we insert them into the following expression giving the energy loss per unit frequency interval ( see , e.g. , @xcite ) , @xmath121 where @xmath122 represents the fourier transform of the second derivative of the traceless quadrupole tensor . for point - mass inspiral , the energy spectrum takes the power - law form @xmath123 @xcite , the slow _ decrease _ with increasing frequency coming from the acceleration of the orbital decay : although more energy is emitted per cycle , fewer cycles are spent in any particular frequency interval as the frequency sweeps up . near the final merger , large deviations from this simple power - law spectrum are expected .
however , our initial binary configurations are still reasonably described by a point - mass model .
therefore , to construct the complete gw signal , we attach a point - mass inspiral waveform ( hereafter referred to as the inspiral subcomponent ) onto the beginning of the signal calculated numerically with sph ( referred to as the merger subcomponent ) .
the quadrupole tensor for the inspiral subcomponent is assumed to have the form @xmath124 where @xmath125 is the time _ before _ our dynamical calculation starts , and the amplitude and phase given by @xmath126 } \\
\phi(t')&=&-\int_{t'}^0 \omega(t ' ) dt'\\ \omega(t')&=&(1.0+\epsilon')\sqrt{\frac{m}{[r_0(1-\frac{t'}{t_{\rm mgr}})^{0.25}]^3}},\end{aligned}\ ] ] where @xmath60 is the initial binary separation and @xmath17 and @xmath127 are the total and reduced masses of the system , respectively . the time constant @xmath128 is given by the familiar expression @xmath129 with @xmath130 , these expressions correspond to the well - known quasi - newtonian point - mass inspiral results@xcite . here
the correction factor @xmath131 is used to account for both finite - size and pn effects and is determined by matching the amplitude of the point - mass signal to the initial amplitude calculated from our sph initial condition .
typically , it is no larger than about @xmath132 .
the correction factor @xmath133 is used in a similar way to match the initial angular velocity of the system , and is of similar magnitude . with these corrections ,
[ eq : qxx]-[eq : qxz ] describe well the inspiral subcomponent at late times , but not , of course , at earlier times where one should have @xmath134 and @xmath135 as @xmath136 . the discrepancy , however , is no more than @xmath101 , and affects only the nearly featureless low - frequency part of the spectrum . in figs .
[ fig : psn ] and [ fig : psp ] , we show a comparison between the energy spectra computed from a newtonian calculation with radiation reaction and from a pn calculation . the newtonian calculation is the n run from paper 1 ( referred to in paper 2 as run a1 ) , which was performed for equal - mass ns with @xmath1 eos , started at a separation @xmath137 in a synchronized initial state .
the pn calculation is the run e1 described in sec .
[ sec : sep0 ] , with equal - mass ns , a @xmath1 eos , and an irrotational initial condition with separation @xmath30 . in both figures ,
the dashed and dotted lines represent the merger and inspiral subcomponents , respectively , with the heavy solid line representing the total combined spectrum .
we see immediately that there is a significant difference in the energy emitted between @xmath138 ( @xmath139 for our adopted ns parameters ; see eq .
[ eq : fdyn ] ) .
this is directly attributable to the earlier dynamical instability and faster inspiral rates found in pn calculations .
since the binary system spends less time in this frequency interval , the energy emitted is greatly suppressed by the addition of 1pn effects .
the characteristic `` cliff frequency , '' at which the energy spectrum plunges below the point - mass power - law , is the best indicator of the start of dynamical instability .
a measurement of this frequency , combined with theoretical calculations such as those presented here , would lead directly to the determination of the ns radii , since all frequencies scale with @xmath140 and the destabilizing 1pn effects scale with @xmath141 . for the system considered in fig .
[ fig : psp ] this `` cliff frequency '' is about @xmath142 ( @xmath143hz ) , almost within the reach of broad - band interferometers . had the 1pn effects been taken into account at full strength , this frequency would likely have been even lower , suggesting that deviations from point - mass behavior may even become measurable within the frequency range of current theoretical point - mass inspiral templates .
two prominent features appear in the pn spectra at much higher frequencies . in general , the characteristic frequency of gw emission during mergers sweeps upward monotonically throughout the evolution .
the sharp peaks at @xmath144 and @xmath145 ( @xmath146 and @xmath147 ) in fig .
[ fig : psp ] are then clearly seen to result from emission at the time of the gw luminosity peak and during the remnant oscillation ( `` ring down '' ) phase , respectively .
in contrast to some previous newtonian results @xcite , we find the amplitude of these peaks to be well below the point - mass power - law ( by a factor of almost 3 and 5 , respectively ) .
no calculation we have performed with our pn formalism has ever produced energy above the point - mass power - law in any frequency range , confirming that the pn effects not only accelerate the dynamical instability of the system , but also cause a suppression of the total gw emission during the entire merger ( as we first showed in paper 1 ) .
measurements of these peak frequencies would also provide independent constraints on the eos , but this will certainly require advanced detectors operated in narrow - band mode in order to beat the very high laser shot noise above @xmath9khz .
we now examine the dependence of these calculated gw energy spectra on initial binary separation and numerical resolution . in the top panel of fig .
[ fig : pssep0 ] , we compare the spectra computed from our @xmath1 eos , equal - mass calculations started at initial separations of @xmath37 ( run t2 ) and @xmath30 ( run e1 ) .
we see excellent agreement above @xmath148 ( @xmath149 ) , but a slight discrepancy near the cliff frequency . at separations in the range @xmath150
, the binary does inspiral faster than a point - mass model would predict . since our code has been extensively tested and shown to reproduce newtonian results in this separation range ( see fig . 4 of paper 1 )
, we attribute the difference to the 1pn effects . assuming that a quasi - equilibrium description of the system is appropriate in this range of separations , the inspiral rate should be given by @xmath151 where @xmath152 is the energy loss rate to gravitational radiation , and @xmath153 is the rate of change of total energy as a function of separation along an equilibrium sequence of binary ns models @xcite . since the first factor is very insensitive to pn effects ,
we conclude that the slope of the equilibrium energy curve must be made smaller by the addition of 1pn corrections , in agreement with the results of pn equilibrium calculations @xcite . while there must exist a _
minimum _ in the equilibrium energy curve at some critical separation , formally representing the innermost stable circular orbit ( isco ) of the system @xcite , this may be less relevant to the onset of dynamical coalescence , which will occur earlier , as soon as the inspiral timescale becomes comparable to the orbital period . in the bottom panel of fig .
[ fig : pssep0 ] we compare the energy spectra from runs t1 ( with @xmath64 particles ) and t2 ( @xmath3 ) , used to study the dependence of the gw signal on numerical resolution .
we find that the spectra are nearly identical in the frequency range @xmath154 ( @xmath155 ) , characteristic of the peak emission and the following remnant oscillations , but disagree by a significant amount at lower frequencies .
recall from fig .
[ fig : gwnp ] that the lowest resolution run led to larger initial oscillations around equilibrium .
while these ( spurious ) oscillations are at too high a frequency to appear in the energy spectrum , they do affect the initial inspiral rate significantly .
the extra energy present in the system , combined with the tendency for the binary orbit to become slightly eccentric , leads to an artificially accelerated inspiral , and a further reduction of the energy above the cliff frequency .
further comparisons are shown in fig .
[ fig : pseos ] . in the top panel ,
we show the energy spectra for two different values of @xmath59 .
these were computed from the two equal - mass runs started from @xmath30 , run e1 with @xmath1 , and run f1 with @xmath0 .
note that these two different values of @xmath59 are compared here at constant @xmath17 and @xmath18 , which is of course not realistic .
the dominant dependence of the gw emission on the eos is likely to be from the scaling of finite - size effects with @xmath18 , and from the scaling of pn effects with @xmath141 , which we can not study realistically given the limitations of our pn formalism ( sec .
[ sec : formal ] ) . at low frequencies
the two spectra are nearly identical .
for the softer eos , the peak in the energy spectrum corresponding to maximum gw emission is of slightly lower amplitude and frequency .
the peak corresponding to remnant oscillations is greatly suppressed , since the soft eos can not support a stable triaxial configuration ( see rs2 ) . in the bottom panel of fig .
[ fig : pseos ] , we compare the energy spectra from synchronized binaries ( run b1 ) and irrotational binaries ( run e1 ) .
we find a sharper `` cliff '' from the synchronized system , leading to lower energy in the frequency range up to about @xmath156 ( @xmath157 ) . at higher frequencies , however , the two spectra are remarkably similar .
in particular , the spectral peak frequencies appear to be affected much more strongly by the eos than by the details of the initial velocity configuration . in fig .
[ fig : psq8 ] , we compare gw spectra for systems with different binary mass ratios
. in the top panel , we compare runs t2 and e2 , with mass ratios of @xmath72 and @xmath61 , respectively ( both with a @xmath1 eos and both started from an initial separation @xmath37 ) .
we see a clear difference in the overall amplitude of the two spectra , but in general the low - frequency behavior of the spectra is not qualitatively different . in both cases ,
we see a smooth decline in the gw energy which indicates the onset of dynamical instability , and levels off as we reach the characteristic frequency of the maximum emission .
a clear difference is the lack of a true peak in the spectrum for the @xmath61 binary , which reflects the suppression of gw emission in unequal - mass mergers , discussed in sec .
[ sec : q8 ] .
there is a well defined peak in the energy spectrum characteristic of emission from the remnant , nearly equal in amplitude to the first peak , as was found for the equal - mass binary with the same eos .
the frequencies of both peaks in the @xmath61 spectrum are shifted lower relative to their location in the equal - mass case . in the bottom panel of fig .
[ fig : psq8 ] , we show the same comparison for runs f1 and f2 with @xmath0 .
the results are similar except that now the first peak for the binary with @xmath61 is completely absent .
using the pn lagrangian sph code described in our previous papers , which is complete to 1pn order and includes radiation reaction effects , we have investigated a wide parameter space of binary ns mergers started from an irrotational initial condition .
this initial configuration represents the most realistic approximation for ns binaries at separations of @xmath158 , corresponding to the onset of dynamical coalescence .
based on these calculations , as well as those from our previous papers , we have calculated the energy spectrum of the gravitational wave emission for a variety of systems .
the key result is the existence of a `` cliff frequency '' in all of the energy spectra we have studied , i.e. , a frequency above which the energy emitted dips dramatically beneath the point - mass approximation .
we attribute this effect to the onset of dynamical instability in binary systems , which leads to a much more rapid inspiral than the point - mass formula predicts , an effect amplified when pn terms are taken into account .
if the cliff frequencies of binary systems are even smaller than those found here ( typically around @xmath159 for our standard ns parameters ) when general relativity is treated consistently , they may lie within the frequency band accessible to broad - band laser interferometers .
for example , proposals for ligo ii @xcite place its upper frequency limit around @xmath160 ( where the sensitivity in terms of a characteristic gw strain has been degraded by a factor @xmath161 due to photon shot noise ) , meaning that cliff frequencies should be seen if the true physical ns radius is sufficiently large , @xmath162 .
our results also suggest that the high frequency features in the gw energy spectrum , which result from emission during the merger itself ( @xmath163 ) and from late - time oscillations of the remnant ( `` ring down '' @xmath164 ) , will be observable only by more advanced narrow - band detectors , but if observed , could place strong constraints on the ns eos .
we also find from our calculations that initially irrotational binaries evolve in a qualitatively different way than do initially synchronized systems .
regardless of the choice of eos , runs started from an irrotational configuration result in much less mass shedding than do synchronized runs , depositing no more than @xmath55 of the total system mass in an outer halo which remains bound to the merger remnant at the center of the system .
additionally , there is a significant difference in the inspiral rate of such systems immediately prior to merger , with synchronized systems merging much more rapidly .
thus , since real ns binaries should be essentially irrotational , it is important to exercise caution when interpreting the results drawn from calculations for initially synchronized systems , both before and after the merger occurs .
the lack of mass shedding seen in our pn calculations of irrotational systems leads us to conclude that essentially no mass should be shed from realistic binary systems , since general relativistic effects further suppress the mass shedding instability @xcite . from a calculational standpoint
, we find competing arguments for the ideal initial binary separation of an irrotational system .
runs started from larger separations ( here @xmath165 ) require greater computational resources , and show more spurious synchronization because of the numerical viscosity of the sph method .
such calculations do provide a better treatment of the deviations from point - mass inspiral prior to merger which result from both newtonian and pn finite - size effects .
however , runs started from closer in ( @xmath166 ) are more reliable for drawing conclusions about the final state of the system , with regard to mass shedding as well as the rotational profile of the remnant .
we find , reassuringly , that the phase and amplitude evolution of the gw signal during the primary luminosity peak is unaffected by the initial separation .
we believe that our results are unaffected by the limited numerical resolution of 3d calculations .
although merging binary ns systems develop small - scale instabilities , whose evolution we can not follow exactly , we see little effect on the gw signals we compute , so long as we use a sufficiently large number of sph particles .
this is especially true for the phase of the gw signal .
we conclude that numerical convergence for a given set of initial conditions and physical assumptions is possible without requiring excessive computational resources . based on these results , we believe that the fundamental limits of our method are not set by the numerical resolution or available computational resources , but rather by shortcomings in the pn formalism itself .
all our pn calculations are limited by the magnitude of the 1pn terms that appear in the hydrodynamic equations .
since we can not treat physically realistic ns models , we deal with the 1pn terms at reduced strength .
this hybrid method does in some sense approximate the cancellations found in gr between 1pn and higher - order terms , but in the end can not fully model the non - linear nature of relativistic gravity . to do so in a more complete manner ,
two different approaches have been employed .
the first is to attempt to integrate the hydrodynamics equations in full gr , as done recently by shibata and uryu @xcite .
starting from quasi - equilibrium irrotational binary systems at the moment of first contact , they calculate the fully relativistic evolution of the system .
such calculations represent a remarkable step forward , and can be considered the forefront of the current efforts to understand binary ns coalescence .
they are limited only by the accuracy to which they can prepare their initial conditions and by the difficulty of extracting accurate gw signals from the boundary of 3d grids at a finite distance , usually well within the near zone of the source .
as of yet no formalism has proven stable enough to handle a calculation which starts from the dynamically stable region and ends after the merger is complete . a second approach is to use an approximation of full gr which is known to be numerically stable , known as the _ conformally flat _ (
cf ) approximation @xcite . in this approximation to general relativity , assuming a specific form for the spatial part of the metric allows the equations of gr to be reduced to a set of linked non - linear _ elliptic _ equations .
the drawback to the method is that the cf approximation is time - symmetric , and thus does not include the dissipative gravitational wave effects seen in full gr .
such terms can be added externally to the formalism , though , to give the proper dynamical behavior @xcite .
the cf approximation has been used successfully to compute quasi - equilibrium binary sequences to high accuracy for both synchronized @xcite and irrotational @xcite binaries , showing good agreement with fully relativistic calculations of synchronized @xcite and irrotational binaries @xcite , except for the case of extremely compact ns at small separations .
additionally , a pn variant of the cf approximation has been used to study rapidly rotating single - star configurations @xcite , giving excellent agreement with fully gr calculations @xcite .
in addition to equilibrium studies , dynamical sph calculations have recently been performed using relaxed , initially synchronized binary configurations @xcite .
the authors are currently working on a code that will take as an initial condition fully relativistic irrotational , quasi - equilibrium models for binary ns systems , calculated using spectral methods @xcite .
the system will then be evolved in the cf approximation .
as the formalism is known to work for both binary configurations and rapidly rotating single - star configurations , reflective of our merger remnants , the hope is that such a method will allow us to calculate the evolution of a physically realistic binary system from the dynamically stable regime through merger and formation of a merger remnant , in a way that is consistent with gr throughout .
the comparison of such calculations with those performed in full gr will serve as an important check of the shortcomings and successes of either approach .
the authors wish to thank thomas baumgarte , joan centrella , roland oechslin , stu shapiro and kip thorne for useful discussions .
this work was supported by nsf grants phy-0070918 and phy-0133425 .
our computational work was supported by the national computational science alliance under grant ast980014n and utilized the ncsa sgi / cray origin2000 .
h. lck _ et al .
_ , class .
quantum grav .
* 14 * , 1471 ( 1997 ) ; k. danzmann , in _ relativistic astrophysics _ , proceedings of the 162nd w.e
. heraeus seminar , edited by h. riffert _ et al . _
( wiesbaden : vieweg verlag , 1998 ) , p.48 .
k. kuroda _ et al .
_ , in _ proceedings of the international conference on gravitational waves : sources and detectors _ , edited by i. ciufolini and f. fidecard ( world scientific , 1997 ) , p.100 ; h. tagoshi _ et al .
_ , phys .
d * 63 * , 062001 ( 2001 ) .
meers , phys .
d * 38 * , 2317 ( 1988 ) ; k.a .
strain , and b.j .
meers , b.j . , phys .
lett . * 66 * , 1391 ( 1991 ) ; g. heinzel , _ et al .
_ , phys .
lett . * 81 * , 5493 ( 1998 ) ; a. freise _ et al .
_ , phys .
lett . a * 277 * , 135 ( 2000 ) .
k. oohara and t. nakamura , prog .
phys . * 82 * , 535 ( 1989 ) ; t. nakamura and k. oohara , _ ibid . _ * 82 * , 1066 ( 1989 ) ; k. oohara and t. nakamura , _ ibid . _ * 83 * , 906 ( 1990 ) ; t. nakamura and k. oohara , _ ibid . _ * 86 * , 73 ( 1991 ) .
s. rosswog _ et al .
_ , astron .
astrophys . *
341 * , 499 ( 1999 ) ; s. rosswog , m.b .
davies , f .- k .
thielemann , and t.piran , astron .
astrophys . * 360 * , 171 ( 2000 ) ; s. rosswog and m.b .
davies , preprint , astro - ph/0110180 .
m. ruffert , h .- th .
janka , and g. schfer , astron .
astrophys . * 311 * , 532 ( 1996 ) ; m. ruffert , h .- th .
janka , k. takahashi , and g. schfer , astron . astrophys . * 319 * , 122 ( 1997 ) ; m. ruffert , m. rampp , and h .- th .
janka , astron .
astrophys . * 321 * , 991 ( 1997 ) .
m. shibata and k. uryu , phys .
d * 61 * , 064001 ( 2000 ) ; m. shibata and k. uryu in _ proceedings of the 20th texas symposium on relativistic astrophysics _ , edited by h. martel and j.c .
wheeler , ( new york , aip press , 2001 ) , p. 717 .
wilson , g.j .
mathews , and p. marronetti , phys .
d * 54 * , 1317 ( 1996 ) ; g.j .
mathews , p. marronetti , and j.r .
wilson , _ ibid .
_ * 58 * , 043003 ( 1998 ) ; p. marronetti , g.j
. mathews , and j.r .
wilson , _ ibid . _ * 60 * , 087301 ( 1999 ) ; g.j .
mathews and j.r .
wilson _ ibid . _ * 61 * , 127304 ( 2000 ) .
.input parameters for the runs described in this paper : @xmath59 is the adiabatic exponent , @xmath14 the binary mass ratio , and @xmath60 the initial separation . all runs used @xmath167 sph particles , except testing runs t1 and t3 , with @xmath111 and @xmath112 sph particles , respectively .
testing run t2 differs from run e1 only in the initial separation .
all runs in this paper were started from an irrotational initial condition . [ cols="^,^,^,^,^ " , ] | gravitational wave ( gw ) signals from coalescing binary neutron stars may soon become detectable by laser - interferometer detectors .
using our new post - newtonian ( pn ) smoothed particle hydrodynamics ( sph ) code , we have studied numerically the mergers of neutron star binaries with irrotational initial configurations .
these are the most physically realistic initial conditions just prior to merger , since the neutron stars in these systems are expected to be spinning slowly at large separation , and the viscosity of neutron star matter is too small for tidal synchronization to be effective at small separation . however , the large shear that develops during the merger makes irrotational systems particularly difficult to study numerically in 3d .
in addition , in pn gravity , accurate irrotational initial conditions are much more difficult to construct numerically than corotating initial conditions . here
we describe a new method for constructing numerically accurate initial conditions for irrotational binary systems with circular orbits in pn gravity .
we then compute the 3d hydrodynamic evolution of these systems until the two stars have completely merged , and we determine the corresponding gw signals .
we present results for systems with different binary mass ratios , and for neutron stars represented by polytropes with @xmath0 or @xmath1 . compared to mergers of corotating binaries
, we find that irrotational binary mergers produce similar peak gw luminosities , but they shed almost no mass at all to large distances .
the dependence of the gw signal on numerical resolution for calculations performed with @xmath2 sph particles is extremely weak , and we find excellent agreement between runs utilizing @xmath3 and @xmath4 sph particles ( the largest sph calculation ever performed to study such irrotational binary mergers ) .
we also compute gw energy spectra based on all calculations reported here and in our previous works .
we find that pn effects lead to clearly identifiable features in the gw energy spectrum of binary neutron star mergers , which may yield important information about the nuclear equation of state at extreme densities . |
ice giants , the cores of gas giants and protoplanets that later form terrestial planets , are generally believed to have formed by coagulation from small planetesimals . understanding the evolution of the size distribution of growing protoplanets , their velocity dispersion and the interplay between the two is crucial for shedding light onto the planet formation process . here
, we study the runaway growth of protoplanets ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and their subsequent velocity evolution .
runaway growth can occur when the accretion cross section of protoplanets is enhanced by gravitational focusing . since , gravitational focusing is strongest for the largest bodies , the radii of larger protoplanets run away from that of smaller .
this leads to a size distribution that develops a tail containing a small number of large protoplanets .
it is however not always clear what fraction of the total mass participates in this runaway growth and what protoplanet size distribution such runaway growth gives rise to @xcite . in this paper
we address both of these questions . in the following sections , we discuss runaway growth in the context of the kuiper belt , which is an ideal laboratory to test our results , since it is a remnant of the primordial solar system , where planet formation never reached completion .
the results however , also apply to runaway growth during planet formation and growth in debris disk , as long as gas plays no significant role in the accretion and damping of the velocity dispersion .
the kuiper belt consists of a disk of icy bodies located at the outskirts of our planetary system , just beyond the orbit of neptune and contains some of the least processed bodies in our solar system .
motivated by the discovery of the first kuiper belt object @xcite after pluto and charon , several groups conducted large scale surveys to characterize the kuiper belt .
these efforts led to the discovery of more than 1200 objects in the kuiper belt to date .
the kuiper belt size distribution contains many important clues concerning the formation of kuiper belt objects ( kbos ) , their effective strength and their collisional evolution @xcite .
it also provides a snapshot of an earlier stage of planet formation , which was erased elsewhere in the solar system where planet formation proceeded all the way to completion .
the cumulative size distribution of kbos larger than @xmath8 is well described by a single power - law given by @xmath9 where @xmath5 is the number of objects with radii greater than @xmath6 , and @xmath10 is the power - law index .
kuiper belt surveys find that the size distribution for kbos with radii greater than about 50 km follows this power - law with @xmath11 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which implies roughly equal mass per logarithmic mass interval .
this size distribution is a relic of the accretion history in the kuiper belt and therefore provides valuable insights into the formation of large kbos ( @xmath12 ) ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
observations suggest that there is a break in the power - law size distribution at smaller kbo sizes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the break in the size distribution is generally attributed to collisions that break - up small kbos ( i.e.,@xmath13 ) and modify their size distribution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the kbo size distribution below the break is still poorly constrained , although some encouraging progress has been made recently in probing the abundance of sub - km - sized kbos by stellar occultations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the work presented in this paper focuses on the size distribution of large kbos ( @xmath12 ) , which is well constrained by observations and which sheds light onto the formation of kbos , protoplanets and accretion processes that could be ongoing in other debris disks .
numerical coagulation simulations have been successful in reproducing the observed kbo size distribution .
such simulations typically find that the accretion processes of kbos yield a power - law size distribution with @xmath14 for 10 - 100 km and larger objects @xcite , which is consistent with the observed power - law size distribution . despite their success ,
the reason for the actual slope of the distribution has so far not been explained by such simulations . in this paper
, we offer an explanation for the slope of the kbo size distribution and for the amount of mass in the large kbos that are observed in today s kuiper belt .
specifically we find a power - law index of @xmath15 and a total mass in large kbos of @xmath16 of the initial planetesimal mass , which is consistent with the current observed mass in the kuiper belt .
we also make a prediction for the maximum mass ratio of kuiper belt binaries that formed by dynamical processes , for example , by three body interactions , and show that our prediction is in good agreement with the observations .
although our work focuses on the kuiper belt , the results also apply to early stages of planet formation and protoplanetary growth in debris disks .
our paper is structured as follows : in 2 we analytically describe the growth of large kbos , including their velocity dispersion and derive the slope of the kbo size distribution .
we confirm our analytic results in 3 with coagulation simulations . in 4 , we discuss how semi - collisional accretion , binary mergers and frequent planetesimal collisions would affect our results .
we show that our results on the kbo growth and velocity dispersion have interesting implications for the formation of kuiper belt binaries and predict the maximum mass ratio for binaries that formed by dynamical processes in 5 .
discussion and conclusions follow in 6 .
in order to gain an analytic understanding of the growth processes of large kbos and the associated velocity evolution we use the two - groups approximation @xcite . the ` two - groups approximation ' consists of the identification of two groups of objects , small ones , that contain most of the total mass with mass surface density @xmath0 , and large ones , that contain only a small fraction of the total mass with mass surface density @xmath17 .
we define @xmath2 as the mass surface density in a single logarithmic mass interval , that includes the largest bodies formed at a given time .
in contrast @xmath0 is defined as the total mass in small objects . within the framework of the ` two - groups approximation '
we arrive at the following picture for kbo growth .
initially all the mass is in small bodies . as the small bodies start to accrete each other , large bodies begin to form . to simplify the argument
, we only consider the mass surface density of the small and large bodies here , ignoring intermediate size bodies for now . as we show later , the large and small bodies alone determine the velocity dispersion for bodies of all sizes and only the large bodies grow significantly . in the initial stage
, @xmath2 grows due to the accretion of small bodies .
therefore the size of the largest bodies and the total mass in large bodies increases with time . during this growth
phase the velocity dispersion of the small bodies increases due to viscous stirring by the large bodies .
the velocity dispersion of large bodies is damped by dynamical friction provided by the small bodies .
@xmath2 continues to grow until the growth of large kbos by accretion of comparable size objects starts to compete with growth by accretion of small bodies . from then on
, @xmath2 remains roughly constant in a given logarithmic mass interval , while the size of the large kbos grows linearly with time .
how the kbo growth ended and how exactly the small bodies were lost from the kuiper belt are still the subject of ongoing research and are unimportant for the purpose of this paper and we therefore will not discuss them here further . we confirm the outlined kbo growth analytically and with numerical simulations .
we show that the mass - ratio , @xmath18 , is not arbitrary but an outcome of kbo growth .
large kbos viscously stir the small bodies , increasing the small bodies velocity dispersion @xmath19 .
as a result @xmath19 grows on the same timescale as @xmath6 , as long as the small bodies experience no significant damping by either gas or mutual collisions , which are , most likely , not yet important ( see section [ s3 ] ) .
we can therefore write the evolution of the small bodies velocity dispersion as @xmath20 where @xmath21 is the orbital frequency around the sun , @xmath6 the radius of the the large bodies and @xmath22 is their material density .
the hill velocity of the large bodies is @xmath23 , which is given by @xmath24 where @xmath25 is the hill radius , @xmath26 , where @xmath27 is the mass of the sun , @xmath28 and @xmath29 are the mass and semi - major - axis of the large kbos .
finally , @xmath30 and is @xmath31 at the distance of the kuiper belt .
we assumed in writing the expression for @xmath19 that @xmath32 , we will verify that this is the correct and self - consistent velocity regime for @xmath19 at the end of this section .
initially , the large bodies grow by accreting the small ones . their growth rate is given by @xmath33 where we assumed that @xmath34 , which we verify later in this section .
equating the rates from equations ( [ e02 ] ) and ( [ e03 ] ) and solving for @xmath19 we find @xmath35 @xcite .
the velocity @xmath36 of large kbos increases due to mutual viscous stirring , but is damped by dynamical friction from the sea of small bodies such that @xmath37 .
the competition between the stirring and damping can be written as @xmath38 balancing the stirring and damping rates for @xmath36 and substituting for @xmath19 from equation ( [ e2 ] ) into equation ( [ e2_3 ] ) , we find @xmath39 having derived expressions for the large and small bodies velocity dispersions we now turn to examining the growth in more detail .
expressions similar to the ones above have been widely used in the literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , in these works @xmath18 was treated as a free parameter . here
we present our method for deriving @xmath18 and calculate its value during runaway growth . within our ` two - groups approximation '
large bodies have two distinct modes for growth . in the first
, they can grow by the accretion of small bodies . in the second ,
they grow by accreting objects comparable to their own size .
the growth rate for large kbos is given by @xmath40 where we used the accretion rate corresponding to sub - hill velocity dispersions for large bodies and the corresponding rate for @xmath41 for small bodies .
the first term in expression ( [ e4 ] ) describes the growth of large kbos by the accretion of small bodies and the second corresponds to the growth of large kbos by accreting objects of their own size . substituting the expression for @xmath19 from equation ( [ e2 ] )
we have @xmath42 comparing the two terms in expression ( [ e5 ] ) we find that they contribute about equally to the growth of large kbos , if @xmath43 therefore , growth of large objects will be dominated by the accretion of small bodies , if @xmath44 . if on the other hand @xmath45 , then accretion of comparable size objects will be the dominant mode of growth provided that @xmath36 remains less than the hill velocity . since initially @xmath44 it follows that the growth of large kbos was at the beginning dominated by the accretion of small bodies , as assumed in equation ( [ e03 ] ) .
this mode of growth continues until @xmath46 at which stage accretion of comparable size bodies starts to compete with growth by accretion of small bodies . from then on ,
large kbos grow in roughly equal amounts by accreting small bodies and by merging with comparably - sized kbos . as a result @xmath2 remains constant , since its increase due to accretion of small bodies
is counteracted by a decrease due to the accretion of large kbos .
the radii of large kbos grows linearly with time ( see equation [ [ e5 ] ] ) .
we find it very encouraging that the observed mass in large kbos in each logarithmic size interval estimated from recent kuiper belt surveys ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) is about @xmath47 of that of the minimum mass solar nebula ( mmsn ) @xcite extrapolated to a heliocentric distance of 40 au .
the observed mass in large kbos ( @xmath12 ) is therefore consistent with @xmath48 and with the hypothesis that they formed by coagulation in the kuiper belt from an mmsn . the growth of kbos ended in this runaway phase with @xmath49 . in the context of planet formation
this phase of runaway growth is terminated by the onset of oligarchic growth , when each large body dominates the stirring in its own feeding zone , allowing @xmath2 to become comparable to @xmath0 .
given this understanding of the kbo growth we now show that our choices for the velocity regimes of @xmath19 and @xmath36 are self - consistent with this picture . substituting @xmath48 into equation ( [ e2 ] ) , we find that @xmath50 .
this implies that @xmath19 is about a few times the hill velocity of the large bodies .
we note here that our derivation of the expression for @xmath19 , specifically equating equations ( [ e03 ] ) and ( [ e02 ] ) , remains valid even when the large bodies contribute significantly to the growth of large kbos , because the growth rates due to the accretion of small and large bodies are comparable . evaluating equation ( [ e3 ] )
we find that @xmath51 .
this implies that the velocity dispersion of large bodies was sub - hill ( i.e. @xmath34 ) during the formation of large 100-km - sized kbos while that of the small bodies was super - hill ( i.e. @xmath41 ) .
this confirms that we used the correct velocity regime for @xmath19 and @xmath36 in the derivation of the kbo accretion rates above , ensuring that our treatment is self - consistent . rewriting equation ( [ e6 ] ) as @xmath52 as we have shown above , @xmath2 of the largest bodies is constant with a value of @xmath53 .
once large bodies form , their mass per logarithmic mass interval is , apart from a very brief period ( see section [ subhill2 ] ) , preserved .
this is because bodies smaller than the largest objects , do not grow significantly on the growth timescale of the largest objects , i.e. the growth is in the runaway regime ( see section [ superhill ] ) .
furthermore , such smaller bodies are also not efficiently consumed by larger objects .
their mass surface density and size distribution are therefore frozen ( see section [ superhill ] ) .
thus , @xmath2 is constant in time , which results in @xmath54 since @xmath55 .
this implies a power - law index @xmath56 ( see equation ( [ e1 ] ) ) for large kbos .
our work suggests that the growth of large kbos resulted from the accretion of small and large kbos alike and that this mode of growth in the runaway regime gave rise to the observed size distribution of large kbos .
we confirm this result in section [ s2 ] using coagulation simulations .
our findings are consistent with direct observations of the size distribution of large kbos @xcite and also agree with results from numerical coagulation simulations that model the growth of kbos carried out by other groups ( e.g. * ? ? ?
* ; * ? ? ?
so far we have only considered two sizes of bodies , small ones and large ones .
we now turn our attention to intermediate size bodies with radii @xmath57 , mass surface density @xmath58 and velocity dispersion @xmath59 . in sections
[ superhill ] and [ subhill2 ] , we show that @xmath58 is of the order @xmath2 .
this is because large bodies form by accreting small bodies until their mass surface density reaches @xmath60 . from then on , the mass surface density remains roughly constant in a given logarithmic mass interval .
@xmath61 because once larger bodies form , the velocity dispersion of the now intermediate sized bodies quickly grows to super hill velocities ( i.e. @xmath62 ) , which implies that such intermediate sized bodies do not grow significantly on the growth timescale of the largest kbos and that they are not consumed efficiently by the larger objects .
the size distribution of such intermediate sized bodies and @xmath58 therefore remain constant . during the brief period
over which @xmath63 , intermediate sized bodies are efficiently accreted by the largest objects , but we show in section [ subhill2 ] that this phase is very short and that it extends over less than a factor of 2 in radius . as a result , @xmath58 is of the order of , but slightly depleted compared to , @xmath2 .
there are three different velocity regimes for intermediate sized bodies that we have to consider separately . in the first regime
the intermediate sized bodies velocity dispersion exceeds the hill velocity of the largest bodies , i.e. @xmath62 , and the bodies themselves are sufficiently small such that their velocity dispersion is not efficiently damped by dynamical friction . in other words ,
the dynamical friction timescale for these intermediate sized bodies exceeds the growth timescale of the large bodies .
the velocity dispersion of these bodies is therefore dominated by gravitational stirring from the large bodies and it grows on the same timescales as the size of the large bodies .
this implies that @xmath65 .
@xmath66 where we substituted @xmath46 from equation ( [ e6 ] ) in the last step . equating the dynamical friction timescale for bodies of size @xmath57 to the growth timescale of the large bodies and solving for @xmath57
, we have @xmath67 this implies that @xmath65 for intermediate sized bodies with @xmath68 and smaller . for bodies above this size , damping by dynamical friction
is important . in the second velocity regime , which applies for bodies larger than @xmath69 and that have @xmath62
, the evolution of the velocity dispersion is dominated by gravitational stirring from the large bodies and damping by dynamical friction generated by the small bodies .
this yields a velocity dispersion that is governed by @xmath70 balancing the stirring and damping rates and substituting @xmath71 , we find that @xmath59 is given by @xmath72 where we substituted for @xmath19 from equation ( [ e2 ] ) and used @xmath73 in the last step . equation ( [ e10 ] ) yields that @xmath74 for kbos with @xmath75 .
therefore , @xmath76 for bodies with @xmath77 .
this implies that kbos with radii smaller than @xmath78 have super - hill velocities ( i.e. , @xmath64 ) and those with radii larger than @xmath78 will have sub - hill velocities ( i.e. , @xmath63 ) .
the growth of intermediate sized bodies that have @xmath62 and @xmath79 is given by @xmath80 substituting for @xmath19 from equation ( [ e2 ] ) and using again the result that @xmath46 we can write the above expression as @xmath81 where the first term corresponds to growth by the accretion of small bodies and the second term to growth by mergers of comparable sized bodies . since @xmath82 , we find when comparing the magnitude of the two growth terms in equation ( [ e12 ] ) , that the growth is dominated by the accretion of small bodies .
we assumed that @xmath58 is of the order of @xmath2 , which we confirm in section [ subhill2 ] .
intermediate sized bodies with @xmath83 therefore grow predominantly by accreting small bodies and their growth rate is reduced by a factor of @xmath84 compared to bodies of size @xmath6 .
this implies that bodies of size @xmath57 do not get the chance to grow , compared to the growth time scale of bodies of size @xmath6 .
these intermediate sized bodies do not contribute significantly to the growth of the large bodies . when examining the contributions from bodies with @xmath62 to the growth of bodies of size @xmath6 , we have @xmath85 where the first and second term correspond to growth by merging with bodies of size @xmath6 and @xmath57 , respectively .
since the first term in equation ( [ e41 ] ) exceeds the second , large bodies grow predominantly by accreting small bodies and large bodies . intermediate sized bodies with @xmath62 are not important for this growth .
furthermore , such intermediate sized bodies are only inefficiently accreted by bodies of size @xmath6 .
this is apparent when examining the rate of change of their surface density due to accretion onto bodies of size @xmath6 : @xmath86 comparing equation ( [ e42 ] ) with equation ( [ e4 ] ) , we find that @xmath58 does not change significantly on the growth timescale of the large bodies , which implies that their surface density is not altered due to accretion onto large bodies . as a result
the mass surface density per logarithmic mass interval of intermediate sized bodies with @xmath62 remains constant in time .
intermediate sized bodies that are about half the size of the large bodies and larger will have a velocity dispersion that is smaller than the large bodies hill velocity .
the evolution of the velocity dispersion of such intermediate sized bodies is determined by gravitational stirring from large bodies and damping by dynamical friction generated by the small bodies .
the expression for the evolution of @xmath59 can be written as @xmath88 balancing the stirring and damping rates and substituting @xmath71 , we can write @xmath59 as @xmath89 where we substituted for @xmath19 from equation ( [ e2 ] ) and used @xmath73 in the last step .
strictly speaking , we should multiply the right hand - side of equation ( [ e14 ] ) by a logarithmic factor , which is given by @xmath90 , where @xmath91 is equal to @xmath57 , if @xmath92 . if @xmath93 then @xmath94 is the radius that corresponds to bodies with a velocity dispersion that is less than or equal to @xmath95 .
the base of the logarithm is the same as the base of the logarithmic mass interval over which @xmath2 is defined .
this logarithmic factor arises because all bodies larger than or equal to @xmath57 that have a velocity dispersion less than @xmath96 contribute to the gravitational stirring for @xmath56 .
this is a key difference compared to the super - hill velocity regime where the stirring is dominated by the large bodies alone .
the growth of intermediate sized bodies with @xmath97 is given by @xmath98 where @xmath99 the radius , @xmath100 , for which @xmath101 can be found from equation ( [ e14 ] ) when including the logarithmic factor and it is given by @xmath102)^{1/4}$ ] .
the @xmath93 regime given in equation ( [ e15 ] ) yields the same expression for the accretion rate as given by equation ( [ e12 ] ) . the growth in this case , as shown above ,
is dominated by the accretion of small bodies . in the second case , which @xmath103 , the rate of growth of bodies with radii @xmath57 is @xmath104 where we substituted again for @xmath19 from equation [ e2 ] and used @xmath46 .
the first term corresponds to growth by the accretion of small bodies and the second to growth by merger of comparable sized bodies . comparing the magnitude of the two growth terms in equation ( [ e18 ] )
we find that , unlike in the regime discussed above , the growth of bodies with radius @xmath57 is dominated by the accretion of similar sized bodies rather than by the accretion of small bodies , again assuming that @xmath58 is of the order of @xmath2 , which we show below .
we note here that similar to the sub - hill velocity excitation rate , the mass accretion rate that corresponds to the accretion of comparable sized bodies , should have been multiplied by a logarithmic factor , which is given by @xmath105 $ ] . the maximum value for @xmath106 is given by @xmath107 .
strictly speaking , we should also have included this logarithmic factor in the derivation of equation ( [ e6 ] ) .
however , since this logarithmic factor only provides a small correction it does not warrant to be included here , since we have been neglecting factors of order unity throughout .
although the overall growth proceeds in the runaway regime ( i.e , @xmath108/dt)/(d \rm{log}[r']/dt)=r / r'$ ] ) , large bodies with @xmath92 grow in an orderly fashion with respect to each other if @xmath56 , which implies that their radii converge .
this leads to a steepening in the size distribution at very large radii , but only for bodies with radii that have a corresponding velocity dispersion such that @xmath92 , which corresponds to less than a factor of 2 in radius .
the largest effect this can have is to reduce @xmath2 by a factor of @xmath109 , such that @xmath110 , since orderly growth will cease when @xmath111 .
intermediate sized bodies with @xmath63 significantly contribute to the growth of large bodies , since @xmath112 where the first and second term correspond to growth by merging with bodies of size @xmath6 and @xmath57 , respectively .
this implies that for @xmath56 , i.e. @xmath113 , large bodies of size @xmath6 grow at comparable rates by accreting bodies of their own size and bodies of size @xmath114 . because these intermediate sized bodies are efficiently accreted by large bodies , their surface density is altered at a rate @xmath115 comparing equation ( [ e43 ] ) with equation ( [ e4 ] ) , we find that @xmath58 changes on the order of the growth timescale of the large bodies . as a result ,
the mass surface density per logarithmic mass interval of intermediate sized bodies with @xmath63 gets depleted , because such bodies are efficiently consumed by large bodies . however , since the size range is very small , such bodies are only depleted by factor of a few before their velocity dispersions become super - hill at which point they are no longer efficiently accreted .
the orderly growth of bodies with @xmath92 relative to each other and the efficient consumption of bodies with @xmath63 are responsible for the steepening of the kbo size distribution seen at @xmath116 in figure [ fig2 ] .
the aim of our coagulation simulation is to test our analytic results outlined above .
its purpose is to capture the dominating physical processes that give rise to the kbo size distribution .
we attempt by no means to present the most detailed or precise kbo formation simulation , since several such works already exist in the current literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we therefore neglect factors of order unity in the accretion , stirring and damping rates .
in addition , we neglect effects of gas damping and possible dynamical stirring from neptune .
we investigate the kbo growth in a single annulus located at roughly 40 au from the sun with a width of about 10 au .
most of our simulations start with a total mass of about 40 earth masses in small planetesimals , which corresponds to @xmath117 .
this mass surface density is consistent with extrapolations of the mmsn @xcite to 40 au , after it has been enhanced about 6 fold as required for the formation of uranus and neptune ( e.g. * ? ? ?
* ; * ? ? ?
we follow the mass growth and the evolution of the velocity dispersion of the kbos using safronov s statistical approach @xcite .
since we are primarily concerned with the initial growth phase , we assume here that all collisions lead to accretion .
this assumption is justified during the initial kbo growth and remains justified even once the planetesimal s velocity dispersion has been excited above its own escape velocity as long as the growth timescale of the large kbos is short compared to the planetesimal - planetesimal collision time .
this condition is generally fulfilled , if the initial planetesimals are of the order of a kilometer in size or larger , since , for 1-km - sized bodies , the planetesimal - planetesimal collision time is comparable to the formation time of pluto .
we therefore neglect the effect of destructive planetesimal collisions here , but we discuss how they would effect our results , had they been important , in section 4 . below we give the relevant stirring , damping and accretion rates that determine the growth and velocity evolution in the kuiper belt .
a detailed derivation of the these rates can , for example , be found in @xcite .
the accretion rate between bodies of two different mass bins is given by @xmath118 where the subscripts ` s ' and ` b ' correspond to the mass bin with the smaller and larger bodies , respectively .
the number of bodies in a given mass bin is given by @xmath119 and @xmath120 is the relative velocity , which was approximated by @xmath121 $ ] .
note , we quote here the sub - hill accretion rate applicable to large objects , since the velocity dispersion of the small bodies is super - hill throughout the growth .
the corresponding velocity evolution for each mass bin is dominated by the following processes .
the velocity dispersion of the mass bin that corresponds to the smaller objects is viscously stirred by the object in the larger mass bin at a rate given by @xmath122 the velocity dispersion of the bodies in the larger of the two mass bin is in turn damped by the smaller bodies .
the damping rate of the large bodies velocity dispersion is given by @xmath123 we implemented the above equations in our coagulation code and followed the growth and velocity evolution of the various mass bins .
as initial conditions we start with an equivalent of about 40 earth masses , in 1-km sized planetesimals with an initial velocity dispersion of 3 times their hill velocity .
we follow the kbo growth and velocity evolution of the different mass bins for 70 myrs .
the results of our coagulation simulation are shown in figures [ fig1 ] , [ fig2 ] , and [ fig3 ] .
figure [ fig1 ] shows the evolution of the cumulative mass distribution , the cumulative number distribution and the velocity dispersion during kbo growth as a function of time .
it is apparent from figure [ fig1 ] that the growth and velocity evolution become self - similar once @xmath124 , i.e. the shape of the size - distribution and velocity - distribution remains unchanged , while the maximum kbo size continues to grow .
the two upper panels of figure [ fig1 ] and figure [ fig2 ] show that the size - distribution of large kbos ( @xmath12 ) indeed follows a power - law with @xmath56 as predicted by our analytic treatment in section [ s1 ] .
this implies a roughly equal amount of mass per logarithmic mass bin for large kbos .
moreover , the middle panel of figure [ fig1 ] and figure [ fig2 ] show that the mass ratio of large to small kbos , i.e. @xmath18 , found in our coagulation simulation agrees very well with our analytic prediction that @xmath125 .
the simulations confirm our analytic results and suggest that the total mass in large objects that we see in the kuiper belt is not arbitrary but an outcome of the kbo growth and that it is roughly @xmath47 of the initial planetesimals mass .
this result is in excellent agreement with the actual observed mass in large kbos and formation from a mmsn type disk .
our work therefore suggests that the kuiper belt did not contain two to three order of magnitude more mass in large kbos as has been proposed by some models @xcite .
figure [ fig3 ] shows a comparison of the velocity dispersion from our coagulation simulation and our analytic results derived in section [ s1 ] .
it displays very good agreement between the velocity dispersion that we derived analytically for the various size regimes , and the results from our coagulation simulation .
figures [ fig1]-[fig3 ] show that our analytic work captures the essential features of kbo growth and that analytic theory and the numerical coagulation results are in excellent agreement .
we are able to successfully explain the slope ( @xmath56 ) and amplitude ( @xmath126 ) of the large kbo size distribution and the evolution of the velocity dispersion in the various velocity and size regimes .
[ htp ] [ htp ] we performed an additional set of coagulation simulations with different initial conditions .
in the first set we started with most of the mass in small 1-km - sized objects , just as before , but added a second population of larger 10-km - sized kbos that contained @xmath47 of the total mass .
all bodies were started with a velocity dispersion equal to 3 times their hill velocity and we followed their growth and velocity evolution .
figure [ fig4 ] shows the result of this growth ( points ) and a comparison with the cumulative mass distribution at various times for kbos that grew from a single population of 1-km sized planetesimals ( lines ) .
the similarity between the two distributions is striking .
the same power - law for the large kbo size distribution emerges and the mass ratio in large and small kbos becomes also the same in both simulations .
these results highlight the power of the ` two groups approximation ' that we used to derive the analytic results in section [ s1 ] and validate our assertion that the growth of large kbos develops towards a state where small and large bodies contribute about equally to the growth .
the overall growth timescale to reach pluto - mass objects differs in these two simulations .
starting from 40 earth masses in small planetesimals , it takes in our simulations about 70 - 80 myrs to form pluto - sized objects if initially all kbos are 1-km in radius but only 40 - 50 myrs if , in addition to the 1-km - sized objects , there also existed a small population of 10-km - sized kbos .
the difference in the growth time results from the timescale it takes 1-km - sized objects to grow into 10-km sized bodies , if none are already present .
when only comparing the growth timescale from 50-km - sized objects to 1000-km sized objects , we find that they are the same in the two scenarios ( see figure [ fig4 ] ) , confirming that it is the initial growth of objects to tens of kilometers in size that give rise to the overall difference in pluto - formation time between the two simulations .
the final shape of the kbo size - distribution and the total mass in large kbos ( as a fraction of the initial mass ) is therefore independent of the initial size distribution of the planetesimals .
we also find the same kbo size distribution and overall mass in large kbos , if the initial planetesimals have a power - law size distribution with various power - law indexes and radii ranging from 1 to 10-km .
in addition , we confirm that our results are independent of the details of the initial planetesimal velocity dispersion , as long as it is below their escape velocity .
[ htp ] we also performed a second set of coagulation simulations , which we started with the same total planetesimal mass but with an initial planetesimal size of 500 m in radius rather than 1 km .
figure [ fig5 ] shows that the shape of the size distribution does not depend on the initial planetesimal size .
this result remains valid as long as the initial planetesimal size is large enough such that the planetesimal collision timescale exceeds the kbo growth timescale .
[ htp ] these results confirm earlier findings by @xcite , who note that the power - law slope for large objects is remarkably independent of the input parameters and initial conditions .
our work offers an explanation of why this is so .
since the size distribution of large kbos reaches an equilibrium state that evolves self - similarly , the signature of the initial planetesimal size - distribution and their initial velocity dispersion are erased .
the slope of the kbo size distribution and the formation timescales that we find in our simulations for large kbos is in agreement with results from previous coagulation simulations @xcite . since our analytic result for the mass ratio between large and small kbos ( i.e. , @xmath46 )
can be expressed as a function of @xmath3 alone , we performed an additional set of coagulation simulations for a value of @xmath3 that was decreased by a factor of 100 .
figure [ fig6 ] shows the coagulation results for @xmath127 ( dashed lines ) , which is roughly the value for the kuiper belt , and @xmath128 ( solid lines ) .
we increased the initial planetesimal mass in the @xmath129 simulation by a factor of @xmath130 to speed up the planetesimal growth . due to this mass increase
, the resulting mass per logarithmic mass interval in large bodies should be the same in both simulations .
this is indeed what we see in figure [ fig6 ] , confirming our analytic expression for @xmath2 .
figure [ fig6 ] shows also that the equilibrium growth state , where growth by accretion of small bodies is comparable to the growth due to similar sized mergers , is reached later for @xmath128 compared to @xmath127 .
we confirmed that this is not caused by the increased planetesimal mass that was used in the @xmath129 simulation .
we have so far neglected collisions among the small bodies . in this section ,
we first derive the planetesimal - planetesimal collision timescale and show that planetesimal collisions can be neglected during the kbo growth for planetesimal sizes of about a kilometer and larger .
we then discuss how our results would be altered , if planetesimal collisions had been important during the kbo growth .
in addition , we examine the effect of semi - collisional accretion @xcite and frequent binary mergers on the growth of kbos .
the planetesimal - planetesimal collision time for objects with radius , @xmath131 , and with a velocity dispersion that has been excited above their escape velocity is given by @xmath132 comparing equation ( [ ec1 ] ) to the pluto formation timescale , which is given by the inverse of equation ( [ e5 ] ) we find that @xmath133 which is @xmath134 for 1-km - sized planetesimals and @xmath135 .
1-km is about the size that planetesimals are expected to have , if they formed by gravitational instability without dissipation of internal angular momentum @xcite .
this implies that if kbos grew from a population of 1-km sized planetesimals then mutual planetesimal collisions become important only once kbos comparable to the size of pluto have formed .
if planetesimal - planetesimal collisions are important , then they would , most likely , break - up the planetesimals and lead to a damping of the small bodies velocity dispersion , which would therefore no longer grow as @xmath6 , as we assumed in section [ s1 ] .
instead , it would be set by balancing the dynamical stirring from large bodies by collisional cooling .
this yields @xmath136 for @xmath137 , where @xmath138 is the radius of the small bodies @xcite .
we note here that @xmath137 is most likely the relevant case to consider , since @xmath19 is only slightly super - hill when collisional cooling is neglected ( see section [ s1 ] ) . in this case , the growth of kbos is given by @xmath139 where we assumed that @xmath140 . substituting for @xmath19 from equation ( [ ce1 ] ) and comparing the two growth rates in equation ( [ ce2 ] ) , we find that the two growth rates become equal when @xmath141 this implies , that in this case , the growth of large bodies is dominated by the accretion of small bodies until more mass is converted into @xmath2 , compared to the case without frequent planetesimal collisions .
the total mass in large kbos and the slope of their size distribution are therefore likely to be different from our results in section [ s1 ] and [ s2 ] , if collisions were important
. the kbo growth would be halted , if collisions lead to the onset of a collisional cascade and if the small collisional fragments are efficiently lost from the kuiper belt .
so far , we have treated the planetesimal accretion as collisionless , meaning that collisions among planetesimals can be neglected while they are inside the hill sphere of a growing kbo .
this assumption is valid , if kbos formed from km - sized planetesimals .
however , if planetesimals are of the order of a meter in size or smaller , either because they formed small , or because they were broken into small pieces , then they are more likely to collide with each other inside a kbo s hill sphere than to accrete onto the growing kbo directly @xcite .
we call this semi - collisional accretion . such planetesimal collisions inside the hill sphere lead to the formation of an accretion disk around the kbos .
if kbos grew by semi - collisional accretion than their growth could have been very fast , because the effective radius for accretion , in this case , is of the order of their hill sphere .
the kbo growth would be dominated by the accretion of small planetesimals for much longer in the semi - collisional regime , compared to the growth scenario discussed and investigated in sections [ s1 ] and [ s2 ] .
finally , the growth of kbos might also have been aided by merging comparable mass kuiper belt binaries .
as we discuss in section [ sec - binaries ] in detail , a significant fraction of kbos reside in comparable mass binary systems and such systems likely formed by dynamical processes such as by three body capture , or by binary formation aided by dynamical friction @xcite .
once formed , the mutual semi - major axis of such binary systems starts to shrink due to dynamical friction generated by the small planetesimals @xcite .
this eventually leads to merging between the binary components .
this channel of growth could be important , because the binary formation time and the time it take for a binary to spiral in until contact occurs are comparable to the timescale on which large kbos grow @xcite . if merging of binary components is important in the overall kbo growth then the growth rate would be enhanced compared to equation ( [ e4 ] ) .
this results in a reduced mass surface density for large kbos , but the power - law slope of the large kbo size distribution will remain unchanged .
our coagulation results also have interesting implications for the formation of kuiper belt binaries . a significant fraction of kbos are part of a binary system .
the binary fraction varies for different dynamical classes and it is highest in the cold classical belt @xcite , where it is about 30% .
more than 70 binaries have been discovered in the kuiper belt to date and their number continues to rise .
high - mass ratio binary systems , including pluto / charon , likely formed in a collision and subsequent tidal evolution . however , the majority of kuiper belt binaries consist of comparable mass companions with wide separations and have too much angular momenta to have formed by the same mechanism .
instead these systems most likely have a dynamical origin like , for example , binary formation by three body capture , or binary formation aided by dynamical friction @xcite .
the hill sphere , the region interior to an object s hill radius , sets the maximum phase space available for binary formation by such dynamical processes .
dynamical binary formation scenarios take advantage of the increased size of the hill radius , which is more than an order of magnitude larger in the kuiper belt than for similar sized objects in the asteroid belt and therefore make the kuiper belt the ideal place for the formation of wide , comparable mass binaries .
dynamical binary formation scenarios require @xmath63 for efficient formation , since binary formation rates quickly exceed the age of the solar system , once the velocity dispersion significantly exceeds the hill velocity @xcite .
since kbo growth and binary formation occurs concurrently , we can predict the maximum mass ratio for such binary systems because we know the range of kbo radii that have sub - hill velocities at any given stage during the growth . from equation ( [ e10 ] ) we have @xmath63 for kbos with radii greater than @xmath75 .
this implies a maximum mass ratio of @xmath142 between the primary and secondary and translates into a maximum magnitude difference between the binary components , assuming similar albedo , of about @xmath143 .
this provides an explanation for the strong clustering in @xmath144 of binaries in the cold classical belt .
this clustering was first pointed out by @xcite and is shown in figure [ fig8 ] .
figure [ fig8 ] shows that all binaries in the cold classical belt have @xmath145 , i.e. they lie in the region enclosed by the solid red lines .
our coagulation work therefore suggests that the observed binaries in the classical belt formed by dynamical processes during the growth of the kbos themselves and that this formation most likely took place in situ .
the cold classical belt is therefore not only distinct in its inclination distribution @xcite , color @xcite and binary fraction @xcite , but also in terms of the type of its binaries , i.e. they are exclusively comparable mass ratio binaries with @xmath146 @xcite .
figure [ fig8 ] shows that , in contrast to the cold classical population , other dynamical classes have binaries with a wide range of mass ratios and they therefore most likely contain dynamically and collisionally formed binary systems . the fact that the other dynamical classes have comparable mass binaries as well as systems with a wider range of component sizes is consistent with the idea that they are a superposition of two original populations , namely a dynamical cold ( low inclination ) population that formed close to its current location and a dynamical hot ( high inclination ) population that formed closer to the sun .
kuiper belt binaries may therefore prove to be useful probes for entangling the two original dynamical populations , if they existed ( murray - clay & schlichting 2010 , submitted ) .
we carried out an analytic and numerical investigation of runaway growth of protoplanets with a focus on the kuiper belt .
since , the kuiper belt is a remnant of the primordial solar system where planet formation never reached completion , it contains some of the least processed bodies . as a result
, it provides a snapshot of earlier stages of planet formation , which were erased elsewhere in the solar system where planet formation proceeded to completion .
our results for runaway growth , which we summarize below , therefore , do not only apply to the kuiper belt but also to planet formation before the onset of oligarchic growth .
in addition , our findings apply to proto planetary growth in debris disks around other stars , as long as the protoplanets growth time is shorter than the planetesimal - planetesimal collision time , such that dynamical cooling and possible mass loss due to frequent planetesimal collisions can be neglected . in this paper
, we presented analytic work that describes the growth of kbos , the evolution of their velocity dispersion , and that provides insights into the underlying physical processes that give rise to the kbo size distribution .
our work successfully explains the observed slope of the kbo size distribution as well as the total mass that is present in large kbos today .
in addition it predicts the maximum mass - ratio of kuiper belt binaries that formed by dynamical processes , which explains the observed clustering in binary companion sizes that is seen in the cold classic belt .
we confirmed our analytic results with numerical coagulation simulations .
we find that the kbo growth proceeds as follows .
initially all the mass resides in small planetesimals and large kbos start to form by accreting small planetesimals .
this growth continues until growth by merging with comparable sized kbos become comparable to growth by accreting small bodies .
we show that this condition sets in when @xmath147 . from that time
onwards , the growth and the evolution of the velocity dispersion become self - similar and @xmath2 remains roughly constant , since the increase in @xmath2 by the accretion of small planetesimals is balanced by a decease due to the accretion of large bodies .
we showed that this mode of growth leads to a kbo size distribution with power - law index @xmath56 .
this is in good agreement with observations of the kuiper belt size distribution , which is well described by a power - law with an index that is consistent with @xmath56 within @xmath148 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a single albedo is assumed for all sizes when converting from the observed magnitude distribution to the kbo size distribution .
therefore , possible albedo variations as a function of size could introduce a significant uncertainty in the estimate of @xmath10 .
the best fit value for the power - law index is typically found to somewhat exceed @xmath56 .
we note that this could be due to fitting the high end of the size distribution that corresponds to the largest kbos , which is somewhat steeper than the rest of the distribution ( see figure [ fig1 ] ) .
this steeper end of the kbo size distribution is due to the fact that the largest objects did not have enough time to grow to their steady - state abundance . if kbos formed by coagulation from km - sized planetesimals , then there could not have been significantly more mass in large kbos than what is observed today , unless the mmsn was initially enhanced by several orders of magnitude .
this result is in good agreement with the current mass in large kbos , since it is about @xmath149 of the mmsn that was enhanced by a factor of a few , as required for the formation of uranus and neptune .
the observed mass in large kbos and their size distribution therefore support the hypothesis that kbos formed from a mmsn type disk by coagulation from kilometer sized planetesimals .
the growth of kbos ended in this runaway phase with @xmath150 . in the context of planet formation
this phase of runaway growth was terminated by the onset of oligarchic growth when each large body dominates the stirring in its own feeding zone . during oligarchic growth @xmath2
continues to grow until it becomes comparable to @xmath0 .
our understanding of the growth , the value of @xmath18 and the size distribution of large bodies differs from previous works that investigated the details of runaway growth @xcite .
this is mainly due to the fact that these works did not take into account the simultaneous evolution of the velocity dispersion during the growth but assumed either equipartition between the different mass bins @xcite and/or a constant velocity dispersion for all sizes @xcite . since efficient binary formation by dynamical processes can only proceed among kbos that have sub - hill velocity dispersions @xcite and because we follow the evolution of the kbo velocity dispersion as well as their growth
, we can predict the maximum mass ratio for such binary systems .
our work yields a maximum binary mass ratio of @xmath151 , or , in other words , a maximum magnitude difference between binary components , assuming similar albedos , of @xmath143 .
this explains the clustering in @xmath144 of all the observed kuiper belt binaries that are part of the cold classic belt , which make up about half of all known binary systems in the kuiper belt .
there exists some tentative observational evidence that the power - law size distribution might be different for kbos with inclinations greater and smaller than @xmath152 @xcite .
however , @xcite and @xcite found no conclusive evidence supporting such a difference .
it will be interesting to see the results of future kbo surveys that address this question .
should there indeed exist a difference between various dynamical classes then this could have very interesting implications for the formation and collisional evolution of the kuiper belt . ,
f. b. , zhang , z. , lehner , m. j. , mondal , s. , king , s. , giammarco , j. , holman , m. j. , coehlo , n. k. , wang , j. , alcock , c. , axelrod , t. , byun , y. , chen , w. p. , cook , k. h. , dave , r. , de pater , i. , kim , d. , lee , t. , lin , h. , lissauer , j. j. , marshall , s. l. , protopapas , p. , rice , j. a. , schwamb , m. e. , wang , s. , & wen , c. 2010 , , 139 , 1499 , j .-
. , kavelaars , j. , gladman , b. , margot , j. l. , nicholson , p. , jones , r. , parker , j. , ashby , m. , & campo bagatin , a. 2009 , in aas / division for planetary sciences meeting abstracts , vol .
40 , aas / division for planetary sciences meeting abstracts , 47.11+ | runway growth is an important stage in planet formation during which large protoplanets form , while most of the initial mass remains in small planetesimals .
the amount of mass converted into large protoplanets and their resulting size distribution are not well understood . here
, we use analytic work , that we confirm by coagulation simulations , to describe runaway growth and the corresponding evolution of the velocity dispersion .
we find that runaway growth proceeds as follows : initially all the mass resides in small planetesimals , with mass surface density @xmath0 , and large protoplanets start to form by accreting small planetesimals .
this growth continues until growth by merging large protoplanets becomes comparable to growth by planetesimal accretion .
this condition sets in when @xmath1 , where @xmath2 is the mass surface density in protoplanets in a given logarithmic mass interval and @xmath3 is the ratio of the size of a body to its hill radius . from then on , protoplanetary growth and the evolution of the velocity dispersion become self - similar and @xmath2 remains roughly constant , since an increase in @xmath2 by accretion of small planetesimals is balanced by a decrease due to merging with large protoplanets .
we show that this growth leads to a protoplanet size distribution given by @xmath4 where @xmath5 is the number of objects with radii greater than @xmath6 ( i.e. , a differential power - law index of 4 ) .
since only the largest bodies grow significantly during runaway growth , @xmath2 and thereby the size distribution is preserved .
we apply our results to the kuiper belt , which is a relic of runaway growth where planet formation never proceeded to completion .
our results successfully match the observed kuiper belt size distribution , they illuminate the physical processes that shaped it and explain the total mass that is present in large kuiper belt objects ( kbos ) today .
this work suggests that the current mass in large kbos is primordial and that it has not been significantly depleted .
we also predict a maximum mass - ratio of kuiper belt binaries that formed by dynamical processes of @xmath7 , which explains the observed clustering in binary companion sizes that is seen in the cold classical belt .
finally , our results also apply to growth in debris disks , as long as frequent planetesimal - planetesimal collisions are not important during the growth . |
as you may conclude from the different concluding remarks at this meeting there are various approaches to doing them .
most of my esteemed colleagues spoke on several topic that they consider important now and in the future .
i decided to concentrate on one single topic that excited not only me but a large number of astrophysicists that were never interested in ultra high energy cosmic rays ( uhecr ) - the observed correlation of the highest energy events detected by the pierre auger observatory in argentina with active galactic nuclei ( agn ) @xcite .
this discovery marked the beginning of a new type of astronomy - cosmic ray astronomy - that may become as important in the future as the tev gamma ray astronomy has recently become .
tev gamma ray astronomy does not only reveal the type of the gamma ray sources , it also provides an indirect measurement of the infrared / optical background whose estimates are based on theoretical calculations involving the light emission of different types of stars and galaxies and the cosmological evolution of these objects and the matter density around them in the universe .
cosmic ray astronomy also involves some general features of the nearby universe that are otherwise extremely difficult to measure such as the local ( within 200 mpc ) distribution of matter and different powerful astrophysical objects , the average strength of the intergalactic magnetic fields ( that could even be mapped when the sources are confirmed ) and the galactic ones .
i will discuss these results and report on the small contribution that i have made to this topic .
it is not connected to the type of the uhecr sources , but includes a better than 3@xmath0 confirmation of the anisotropy detected by auger @xcite .
the auger collaboration reported a correlation of their 27 highest energy events ( e @xmath1 5.7@xmath210@xmath3 ev ( 57 eev ) with active galactic nuclei ( agn ) at redshifts @xmath4 less than 0.017 @xcite from the vron - cetty and vron catalog @xcite ( v - c ) .
twenty out of 27 events are within 3.1@xmath5 of individual agn , while for isotropic arrival distribution one expects on average 5 coincidences .
five of the non - correlating events come from less than 12@xmath5 galactic latitude which may be understood as larger deflections in the galactic magnetic field .
the scattering in the galactic magnetic field is consisten with these events being protons , not heavier nuclei .
it is surprising that there are no events coming from the direction of the virgo cluster , that includes a large number of powerful galaxies in addition to m87 , as stated in ref .
@xcite .
a better description of the statistical analysis and a broad discussion if its features is given in the longer second paper @xcite .
the self correlation between the 27 events was studied and found to be consistent with the existence of many ( more than 61 ) sources within the 71 mpc radius , that leads to a source density of 3.5@xmath210@xmath6 mpc@xmath7 .
the auger collaboration warns us that because most of the correlating agn are not very powerful the observed anisotropy may be due to a different type of source .
it is quite interesting that two auger events coincide with the close by ( less than 4 mpc ) powerful radio galaxy cen a , and several other events are close to it . among
the criticism toward this analysis is the statement that redshift @xmath4 of 0.017 is much lower than the expected contribution of redshifts for extragalactic protons and nuclei .
protons are supposed to come to us from redshifts up to 0.05 , roughly distances to 200 mpc .
other authors complain about the types of agn in the v - c catalog that are low luminosity seyfert and liner galaxies .
since i do not have much knowledge about agn i decided to check the discovered anisotropy by studying the correlation of the auger events with the supergalactic plane ( sgp ) .
a report on the correlation of the then existing statistics of uhecr of energy above 4@xmath210@xmath3 ev with the supergalactic plane @xcite was published in 1995 @xcite .
the data set consisted mostly of events detected by the haverah park detector with the addition of events detected by agasa , yakutsk and volcano ranch air shower arrays .
the anisotropy of that data set was studied by a comparison of the average and rms supergalactic latitudes @xmath8 of the experimental events to that of an isotropic monte carlo sample .
the significance of the correlation was at the 3@xmath0 level .
later on , when the agasa detector dominated the world s statistics , the significance of the correlation decreased .
we decided to check on the correlation of the 27 auger events with the same definition of the nearby large scale mass distribution . the result is shown in fig . [ auger17 ] .
the probability for such a correlation was calculated with a monte carlo simulation where 100,000 sets of 27 events were injected isotropically in the auger field of view .
we arbitrarily chose to bands of 10@xmath5 and 15@xmath5 in supergalactic latitude that surround the 1@xmath0 deviation of the agn from the v - c catalog to the sgp .
nine ( 13 ) of the 27 experimental events fall in these two bands .
the random coincidence probability derived from the monte carlo is 0.024 ( 0.008 ) corresponding to 2.0 ( 2.4 ) @xmath0 . until now we dealt with the original definition @xcite of the supergalactic plane by de vaucouleurs .
this definition was studied in 2000 by lahav et al . in terms of redshift
@xcite on the basis of the optical redshift survey ( ors ) ( 8457 galaxies , 98% with redshift ) . since ors had a zone of avoidance of @xmath9 20@xmath5 it was complemented with iras galaxies in order to describe better the intersection of sgp with the galactic plane .
this study introduces correction to the definition of sgp for different distances from 0 to 80 mpc .
the one that is most suitable for analysis of the auger events is for distances up to 70 mpc , i.e. identical to the redshift of 0.017 .
the corrected shape of the sgp is plotted in fig .
[ auger16 ] the same way as the original definition is plotted in fig .
[ auger17 ] . respectively .
the probability to have so many events from isotropic arrival direction distribution drops to 1.6@xmath210@xmath10 ( 6.0@xmath210@xmath10 ) for distances from the supergalactic plane of 10@xmath5(15@xmath5 ) .
the correlation of the auger events with the updated definition of the sgp within 70 mpc is indeed much more significant - the probabilities above correspond to 3.6(3.2)@xmath0 . since the rotation angle of sgp in galactic coordinates in the analysis of lahav et al .
depends on the the redshifts of the astrophysical objects the one we used certainly defines better the plane of weight of the mass distribution within 70 mpc and of the uhecr sources if these cosmic rays indeed are produced within this distance .
even if the gzk horizon is larger we expect some degree of correlation with the nearby astrophysical objects .
it is important to note that the hires collaboration does not confirm the correlation with agn @xcite . out of 13 highest energy events
only two correlate with agn from the v - c catalog within 3.1@xmath5 .
the hires paper does not describe well their field of view and it is difficult to understand if the difference in the field of view is the reason for this discreapancy . there is also a lack of obvious correlation with sgp .
three ( 5 ) of the hires events are within 10@xmath5 ( 15@xmath5 ) of the sgp in the definition of ref .
@xcite .
we still believe that the anisotropy published by the auger collaboration is strong and with the increase of the experimental statistics , which must have doubled by august 2008 , it will lead us to identification of the sources of the highest energy cosmic rays .
99 the auger collaboration ( j. abraham et al . ) , science 318:938 ( 2007 ) the auger collaboration ( j. abraham et al . ) , astropart .
phys . 29:188 ( 2008 )
t. stanev , arxiv:0805.1746 m .- p .
vron - cetty & p. vron , a&a * 455 * , 773 ( 2006 ) d.s .
gorbunov et al . , arxiv:0804.1088 ( 2008 ) g. de vaucouleurs et al . , the third catalogue of bright galaxies ( rc3 ) university of texas press , austin ( 1991 ) t. stanev et al .
75*:305 ( 1995 ) o. lahav et al . , mnras , * 312 * 166 ( 2000 ) hires collaboration ( r.u .
abbasi et al ) , arxiv:0804.0382 | we discuss the anisotropic arrival directions of the ultra high energy cosmic rays detected by auger which i consider one of the biggest discoverie in astrophysics during the last year . |
the driven - diffusive lattice gas models are of current interest in non - equilibrium statistical mechanics @xcite . depending on the nature of the drive
, these driven - dissipative systems can attain steady - states that are far from equilibrium . the simplest driven - diffusive lattice gas model that incorporates inter - particle interactions
is the katz - lebowitz - spohn model @xcite ( from now onwards referred to as * kls * ) .
some of the particle - hopping models of vehicular traffic @xcite are closely related to some special limits of the kls model in one dimension .
therefore , in order to compare and contrast the spatio - temporal organizations and the flow properties of the kls model with those in the particle - hopping models of vehicular traffic , we calculate here those properties of the kls model which are important from the perspective of vehicular traffic . over the last decade
extensive investigations of vehicular traffic have been made using the so - called particle - hopping models which represent each vehicle by a particle @xcite .
all these traffic models are defined on discrete lattices each site of which , in the spirit of the lattice gas models , represents a cell that can accommodate at most one particle at a time . in almost all the standard particle - hopping models of vehicular traffic
the only non - vanishing inter - particle interaction is the mutual hard - core repulsion which is usually implemented through the condition of exclusion principle : no two particles are allowed to occupy the same lattice site simultaneously .
therefore , a comparison of our results on the kls model with the corresponding results for the particle - hopping models of vehicular traffic will show the effects of inter - particle interactions other than mere hard - core repulsion . the * flux * ( per lane ) is defined to be the number of particles ( per lane ) crossing a detector site per unit time . in the context of vehicular traffic @xcite , the most important quantity of interest is the so - called _ fundamental relation _ which depicts the dependence of the flux on the density of the vehicles .
the number of empty sites in between a pair of particles is usually taken as a measure of the corresponding distance - headway * dh*. in this paper we theoretically calculate the dh distributions and the flux in the steady - states of the kls model , separately for attractive and repulsive inter - particle interactions , within the framework of a _ cluster mean - field theory _ ( * mft * ) @xcite which has been very successful also in the theoretical treatment of the particle - hopping models of vehicular traffic @xcite .
we also indicate the level of the accuracy of our cluster - mft results by comparing these with the corresponding numerical data obtained from our computer simulations of the kls model .
the organization of this paper is as follows : in section ii we define the kls model and some related particle - hopping models which are relevant for our discussion in the subsequent sections .
we summarize in section iii the methods of the cluster mft we use for our theoretical calculations as well as those of computer simulation . in sections iv , v and vi , we present our theoretical results for the one - dimensional kls model ( both with attractive and repulsive interactions ) in the @xmath2-cluster , @xmath3-cluster and @xmath4-cluster approximations , respectively , together with the corresponding numerical data from our computer simulations .
we present our results for the two - dimensional kls model in section vii .
we compare and contrast the results for the kls model with the corresponding results for the particle - hopping models of vehicular traffic in section viii before summarizing the main results in the concluding section ix .
suppose the variable @xmath5 describes the state of occupation of the site @xmath6 ( @xmath7 ) on a discrete lattice ; @xmath5 is allowed to take one of the only two values , namely , @xmath8 if the site @xmath6 is occupied by a particle and @xmath9 if it is empty ( or , equivalently , occupied by a hole " ) .
the hamiltonian for the system , in the absence of any external driving field , is given by @xmath10 where the summation on the right hand side is to be carried out over all the nearest - neighbor pairs and @xmath11 takes into account the corresponding inter - particle interactions .
the kls model can be recast in the language commonly used in the theory of magnetism by using classical ising spin variables @xmath12 where @xmath13 and @xmath14 represent the particles and holes , respectively , and the corresponding hamiltonian , in the absence of the external drive , is given by @xmath15 the _ attractive _ and _ repulsive _ inter - particle interactions , captured by @xmath16 and @xmath17 , respectively , in the hamiltonian ( [ eq-1 ] ) correspond to the ferromagnetic and antiferromagnetic interactions in the form ( [ eq-2 ] ) of the hamiltonian .
however , throughout the rest of this paper , we shall use the particle - hole picture , where the instantaneous state ( configuration ) of the system at time @xmath18 is completely described by @xmath19 .
for example , in case of a system of length @xmath20 in dimension @xmath21 , @xmath22 .
similarly , for the @xmath23 square lattice @xmath24 .
the average density @xmath25 of the particles is given by @xmath26 where @xmath27 , the total number of available sites is @xmath20 for a linear chain and @xmath28 for a square lattice of size @xmath29 .
note that , because of the conservation of the particles , the density @xmath25 is conserved by the dynamics .
the dynamics of the system is governed by the well - known kawasaki dynamics : at any non - zero temperature @xmath30 , a randomly chosen nearest - neighbor particle - hole pair is exchanged with the probability @xmath31 $ ] where @xmath32 ( @xmath33 being the boltzmann constant ) and @xmath34 is the difference in the energy of the new and old configurations while @xmath35 for jumps , respectively , along , transverse to , against the direction of the driving field @xmath36 . throughout this paper
we take @xmath37 and express the temperature @xmath30 in the units of @xmath11 . for the kls model with
_ attractive _ inter - particle interactions ( @xmath16 ) on a square lattice , there is not only an ordered state at all @xmath38 , but the critical temperature @xmath39 increases with @xmath40 , saturating at a value @xmath41 where @xmath42 is the critical temperature of the corresponding ising model in thermodynamic equilibrium @xcite . on the other hand
, @xmath39 decreases with @xmath40 when the inter - particle interactions are _ repulsive _
( i.e. , @xmath17 ) ; the ordering is altogether destroyed by sufficiently large @xmath40 .
however , there is no ordered structure at any non - zero temperature in the one - dimensional kls model , irrespective of the sign of the interaction @xmath11 . because of this intrinsic qualitative difference in the nature of the ordering in the steady states in @xmath43 and @xmath44 we present the corresponding results in separate sections . in the totally asymmetric simple exclusion process
( from now onwards referred to as * tasep * ) @xcite , initially , @xmath45 classical particles occupy _ randomly _ the sites of a one - dimensional lattice of length @xmath20 ( @xmath46 ) . one time step of the dynamics consists of updating the position of @xmath45 particles picked up in a random - sequential manner ; each randomly chosen particle moves forward , with probability @xmath47 , if the lattice site immediately in front of it is empty . for this model , the simple single - site ( i.e. , @xmath2-cluster ) mft @xcite gives the _ exact _ flux @xmath48 for all densities @xmath25 of the particles . in the special case
@xmath49 the kls model reduces to the corresponding standard ising model in thermodynamic equilibrium .
note that in @xmath43 , in the opposite limit @xmath50 the hopping against the driving field becomes impossible and , moreover , the field @xmath40 dominates so overwhelmingly over @xmath51 at all non - zero temperatures @xmath30 that , in this limit , the one - dimensional kls model reduces to the tasep , with @xmath52 , irrespective of the sign of the interaction @xmath11 , provided @xmath11 remains _
finite_. however , in @xmath53 , in the limit @xmath50 the hopping probabilities in the directions transverse to the driving field remain finite and the ratio of the hopping probabilities in the direction of the field and those in the transverse direction diverges @xcite .
for all the non - vanishing finite @xmath40 , in the limit @xmath54 , the one - dimensional kls model reduces to the tasep with the hopping probability @xmath55 .
finally , at infinitely high temperatures each particle moves completely randomly , independent of each other , with equal probability in all directions .
in this section we briefly outline the methods of our analytical as well as numerical calculations .
the dynamical cluster mft has been used successfully in the analytical treatments of several non - equilibrium models including , for example , surface - reaction models @xcite and particle - hopping models of vehicular traffic @xcite . however , in all the traffic models there is no inter - particle interaction except , of course , the hard - core repulsion . moreover , unlike the traffic models , the particles in the kls model can also move against the drive . in this paper
we extend the approach in appropriate manner to calculate the flux @xmath56 as a function of @xmath25 in the kls model for arbitrary @xmath57 , @xmath58 and @xmath59 .
we define a @xmath0-cluster ( @xmath60 ) to be a collection of @xmath0 sites each of which is the nearest - neighbor of at least another site belonging to the same cluster . for simplicity of notation ,
let us consider @xmath43 .
we denote the probability of finding an @xmath0-cluster in the state @xmath61 at time @xmath18 by the symbol @xmath62 .
we treat an @xmath0-cluster exactly and approximate all the @xmath63-cluster probabilities by a product of @xmath0-cluster probabilities in a manner so as to couple the @xmath0-cluster to the rest of the system self - consistently ( see , for example , @xcite for a pedagogical introduction and the existing literature ) .
it is straightforward to see that the state of the @xmath3-cluster @xmath64 at time @xmath65 depends on the state of the @xmath4-cluster @xmath66 at time @xmath18 so that the exact master equation @xmath67 \nonumber \\ & = & \sum_{\tau_{i-1},\tau_{i+2 } } \biggl[p_4(\tau_{i-1},c_{i+1},c_i,\tau_{i+2 } ) w(\tau_{i-1},c_{i+1},c_i,\tau_{i+2 } ) \nonumber \\ & -&p_4(\tau_{i-1},c_i , c_{i+1},\tau_{i+2 } ) w(\tau_{i-1},c_i , c_{i+1},\tau_{i+2})\biggr ] \nonumber \\ & + & \sum_{\tau_{i-1},\tau_{i-2 } } \biggl[p_4(\tau_{i-2},c_i,\tau_{i-1},c_{i+1 } ) w(\tau_{i-2},c_i,\tau_{i-1},c_{i+1 } ) \nonumber \\ & - & p_4(\tau_{i-2},\tau_{i-1},c_i , c_{i+1 } ) w(\tau_{i-2},\tau_{i-1},c_i , c_{i+1})\biggr ] \nonumber \\ & + & \sum_{\tau_{i+2},\tau_{i+3 } } \biggl[p_4(c_i,\tau_{i+2},c_{i+1},\tau_{i+3 } ) w(c_i,\tau_{i+2},c_{i+1},\tau_{i+3 } ) \nonumber \\ & -&p_4(c_i , c_{i+1},\tau_{i+2},\tau_{i+3 } ) w(c_i , c_{i+1},\tau_{i+2},\tau_{i+3})\biggr ] \label{eq-4}\end{aligned}\ ] ] governing the time evolution of the @xmath3-cluster probabilities @xmath68 involves the @xmath4-cluster probabilities @xmath69 for all those configurations which can lead to the @xmath3-cluster configuration @xmath70 under consideration ; @xmath71}\biggr ) \label{eq-4a}\end{aligned}\ ] ] are the corresponding transition probabilities
. however , the the master equation governing the time evolution of the @xmath4-cluster probabilities @xmath69 involve @xmath72-cluster probabilities , and so on . a few concrete examples of such exact master equations for @xmath0-cluster probabilities are given in the appendix a. in the spirit of the cluster - mean - field approach , we truncate this hierarchy of exact master equations by expressing , albeit approximately , all the @xmath63-cluster probabilities in terms of the @xmath0-cluster probabilities . according to the definition of dh , the probability for a dh of @xmath73 is given by @xmath74 we evaluate the right hand side of the equation ( [ eq - dh1 ] ) in the @xmath3-cluster and @xmath4-cluster approximations . in our computer simulations , we begin with initial conditions where the particles are arranged randomly on a lattice of linear size @xmath20 ( i.e. , the system is a linear chain of size @xmath20 in @xmath43 and square lattice of size @xmath29 in @xmath53 ) .
the system is then allowed to evolve , following the kawasaki exchange algorithm with the metropolis probabilities min@xmath75 $ ] mentioned above , for time steps of the order of @xmath76 , so as to allow it to reach its steady - state and , then , for further @xmath77 steps during which the time - averaged flux is measured . finally , for the same set of values of the parameters , this process is repeated for @xmath78 different initial configurations to compute the configuration - averaged flux .
since we did not observe any significant difference between the results for @xmath79 and @xmath80 in @xmath43 and between those for @xmath81 and @xmath82 in @xmath44 all the data presented in this paper were generated using @xmath79 in @xmath43 and @xmath83 in @xmath44 .
it has been realized for some time @xcite that the smallest cluster one must consider in a dynamical cluster mft depends on the nature of the dynamics .
since the kawasaki dynamics conserves the number of particles , in principle , the smallest cluster must consist of at least one _ pair _ of sites .
however , it is also known that for the @xmath50 limit of the kls model , which is exactly identical with the tasep ( with @xmath52 ) , the single - site mft gives the exact result .
therefore , in this short section we not only establish explicitly the limitations of the @xmath2-cluster mft at weak @xmath40 but also demonstrate how the accuracy of the @xmath2-cluster mft increases with increasing @xmath40 in the kls model in @xmath43 .
the net flux is obtained from @xmath84 where the forward flux ( i.e. , flux in the direction of @xmath36 ) is given by @xmath85 while the reverse flux ( i.e. , the flux against @xmath36 ) is given by @xmath86 in the @xmath2-cluster approximation the @xmath4-cluster probabilities @xmath87 are approximated by the products of corresponding @xmath2-cluster probabilities .
therefore , utilizing the facts that @xmath88 and @xmath89 , in the @xmath2-cluster approximation , the equations ( [ eq-5 ] ) and ( [ eq-6 ] ) give , respectively , @xmath90 and @xmath91 for the forward and reverse flux .
hence , the net flux is @xmath92 - \min[1,e^{-\beta e } ] ) \nonumber \\ & + & c(1-c)\biggl(\min[1,e^{\beta(e-4j ) } ] - \min[1,e^{\beta(-e+4j ) } ] \nonumber \\ & - & \min[1,e^{\beta(-e-4j ) } ] + \min[1,e^{\beta(e+4j)}]\biggr)\biggr ] .
\label{eq-9}\end{aligned}\ ] ] the expression ( [ eq-9 ] ) predicts that the net flux @xmath56 is independent of the sign of the interaction @xmath11 , at all @xmath25 and @xmath30 , irrespective of the strength of @xmath40 ; this is certainly not true in general , except at very large @xmath40 ( fig.1 ) .
a comparison between the theoretical prediction ( [ eq-9 ] ) and the results of our computer simulations ( fig.1 ) exposes not only the quantitative inaccuracy of the @xmath2-cluster mft but also its failure to account for the qualitative features of @xmath93 in the case of repulsive inter - particle interactions . with the density @xmath25 of the particles in the _ steady - state _ of the one - dimensional kls model with ( a ) attractive ( i.e. , ferromagnetic ) interaction @xmath94 and ( b ) repulsive ( i.e. , anti - ferromagnetic ) interaction @xmath95 , both at @xmath96 . in both ( a ) and ( b )
the discrete data points have been obtained from our computer simulations with @xmath97 and @xmath98 , respectively .
the lines represent the predictions of the @xmath2-cluster approximation for @xmath99 ( solid line ) , @xmath100 ( dotted line ) , @xmath101 ( dashed line ) , @xmath102 ( long dashed ) and @xmath103 ( dot - dashed ) , respectively . ] the fact that the @xmath2-cluster mft works better for stronger @xmath40 is not surprising as the one - dimensional kls model reduces to the tasep ( with @xmath52 ) in the limit @xmath50 and it is well known that @xmath2-cluster mft gives exact result for the tasep .
thus , fig.[fig-1 ] establishes the existence of strong correlations which are neglected by the @xmath2-cluster mft .
in the @xmath3-cluster approximation , we approximate the @xmath4-cluster probabilities , appearing in the exact master equation for the 2-cluster probabilities , by a product of @xmath3-cluster probabilities , i.e. , @xmath104 or , more precisely , @xmath105 next , we parametrize the @xmath3-cluster probabilities as follows : + @xmath106 @xmath107 where @xmath108 this parametrization enables us to satisfy the three constraints @xcite namely , @xmath109 , @xmath110 , and @xmath111 .
equivalently , these constraints also imply that @xmath112 .
so , only one of the four equations represented by equation ( [ eq-4 ] ) can be taken as an independent equation .
thus , the calculation of the four @xmath3-cluster probabilities boils down to the calculation of the single parameter @xmath113 . using _ mathematica _ for an automated generation of the equations and simplification
, we find that , in the steady state , @xmath113 satisfies the quadratic equation @xmath114 where @xmath115 + \min[1,e^{\beta(e-4j ) } ] \nonumber \\ & - & \min[1,e^{\beta(-e+4j ) } ] - \min[1,e^{\beta(e+4j ) } ] , \label{eq-16}\end{aligned}\ ] ] @xmath116 + \min[1,e^{\beta(e-4j)}]\biggr ) , \label{eq-17}\ ] ] @xmath117 + \min[1,e^{\beta(e-4j)}]\ } , \label{eq-18}\ ] ] and , hence , taking the physical solution that allows the @xmath3-cluster probabilities to be between @xmath118 and @xmath2 , we get @xmath119 the corresponding form of @xmath120 for an alternative choice of @xmath121 was derived by szabo et al.@xcite . both the forms ( [ eq-19 ] ) and the equation ( 13 ) in reference @xcite are special cases of the general form @xcite @xmath122 where @xmath123 ; \label{eq-19b}\ ] ] @xmath124 being the hopping probabilities against and along the field @xmath40 , respectively .
finally , in the @xmath3-cluster mft , the forward flux and the reverse flux are given by @xmath125 and @xmath126 respectively ; the net flux is obtained from @xmath84 . in the next few subsections
we compare the predictions of this @xmath3-cluster mft with the corresponding numerical data obtained from our extensive computer simulations of the kls model .
it is straightforward to verify that at an infinitely high temperature the expressions ( [ eq-20 ] ) and ( [ eq-21 ] ) for @xmath128 and @xmath129 become identical and , therefore , the net flux vanishes ; hign temperature not only washes away the effects of @xmath11 , as it is known to do even in equilibrium , but also the effects of @xmath40 making particle movement in both directions equally probable .
in fig.2 we plot the flux as a function of the particle density @xmath25 , at a fixed non - zero finite temperature @xmath30 , for five different values of @xmath40 ; the agreement between the theoretical prediction and computer simulation is very good for both attractive ( fig.2a ) as well as repulsive ( fig.2b ) interactions .
the @xmath93 curves are the analogues of the fundamental relations for the traffic models .
the shape of the curve @xmath93 in the kls model with _ attractive _ ( @xmath16 ) is qualitatively similar to those observed in the particle - hopping models of vehicular traffic . in sharp contrast
, we find a qualitatively different shape of the curve @xmath93 in the kls model with _ repulsive _ interactions ( @xmath17 ) ; there is a minimum , rather than maximum , at @xmath130 provided the strength of @xmath11 is comparable to that of @xmath40 .
moreover , for the same @xmath30 and @xmath40 , the flux is higher in the case of _ repulsive _ inter - particle interactions than that in the case of _ attractive _ inter - particle interactions . nevertheless , because of the particle - hole symmetry , the curves in both the figures [ fig-2](a ) and ( b ) are symmetric about @xmath130 irrespective of the sign of the inter - particle interactions . with the density @xmath25 of the particles in the _ steady - state _ of the one - dimensional kls model with ( a ) attractive interaction @xmath94 and ( b ) repulsive interaction @xmath95 , both at @xmath96
( a ) and ( b ) the discrete data points , have been obtained from our computer simulations with @xmath97 and @xmath98 , respectively .
the lines represent the corresponding predictions of the @xmath3-cluster approximation . ]
also note that in both the figures [ fig-2](a ) and ( b ) @xmath131 as @xmath50 . this is a consequence of the fact that , as stated before , the kls model reduces to the asep , with @xmath52 , in the limit @xmath50 ( in both the cases of attractive and repulsive inter - particle interactions ) as long as @xmath11 and @xmath30 remain finite .
in fact , in all the figures [ fig-1](a ) , ( b ) and [ fig-2](a ) , ( b ) , the full curve @xmath132 is given by exact expression @xmath133 .
thus , unlike the @xmath2-cluster mft , the @xmath3-cluster mft reproduces the qualitative features of the @xmath93 curves for all @xmath40 and for both attractive as well as repulsive interactions .
moreover , the flux predicted by the @xmath3-cluster mft is also in good quantitative agreement with the corresponding computer simulation data , except for a narrow range of @xmath25 about @xmath134 .
this indicates that the predictions of @xmath3-cluster mft , although quite accurate for arbitrary @xmath40 , is _ not exact _ ( except , of course , @xmath50 ) for the kls model in @xmath43 .
we shall improve our theory further by developing a @xmath4-cluster mft in the next section . in order to get a deeper insight into the dependence of the flux @xmath135 on @xmath136 as well as on the sign of @xmath11
we analyze the results of the @xmath3-cluster mft in detail in a few special limits in the following subsections .
let us investigate the dependence of the flux @xmath137 on the driving field @xmath40 at @xmath138 for arbitrary values of @xmath25 . for _ repulsive _ inter - particle interactions , the @xmath3-cluster expressions ( [ eq-16])-([eq-18 ] ) for @xmath139 and @xmath140 reduce to the simple forms @xmath141 , @xmath142 , @xmath143 for all @xmath144 , @xmath145 , @xmath142 , @xmath143 at @xmath146 and to the forms @xmath147 , @xmath148 , @xmath149 for all @xmath150 .
substituting the zero - temperature values of @xmath139 and @xmath140 into the equation ( [ eq-19 ] ) , we find that , at @xmath138 , for all @xmath144 @xmath151 for @xmath152 , and @xmath153 for @xmath154 ; physically , this means that when less than half of the sites are occupied by particles ( holes ) , both the nearest - neighbors of each particle ( hole ) are _ certainly _ holes ( particles ) because of the repulsive nature of the inter - particle interaction . on the other hand , at @xmath138 , for all @xmath150 , @xmath155 for any arbitrary @xmath25
; this implies that at @xmath138 the effects of all @xmath150 is equivalent to those of infinitely large @xmath40 at all finite non - zero @xmath30 so that @xmath2-cluster mft becomes exact .
moreover , at @xmath138 , for @xmath146 , @xmath156 for all @xmath25 . substituting the appropriate expression of @xmath113 , derived above for @xmath138 , into the limiting form of the net flux @xmath137 , obtained from the @xmath3-cluster expressions ( [ eq-20 ] ) and ( [ eq-21 ] ) , we get @xmath157 for the kls model with _ repulsive _ inter - particle interactions at @xmath138 , where @xmath158 \nonumber \\
+ \theta(c-0.5 ) ~\biggl[\frac{1-c}{c}(2c-1)\biggr ] , \label{eq-23}\end{aligned}\ ] ] @xmath159 and @xmath160 @xmath161 being the step function , namely , @xmath162 for all @xmath163 and @xmath164 for all @xmath165 .
in the special case of half - filling , the particles remain pinned " to their respective positions by the inter - particle interactions @xmath11 and no @xmath144 is strong enough to cause any de - pinning " .
such switching " of the flux from zero to a non - zero value _
discontinuously _ , in response to the driving field @xmath40 , has been reported earlier for the kls model on a linear chain @xcite as well as on a square lattice @xcite . in sharp contrast to the @xmath25-dependence of @xmath166 in the case of _ repulsive _ inter - particle interactions , we have @xmath167 for all @xmath25 when the inter - particle interaction is _ attractive _ ; therefore , in the latter case , @xmath168 for the kls model with _ attractive _ inter - particle interactions at @xmath138 , where @xmath169 @xmath170 and @xmath171 at @xmath138 , all @xmath150 is equivalent to @xmath50 , irrespective of the sign of @xmath11 and , hence , the corresponding flux is @xmath172 . in this subsection
we consider the dependence of @xmath56 on @xmath30 at a few special values of @xmath25 . as demonstrated in the figures ( [ fig-3])-([fig-5 ] ) , the predictions of the @xmath3-cluster theory agree very well with the corresponding numerical data obtained from our computer simulations .
the non - monotonic variation of the flux with temperature at small and intermediate values of @xmath40 is an interesting phenomenon .
so long as @xmath30 is non - zero but much smaller than @xmath40 , it works against @xmath11 and helps in de - pinning " the particles which can move forward under the influence of the driving field @xmath40 .
however , when @xmath30 becomes much larger than @xmath40 , then it washes out the effect of @xmath40 allowing particles to move against @xmath36 as often as along @xmath36 .
with the temperature @xmath30 in the _ steady - state _ of the one - dimensional kls model with ( a ) attractive interaction @xmath94 and ( b ) repulsive interaction @xmath95 , both at @xmath173 . in both ( a ) and ( b ) the discrete data points , have been obtained from our computer simulations with @xmath97 and @xmath98 , respectively .
the lines represent the corresponding predictions of the @xmath3-cluster approximation . ] ) , except that the density of the particles is @xmath174 . ] in the @xmath2-cluster approximation , the dh distribution is given by @xmath175 however , in the @xmath3-cluster approximation , we write @xmath176 and @xmath177 hence , in the @xmath3-cluster approximation , we get @xmath178 whereas @xmath179^{j-1 } , ~{\rm for } ~j \geq 1 .
\label{eq - dh6}\ ] ] the variations of the dh distribution with @xmath40 ( for fixed @xmath25 ) and with @xmath25 ( for fixed @xmath40 ) , as predicted by ( [ eq - dh5 ] ) and ( [ eq - dh6 ] ) of the @xmath3-cluster approximation , are compared with the corresponding computer simulation data in fig.[dhfig1 ] . and
fig.[dhfig2 ] , respectively .
note that for @xmath173 , in the absence of @xmath40 , the most probable dh is @xmath180 or @xmath181 depending on whether the interaction is attractive ( i.e. , ferromagnetic , in the language of magnetism ) or repulsive ( i.e. , antiferromagnetic ) .
this type of spatial organization of the particles persists even in the presence of @xmath40 as long as @xmath40 is much weaker than the strength of the interaction @xmath11 ( see fig.[dhfig1 ] ) .
however , deviation from this spatial organization increases gradually with increasing strength of @xmath40 . in the limit @xmath50 ,
for all finite @xmath182 , the dh distribution approaches the exact dh distribution of the tasep and , as expected , is independent of the sign of the interaction @xmath11 . for a given @xmath40 ,
which is comparable with the strength @xmath182 of the interaction , increasing @xmath25 leads to more congestion and , therefore , the probability of having a dh = 0 becomes larger for higher @xmath25 ( se fig.[dhfig2 ] ) . and ( b ) repulsive interaction @xmath95 , both at for @xmath130 and @xmath96 . in both ( a ) and ( b )
the discrete data points have been obtained from our computer simulations with @xmath97 and @xmath98 , respectively .
the lines represent the corresponding predictions of the @xmath3-cluster approximation . ] and ( b ) repulsive interaction @xmath95 , both at for @xmath183 and @xmath96 . in both ( a ) and ( b )
the discrete data points have been obtained from our computer simulations with @xmath184 and @xmath185 , respectively .
the lines represent the corresponding predictions of the @xmath3-cluster approximation . ]
the exact master equation for the @xmath4-cluster probabilities ( given in the appendix b ) , as expected on general grounds , involve @xmath72-cluster probabilities . in the @xmath4-cluster approximation
we break up the @xmath72-cluster probabilities in terms of the products of the @xmath4-cluster probabilities using the prescription @xmath186 due to the constraints of conditional probabilities and conservation of total probabilities , at @xmath4-cluster level , we have 7 independent equations and the same number of variables .
the solution is obtained by solving a set of non - linear equations .
we find that although the @xmath4-cluster results are , clearly , an improvement over the @xmath3-cluster results , the difference in the actual numerical values of the flux in the two approximations , for the same set of parameters , is extremely small . the flux obtained in these two approximations for a typical set of values of the parameters
are compared in fig.[fig-5 ] ( the corresponding differences in the dh distributions are too small to be shown in a figure ) .
it is worth pointing out that the @xmath4-cluster approximation is not only a _ quantitative _ improvement over the @xmath3-cluster approximation .
the @xmath4-cluster approximation can account for the forward - backward symmetry breaking , an interesting phenomenon @xcite , which the @xmath3-cluster approximation fails to capture .
it is well known @xcite that , usually , the cluster mft fails to account for the properties of the driven - diffusive lattice gases below the ordering temperature @xmath39 .
therefore , we confine our discussions in this section to temperatures @xmath187 . as in @xmath43 , for the same set of parameters , the flux in the two - dimensional kls model with attractive inter - particle interactions ( fig.[fig-6](a ) ) is lower than that in the same model with repulsive inter - particle interactions ( fig.[fig-6](b ) ) .
-cluster and @xmath4-cluster mfts on the variation of the _ net _ flux @xmath56 with ( a ) the density and ( b ) temperature in the _ steady - state _ of the one - dimensional kls model with repulsive interaction @xmath95 .
the parameters in ( a ) are @xmath188 while those in ( b ) are @xmath189 the dashed and solid lines are the theoretical results obtained in the @xmath3-cluster and @xmath4-cluster approximations , respectively , whereas the discrete data points are the numerical data obtained from our computer simulations . ]
a comparison of the predictions of the cluster mft with the computer simulation data ( fig.[fig-6 ] ) establishes that , in the case of _ attractive _ interactions both the @xmath2-cluster and @xmath3-cluster mft over - estimate the flux , although the prediction of the @xmath3-cluster theory is closer to the computer simulation data . on the other hand , in the case of _
repulsive _ interactions , the @xmath2-cluster mft gives an under - estimate whereas the @xmath3-cluster mft provides an over - estimate of the flux . the level of accuracy of the @xmath3-cluster mft can be estimated from the plots in fig.([fig-7 ] ) ; a close inspection , thus , reveals that the @xmath3-cluster mft does not reproduce the dip in the flux around @xmath130 in the case of repulsive inter - particle interactions . with the density @xmath25 of the particles in the _ steady - state _ of the two - dimensional kls model with ( a ) attractive interaction @xmath94 and ( b ) repulsive interaction @xmath95 , both at @xmath190 . in both ( a ) and ( b )
the discrete data points have been obtained from our computer simulations with @xmath191 and @xmath192 , respectively . in both ( a ) and ( b ) the dotted and dashed lines represent the predictions of the @xmath2-cluster mft for @xmath99 and @xmath102 , respectively , while the long - dashed and solid lines represent the corresponding predictions of the @xmath3-cluster mft . ]
it is well established @xcite that the @xmath2-cluster mft result ( eq.[eq-3 ] ) is the _ exact _ expression for the flux in tasep . if the random - sequential updating of the tasep is replaced by the parallel updating it becomes identical with the nagel - schreckenberg ( ns ) model @xcite of vehicular traffic with @xmath193 , where @xmath194 is the largest ( integer ) speed allowed for each of the vehicles .
-cluster and @xmath3-cluster mfts on the _ net _ flux @xmath56 in the _ steady - state _ of the two - dimensional kls model with ( a ) attractive interaction @xmath94 and ( b ) repulsive interaction @xmath94 , all for @xmath99 . in both ( a ) and ( b )
the discrete data points have been obtained from our computer simulations at @xmath195 and @xmath196 , respectively . in both ( a ) and ( b ) the dotted and solid lines represent the predictions of the @xmath3-cluster mft at @xmath197 and @xmath190 , respectively . ] in the case of the ns model the @xmath2-cluster mft makes an under - estimate of the flux but the @xmath3-cluster mft treatment is adequate to take into account the correlations introduced purely by the parallel dynamics and , therefore , it gives the exact result @xcite .
the kls model can be regarded as an extension of the tasep by incorporating non - vanishing inter - particles interactions through non - zero @xmath11 .
our results reported in this paper show that neither the @xmath2-cluster mft nor the @xmath3-cluster mft yield exact flux in the kls model ; although the @xmath3-cluster results are accurate to order @xmath198 it is the @xmath4-cluster results that are practically indistinguishable from the corresponding computer simulation data .
although it is possible that the results of the @xmath4-cluster mft might be exact for the kls model we refrain from making such a claim as we do not have any rigorous proof .
note that in the special case @xmath199 we have @xmath200 , @xmath142 and @xmath201 where @xmath202 . in this case , the quadratic equation ( [ eq-16 ] ) reduces to the form @xmath203 which was derived @xcite directly from the @xmath3-cluster mf treatment of the ns model with @xmath193 .
interestingly , the @xmath3-cluster result is not exact for the kls model , but it gives exact result for the ns model .
this is a consequence of the fact that the @xmath3-cluster mft gives , in general exact results for the one - dimensional ising model in equilibrium . since @xmath204 , the relation ( [ eq-28 ] ) maps the ns model , with @xmath193 , onto an _
antiferro_-magnetic ising model , ( or , equivalently , to the kls model , with _ repulsive _ inter - particle interactions , in the absence of external drive ) so that the steady - state of the former is identical to the equilibrium state of the latter .
consequently , for all densities @xmath25 , we observe perfect agreement of the dh distributions in the ns model and that in the kls model with @xmath49 , @xmath205 ( fig.[figns ] ) . , for @xmath130 at three different values of @xmath206 corresponding , respectively , to @xmath207 ( @xmath208 ) , @xmath209 ( @xmath210 ) and @xmath211 ( @xmath212 ) , all for @xmath213 .
the lines are merely guides to the eye connecting the points corresponding to the dh distributions in the ns model with @xmath193 and @xmath207 ( solid line ) , @xmath209 ( dotted line ) and @xmath211 ( dashed line ) , respectively .
all the results in this figure have been obtained in the @xmath3-cluster approximation . ] since the relation ( [ eq-28 ] ) maps the ns model ( with @xmath193 ) onto the kls model only for @xmath49 , this mapping can not relate the properties of the kls model for any non - zero @xmath40 with those of the ns model .
interestingly , in the ns model with @xmath193 the flux is _ maximum _ at @xmath130 for all @xmath47 . in sharp contrast
, the flux is _ minimum _ at @xmath130 in the kls model with repulsive ( anti - ferromagnetic ) interactions so long as @xmath40 is not much stronger than @xmath182 ; as @xmath40 increases the depth of the well at @xmath130 in fig.[fig-2 ] decreases and , eventually , for sufficiently large @xmath40 the flux exhibits its maximum at @xmath130 .
moreover , the location of the maximum in the dh distribution depends crucially on the sign of the interaction @xmath11 provided @xmath40 is not much larger than @xmath182 . the time - interval between the arrivals ( or departures ) of the successive particles at a detector site
is defined to be the corresponding time - headway * th*. the exact th distribution for the ns model with @xmath193 has been calculated @xcite . however , because of the possibility of hopping of the particles against @xmath40 in the kls model , the analytical calculation of th distribution is extremely difficult and will not be reported here .
in this paper we have reported the results of cluster - mean - field theoretic treatments of a driven - diffusive lattice gas model to calculate the flux of the particles under the influence of the driving field .
although we have considered the standard model , namely , the katz - lebowitz - spohn model in this paper , our technique is sufficiently general that it can be used , in principle , to calculate the flow properties of any other driven - diffusive lattice gas .
we have shown interesting novel features of the flux - density relation of the kls model with _ repulsive _ inter - particle interactions .
our investigation has helped in elucidating the roles of inter - particle interactions @xmath11 , temperature @xmath30 and the driving field @xmath40 in determining the trend of variation of the flux with the density of the particles in the driven - diffusive lattice gas models .
one of us ( dc ) would like to thank the department of computational science of the national university of singapore for hospitality during a short visit where this paper was completed .
we also thank a. schadschneider and a. szolnoki for useful correspondence .
concrete examples of exact master equations for @xmath0-cluster probabilities in the one - dimensional kls model : + @xmath214 p_4(0,0,1,1 ) \nonumber \\ & + & \min[1,e^{\beta(e+4j ) } ] p_4(0,1,0,1 ) \nonumber \\ & + & \min[1,e^{\beta(-e+4j ) } ] p_4(1,0,1,0 ) \nonumber \\ & - & \min[1,e^{\beta(e-4j ) } ] p_4(1,1,0,0 ) , \label{eq-29 } \end{aligned}\ ] ] @xmath215 p_5(0,0,0,1,0 ) \nonumber \\ & -&\min[1,e^{-\beta(e+4j ) } ] p_5(0,0,0,1,1 ) \nonumber \\ & + & \min[1,e^{-\beta e } ] p_5(0,0,1,0,0 ) \nonumber \\ & + & \min[1,e^{\beta e } ] p_5(0,0,1,0,0 ) \nonumber \\ & + & \min[1,e^{\beta(e+4j ) } ] p_5(0,0,1,0,1 ) \nonumber \\ & -&\min[1,e^{\beta e } ] p_5(0,1,0,0,0 ) \nonumber \\ & + & \min[1,e^{\beta(-e+4j ) } ] p_5(1,0,1,0,0 ) \nonumber \\ & - & \min[1,e^{\beta(e-4j ) } ] p_5(1,1,0,0,0 ) .
\label{eq-30 } \end{aligned}\ ] ]
exact master equations for the @xmath4-cluster probabilities in the one - dimensional kls model : + @xmath216 \nonumber \\ & = & \sum_{\tau_{i-2},\tau_{i+3 } } \biggl[p_6(\tau_{i-2},c_{i-1},c_{i+1},c_i , c_{i+2},\tau_{i+3 } ) \nonumber \\ & \times & w(c_{i-1},c_{i+1},c_i , c_{i+2 } ) \nonumber \\ & - & p_6(\tau_{i-2},c_{i-1},c_i , c_{i+1},c_{i+2},\tau_{i+3 } ) \nonumber \\ & \times & w(c_{i-1},c_i , c_{i+1},c_{i+2})\biggr ] \nonumber \\ & + & \sum_{\tau_{i-2},\tau_{i+3 } } \biggl[p_6(\tau_{i-2},c_{i-1},c_i , c_{i+2},c_{i+1},\tau_{i+3 } ) \nonumber \\ & \times & w(c_{i-1},c_i , c_{i+2},c_{i+1 } ) \nonumber \\ & - & p_6(\tau_{i-2},c_{i-1},c_i , c_{i+1},c_{i+2},\tau_{i+3 } ) \nonumber \\ & \times & w(c_{i-1},c_i , c_{i+1},c_{i+2})\biggr ] \nonumber \\ & + & \sum_{\tau_{i-2},\tau_{i+3 } } \biggl[p_6(\tau_{i-2},c_i , c_{i-1},c_{i+1},c_{i+2},\tau_{i+3 } ) \nonumber \\ & \times & w(c_i , c_{i-1},c_{i+1},c_{i+2 } ) \nonumber \\ & - & p_6(\tau_{i-2},c_{i-1},c_i , c_{i+1},c_{i+2},\tau_{i+3 } ) \nonumber \\ & \times & w(c_{i-1},c_i , c_{i+1},c_{i+2})\biggr ] \nonumber \\ & + & \sum_{\tau_{i+3},\tau_{i+4 } } \biggl[p_6(c_{i-1},c_i , c_{i+1},\tau_{i+3},c_{i+2},\tau_{i+4 } ) \nonumber \\ & \times & w(c_{i+1},\tau_{i+3},c_{i+2},\tau_{i+4 } ) \nonumber \\ & - & p_6(c_{i-1},c_i , c_{i+1},c_{i+2},\tau_{i+3},\tau_{i+4 } ) \nonumber \\ & \times & w(c_{i+1},c_{i+2},\tau_{i+3},\tau_{i+4})\biggr ] \nonumber \\ & + & \sum_{\tau_{i-2},\tau_{i-3 } } \biggl[p_6(\tau_{i-3},c_{i-1},\tau_{i-2},c_i , c_{i+1},c_{i+2 } ) \nonumber \\ & \times & w(\tau_{i-3},c_{i-1},\tau_{i-2},c_i)\nonumber \\ & - & p_6(\tau_{i-3},\tau_{i-2},c_{i-1},c_i , c_{i+1},c_{i+2 } ) \nonumber \\ & \times & w(\tau_{i-3},\tau_{i-2},c_{i-1},c_i)\biggr ] .
\label{eq-31}\end{aligned}\ ] ] b. schmittmann and r.k.p .
zia , in : _ phase transitions and critical phenomena _ , vol.17 ,
eds . c. domb and j.l .
lebowitz ( academic press , 1995 ) ; phys .
rep . * 301 * , 45 ( 1998 ) ; r.k.p .
zia , l.b .
shaw , b. schmittmann and r.j .
astalos , cond - mat/9906376 .
wolf , m. schreckenberg and a. bachem ( eds . ) , _ traffic and granular flow _
( world scientific , singapore , 1996 ) ; m. schreckenberg and d.e . wolf ( eds . ) , _ traffic and granular flow 97 _ ( springer , singapore , 1998 ) ; d. helbing , h.j .
herrmann , m. schreckenberg and d.e .
wolf ( eds . ) _ traffic and granular flow 99 _ ( springer , berlin , 2000 ) .
m. schreckenberg , a. schadschneider , k. nagel and n. ito , phys .
e * 51 * , 2939 ( 1995 ) ; a. schadschneider and m. schreckenberg , j. phys .
a * 26 * , l679 ( 1993 ) ; a. schadschneider , eur .
j. b * 10 * , 573 ( 1999 ) . | we develop @xmath0-cluster mean - field theories ( @xmath1 ) for calculating the _ flux _ and the _ gap distribution _ in the non - equilibrium steady - states of the katz - lebowitz - spohn model of the driven diffusive lattice gas , with attractive and repulsive inter - particle interactions , in both one and two dimensions for arbitrary particle densities , temperature as well as the driving field .
we compare our theoretical results with the corresponding numerical data we have obtained from the computer simulations to demonstrate the level of accuracy of our theoretical predictions .
we also compare our results with those for some other prototype models , notably particle - hopping models of vehicular traffic , to demonstrate the novel qualitative features we have observed in the katz - lebowitz - spohn model , emphasizing , in particular , the consequences of _ repulsive _ inter - particle interactions . |
in this paper , we focus on the border between physics and number theory .
it can be regarded as a first step to investigate , in particular , whether quantum field theory can shed any light on some long - standing , yet unproved , mathematical problems in number theory .
indeed it seems reasonable that the `` translation '' of certain difficult mathematical questions into the physical language might provide essential clues to better understand ( or maybe help to solve ) them .
for instance , it is well - known that the hamiltonian describing quantum - mechanically a one - dimensional harmonic oscillator has an equally spaced spectrum ( in the appropriate units and after a shift corresponding to the zero point energy ) given by the set of positive integers .
this remarkable property naturally leads to the following question : is it possible to use spectral properties of this or other operators to study and analyze problems of number theory , e.g. the distribution of primes or the goldbach conjecture ? at the beginning of the 20th century , plya and hilbert independently suggested the possibility of proving the riemann hypothesis on the distribution of non - trivial zeroes of the riemann function @xmath2 by finding a hermitian operator with eigenvalues that could be put in correspondence with those of @xmath2 @xcite .
moreover , a remarkable connection between number theory and nuclear physics emerged ( see @xcite for a review ) as a consequence of a fortuitous meeting at princeton in the early s 1970s between the mathematician hugh montgomery and the physicist freeman dyson .
it was then noticed that random matrix theory ( introduced in the physical sciences by eugene wigner in the fifties @xcite ) could model apparently unrelated problems of both fields .
the level spacing of heavy nuclei and the non - trivial zeroes of the riemann zeta function showed a striking statistical similarity .
this example is not the only one to suggest a link between physics and prime numbers . in 1990
bernard julia put forward the idea of a fictitious , non - interacting , boson gas @xcite whose constituents were called _
primons_. the partition function of the primon gas was found to be the riemann - zeta function @xmath2 .
the idea of the riemann gas can also be extended to fermions leading to a relatively simple partition function of the form @xmath3 .
furthermore , it has been argued that more intriguing relations exist between the riemann gas , quark confinement and even string theory @xcite .
this is a clear example of how studies initially started as a rather unorthodox approach to the riemann hypothesis can lead , in turn , to new and eventually fertile perspectives in physics and vice versa .
a thorough summary of several suggestive relationships between different branches of physics and the riemann hypothesis can be found in @xcite . in this paper
we develop a ( quite simple ) physical model involving a free fermionic field defined on a cylindrical space - time manifold , with prime excitation modes only . no relation with p - adic theories in physics
@xcite has been envisaged in this work .
rather , one of our goals is to introduce prime numbers within the standard language of quantum field theory ( qft ) and investigate its primary consequences . in particular
, we will cast goldbach s conjecture into the standard language of quantum field theory ( qft ) .
interestingly , as will be shown , the consistency analysis for the underlying qft ultimately involves the riemann hypothesis in the computation of the renormalized vacuum energy . in june 1742
the mathematician christian goldbach wrote a letter to his friend , leonhard euler , where he made his famous conjecture about prime numbers which can be split into the following two statements : every even integer greater than 2 can be expressed as the sum of two primes every odd integer greater than 5 can be expressed as the sum of three primes the first statement is known as the _ binary _ or _ strong _ goldbach s conjecture ( gc ) .
the second is known as the _ ternary _ or _ weak _ gc .
note that if the strong gc is true then the weak gc is also true .
if not otherwise stated , we will denote the strong gc as the gc for short .
the weak gc was proved in 1923 by hardy and littlewood , for sufficiently large odd numbers as an asymptotic " result , assuming the truth of the riemann hypothesis . in 1937
vinogradov again proved the weak conjecture for a sufficiently large ( but indeterminate ) odd number using analytic methods . in 1966
chen jing - run proved that every sufficiently large even number can be expressed as the sum of a prime and a number with no more than two prime factors . at the present moment
, the gc has been checked up to integers @xmath4 @xcite by using computers . in 1849 alphonse
de polignac ( 1849 ) made the general conjecture that there are infinitely many cases of two consecutive prime numbers such that their difference is any even number @xmath5 .
for @xmath6 it reduces to the twin prime conjecture . for @xmath7 there should be infinitely many cousin primes ... based on heuristic arguments , hardy and littlewood found in 1922 a law to estimate the density of twin primes @xcite , also discussed by clement in 1949 @xcite
. the following weakened form of polignac s conjecture ( pc ) can be enunciated by relaxing the condition on the necessity that both primes are consecutive : any even number can be expressed as the difference of two primes in infinitely many ways .
polignac s weakened conjecture can be seen as complementary to goldbach s conjecture .
both polignac s and goldbach s conjectures have never been proved . at first glance the riemann hypothesis ( likewise the two preceding conjectures )
does not seem very complicated to state , but in fact it lies deep in the roots of number theory , in connection with the mysteries of the distribution of the prime numbers : all non - trivial zeros of the zeta function have real part equal to one - half .
the prime number theorem is equivalent to the assertion that no zeros of @xmath2 on the complex plane ( @xmath8 ) lie on the boundary of the critical strip ( @xmath9 ) .
understanding how the non - trivial zeros are located relative to the critical line ( @xmath10 ) , remains today one of the deepest unsolved questions in mathematics .
moreover , note that many developments in number theory are contingent upon the validity of the riemann hypothesis ( rh ) . in this paper
we explore the connection between the rh and the gc and pc on the light of a quantum field theoretical model to be developed below .
our starting point is the lagrangian density of a classical free real scalar field @xmath11 in a two - dimensional spacetime : @xmath12 where @xmath13 denotes the space coordinate , leading to the well - known massless klein - gordon equation : [ kg ] _
^ ( t , x ) ( t , x)=0 the inverse metric tensor @xmath14 has been defined such that @xmath15 , @xmath16 , all other components vanishing . in addition we will set @xmath17 throughout the paper . since we have assumed a zero mass the theory inherits the two - dimensional conformal symmetry @xcite , given by the spacetime transformations @xmath18 with @xmath19 .
as we ultimately want to deal with integer numbers we will let the spatial dimension @xmath20curling up into a circle , becoming cyclic thereby introducing a discrete spectrum in the theory . in such @xmath21 dimensional space , the scalar field @xmath11 has to be periodic , satisfying the boundary condition : [ period ] ( t , x)=(t , x+2r ) let us now expand this field as [ fourier ] ( t , x)= _ n=-^_n^(t)e^inx / r where @xmath22 are complex fourier coefficients in the above expansion , corresponding to a given radius @xmath23 .
the coefficient in front of the sum is introduced for normalization purposes .
using the equation of motion ( [ kg ] ) one can fix immediately the ( normalized ) positive frequency solutions _ n^(t)=
e^-i_nt , where @xmath24 ( for simplicity we neglect the zero - mode solutions ) .
the modes labeled by positive values of @xmath25 represent complex waves that move from left to right , while negative values represent left - moving modes . obviously ,
owing to the linear character of the klein - gordon equation ( [ kg ] ) , the @xmath26 field can be split into left and right moving ( chiral ) fields , @xmath27 and @xmath28 , respectively , both obeying the periodic constraint ( [ period ] ) , as : [ split+- ] ( t , x)=_-(t , x)+_+(t , x ) in such a way that the former ( latter ) only contains ( positive frequency ) fourier components labeled by positive ( @xmath29 ) and negative ( @xmath30 ) integer numbers , respectively . at the quantum level
the fourier expansion is re - expressed in terms of creation and annihilation operators [ positive ] _ -(t - x ) & = & _
n>0 + _ + ( t+x ) & = & _
n<0 , [ negative ] where @xmath31=\delta_{nm}$ ] , @xmath32= 0 = [ a^{-\dagger}_n , a^{-\dagger}_m]$ ] , and also @xmath33=\delta_{nm}$ ] , @xmath34= 0 = [ a^{+\dagger}_n , a^{+\dagger}_m]$ ] .
the vacuum state @xmath35 is defined by the relations @xmath36 .
we go further and generalize the above splitting in terms of even and odd integer numbers [ split ] ( t , x)=^(even)(t , x)+^(odd)(t , x ) with [ evenodd ] ^(even)(t , x ) & = & _ n_even + ^(odd)(t , x ) & = & _ n_odd , [ odd ] with analogous commutation relations for the creation and annihilation operators . from the point of view of cft
the decomposition ( [ split ] ) is somewhat bizarre , since it partially breaks conformal invariance .
only the virasoro generators @xmath37 , with @xmath25 being an even integer , do preserve the above decomposition .
however , in our considerations conformal symmetry will play a rather heuristic role . now , with the goal of restating the gc and the pc in a field - theoretic language
, we write the even ( positive ) numbers as either the sum or the difference of two ( non necessarily consecutive ) primes @xmath1 and @xmath38 : [ goldbach2 ] n_even & = & p+q + n_even & = & |p - q| from eq.([evenodd ] ) and taking into account the elementary property of the exponential function , one could make the tentative hypothesis that the chiral field @xmath39 _ -^(even)(t , x ) = _ n_even>0 can be reconstructed as a composite field according to @xmath40 where the chiral field @xmath41 is expressed by definition as a fourier expansion over prime modes exclusively .
this ansatz does not fit smoothly with the underlying conformal symmetry .
neither @xmath42 and @xmath43 have well - defined properties under conformal transformations .
however , this proposal can be improved on the basis of the classical formulae on the abelian bosonization in two - dimensions @xcite .
it is well - known that the light - cone derivatives of the scalar field @xmath44 are conformal fields which can be reconstructed in terms of fermion bilinears , as we review next ( see also @xcite ) . to describe massless fermions in two dimensions
it is useful to use a real representation of @xmath45-matrices ^0= ( cc 0 & 1 + 1 & 0 ) ^1= ( cc 0 & -1 + 1 & 0 ) ^5=^0 ^ 1= ( cc 1 & 0 + 0 & -1 ) and write a dirac field using two real components =( c _ - + _
the equation of motion @xmath46 implies immediately that @xmath47 and @xmath48 .
moreover , the theory turns out to be invariant under conformal transformations @xmath18 , where the chiral fermi fields transform as follows _
( y^)= ( ) ^1/2 _ ( y^(x^ ) ) . this means that the conformal algebra for the dirac field has central charge @xmath49 .
it is possible to reconstruct a scalar field @xmath50 , which has central charge @xmath51 , from two real dirac fields @xmath52 , with @xmath53 .
the currents @xmath54 acquire a simple form in terms of derivatives of @xmath50 : _ ^1_^2:= _ , where one introduces normal ordering to avoid divergences .
we will also assume for simplicity that the fermionic fields satisfy the same periodic conditions of ( [ period ] ) , i.e. @xmath55 .
expanding the fermion fields in modes ( for simplicity we restrict to the right - moving sector ) [ positive2 ] ^j_-(t - x ) = _ nb^j_n e^-in(t - x)/r , where @xmath56 , one recovers the scalar annihilation and creation operators as bilinears involving all fermion operators @xmath57 ( we use the usual definitions @xmath58 and @xmath59 ) [ eq : a ] a^-_n= sgn(n)_r i : b^1_rb^2_n - r : , with the commutation relations @xmath60= \delta_{n , m}$ ] , with @xmath29 , or , equivalently [ a_n^- , a_-m^-]= _
n , m .
the trivial fact that any even positive number can be written as a sum of two odd numbers suggests the possibility of extending the reduction of an even boson field in terms of two odd fermion fields .
in fact , one can easily obtain a relation ( and a similar one for the opposite chirality ) equivalent to eq.([eq : a ] ) this time for entire @xmath25 such that : a^-_n= sgn(n)_r_odd i : b^1_rb^2_n - r : where @xmath61 in the above expression are annihilation and creation operators associated with fermion fields with fourier expansion _ over odd modes exclusively _ _ -^(odd)j= _ n_oddb^j_n e^-in(t - x)/r .
the difference with respect to the standard fields is that the commutation relations for the bosonic creation / annihitation operators are now of the form ( with @xmath25 and @xmath62 even entires ) : [ a^-_n , a^-_-m]= _
n , m .
the above mechanism and the gc ( and also pc ) suggest to continue the reduction process and consider fermion fields @xmath63 whose fourier expansion _ exclusively contains ( odd ) prime modes _ the set of all ( odd ) prime numbers , including the number one . ] [ psi]^j= _
pb^j_p e^-ip(t - x)/r .
the ( even ) bosonic field can be reconstructed from the expression a^-_n= sgn(n ) _
r i : b^1_rb^2_n - r : , and it has now the following non - canonical commutation relations . ]
[ a^-_n , a^-_-m]= _
n , m , with @xmath25 and @xmath62 even entires ; @xmath64 stands for the number of times the positive even integer @xmath25 can be written as the sum of two primes ( the so - called goldbach s partition ) .
notice that the goldbach conjeture has been invoked in order to reconstruct the field @xmath65 as a product of two fields @xmath63 with only prime fourier components .
this becomes explicit in the assumed non - vanishing function @xmath64 in the above commutation relations . according to quantum field theory ,
@xmath66 and @xmath67 can be interpreted as annihilation and creation operators acting on the vacuum state denoted as @xmath68 such that [ vacuum ] b_p^j
| 0_r = 0 once the creation and annihilation operators have been defined , one can readily obtain two - prime states as [ 2p ] |p_1,p_2 = nb_-p_1 ^ 1 b_-p_2 ^ 2 |0_rwhere @xmath69 ; and @xmath70 stands for the normalization of states : @xmath71 if @xmath72 , or @xmath73 if @xmath74 . similarly , a three - particle state results from [ 3p ] @xmath75 satisfying @xmath76 .
the validity of the gc implies that every even state , whose @xmath20mass is given by @xmath77 , can be interpreted as a two - particle ( two - prime ) state @xmath78 , where @xmath79 .
correspondingly , every odd state , with mass @xmath80 can be interpreted as either a one - prime state ( for @xmath81 a prime number itself ) or a three - prime state @xmath82 , where @xmath83 .
leaving aside the fact that we are dealing with a noninteracting field theory , states such as the two- and three - prime excitations of the vacuum bear some resemblance to hadrons , i.e. mesons and baryons , interpreted in non - relativistic models of particle physics as quark - antiquark and three - quark ( or antiquark ) states , respectively .
naturally , four - prime states ( or higher ) can be formed by acting four times with the creation operation on the vacuum : [ 4p ] any even combination of creation operators .
given an even integer @xmath5 in our physical model , the gc implies that all those fock states such that @xmath84 , are in fact degenerate .
( of course , similar extensions can be applied to odd combinations of creation operators . )
would the gc fail for a given @xmath5 , the corresponding state could not be represented as a two - prime state anymore but likely as a _ higher _ fock state . in particle physics jargon
, one would refer to such states as @xmath20exotic .
the particle interpretation of the model presented above requires a consistent definition of the vacuum energy on a two - cylindrical spacetime .
we examine in detail this issue in the following sections .
specifically , we will study the regularization and renormalizability properties of both a bosonic field and a fermionic field on the circle , the latter with vanishing fourier coefficients for modes with non - prime labels .
we will pay special attention to the two point functions and the expectation value of the hamiltonian operator ( casimir energy ) .
the expectation value of the the hamiltonian @xmath85 of the free scalar field @xmath26 , can be written as [ sumharm ] 0_r|h_| 0_r = _ nw_n , w_n=|n|/r where the sum runs over all integer numbers and the vacuum state corresponds to a given radius @xmath23 .
this quantity obviously diverges as @xmath86 . on the other hand , in the absence of gravity
the zero - point energy actually has no physical meaning in itself .
this motivates the removal of the infinite vacuum energy of free minkowski space by defining the casimir energy @xmath87 as in @xcite : [ casimir ] e_c=0_r|h_|0_r - _
r0_r|h_|0_r nowadays it has become standard practice the use of a regularization method to determine the casimir energy based on the riemann @xmath88 function @xcite defined as [ zeta ] ( s ) = _ nn^-s analytic continuation of @xmath2 from the region of the complex plane such that @xmath89 , to the whole complex plane including the real negative axis ( and therefore @xmath90 ) permits the regularized summation of the divergent series ( [ sumharm ] ) .
thus one easily gets the textbook result for the vacuum energy density @xmath91 : @xmath92 moreover , the vacuum energy energy for the chiral fields @xmath93 is , obviously , half of the vacuum energy density obtained above for @xmath50 .
we also note that the casimir energy density when the sum over the modes is restricted to even ( odd ) integers is @xmath94 ( @xmath95 ) . ]
leads to @xmath96 and @xmath97 for @xmath90 , respectively @xcite . ]
for the massless dirac field @xmath98 considered before the expectation value of the hamiltonian @xmath99 is minus the corresponding scalar field result [ sumharmspsi ] 0_r|h_| 0_r = -_nw_n , w_n=|n|/r therefore , the regularized vacuum energy density turns out to be positive , with value @xmath100 . on the other hand
, we are considering @xmath101 introduced in eq.([psi ] ) as the fundament fields in our model .
consequently , the expectation value of the hamiltonian @xmath102 reads now [ sumprimes ] 0_r|h_| 0_r = - _
p w_p , w_p = p / r where the sum runs over all ( odd ) prime numbers . similarly to ( [ sumharm ] ) and ( [ sumharmspsi ] ) , this quantity also diverges as @xmath103 and requires a regularization procedure .
in fact , the formal similarity of summations ( [ sumharm ] ) and ( [ sumprimes ] ) suggests trying a similar regularization as before but this time using the the riemann prime zeta function , defined as implies that any successful regularization that can obtained with @xmath104 can also apply to @xmath105 , and viceversa . ]
[ p ] p(s)=_p p^-s .
unfortunately , @xmath106 can not be analytically continued to the negative real half - plane @xmath107 because of a natural boundary along the line @xmath108 @xcite .
therefore , the above method to obtain the casimir energy , though successful for integer modes as shown before , can not be applied to our model with prime modes only .
this failure should not be seen as exceptional as the casimir effect regularization based on the use of the analytic continuation of the riemann function is known to be not always applicable .
this happens , for instance , in the case of a massive field on a two - dimensional cylinder @xcite . in the next section
we will approach this important question by a more systematic method based on the point - splitting regularization and renormalization tools of quantum field theory @xcite .
it is instructive to consider the calculation of the vacuum energy of the scalar field @xmath11 by means of the so - called abel summation of the divergent series ( [ sumharm ] ) . to motivate this procedure we start with the two - point function of the field , given by the sum in modes 0_r|(t , x ) ( t , x)|0_r = _ n=1^ [ e^-in[(t - t -i)-(x - x)]/r+ e^-in[(t - t
-i)+(x - x)]/r ] , where , as usual , the infinitesimal @xmath109 has been appropriately inserted to make the two - point function well - defined . as is customary in quantizing a field in two - dimensions , we have discarded the zero mode @xmath110 in the above summation . at coincidence
@xmath111 we have 0_r|^2(t , x)|0_r = _ n=1^ e^-n / r , which turns out to be the abel summation of a divergent series @xcite . from the sum _
n=1^ n^-1e^-n / r= -(1-e^-/r ) , one can easily derive the expression for the two - point function @xcite [ tpf ] 0_r|(t , x ) ( t , x)|0_r = -(1-e^-i(u -i)/r)(1-e^-i(v -i)/r ) , where it is understood that the limit @xmath112 should be taken and @xmath113 and @xmath114 . according to the above discussion
, the energy density of the field , which classically is given by t_tt= , can be formally evaluated , at the quantum level , as 0_r|t_tt|0_r = _ 0^+ ( 2r)^-1_n=1^w_ne^-n / r . in the limit @xmath112 , the quantity [ sumdamping](2r)^-1_n=1^w_ne^-n / r= .
is divergent , reflecting the well - known divergent vacuum energy .
however , as is usual in quantum field theory one can invoke _ renormalization _ to properly find a finite quantity for the physical vacuum energy density .
to renormalize @xmath115 one can perform an inverse radius expansion in [ texpansion ] 0_r|t^()_tt|0_r : = ( 2r)^-1_n=1^w_ne^-n / r= -+ o(r^-4 ) , and renormalize by subtracting , from the exact expression ( [ sumdamping ] ) , the terms in the expansion ( [ texpansion ] ) that diverge when @xmath112 [ tren ] 0_r|t_tt|0_r _ ren = _ 0^+= - . in the cft language ,
this result is somewhat related to the calculation of the central charge @xmath51 of the scalar field @xcite .
it is now illustrative to consider the restriction of the modes of the original field @xmath50 , parameterized by @xmath25 , to the set of even ( or odd ) integer numbers . in this case
we have 0_r|t^(even)_tt|0_r = ( 2r)^-1_n = even^w_ne^-n / r= -+ o(r^-4 ) , and , after renormalization , we get 0_r|t^(even)_tt|0_r _ ren = - .
similarly , for the quantum field defined on the odd integer numbers modes one gets 0_r|t^(odd)_tt|0_r _ ren = .
we reproduce in this way the results obtained in the previous section with the zeta function regularization .
let us now extend the above analysis to the fields @xmath63 introduced in section @xmath116 . since , by definition , the ( chiral ) field @xmath117 ( for simplicity we omit the index @xmath118 ) has only ( odd ) prime number modes in its fourier expansion , the corresponding two - point function is then given by [ tpfprimes ] 0_r|(t - x ) ( t-x)|0_r = _ p e^-ip[(t - t -i)-(x - x)]/r . at coincidence , one should deal with the following abel sum 0_r|^2 ( ) |0_r = _ p e^-p / r , which , as expected , is divergent when @xmath112 .
moreover , in the calculation of the vacuum energy density one faces directly the abel sum of all ( odd ) prime numbers 0_r|t^prime ( ) _ tt|0_r = -_p p e^-p / r . in sharp contrast
with the field @xmath50 , the restriction to the prime numbers turns the above sums into rather complicated functions of @xmath119 .
our next and crucial step is to carry out a renormalization procedure to properly associate finite quantities to the above divergent sums . as in the previous analysis of the field
@xmath50 , this requires the introduction of suitable asymptotic expansions for large radius .
the leading terms in the asymptotic expansions of @xmath120 is essentially fixed by the corresponding ones in minkowski space : @xmath121 .
however , the definition of @xmath117 requires of a finite radius @xmath23 and thus it is not possible to define @xmath117 in minkowski space .
therefore , the asymptotic expansions needed to identify the subtracting terms can not be based on the results in minkowski space and one has to construct them according to new criteria that we investigate in the following .
since our considerations should necessarily involve some conventional but sophisticated mathematics we will study the above abel sums on the more conventional set @xmath104 , instead of @xmath105 . inasmuch as the fundamental aspects of the renormalization process only involves the behavior of the abel sum as @xmath103 , the shift from @xmath105 to @xmath104 is irrelevant ( one should add the quantity @xmath122 to the result obtained with @xmath104 ) .
let us now focus on the renormalization of the expectation value of the energy density .
to perform renormalization we should expand @xmath123 for large radius .
the desired expansion can be computed either directly or by differentiating the expansion of @xmath124 ( @xmath125 ) with respect to the parameter @xmath126 twice and taking into account the additional factor @xmath127 .
the result turns out to be the same . after an involved calculation it is possible to show that the asymptotic expansion of @xmath124 when @xmath128 satisfies ( without invoking the riemann hypothesis ) [ fernandoexpansion1 ]
+ o(1 ) , where @xmath129 is mertens constant .
this result is heuristically suggested by mertens theorem , but a rigorous proof involves non - trivial computations . for more details of this and other calculations
we refer the reader to the appendix at the end of the paper .
note that the growth of the first ( and only ) divergent term on the r.h.s . of eq.([fernandoexpansion1
] ) is very slow , due to the natural decrease of the density of prime numbers as @xmath103 . to go further in the analysis
we are forced to know the subleading terms in the expansion ( [ fernandoexpansion1 ] ) .
for instance , terms of order @xmath130 in the expansion of @xmath124 are converted , after derivation , into divergent terms .
as explained in the appendix one can obtain the following expansion if the rh is assumed to be true [ fernandoexpansion2 ] _
p e^-ap ~(-a)+b_1-_k=1^ , where @xmath131 stands for the @xmath132-order derivative of the @xmath133 function .
for example , the first five terms in this expansion are _
p e^-ap & ~&(-a)+b_1 - - + & -&+the expansion for @xmath134 can be expressed then as [ fernandoexpansion3 ] 0_r|t^prime ( ) _ tt|0_r ~
[ asymtoticfernando2]this series has an infinite number of divergent terms ( as @xmath112 ) and their sum does not converge .
therefore , one can not define at this point a proper subtracting term to cancel out the divergences and , hence , one could be tempted to conclude that renormalization is not possible .
however , a more careful inspection shows that this situation could be conceivably improved . to this end
let us go back to the origin of ( [ fernandoexpansion2 ] ) . as explained in the appendix , by accepting the truth of the _ riemann hypothesis _
it is possible to derive the following asymptotic expansion [ fernandoexpansion4 ] _
p e^-ap = ( 1+(-a)^2)+b_1- _ 0^ e^i t a ( ( -i t)-)dt+ o(a^1/2- ) , that , by combining a further expansion of the first term with the third , leads to ( [ fernandoexpansion2 ] ) . similarly , assuming the riemann hypothesis , the asymptotic expansion for @xmath134 can be cast in the form [ expansiontprime ] 0_r|t^prime ( ) _ tt|0_r : = _ p p e^-p / r = ( f(/r)+ o((/r)^-3/2- ) ) , where all divergences in ( [ fernandoexpansion3 ] ) are captured by the single function @xmath135f(a)=-_0^e^ita(2-it)t .
it is important to notice , however , that the terms of the form @xmath136 in ( [ expansiontprime ] ) are potentially divergent .
one would need to know the form of the non - leading contributions in ( [ expansiontprime ] ) to see whether or not the asymptotic expansion can be expressed as a finite sum of hierarchical divergent terms . in order to do this
it would be necessary to go beyond the asymptotic analysis presented in the appendix and study in detail the contributions to the asymptotic expansions coming from all the singularities in the integrands of ( [ intrep ] ) and ( [ intrep2 ] ) .
the very complicated analytic structure of the relevant functions precludes us from making a definitive statement but , the presence of an infinite number of branch points originating in the non trivial zeros of the zeta function , strongly suggests that an infinite number of counterterms would be necessary to subtract all the infinities appearing in the expectation value of the stress - energy tensor .
this would mean that the model is non - renormalizable .
motivated by the goldbach and polignac conjectures in number theory and the classical results on abelian bosonization in two - dimensions we study in this paper the `` factorization '' of a real scalar field ( whose fourier expansion is made out of even modes ) as a product of two ( so - called prime ) fermion fields whose fourier expansion exclusively contains prime modes .
we have subsequently investigated the quantum properties of this theory by studying the vacuum energy density . as expected , the short - distance behavior of the prime field is highly non - trivial and quite different from a conventional field theory on the cylinder . due to the restriction to the prime modes ,
the short - distance singularity is no longer purely of the form @xmath137 , but rather of form @xmath138 this feature makes the calculation of the renormalized expectation value of the energy density quite involved .
the computation has turned out to be very elusive since an infinite number of divergences emerges naturally , showing the non - renormalizability of the underlying qft .
let us stress that the riemann hypothesis plays an important role in the calculation of the vacuum energy density .
this is , in our opinion , a remarkable outcome of this work .
the possibility of relating consistency properties of a particular field theory ( such as the one considered here ) to the riemann hypothesis would definitely be very intriguing .
the non - renormalizability of the theory in the standard sense would suggest that there is some scale at which its properties change dramatically . for instance , the two - particle sector of the fock space reflects the gc at low energies . accordingly , the set of all two - particle states span all energies of the form @xmath139 .
the non - renormalizability of the theory could signal a breakdown of the gc above some number density scale @xmath140 , acting as a sort of ultraviolet cut - off . by breakdown of the gc we should understand in this context that infinitely many even numbers can not be written as the sum of two primes . on the other hand , since the value of the radius @xmath23 can be freely chosen , the ultraviolet cut - off and therefore @xmath70 should be in principle also as large as desired .
this view actually suggests that the gc might be satisfied for arbitrarily large even numbers .
one can envisage this possibility on deeper physical grounds by extending our study beyond qft , e.g. to string - like models , where dualities involving the size of @xmath23 ( like @xmath141-duality , see for example @xcite ) could allow one to exploit this fascinating idea .
let us finally stress that no actual claim is made in this paper regarding the proof or disproof of the goldbach conjecture .
our main result is to point out that a simple theory , a free field on a cylindrical spacetime with modes restricted to prime integers , is full of divergences , making impossible to properly define the vacuum energy .
we believe that the truth of the goldbach conjecture may be deeply related with this result , but in a way not unraveled yet .
we give here some basic results about the asymptotic behavior as @xmath142 of the functions @xmath143 where the sums are extended to the set of prime numbers @xmath104 . as a way to highlight the role of the riemann conjecture in the final form of the asymptotic expansions we will use complex variable methods .
the functional series @xmath145 is convergent for @xmath146 with @xmath147 and uniformly convergent in every compact subset of this set , hence it defines a holomorphic function in the complex half plane @xmath148 .
the best way to study the asymptotic behavior of @xmath149 as @xmath142 is to use mellin transform methods . in the present case
the integral defining the mellin transform converges if @xmath150 and is given by @xmath151=\sum_{p\in \mathbb{p}}\frac{1}{p}m[e^{-a p};s]=\gamma(s)p(s+1 ) \label{mellintransformoff}\ ] ] where @xmath152 denotes the prime zeta function .
this function is defined , in analogy with the standard zeta function , by the functional series @xmath153 for complex values of @xmath154 with @xmath155 .
the mellin inversion formula provides us with the following integral representation for @xmath149 .
@xmath156 with @xmath157 .
this representation is the main tool to obtain the sought for asymptotic formulas for @xmath149 and @xmath158 . by using the expansion
( see for example @xcite ) @xmath159 that can be obtained by using mbius inversion denotes the mbius function and @xmath88 is the standard zeta function .
] , we get @xmath160 that can be equivalently written as let us start by considering the first integral in ( [ tresintegrales ] ) . without using the riemann hypothesis this integral can be easily seen to be @xmath163 by moving the integration contour to the line @xmath164 .
this is possible because there are no zeros of the @xmath88 function on the line @xmath165 .
also , we have @xmath166 and @xmath167 and , hence , the contributions of the segments needed to close the integration contour and displace it to the line @xmath164 go to zero as @xmath168 .
the previous result can be improved by noting that the riemann - lebesgue lemma actually shows that the limit of the integral when @xmath169 vanishes [ or , equivalently , the integral is @xmath170 .
if the riemann hypothesis is true the integral can be shown to be @xmath171 for all @xmath172 by moving the contour to the line @xmath173 . again
this result can be slightly improved and the integral shown to be @xmath174 as a consequence of the riemann - lebesgue lemma . * the ones with
the largest real part correspond to the pole of @xmath133 at @xmath176 .
the residues are @xmath177 for @xmath178 . *
the pole of @xmath88 gives rise to branch points at @xmath179 , with @xmath178 .
* the non - trivial zeros of the @xmath88 function , if the riemann hypothesis is true , give rise to branch points with @xmath180 , and @xmath178 as in the integral considered in the first step , the integration contour can be moved parallel to itself due to the exponential fall - off of the integrand and the mild growth of the absolute value of the @xmath88 function . by moving the contour past the origin
we get contributions , from the residues at the poles of the @xmath181 factor , that add up to the constant @xmath182 this , combined with the euler constant @xmath45 , will give a final contribution equal to the mertens constant @xmath183 .
once past the @xmath176 singularity , the integration contour can be displaced to the line @xmath184 and it is straightforward to show that the sum of the integrals on this contour is @xmath174 without invoking the riemann hypothesis .
the singularities in the integrand are determined now by the poles of the gamma function at the non - positive integers , and the branching point of the logarithm at @xmath176 .
the quick fall - off of the integrand provided by the gamma function allows us to freely move the integration contour in the negative direction of the real axis while wrapping around the singularities of the integrand . in order to compute the integral
it is useful to somehow disentangle the singularity structure by choosing a branch cut for the logarithm such that @xmath186 $ ] ( see figure [ fig : contour ] ) .
as long as the choice does not change the value of the integrand in the region where the initial integration contour lies this is allowed .
the advantage of proceeding in this way is that the poles of the gamma function will not lie on top of the branch cut .
a straightforward computation gives now @xmath187 it is important to realize that despite the relative complication of ( [ re1 ] ) it is actually easy to find its asymptotic behavior when @xmath142 .
in particular , the last integral can be easily seen to be @xmath188 by integration by parts and its full asymptotic expansion can be obtained directly by this procedure .
this gives @xmath189\big(\gamma(-i t)-\frac{ie^{-t}}{t}\big)\\ & & = -\sum_{k=1}^\infty\frac{(-1)^k\gamma^{(k)}(1)}{k}\frac{1}{(-\log a)^{k}}+\frac{1}{2}\sum_{k=1}^\infty\frac{(-1)^k}{k}\frac{1}{(-\log a)^{2k}}\,.\end{aligned}\ ] ] in the limit @xmath142 the asymptotic behavior is controlled by the divergence of @xmath190 that behaves as @xmath191 for small enough values of @xmath126 .
@xmath192 by taking together the results obtained for each of the integrals in ( [ tresintegrales ] ) we conclude that , without invoking the riemann hypothesis , we have @xmath193 whereas , if the riemann hypothesis is true we have @xmath194 for every @xmath195 . here
@xmath196 is the mertens constant mentioned above . as a side remark we would like to mention here that the asymptotic expansion ( [ as - exp ] ) is the same @xcite as the one corresponding to the truncated series @xmath197 in the limit @xmath169 .
the function @xmath198 can be studied by following the same steps as before .
now the functional series @xmath199 is convergent for @xmath146 with @xmath147 and uniformly convergent in every compact subset of this set , hence it defines a holomorphic function in the complex half plane @xmath148 .
it is straightforward to find the following representation for @xmath198 by using mellin transforms .
@xmath200 with @xmath201 .
this expression can be written as @xmath202 the singularity structure of the integrand suggests to split the previous expression for @xmath198 as @xmath203 for a certain value of @xmath162 . without invoking the riemann hypothesis the first integral in ( [ intsforg ] )
can be easily shown to be @xmath204 by moving the integration contour ; actually , by invoking the riemann - lebesgue lemma it can be seen to be @xmath205 .
if the riemann hypothesis is true then this integral can be seen to be @xmath206 ( a bound that can , again , be slightly improved to @xmath207 by using the riemann - lebesgue lemma ) .
let us consider now the sum of integrals @xmath208 at variance with the analysis presented for @xmath144 , in this case the singularities of @xmath181 do not play any role to determine the asymptotic behavior of @xmath198 .
this expression can be seen to be @xmath207 regardless of the truth of the riemann hypothesis .
finally let us study the behavior of @xmath209 the singularities of the integrand and the the integration contours used to study the asymptotic behavior of this integral are shown in fig .
[ fig : contour2 ] .
a simple computation gives now @xmath210 the asymptotic behavior of this last integral can be easily obtained by integration by parts @xmath211\gamma(2-i t)=\frac{1}{a^2}\sum_{k=0}^\infty \frac{(-1)^k\gamma^{(k)}(2)}{(-\log a)^{k+1}}\,.\ ] ] gathering the preceding results we conclude that without invoking the riemann hypothesis we can only state that @xmath212 whereas , if the riemann hypothesis is true , we have that @xmath213 for all @xmath195 .
it is interesting to point out here that the leading asymptotics of @xmath198 can be obtained by differentiating twice the asymptotic behavior of the leading terms in @xmath144 , i.e. @xmath214 @xmath215 this work is supported by the meyc research grants fpa2011 - 23596 , fis2009 - 11893 , fis2008 - 06078-c03 - 02 , the consolider - ingenio 2010 program cpan ( csd2007 - 00042 ) and the generalitat valenciana grant gvprometeo2010 - 056 .
we thank l. lvarez - gaum and e. elizalde for useful comments . | motivated by the goldbach conjecture in number theory and the abelian bosonization mechanism on a cylindrical two - dimensional spacetime we study the reconstruction of a real scalar field as a product of two real fermion ( so - called _ prime _ ) fields whose fourier expansion exclusively contains prime modes .
we undertake the canonical quantization of such prime fields and construct the corresponding fock space by introducing creation operators @xmath0 labeled by prime numbers @xmath1 acting on the vacuum .
the analysis of our model , based on the standard rules of quantum field theory and the assumption of the riemann hypothesis , allow us to prove that the theory is not renormalizable .
we also comment on the potential consequences of this result concerning the validity or breakdown of the goldbach conjecture for large integer numbers .
report : ific/12 - 03 , ftuv-12 - 0131 + + * miguel - angel sanchis - lozano + 0.1 cm _ instituto de fsica corpuscular ( ific ) and departamento de fsica terica + _ centro mixto universitat de valncia - csic + dr . moliner 50 , 46100 burjassot , valencia ( spain ) _ _ * 0.4 cm * j. fernando barbero g. * + 0.1 cm _ instituto de estructura de la materia , csic + _ serrano 123 , 28006 madrid ( spain ) _ _ 0.4 cm * jos navarro - salas * 0.1 cm _ instituto de fsica corpuscular ( ific ) and departamento de fsica terica + _ centro mixto universitat de valncia - csic + dr .
moliner 50 , 46100 burjassot , valencia ( spain ) _ _ _ omnia disce , videbis postea nihil esse superfluum _ + ( learn everything , later you will see that nothing is superfluous . ) + hugh de saint victor ( c. 1078 - 1141 ) , french or german philosopher 1.4 cm |
the search for the higgs boson / bosons and the study of their properties tops the priority list of the on going experiments at the lhc at 7 tev as well as of the upcoming experiments at the highest attainable energy of 14 tev .
this will shed light on the mechanism of electroweak symmetry breaking ( ewsb ) and the generation of the particle masses .
the higgs sector of the very successful standard model ( sm ) of particle physics responsible for ewsb is indeed very simple .
it consists of a single neutral , scalar higgs particle .
however , this economical higgs sector turns out to be the achilles heel of the sm .
the scale of ewsb is expected to be around 100 gev . on the other hand the naturalness or the hierarchy problem
@xcite comes into play , should there be some new physics beyond the sm characterized by a much higher energy scale .
the very existence of the new scale tends to push up the ewsb scale far beyond the expected magnitude .
for example , the higgs boson mass blows up to values closer to the new scale , in contrast to the preferred range @xcite @xmath3 suggested by ew precision data ( the hierarchy problem ) , unless the parameters of the sm are extremely fine tuned ( the naturalness problem ) .
even the absence of any compelling evidence in favour of a new physics model does not allow us to overlook the above issues , since one can not wish away the planck scale , the scale of gravity which is always there . the above theoretical inconsistency has led to several extensions of the sm .
the minimal supersymmetric standard model(mssm ) @xcite removes the above problem elegantly in the limit of exact supersymmetry ( susy ) and stabilizes the higgs mass .
however , susy must be broken by the soft breaking terms , since the negative results of sparticle searches at various colliders indicate that the superpartners ( or the sparticles ) must be considerably heavier than the corresponding particles . in spite of susy breaking ,
the naturalness problem remain under control , provided the masses of the sparticles are @xmath4 1 tev @xcite or smaller . in the mssm the higgs sector must necessarily be extended and the spectrum consists of two neutral scalars ( @xmath5 and @xmath6 ) , one neutral pseudo scalar ( @xmath7 ) and two charged scalars ( @xmath8 ) @xcite .
enormous amount of work has been done in developing the strategies for searching these bosons at the lhc @xcite . adding any new dimension to the above studies is not the aim of this paper . however , at the lhc these higgs bosons are likely be produced along with a variety of superpartners of the standard model particles - the sparticles .
in particular the strongly interacting sparticles may indeed be produced in numbers much larger than that of typical higgs induced events . even if a small fraction of such events pass through the selection criteria for the dedicated search of any of the above higgs bosons , then the shape of certain distributions expected for higgs production alone may change significantly , indicating that the higgs boson may be part of a framework larger than the sm .
such an unexpected shape is likely to be prominent near the tails of the distributions under study where contributions from the higgs signal as well as the sm background are naturally small . in this paper
we shall illustrate this possibility with a specific example .
some of the above issues have recently been addressed in the context of charged higgs search by a counting experiment during the lhc-14 tev run @xcite .
it has been demonstrated that both in the unconstrained mssm and in the minimal supergravity(msugra ) model @xcite , susy events can seriously affect the size of the charged higgs signal unless additional selection criteria are introduced . on the other hand , the combined squark - gluino and charged higgs events may establish physics beyond the standard model ( bsm ) physics at a confidence level higher than that attainable by the higgs signal alone .
this enhancement would disfavour not only the sm but also its extensions with two higgs doublets .
after this important hint regarding bsm physics , a clean charged higgs sample may be recovered by additional kinematical cuts .
it needs to be emphasized that the above issues are relevant in any extension of the sm with a new particle sector as well as an extended higgs sector . in this paper
we focus our attention on the search for the lighter cp even scalar higgs boson(h ) .
the h boson in the mssm has properties very similar to the sm higgs boson in a large volume of the parameter space ( the so called decoupling region ) .
thus the discovery of a single neutral higgs scalar , which is a likely scenario in the early stages of the lhc experiment , will yield very little information regarding the underlying theory . the situation may change dramatically if susy is realized in nature in such a way that a sizable number of sparticle induced events change the characteristics of the expected h signal significantly .
this can convincingly establish the bsm origin of the discovered boson .
even the discovery of one or more heavier higgs boson with no counterpart in the sm , would indicate only an extended higgs sector but not the existence of new heavy sparticles .
the mass of h is always bounded from above ( @xmath9 gev ) @xcite in supersymmetric theories . as a consequence
the search strategy for it is very similar to that for the standard higgs boson with mass within the above bound especially in the decoupling region .
if @xmath10 gev , an important discovery channel would be the inclusive di - photon + x channel with a relatively modest integrated luminosity ( @xmath11 ) of a few tens of @xmath12 at the 14 tev run of the lhc ( see figure 10.38 of @xcite ) . here
x stands for any other particle accompanying the h signal .
this signal stems from h production at the lhc via several channels , followed by the one loop decay @xmath13 .
the suppression of the signal due to the tiny branching ratio ( br ) of this decay ( @xmath14 ) is adequately compensated by the relatively low sm background . in the mssm also a small peak in the @xmath15 invariant mass distribution around the h mass would establish the h signal ( see figure 11.37 of @xcite ) .
among various models of softly broken supersymmetry the gauge mediated supersymmetry breaking ( gmsb ) models @xcite ( to be briefly reviewed below ) predict a large number of di - photon + x events at lhc energies coming from the production of strongly interacting sparticles . in this paper
we wish to focus on the impact of these events on the di - photon invariant mass distribution to be studied by the dedicated h - search experiments with utmost care . in these models
the gravitino with negligible mass turns out to be the lightest supersymmetric particle ( lsp ) and , hence , the carrier of missing energy .
if the lightest neutralino ( @xmath16 ) is the next lightest supersymmetric particle ( nlsp ) , it can decay into the lsp and a photon with a large br . in r - parity conserving supersymmetry , sparticles are produced in pairs and each of them eventually decay , through a cascade , into a gravitino - photon pair . each susy event in this scenario will , therefore , have the @xmath17 + x topology . thus the pair production of squarks and gluinos in any combination having a large cross section may leave their signature in the observed @xmath15 invariant mass distribution .
the plan of the paper is as follows . in the next section
we briefly describe different gmsb scenarios and the present collider constraints on them .
we also define our benchmark points for the analysis . in section 3 ,
we present our main results on the di - photon invariant mass distribution in gmsb models at the lhc
. our emphasis will be on the ongoing lhc experiments .
however , the possibilities for the 14 tev experiments will also be briefly touched upon .
our conclusions will be summarized in section 4 .
to begin with we shall review the constraints from different collider experiments on the parameter space of the minimal gmsb ( mgmsb ) model @xcite .
we also wish to add a few points regarding the approximations in obtaining these bounds and the resulting uncertainties which have not been sufficiently elucidated in the current literature . in this model
the soft breaking parameters are generated via the gauge interactions of the sm . as a result the sparticles with the same sm quantum numbers but different flavours acquire the same mass .
this keeps the potentially dangerous flavour changing neutral current induced processes under control . in the mgmsb model
there is an observable sector consisting of the mssm fields .
the supersymmetry breaking sector consists of a gauge singlet chiral superfield y , identified with the goldstino superfield in the simplest version of the model .
the scalar and auxiliary components of y are assumed to develop vacuum expectation values ( vevs ) denoted by s and f respectively . in the early versions of this model
the dynamics of generating these vevs were not specified .
the model also has a messenger sector consisting of superfields @xmath18 having the gauge interaction of the sm .
the messenger superfields interact with y via a tree level interaction in the superpotential .
this generates a supersymmetric mass of @xmath19(m ) of the messenger fields , usually referred to as the messenger scale .
a susy breaking mass squared splittings of the order f is also generated within the messenger supermultiplets .
this susy breaking is communicated to the observable sector by gauge interactions between the messenger and the observable fields via higher order processes .
a restriction in the messenger sector comes from the requirement of the unification of the coupling constants of the standard model . if the messenger superfields form complete gut multiplets ( e.g. , 5 or @xmath20 of su(5 ) ) , the value of the unification scale or the gut scale ( @xmath21 ) does not change .
however , coupling constant unification constrains the number of messenger superfields(n ) to be @xmath22 5 .
the simple mgmsb model is characterized by five parameters @xcite @xmath23 , m , @xmath24 , tan @xmath25 and sign(@xmath26 ) where @xmath23 = f / s is the susy breaking scale in the observable sector , m is the messenger mass scale , @xmath24 is the number of messenger multiplets belonging to the 5 + @xmath20 representation of su(5 ) , tan @xmath25 is the ratio of the vacuum expectation values of the neutral higgs fields in the observable sector and @xmath26 is the higgs mixing parameter in the superpotential with magnitude fixed by the radiative symmetry breaking condition .
it bears recall that the soft breaking trilinear ( a ) and bilinear ( b ) parameters are generated by higher order processes and are small at the messenger scale m. the weak scale parameters are then obtained by the standard renormalization group ( rg ) evolutions .
the sparticle spectrum at the weak scale can be computed by both suspect version 2.41 @xcite and isajet @xcite .
the phenomenology of this model has been discussed by various authors @xcite the current experimental lower bound on the higgs boson mass from lep @xmath27 gev @xcite poses the strongest constraint on the parameter space of the mgmsb model .
given the input parameters one can compute the higgs mass including higher order corrections .
in this paper @xmath27 gev will be henceforth referred to as the stronger bound on the computed h - mass .
however , there is an estimated theoretical uncertainty of about 3 gev on the computed h - mass due to yet unknown higher order effects@xcite . in view of this a point in the parameter space with computed @xmath28 111.4 gev ( the weaker h - mass bound ) may still be acceptable . in figure
[ fig : mh_tb ] we present @xmath29 computed as a function of @xmath23 .
we take @xmath30=173 gev , @xmath31=@xmath32 , @xmath24=1 .
it follows that for tan @xmath25 = 5 ( tan @xmath25 = 20 ) the stronger h - mass bound yields a lower bound @xmath33 257 tev ( 118 tev ) .
if , however , the theoretical uncertainties are taken into account and the weaker @xmath29 bound is used instead , relaxed bounds @xmath33 161 tev ( 79 tev ) are obtained .
it is interesting to note that for @xmath34 , @xmath35 gev which corresponds to gluino mass ( @xmath36 ) and average squark mass ( @xmath37 ) of approximately 2 tev and 3 tev respectively .
thus should the higgs mass bound continues to be pushed upwards , the squark - gluino search at the lhc , as predicted by mgmsb , will be quite challenging .
the @xmath38 and cdf collaborations have put strong constraints on nlsp ( @xmath39 ) mass . in order to obtain quantitative results
they consider sps8 slope in the mgmsb model which has only one parameter @xmath23 .
the other parameters are fixed and given by @xmath31=2@xmath23 , @xmath40=15 , @xmath24=1 and @xmath41 .
the lifetime of the @xmath39 is not fixed by these parameters and it is assumed to be sufficiently short .
it means that the photon coming from the decay of @xmath39 is prompt .
the gaugino pair production processes are expected to be dominant at the tevatron .
the decay of two @xmath39 produce two photons and gravitinos that give rise to missing energy .
the latest constraint is from @xmath38 collaboration who have obtained a lower bound 124 tev on the scale @xmath23 with 6.3 @xmath12 of data at run ii @xcite .
we have computed the leading order ( lo ) cross sections for the dominant susy processes at tevatron energies ( mainly electroweak gaugino pair production ) using pythia @xcite . the next to leading order(nlo ) cross section and the k factor with renormalization scale(@xmath42)= factorization scale ( @xmath43)= 2@xmath44 is computed by prospino @xcite .
our cross section for @xmath23=124 tev ( assuming sps8 point ) is in good agreement with the result in @xcite .
we find that the above bound corresponds to an upper bound on the susy production cross section of 4.5 fb . for promptly decaying @xmath16 this
@xmath45 yields @xmath46 174 gev .
this bound is indicated in the @xmath23 - tan@xmath25 plane ( figure [ fig : scan ] ) by the horizontal line . for comparison the stronger and weaker h
- mass bounds are also indicated in the same figure .
it follows that the d0 bound on @xmath45 supersedes that from the stronger ( relaxed ) @xmath29 bound for tab @xmath47 15 ( 6.5 ) .
it follows that irrespective of the uncertainties in @xmath29 due to higher order corrections and the choice of tan @xmath25 , values of @xmath23 smaller than 124 tev are disfavoured .
this bound also supersedes the earlier cdf lower bound on susy scale 107 tev@xcite .
the assumptions underlying this bound will be critically examined below . strictly speaking
the @xmath38 bound is valid for the snowmass slope sps8 with variable @xmath23 .
the other parameters are m = 2 @xmath23 , @xmath24 = 1 , tan @xmath25 = 15 and sign ( @xmath48 0 . for @xmath49 2 ,
the scenario reduces to a @xmath50 nlsp scenario which is not under consideration in the present paper .
however , keeping other parameters fixed and varying tan @xmath25 we have checked that neither the cross sections of the dominant processes ( @xmath51 and @xmath52 pair production @xcite ) at the tevatron energy nor the @xmath53 and @xmath54 masses vary appreciably with tan @xmath25 .
this is , however , expected since over the scanned parameter space the above gauginos are wino like to a very good approximation and have masses controlled by the @xmath55 gaugino mass @xmath56 at the weak scale .
the cross section , therefore , shows only a mild tan @xmath25 dependence .
the bino - like @xmath16 mass on the other hand is determined by the @xmath57 gaugino mass parameter ( @xmath58 ) at the weak scale related to @xmath56 by the gaugino mass unification condition .
we shall , therefore , use the @xmath38 bound throughout this analysis unless the electroweak gauginos happen to be significantly mixed , which is often the case for non - minimal models to be discussed below .
the neutralino mass bound from tevatron has an important advantage .
the predicted @xmath29 depends sensitively on @xmath30 ( recall the @xmath59 dependence in the radiative corrections to @xmath29 ) .
the minimum value of @xmath23 consistent with the weaker or stronger h - mass bound as a function of @xmath30 is presented in figure [ fig : top_sen ] for two values of tan @xmath25 : 5 and 20 .
the top masses on the x - axis are allowed by the current bound obtained by a recent cdf measurement @xmath60 @xcite .
these curves illustrate the uncertainty in @xmath45 due to the present error in @xmath30 .
the neutralino mass bound , on the other hand , shows no strong @xmath30-dependence . throughout this paper
we shall use @xmath30 = 173 gev . in summary , while choosing the representative points for calculating the susy contribution to the di - photon invariant mass spectrum in the gmsb model , the @xmath45 from the weaker h - mass bound ( neutralino mass bound ) should be used for tan @xmath61 6.5 ( tan @xmath62 6.5 ) .
another theoretical uncertainty comes from the choice of the messenger scale m. in table [ m_variation ] we present the variation of the sparticle spectrum with m. we find that the variation of m within moderate ranges does not affect the spectrum and , hence , the collider signatures drastically . [ 0.90 ] .__variation of sparticle masses as a function of the messenger scale ( @xmath63 ) for fixed susy scale and @xmath64 in the minimal gmsb model . _ _ [ cols="^,^,^,^,^,^,^ " , ] of course a suitable cut can easily eliminate these unwanted events .
for example , the distribution in figure [ fig : eff ] suggests that a cut of @xmath65 300 can eliminate the susy contributions around the higgs mass peak , while the higgs signal and the sm backgrounds remain practically unaffected . before the cut
, however , the non - standard origin of the higgs boson will be revealed by the @xmath66 distribution .
the susy events around the higgs peak can also be eliminated by putting an upper cut on the @xmath0 . however , as already discussed the @xmath66 distribution provides a strong hint for the presence of new heavy particles , whereas large @xmath0 can come from massless particles and/or jet energy mis - measurements .
in this paper we analyze the invariant mass distribution of the di - photons to be minutely studied by the dedicated higgs search experiments in the di - photon + x channel at the lhc . in the sm only a tiny peak at @xmath29 above the sm background is expected . in gmsb models , in contrast , the di - photons stemming from squark - gluino events may distort the tail of the distribution far away from the higgs mass peak revealing the non - standard origin of the higgs boson .
in this paper we explore this possibility both at lhc 7 tev and 14 tev experiments by considering benchmark points in different gmsb models - minimal as well as non - minimal ( table 5 ) @xcite .
we also consider a simplified model recently employed by the cms collaboration to interpret their negative results for susy search in the ggmsb models and illustrate the above point through a detailed parameter space scan .
we first consider the minimal gmsb @xcite model subject to the lower bounds on @xmath29 from lep and on the lightest neutralino ( nlsp ) mass ( or equivalently on the scale @xmath23 ) from tevatron ( see figures [ fig : mh_tb ] and [ fig : scan ] ) .
the resulting lower bounds on the squark - gluino masses are so strong that di - photon events induced by sparticle production with tiny cross - sections are unlikely to affect the h - signal .
this has been illustrated with the help of two benchmark points ( bp i and ii , table [ bp_mgmsb ] ) in section 3 .
the @xmath38 and the cdf experiments actually constrain quantity @xmath67 , where @xmath68 is the total sparticle production cross section within the kinematic reach of the tevatron , br ( @xmath69 is the branching ratio underlying each single photon event . the bounds on @xmath23 or the nlsp mass in the mgmsb model are derived with the assumption that the above br is one . if br @xmath70 is significantly smaller than 1 , the limit on @xmath68 and , consequently , on @xmath23 or the nlsp mass become weaker .
this can happen in the mgmsb model ( table [ branching ] ) and , more importantly , in its generalizations ( table 6 ) . while extracting @xmath45 from the tevatron data in different models we have taken this possible reduction into account and have selected our benchmark points accordingly ( see table [ bp_nonminimal ] ) . in some of the models the total squark - gluino cross sections are significantly larger than the expected values in the mgmsb model and the relevant brs are not too small either ( see table [ cross ] ) .
more recently it has been shown that the distinct features of these phenomenological models can be incorporated into theoretically well - motivated models@xcite . 1 .
the di - photon + x events from sparticle production may change the shape of the tail of the @xmath71 invariant mass distributions which will be analyzed with utmost care by the lhc experiments .
this distortion was shown to be statistically significant in several benchmark scenarios of the non - minimal gmsb models for both lhc 7 tev ( table [ result_7 ] ) and 14 tev ( table [ result_14 ] ) experiments . in both cases
the traces of new physics beyond the higgs sector may show up at the early stages of the experiments with @xmath11 insufficient for the discovery of the h boson in the di - photon + x channel .
similar conclusions follow by analyzing a simplified model @xcite ( figures [ fig : reach7 ] and [ fig : reach14 ] ) .
other observables like the effective mass distribution of the di - photon events may also contain hints of new physics consisting of heavy particles ( figure [ fig : eff ] ) .
unexpected shapes of this distribution may disfavour not only the sm but i ) extensions of it with larger higgs sectors but no new heavy particles , ii ) models with other susy breaking mechanisms like msugra which do not have natural sources of di - photon events from sparticle production and iii ) the mgmsb model subject to the constraints from lep and tevatron .
3 . the squark - gluino induced @xmath71 events may contaminate the peak of the invariant mass distribution at @xmath29 ( see table [ inv_h ] ) . to improve the purity of the h - signal one may implement cuts on suitable kinematical variables like an upper cut on the effective mass ( see figure [ fig : eff ] ) or an upper cut on the missing energy which eliminate the sparticle induced events in the higgs peak .
99 l. susskind , ( 20,2619,1979 ) ; g. t hooft in _ recent developments in gauge theories _ , proceedings of the cargse summer institute , cargse , france , eds .
g. t hooft _ et al .
_ ; m. veltman , ( 12,437,1981 ) . h. p. nilles , phys .
* 110 * , 1 ( 1984 ) ; h. e. haber and g. l. kane , phys .
* 117 * , 75 ( 1985 ) ; s. p. martin , arxiv : hep - ph/9709356 ; m. dress , r. m. godbole , and p. roy , theory and phenomenology of sparticles ( world scientific , singapore , 2005 ) .
r. arnowitt , a. h. chamseddine and p. nath , ( 49,970,1982 ) ; r. barbieri , s. ferrara and c. a. savoy , ( 119,343,1982 ) ; l. j. hall , j. lykken and s. weinberg , ( 27,2359,1983 ) ; p.
nath , r. arnowitt and a. h. chamseddine , ( 227,121,1983 ) ; n. ohta , prog .
. phys . * 70 * , 542 ( 1983 ) .
p. fayet , phys .
lett . * b70 * , 461 ( 1977 ) ; m. dine , w. fischler and m. srednicki , ( 189,575,1981);s .
dimopoulos and s. raby,(192,353,1981);m .
dine and a. nelson , ( 48,1277,1993);m .
dine , a. e. nelson , y. shirman , phys .
* d51 * , 1362 - 1370 ( 1995 ) ; m. dine , a. e. nelson , y. nir and y. shirman , phys .
d * 53 * , 2658 ( 1996 ) .
s. dimopoulos , s. d. thomas , j. d. wells , nucl .
* b488 * , 39 - 91 ( 1997 ) .
a. djouadi , j. l. kneur and g. moultaka , comput .
commun . * 176 * , 426 ( 2007 ) .
f. e. paige , s. d. protopopescu , h. baer and x. tata , arxiv : hep - ph/0312045 .
for online version see http://www.phys.ufl.edu/ jblender / isajet / isajet.html j. a. bagger , k. t. matchev , d. m. pierce , r. -j .
zhang , phys .
rev . * d55 * , 3188 - 3200 ( 1997 ) ; h. baer , m. brhlik , c. -h .
chen , x. tata , phys .
rev . * d55 * , 4463 - 4474 ( 1997 ) ; h. baer , p. g. mercadante , x. tata , y. -l .
wang , phys .
rev . * d60 * , 055001 ( 1999 ) ; j. l. diaz - cruz , d. k. ghosh , s. moretti , phys .
* d68 * , 014019 ( 2003 ) .
r. barate _
[ lep working group for higgs boson searches and aleph and delphi and l3 and opal collaborations ] , phys .
* b565 * , 61 - 75 ( 2003 ) . s. heinemeyer , w. hollik and g. weiglein , ( 425,265,2006 ) , s. heinemeyer , int .
j. mod .
a * 21 * , 2659 ( 2006 ) .
v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
lett . * 105 * , 221802 ( 2010 ) .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) [ arxiv : hep - ph/0603175 ] . w. beenakker , r. hopker and m. spira , arxiv : hep - ph/9611232 .
please see see http://www.thphys.uni-heidelberg.de/plehn/prospino/ or http://people.web.psi.ch/spira/prospino/ k. i. izawa , y. nomura , k. tobe and y. yanagida , ( 56,2886,1997);r .
kitano , h. ooguri and y. ookouchi , ( 75,045022,2007);c .
csaki , y. shirman and j. terning , ( 0705,099,2007);s .
abel , c. durnford , j. jaeckel and v. v. khoze , ( 661,201,2008 ) .
l. m. carpenter , arxiv:0812.2051 ( hep - ph ) ; a. rajaraman , y. shirman , j. smidt and f. yu , ( 678,367,2009 ) ; j. d. mason , d. e. morrissey , d. poland , phys .
rev . * d80 * , 115015 ( 2009 ) ; p. meade , m.reece , d. shih , ( 1005,105,2010),(1010,067,2010);j .
t.ruderman and d. shih , ( 1011,046,2010);s .
a. abel , m. j. dolan , j. jaekel and v. v. khoze,(0912,001,2009 ) ; r. sato and s. shirai , phys .
b * 692 * , 126 ( 2010 ) ; j. jaeckel , v. v. khoze and c. wymant , jhep * 1104 * , 126 ( 2011 ) ; s. biswas , j. chakrabortty and s. roy , arxiv:1010.0949 [ hep - ph ] ; t.ruderman and d. shih , arxiv:1103.6083 ; m. j. dolan , d. grellscheid , j. jaeckel , v. v. khoze , p. richardson , [ arxiv:1104.0585 [ hep - ph ] ] . | the invariant mass distribution of the di - photons from the decay of the lighter scalar higgs boson(h ) to be carefully measured by dedicated h search experiments at the lhc may be distorted by the di - photons associated with the squark - gluino events with much larger cross sections in gauge mediated supersymmetry breaking ( gmsb ) models .
this distortion if observed by the experiments at the large hadron collider at 7 tev or 14 tev , would disfavour not only the standard model but various two higgs doublet models with comparable h - masses and couplings but without a sector consisting of new heavy particles decaying into photons .
the minimal gmsb ( mgmsb ) model constrained by the mass bound on h from lep and that on the lightest neutralino from the tevatron , produce negligible effects .
but in the currently popular general gmsb(ggmsb ) models the tail of the above distribution may show statistically significant excess of events even in the early stages of the lhc experiments with integrated luminosity insufficient for the discovery of h. we illustrate the above points by introducing several benchmark points in various gmsb models - minimal as well as non - minimal .
the same conclusion follows from a detailed parameter scan in a simplified ggmsb model recently employed by the cms collaboration to interpret their searches in the di - photon + @xmath0 channel .
other observables like the effective mass distribution of the di - photon + x events may also reveal the presence of new heavy particles beyond the higgs sector .
the contamination of the h mass peak and simple remedies are also discussed .
= 100000 tifr / th/11 - 16 + iiser / hep/02/11 + * revealing the footprints of squark gluino production through higgs search experiments at the large hadron collider at 7 tev and 14 tev * 0.4 cm biplob bhattacherjee@xmath1 and amitava datta@xmath2 0.1 cm _
@xmath1 department of theoretical physics , tata institute of fundamental research , + 1 , homi bhabha road , mumbai 400 005 , india .
+ _ @xmath2 indian institute of science education and research , kolkata , + mohanpur campus , po : bckv campus main office , + mohanpur - 741252 , india . + _ _ pacs no : 14.80.da , 12.60.jv |
we start by collecting some classical elliptic estimates and their applications to eigenfunctions .
first , the general sobolev estimate : let @xmath39 where @xmath40 with @xmath41 .
then there exists @xmath42 : @xmath43 next , we recall the bernstein gradient estimates : @xcite local eigenfunctions of a riemannian manifold satisfy : 1 .
@xmath44 bernstein estimate : @xmath45 2 .
@xmath22 bernstein estimate : there exists @xmath46
so that @xmath47 for @xmath48 another well - known estimate is the doubling estimate : [ double ] ( donnelly - fefferman , lin ) and @xcite ( lemma 6.1.1 ) let @xmath12 be a global eigenfunction of a @xmath26 @xmath1 there exists @xmath49 and @xmath50 such that for @xmath51 , @xmath52 and suppose that @xmath53 . by the vanishing order @xmath54 of @xmath55 at @xmath56 is meant the largest positive integer such that @xmath57 for all @xmath58 .
suppose that @xmath28 is compact and of dimension @xmath2 .
then there exist constants @xmath59 depending only on the dimension such that the the vanishing order @xmath54 of @xmath55 at @xmath60 satisfies @xmath61 for all @xmath62 . in the case of a global eigenfunction , @xmath63 we now recall quantitative lower bound estimates . they follow from doubling estimates and also from carleman inequalities .
[ dfed ] suppose that @xmath28 is compact and that @xmath12 is a global eigenfunction , @xmath64 .
then for all @xmath65 , there exist @xmath66 so that @xmath67 local lower bounds on @xmath68 follow from doubling estimates .
they imply that there exists @xmath69 so that , for any @xmath70 , @xmath71 indeed , there of course exists a point @xmath72 so that @xmath73 .
any point of @xmath74 can be linked to this point by a smooth curve ofuniformly bounded length .
we then choose @xmath75 sufficiently small so that the @xmath75-tube around the curve lies in @xmath76 and link @xmath77 to @xmath78 by a chain of @xmath75-balls in @xmath76 where the number of links in the chain is uniformly bounded above as @xmath79 varies in @xmath76 . if the balls are denoted @xmath80 we have @xmath81 since @xmath82 .
the growth estimate implies that for any ball @xmath83 , @xmath84 .
since the number of balls is uniformly bounded , @xmath85 and we get a contradiction if no such @xmath86 exists . as an illustration ,
gaussian beams such as highest weight spherical harmonics decay at a rate @xmath87 away from a stable elliptic orbit @xmath88 .
hence if the closure of an open set is disjoint from @xmath88 , one has a uniform exponential decay rate which saturate the lower bounds .
the classical sogge estimates state that , for any compact riemannian manifold of dimension @xmath2 , we have @xmath93 where @xmath94 since we often use surfaces as an illustrantion , we note that in dimension @xmath95 one has for @xmath96 , @xmath97 and @xmath98 these estimates are also sharp for the round sphere @xmath99 . the first estimate , ,
is saturated by highest weight spherical harmonics .
the second estimate , , is sharp due to the zonal functions on @xmath99 , which concentrate at points .
we go over these examples in [ s2 ] .
we begin the survey by stating some of the principal problems an results regarding nodal sets and more general level sets . some of the problems are intentionally stated in vague terms that admit a number of rigorous formulations .
one of the principal problems on nodal sets is to measure their hypersurface volume . in the real analytic case , donnelly - fefferman ( @xcite ( see also @xcite ) ) proved : [ df ] let @xmath1 be a compact real analytic riemannian manifold , with or without boundary .
then there exist @xmath100 depending only on @xmath1 such that @xmath101 the bounds were conjectured by s. t. yau @xcite for all @xmath26 @xmath102 , but this remains an open problem .
the lower bound was proved for all @xmath26 metrics for surfaces , i.e. for @xmath103 by brning @xcite . for general @xmath26 metrics the sharp upper and lower bounds
are not known , although there has been some recent progress that we consider below .
the nodal hypersurface bounds are consistent with the heuristic that @xmath12 is the analogue on a riemannian manifold of a polynomial of degree @xmath21 , since the hypersurface volume of a real algebraic variety is bounded by its degree .
the equidistribution problem for nodal sets is to study the behavior of the integrals of general continuous functions @xmath104 over the nodal set . here
, we normalize the delta - function on the nodal set by the conjectured surface volume of [ area ] . more precisely : * problem * find the weak * limits of the family of measures @xmath105 .
note that in the @xmath26 case we do not even know if this family has uniformly bounded mass .
the high - frequency limit is the semi - classical limit and generally signals increasing complexity in the ` topography ' of eigenfunctions .
heuristics from quantum chaos suggests that eigenfunctions of quantum chaotic systems should behave like random waves .
the random wave model is defined and studied in @xcite ( see [ rwonb ] ) , and it is proved ( see theorem [ zdsm ] ) that if one picks a random sequence @xmath106 of random waves of increasing frequency , then almost surely @xmath107 i.e. their nodal sets become equidistributed with respect to the volume form on @xmath28 .
hence the heuristic principle leads to the conjecture that nodal sets of eigenfunctions of quantum chaotic systems should become equidistributed according to the volume form .
the conjecture for eigenfunctions ( rather than random waves ) is far beyond any current techniques and serves mainly as inspiration for studies of equidistribution of nodal sets . a yet more speculative conjecture in quantum chaosis that the nodal sets should tend to @xmath108 curves in critical percolation .
cle refers to conformal loop ensembles , which are closed curves related to @xmath109 curves .
as above , this problem is motivated by a comparision to random waves , but for these the problem is also completely open . in [ perc ]
we review the heuristic principles which started in condensed matter physics @xcite before migrating to quantum chaos @xcite .
it is dubious that such speculative conjectures can be studied rigorously in the forseeable future , but we include them to expose the reader to the questions that are relevant to physicists . besides nodal sets it is of much current interest to study @xmath0 norms of eigenfunctions globally on @xmath1 and also of their restrictions to submanifolds .
in fact , recent results show that nodal sets and @xmath0 norms are related .
for instance , in [ lb ] we will use the identity @xmath111 relating the @xmath110 norm of @xmath12 to a weighted integral over @xmath18 to obtain lower bounds on @xmath112 see . obtaining lower bounds on @xmath110 norms of eigenfunctions
is closely related to finding upper bounds on @xmath113 norms .
the current bounds are nowhere near sharp enough to improve nodal set bounds .
a closely related problem in the ` topography ' of laplace eigenfunctions @xmath12 is to determine the asymptotic distribution of their critical points @xmath114 this problem is analogous to that of measuring the hypersurface area @xmath115 of the nodal ( zero ) set of @xmath12 , but it is yet more complicated due to the instability of the critical point set as the metric varies . for a generic metric ,
all eigenfunctions are morse functions and the critical point set is discrete . one may ask to count the number of critical points asymptotically as @xmath16 . but there exist metrics ( such as the flat metric on the torus , or the round metric on the sphere ) for which the eigenfunctions have critical manifolds rather than points . to get around this obstruction
, we change
the problem from counting critical points to counting critical values @xmath116 since a real analytic function on a compact real analytic manifold has only finitely many critical values , eigenfunctions of real analytic riemannian manifolds @xmath1 have only finitely many critical values and we can ask to count them . moreover for generic real analytic metrics , all eigenfunctions are morse functions and there exists precisely one critical point for each critical value .
thus , in the generic situation , counting critical values is equivalent to counting critical points . to our knowledge , there are no results on this problem , although it is possible to bound the @xmath14-measure of @xmath117 ( see theorem @xcite ) .
however @xmath118 in the generic case and in special cases where it is not zero the method is almost identical to bounds on the nodal set .
thus , such results bypass all of the difficulties in counting critical values .
we will present one new ( unpublished ) result which generalizes eqrefid to critical points .
but the resulting identity is much more complicated than for zeros .
singular points are critical points which occur on the nodal sets .
we recall ( see @xcite ) that the the singular set @xmath119 satisfies @xmath120 .
thus , outside of a codimension one subset , @xmath18 is a smooth manifold , and the riemannian surface measure @xmath121 on @xmath18 is well - defined .
we refer to @xcite for background .
it is known that in dimension two , the minimal possible area of a nodal domain of a euclidean eigenfunction is @xmath122 .
this follows from the two - dimensional faber - krahn inequality , @xmath123 where @xmath124 is a nodal domain in @xmath125 . in higher dimensions
, the faber - krahn inequality shows that on any riemannian manifold the volume of any nodal domain is @xmath126 @xcite .
another size measure of a nodal domain is its inradius @xmath127 , i.e. the radius of the largest ball contained inside the nodal domain .
as can be seen from computer graphics ( see e.g. @xcite ) , there are a variety of ` types ' of nodal components . in @xcite
, mangoubi proves that @xmath128 where @xmath129 ; note that eigenvalues in @xcite are denoted @xmath21 while here we denote them by @xmath37 . in dimension
@xmath95 , it is known ( loc.cit . ) that @xmath130 there are two natural decompositions ( partitions ) of @xmath28 associated to an eigenfunction ( or any smooth function ) . *
( i ) * nodal domain decomposition . first is the decomposition of @xmath28 into nodal domains of @xmath12 . as in @xcite
we denote the collection of nodal domains by @xmath131 and denote a nodal domain by @xmath86 .
thus , @xmath132 when @xmath133 is a regular value of @xmath12 the level sets are smooth hypersurfaces and one can ask how many components of @xmath18 occur , how many components of the complement , the topological types of components or the combinatorics of the set of domains .
when @xmath133 is a singular value , the nodal set is a singular hypersurface and can be connected but one may ask similar questions taking multiplicities of the singular points into account . to be precise ,
let @xmath134 the best - known problem is to estimate @xmath135 . according to the courant nodal domain theorem , @xmath136 . in the case of spherical harmonics , where many orthonormal bases are possible , it is better to estimate the number in terms of the eigenvalue , and the estimate has the form @xmath137 where @xmath138 and @xmath139 is a constant depending on @xmath33 . in dimension 2
, pleijel used the faber - krahn theorem to improve the bound to @xmath140 where @xmath141 is the smallest zero of the @xmath142 bessel function .
a wide variety of behavior is exhibited by spherical harmonics of degree @xmath143 .
we review the definitions below .
the even degree harmonics are equivalent to real projective plane curves of degree @xmath144 .
but each point of @xmath145 corresponds to a pair of points of @xmath99 and at most one component of the nodal set is invariant under the anti - podal map . for other components ,
the anti - podal map takes a component to a disjoint component .
thus there are essentially twice the number of components in the nodal set as components of the associated plane curve .
as discussed in @xcite , one has ; * harnack s inequality : the number of components of any irreducible real projective plane curve is bounded by @xmath146 where @xmath33 is the genus of the curve . *
if @xmath147 is a real projectove plane curve of degree @xmath143 then its genus is given by noether s formula @xmath148 where @xmath149 is the order of vanishing of @xmath12 at @xmath150 .
thus , the number of components is @xmath151 for a non - singular irreducible plane curve of degree @xmath143 .
curves which achieve the maximum are called @xmath28-curves .
also famous are harnack curves , which are @xmath28 curves for which there exist three distinct lines @xmath152 of @xmath145 and three distinct arcs @xmath153 of the curve on one component so that @xmath154 .
it follows from pleijel s bound that nodal sets of spherical harmonics can not be maximal for large @xmath143 , since half of the pleijel bound is roughly @xmath155 which is below the threshold @xmath156 for maximal curves .
associated to the collection of nodal domains is its incidence graph @xmath157 , which has one vertex for each nodal domain , and one edge linking each pair of nodal domains with a common boundary component .
here we assume that @xmath133 is a regular value of @xmath12 so that the nodal set is a union of embedded submanifolds .
the euler characteristic of the graph is the difference beween the number of nodal domains and nodal components . in the non - singular case
, one can convert the nodal decomposition into a cell decomposition by attaching a one cell between two adjacent components ,
and then one has @xmath158 ( see lemma 8 of @xcite ) .
the possible topological types of arrangements of nodal components of spherical harmonics is studied in @xcite .
they prove that for any @xmath159 with @xmath160 even and for every set of @xmath161 disjoint closed curves whose union is invariant with respect to the antipodal map , there exists an eigenfunction whose nodal set has the topological type of the union of curves . note that these spherical harmonics have relatively few nodal domains compared to the pleijel bound .
it is proved in @xcite that random spherical harmonics have @xmath162 nodal components for some ( undetermined ) @xmath163 . * morse - smale decomposition * for generic metrics , all eigenfunctions are morse functions @xcite .
suppose that @xmath164 is a morse function .
for each critical point @xmath147 let @xmath165 ( the stable or descending cell through p ) denote the union of the gradient flow lines which have @xmath147 as their initial point , i.e. their @xmath166-limit point .
then @xmath167 is a cell of dimension @xmath168 number of negative eigenvalues of @xmath169 . by the morse - smale decomposition
we mean the decomposition @xmath170 it is not a good cell decomposition in general . if we change @xmath104 to @xmath171 we get the decomposition into ascending ( unstable ) cells @xmath172 if the intersections @xmath173 are always transversal then @xmath174 is said to be transversal . in this case @xmath175 and the number of gradient curves joining two critical points whose morse index differs by 1 is finite .
we are mainly interested in the stable cells of maximum dimension , i.e. basins of attraction of the gradient flow to each local minimum .
we then have the partition @xmath176 this decomposition is somtimes used in condensed matter physics ( see e.g. @xcite ) and in computational shape analysis @xcite . in dimension two , the surface is partitioned into ` polygons ' defined by the basins of attraction of the local minima of @xmath177 .
the boundaries of these polygons are gradient lines of @xmath177 which emanate from saddle points .
the vertices occur at local maxima .
an eigenfunction is a neumann eigenfunction in each basin since the boundary is formed by integral curves of @xmath178 .
possibly it is ` often ' the first non - constant neumann eigenfunction ( analogously to @xmath12 being the lowest dirichlet eigenfunction in each nodal domain ) , but this does not seem obvious . hence it is not clear how to relate the global eigenvalue @xmath37 to the neumann eigenvalues of the basins , which would be useful in understanding the areas or diameters of these domains .
note that @xmath179 where @xmath180 is the unit normal to @xmath181 , since @xmath182 is tangent to the boundary .
in particular , the intersection @xmath183 is non - empty and is a connected hypersurface which separates @xmath184 into two components on which @xmath12 has a fixed sign . to our knowledge , there do not exist rigorous results bounding the number of local minima from above or below , i.e. there is no analogue of the courant upper bound for the number of local minima basins .
it is possible to obtain statstical results on the asymptotic expected number of local minima , say for random spherical harmonics of degree @xmath143 .
the methods of @xcite adapt to this problem if one replaces holomorphic szeg kernels by spectral projections ( see also @xcite . ) thus , in a statistical sense it is much simpler to count the number of neumann domains " or morse - smale basins than to count nodal domains as in @xcite .
before proceeding to rigorous results , we go over a number of explicitly solvable examples . almost by definition , they are highly non - generic and in fact represent the eigenfunctions of quantum integrable systems . aside from being explicitly solvable ,
the eigenfunctions of this section are extremals for a number of problems .
the basic real valued eigenfunctions are @xmath185 or @xmath186 ( @xmath187 ) on the flat torus @xmath188 .
the zero set consists of the hyperplanes @xmath189 mod @xmath190 or in other words @xmath191 .
thus the normalized delta function @xmath192 tends to uniform distribution along rays in the lattice @xmath193 .
the lattice arises as the joint spectrum of the commuting operators @xmath194 and is a feature of quantum integrable systems .
the critical point equation for @xmath186 is @xmath195 and is thus the same as the nodal equation . in particular ,
the critical point sets are hypersurfaces in this case .
there is just one critical value @xmath196 . instead of the square torus
we could consider @xmath197 where @xmath198 is a lattice of full rank .
then the joint spectrum becomes the dual lattice @xmath199 and the eigenfunctions are @xmath200 with @xmath201 . the real eigenspace @xmath202 and other rational lattices . in that case
, one may take linear combinations of the basic eigenfunctions and study their nodal and critcal point sets . for background , some recent results and further references we refer to @xcite .
the spectral decomposition for the laplacian is the orthogonal sum of the spaces of spherical harmonics of degree @xmath143 , @xmath203 the eigenvalues are given by @xmath204 and the multiplicities are given by @xmath205 . a standard basis is given by the ( complex valued ) spherical harmonics @xmath206 which transform by @xmath207 under rotations preserving the poles . the @xmath206 are complex valued , so we study the nodal sets of their real and imaginary parts .
they are separable , i.e. factor as @xmath208 ( resp .
@xmath209 where @xmath210 is an associated legendre function .
thus the nodal sets of these special eigenfunctions form a checkerboard pattern that can be explicitly determined from the known behavior of zeros of associated legendre functions .
see the first image in the illustration below
. among the basic spherical harmonics , there are two special ones : the zonal spherical harmonics ( i.e. the rotationally invariant harmonics ) and the highest weight spherical harmonics .
their nodal sets and intensity plots are graphed in the bottom two images , respectively .
since the zonal spherical harmonics @xmath211 on @xmath99 are real - valued and rotationally invariant , their zero sets consist of a union of circles , i.e. orbits of the @xmath212 rotation action around the third axis .
it is well known that @xmath213 , where @xmath214 is the @xmath143th legendre function and the normalizing constant is chosen so that @xmath215 , i.e. @xmath216 so that @xmath217 . all zeros of @xmath218 are real and it has @xmath143 zeros in @xmath219 $ ] .
it is classical that the zeros @xmath220 of @xmath221 in @xmath222 become uniformly distributed with respect to @xmath223 @xcite .
it is also known that @xmath214 has @xmath224 distinct critical points @xcite and so the critical points of @xmath225 is a union of @xmath224 lattitude circles .
we now consider real or imaginary parts of highest weight spherical harmonics @xmath226 . up to a scalar multiple , @xmath227 as a harmonic polynomial on @xmath228 .
it is an example of a gaussian beams along a closed geodesic @xmath88 ( such as exist on equators of convex surfaces of revolution ) .
see @xcite for background on gaussian beams on riemannian manifolds .
the real and imaginary parts are of the form @xmath229 where @xmath230 is a constant multiple of @xmath231 so @xmath232 the factors @xmath233 have @xmath143 zeros on @xmath234 .
the legendre funtions satisfy the recursion relation @xmath235 with @xmath236 and therefore have no real zeros away from the poles .
thus , the nodal set consists of @xmath143 circles of longitude with equally spaced intersections with the equator .
the critical points are solutions of the pair of equations @xmath237 .
since @xmath238 has no zeros away from the poles , the second equation forces the zeros to occur at zeros of @xmath239 .
but then @xmath240 so the zeros must occur at the zeros of @xmath241 .
the critical points only occur when @xmath242 or @xmath243 on @xmath222
. there are critical points at the poles where @xmath244 vanishes to order @xmath143 and there is a local maximum at the value @xmath245 of the equator .
thus , @xmath246 has @xmath144 isolated critical points on the equator and multiple critical points at the poles .
we note that @xmath247 is a gaussian bump with peak along the equator in the radial direction .
its radial gaussian decay implies that it extremely small outside a @xmath248 tube around the equator .
the complement of this tube is known in physics as the classically forbidden region .
we see that the nodal set stretches a long distance into the classically forbidden region .
this creates problems for nodal estimates since exponentially small values ( in terms of the eigenvalue ) are hard to distinguish from zeros . on the other hand
, it has only two ( highly multiple ) critical points away from the equator . the examples above exhibit quite disparate behavior but all are eigenfunctions of quantum integrable systems .
we do not review the general results in this case but plan to treat this case in an article in preparation @xcite .
we now contrast the nodal set behavior with that of random spherical harmonics ( left ) and a chaotic billiard domain ( the graphics are due to e. j. heller ) .
in this section we review some recent lower bounds on @xmath38 from @xcite .
before we begin , we recall the co - area formula : let @xmath164 be lipschitz .
then for any continuous function @xmath55 on @xmath28 , @xmath249 equivalently , @xmath250 we refer to @xmath251 as the leray form " on the level set @xmath252 . unlike the riemannian surface measure @xmath253 it depends on the choice of defining function @xmath104 .
the surface measures are related by @xmath254 for background , see theorem 1.1 of @xcite .
[ theorem ] as above assume that @xmath255 .
then @xmath256 for some uniform constant @xmath257 .
consequently , @xmath258 inequality follows from and the lower bounds in @xcite @xmath259 the lower bound was first proved by colding and minicozzi @xcite .
a slightly weaker result was proved in @xcite by a different method inspired by the article of r. t. dong @xcite which was successively improved in @xcite to the same bound as @xcite .
the @xmath110-lower bounds in were proved in @xcite using hlder s inequality and the @xmath0 eigenfunction estimates of sogge @xcite for the range where @xmath260 . as will be seen below
, the estimate is sharp for gaussian beams such as highest weight spherical harmonics .
the random wave model would predict that @xmath261 and that would imply yau s conjectured lower bound in the ` chaotic case ' .
this provides ample motivation to study @xmath110 norms of eigenfunctions on manifolds with ergodic geodesic flow .
the proof of theorem [ theorem ] is based on an identity from @xcite ( inspired by an identity in @xcite ) : @xmath262 we recall that @xmath263 is the riemannian surface measure on @xmath18 . in @xcite , we substituted @xmath264 and used to obtain a power lower bound .
this was improved in @xcite by putting @xmath265 but applying schwarz s inequality to get @xmath266 the final important step is to show that @xmath267 by choosing @xmath268 from it follows that @xmath269 by the @xmath270-sobolev bounds @xmath271 it follows that the second term " of the right side satisfies @xmath272 and thus to prove , it suffices to show that the first terms " satisfies @xmath273 we refer to @xcite for further details on this bound .
a simpler approach was suggested by w. minicozzi , who pointed out that also follows from the identity @xmath274 this approach is used in @xcite to generalize the nodal bounds to dirichlet and neumann eigenfunctions of bounded domains . in the next setction
we explain how to obtain more general identities .
there are several ways to prove the identity in .
one way to see it is that @xmath275 away from @xmath276 .
hence this distribution is a positive measure supported on @xmath277 to determine the coefficient of the surface measure @xmath263 we calculate the limit as @xmath278 of the integral latexmath:[\[\int_m f ( \delta + \lambda^2 ) |{\varphi}_{\lambda}| dv = \int_{|{\varphi}_{\lambda}| \leq \delta } f ( \delta + \lambda^2 ) no loss of generality we may assume that @xmath75 is a regular value of @xmath12 ( by sard s theorem ) . by the gauss - green theorem ,
@xmath280 here , @xmath180 is the outer unit normal and @xmath281 is the associated directional derivative . for @xmath282 , we have @xmath283 letting @xmath278 ( through the sequence of regular values ) we get @xmath284 since @xmath285 on @xmath286 and by ( [ normal ] ) , we see that @xmath287 the gauss - green formula and limit are justified by the fact that the singular set @xmath288 has codimension two .
we refer to @xcite for further details .
the @xmath110 lower bound of follows from eigenfunction estimates in @xcite , which say that @xmath289 if we pick such a @xmath290 , then by hlder s inequality , we have @xmath291 which implies @xmath292 , since @xmath293 .
one can also integrate the identity over a basin of attraction of a local minimum ( or maximum ) , since the boundary term vanishes .
thus we get an identity between the @xmath110 norm of @xmath12 on each basin and the @xmath294-measure of the nodal line inside the basin . for any function @xmath295
, we have @xmath296 we then take @xmath295 to be the meromorphic family of homogeneous distribution @xmath297 .
we recall that for @xmath298 , @xmath299 the family extends to @xmath300 as a meromorphic family of distributions with simple poles at @xmath301 using the equation @xmath302 to extend it one unit strip at a time .
one can convert @xmath297 to the holomorphic family @xmath303 the identity we used above belongs to the family , @xmath304 here @xmath305 has poles at @xmath306 .
the calculation in used @xmath307 but is equivalent to using when @xmath308 . then @xmath309 has a pole when @xmath308 with residue @xmath310 ; it is cancelled by the factor @xmath311 and we obtain . this calculation is formal because the pullback formulae are only valid when @xmath312 when @xmath313 , but as above they can be justified because the singular set has codimension 2 .
the right side also has a pole at @xmath314 and we get @xmath315 , which is equivalent to the divergence identity above . there are further poles at @xmath316 but they now occur on both sides of the formulae .
it is possible that they have further uses .
the question arises of how such identities are related to the bernstein - kashiwara theorem that for any real analytic function @xmath104 one may meromorphically extend @xmath317 to @xmath318 by constructing a family @xmath319 of differential operators with analytic coefficients and a meromorphic function @xmath320 so that @xmath321 in the case @xmath322 , the operator @xmath323 accomplishes something like this , but it does not have analytic coefficients due to poles at the critical points of @xmath177 .
one wonders what @xmath324 might be when @xmath322 .
these results generalize easily to any level set @xmath325 .
let @xmath326 .
[ boundsc ] for any @xmath26 riemannian manifold , and any @xmath327 we have ,
@xmath328 this identity has similar implications for @xmath329 and for the equidistribution of level sets .
note that if @xmath330 then indeed both sides are zero .
[ cintro ] for @xmath331 @xmath332 consequently , if @xmath333 @xmath334 the corollary follows by integrating @xmath8 by parts , and by using the identity , @xmath335 since @xmath336 .
the lower bound of theorem [ theorem ] is far from the lower bound conjectured by yau , which by theorem [ df ] is correct at least in the real analytic case . in this section
we go over the model examples to understand why the methds are not always getting sharp results .
we have , @xmath337 .
since @xmath338 when @xmath339 the integral is simply @xmath340 times the surface volume of the nodal set , which is known to be of size @xmath340 . also , we have @xmath341 .
thus , our method gives the sharp lower bound @xmath342 in this example .
so the upper bound is achieved in this example .
also , we have @xmath341 .
thus , our method gives the sharp lower bound @xmath342 in this example .
since @xmath338 when @xmath339 the integral is simply @xmath340 times the surface volume of the nodal set , which is known to be of size @xmath340 .
the @xmath110 of @xmath225 norm can be derived from the asymptotics of legendre polynomials @xmath343 where the remainder is uniform on any interval @xmath344 .
we have @xmath345 i.e. the @xmath110 norm is asymptotically a positive constant . hence @xmath346 . in this example
@xmath347 saturates the sup norm bound .
so the estimate of produces the lower bound @xmath348 .
the accurate lower bound is @xmath21 , as one sees from the rotational invariance and by the fact that @xmath214 has @xmath143 zeros .
the defect in the argument is that the bound @xmath349 is only obtained on the nodal components near the poles , where each component has length @xmath350 .
* gaussian beams * gaussian beams are gaussian shaped lumps which are concentrated on @xmath351 tubes @xmath352 around closed geodesics and have height @xmath353 .
we note that their @xmath110 norms decrease like @xmath354 , i.e. they saturate the @xmath0 bounds of @xcite for small @xmath147 . in such cases we have latexmath:[$\int_{z_{{\varphi}_{\lambda } } } is likely that gaussian beams are minimizers of the @xmath110 norm among @xmath270-normalized eigenfunctions of riemannian manifolds . also , the gradient bound @xmath356 is far off for gaussian beams , the correct upper bound being @xmath357 .
if we use these estimates on @xmath358 and @xmath359 , our method gives @xmath360 , while @xmath21 is the correct lower bound for gaussian beams in the case of surfaces of revolution ( or any real analytic case ) .
the defect is again that the gradient estimate is achieved only very close to the closed geodesic of the gaussian beam .
outside of the tube @xmath352 of radius @xmath361 around the geodesic , the gaussian beam and all of its derivatives decay like @xmath362 where @xmath363 is the distance to the geodesic .
hence @xmath364 applying the gradient bound for gaussian beams to the latter integral gives @xmath365 , which is sharp since the intersection @xmath366 cuts across @xmath88 in @xmath367 equally spaced points ( as one sees from the gaussian beam approximation ) . in this section , we prove a rather simple ( unpublished ) result on nodal sets when the geodesic flow of @xmath1 is ergodic . since there exist many expositions of quantum ergodic eigenfunctions , we only briefly recall the main facts and definitions and refer to @xcite for further background .
quantum ergodicity concerns the semi - classical ( large @xmath21 ) asymptotics of eigenfunctions in the case where the geodesic flow @xmath30 of @xmath1 is ergodic .
we recall that the geodesic flow is the hamiltonian flow of the hamiltonian @xmath368 ( the length squared ) and that ergodicity means that the only @xmath30-invariant subsets of the unit cosphere bundle @xmath369 have either full liouville measure or zero liouville measure ( liouville measure is the natural measure on the level set @xmath370 induced by the symplectic volume measure of @xmath31 ) .
we will say that a sequence @xmath371 of @xmath270-normalized eigenfunctions is _ quantum ergodic _
if @xmath372 here , @xmath373 denotes the space of pseudodifferential operators of order @xmath374 , and @xmath375 denotes liouville measure on the unit cosphere bundle @xmath29 of @xmath1 .
more generally , we denote by @xmath376 the ( surface ) liouville measure on @xmath377 , defined by @xmath378 we also denote by @xmath166 the canonical action @xmath379-form of @xmath380 .
the main result is that there exists a subsequence @xmath371 of eigenfunctions whose indices @xmath381 have counting density one for which @xmath382 ( where as above @xmath383 is the normalized liouville average of @xmath384 ) .
such a sequence of eigenfunctions is called a sequence of ` ergodic eigenfunctions ' .
the key quantities to study are the quantum variances @xmath385 the following result is the culmination of the results in @xcite .
[ qe ] let @xmath102 be a compact riemannian manifold ( possibly with boundary ) , and let @xmath386 be the spectral data of its laplacian @xmath387 then the geodesic flow @xmath30 is ergodic on @xmath388 if and only if , for every @xmath389 , we have : 1 .
@xmath390 2 .
@xmath391 since all the terms in ( 1 ) are positive , no cancellation is possible , hence ( 1 ) is equivalent to the existence of a subset @xmath392 of density one such that @xmath393 has only @xmath394 as a weak * limit point .
we now consider nodal sets of quantum ergodic eigenfunctions .
the following result says that if we equip nodal sets with the measure @xmath395 , then nodal sets can not ` scar ' , i.e. concentrate singularly as @xmath396 .
[ l1rega ] suppose that @xmath397 is a quantum ergodic sequence .
then any weak limit of @xmath398 must be absolutely continuous with respect to @xmath399 .
we recall that , for any @xmath400 , @xmath401 the identity for general @xmath400 can be used to investigate the equidistribution of nodal sets equipped with the surface measure @xmath402 .
we denote the normalized measure by @xmath403the weak * limits of the sequence @xmath404 of bounded positive measures are the same as the weak * limits of @xmath405 ( against @xmath327 . ) we let @xmath400 and multiply the identity ( [ dongtype ] ) by @xmath406 .
we then integrate by parts to put @xmath8 on @xmath104 .
this shows that for @xmath407 , we have @xmath408 letting @xmath264 , we see that the family of measures @xmath409 is bounded . by uniform approximation of @xmath410 by elements of @xmath411
, we see that the weak * limit formula extends to @xmath412 .
[ l1reg ] suppose that @xmath397 is a quantum ergodic sequence .
then any weak limit of @xmath413 must be absolutely continuous with respect to @xmath399 .
we recall that a sequence of measures @xmath414 converges weak * to @xmath415 if @xmath416 for all continuous @xmath104 .
a basic fact about weak * convergence of measures is that @xmath417 for all @xmath327 implies that @xmath418 for all sets @xmath419 with @xmath420 ( portmanteau theorem ) .
we also recall that a sequence of eigenfunctions is called quantum ergodic ( in the base ) if @xmath421 in the weak * topology , i.e. the vague topology on measures .
we now prove lemma [ l1reg ] .
suppose that @xmath422 and assume that @xmath423 where @xmath424 is singular with respect to @xmath399 .
let @xmath425 , and let @xmath426 .
let @xmath427 be the @xmath428-tube around @xmath429 .
then @xmath430 but for any set @xmath431 , @xmath432 hence if @xmath433 , @xmath434 . letting @xmath435 we get @xmath436 since @xmath437 .
letting @xmath438 gives a contradiction .
of course , it is possible that the only weak * limit is zero . to our knowledge
, the question whether the limit ( [ qe ] ) holds @xmath439 when @xmath102 has ergodic geodesic flow has not been studied .
it is equivalent to strengthening the portmanteau statement to all measurable sets @xmath419 , and is equivalent to the statement that @xmath440 weakly in @xmath110 .
we call such sequences @xmath22 quantum ergodic on the base .
the term ` on the base ' refers to the fact that we only demand quantum ergodicity for the projections of the ` microlocal lifts ' to the base @xmath28 .
for instance , the exponential eigenfunctions of flat tori are @xmath22 quantum ergodic in this sense .
[ eg ] suppose that @xmath3 is an @xmath22- quantum ergodic sequence .
then there exists @xmath441 so that @xmath442 for all @xmath443 .
we argue by contradiction . if the conclusion were false , there would exist a subsequence @xmath444 strongly in @xmath110 , but with @xmath445 weakly in @xmath110 .
the first assumption implies the existence of a subsequence ( which we continue to denote by @xmath446 ) satisfying @xmath444 a.e .
but @xmath110 has the weak banach - saks property : any weakly convergent sequence in @xmath110 has a subsequence whose arithmetic means converge strongly ( szlenk s weak banach - saks theorem for @xmath110 ) .
we choose such a subsequence for @xmath446 and continue to denote it as @xmath446 .
this subsequence has the properties that 1 .
@xmath444 a.e .
2 . @xmath447 strongly in @xmath110 .
but @xmath448 on the same set where @xmath449 , hence by ( 1 ) @xmath450 a.s .
this contradicts ( 2 ) and completes the proof . combining with the above ,
we have suppose that @xmath397 is an @xmath22 quantum ergodic sequence on the base .
then the conjectured yau lower bound holds : @xmath451 for some @xmath452 .
we also see that the limits in proposition [ l1rega ] are non - zero : suppose that @xmath397 is an @xmath22 quantum ergodic sequence on the base .
then there exists @xmath453 so that any weak limit of the sequence @xmath454 has mass @xmath455 .
of course , such an abstract functional analysis argument only serves a purpose if we can prove that eigenfunctions of @xmath8 are @xmath22 quantum ergodic on the base in interesting cases .
it is natural to conjecture that this condition holds on negatively curved manifolds , since the expected @xmath110 norm of a random wave is bounded below by a positive constant .
the main problem is that @xmath456 is a non - separable banach space .
the standard quantum ergodicity arguments show that ( when quantum ergodicity is valid ) , for any borel set @xmath419 there exists a subsequence @xmath457 of density one so that @xmath458 however , the non - separability of @xmath456 means that one can not use the diagonalization argument of @xcite to show that there exists a density one subsequence independent of @xmath419 so that holds .
if @xmath22 quantum ergodicity fails , then zero - density subsequences of eigenfunctions would ` scar ' along cantor sets c of positive measure .
that is , the mass @xmath459 may tend to a larger value than @xmath460 .
* equidistributed sums of gaussian beams and quantum ergodicity * we briefly consider the question whether it is possible to have a quantum ergodic sequence of eigenfunctions for which @xmath461 first , we observe that there do exist sequences of quantum ergodic functions ( not eigenfunctions ) with this property : @xmath462 } \to 0 $ ] in @xmath463 , dx)$ ] as long as @xmath464 but its square is the probability measure @xmath465 } $ ] and if the @xmath466 are uniformly distribution in @xmath467 $ ] ( w.r.t .
@xmath468 ) , this tends weakly to @xmath468 . it is tempting to construct sequences of eigenfunctions with the same property : a gaussian beam @xmath469 on the standard @xmath99 associated to a closed geodesic @xmath88 ( i.e. a rotate of @xmath226 ) is of height @xmath470 in a tube of radius @xmath471 around @xmath88 .
if we let @xmath472 and choose @xmath473 closed geodesics which are @xmath474separated , and become equidistributed in the space of closed geodesics , then @xmath475 is an eigenfunction whose @xmath110-norm tends to zero like @xmath476 but whose @xmath270 norm is asymptotic to @xmath379 and whose modulus square tends weak * to @xmath379 .
more precisely , @xmath477 weakly . to prove that @xmath478 requires proving that @xmath479 . the sum is over @xmath480 terms which are exponentially outside the tube intersections @xmath481 . in the sum
we may fix @xmath482 and multiply by @xmath473 .
so we need then to show that @xmath483 .
the geodesics are well - separated if the distance in the space of geodesics between them is @xmath484 , which means that the angle between @xmath485 and @xmath486 is at least this amount .
when the angle is @xmath487 then the inner product @xmath488 since the area of @xmath481 is bounded by this amount .
for any @xmath428 the sum over geodesics separated by @xmath428 is @xmath489 .
the remaining number of terms is @xmath490 .
so if @xmath491 both terms tend to zero .
a related question is whether nodal sets of orthogonal eigenfunctions of the same eigenvalue must intersect . of course
, this question only arises when the eigenvalue has multiplicity @xmath492 .
a result of this kind was obtained by v. gichev under a topological condition on @xmath28 .
[ gi ] @xcite suppose that @xmath493 and that @xmath494 are orthogonal eigenfunctions with the same eigenvalue @xmath37 . then @xmath495 .
we briefly sketch the proof : let @xmath496 resp .
@xmath497 be the family of nodal domains of @xmath498 resp .
each union @xmath500 covers @xmath28 up to the nodal set of @xmath501 .
if the nodal sets do not intersect then the nodal set of @xmath502 is contained in @xmath503 , for instance ; similarly if the indices are reversed .
hence the nodal sets have empty intersection if and only if @xmath504 covers @xmath28 . under this condition
, gichev constructs a closed 1-form which is not exact by showing that the incidence graph of the cover obtained from the union of the nodal domains of @xmath498 and @xmath499 contains a cycle .
he then considers a nodal domain @xmath505 of @xmath498 and a nodal domain @xmath506 of @xmath502 which intersect .
let @xmath507 . since @xmath508 there exists a smooth function @xmath104 on @xmath28 such that @xmath509 in a neighborhood of @xmath510 and @xmath511 near @xmath512 .
let @xmath513 be the one form which equals @xmath514 on @xmath505 and @xmath133 on the complement of @xmath505 .
clearly @xmath515 is closed and it is verified in @xcite that @xmath515 is not exact .
givech also proves that for @xmath99 , if @xmath133 is a regular value of @xmath516 then @xmath517 for every orthogonal eigenfunction @xmath499 with the same eigenvalue .
the proof is simply to use green s formula for a nodal domain for @xmath498 and note that the integral of @xmath518 equals zero on its boundary .
a related observation is the curious identity of @xcite , which holds for any @xmath1 : for any pair of eigenfunctions , @xmath519 hence for a pair of orthogonal eigenfunctions of the same eigenvalue , @xmath520
studies of nodal sets often involve dual studies of @xmath0 norms of eigenfunctions . in this section
, we review a number of relatively recent results on @xmath0 norms , both in the global manifold @xmath28 and for restrictions of eigenfunctions to submanifolds .
let @xmath131 denote the collection of nodal domains of @xmath12 .
for @xmath521 let @xmath522 in @xcite the following is proved ( see corollary 1.7 ) : @xcite let @xmath1 be a @xmath26 riemannian surface . for every @xmath12 with @xmath523 , @xmath524 hence , for each @xmath525 , the number of nodal domains @xmath86 of @xmath12 where the maximal bound @xmath526 is achieved in order of magnitude does not exceed @xmath527 .
in particular , for fixed @xmath56 , it remains bounded as @xmath528 .
the proof uses a certain bananch indicatrix , the sogge @xmath529 bounds , and estimates on the inradius of nodal domains . for a continuous function @xmath530 ,
the generalized banach indicatrix is defined by @xmath531 where for a regular value @xmath532 of @xmath104 , @xmath533 is the number of connected components of @xmath534 . in @xcite , the integral @xmath535 )
is bounded from above through the @xmath270-norms of the function @xmath104 and @xmath536 .
i.e .. in theorem 1.3 . for any @xmath537 and any continuous function @xmath55 on @xmath538 , @xmath539
the proof is roughly as follows : let @xmath540 be a point of @xmath541 where th maximum is achieved . by the inradius bound @xcite , there exists
so that the disc @xmath543 . one can then express @xmath12 in @xmath544 by the sum of a green s integral and poisson integral with respect to the euclidean dirichlet green s function of a slightly smaller disc .
in particular one may express @xmath545 by such an integral .
apply hlder s inequality one gets @xmath546 since the discs are disjoint one can sum in @xmath443 and apply the sogge @xmath547 bound to include the proof .
thus , the only fact one used about nodal domains was lower bound on the inradius .
this result bears a curious comparison to the results of @xcite giving new constraints on @xmath1 which are of maximal eigenfunction growth , i.e. possess eigenfunctions such that @xmath548 for some sequence of eigenfunctions @xmath549 with @xmath396 .
the result ( building on older results of sogge and the author ) states that such a sequence can exist only if @xmath1 possesses a ` pole ' @xmath147 for which the set of geodesic loops @xmath550 based at @xmath147 has positive measure in @xmath551 ( with respect to the natural spherical volume measure ) and such that the first return map has a recurrence property .
in fact , the only known surfaces where the bounds are achieved are surfaces of revolution , and in this case the first return map is the identity .
it is quite plausible that if @xmath1 has maximal eigenfunction growth , then the first return map must be the identity map on a set of positive measure in @xmath550 .
combined with the polterovich - sodin result above , we see that such ` poles ' @xmath147 , when they exist , can only occur in a uniformly bounded number of nodal domains of a surface
. it would be interesting to know if there can exist only a finite number of such points at all if one additionally assumes that the set of smoothly closed geodesics has measure zero .
for instance , , in that case , there might be a unique pole in each of the finite number of possible nodal domains .
this finitude problem would be useful in strengthening the condition on @xmath1 of maximal eigenfunction growth .
a problem of current interest is to consider @xmath0 norms of restrictions of eigenfunctions to hypersurfaces or higher codimension submanifolds . for expository purposes
we only consider geodesics on surfaces here . following earlier work of a. reznikov , burq , grard and tzvetkov
@xcite proved [ bgt ] @xcite suppose that @xmath1 is a compact surface , then there exists @xmath552 so that , for any geodesic segment @xmath553 of length @xmath554 and any eigenfunction @xmath12 with @xmath555 we have @xmath556 their estimate is sharp for the round sphere @xmath99 because of the highest weight spherical harmonics they also showed that for all geodesic segments @xmath88 of unit length , @xmath557 the estimate is only known to be achieved when the geodesic is elliptic , and quite likely it can be improved if the geodesic is hyperbolic . a result in this direction is : [ str ] @xcite suppose that @xmath1 is a compact surface of non - positive curvature
. then for all @xmath428 , there exists @xmath552 so that , for any geodesic segment @xmath553 of length @xmath554 and any eigenfunction @xmath12 with @xmath558 we have @xmath559 a related result on @xmath113 norms is , [ theorem1.1 ] @xcite let @xmath102 be a surface and assume that the set @xmath560 of periodic points has liouville measure zero in @xmath29 .
then there is a subsequence of eigenvalues @xmath561 of density one so that @xmath562 the results are based in part on a relatively new kakeya - nikodym maximal function estimate of bourgain @xcite , as improved by sogge @xcite .
we believe that it can be improved the following phase space kakeya - nikodym theorem .
let @xmath563 be the tube of radius @xmath75 around a geodesic arc in @xmath28 , and let @xmath564 be a smooth cutoff to a phase space tube of its lift to @xmath29 .
then for all @xmath428 , there exists @xmath565 such that @xmath566 we expect the sup occurs when @xmath88 is the orbit of @xmath567 .
but then it is easy to estimate the right side and one should be able to get a quantitative improvement of theorem [ theorem1.1 ] .
in this section we briefly review a recent series of results @xcite on quantum ergodic restriction theorems . they are used in section [ nodalgeos ] to determine the limit distribution of intersections of nodal lines and geodesics on real analytic surfaces ( in the complex domain ) .
let @xmath568 be a hypersurface and consider the cauchy data @xmath569 of eigenfunctions along @xmath570 ; here @xmath281 is the normal derivative .
we refer to @xmath571 as the dirichlet data and to @xmath572 as the neumann data .
a qer ( quantum ergodic restriction ) theorem seeks to find limits of matrix elements of this data along @xmath570 with respect to pseudo - differential operators @xmath573 on @xmath570 .
the main idea is that @xmath574 , the set of unit covectors with footpoints on @xmath570 , is a cross - section to the geodesic flow and the first return map of the geodesic flow for @xmath574 is ergodic .
the cauchy data should be the quantum analogue of such a cross section and therefore should be quantum ergodic on @xmath570 . for applications to nodal sets and other problems
, it is important to know if the dirichlet data alone satisfies a qer theorem .
the answer is obviously ` no ' in general .
for instance if @xmath1 has an isometric involution and with a hypersurface @xmath570 of fixed points , then any eigenfunction which is odd with respect to the involution vanishes on @xmath570 .
but in @xcite a sufficient condition is given for quantum ergodic restriction , which rules out this and more general situations .
the symmetry condition is that geodesics emanating from the ` left side ' of @xmath570 have a different return map from geodesics on the ` right side ' when the initial conditions are reflections of each other through @xmath575 . to take
the simplest example of the circle , the restriction of @xmath576 to a point is never quantum ergodic but the full cauchy data @xmath577 of course satisfies @xmath578 . in @xcite
it is proved that cauchy data always satisfies qer for any hypersurface .
this has implications for ( at least complex ) zeros of even or odd eigenfunctions along an axis of symmetry , e.g. for the case of maass forms for the modular domain @xmath579 ( see [ nodalgeos ] ) . to state the qer theorem
, we introduce some notation .
we put @xmath580 we further denote by @xmath581 the restriction map , @xmath582 for any orientable ( embedded ) hypersurface @xmath568 , there exists two unit normal co - vector fields @xmath583 to @xmath570 which span half ray bundles @xmath584 . infinitesimally , they define two ` sides ' of @xmath570 , indeed they are the two components of @xmath585 .
we use fermi normal coordinates @xmath586 along @xmath570 with @xmath587 and with @xmath588 and let @xmath589 denote the dual symplectic coordinates . for @xmath590 ( the co - ball bundle ) , there exist two unit covectors @xmath591 such that @xmath592 and @xmath593 . in the above orthogonal decomposition ,
they are given by @xmath594 we define the reflection involution through @xmath595 by @xmath596 its fixed point set is @xmath595 .
we denote by @xmath30 the homogeneous geodesic flow of @xmath1 , i.e. hamiltonian flow on @xmath597 generated by @xmath598 .
we define the _ first return time _
@xmath599 on @xmath574 by , @xmath600 by definition @xmath601 if the trajectory through @xmath602 fails to return to @xmath570 .
inductively , we define the jth return time @xmath603 to @xmath574 and the jth return map @xmath604 when the return times are finite .
we define the first return map on the same domain by @xmath605 when @xmath30 is ergodic , @xmath606 is defined almost everywhere and is also ergodic with respect to liouville measure @xmath607 on @xmath574 .
[ anc ] we say that @xmath570 has a positive measure of microlocal reflection symmetry if @xmath608 otherwise we say that @xmath570 is asymmetric with respect to the geodesic flow . the qer theorem we state below holds for both poly - homogeneous ( kohn - nirenberg ) pseudo - differential operators as in @xcite and also for semi - classical pseudo - differential operators on @xmath570 @xcite with essentially the same proof .
to avoid confusion between pseudodifferential operators on the ambient manifold @xmath28 and those on @xmath570 , we denote the latter by @xmath573 where @xmath609 by kohn - nirenberg pseudo - differential operators we mean operators with classical poly - homogeneous symbols @xmath610 @xmath611 as @xmath612 on @xmath595 as in @xcite . by semi - classical pseudo - differential operators we mean @xmath613-quantizations of semi - classical symbols @xmath614)$ ] of the form @xmath615 as in @xcite .
we further introduce the zeroth order homogeneous function @xmath616 on @xmath617 and also denote by @xmath618 its restriction to @xmath619 . for homogeneous pseudo - differential operators ,
the qer theorem is as follows : [ phgtheorem ] @xcite let @xmath1 be a compact manifold with ergodic geodesic flow , and let @xmath620 be a hypersurface .
let @xmath621 denote the @xmath44-normalized eigenfunctions of @xmath23 .
if @xmath570 has a zero measure of microlocal symmetry , then there exists a density - one subset @xmath622 of @xmath623 such that for @xmath624 and @xmath625 @xmath626 where @xmath627 alternatively , one can write @xmath628 note that @xmath629 is bounded but is not defined for @xmath630 , hence @xmath631 is not defined for @xmath632 if @xmath633 is homogeneous of order zero on @xmath595 .
the analogous result for semi - classical pseudo - differential operators is : @xcite [ sctheorem ]
let @xmath1 be a compact manifold with ergodic geodesic flow , and let @xmath620 be a hypersurface .
if @xmath570 has a zero measure of microlocal symmetry , then there exists a density - one subset @xmath622 of @xmath623 such that for @xmath634 @xmath635 where @xmath627 examples of asymmetric curves on surfaces in the case where @xmath1 is a finite area hyperbolic surface are the following : * @xmath570 is a geodesic circle ; * @xmath570 is a closed horocycle of radius @xmath636 , the injectivity radius . *
@xmath570 is a generic closed geodesic or an arc of a generic non - closed geodesic .
in this section , we briefly discuss some analogues of and for critical points on surfaces . to be sure , it is not hard to generate many identities ; the main problem is to derive information from them .
we denote the gradient of a function @xmath177 by @xmath637 and its hessian by @xmath638 , where @xmath639 is the riemannian connection .
we also denote the area form by @xmath640 and the scalar curvature by @xmath641 .
the results are based on unpublished work in progress of the author .
it is often said that measuring critical point sets and values is much more difficult than measuring nodal sets , and in a sense the identities reflect this difficulty , and we immediately see one difficulty in that the identities become signed : [ mainprop ] suppose that @xmath1 is a riemannian surface , and that @xmath177 is a morse eigenfunction with @xmath642 .
. then @xmath644 here , @xmath645 if @xmath147 is a local maximum or minimum and @xmath646 if @xmath147 is a saddle point .
when @xmath647 , the identity reduces to the gauss - bonnet theorem @xmath648 and the hopf index formula @xmath649 as this indicates , the main problem with applying the identity to counting critical points is that the left side is an alternating sum over critical points rather than a positive sum . in @xcite
a related identity using @xmath650 produced a sum of constant sign over the singular points of @xmath177 , but singular points are always saddle points of index @xmath651 and hence of constant sign .
note that under the morse assumption , @xmath652 , so that the right side is a well defined measure integrated against @xmath506 .
we now make some interesting choices of @xmath506 . as mentioned above , ( weighted ) counting of critical values should be simpler than weighted counting of critical points .
hence we put @xmath653 for smooth @xmath104 . this choice does give cancellation of the ` bad factor ' @xmath654 and ( using that @xmath655 ) we get [ v = fphi ] with the assumptions of proposition [ mainprop ] , if @xmath656 , then @xmath657 of course , this still has the defect that the left side is an oscillating sum , and the factor @xmath658 in the sum damps out the critical points in regions of exponential decay .
to illustrate , if @xmath659 we get @xmath660 to highlight the sign issue , we break up the sum into the sub - sum over maxima / minima and the sub - sum over saddle points , denoting the set of local maxima ( resp .
minima ) by @xmath661 ( resp .
@xmath662 ) and the set of saddle points by @xmath663 .
of course we have @xmath664 .
then is equivalent to @xmath665 we write @xmath666 where @xmath667 .
we note that on any compact riemannian manifold , @xmath668 uniformly in @xmath150 as @xmath16 while @xmath669 can be quite complicated to estimate . when @xmath670 we get , @xmath671 assuming @xmath177 is a morse eigenfunction , this implies @xmath672 to get rid of the signs in the sum , we could choose @xmath673 , where the determinant is defined by the metric .
since @xmath674 we obtain @xmath675 but the first term appears to be difficult to estimate .
we now take up the theme mentioned in the introduction of analytically continuing eigenfunctions on real analytic @xmath1 to the complex domain . in the next sections we apply the analytic continuation to the study of nodal of eigenfunctions in the real analytic case . for background
we refer to @xcite .
a real analytic manifold @xmath28 always possesses a unique complexification @xmath676 generalizing the complexification of @xmath677 as @xmath678 .
the complexification is an open complex manifold in which @xmath28 embeds @xmath679 as a totally real submanifold ( bruhat - whitney ) . as examples
, we have : * @xmath680 is @xmath681 . *
the unit sphere @xmath682 defined by @xmath683 in @xmath684 is complexified as the complex quadric @xmath685 * the hyperboloid model of hyperbolic space is the hypersurface in @xmath684 defined by @xmath686 then , @xmath687 * any real algebraic subvariety of @xmath677 has a similar complexification . *
any lie group @xmath688 ( or symmetric space ) admits a complexification @xmath689 .
the riemannian metric determines a special kind of distance function on @xmath676 known as a grauert tube function .
it is the plurisubharmonic function @xmath690 on @xmath676 defined as the unique solution of the monge - ampre equation @xmath691 here , @xmath692 is the delta - function on the real @xmath28 with respect to the volume form @xmath6 , i.e. @xmath693 . in fact
, it is observed in @xcite that the grauert tube function is obtained from the distance function by setting @xmath694 where @xmath695 is the squared distance function in a neighborhood of the diagonal in @xmath696 .
one defines the grauert tubes @xmath697 .
there exists a maximal @xmath698 for which @xmath699 is well defined , known as the grauert tube radius . for @xmath700 ,
@xmath76 is a strictly pseudo - convex domain in @xmath676 .
the complexified exponential map @xmath701 defines a diffeomorphism from @xmath702 to @xmath76 and pulls back @xmath699 to @xmath598 .
the one - complex dimensional null foliation of @xmath703 , known as the ` monge - ampre ' or riemann foliation , are the complex curves @xmath704 , where @xmath88 is a geodesic , where @xmath705 and where @xmath706 denotes multiplication of the tangent vector to @xmath88 by @xmath707 .
we refer to @xcite for further discussion .
the half - wave group of @xmath1 is the unitary group @xmath708 generated by the square root of the positive laplacian .
its schwartz kernel is a distribution on @xmath709 with the eigenfunction expansion @xmath710 by the poisson operator we mean the analytic continuation of@xmath711 to positive imaginary time , @xmath712 the eigenfunction expansion then converges absolutely to a real analytic function on @xmath713 .
let @xmath714 denote the operator of analytic continuation of a function on @xmath28 to the grauert tube @xmath76 . since @xmath715 it is simple to see that @xmath716 where @xmath717 is the analytic continuation of the poisson kernel in @xmath150 to @xmath76 . in terms of the eigenfunction expansion , one has @xmath718 this is a very useful observation because @xmath719 is a fourier integral operator with complex phase and can be related to the geodesic flow .
the analytic continuability of the poisson operator to @xmath76 implies that every eigenfunction analytically continues to the same grauert tube .
the analytic continuation of the possion - wave kernel to @xmath76 in the @xmath150 variable is discussed in detail in @xcite and ultimately derives from the analysis by hadamard of his parametrix construction .
we only briefly discuss it here and refer to @xcite for further details . in the case of euclidean @xmath720 and its wave kernel
@xmath721 which analytically continues to @xmath722 as the integral @xmath723 the integral clearly converges absolutely for @xmath724 exact formulae of this kind exist for @xmath725 and @xmath726 . for a general real analytic riemannian manifold
, there exists an oscillatry integral expression for the wave kernel of the form , @xmath727 where @xmath728 is a polyhomogeneous amplitude of order @xmath133 .
the holomorphic extension of ( [ paraone ] ) to the grauert tube @xmath729 in @xmath150 at time @xmath730 then has the form @xmath731 thus , a function @xmath732 has a holomorphic extension to the closed tube @xmath733 if and only if @xmath734 where @xmath735 is the backwards ` heat operator ' generated by @xmath736 ( rather than @xmath8 ) .
that is , @xmath737 admits an analytic continuation to the open grauert tube @xmath76 if and only if @xmath104 is in the domain of @xmath735 , i.e. if @xmath738 .
indeed , the analytic continuation is @xmath739 .
the subtlety is in the nature of the restriction to the boundary of the maximal grauert tube .
this result generalizes one of the classical paley - wiener theorems to real analytic riemannian manifolds @xcite . in the simplest case of @xmath740 , @xmath741 is the restriction of a holomorphic function @xmath742 on the annulus @xmath743 and with @xmath744 if and only if @xmath745 . the case of @xmath677
is more complicated since it is non - compact .
we are mainly concerned with compact manifolds and so the complications are not very relevant here .
but we recall that one of the classical paley - wiener theorems states that a real analytic function @xmath104 on @xmath720 is the restriction of a holomorphic function on the closed tube @xmath746 which satisfies @xmath747 for @xmath748 if and only if @xmath749 .
let us consider examples of holomorphic continuations of eigenfunctions : * on the flat torus @xmath750 the real eigenfunctions are @xmath751 with @xmath752 the complexified torus is @xmath753 and the complexified eigenfunctions are @xmath754 with @xmath755 * on the unit sphere @xmath725 , eigenfunctions are restrictions of homogeneous harmonic functions on @xmath756 .
the latter extend holomorphically to holomorphic harmonic polynomials on @xmath757 and restrict to holomorphic function on @xmath758 .
* on @xmath726 , one may use the hyperbolic plane waves @xmath759 , where @xmath760 is the ( signed ) hyperbolic distance of the horocycle passing through @xmath761 and @xmath762 to @xmath133 .
they may be holomorphically extended to the maximal tube of radius @xmath763 . * on compact hyperbolic quotients @xmath764 , eigenfunctions can be then represented by helgason s generalized poisson integral formula @xcite , @xmath765 here , @xmath766 ( the unit disc ) , @xmath767 , and @xmath768 is the boundary value of @xmath12 , taken in a weak sense along circles centered at the origin @xmath133 . to analytically continue
@xmath12 it suffices to analytically continue @xmath769 writing the latter as @xmath770 we have : @xmath771 the next step is to holomorphically extend the spectral projectors @xmath772}(x , y ) = \sum_j \delta(\lambda - \lambda_j ) { \varphi}_j(x ) { \varphi}_j(y ) $ ] of @xmath736 .
the complexified diagonal spectral projections measure is defined by @xmath773}^{\c}(\zeta , \bar{\zeta } ) = \sum_j \delta(\lambda - \lambda_j ) |{\varphi}_j^{\c}(\zeta)|^2.\ ] ] henceforth , we generally omit the superscript and write the kernel as @xmath774}^{\c}(\zeta , \bar{\zeta})$ ] .
this kernel is not a tempered distribution due to the exponential growth of @xmath775 .
since many asymptotic techniques assume spectral functions are of polynomial growth , we simultaneously consider the damped spectral projections measure @xmath776 } ^{\tau}(\zeta , \bar{\zeta } ) = \sum_j \delta(\lambda - \lambda_j ) e^{- 2 \tau \lambda_j } |{\varphi}_j^{\c}(\zeta)|^2,\ ] ] which is a temperate distribution as long as @xmath777 when we set @xmath778 we omit the @xmath707 and put @xmath779 } ( \zeta , \bar{\zeta } ) = \sum_j \delta(\lambda - \lambda_j ) e^{- 2 \sqrt{\rho}(\zeta ) \lambda_j } |{\varphi}_j^{\c}(\zeta)|^2.\ ] ] the integral of the spectral measure over an interval @xmath780 gives @xmath781 its complexification gives the kernel ( [ cxspecproj ] ) along the diagonal , latexmath:[\[\label{cxsp } \pi_{i}(\zeta , \bar{\zeta } ) = \sum_{j : \lambda_j \in i } gives its temperate version latexmath:[\[\label{cxdsp } p^{\tau}_{i}(\zeta , \bar{\zeta } ) = \sum_{j : \lambda_j \in i } e^{- 2 \tau \lambda_j } the damped spectral projection measure @xmath784}^{\tau}(\zeta , \bar{\zeta})$ ] ( [ spprojdamped ] ) is dual under the real fourier transform in the @xmath785 variable to the restriction @xmath786 to the anti - diagonal of the mixed poisson - wave group .
the adjoint of the poisson kernel @xmath787 also admits an anti - holomorphic extension in the @xmath788 variable .
the sum ( [ cxwvgp ] ) are the diagonal values of the complexified wave kernel @xmath789 we obtain ( [ eform ] ) by orthogonality of the real eigenfunctions on @xmath28 .
since @xmath790 takes its values in the cr holomorphic functions on @xmath791 , we consider the sobolev spaces @xmath792 of cr holomorphic functions on the boundaries of the strictly pseudo - convex domains @xmath793 , i.e. @xmath794 where @xmath795 is the @xmath374th sobolev space and where @xmath796 is the space of boundary values of holomorphic functions .
the inner product on @xmath797 is with respect to the liouville measure @xmath798 we then regard @xmath799 as the kernel of an operator from @xmath800 .
it equals its composition @xmath801 with the szegprojector @xmath802 for the tube @xmath803 , i.e. the orthogonal projection onto boundary values of holomorphic functions in the tube .
this is a useful expression for the complexified wave kernel , because @xmath804 is a complex fourier integral operator with a small wave front relation .
more precisely , the real points of its canonical relation form the graph @xmath805 of the identity map on the symplectic one @xmath806 spanned by the real one - form @xmath807 , i.e. @xmath808 we note that for each @xmath809 there exists a symplectic equivalence @xmath810 by the map @xmath811 , where @xmath812 is the action form ( cf .
@xcite ) .
the following result was first stated by boutet de monvel ( for more details , see also @xcite ) .
[ boufio ] @xcite [ bdm ] @xmath813 is a complex fourier integral operator of order @xmath814 associated to the canonical relation @xmath815 moreover , for any @xmath374 , @xmath816 is a continuous isomorphism . in @xcite
we give the following sharpening of the sup norm estimates of @xcite : [ pw ] suppose @xmath1 is real analytic .
then @xmath817 the proof follows easily from the fact that the complexified poisson kernel is a complex fourier integral operator of finite order .
the estimates can be improved further . in @xcite , we discussed analogues in the setting of gruaert tubes for the basic notions of pluripotential theory on domains in @xmath678 . of relevance here
is that the grauert tube function @xmath699 is the analogue of the pluri - complex green s function .
we recall that the maximal psh function ( or pluri - complex green s function ) relative to a subset @xmath819 is defined by @xmath820 is the space @xmath821 spanned by eigenfunctions with frequencies @xmath822 . rather than using the sup norm , it is convenient to work with @xmath270 based norms than sup norms , and so we define @xmath823 and the extremal function by @xmath824 the extremal psh function is defined by @xmath825 and that @xmath826 the proof is based on the properties of . by using a bernstein - walsh inequality @xmath827}(\zeta , \bar{\zeta})}{\phi_m^{\lambda}(\zeta)^2 } \leq c n(\lambda)\ ; e^{\epsilon n(\lambda)},\ ] ]
it is not hard to show that @xmath828 to evaluate the logarithm , one can show that the kernel is essentially @xmath829 times the temperate projection defined by the poisson operator , @xmath830}(\zeta , \bar{\zeta } ) = \sum_{j : \lambda_j \in [ 0 , \lambda ] } e^{- 2 \sqrt{\rho}(\zeta)\lambda_j } fact that @xmath831}(\zeta , \bar{\zeta } ) = 0 $ ] .
we now return to nodal sets , where we will see the same extremal functions arise .
it is often possible to obtain more refined results on nodal sets by studying their intersections with some fixed ( and often special ) hypersurface .
this has been most successful in dimension two . in this section ,
we review the results of @xcite giving upper bounds on the number of intersections of the nodal set with the boundary of an analytic ( or more generally piecewise analytic ) plane domain .
one may expect that the results of this section can also be generalized to higher dimensions by measuring codimension two nodal hypersurface volumes within the boundary .
thus we would like to count the number of nodal lines ( i.e. components of the nodal set ) which touch the boundary .
here we assume that @xmath133 is a regular value so that components of the nodal set are either loops in the interior ( closed nodal loops ) or curves which touch the boundary in two points ( open nodal lines ) .
it is known that for generic piecewise analytic plane domains , zero is a regular value of all the eigenfunctions @xmath549 , i.e. @xmath832 on @xmath833 @xcite ; we then call the nodal set regular . since the boundary lies in the nodal set for dirichlet boundary conditions , we remove it from the nodal set before counting components .
henceforth , the number of components of the nodal set in the dirichlet case means the number of components of @xmath834 in the following , and henceforth , @xmath835 denotes a positive constant depending only on the domain @xmath125 .
[ cor ] let @xmath125 be a piecewise analytic domain and let @xmath836 be the number of components of the nodal set of the @xmath443th neumann or dirichlet eigenfunction which intersect @xmath837 .
then there exists @xmath838 such that @xmath839 by a piecewise analytic domain @xmath840 , we mean a compact domain with piecewise analytic boundary , i.e. @xmath841 is a union of a finite number of piecewise analytic curves which intersect only at their common endpoints .
such domains are often studied as archtypes of domains with ergodic billiards and quantum chaotic eigenfunctions , in particular the bunimovich stadium or sinai billiard .
their nodal sets have been the subject of a number of numerical studies ( e.g. @xcite ) . in general
, there does not exist a non - trivial lower bound for the number of components touching the boundary .
e.g. in a disc , the zero sets of the eigenfunctions are unions of circles concentric with the origin and spokes emanating from the center . only the spokes intersect the boundary and their number reflects the angular momentum rather than the eigenvalue of the eigenfunction .
but we conjecture that for piecewise analytic domains with ergodic billiards , the the number of complex zeros of @xmath842 is bounded below by @xmath843 .
we discuss work in progress on this conjecture in [ nodalgeos ] . in comparison to the order @xmath844 of the number of boundary nodal points , the total number of connected components of @xmath833 has the upper bound @xmath845 by the courant nodal domain theorem .
it is not known in general whether the courant upper bound is achieved , but we expect that it is often achieved in order of magnitude . in @xcite
it is proved that the average number of nodal components of a random spherical harmonic is of order of magnitude @xmath10 .
thus , the number of components touching the boundary is one order of magnitude below the total number of components .
the article @xcite also contains a similar estimate on the number of critical points of @xmath549 which occur on the boundary .
we denote the boundary critical set by @xmath846 in the case of neumann eigenfunctions , @xmath847 since the normal derivative automatically equals zero on the boundary , while in the dirichlet case @xmath848 since the boundary is a level set . we observe that radial eigenfunctions on the disc are constant on the boundary ; thus , boundary critical point sets need not be isolated .
we therefore impose a non - degeneracy condition on the tangential derivative @xmath849 to ensure that its zeros are isolated and can be counted .
we say that the neumann problem for a bounded domain has the _ asymptotic schiffer property _
if there exists @xmath850 such that , for all neumann eigenfunctions @xmath549 with sufficiently large @xmath11 , @xmath851 here , @xmath852 is the unit tangential derivative , and the @xmath270 norms refer to the restrictions of the eigenfunction to @xmath837 .
[ critsb ] let @xmath853 be piecewise real analytic .
suppose that @xmath854 satisfies the asymptotic schiffer condition ( [ neund ] ) in the neumann case .
then the number of @xmath855 of critical points of a neumann or dirichlet eigenfunction @xmath549 which lie on @xmath837 satisfies @xmath856 for some @xmath835 in the case of dirichlet eigenfunctions , endpoints of open nodal lines are always boundary critical points , since they must be singular points of @xmath549 .
hence , an upper bound for @xmath857 also gives an upper bound for the number of open nodal lines .
[ bnpd ] suppose that @xmath853 is a piecewise real analytic plane domain .
let @xmath836 be the number of open nodal lines of the @xmath443th dirichlet eigenfunction , i.e. connected components of @xmath858 whose closure intersects @xmath837 .
then there exists @xmath859 such that @xmath860 there does not exist a non - trivial lower bound on the number of interior critical points @xcite . for the neumann problem ,
the boundary nodal points are the same as the zeros of the boundary values @xmath861 of the eigenfunctions .
the number of boundary nodal points is thus twice the number of open nodal lines .
hence in the neumann case , theorem [ cor ] follows from : [ bnp ] suppose that @xmath853 is a piecewise real analytic plane domain .
then the number @xmath862 of zeros of the boundary values @xmath861 of the @xmath443th neumann eigenfunction satisfies @xmath863 , for some @xmath835 .
this is a more precise version of theorem [ cor ] since it does not assume that @xmath133 is a regular value . in keeping with the theme of this survey , we prove theorem [ bnp ] by analytically continuing the boundary values of the eigenfunctions and counting _ complex zeros and critical points _ of analytic continuations of cauchy data of eigenfunctions .
when @xmath864 , the eigenfunctions can be holomorphically continued to an open tube domain in @xmath865 projecting over an open neighborhood @xmath866 in @xmath867 of @xmath125 which is independent of the eigenvalue .
we denote by @xmath868 the points @xmath869 with @xmath870 .
then @xmath871 extends to a holomorphic function @xmath872 where @xmath873 and where @xmath874 for some @xmath875 .
assuming @xmath837 real analytic , we define the ( interior ) complex nodal set by @xmath876 and the ( interior ) complex critical point set by @xmath877 [ mainthm ] suppose that @xmath853 is a piecewise real analytic plane domain , and denote by @xmath878 the union of the complexifications of its real analytic boundary components . 1 .
let @xmath879 be the number of complex zeros on the complex boundary .
then there exists a constant @xmath835 independent of the radius of @xmath878 such that @xmath880 2 .
suppose that the neumann eigenfunctions satisfy ( [ neund ] ) and let @xmath881 .
then there exists @xmath835 independent of the radius of @xmath878 such that @xmath882 the theorems on real nodal lines and critical points follow from the fact that real zeros and critical points are also complex zeros and critical points , hence @xmath883 all of the results are sharp , and are already obtained for certain sequences of eigenfunctions on a disc ( see [ examples ] ) .
if the condition ( [ neund ] ) is not satisfied , the boundary value of @xmath549 must equal a constant @xmath884 modulo an error of the form @xmath885 .
we conjecture that this forces the boundary values to be constant .
the method of proof of theorem [ mainthm ] generalizes from @xmath837 to a rather large class of real analytic curves @xmath886 , even when @xmath837 is not real analytic .
let us call a real analytic curve @xmath257 a _ good _ curve if there exists a constant @xmath163 so that for all @xmath11 sufficiently large , @xmath887 here , the @xmath270 norms refer to the restrictions of the eigenfunction to @xmath257 and to @xmath837 .
the following result deals with the case where @xmath888 is an _ interior _ real - analytic curve .
the real curve @xmath257 may then be holomorphically continued to a complex curve @xmath889 obtained by analytically continuing a real analytic parametrization of @xmath257 .
[ cnpthm ] suppose that @xmath853 is a @xmath26 plane domain , and let @xmath886 be a good interior real analytic curve in the sense of ( [ good ] ) .
let @xmath890 be the number of intersection points of the nodal set of the @xmath443-th neumann ( or dirichlet ) eigenfunction with @xmath257 .
then there exists @xmath891 depending only on @xmath892 such that @xmath893 .
a recent paper of j. jung shows that many natural curves in the hyperbolic plane are ` good ' @xcite .
we also note an interesting application due to i. polterovich @xcite of theorem [ cor ] to an old conjecture of a. pleijel regarding courant s nodal domain theorem , which says that the number @xmath894 of nodal domains ( components of @xmath895 ) of the @xmath896th eigenfunction satisfies @xmath897 .
pleijel @xcite improved this result for dirichlet eigefunctions of plane domains : for any plane domain with dirichlet boundary conditions , @xmath898 , where @xmath899 is the first zero of the @xmath142 bessel function .
he conjectured that the same result should be true for a free membrane , i.e. for neumann boundary conditions .
this was recently proved in the real analytic case by i. polterovich @xcite .
his argument is roughly the following : pleijel s original argument applies to all nodal domains which do not touch the boundary , since the eigenfunction is a dirichlet eigenfunction in such a nodal domain .
the argument does not apply to nodal domains which touch the boundary , but by theorem [ cor ] the number of such domains is negligible for the pleijel bound .
we now consider global results when hypotheses are made on the dynamics of the geodesic flow .
use of the global wave operator brings into play the relation between the geodesic flow and the complexified eigenfunctions , and this allows one to prove gobal results on nodal hypersurfaces that reflect the dynamics of the geodesic flow . in some cases ,
one can determine not just the volume , but the limit distribution of complex nodal hypersurfaces .
since we have discussed this result elsewhere @xcite we only briefly review it here .
the complex nodal hypersurface of an eigenfunction is defined by @xmath900 there exists a natural current of integration over the nodal hypersurface in any ball bundle @xmath901 with @xmath902 , given by @xmath903 , { \varphi}\rangle = \frac{i}{2 \pi } \int_{b^*_{\epsilon } m } { \partial{\bar\partial}}\log \int_{z_{{\varphi}_{\lambda}^{\c } } } { \varphi},\;\;\ ; { \varphi}\in { \mathcal{d}}^ { ( m-1 , m-1 ) } ( b^*_{\epsilon } m).\ ] ] in the second equality we used the poincar - lelong formula .
the notation @xmath904 stands for smooth test @xmath905-forms with support in @xmath906 the nodal hypersurface @xmath907 also carries a natural volume form @xmath908 as a complex hypersurface in a khler manifold . by wirtinger s formula
, it equals the restriction of @xmath909 to @xmath907 .
hence , one can regard @xmath907 as defining the measure @xmath910 we prefer to state results in terms of the current @xmath911 $ ] since it carries more information .
[ zero ] let @xmath1 be real analytic , and let @xmath371 denote a quantum ergodic sequence of eigenfunctions of its laplacian @xmath8 .
let @xmath912 be the maximal grauert tube around @xmath28 with complex structure @xmath913 adapted to @xmath33 .
let @xmath914 .
then : @xmath32 \to \frac{i } { \pi } { \partial{\bar\partial}}\sqrt{\rho}\;\ ; \mbox{weakly in } \;\ ; { \mathcal{d}}^ { ' ( 1,1 ) } ( b^*_{\epsilon } m),\ ] ] in the sense that , for any continuous test form @xmath915 , we have @xmath916 equivalently , for any @xmath917 , @xmath918 [ zerocor ] let @xmath1 be a real analytic with ergodic geodesic flow .
let @xmath371 denote a full density ergodic sequence .
then for all @xmath919 , @xmath32 \to
\frac{i } { \pi } { \partial{\bar\partial}}\sqrt{\rho},\;\ ; \mbox{weakly in}\;\ ; { \mathcal{d}}^ { ' ( 1,1 ) } ( b^*_{\epsilon } m).\ ] ] the proof consists of three ingredients : 1 . by the poincar - lelong formula , @xmath911 = i { \partial{\bar\partial}}\log |{\varphi}_{\lambda}^{\c}| . $ ]
this reduces the theorem to determining the limit of latexmath:[$\frac{1}{\lambda } \log 2 .
@xmath68 is a sequence of psh functions which are uniformly bounded above by @xmath699 . by a standard compactness theorem ,
the sequence is pre - compact in @xmath110 : every sequence from the family has an @xmath110 convergent subsequence .
@xmath921 , when properly @xmath270 normalized on each @xmath922 is a quantum ergodic sequence on @xmath922 .
this property implies that the @xmath270 norm of @xmath921 on @xmath837 is asymtotically @xmath699 .
4 . ergodicity and the calculation of the @xmath270 norm imply that the only possible @xmath110 limit of @xmath68 .
this concludes the proof .
we note that the first two steps are valid on any real analytic @xmath1 .
the difference is that the @xmath270 norms of @xmath818 may depend on the subsequence and can often not equal @xmath699 .
that is , latexmath:[$\frac{1}{\lambda } the ergodic case , but not in general .
for instance , on a flat torus , the complex zero sets of ladders of eigenfunctions concentrate on a real hypersurface in @xmath676 .
this may be seen from the complexified real eigenfunctions @xmath924 , which vanish if and only if @xmath925 and @xmath926 . here
, @xmath927 is a lattice point .
the exact limit distribution depends on which ray or ladder of lattice points one takes in the limit .
the result reflects the quantum integrability of the flat torus , and a similar ( but more complicated ) description of the zeros exists in all quantum integrable cases .
the fact that @xmath928 is pre - compact on a grauert tube of any real analytic riemannian manifold confirms the upper bound on complex nodal hypersurface volumes .
in [ planedomain ] we discussed upper bounds on the number of intersection points of the nodal set with the bounary of a real analytic plane domain and more general ` good ' analytic curves . in this section , we discuss work in progress on intersections of nodal sets and geodesics on surfaces with ergodic geodesic flow .
of course , the results are only tentative but it seems worthwhile at this point in time to explain the role of ergodicity in obtaining lower bounds and asymptotics .
we restrict to geodesic curves because they have rather special properties that makes the analysis somewhat different than for more general curves such as distance circles .
the dimensional restriction is due to the fact that the results are partly based on the quantum ergodic restriction theorems of @xcite , which concern restrictions of eigenfunctions to hypersurfaces .
nodal sets and geodesics have complementary dimensions and intersect in points , and therefore it makes sense to count the number of intersections .
we fix @xmath929 and let @xmath930 denote the corresponding parametrized geodesic .
our goal is to determine the asymptotic distribution of intersection points of @xmath931 with the nodal set of a highly eigenfunction . as usual
, we can not cope with this problem in the real domain and therefore analytically continue it to the complex domain .
thus , we consider the intersections @xmath932 of the complex nodal set with the ( image of the ) complexification of a generic geodesic if @xmath933 then @xmath931 admits an analytic continuation @xmath934 in other words , we consider the zeros of the pullback , @xmath935 we encode the discrete set by the measure @xmath936 = \sum_{(t + i \tau):\ ; { \varphi}_j^{\c}(\gamma_{x , \xi}^{\c}(t + i \tau ) ) = 0 } \delta_{t + i \tau}.\ ] ] we would like to show that for generic geodesics , the complex zeros on the complexified geodesic condense on the real points and become uniformly distributed with respect to arc - length .
this does not always occur : as in our discussion of qer theorems , if @xmath931 is the fixed point set of an isometric involution , then odd " eigenfunctions under the involution will vanish on the geodesic .
the additional hypothesis is that qer holds for @xmath931 , i.e. that theorem [ sctheorem ] is valid .
the following conjecture appears to be proved ( @xcite ) , but to be conservative , we state it here only as a conjecture : [ maincor ] let @xmath937 be a real analytic riemannian surface with ergodic geodesic flow .
let @xmath931 satisfy the qer hypothesis .
then there exists a subsequence of eigenvalues @xmath561 of density one such that for any @xmath938 , @xmath939 in other words , @xmath940 = \delta_{\tau = 0},\ ] ] in the sense of weak * convergence on @xmath941 .
thus , the complex nodal set intersects the ( parametrized ) complexified geodesic in a discrete set which is asymptotically ( as @xmath16 ) concentrated along the real geodesic with respect to its arclength .
this concentration- equidistribution result is a ` restricted ' version of the result of [ ergodicnodal ] . as noted there , the limit distribution of complex nodal sets in the ergodic case is a singular current @xmath942 .
the motivation for restricting to geodesics is that restriction magnifies the singularity of this current . in the case of a geodesic
, the singularity is magnified to a delta - function ; for other curves there is additionally a smooth background measure .
the assumption of ergodicity is crucial .
for instance , in the case of a flat torus , say @xmath943 where @xmath944 is a generic lattice , the real eigenfunctions are @xmath945 where @xmath946 , the dual lattice , with eigenvalue @xmath947 . due to the flatness , the restriction @xmath948 of the eigenfunction to a geodesic is an eigenfunction of the laplacian @xmath949 of submanifold metric along the geodesic with eigenvalue @xmath950 .
the complexification of the restricted eigenfunction is @xmath951 and its exponent of its growth is @xmath952 , which can have a wide range of values as the eigenvalue moves along different rays in @xmath199 .
the limit current is @xmath953 applied to the limit and thus also has many limits the proof involves several new principles which played no role in the global result of [ ergodicnodal ] and which are specific to geodesics .
however , the first steps in the proof are the same as in the global case . by the poincar - lelong formula
, we may express the current of summation over the intersection points in in the form , @xmath954 = i { \partial{\bar\partial}}_{t + i \tau } \log \left| \gamma_{x , \xi}^ * { \varphi}_{\lambda_j}^{\c } ( t + i \tau ) \right|^2.\ ] ] thus , the main point of the proof is to determine the asymptotics of @xmath955 .
when we freeze @xmath707 we put @xmath956 [ mainpropa ] ( growth saturation ) if @xmath371 satisfies qer along any arcs of @xmath931 , then in @xmath957 we have @xmath958 proposition [ mainpropa ] immediately implies theorem [ maincor ] since we can apply @xmath959 to the @xmath110 convergent sequence @xmath960 to obtain @xmath961 .
the upper bound in proposition [ mainpropa ] follows immediately from the known global estimate @xmath962 on all of @xmath922 .
hence the difficult point is to prove that this growth rate is actually obtained upon restriction to @xmath963 .
this requires new kinds of arguments related to the qer theorem .
* complexifications of restrictions of eigenfunctions to geodesics have incommensurate fourier modes , i.e. higher modes are exponentially larger than lower modes . *
the quantum ergodic restriction theorem in the real domain shows that the fourier coefficients of the top allowed modes are ` large ' ( i.e. as large as the lower modes ) .
consequently , the @xmath270 norms of the complexified eigenfunctions along arcs of @xmath963 achieve the lower bound of proposition [ mainpropa ] .
* invariance of wigner measures along the geodesic flow implies that the wigner measures of restrictions of complexified eigenfunctions to complexified geodesics should tend to constant multiples of lebesgue measures @xmath964 for each @xmath705 .
hence the eigenfunctions everywhere on @xmath963 achieve the growth rate of the @xmath270 norms .
these principles are most easily understood in the case of periodic geodesics .
we let @xmath965 parametrize the geodesic with arc - length ( where @xmath966 where @xmath967 is the length of @xmath931 ) .
first , we use theorem [ sctheorem ] to prove [ l2normintro ] assume that @xmath3 satsifies qer along the periodic geodesic @xmath931 .
let @xmath968 be the @xmath270-norm of the complexified restriction of @xmath969 along @xmath970 .
then , @xmath971 to prove lemma [ l2normintro ] , we study the orbital fourier series of @xmath972 and of its complexification .
the orbital fourier coefficients are @xmath973 and the orbital fourier series is @xmath974 hence the analytic continuation of @xmath972 is given by @xmath975 by the paley - wiener theorem for fourier series , the series converges absolutely and uniformly for @xmath976 . by
energy localization " only the modes with @xmath977 contribute substantially to the @xmath270 norm .
we then observe that the fourier modes decouple , since they have different exponential growth rates .
we use the qer hypothesis in the following way : [ fcsat ] suppose that @xmath397 is qer along the periodic geodesic @xmath931
. then for all @xmath441 , there exists @xmath978 so that @xmath979 lemma [ fcsat ] implies lemma [ l2normintro ] since it implies that for any @xmath441 , @xmath980 to go from asymptotics of @xmath270 norms of restrictions to proposition [ mainpropa ] we then use the third principle : [ ll ] ( lebesgue limits ) if @xmath981 ( identically ) , then for all @xmath705 the sequence @xmath982 is que with limit measure given by normalized lebesgue measure on @xmath212 . the proof of proposition [ mainpropa ] is completed by combining lemma [ l2normintro ] and proposition [ ll ] . conjecture [ maincor ] follows easily from proposition [ mainprop ] .
the proof for non - periodic geodesics is considerably more involved , since one can not use fourier analysis in quite the same way .
we mentioned above that riemannian random waves provide a probabilistic model that is conjectured to predict the behavior of eigenfunctions when the geodesic flow of @xmath1 is ergodic . in this section
, we define the model precisely as in @xcite ( see also @xcite for a similar model ) and survey some of the current results and conjectures . we should emphasize that some of the rigorous results on zeros or critical points of riemannian random waves , both in the real and complex domain , are much simpler than for individual eigenfuntions , and therefore do not provide much guidance on how to prove results for an orthonormal basis of eigenfunctions .
but the relative simplicity of random waves and their value as predictors provide the motivation for studying random waves . and
there are many hopelessly difficult problems on random waves as well , which we will survey in this section . for expository simplicity
we assume that the geodesic flow @xmath30 of @xmath1 is of one of the following two types : 1 .
_ aperiodic : _ the liouville measure of the closed orbits of @xmath30 , i.e. the set of vectors lying on closed geodesics , is zero ; or 2 . _ periodic = zoll : _ @xmath983 for some @xmath984 ; henceforth @xmath985 denotes the minimal period .
the common morse index of the @xmath985-periodic geodesics will be denoted by @xmath986 . in the real analytic case
, @xmath1 is automatically one of these two types , since a positive measure of closed geodesics implies that all geodesics are closed .
the two - term weyl laws counting eigenvalues of @xmath736 are very different in these two cases . 1 .
in the _ aperiodic _ case , ivrii s two term weyl law states @xmath987 where @xmath988 and where @xmath989 is a universal constant .
2 . in the _ periodic _ case , the spectrum of @xmath736 is a union of eigenvalue clusters @xmath990 of the form @xmath991 with @xmath992 .
the number @xmath993 of eigenvalues in @xmath990 is a polynomial of degree @xmath994 .
we refer to @xcite for background and further discussion . to define riemannian random waves , we partition the spectrum of @xmath995 into certain intervals @xmath996 of width one and denote by @xmath997 the spectral projections for @xmath995 corresponding to the interval @xmath996 .
the choice of the intervals @xmath996 is rather arbitrary for aperiodic @xmath1 and as mentioned above we assume @xmath998 $ ] . in the zoll case , we center the intervals around the center points @xmath999 of the @xmath143th cluster @xmath990 .
we call call such a choice of intervals a cluster decomposition .
we denote by @xmath993 the number of eigenvalues in @xmath996 and put @xmath1000 ( the range of @xmath1001 ) .
we choose an orthonormal basis @xmath1002 for @xmath1003 .
for instance , on @xmath99 one can choose the real and imaginary parts of the standard @xmath206 s .
we endow the real vector space @xmath1004 with the gaussian probability measure @xmath1005 defined by @xmath1006 here , @xmath1007 is @xmath1008-dimensional real lebesgue measure .
the normalization is chosen so that @xmath1009 , where @xmath1010 is the expected value with respect to @xmath1005
. equivalently , the @xmath993 real variables @xmath1011 ( @xmath1012 ) are independent identically distributed ( i.i.d . )
random variables with mean 0 and variance @xmath1013 ; i.e. , @xmath1014 we note that the gaussian ensemble is equivalent to picking @xmath1015 at random from the unit sphere in @xmath1016 with respect to the @xmath270 inner product .
depending on the choice of intervals , we obtain the following special ensembles : * the asymptotically _ fixed frequency _
ensemble @xmath1017 , where @xmath1018 $ ] and where @xmath1017 is the vector space of linear combinations @xmath1019 } c_j\;\ ; { \varphi}_{\lambda_j},\ ] ] of eigenfunctions with @xmath11 ( the frequency ) in an interval @xmath1020 $ ] of fixed width .
( note that it is the square root of the eigenvalue of @xmath8 , not the eigenvalue , which is asymptotically fixed ) .
* the _ high frequency cut - off _ ensembles @xmath1021}$ ] where the frequency is cut - off at @xmath21 : @xmath1022 * the _ cut - off gaussian free field _ , @xmath1023 one could use more general weights @xmath1024 on a sobolev space of functions or distributions on @xmath28 . in the physics terminology , @xmath1024 ( or it square )
is referred to as the power spectrum .
the key reason why we can study the limit distribution of nodal sets in this ensemble is that the covariance kernel @xmath1025 is the spectral projections kernel for @xmath1026 the real zeros are straightforward to define . for each @xmath1027}$ ] or @xmath1028 we associated to the zero set @xmath1029 the positive measure @xmath1030 where @xmath1031 is the induced ( hausdorff ) hypersurface measure .
the main result we review is the limit law for random sequences of random real riemannian waves . by a random sequence ,
we mean an element of the product probability space @xmath1032 [ zdsm ] @xcite let @xmath1 be a compact riemannian manifold , and let @xmath1033 be a random sequence in ( [ hinfty ] ) .
then @xmath1034 we first show that the normalized expected limit distribution @xmath1035 of zeros of random riemannian waves tends to the volume form @xmath6 as @xmath16 .
that is , we define the ` linear statistic ' , @xmath1036 and then define @xmath1037 [ zmeasure ] let @xmath1 be a compact riemannian manifold , let @xmath1038}$ ] be the cutoff ensemble and let @xmath1039 be the ensemble of riemannian waves of asymptotically fixed frequency .
then in either ensemble : 1 . for any @xmath26 @xmath1 , @xmath1040 .
2 . for a real analytic @xmath1 ,
@xmath1041 we restrict to real analytic metrics in ( 2 ) for the sake of brevity . in that case
, the variance estimate follows easily from theorem [ df ] .
the formula for the density of zeros of random elements of @xmath1016 can be derived from the general kac - rice formula @xcite : @xmath1042 here , @xmath1043 is the joint probability distribution of the gaussian random variables @xmath1044 , i.e. the pushforward of the gaussian measure on @xmath1045 under the map @xmath1046 .
note that the factor @xmath1047 in @xcite equals @xmath1048 in the codimension one case .
indeed , let @xmath1049 be the adjoint map with respect to the inner product @xmath33 on @xmath1050 .
let @xmath1051 be the composition .
by @xmath1052 is meant the determinant with respect to the inner product on @xmath1050 ; it clearly equals @xmath1053 in the codimension one case .
the formulae of @xcite ( the ` kac - rice ' formulae ) give that @xmath1054 where @xmath1055 is the gaussian density with covariance matrix @xmath1056 and @xmath1057 here , @xmath1058 making a simple change of variables in the integral ( [ d2 ] ) , we have [ bsz ] @xcite on a real riemannian manifold of dimension @xmath161 , the density of zeros of a random riemannian wave is @xmath1059 where @xmath1060 is a symmetric form on @xmath1061 . for the asymptotically fixed freqency ensembles , it is given by @xmath1062 in the cutoff ensemble the formula is the same except that @xmath1001 is replaced by @xmath1063}$ ] .
we then need the asymptotics of the matrix elements of @xmath1064 .
they are simplest for the round sphere , so we state them first in that case : [ pinsm ] let @xmath1065 be the orthogonal projection .
then : * \(a ) @xmath1066 ; * \(b ) @xmath1067 ; * \(c ) @xmath1068 we refer to @xcite for the calculation , which is quite simple because of the invariance under rotations .
the expected density of random nodal hypersurfaces is given as follows [ realdensity ] in the case of @xmath725 , @xmath1069 where @xmath1070 by propositiosn [ bsz ] and [ pinsm ] , we have @xmath1071 where @xmath1072
we now generalize the result to any compact @xmath26 riemannian manifold @xmath1 which is either aperiodic or zoll .
as in the case of @xmath725 , the key issue is the asymptotic behavior of derivatives of the spectral projections @xmath1073 [ ping ] assume @xmath1 is either aperiodic and @xmath998 $ ] or zoll and @xmath996 is a cluster decomposition .
let @xmath1074 be the orthogonal projection . then : * \(a ) @xmath1075 ; * \(b ) @xmath1076 ; * \(c ) @xmath1077 in the aperiodic case , 1 .
@xmath774 } ( x , x ) = c_m \lambda^m + o(\lambda^{m - 1 } ) ; $ ] 2 .
@xmath1078}(x , y ) |_{x = y } = c_m \lambda^{m + 2 } g_x + o(\lambda^{m + 1 } ) .
$ ] in the zoll case , one adds the complete asymptotic expansions for @xmath1001 over the @xmath143 clusters to obtain expansions for @xmath1079 .
we then have : [ realdensityg ] for the asymptotically fixed frequency ensemble , and for any @xmath1080 which is either zoll or aperiodic ( and with @xmath996 as in proposition [ ping ] ) , we have @xmath1081 where @xmath1082
the same formula holds for the cutoff ensemble . both on a sphere @xmath725 or on a more general @xmath1 which is either zoll or aperiodic , we have by propositions [ pinsm ] resp .
[ ping ] and the general formula for @xmath1083 in [ den ] that @xmath1084 it follows that @xmath1085 thus , we have @xmath1086 where @xmath1087 denotes a matrix whose norm is @xmath1087 , as as @xmath1088 we obtain the stated asymptotics .
so far , we have only determined the expected values of the nodal hypersurface measures . to complete the proof of theorem [ zmeasure ] , we need to prove : [ var ] if @xmath1 is real analytic , then the variance of @xmath1089 is bounded . by theorem [ df ] , for @xmath1090 , @xmath1091 has bounded mass .
hence , the random variable @xmath1089 is bounded , and therefore so is its variance .
the variance of @xmath1089 is given by @xmath1092 where @xmath1093 is the pair correlation function for zeros .
hence , boundedness would follow from @xmath1094 there is a formula similar to that for the density in proposition [ bsz ] for @xmath1095 and it is likely that it could be used to prove boundedness of the variance for any @xmath26 riemannian manifold .
we recall that the set of random sequences of riemannian waves of increasing frequency is the probability space @xmath1096 with the measure @xmath1097 .
an element in @xmath1098 will be denoted @xmath1099 .
we have , @xmath1100 by a density argument it suffices to prove that the linear statistics @xmath1101 almost surely in @xmath1098 we know that \(i ) @xmath1102 \(ii ) @xmath1103 is bounded on @xmath1098 .
since @xmath1089 for @xmath1104 is a sequence of independent random variables in @xmath1098 with bounded variances , the kolmogorov strong law of large numbers implies that @xmath1105 almost surely .
we now state a complex analogue of the equidistribution of real nodal sets and show that it agrees with the the limit formula of theorem [ zero ] .
we complexify riemannian random waves as @xmath1106 we note that the coefficients @xmath1107 are real and that the gaussian measure on the coefficients remains the real gaussian measure @xmath1005 .
the two point function is the analytic extensions to the totally real anti - diagonal in @xmath1108 is therefore @xmath1109 as in the proof of theorem [ zero ] , the current of integration over the complex zero set @xmath1110 is the @xmath1111 current defined by @xmath1112 , \psi \rangle = \int_{z_{f_n^{\c } } } \psi , \;\ ; \psi \in { \mathcal{d}}^{m-1 , m-1}(m_{\c}),\ ] ] for smooth test forms of bi - degree @xmath1113 . in terms of scalar functions
@xmath1114 we may define @xmath1115 as the measure , @xmath1112 , \psi \rangle = \int_{z_{f_n^{\c } } } \psi \omega_g^{m-1}/(m-1)!,\ ] ] where @xmath1116 is the khlermetric adapted to @xmath1117 the proof of the next result is close to the proof of theorem [ zero ] and we therefore refer to @xcite for the details : [ zeroran ] @xcite let @xmath1 be a real analytic compact riemannian manifold .
then for either of the riemannian random wave ensembles @xmath1118 \right ) \to
\frac{i } { \pi } { \partial{\bar\partial}}|\xi|_g,\;\ ; \mbox{weakly in}\;\ ; { \mathcal{d}}^ { ' ( 1,1 ) } ( b^*_{\epsilon } m).\ ] ] as mentioned above , this result shows that the complex zeros of the random waves have the same expected limit distribution found in @xcite for real analytic compact riemannian manifolds with ergodic geodesic flow .
in this final section , we review some of the more speculative conjectures relating nodal sets of both eigenfunctions and random waves to percolation theory .
the conjectures are often quoted and it therefore seems worthwhile to try to state them precisely . the only rigorous result to date regarding eigenfunctions is the theorem of nazarov - sodin on the expected number of nodal domains for random spherical harmonics @xcite ( see @xcite for a brief over - view ) .
the percolation conjectures concern the statistics of sizes of nodal domains or nodal components .
they are based on the idea that the nodal domains resemble percolation clusters .
one might measure the ` size ' of a nodal component @xmath1119 by its hypersurface area @xmath1120 , and a nodal domain @xmath1121 by its volume @xmath1122 .
let us restrict to the case of surfaces .
for the purposes of this article , we introduce the term _ length spectrum _ of the nodal set as the set @xmath1123 of lengths of its components , counted with multiplicity .
it is encoded by the empirical measure of surface areas@xmath1124 ( where @xmath1125 is the set of probability measures on @xmath125 ) , or equivalently by the length distribution function , @xmath1126 we also consider the _ area spectrum _
, @xmath1127 encoded by its empirical measure it is encoded by the empirical measure of surface areas@xmath1128 or by the area distribution function , @xmath1129 of course , there are some obvious constraints on such spectra ; e.g. in the analytic case , there could only exist @xmath1130 components with @xmath1131-length of order @xmath379 , and only a bounded number of order @xmath21 . in computer graphics of eigenfunctions on plane domains or surfaces
, one sees many ` small ' components @xmath1132 of the nodal set whose length appears to be of order @xmath1133 .
but one also sees long snaky nodal lines .
how long are they ?
do they persist as @xmath16 ? roughly speaking , one may ask what proportion of the components come in sizes with different orders of magnitude . of course , this depends on how many components there are ,
so it could be simpler to work with @xmath1134 . *
how many components have @xmath14-surface measure which is @xmath1135 for some given @xmath1136 .
it is possible that some individual nodal component has @xmath14-surface area commensurate with that of the entire nodal set , as in the lewy spherical harmonics with just two or three nodal components @xcite .
* how many components have @xmath14-surface measure ( i.e. length in dimension two ) which is bounded below by a constant @xmath1137 independent of @xmath21 ?
such components are sometimes termed percolating nodal lines " since their hypersurface volume is commensurate with the size of the macroscopic object ( i.e. @xmath28 ) .
* how many components have @xmath14-surface measure of the minimal order @xmath1133 ?
the percolation conjectures relate the asymptotic distribution of lengths of nodal components and areas of nodal domains of eigenfunctions as defined in - to lengths of boundaries and areas of percolation clusters at criticality .
there are different types of conjectures for the _ fixed frequency _ ensemble and the _ high frequency cutoff _ ensemble ( see [ rwonb ] for the definitions ) . according to the random wave hypothesis ,
the conjectures concerning the fixed frequency ensemble ( e.g. random spherical harmonics of fixed degree ) should also apply to nodal sets of eigenfunctions of quantum chaotic systems .
percolation theory is concerned with connectivity and transport in a complex system . in particular ,
it studies connected clusters of objects in a random graph . in bond percolation
the edges of the graph are independently open or closed with some probability @xmath147 .
the open edges form a subgraph whose connected components form the clusters . in site
percolation the vertices are open or closed and an open path is a path through open vertices .
the open cluster @xmath1138 of a vertex is the set of all open vertices which are connected to @xmath1139 by an open path .
there also exists an analogous continuum percolation theory for level sets of random functions
. we will assume the random functions are gaussian riemannian random waves on a surface .
the main problem is to study the connectivity properties of level sets @xmath1140 .
one imagines a random landscape of lakes and islands depending on the variable height @xmath785 of the water , the islands being the super - level sets @xmath1141 of the random functions . for high water levels ,
the islands are disconnected , but as the water level is lowered the islands become more connected . at a critical level @xmath1142
they ` percolate ' , i.e. it is possible to traverse the landscape while remaining on the land .
a review with many illustrations is given by isichenko @xcite ( see section e ( c ) , pages 980 - 984 ) .
as explained in @xcite page 984 , the contour lines of a random potential are associated to hulls of percolation clusters .
hence the area spectrum is similar to the set of sizes of connected clusters in a percolation model . in the physics literature ,
the random functions are usually functions on @xmath867 ( or possibly higher dimensional @xmath720 ) and the gaussian measure on the space of functions corresponds to a hilbert space inner product .
the hilbert space is usually taken to be a sobolev space , so that the inner product has the form @xmath1143 ( where @xmath1144 is the fourier transform of @xmath104 ) and @xmath1145 .
the case @xmath1146 is known as the gaussian free field ( or massless scalar field ) and is quite special in two dimensions since then the inner product @xmath1147 is conformally invariant .
there are rigorous results on level sets of discretizations of the gaussian free field and their continuum limits in @xcite , with authoritative comments on the physics literature . for purposes of this exposition
, we assume the riemannian random waves fall are of the types discussed in
[ rwonb ] . in all cases
, we truncate the frequency above a spectral parameter @xmath21 and consider asymptotics as @xmath16 . in this high frequency limit ,
the random waves oscillate more rapidly on the length scale @xmath1133 .
since the conjectures and results depend strongly on the chosen weight @xmath1148 , we break up the discussion into two cases as in [ rwonb ] : the high frequency cutoff ensemble and the fixed frequency ensemble . for each ensemble
we let @xmath1149 denote the expectation with respect to the gaussian measure on the relevant space of linear combinations
. then we may ask for the asymptotic behavior of the expected distribution of lengths of nodal lines , resp .
area of nodal domains @xmath1150 where @xmath1151 resp .
@xmath1152 are the empirical measures of lengths of nodal lines , resp
. areas of nodal domains .
the distribution of contour lengths of certain gaussian random surfaces over @xmath867 was studied at the physics level of rigor in @xcite .
they define the gaussian measure as @xmath1153 where the ` free energy ' is defined by latexmath:[\[f_{\zeta}(h ) = \frac{k}{2 } \int_{\r^2 } \chi(\frac{|\xi|}{\lambda } ) \ ; |\hat{h}(\xi)|^2 function to @xmath467 $ ] ( they use the notation @xmath56 for @xmath1155 in our notation ) .
when @xmath1146 , this is a truncated gaussian free field ( truncated at frequencies @xmath822 ) and its analogue on a surface @xmath1 is the riemannian random wave model with spectral interval @xmath1156 $ ] and weight @xmath1157 .
the parameter @xmath1158 is referred to as the roughness exponent in the physics literature . in the case of the gaussian free field
@xmath1146 the inner product is the dirichlet inner product @xmath1147 .
an important feature of the ensembles is scale - invariance . in the special case @xmath1146 ( and dimension two ) , the dirichlet inner product @xmath1159 is conformally invariant , i.e. invariant under conformal changes @xmath1160 of the riemannian metric .
when @xmath1161 this is not the case , but it is assumed in @xcite that the fluctuations of the random gaussian surface with height function @xmath613 are invariant under the rescaling @xmath1162 for any @xmath1163 .
the authors of @xcite then make a number of conjectures concerning the distribution of contour lengths , which we interpret as conjectures concerning @xmath1164 .
first , they consider contours ( i.e. level sets ) through a fixed point @xmath1165 and measure its length with the re - scaled arc - length measure @xmath1166 , i.e. with arclength @xmath374 in units of @xmath1133 .
they define the fractal dimension of a nodal line component as the dimension @xmath124 so that @xmath1167 where @xmath1168 is the radius of the nodal component ( i.e. half the diameter ) .
they define @xmath1169 as the probability density that the contour through @xmath1165 has length @xmath374 .
the principal claim is that @xmath1170 satisfies a power law for some exponent @xmath707 ( @xcite ( 4 ) ) .
they also defines the distribution of loop lengths @xmath1171 as the probability density that a random component has length @xmath374 .
we interpret their @xmath1172 as the density of @xmath1173 with respect to @xmath1174 on @xmath1175 .
we thus interpret their conjecture as saying that a unique weak * limit of this family of measures exists and has a density relative to @xmath1174 with a power law decay as above .
the claims are based in part on scaling properties of the contour ensemble .
they also are based in part on the expectation that , at ` criticality ' , the key percolation ` exponents ' of power laws are universal and therefore should be the same for the discrete and continuum percolation theories ( see e.g. @xcite ) . in @xcite , the authors suggest that when a certain roughness exponent @xmath79 vanishes ( the critical models ) , the continuum problem is related to the four - state potts model .
the @xmath1176-state potts model is an ising type spin model on a lattice where each spin can take one of @xmath1176 values .
it is known to be related to connectivity and percolation problems on a graph @xcite .
they compute @xmath1177 by relating both to another exponent @xmath1178 defined by a contour correlation function " @xmath1179 , which measures the probability that points at @xmath1180 lie on the same contour loop . they claim that @xmath1181 .
they claim that @xmath1182 and @xmath1183 . as a result , @xmath1184 . from the mapping to the four - state potts model , they conclude that @xmath1185
. there exist rigorous results in @xcite relating discretizations of the gaussian free field ( rather than high frequency truncations ) to the percolation models .
they prove that in various senses , the zero set of the discrete gaussian free field tends to an @xmath1186 curve .
it does not seem to be known at present if zero sets of the high frequency truncation of the gaussian free field also tends in the same sense to an @xmath1186 curve .
note that the sle curves are interfaces and that one must select one component of the zero set that should tend to an sle curve . there might exist modified conjectures regarding cle curves . to determine the ` critical exponents ' in continuum percolation , it is tempting to find a way to ` map ' the continuum problem to a discrete percolation model . a geometric ` map ' from a random wave to a graph is to associate to the random function its morse - smale decomposition , known in the physics literature as the morse skeleton " ( see [ decomp ] or @xcite for an extensive exposition ) . as discussed in @xcite , and as illustrated in figure 10 of @xcite , the morse complex of the random function plays the role of the lattice in lattice percolation theory .
we now consider riemannian random waves of asymtotically fixed frequency @xmath21 , such as random spherical harmonics of fixed degree or euclidean random plane waves of fixed eigenvalue . in this case
the weight is a delta function at the frequency .
one would expect different behavior in the level sets since only one frequency is involved rather than the superposition of waves of all frequencies @xmath822 .
a recent exposition in the specific setting of random euclidean eigenfunctions of fixed frequency is given by @xcite .
the level sets play the role of open paths .
super - level sets are compared to clusters of sites in a critical 2d percolation model , such as bond percolation on a lattice .
each site may of the percolation model may be visualized as a disc of area @xmath1187 , i.e. as a small component .
the nodal domains may be thought of as connected clusters of a number @xmath2 such discs . since nodal domains are connected components in which the eigenfunction is either positive @xmath1188 or negative @xmath1189 , they are analogous to clusters of ` open ' or ` closed ' vertices .
the main conjectures in this fixed frequency ensemble are due to e. bogolmony and c. schmidt @xcite .
they conjecture that the continuum percolation problem should belong to the same universality class as the potts model at a certain critical point ( where @xmath1176 is related to a certain temperature ) for a large rectangular lattice and that the nodal lines in the @xmath1190 limit tend to @xmath1191 curves .
this is similar to the predictions of @xcite but for a very different ensemble where there is little apriori reason to expect conformal invariance in the limit .
there are parallel conjectures in @xcite for zero - vorticity isolines in 2d turbulence , which are also conjectured to tend to @xmath1191 curves .
they remark ( page 127 ) that this limit is surprising since continuous percolation models assume short - correlations in the height functions whereas the vorticity field correlations decay only like @xmath1192 .
they write , when the pair correlation function falls off slower than @xmath1193 , the system is not expected generally to belong to the universality class of uncorrelated percolation and to be conformally invariant " .
the same remarks apply to the fixed frequency ensemble , where the correlation function is the spectral projection @xmath1194}(x , y)$ ] for a fixed frequency . in this case , the correlations decay quite slowly as @xmath1195 ; we refer to @xcite for this background and also for an argument why the nodal sets should nevertheless resemble conformally invariant @xmath1196 curves
. if the nodal lines in the fixed frequency model are equivalent to the critical percolation model , then the ` probability ' of finding a nodal domain of area @xmath374 should decay like @xmath1197 where @xmath1198 ( see @xcite , p. 52
for the percolation theory result ) . under some shape assumptions adopted in @xcite ,
it is equivalent that the probability of finding clusters consisting of @xmath2 discs is of order @xmath1199 . for random spherical harmonics
, one may ask for the probability that a spherical harmonic of degree @xmath143 has size @xmath2 . for a fixed @xmath1200 with simple eigenvalues ,
this notion of probability from percolation theory does not make sense , but we might assume that the number of of nodal components is of order @xmath37 and ask what proportion of the nodal components has size @xmath379 . to obtain a percolating nodal line , one would need a cluster with @xmath1201 sites , and thus the proportion of such nodal components to the total number would be of order @xmath1202 .
thus , if there are @xmath1203 total components , the number of such components would be around @xmath1204 , so the model seems to predict that such macroscopic nodal lines are quite rare .
it also predicts that the ` vast majority ' of nodal components are close to the minimal size , which does not seem so evident from the computer graphics .
p. bleher , b shiffman , and s. zelditch , universality and scaling of zeros on symplectic manifolds .
_ random matrix models and their applications _ ,
3169 , math .
, 40 , cambridge univ . press ,
cambridge , 2001 .
j. bourgain , _ geodesic restrictions and @xmath0-estimates for eigenfunctions of riemannian surfaces _ , linear and complex analysis , 2735 , amer .
ser . 2 , 226 , amer .
soc . , providence , ri , 2009 .
s. y. cheng , eigenfunctions and eigenvalues of laplacian .
differential geometry ( proc . sympos .
pure math .
xxvii , stanford univ . ,
stanford , calif . , 1973 ) ,
part 2 , pp .
soc . , providence , r.i . , 1975
. h. donnelly and c. fefferman , growth and geometry of eigenfunctions of the laplacian .
analysis and partial differential equations , 635655 , lecture notes in pure and appl . math .
, 122 , dekker , new york , 1990 .
f. golse , e. leichtnam , and m. stenzel , intrinsic microlocal analysis and inversion formulae for the heat equation on compact real - analytic riemannian manifolds .
sci . cole norm .
( 4 ) 29 ( 1996 ) , no . 6 , 669736 .
d. jerison and g. lebeau , nodal sets of sums of eigenfunctions .
harmonic analysis and partial differential equations ( chicago , il , 1996 ) , 223239 , chicago lectures in math .
chicago press , chicago , il , 1999 .
n. nadirashvili , d. jakobson , and j.a .
toth , geometric properties of eigenfunctions .
( russian ) uspekhi mat .
nauk 56 ( 2001 ) , no .
6(342 ) , 6788 ; translation in russian math .
surveys 56 ( 2001 ) , no .
6 , 10851105 , open problems in geometry . _
differential geometry : partial differential equations on manifolds _
( los angeles , ca , 1990 ) , 128 , proc .
pure math . , 54 , part 1 , amer .
soc . , providence , ri , 1993 . | this is a survey of recent results on nodal sets of eigenfunctions of the laplacian on riemannian manifolds .
the emphasis is on complex nodal sets of analytic continuations of eigenfunctions .
key words : laplacian , eigenvalues and eigenfunctions , quasi - mode , wave equation , frequency function , doubling estimate , nodal set , quantum limit , @xmath0 norm , geodesic flow , quantum complete integrable , ergodic , anosov , riemannian random wave .
ams subject classification : 34l20 , 35p20 , 35j05 , 35l05 , 53d25 , 58j40 , 58j50 , 60g60 .
let @xmath1 be a ( usually compact ) riemannian manifold of dimension @xmath2 , and let @xmath3 denote an orthonormal basis of eigenfunctions of its laplacian , @xmath4 here @xmath5 where @xmath6 is the volume form of @xmath1 . if @xmath7 we impose dirichlet or neumann boundary conditions .
when @xmath1 is compact , the spectrum of @xmath8 is discrete and can be put in non - decreasing order @xmath9 .
the eigenvalues @xmath10 are often termed energies while their square roots @xmath11 are often termed the frequencies . the nodal set of an eigenfunction @xmath12
is the zero set @xmath13 the aim of this survey is to review some recent results on the @xmath14-surface measure and on the yet more difficult problem of the spatial distribution of the nodal sets , i.e. the behavior of the integrals @xmath15 as @xmath16 . here
, @xmath17 denotes the riemannian hypersurface volume form on @xmath18.more generally , we consider the same problems for any level set @xmath19 where @xmath20 is a constant ( which in general may depend on @xmath21 ) .
nodal sets are special level sets and much more attention has been devoted to them than other level sets , but it is often of interest to study general level sets and in particular ` high level ' sets or excursion sets . we have recently written surveys @xcite on the global harmonic analysis of eigenfunctions , which include some discussion of nodal sets and critical point sets . to the extent
possible , we hope to avoid repeating what is written there , but inevitably there will be some overlap .
we refer there and @xcite for background on well - established results .
we also decided to cover some results of research in progress ( especially from @xcite , but also on @xmath22 quantum ergodic theory ) .
we generally refer to the results as ` conjectures ' even when detailed arguments exist , since they have not yet been carefully examined by others .
there are two basic intuitions underlying many of the conjectures and results on eigenfunctions : * eigenfunctions of @xmath23-eigenvalue @xmath24 are similar to polynomials of degree @xmath21 .
in particular , @xmath25 is similar to a real algebraic variety of degree @xmath21 .
+ of course , this intuition is most reliable when @xmath1 is real analytic .
it is quite unclear at this time how reliable it is for general @xmath26 metrics , although there are some recent improvements on volumes and equidistribution in the smooth case . *
high frequency behavior of eigenfunctions reflects the dynamics of the geodesic flow @xmath27 of @xmath28 . here , @xmath29 is the unit co - sphere bundle of @xmath1 .
+ when the dynamics is chaotic " ( highly ergodic ) , then eigenfunctions are de - localized and behave like gaussian random waves of almost fixed frequency .
this motivates the study of gaussian random wave models for eigenfunctions , and suggests that in the ` chaotic case ' nodal sets should be asympotically uniformly distributed .
+ when @xmath30 is completely integrable , model eigenfunctions are highly localized and their nodal sets are often exhibit quite regular patterns . the latter heuristic is not necessarily expected when there exist high multiplicities , as for rational flat tori , and then some weaker randomness can enter . both of these general intuitions lead to predictions about nodal sets and critical point sets .
most of the predictions are well beyond current or forseeable techniques to settle .
a principal theme of this survey is that the analogues of such ` wild ' predictions can sometimes be proved for real analytic @xmath1 if one analytically continues eigenfunctions to the complexification of @xmath28 and studies complex nodal sets instead of real ones . _ as with algebraic varieties , nodal sets in the real analytic case are better behaved in the complex domain than the real domain .
that is , zero sets of analytic continuations of eigenfunctions to the complexification of @xmath28 behave like complex algebraic varieties and also reflect the dynamics of the geodesic flow . _
it is well - known that the complexification of @xmath28 can be identified with a neighborhood of the zero - section of the phase space @xmath31 .
that is one reason why dynamics of the geodesic flow has greater impact on the complex nodal set .
we will exhibit a number of relatively recent results ( some unpublished elsewhere ) which justify this viewpoint : * theorem [ bnp ] , which shows that complex methods can be used to give upper bounds on the number of nodal components of dirichlet or neumann eigenfunctions which touch the boundary " of a real analytic plane domain . * theorem [ zero ] on the limit distribution of the normalized currents of integration @xmath32\ ] ] over the complex zero sets of `` ergodic eigenfunctions '' in the complex domain .
* theorem [ zmeasure ] and corollary [ zdsm ] , which show that the similar currents for analytic continuations of `` riemannian random waves '' tend to the same limit almost surely .
thus , the prediction that zero sets of ergodic eigenfunctions agrees with that of random waves is correct in the complex domain . * sharper results on the distribution of intersections points of nodal sets and geodesics on complexified real analytic surfaces ( theorem [ maincor ] ) .
our analysis of nodal sets in the complex domain is based on the use of complex fourier integral techniques ( i.e. generalized paley - wiener theory ) .
the principal tools are the analytic continuation of the poisson - wave kernel and the szeg kernel in the complex domain .
they become fourier integral operators with complex phase and with wave fronts along the complexified geodesic flow .
one can read off the growth properties of complexified eigenfunctions from mapping properties of such operators .
log moduli of complexified spectral projectors are asymptotically extremal pluri - subharmonic functions for all @xmath1 .
these ideas are the basis of the articles @xcite .
such ideas have antecedents in work of s. bernstein , baouendi- goulaouic , and donnelly - fefferman , guillemin , f.h .
lin ( among others ) .
we note that the focus on complex nodal sets only makes sense for real analytic @xmath1 .
it is possible that one can study `` almost analytic extensions '' of eigenfunctions for general @xmath26 metrics in a similar spirit , but this is just a speculation and certain key methods break down when @xmath33 is not real analytic .
hence the results in the @xmath26 case are much less precise than in the real analytic case .
it should also be mentioned that much work on eigenfunctions concerns ground states , i.e. the first and second eigenfunctions .
unfortunately , we do not have the space or expertise to review the results on ground states in this survey . for a sample
we refer to @xcite .
further , many if not all of the techniques and results surveyed here have generalizations to schrdinger operators @xmath34 . for the sake of brevity
we confine the discussion to the laplacian .
the first notational issue is whether to choose @xmath23 to be the positive or negative laplacian .
the traditional choice @xmath35 makes @xmath23 is negative , but many authors call @xmath36 the laplacian to avoid the minus signs . also , the metric @xmath33 is often fixed and is dropped from the notation .
a less traditional choice is to denote eigenvalues by @xmath37 rather than @xmath21 .
it is a common convention in microlocal analysis and so we adopt it here .
but we warn that @xmath21 is often used to denote @xmath8-eigenvalues as is @xcite .
we sometimes denote eigenfunctions of eigenvalue @xmath24 by @xmath12 when we only wish to emphasize the corresponding eigenvalue and do not need @xmath12 to be part of an orthonormal basis . for instance , when @xmath23 has multiplicities as on the standard sphere or rational torus , there are many possible orthonormal bases . but estimates on @xmath38 do not depend on whether @xmath12 is included in the orthonormal basis .
thanks to c. d. sogge and b. shiffman for helpful comments on the exposition , and to s. dyatlov for a stimulating discussion of @xmath22 quantum ergodicity . |
in the quiet sun , the most prominent photospheric feature is the surface granulation pattern .
observations in white light show bright granules with a typical size of @xmath1 mm , embedded in a network of dark intergranular downflow lanes and vertices .
the granulation pattern is non - stationary : studies of time - sequences of the granulation pattern in the quiet sun show that individual features evolve with an average lifetime of about @xmath2 minutes ( e.g. * ? ? ?
while intensity images in the visible continuum taken near disk - center show bright granules and comparatively dark intergranular boundaries , observations taken in the wings of the ca ii h & k lines exhibit a similar pattern , but with reversed intensity contrast @xcite .
this effect , known as the _ reversed granulation _ , has also been detected in intensity measurements of photospheric spectral lines @xcite .
these observations suggest that the reversal of the intensity contrast occurs already at a height of about @xmath3 km above the mean geometric level where @xmath4 .
recently , @xcite acquired simultaneous , speckle - reconstructed images of quiet - sun granulation in the g - band and the ca ii h line .
on the basis of cross - correlations between time sequences of intensity images in these two spectral ranges they found that the anticorrelation between the two patterns is highest if the ca ii h intensity maps are time - delayed by @xmath5 minutes with respect to the g - band intensity maps . by computing anticorrelation coefficients between ( 1 ) intensity images in the continuum and in the fe i 7090 line and ( 2 ) intensity images in the continuum and in the wing of the ca ii 8542 line , @xcite also found that the strongest anticorrelation occurs for a time delay of @xmath6 min .
@xcite extended the work of @xcite by calculating synthetic intensity images in the blue continuum and in the ca ii h wing ( in the lte approximation ) from a 3-dimensional radiative hydrodynamic simulation . from the unsmoothed synthetic intensity images , they found that the anticorrelation between the continuum images and the ca ii h wing images does not vary significantly between a time delay of @xmath7 and @xmath8 minutes .
the anticorrelation decreased for longer time delays . when the images were smoothed to @xmath9 arcsec resolution , a time - delay of @xmath5 minutes yielded a significantly higher value of the anticorrelation compared to zero time delay .
@xcite speculated that magnetic fields play no major role in the formation of the reversed granulation .
the results of @xcite is evidence supporting this view .
numerical simulations have firmly established the solar granulation as a radiative - convective phenomenon @xcite .
granules are upwellings of hot , high - entropy material originating from the convection zone and overshooting into the stably stratified photosphere .
radiative losses make the overshooting material relatively cold and dense .
its overturning motion supplies the network of intergranular downflows .
simulations have also revealed that the pattern of horizontal temperature fluctuations in the middle / upper photosphere is inverted compared to the corresponding pattern at the base of the photosphere . whereas at the base of the photosphere , the granules are relatively hot compared to the intergranular network , the former are relatively cooler for @xmath10 .
it is well known that this reversal in the temperature fluctuations in the photosphere is responsible for the reversal in the observed intensity pattern @xcite .
what is less clear is the physical mechanism which leads to the reversal of the temperature fluctuations with height .
it is sometimes incorrectly assumed that fluid travelling in the optically thin layers of the photosphere does so adiabatically . as pointed out by @xcite
, this can not be true because material with an original temperature of @xmath11 k penetrating and expanding adiabatically in the photosphere would reach a temperature of @xmath12 k after traversing only three pressure scale heights .
such low temperatures are not detected in the photosphere .
@xcite indeed find that fluid rising in the photosphere is typically reheated by the radiation field .
this leads one to conclude that non - adiabatic effects , such as radiative energy exchange , must play a crucial role in the formation of the reversed temperature fluctuations .
the aim of the present paper is to investigate the physical origin of the reversal of temperature fluctuations .
as we shall see in the following sections , radiative energy exchange is indeed key to the temperature fluctuation reversal . by following in detail the time histories of fluid elements penetrating into the photosphere , we identify how radiative heating and cooling , in tandem with the horizontal pressure fluctuations , causes the reversed granulation .
the paper is structured as follows . in [ sec : setup ] , we describe the set of governing equations , the numerical methods used and the simulation setup . in
[ sec : results ] , we discuss results from the numerical simulation , including the mean stratification of the 3d model ( [ subsec : mean_stratification ] ) , the temperature structure in the photosphere ( [ subsec : reversed_granulation_structure ] ) and reveal the physical origin of the reversed granulation ( [ subsec : origin_reversed_granulation ] ) . finally , in
[ sec : conclusion ] , we discuss how the present work dispels previous misconceptions and improves our understanding of the phenomenon .
for a realistic treatment of the dynamics and energetics in the near - surface layers of the non - magnetic solar atmosphere , it is important to consider the effects of compressibility , energy exchange via radiative transfer and partial ionization @xcite .
the hydrodynamics equations , the radiative transfer equation ( rte ) and the equation of state ( eos ) including the effects of partial ionization constitute the system of governing equations for our model . since we are confining our attention to the near - surface layers of the convection zone and the photosphere , the approximation of local thermodynamic equilibrium ( lte ) suffices for the treatment of the radiative transfer .
the hydrodynamics equations include the continuity equation , the momentum equation and the energy equation , viz .
@xmath13 \right ) = \nonumber\\ & & \nabla\cdot(\mathbf{v}\cdot\underline{\underline{\tau } } ) + \nabla\cdot(k\nabla t ) + \varrho ( \mathbf{g}\cdot\mathbf{v } ) + q_{\rm rad } \label{eqn : energy},\end{aligned}\ ] ] where @xmath14 is the fluid velocity , @xmath15 the mass density and @xmath16 the internal energy density per unit mass .
the operator @xmath17 denotes the tensor product . in the near - surface layers of the sun ,
the gravitational acceleration is almost constant and we take @xmath18 cm s@xmath19 . the hydrodynamics equations ( [ eqn : continuity])-([eqn : energy ] ) must be supplemented by the appropriate constitutive relations describing the material properties of the gas .
these include the equation of state ( eos ) , functions specifying the viscous stress tensor @xmath20 and thermal conductivity @xmath21 , as well as functions specifying the radiative properties of the gas ( e.g. source function @xmath22 , opacity @xmath23 etc . ) . the frequency - integrated radiative heating term appearing on the r.h.s . of eq .
( [ eqn : energy ] ) is given by @xmath24 where @xmath25 is the angle - averaged specific intensity and @xmath26 the angular - averaged source function .
for our radiative hydrodynamics simulations , we have used the _ _ m__ps/__u__niversity of chicago _ _ ra__diative _ _ m__hd ( muram ) code @xcite . in the absence of magnetic fields ,
the equations solved by muram reduce to the equations given in the preceding section . in the following ,
we briefly describe the numerical methods implemented in muram . for further details ,
we refer the reader to @xcite and @xcite .
muram integrates the discretized equations ( [ eqn : continuity])-([eqn : energy ] ) on a 3-dimensional cartesian grid , with constant grid - spacing in each cartesian direction .
the spatial discretization of the equations is based on a fourth - order centered - difference scheme on a five - point stencil .
time - stepping is carried out using an explicit fourth - order runge - kutta scheme , with the maximum size of the time step determined by a modified cfl - criterion . in order to prevent the nonlinear cascade of energy towards the grid scale ,
artificial diffusivities are implemented to stabilize the simulations . a combination of _ hyper- _ and _ shock - resolving _ diffusivities are used @xcite .
hyperdiffusivities are implemented in such a way , that the diffusion terms associated with them are comparable in magnitude to the inertial terms only at length scales close to the grid - spacing . for the thermal and viscous coefficients @xmath21 and @xmath27 , the hyperdiffusivities are defined using the scheme described in @xcite .
the rte is solved with the short - characteristics method on a 3d cartesian grid , with the grid centers of this radiative grid defined by the cell corners of the grid used for discretizing the hydrodynamics equations ( [ eqn : continuity])-([eqn : energy ] ) . for each point on this radiative grid ,
the rte is solved over @xmath28 ray directions ( @xmath29 per octant ) .
the first moment of @xmath30 over these rays provides the radiative flux density @xmath31 , which in turn allows @xmath32 to be calculated .
muram can treat the radiative transfer in the grey or non - grey case ( see * ? ? ?
* and references therein ) . for this study
, we have carried out two simulations , namely a grey run and a non - grey run .
the results presented in the rest of this paper correspond to the non - grey case .
the main difference between the results of the two cases it that the non - grey case yields smaller temperature fluctuations in the photosphere than the grey case .
this difference arises because the grey case underestimates the smoothing effects of radiative transfer in the optically thin layers of the photosphere in comparison with the non - grey case @xcite . our explanation for the reversed granulation phenomenon ( see [ subsec : reversed_granulation_structure ] )
, however , applies to both cases .
the eos relations @xmath33 and @xmath34 are used by muram in the form of look - up tables .
the look - up tables were evaluated by solving the saha - boltzmann equations .
the first ionization of the @xmath35 most abundant elements was considered .
for a detailed account of how the tables were evaluated , we refer the reader to appendix a of @xcite . in order to obtain a thermally relaxed 3d model of the upper convection zone and the photosphere ,
we have chosen the following setup .
the horizontal size of the simulation domain is @xmath28 mm @xmath36 mm , with each direction spanned by @xmath37 and @xmath38 grid cells , respectively .
the height of the domain is @xmath39 mm , and is spanned by @xmath40 grid cells .
the grid - spacing in the horizontal and vertical directions is @xmath41 and @xmath42 km , respectively .
periodic boundary conditions are assumed for the vertical side boundaries .
the bottom boundary is open and allows mass to flow smoothly into and out of the domain ( see * ? ? ?
* ) . for the initial condition
, we took the 1d height profiles of @xmath15 and @xmath43 from the mixing length model of @xcite and introduced a plane - parallel stratification into the simulation domain . to break the symmetry
, we imposed pressure perturbations on the order of a few percent .
we then ran the simulation until we obtained a thermally relaxed , statistically stationary state .
the horizontally averaged profiles of pressure and temperature as functions of height are shown in fig .
[ fig : pt_profiles ] , respectively , as solid and dashed curves .
the geometrical height @xmath44 corresponds to the geometrical level where the continuum optical depth at @xmath45 nm is , on average , unity .
the height dependence of the mean profiles are in general agreement with the simulation results of @xcite . figure [ fig : nabla_profiles ] shows the logarithmic temperature gradients determined from the mean profiles : @xmath46 @xmath47 ( dotted curve ) is the actual average temperature gradient in the simulation and @xmath48 ( solid black curve ) is the adiabatic temperature gradient .
@xmath48 describes the variation of temperature of a fluid element undergoing adiabatic expansion or compression .
.,scaledwidth=70.0% ] ( dotted curve ) , @xmath48 ( solid black ) and the superadiabaticity @xmath49 ( dashed curve).,scaledwidth=70.0% ] the stability of the stratification to convective motion is affected by changes in the ionization state .
the superadiabaticity of the stratification is defined as @xmath50 and is plotted as a dashed curve in fig .
[ fig : nabla_profiles ] . in the photosphere ,
the chemical species are almost completely monatomic .
for such a gas , @xmath51 and hence , @xmath52 . with increasing depth into the convection zone , the ionization fraction of hydrogen ( and traces of other species ) increases .
this has the effect of decreasing the adiabatic temperature gradient , so that @xmath53 while @xmath54 remains positive . in other words , changes in the ionization state of the gas reduces @xmath48 and thus enhances buoyancy driving . in this section ,
we study the reversed granulation in our 3d model . for lines that form under approximate lte conditions , such as some fe i lines and the wings of the ca ii h & k lines , fluctuations in line intensity are a proxy for diagnosing temperature fluctuations .
the intensity image therefore maps the temperature distribution in the layer of the atmosphere in which the corresponding part of the line is formed .
an inspection of the temperature fluctuations at different heights of the atmosphere allows one to determine the height at which the temperature fluctuations start to reverse . the upper row of fig .
[ fig : reverse_granulation ] shows the relative temperature fluctuations ( @xmath55 ) at the surfaces @xmath56 ( left column ) and @xmath57 ( right column ) .
the lower row gives the vertical velocity at the same surfaces . on average ,
the @xmath57 surface is elevated above the @xmath56 surface by @xmath58 km . at @xmath56 ,
one finds the normal granulation pattern : the granules are hotter than the intergranular lanes . at @xmath59 ,
the reversed granulation is already clearly discernible : the lanes are regions of temperature excess . when we inspect the velocity structure , we find that the pattern at @xmath57 is very similar to that at @xmath60 , albeit with smaller flow speeds .
the upflows in the higher layers result from the overshooting of upwelling gas from the convection zone .
thus , the reversal of the pattern of temperature ( and intensity ) fluctuations is _ not _ accompanied by a reversal in the velocity pattern ( e.g. see * ? ? ?
in order to determine the geometrical height at which the reversed granulation begins , we consider the vertical velocity and temperature as functions of @xmath61 for vertical lines of sight ( l.o.s . ) . distinguishing the l.o.s .
profiles according to the sign of the vertical velocity at @xmath56 , we separately determine averages for profiles of the temperature fluctuations in granules and in the intergranular downflow network , respectively .
the fluctuations are evaluated relative to the overall average temperature profile .
these two quantities are plotted as functions of @xmath61 in fig .
[ fig : reversal_geometric ] .
diamonds indicate the mean temperature fluctuation in the granules whereas crosses indicate the mean temperature fluctuation in the intergranular lanes .
these two are not simply the negative of each other because the cells have a larger area fraction than the boundaries . in the convection zone and in the lower photosphere ,
upflowing material in the granules are relatively hot compared to downflowing material in the lanes .
the reversal of the horizontal temperature fluctuations occurs , on average , in the middle photosphere at a height of @xmath0 km .
above that height , the intergranular lanes are hotter than the granules with a temperature excess of @xmath62-@xmath63 k. ) of granules ( diamonds ) and of intergranular lanes ( crosses ) as a functions of geometrical height . at @xmath0 km , the cells and boundaries have , on average , the same temperature . above this geometrical height ,
the intergranular lanes are relatively hotter than the granules.,scaledwidth=70.0% ] ) of granules ( diamonds ) and of intergranular lanes ( crosses ) as a functions of optical depth .
the reversed granulation appears at optical depths of about @xmath64.,scaledwidth=70.0% ] we have also determined the temperature profiles in granules and in the intergranular network as functions of optical depth along vertical lines of sight .
the average temperature fluctuations of the granules and the intergranular lanes as functions of @xmath65 are indicated by diamonds and crosses respectively in fig .
[ fig : reverse_temperature_fluctuation ] . in the convection zone ( @xmath66 ) , the temperature fluctuations at a given geometrical height are much larger than the temperature fluctuations at constant optical depth . as pointed out by @xcite , this is due to the high temperature sensitivity of the h@xmath67 contribution to the continuum opacity . in this figure ,
the two curves cross each other at @xmath68 . on average ,
the surface of constant optical depth at @xmath68 is elevated above the @xmath56 surface by @xmath69 km .
the fact that the two different averaging procedures yield somewhat different heights ( @xmath0 km and @xmath70 km , respectively ) for the layer of the photosphere where the temperature fluctuations reverse their sign is easily understood . in the former case , we calculated the horizontal temperature fluctuations relative to the mean temperature of the layer at different geometrical heights . in the latter case , the temperature fluctuations correspond to deviations from the mean temperature at surfaces of constant optical depth .
since these surfaces are corrugated ( e.g. the rms fluctuation of the geometrical height of the @xmath56 surface is @xmath71 km ) , temperature fluctuations across such a surface only approximate the horizontal temperature fluctuations at the same geometrical height . in terms of both geometrical height and optical depth ,
our determination of the level at which the reversal occurs is in good agreement with previous observational work @xcite .
the analysis in the previous section confirmed the reversal of horizontal temperature fluctuations in the middle photosphere .
now we proceed to explain the physics behind this phenomenon .
there are two important aspects to the explanation .
firstly , we need to understand the nature of convective overshoot of granular fluid into the photosphere : namely , what is the balance of forces controlling the dynamics of convective penetration and overturn ? secondly , we need to understand the thermodynamical processes experienced by the overshooting material . in order to address these issues , we have examined the histories of tracer fluid elements in our simulation . at time @xmath72 ( corresponding to the snapshot shown in fig .
[ fig : reverse_granulation ] ) , tracer fluid elements were introduced just beneath the surface inside a granule and propagated according to the simulated velocity field .
the physical properties of the fluid elements ( e.g. @xmath73 etc . ) at their actual positions were determined by tricubic interpolation from the values in the adjacent cells @xcite .
figure [ fig : fountain_qtot ] shows a subvolume of the simulation domain in the neighbourhood of the granule centered at @xmath74=[7,8]$ ] mm in fig .
[ fig : reverse_granulation ] .
the greyscale image at @xmath44 indicates the vertical velocity at @xmath44 at the instant when the tracer fluid elements were released ( @xmath72 ) .
the colour - coding of the trajectories indicate the sign of @xmath32 : dark blue means @xmath75 ( cooling ) , yellow means @xmath76 ( heating ) .
the set of trajectories resembles a fountain : the fluid elements originate from the granule interior , travel up to the photosphere and eventually overturn and descend into the downflow network in the convection zone .
first we concentrate on the motion and dynamics of the fluid elements . as fig .
[ fig : fountain_qtot ] shows , in spite of the strongly stable average stratification , the fluid elements are able to penetrate into the photosphere .
the shape of the trajectories tell us that in the photosphere , they experience a deceleration in the vertical direction and a deflection towards the edge of the granule . in the following ,
we give a brief description of the force balance which drives the horizontal granular outflows and the feeding of the downflows in the intergranular lanes .
inspection of the horizontal pressure fluctuations in the photospheric layers of our simulation reveal that local pressure hills in the interior of granules are responsible for the buoyancy braking of the upflows and the deflection of material horizontally outwards from the centres of granules . at the interface between two granules , the horizontal outflows from the granules
meet and develop localized pressure ridges .
these localized regions of pressure enhancement serve to retard the horizontal granular outflows .
so , when a fluid element in the granular outflow approaches the intergranular boundary , it is decelerated by the pressure gradient .
finally , the pressure enhancement at the boundaries , together with gravity , work in tandem to accelerate the fluid element downwards , thereby feeding the intergranular downflow network .
these results are in accordance with results from 2d idealized simulations of compressible convection @xcite and 3d near - surface convection simulations @xcite . to avoid confusion in the following discussion , at this point we would like to clarify the following terminology : by heating ( cooling ) , we refer to a non - adiabatic process which increases ( decreases ) the specific entropy of a fluid element . for example , when a fluid element absorbs more radiation than it emits , it is heated and gains entropy . a fluid element which is compressed adiabatically , however , will have a higher final temperature but is not heated .
in the absence of viscous dissipation or thermal conduction , the equation governing the evolution of the specific entropy of a fluid element is @xmath77 where @xmath78 denotes the lagrangian derivative @xcite .
the corresponding equation governing the temperature of a fluid element is @xmath79 where @xmath80 is chandrasekhar s third adiabatic exponent .
two effects contribute to changes in the gas temperature of a fluid element .
the first term on the r.h.s . of eq .
( [ eqn : temperature_evolution2 ] ) describes the role of adiabatic expansion or compression .
clearly , the effect of adiabatic expansion ( @xmath81 ) is to decrease the gas temperature .
the second term on the r.h.s . of eq .
( [ eqn : temperature_evolution2 ] ) describes the role of radiative heating .
as expected , radiative heating ( cooling ) leads to an increase ( decrease ) of the gas temperature .
the relative importance of the two effects in controlling the fluid temperature depends on their characteristic timescales .
inspection of eq .
( [ eqn : temperature_evolution2 ] ) motivates us to define the dynamical and radiative timescales respectively as @xmath82 in the case that @xmath83 , the motion of a fluid element is essentially adiabatic . in the opposite extreme , where @xmath84 , the fluid element is always in radiative equilibrium with its surroundings .
now we proceed to discuss the thermodynamic history of fluid elements overshooting into the photosphere . owing to the large opacity for @xmath66 , fluid in the upwellings in the convection zone
ascend almost adiabatically .
the decrease of a fluid element s temperature during its rise in the convection zone is due to adiabatic expansion .
as it approaches the photosphere , the opacity drops rapidly and the fluid element loses entropy by radiative cooling .
thereafter , the fluid element overshoots into the stably stratified photosphere . although the fluid element is now optically thin , it is not completely transparent to radiation and continues to exchange energy with its surroundings . in regions above granules , ( consisting of overshooting material ) ,
radiative energy exchange predominantly leads to heating .
this effect can be seen in the colour - coding of the trajectories in fig .
[ fig : fountain_qtot ] .
this figure clearly indicates that as the fluid elements penetrate into the photosphere ( and as they are diverted toward the intergranular lanes ) , they experience radiative heating .
@xcite also found that rising fluid in the photosphere is systematically reheated by the radiation field ( see fig .
18 of their paper ) .
the reason for this behaviour is the thermostatic action of the radiation field to keep the fluid elements close to radiative equilibrium .
the radiative equilibrium temperature @xmath85 of a fluid element embedded in a radiation field is defined as the temperature at which the fluid element emits and absorbs radiation in equal amounts ( i.e. @xmath86 ) .
one can make a simple estimate for the _ lower _ limit of @xmath85 by assuming the fluid element to be located at @xmath87 . in the approximation of grey radiative transfer ,
@xmath23 is replaced by the mean opacity @xmath88 , so that eq .
( [ eqn : qrad_two ] ) becomes @xmath89,\label{eqn : qrad_mean_kappa}\ ] ] where @xmath90 is the frequency - integrated planck function and @xmath91 is the angle - averaged , frequency - integrated intensity .
assume that the solar atmosphere has plane - parallel symmetry .
in this case , @xmath30 is symmetric about the @xmath61-axis . making use of the eddington - barbier approximation
, we obtain @xmath92 { \rm d}\nu \\ & = & \frac{\sigma}{2\pi } \int_{-1}^{1 } ( t|_{\tau=\mu})^4 { \rm d}\mu,\label{eqn : j_expression}\end{aligned}\ ] ] where @xmath93 . here
@xmath94 is the direction cosine vector and @xmath95 is the angle between @xmath94 and @xmath96 . by definition
, @xmath85 is such that @xmath97 . to evaluate @xmath98 from eq .
( [ eqn : j_expression ] ) , we use the average temperature profile @xmath99 from our model .
this yields @xmath100 k. this result tells us why granular material ascending in the photosphere _ must _ be radiatively heated . consider a fluid element that is rising in the photosphere .
if the fluid element rises adiabatically , its temperature would drop so quickly with height , that it becomes cooler than the radiative equilibrium temperature .
as soon as this happens , the fluid element will absorb more radiation than it emits ( @xmath101 ) . by the same token ,
consider a fluid element in the photosphere which has been transported from the interior of a granule towards an intergranular lane .
as it descends , it is compressed .
if it were to descend adiabatically , its temperature would soon be above @xmath85 , which means it must then cool radiatively .
figure [ fig : deviation_from_t_re ] provides an illustration of how overshooting fluid elements become cooler than @xmath85 as they penetrate into the photosphere , and subsequently become hotter than @xmath85 as they descend towards the convection zone . in the figure ,
the height @xmath61 of a fluid element is plotted as a function of the horizontal displacement from its original position ( i.e. @xmath102^{1/2}$ ] ) . in both panels ,
the fluid elements travel from the left to the right of the plot . in the upper panel ,
the trajectories are colour - coded according to the sign of @xmath32 . in the lower panel , they are colour - coded according to the instantaneous , local value of @xmath103 .
@xmath85 is evaluated by keeping track of @xmath104 and @xmath105 ( @xmath106 refers to the index of the frequency bin ) along the trajectories and then finding a temperature @xmath107 such that @xmath108 .
as a rising fluid element approaches the base of the photosphere , its temperature can be up to @xmath109 k higher than @xmath85 .
this causes the fluid element to radiate intensely .
once the fluid element has reached the optically thin layers of the photosphere , adiabatic expansion and radiative heating compete to control the temperature of the fluid element .
the relative importance of these two effects ( measured by the ratio of @xmath110 and @xmath111 ) determine whether the fluid element will be driven towards or away from radiative equilibrium .
for a fluid element penetrating into the middle photosphere in a granular environment , we find that @xmath112 .
this means that the decrease in fluid temperature due to expansion of the fluid element is only partially compensated by radiative heating .
consequently , its temperature decreases and it departs further away from @xmath85 . in fig .
[ fig : deviation_from_t_re ] , we can see a general trend for @xmath103 to become more negative as a fluid element rises . for fluid elements that penetrate beyond @xmath113 km
, @xmath114 can reach a value of @xmath115 k. figure [ fig : hysteresis_curves ] shows the profiles of @xmath116 , @xmath107 and @xmath15 along one of the trajectories shown in fig .
[ fig : fountain_qtot ] .
the colour - coding indicates the sign of @xmath32 experienced by the fluid element along its trajectory ( dark blue means @xmath75 , yellow means @xmath117 ) .
the motion of the fluid element before it emerged at the surface is essentially adiabatic , with @xmath118 , where @xmath119 is the universal gas constant .
as it approaches @xmath120 , radiative cooling leads to a steep drop of @xmath116 and @xmath107 with height . accompanying
this cooling is a simultaneous compression of the fluid element , so that it becomes temporarily denser than the fluid beneath it .
thereafter , the fluid element continues to ascend and the density decreases again . when it has reached @xmath121 mm , the fluid element is cooler than @xmath85 and it begins to be heated radiatively .
this radiative heating leads to a rise in entropy , while the temperature continues to decrease because the effect of expansion overwhelms that of radiative heating .
finally , at @xmath122 mm , the fluid element overturns and begins its descent . at this moment , the fluid element has a gas temperature below @xmath85 , so that it continues to be heated radiatively even though its temperature rises by compression . only at @xmath123 mm ( @xmath124 km below the apex of the fluid element s trajectory )
is the temperature of the fluid element above @xmath85 and @xmath32 becomes negative .
, temperature @xmath107 and density @xmath15 along one of the fluid element trajectories shown in fig .
[ fig : fountain_qtot ] .
as the fluid element initially emerges at the base of the photosphere ( @xmath125 ) , it has relatively high entropy , high temperature and low density .
the colour - coding indicates the sign of @xmath32 experienced by the fluid element along its trajectory ( dark blue means @xmath126 , yellow means @xmath76 ) . above @xmath127 km ,
the fluid element is hotter during descent than during its ascent .
this is a result of radiative heating of the fluid element in the photosphere.,scaledwidth=70.0% ] the profiles of @xmath116 , @xmath107 and @xmath15 in fig .
[ fig : hysteresis_curves ] exhibit hysteresis .
there are two reasons for this behaviour .
firstly , there is a lag between when the fluid element overturns and when @xmath32 switches sign ( see previous paragraph ) .
secondly , the asymmetry in the speed of upflows ( slower ) and downflows ( faster ) means that the cumulative heating experienced by the fluid element over its journey upwards in the photosphere is greater than the accumulative radiative cooling it undergoes on the return journey .
these two factors contribute to the temperature and entropy enhancements of the intergranular boundaries ( downflows ) with respect to the granules ( upflows ) . ) .
the red dashed curve shows the reconstructed entropy profile , @xmath128 , calculated by integrating eq .
( [ eqn : entropy_evolution ] ) .
, scaledwidth=70.0% ] in order to test whether dissipative effects ( viscous and thermal , e.g. in shocks ) play a role in causing the temperature enhancement in the intergranular lanes in the upper photosphere , we have reconstructed the profiles of @xmath116 along the trajectory neglecting hyperdiffusivity effects . by keeping track of @xmath129 , @xmath130 and @xmath131 along the trajectory of a fluid element , we can reconstruct the time history of @xmath116 for a fluid element .
given these profiles and the initial entropy @xmath132 of the fluid element , eq .
( [ eqn : entropy_evolution ] ) can be integrated to yield the reconstructed entropy profile @xmath133 .
this quantity is plotted in fig .
[ fig : hysteresis_prediction ] together with the entropy profile shown in fig .
[ fig : hysteresis_curves ] .
the good match of the two profiles in the layers @xmath134 tells us that radiation is indeed the cause of hysteresis in the entropy and temperature profiles .
the reconstructed profile @xmath128 begins to deviate from the tracer profile once the fluid element approaches the narrow turbulent downflow jets in the convection zone . here
the radiative effects diminish and dissipative effects described by the hyperdiffusive terms become important .
the reversed granulation is a reversal of the pattern of intensity contrast between intensity maps in the continuum and intensity maps in spectral lines forming in the middle to upper photosphere .
it has long been established that this reversal of the intensity fluctuations is a result of the sign change of the horizontal temperature fluctuations ( see fig .
[ fig : reverse_temperature_fluctuation ] ) .
analysis of our 3d numerical model of the near - surface convection zone and photosphere confirms the existence of this reversal in the temperature fluctuations above a height of @xmath0 km , in agreement with previous observational findings . by following in detail the time histories of fluid elements overshooting into the photosphere
, we were able to identify the processes that determine the horizontal temperature fluctuations in the photosphere .
we find that changes in the temperature of fluid elements in the photosphere result from adiabatic expansion and radiative heating . whereas the former decreases the gas temperature of a fluid element , the latter increases it towards the radiative equilibrium temperature @xmath85 .
the ratio of the associated characteristic timescales tells us that , for a fluid element rising in the optically thin layers , the decrease of temperature due to expansion is only partially cancelled by radiative heating .
this disparity allows a fluid element to depart from radiative equilibrium as it penetrates into the photosphere . due to this deviation from @xmath85 , a fluid element overturning in the optically thin layers
still experiences some radiative heating as it begins to descend .
this , together with the fact that downflows have higher vertical speed than upflows , leads to a hysteretic behaviour of the thermodynamic history of the fluid element ( see fig . [
fig : hysteresis_curves ] ) . in various guises in previous works ,
the following explanation has often been invoked to explain the origin of the reversal of temperature fluctuations : since the photosphere is subadiabatically stratified , material rising from granules into this stable layer has lower temperature that its ` mean surroundings ' .
this view is problematic in two respects .
firstly , it implicitly assumes that material penetrating into the photosphere does so in an adiabatic fashion .
secondly , it is not at all clear what are the ` mean surroundings ' of a fluid element embedded in convective flow consisting of disjointed upwellings separated by an intergranular network .
our present efforts highlight the importance of radiative energy exchange in the dynamics of the photosphere and helps dispel possible misconceptions regarding the origin of the reversed granulation in the photosphere .
+ + * acknowledgements * + mcmc thanks prof .
franz kneer and his group at the georg - august university of gttingen for valuable discussions on this topic .
mcmc acknowledges financial support from the international max planck research school at the max planck institute for solar system research . ,
i. , ruiz cobo , b. , collados , m. , & del toro iniesta , j. c. 1999 , in asp conf .
ser . 173 : stellar structure : theory and test of connective energy transport , ed .
a. gimenez , e. f. guinan , & b. montesinos , 313 | we study the structure and reveal the physical nature of the reversed granulation pattern in the solar photosphere by means of 3-dimensional radiative hydrodynamics simulations .
we used the muram code to obtain a realistic model of the near - surface layers of the convection zone and the photosphere .
the pattern of horizontal temperature fluctuations at the base of the photosphere consists of relatively hot granular cells bounded by the cooler intergranular downflow network . with increasing height in the photosphere
, the amplitude of the temperature fluctuations diminishes . at a height of @xmath0 km in the photosphere ,
the pattern of horizontal temperature fluctuations reverses so that granular regions become relatively cool compared to the intergranular network .
detailed analysis of the trajectories of fluid elements through the photosphere reveal that the motion of the fluid is non - adiabatic , owing to strong radiative cooling when approaching the surface of optical depth unity followed by reheating by the radiation field from below .
the temperature structure of the photosphere results from the competition between expansion of rising fluid elements and radiative heating .
the former acts to lower the temperature of the fluid whereas the latter acts to increase it towards the radiative equilibrium temperature with a net entropy gain .
after the fluid overturns and descends towards the convection zone , radiative energy loss again decreases the entropy of the fluid .
radiative heating and cooling of fluid elements that penetrate into the photosphere and overturn do not occur in equal amounts .
the imbalance in the cumulative heating and cooling of these fluid elements is responsible for the reversal of temperature fluctuations with respect to height in the photosphere . |
a _ frame _ over a hilbert space @xmath2 is a collection of vectors @xmath3 with _ frame bounds _
@xmath4 such that @xmath5 for a finite frame @xmath6 in an @xmath0-dimensional hilbert space @xmath7 , the upper frame bound @xmath8 which satisfies trivially exists . in this finite - dimensional
setting , the notion of being a frame solely depends on the lower frame bound @xmath9 being strictly positive , which is equivalent to having the frame elements @xmath6 span the hilbert space . in practice ,
frames are chiefly used in two ways .
the _ synthesis operator _ @xmath10 of a finite frame @xmath11 is defined by @xmath12f_n$ ] . as such
, the @xmath13 matrix representation @xmath14 of the synthesis operator has the frame elements @xmath6 as columns , with the natural identification @xmath15 .
note that here and throughout , we make no notational distinction between a frame @xmath14 and its synthesis operator .
the adjoint of the synthesis operator is called the _ analysis operator _
@xmath16 , defined by @xmath17:=\langle x , f_n\rangle$ ] .
classically , frames are used to redundantly decompose signals , @xmath18 , before synthesizing the corresponding frame coefficients , @xmath19 , and so the _ frame operator _ @xmath20 is often analyzed to determine how well this process preserves information about the original signal @xmath21 . in particular , if the frame bounds are equal , the frame operator has the form @xmath22 , and so signal reconstruction is rather painless : @xmath23 ; in this case , the frame is called _ tight_. oftentimes ,
it is additionally desirable for the frame elements to have unit norm , in which case the frame is a _ unit norm frame_. moreover , the _ worst - case coherence _ between unit norm frame elements @xmath24 satisfies @xmath25 , and equality is achieved precisely when the frame is tight with @xmath26 for all distinct pairs @xmath27
@xcite ; in this case , the frame is called an _ equiangular tight frame _ ( etf ) .
the utility of unit norm tight frames and etfs is commonly expressed in terms of a scenario in which frame coefficients @xmath28\}_{n=1}^n$ ] are transmitted over a noisy or lossy channel before reconstructing the signal : @xmath29 where @xmath30 represents the channel s random and not - necessarily - linear deformation process . using an additive white gaussian noise model ,
goyal @xcite established that , of all unit norm frames , unit norm tight frames minimize mean squared error in reconstruction . for the case of a lossy channel , holmes and paulsen
@xcite established that , of all tight frames , unit norm tight frames minimize worst - case error in reconstruction after one erasure , and that equiangular tight frames minimize this error after two erasures .
we note that the reconstruction process in , namely the application of @xmath31 , is inherently blind to the effect of the deformation process of the channel .
this contrasts with pschel and kovaevi s more recent work @xcite , which describes an adaptive process for reconstruction after multiple erasures ; we will return to this idea later .
another application of frames is sparse signal processing .
this field concerns signal classes which are sparse in some basis . as an example
, natural images tend to be nearly sparse in the wavelet basis @xcite .
some applications have signal sparsity in the identity basis @xcite . given a signal @xmath21 , let @xmath32 represent its sparsifying basis and consider @xmath33 where @xmath14 is the @xmath13 synthesis operator of a frame , @xmath34 has at most @xmath35 nonzero entries , and @xmath36 is some sort of noise .
when given measurements @xmath37 , one typically wishes to reconstruct the original vector @xmath21 ; viewing @xmath14 as a sensing matrix , this process is commonly referred to as _ compressed sensing _ since we often take @xmath38 .
note that @xmath37 can be viewed as a noisy linear combination of a few unknown frame elements , and the goal of compressed sensing is to determine which frame elements are active in the combination , and further estimate the scalar multiples used in this combination .
this problem setup is very related to that of sparse approximation , in which signals are known to be expressible as a sparse combination of elements from an _ overcomplete _ dictionary ; in this case , the dictionary elements form the columns of @xmath14 , and we again wish to determine the scalar multiples used in a given sparse combination . in order to have any sort of inversion process for , even in the noiseless case where @xmath39 , @xmath14 must map @xmath35-sparse vectors injectively .
that is , we need @xmath40 for any distinct @xmath35-sparse vectors @xmath41 and @xmath42 .
subtraction then gives that @xmath43-sparse vectors like @xmath44 can not be in the nullspace of @xmath14 .
this identification led to the following definition @xcite : the _ spark _ of a matrix @xmath14 is the size of the smallest linearly dependent subset of columns , i.e. , @xmath45 using the above analysis , donoho and elad @xcite showed that a signal @xmath21 with sparsity level @xmath46 is necessarily the unique sparsest solution to @xmath47 , and furthermore , there exist signals @xmath21 of sparsity level @xmath48 which are not the unique sparsest solution to @xmath47 .
this demonstrates that matrices with larger spark are naturally equipped to support signals with larger sparsity levels .
one is naturally led to consider matrices that support the largest possible sparse signal class ; we say an @xmath13 matrix @xmath14 is _ full spark _ if its spark is as large as possible , i.e. , @xmath49 .
equivalently , @xmath13 full spark matrices have the property that every @xmath50 submatrix is invertible ; as such , a full spark matrix is necessarily full rank , and therefore a frame .
that being said , while the submatrices of a full spark matrix will be invertible , they may not be well - conditioned
. moreover , the conditioning of these submatrices is an important feature for compressed sensing ; cands @xcite gives an elegant proof that sensing matrices with well - conditioned submatrices , specifically , satisfying the _ restricted isometry property ( rip ) _ , allow for stable and efficient recovery .
unfortunately , for deterministic matrices , it is difficult to determine the conditioning of every submatrix ; to date , no deterministic rip matrix is known to perform optimally @xcite , and in one respect , this difficulty is precisely the barrier to significant progress on the kadison - singer problem @xcite .
however , some work has been done to test whether _ most _ submatrices are well - conditioned @xcite , and better yet , other work has managed to achieve rip - like performance without requiring all submatrices to be well - conditioned @xcite .
regardless , good sensing matrices for compressed sensing must necessarily have large spark to enable reconstruction @xcite , and so building such matrices could serve as one step toward optimal deterministic rip matrices . in sparse signal processing , the specific application of full spark frames has been studied for some time . in 1997 ,
gorodnitsky and rao @xcite first considered full spark frames , referring to them as matrices with the _ unique representation property _
( for reasons discussed above ) .
since @xcite , the unique representation property has been explicitly used to find a variety of performance guarantees for sparse signal processing @xcite .
tang and nehorai @xcite also obtain performance guarantees using full spark frames , but they refer to them as _ non - degenerate measurement matrices_. consider a matrix whose entries are independent continuous random variables .
intuitively , the matrix is full spark with probability one , and this fact was recently proved in @xcite .
however , this random process does not allow one to control certain features of the matrix , such as unit - norm tightness or equiangularity .
indeed , etfs are notoriously difficult to construct , but they appear to be particularly well - suited for sparse signal processing @xcite .
for example , bajwa et al .
@xcite uses steiner etfs to recover the support of @xmath34 in ; given measurements @xmath37 , the largest entries in the back - projection @xmath51 often coincide with the support of @xmath34 . however , steiner etfs have particularly small spark @xcite , and back - projection will correctly identify the support of @xmath21 even when the corresponding columns in @xmath14 are dependent ; in this case , there is no way to estimate the nonzero entries of @xmath34 .
this illustrates one reason to build deterministic full spark frames : their submatrices are necessarily invertible , making it possible to estimate these nonzero entries . for another application of full spark frames , we return to the problem of reconstructing a signal from distorted frame coefficients .
specifically , we consider pschel and kovaevi s work @xcite , which focuses on an internet - like channel that is prone to multitudes of erasures . in this context , they reconstruct the signal after first identifying which frame coefficients were not erased ; with this information , the signal can be estimated provided the corresponding frame elements span . in this sense ,
full spark frames are _ maximally robust to erasures _ , as coined in @xcite .
in particular , an @xmath13 full spark frame is robust to @xmath52 erasures since any @xmath0 of the frame coefficients will uniquely determine the original signal .
yet another application of full spark frames is phaseless reconstruction , which can be viewed in terms of a channel , as in ; in this case , @xmath30 is the entrywise absolute value function .
phaseless reconstruction has a number of real - world applications including speech processing @xcite , x - ray crystallography @xcite , and quantum state estimation @xcite .
as such , there has been a lot of work to reconstruct an @xmath0-dimensional vector ( up to an overall phase factor ) from the magnitudes of its frame coefficients , most of which involves frames in operator space , which inherently require @xmath53 measurements @xcite
. however , balan et al .
@xcite show that if an @xmath13 real frame @xmath14 is full spark with @xmath54 , then @xmath55 is injective , meaning an inversion process is possible with only @xmath56 measurements .
this result prompted an ongoing search for efficient phaseless reconstruction processes @xcite , but no reconstruction process can succeed without a good family of frames , such as full spark frames . despite the fact that full spark frames have a multitude of applications , to date
, there has not been much progress in constructing deterministic full spark frames , let alone full spark frames with additional desirable properties .
a noteworthy exception is pschel and kovaevi s work @xcite , in which real full spark tight frames are constructed using polynomial transforms . in the present paper , we start by investigating vandermonde frames , harmonic frames , and modifications thereof . while the use of certain vandermonde and harmonic frames as full spark frames is not new @xcite , the fruits of our investigation are new : for instance , we demonstrate that certain classes of etfs are full spark , and we characterize the @xmath13 full spark harmonic frames for which @xmath57 is a prime power .
the remainder of the paper proves two results which might be considered folklore - type observations while their proofs are new , the results are not surprising .
first , in section 3 , we show that @xmath13 full spark parseval frames form a dense subset of the entire collection of @xmath13 parseval frames . as such
, full spark frames are abundant , even after imposing the additional condition of tightness . this result is balanced with section 4 ,
in which we prove that verifying whether a matrix is full spark is hard for @xmath1 under randomized polynomial - time reductions .
in other words , assuming @xmath58 ( a computational complexity assumption slightly stronger than @xmath59 and nearly as widely believed ) , then there is no method by which one can efficiently test whether matrices are full spark . as such , the deterministic constructions in section 2 are significant in that they guarantee a property which is otherwise difficult to check .
a square matrix is invertible if and only if its determinant is nonzero , and in our quest for deterministic constructions of full spark frames , this characterization will reign supreme .
one class of matrices has a particularly simple determinant formula : vandermonde matrices .
specifically , vandermonde matrices have the following form : @xmath60 and square vandermonde matrices , i.e. , with @xmath61 , have the following determinant : @xmath62 consider in the case where @xmath63 . since every @xmath50 submatrix of @xmath64 is also vandermonde , we can modify the indices in to calculate the determinant of the submatrices .
these determinants are nonzero precisely when the bases @xmath65 are distinct , yielding the following result : [ lem.vandermonde ] a vandermonde matrix is full spark if and only if its bases are distinct . to be clear , this result is not new .
in fact , the full spark of vandermonde matrices was first exploited by fuchs @xcite for sparse signal processing .
later , bourguignon et al .
@xcite specifically used the full spark of vandermonde matrices whose bases are sampled from the complex unit circle .
interestingly , when viewed in terms of frame theory , vandermonde matrices naturally point to the discrete fourier transform : [ thm.vandermonde ] the only @xmath13 vandermonde matrices that are equal norm and tight have bases in the complex unit circle . among these ,
the frames with the smallest worst - case coherence have bases that are equally spaced in the complex unit circle , provided @xmath66 .
suppose a vandermonde matrix is equal norm and tight .
note that a zero base will produce the zeroth identity basis element @xmath67 . letting @xmath68 denote the indices of the nonzero bases , the fact that the matrix is full rank implies @xmath69 .
also , equal norm gives that the frame element length @xmath70|^2=\sum_{m=0}^{m-1}|\alpha_n^m|^2=\sum_{m=0}^{m-1}|\alpha_n|^{2m}\ ] ] is constant over @xmath71 .
since @xmath72 is strictly increasing over @xmath73 , there exists @xmath74 such that @xmath75 for all @xmath71 .
next , tightness gives that the rows have equal norm , implying that the first two rows have equal norm , i.e. , @xmath76 .
thus @xmath77 , and so the nonzero bases are in the complex unit circle .
furthermore , since the zeroth and first rows have equal norm by tightness , we have @xmath78 , and so every base is in the complex unit circle . now consider the inner product between vandermonde frame elements whose bases @xmath79 come from the complex unit circle : @xmath80 we will show that the worst - case coherence comes from the two closest bases . consider the following function : @xmath81 figure [ figure ] gives a plot of this function in the case where @xmath82 .
we will prove two things about this function : * @xmath83 for every @xmath84 , * @xmath85 for every @xmath86 . defined by in the case where @xmath82 .
observe ( i ) that @xmath87 is strictly decreasing on the interval @xmath88 , and ( ii ) that @xmath89 for every @xmath90 .
as established in the proof of theorem [ thm.vandermonde ] , @xmath87 behaves in this manner for general values of @xmath0 .
[ figure],scaledwidth=50.0% ] first , we claim that ( i ) and ( ii ) are sufficient to prove our result . to establish this ,
we first show that the two closest bases @xmath91 and @xmath92 satisfy @xmath93 . without loss of generality ,
the @xmath94 s are ordered in such a way that @xmath95 are nondecreasing .
define @xmath96 and let @xmath97 be the @xmath94 which minimizes @xmath98 . since the minimum is less than the average , we have @xmath99 provided @xmath100 . note that if we view @xmath101 as members of @xmath102 , then @xmath103 . since @xmath104 is even , then ( i ) implies that @xmath105 is larger than any other @xmath106 in which @xmath107\cup[1-\tfrac{1}{2m},1)$ ] .
next , and ( ii ) together imply that @xmath108 is larger than any other @xmath106 in which @xmath109 , provided @xmath100 .
combined , ( i ) and ( ii ) give that @xmath110 achieves the worst - case coherence of @xmath111 .
additionally , ( i ) gives that the worst - case coherence @xmath110 is minimized when @xmath112 is maximized , i.e. , when the @xmath113 s are equally spaced in the unit interval . to prove ( i ) , note that the geometric sum formula gives @xmath114 where the final expression uses the identity @xmath115 . to show that @xmath87 is decreasing over @xmath116 , note that the base of is positive on this interval , and performing the quotient rule to calculate its derivative will produce
a fraction whose denominator is nonnegative and whose numerator is given by @xmath117 this factor is zero at @xmath118 and has derivative : @xmath119 which is strictly negative for all @xmath84 .
hence , is strictly negative whenever @xmath84 , and so @xmath120 for every @xmath84 .
for ( ii ) , note that for every @xmath121 , we can individually bound the numerator and denominator of what the geometric sum formula gives : @xmath122 consider the @xmath123 discrete fourier transform ( dft ) matrix , scaled to have entries of unit modulus : @xmath124 where @xmath125 .
the first @xmath0 rows of the dft form a vandermonde matrix of distinct bases @xmath126 ; as such , this matrix is full spark by lemma [ lem.vandermonde ] .
in fact , the previous result says that this is in some sense an optimal vandermonde frame , but this might not be the best way to pick rows from a dft . indeed , several choices of dft rows could produce full spark frames , some with smaller coherence or other desirable properties , and so the remainder of this section focuses on full spark dft submatrices .
first , we note that not every dft submatrix is full spark . for example , consider the @xmath127 dft : @xmath128 certainly , the zeroth and second rows of this matrix are not full spark , since the zeroth and second columns of this submatrix form the all - ones matrix , which is not invertible .
so what can be said about the set of permissible row choices ?
the following result gives some necessary conditions on this set : [ thm.closure ] take an @xmath123 discrete fourier transform matrix , and select the rows indexed by @xmath129 to build the matrix @xmath14 . if @xmath14 is full spark , then so is the matrix built from rows indexed by * any translation of @xmath130 , * any @xmath131 with @xmath9 relatively prime to @xmath57 , * the complement of @xmath130 in @xmath132 . for ( i )
, we first define @xmath133 to be the @xmath123 diagonal matrix whose diagonal entries are @xmath126 .
note that , since @xmath134 , translating the row indices @xmath130 by @xmath135 corresponds to multiplying @xmath14 on the right by @xmath133 .
for some set @xmath136 of size @xmath137 , let @xmath138 denote the @xmath50 submatrix of @xmath14 whose columns are indexed by @xmath139 , and let @xmath140 denote the @xmath50 diagonal submatrix of @xmath133 whose diagonal entries are indexed by @xmath139
. then since @xmath140 is unitary , we have @xmath141 thus , if @xmath14 is full spark , @xmath142 , and so @xmath143 is also full spark .
using this fact inductively proves ( i ) for all translations of @xmath130 . for ( ii ) ,
let @xmath144 denote the submatrix of rows indexed by @xmath131 .
then for any set @xmath136 of size @xmath0 , @xmath145 since @xmath9 is relatively prime to @xmath57 , multiplication by @xmath9 permutes the elements of @xmath132 , and so @xmath146 has exactly @xmath0 distinct elements .
thus , if @xmath14 is full spark , then @xmath147 , and so @xmath144 is also full spark . for ( iii ) , we let @xmath144 be the @xmath148 submatrix of rows indexed by @xmath149 , so that @xmath150 we will use contraposition to show that @xmath14 being full spark implies that @xmath144 is also full spark . to this end ,
suppose @xmath144 is not full spark .
then @xmath144 has a collection of @xmath52 linearly dependent columns @xmath151 , and so there exists a nontrivial sequence @xmath152 such that @xmath153 considering @xmath154 , where @xmath155 is the @xmath156th identity basis element , we can use to express this linear dependence in terms of @xmath14 : @xmath157 rearranging then gives @xmath158 here , we note that @xmath21 is nonzero since @xmath152 is nontrivial , and that @xmath159 .
furthermore , whenever @xmath160 , we have from that @xmath161 and so @xmath162 .
thus , the containment @xmath163 is proper , and so @xmath164 since the @xmath50 submatrix @xmath165 is rank - deficient , it is not invertible , and therefore @xmath14 is not full spark .
we note that our proof of ( iii ) above uses techniques from cahill et al .
@xcite , and can be easily generalized to prove that the naimark complement of a full spark tight frame is also full spark .
theorem [ thm.closure ] tells us quite a bit about the set of permissible choices for dft rows .
for example , not only can we pick the first @xmath0 rows of the dft to produce a full spark vandermonde frame , but we can also pick any consecutive @xmath0 rows , by theorem [ thm.closure](i ) .
we would like to completely characterize the choices that produce full spark harmonic frames .
the following classical result does this in the case where @xmath57 is prime : let @xmath57 be prime .
then every square submatrix of the @xmath123 discrete fourier transform matrix is invertible . as an immediate consequence of chebotarv s theorem ,
every choice of rows from the dft produces a full spark harmonic frame , provided @xmath57 is prime .
this application of chebotarv s theorem was first used by cands et al .
@xcite for sparse signal processing .
note that each of these frames are equal - norm and tight by construction .
harmonic frames can also be designed to have minimal coherence ; xia et al .
@xcite produces harmonic equiangular tight frames by selecting row indices which form a difference set in @xmath132 .
interestingly , most known families of difference sets in @xmath132 require @xmath57 to be prime @xcite , and so the corresponding harmonic equiangular tight frames are guaranteed to be full spark by chebotarv s theorem . in the following , we use chebotarv s theorem to demonstrate full
spark for a class of frames which contains harmonic frames , namely , frames which arise from concatenating harmonic frames with any number of identity basis elements : [ thm.harmonic plus identity ] let @xmath57 be prime , and pick any @xmath166 rows of the @xmath123 discrete fourier transform matrix to form the harmonic frame @xmath167 .
next , pick any @xmath168 , and take @xmath133 to be the @xmath50 diagonal matrix whose first @xmath35 diagonal entries are @xmath169 , and whose remaining @xmath170 entries are @xmath171 .
then concatenating @xmath172 with the first @xmath35 identity basis elements produces an @xmath173 full spark unit norm tight frame . as an example , when @xmath174 and @xmath175 , we can pick @xmath176 rows of the @xmath177 dft which are indexed by @xmath178 . in this case
, @xmath133 makes the entries of the first dft row have size @xmath179 and the entries of the remaining rows have size @xmath180 .
concatenating with the first identity basis element then produces an equiangular tight frame which is full spark : @xmath181.\ ] ] let @xmath14 denote the resulting @xmath173 frame .
we start by verifying that @xmath14 is unit norm .
certainly , the identity basis elements have unit norm . for the remaining frame elements ,
the modulus of each entry is determined by @xmath133 , and so the norm squared of each frame element is @xmath182 to demonstrate that @xmath14 is tight , it suffices to show that @xmath183 .
the rows of @xmath172 are orthogonal since they are scaled rows of the dft , while the rows of the identity portion are orthogonal because they have disjoint support .
thus , @xmath184 is diagonal .
moreover , the norm squared of each of the first @xmath35 rows is @xmath185 , while the norm squared of each of the remaining rows is @xmath186 , and so @xmath183 . to demonstrate that @xmath14 is full spark , first note that every @xmath50 submatrix of @xmath172 is invertible since @xmath187 by chebotarv s theorem .
also , in the case where @xmath188 , we note that the @xmath50 submatrix of @xmath14 composed solely of identity basis elements is trivially invertible .
the only remaining case to check is when identity basis elements and columns of @xmath172 appear in the same @xmath50 submatrix @xmath138 . in this case
, we may shuffle the rows of @xmath138 to have the form @xmath189 since shuffling rows has no impact on the size of the determinant , we may further use a determinant identity on block matrices to get @xmath190 since @xmath9 is a multiple of a square submatrix of the @xmath123 dft , we are done by chebotarv s theorem . as an example of theorem [ thm.harmonic plus identity ] ,
pick @xmath57 to be a prime congruent to @xmath191 , and select @xmath192 rows of the @xmath123 dft according to the index set @xmath193 .
if we take @xmath175 , the process in theorem [ thm.harmonic plus identity ] produces an equiangular tight frame of redundancy @xmath194 , which can be verified using quadratic gauss sums ; in the case where @xmath174 , this construction produces . note
that this corresponds to a special case of a construction in zauner s thesis @xcite , which was later studied by renes @xcite and strohmer @xcite .
theorem [ thm.harmonic plus identity ] says that this construction is full spark .
maximally sparse frames have recently become a subject of active research @xcite .
we note that when @xmath188 , theorem [ thm.harmonic plus identity ] produces a maximally sparse @xmath195 full spark frame , having a total of @xmath196 zero entries . to see that this sparsity level is maximal
, we note that if the frame had any more zero entries , then at least one of the rows would have @xmath0 zero entries , meaning the corresponding @xmath50 submatrix would have a row of all zeros and hence a zero determinant .
similar ideas were studied previously by nakamura and masson @xcite .
another interesting case is where @xmath197 , i.e. , when the frame constructed in theorem [ thm.harmonic plus identity ] is a union of the unitary dft and identity bases .
unions of orthonormal bases have received considerable attention in the context of sparse approximation @xcite .
in fact , when @xmath57 is a perfect square , concatenating the dft with an identity basis forms the canonical example @xmath14 of a dictionary with small spark @xcite . to be clear
, the dirac comb of @xmath198 spikes is an eigenvector of the dft , and so concatenating this comb with the negative of its fourier transform produces a @xmath199-sparse vector in the nullspace of @xmath14 . in stark contrast , when @xmath57 is prime , theorem [ thm.harmonic plus identity ] shows that @xmath14 is full spark .
the vast implications of chebotarv s theorem leads one to wonder whether the result admits any interesting generalization . in this direction ,
cands et al .
@xcite note that any such generalization must somehow account for the nontrivial subgroups of @xmath132 which are not present when @xmath57 is prime .
certainly , if one could characterize the full spark submatrices of a general dft , this would provide ample freedom to optimize full spark frames for additional considerations . while we do not have a characterization for the general case
, we do have one for the case where @xmath57 is a prime power . before stating the result
, we require a definition : we say a subset @xmath129 is _ uniformly distributed over the divisors of @xmath57 _ if , for every divisor @xmath200 of @xmath57 , the @xmath200 cosets of @xmath201 partition @xmath130 into subsets , each of size @xmath202 or @xmath203 . at first glance
, this definition may seem rather unnatural , but we will discover some important properties of uniformly distributed rows from the dft . as an example , we take a short detour by considering the _ restricted isometry property ( rip ) _ , which has received considerable attention recently for its use in compressed sensing .
we say a matrix @xmath14 is @xmath204-rip if @xmath205 cands and tao @xcite demonstrated that the sparsest @xmath206 which satisfies can be found using @xmath207-minimization , provided @xmath14 is @xmath208-rip .
later , rudelson and vershynin @xcite showed that a matrix of random rows from a dft and normalized columns is rip with high probability .
we will show that harmonic frames satisfy rip only if the selected row indices are nearly uniformly distributed over sufficiently small divisors of @xmath57 . to this end ,
recall that for any divisor @xmath200 of @xmath57 , the fourier transform of the @xmath200-sparse normalized dirac comb @xmath209 is the @xmath210-sparse normalized dirac comb @xmath211 .
let @xmath14 be the @xmath123 unitary dft , and let @xmath167 be the harmonic frame which arises from selecting rows of @xmath14 indexed by @xmath130 and then normalizing the columns .
in order for @xmath167 to be @xmath204-rip , @xmath130 must contain at least one member of @xmath201 for every divisor @xmath200 of @xmath57 which is @xmath212 , since otherwise @xmath213 which violates the lower rip bound at @xmath214 . in fact , the rip bounds indicate that @xmath215 can not be more than @xmath216 away from @xmath217 .
similarly , taking @xmath21 to be @xmath218 modulated by @xmath219 , i.e. , @xmath220:=\frac{1}{\sqrt{d}}\chi_{\langle\frac{n}{d}\rangle}[n]e^{2\pi ian / n}$ ] for every @xmath221 , gives that @xmath222 is also no more than @xmath216 away from @xmath135 .
this observation gives the following result : select rows indexed by @xmath129 from the @xmath123 discrete fourier transform matrix and then normalize the columns to produce the harmonic frame @xmath167 .
then @xmath167 satisfies the @xmath204-restricted isometry property only if @xmath223 for every divisor @xmath200 of @xmath57 with @xmath224 and every @xmath225 . now that we have an intuition for uniform distribution in terms of modulated dirac combs and rip , we take this condition to the extreme by considering uniform distribution over all divisors .
doing so produces a complete characterization of full spark harmonic frames when @xmath57 is a prime power : [ thm.uniformly distributed ] let @xmath57 be a prime power , and select rows indexed by @xmath129 from the @xmath123 discrete fourier transform matrix to build the submatrix @xmath14 .
then @xmath14 is full spark if and only if @xmath130 is uniformly distributed over the divisors of @xmath57 .
note that , perhaps surprisingly , an index set @xmath130 can be uniformly distributed over @xmath226 but not over @xmath227 , and vice versa . for example , @xmath228 is uniformly distributed over @xmath194 but not @xmath229 , while @xmath230 is uniformly distributed over @xmath229 but not @xmath194 .
since the first @xmath0 rows of a dft form a full spark vandermonde matrix , let s check that this index set is uniformly distributed over the divisors of @xmath57 . for each divisor @xmath200 of @xmath57 , we partition the first @xmath0 indices into the @xmath200 cosets of @xmath201 .
write @xmath231 with @xmath232 .
the first @xmath233 of the @xmath0 indices are distributed equally amongst all @xmath200 cosets , and then the remaining @xmath234 indices are distributed equally amongst the first @xmath234 cosets .
overall , the first @xmath234 cosets contain @xmath235 indices , while the remaining @xmath236 cosets have @xmath237 indices ; thus , the first @xmath0 indices are indeed uniformly distributed over the divisors of @xmath57 .
also , when @xmath57 is prime , _ every _ subset of @xmath132 is uniformly distributed over the divisors of @xmath57 in a trivial sense . in fact , chebotarv s theorem follows immediately from theorem [ thm.uniformly distributed ] . in some ways , portions of our proof of theorem [ thm.uniformly distributed ] mirror recurring ideas in the existing proofs of chebotarv s theorem @xcite . for the sake of completeness
, we provide the full argument and save the reader from having to parse portions of proofs from multiple references .
we start with the following lemmas , whose proofs are based on the proofs of lemmas 1.2 and 1.3 in @xcite .
[ lem.multiple of p ] let @xmath57 be a power of some prime @xmath226 , and let @xmath238 be a polynomial with integer coefficients .
suppose there exists @xmath57th roots of unity @xmath239 such that @xmath240 .
then @xmath241 is a multiple of @xmath226 . denoting @xmath242 , then for every @xmath243
, we have @xmath244 for some @xmath245 .
defining the polynomial @xmath246 , we have @xmath247 by assumption .
also , @xmath248 is a polynomial with integer coefficients , and so it must be divisible by the minimal polynomial of @xmath249 , namely , the cyclotomic polynomial @xmath250 . evaluating both polynomials at @xmath251
then gives that @xmath252 divides @xmath253 .
[ lem.big fraction ] let @xmath57 be a power of some prime @xmath226 , and pick @xmath254 such that @xmath255 is not a multiple of @xmath226 . then the rows indexed by @xmath130 in the @xmath123 discrete fourier transform form a full spark frame .
we wish to show that @xmath256 for all @xmath0-tuples of distinct @xmath57th roots of unity @xmath257 .
define the polynomial @xmath258 .
since columns @xmath156 and @xmath259 of @xmath260 are identical whenever @xmath261 , we know that @xmath133 vanishes in each of these instances , and so we can factor : @xmath262 for some polynomial @xmath238 with integer coefficients . by lemma [ lem.multiple of p ] , it suffices to show that @xmath241 is not a multiple of @xmath226 , since this implies @xmath263 is nonzero for all @xmath0-tuples of distinct @xmath57th roots of unity @xmath257 . to this end ,
we proceed by considering @xmath264 to compute @xmath9 , we note that each application of @xmath265 produces terms according to the product rule . for some terms , a linear factor of the form @xmath266 or @xmath267
is replaced by @xmath268 or @xmath269 , respectively .
for each the other terms , these linear factors are untouched , while another factor , such as @xmath238 , is differentiated and multiplied by @xmath268 .
note that there are a total of @xmath270 linear factors , and only @xmath270 differentiation operators to apply .
thus , after expanding every product rule , there will be two types of terms : terms in which every differentiation operator was applied to a linear factor , and terms which have at least one linear factor remaining untouched .
when we evaluate at @xmath271 , the terms with linear factors vanish , and so the only terms which remain came from applying every differentiation operator to a linear factor .
furthermore , each of these terms before the evaluation is of the form @xmath272 , and so evaluation at @xmath271 produces a sum of terms of the form @xmath241 ; to determine the value of @xmath9 , it remains to count these terms .
the @xmath273 copies of @xmath274 can only be applied to linear factors of the form @xmath275 , of which there are @xmath273 , and so there are a total of @xmath276 ways to distribute these operators .
similarly , there are @xmath277 ways to distribute the @xmath278 copies of @xmath279 amongst the @xmath278 linear factors of the form @xmath280 .
continuing in this manner produces an expression for @xmath9 : @xmath281 for an alternate expression of @xmath9 , we substitute the definition of @xmath282 into @xmath283 . here , we exploit the multilinearity of the determinant and the fact that @xmath284 to get @xmath285 where the final equality uses the fact that @xmath286 is the transpose of a vandermonde matrix . equating to reveals that is an expression for @xmath241 .
thus , by assumption , @xmath241 is not a multiple of @xmath226 , and so we are done .
( @xmath287 ) we will use lemma [ lem.big fraction ] to demonstrate that @xmath14 is full spark . to apply this lemma
, we need to establish that is not a multiple of @xmath226 , and to do this , we will show that there are as many @xmath226-divisors in the numerator of as there are in the denominator .
we start by counting the @xmath226-divisors of the denominator : @xmath288 for each pair of integers @xmath289 , there are @xmath290 factors in of the form @xmath291 . by adding these , we count each factor @xmath292 as many times as it can be expressed as a multiple of a power of @xmath226 , which equals the number of @xmath226-divisors in @xmath292 .
thus , the number of @xmath226-divisors of is @xmath293 next , we count the @xmath226-divisors of the numerator of . to do this ,
we use the fact that @xmath130 is uniformly distributed over the divisors of @xmath57 .
since @xmath57 is a power of @xmath226 , the only divisors of @xmath57 are smaller powers of @xmath226 . also ,
the cosets of @xmath294 partition @xmath130 into subsets @xmath295 .
we note that @xmath296 is a multiple of @xmath297 precisely when @xmath298 and @xmath299 belong to the same subset @xmath300 for some @xmath301 .
to count @xmath226-divisors , we again count each factor @xmath296 as many times as it can be expressed as a multiple of a prime power : @xmath302 write @xmath303 with @xmath304 .
then @xmath305 .
since @xmath130 is uniformly distributed over @xmath297 , there are @xmath234 subsets @xmath300 with @xmath306 elements and @xmath307 subsets with @xmath308 elements .
we use this to get @xmath309 rearranging and substituting @xmath303 then gives @xmath310 thus , there are as many @xmath226-divisors in the numerator as there are in the denominator , and so is not divisible by @xmath226 .
lemma [ lem.big fraction ] therefore gives that @xmath14 is full spark .
( @xmath311 ) we will prove that this direction holds regardless of whether @xmath57 is a prime power .
suppose @xmath129 is not uniformly distributed over the divisors of @xmath57 .
then there exists a divisor @xmath200 of @xmath57 such that one of the cosets of @xmath201 intersects @xmath130 with @xmath312 or @xmath313 indices .
notice that if a coset of @xmath201 intersects @xmath130 with @xmath312 indices , then the complement @xmath149 intersects the same coset with @xmath314 indices . by theorem [ thm.closure](iii ) , @xmath130 produces a full spark harmonic frame precisely when @xmath149 produces a full spark harmonic frame , and so we may assume without loss of generality that there exists a coset of @xmath201 which intersects @xmath130 with @xmath313 indices . to prove that the rows with indices in @xmath130 are not full spark , we find column entries which produce a singular submatrix . writing @xmath231 with @xmath232 , let @xmath139 contain @xmath237 cosets of @xmath315 along with @xmath234 elements from an additional coset .
we claim that the dft submatrix with row entries @xmath130 and column entries @xmath139 is singular . to see this , shuffle the rows and columns to form a matrix @xmath9 in which the row entries are grouped into common cosets of @xmath201 and
the column entries are grouped into common cosets of @xmath315 .
this breaks @xmath9 into rank-1 submatrices : each pair of cosets @xmath316 and @xmath317 produces a submatrix @xmath318 for some index sets @xmath319 and @xmath320 ; this is a rank-1 outer product .
let @xmath321 be the largest intersection between @xmath130 and a coset of @xmath201 .
then @xmath322 is the number of rows in the tallest of these rank-1 submatrices .
define @xmath323 to be the @xmath50 matrix with entries @xmath324=a[i , j]$ ] whenever @xmath325 and zero otherwise .
then @xmath326 since @xmath327 has @xmath328 rows of zero entries , we also have @xmath329 moreover , since we can decompose @xmath323 into a sum of @xmath330 zero - padded rank-1 submatrices , we have @xmath331 . combining this with and
then gives that @xmath332 , and so the dft submatrix is not invertible .
note that our proof of theorem [ thm.uniformly distributed ] establishes the necessity of having row indices uniformly distributed over the divisors of @xmath57 in the general case .
this leaves some hope for completely characterizing full spark harmonic frames .
naturally , one might suspect that the uniform distribution condition is sufficient in general , but this suspicion fails when @xmath333 .
indeed , the following dft submatrix is singular despite the row indices being uniformly distributed over the divisors of @xmath334 : @xmath335 also , just as we used chebotarv s theorem to analyze the harmonic equiangular tight frames from xia et al .
@xcite , we can also use theorem [ thm.uniformly distributed ] to determine whether harmonic equiangular tight frames with a prime power number of frame elements are full spark . unfortunately , none of the infinite families in @xcite have the number of frame elements in the form of a prime power ( other than primes ) .
luckily , there is at least one instance in which the number of frame elements happens to be a prime power : the harmonic frames that arise from singer difference sets have @xmath336 and @xmath337 for a prime power @xmath308 and an integer @xmath338 ; when @xmath339 and @xmath340 , the number of frame elements @xmath341 is a prime power . in this case , the row indices we select are @xmath342 but these are not uniformly distributed over 11 , and so the corresponding harmonic frame is not full spark by theorem [ thm.uniformly distributed ] .
recently , lu and do @xcite showed that full spark frames are dense in the entire set of matrices .
this corresponds to our intuition that a matrix whose entries are independent continuous random variables is full spark with probability one , which was also recently proved by blumensath and davies @xcite . by contrast , as noted by gorodnitsky and rao @xcite , certain classes of frames which arise in practice , such as in physical tomography , are never full spark .
this issue also occurs in frame theory : steiner etfs form one of the largest known classes of etfs , and yet none of them are full spark @xcite . as such , bourguignon et al .
@xcite were prompted to prove that , among all vandermonde frames with bases in the complex unit circle , full spark frames are dense . in this section
, we consider a very important class of frames , namely , those which exhibit parseval tightness , where the frame bound is @xmath135 .
specifically , we show that full spark parseval frames are dense in the entire set of parseval frames . unlike the previous work in this vein ,
our techniques exploit general concepts in algebraic geometry , lending themselves to future application in proving further density results . in order to make our arguments rigorous
, we view each entry @xmath343 of the @xmath13 matrix @xmath14 in terms of its real and imaginary parts : @xmath344 ; this decomposition will become helpful later when we consider inner products between rows of @xmath14 , which are not algebraic operations on complex vectors . recall that @xmath14 is full spark precisely when each @xmath50 submatrix has nonzero determinant .
we note that for each submatrix , the determinant is a polynomial in the @xmath345 s and @xmath346 s , and having this polynomial be nonzero is equivalent to having either its real or imaginary part be nonzero .
this naturally leads us to the following definition from algebraic geometry : a _ real algebraic variety _ is the set of common zeros of a finite set of polynomials , that is , given polynomials @xmath347 $ ] , we define the corresponding real algebraic variety by @xmath348 each submatrix determinant corresponds to a real algebraic variety @xmath349 of two polynomials , and having this determinant be nonzero is equivalent to restricting to the complement of @xmath64 . in general , a variety is equipped with a topology known as the _ zariski topology _ , in which subvarieties are the closed sets . as such , the set of matrices with a nonzero determinant forms a zariski - open set , since it is the complement of a variety . in order to exploit this zariski - openness , we require the additional concept of irreducibility : a variety is said to be _
irreducible _ if it can not be written as a finite union of proper subvarieties .
as an example , the entire space @xmath350 is the variety which corresponds to the zero polynomial ; in this case , every proper subvariety is lower - dimensional , and so @xmath350 is trivially irreducible . on the other hand ,
the variety in @xmath351 defined by @xmath352 is not irreducible because it can be expressed as the union of varieties defined by @xmath118 and @xmath353 .
irreducibility is important because it says something about zariski - open sets : [ thm.irreducible dense ] if @xmath64 is an irreducible algebraic variety , then every nonempty zariski - open subset of @xmath64 is dense in @xmath64 in the standard topology .
for example , the hyperplane defined by @xmath354 is a subvariety of @xmath350 , and since @xmath350 is irreducible , the complement of the hyperplane is dense in the standard topology .
going back to the variety defined by @xmath352 , we know it can be expressed as a union of proper subvarieties , namely the @xmath21- and @xmath37-axes , and complementing one gives a subset of the other , neither of which is dense in the entire variety .
we are now ready to prove the following result : [ thm.overall density ] every matrix is arbitrarily close to a full spark frame . in an @xmath13 matrix
, there are @xmath355 submatrices of size @xmath50 , and the determinant of each of these submatrices corresponds to a variety of two real polynomials .
since the set of @xmath13 full spark frames is defined as the ( finite ) intersection of the complements of these varieties , it is a zariski - open subset of the irreducible variety @xmath356 of all @xmath13 matrices .
moreover , this set is nonempty since it contains the matrix formed by the first @xmath0 rows of the @xmath123 dft , and so we are done by theorem [ thm.irreducible dense ] .
we now focus on the set of _ parseval frames _ , that is , tight frames with frame bound @xmath135 .
note that @xmath13 parseval frames are characterized by the rows forming an orthonormal system of size @xmath0 in @xmath57-dimensional space .
the set of all such orthonormal systems is known as the _ stiefel manifold _
, denoted @xmath357 . in general , a _
manifold _ is a set of vectors with a well - defined tangent space at every point in the set .
since this manifold property is nice , we would like to think of varieties as manifolds , but there exist varieties with points at which tangent spaces are not well - defined ; such points are called _ singularities_. note that our definition of singularity is geometric , i.e. , where the variety fails to be a real differentiable manifold , as opposed to algebraic . for example , the variety defined by @xmath352 is the union of the @xmath21- and @xmath37-axes , and as such , has a singularity at @xmath358 .
certainly with different types of varieties , there are other types of singularities which may arise , but there are also many varieties which do not have singularities at all , and we call these varieties _ nonsingular_. nonsingularity is a useful property for the following reason : [ thm.irreducible ] an algebraic variety which is nonsingular and connected is necessarily irreducible .
this result follows from theorem i.5.1 and remark iii.7.9.1 of hartshorne @xcite , which assumes that the variety is over an algebraically closed field , unlike @xmath359 ; however , the proof is unaffected when removing the algebraically - closed assumption .
note that the orthonormality conditions which characterize parseval frames @xmath14 can be expressed as polynomial equations in the real and imaginary parts of the entries @xmath343 . as such
, we may view the stiefel manifold as a real algebraic variety ; this variety is nonsingular because it is a manifold .
moreover , the variety is connected because the ( connected ) unitary group @xmath360 acts transitively on @xmath357 .
thus by theorem [ thm.irreducible ] , the variety of parseval frames is irreducible .
having established this , we can now prove the following result : [ thm.parseval density ] every parseval frame is arbitrarily close to a full spark parseval frame .
proceeding as in the proof of theorem [ thm.overall density ] , we have that the set of @xmath13 full spark parseval frames is zariski - open in the irreducible variety of all @xmath13 parseval frames .
again considering the first @xmath0 rows of the @xmath123 dft , we know this set is nonempty , and so we are done by theorem [ thm.irreducible dense ] .
we note that theorems [ thm.overall density ] and [ thm.parseval density ] are also true when we further require the frames to be real . in this case
, we can not use the first @xmath0 rows of the @xmath123 dft to establish that the zariski - open sets are nonempty . for the real version of theorem [ thm.overall density ] ,
we use an @xmath13 vandermonde matrix with distinct real bases ; see lemma [ lem.vandermonde ] .
however , this construction must be modified to use it in the proof of the real version of theorem [ thm.parseval density ] , since such vandermonde matrices will not be tight ; see theorem [ thm.vandermonde ] . given an @xmath13 full spark frame @xmath14 , the modification @xmath361 is full spark and parseval ; indeed , @xmath362 , and since @xmath14 is a frame , @xmath363 is full rank , and so the columns of @xmath144 are linearly independent precisely when the corresponding columns of @xmath14 are linearly independent .
another way that the proof of theorem [ thm.parseval density ] changes in the real case is in verifying that the real stiefel manifold is irreducible . just as in the complex case
, this follows from theorem [ thm.irreducible ] since @xmath357 is connected @xcite , but the fact that @xmath357 is connected in the real case is not immediate . by analogy ,
the orthogonal group @xmath364 certainly acts transitively on @xmath357 , but unlike the unitary group , the orthogonal group has two connected components . intuitively , this is resolved by the fact that @xmath365 , granting additional freedom of movement throughout @xmath357 .
in addition to theorems [ thm.overall density ] and [ thm.parseval density ] , we would like a similar result for unit norm tight frames , i.e. , that every unit norm tight frame is arbitrarily close to a full spark unit norm tight frame .
certainly , the set of unit norm tight frames is a real algebraic variety , but it is unclear whether this variety is irreducible . without knowing whether the variety is irreducible , we can follow the proofs of theorems [ thm.overall density ] and [ thm.parseval density ] to conclude that full spark unit norm tight frames are dense in the irreducible components in which they exist a far cry from the density result we seek .
this illustrates a significant gap in our current understanding of the variety of unit norm tight frames . it should be mentioned that strawn @xcite showed that the variety of @xmath13 unit norm tight frames ( over real or complex space ) is nonsingular precisely when @xmath0 and @xmath57 are relatively prime .
additionally , dykema and strawn @xcite proved that the variety of @xmath366 real unit norm tight frames is connected , and so by theorem [ thm.irreducible ] , this variety is irreducible when @xmath57 is odd .
it is unknown whether the variety of @xmath13 unit norm tight frames is connected in general .
finally , we note that a theorem [ thm.parseval density ] gives a weaker version of the result we would like : every unit norm tight frame is arbitrarily close to a full spark tight frame with frame element lengths arbitrarily close to @xmath135 .
in the previous section , we demonstrated the abundance of full spark frames , even after imposing the additional condition of tightness . but how much computation is required to check whether a particular frame is full spark ? at the heart of the matter is computational complexity theory , which provides a rigorous playing field for expressing how hard certain problems are . in this section
, we consider the complexity of the following problem : for the lay mathematician , full spark is `` obviously '' @xmath1-hard because the easiest way he can think to solve it for a given @xmath13 matrix is by determining whether each of the @xmath50 submatrices is invertible ; computing @xmath355 determinants would do , but this would take a lot of time , and so full spark must be @xmath1-hard .
however , computing @xmath355 determinants may not necessarily be the fastest way to test whether a matrix is full spark .
for example , perhaps there is an easy - to - calculate expression for the product of the determinants ; after all , this product is nonzero precisely when the matrix is full spark . recall that theorem [ thm.uniformly distributed ] gives a very straightforward litmus test for full spark in the special case where the matrix is formed by rows of a dft of prime - power order who s to say that a version of this test does not exist for the general case ?
if such a test exists , then it would suffice to find it , but how might one disprove the existence of any such test ? indeed , since we are concerned with the necessary amount of computation , as opposed to a sufficient amount , the lay mathematician s intuition is a bit misguided . to discern how much computation is necessary , the main feature of interest is a problem s _
complexity_. we use complexity to compare problems and determine whether one is harder than the other . as an example of complexity , intuitively , doubling an integer is no harder than adding integers , since one can use addition to multiply by @xmath194 ; put another way , the complexity of doubling is somehow `` encoded '' in the complexity of adding , and so it must be lesser ( or equal ) . to make this more precise , complexity theorists use what is called a _
polynomial - time reduction _ , that is , a polynomial - time algorithm that solves problem @xmath9 by exploiting an oracle which solves problem @xmath8 ; the reduction indicates that solving problem @xmath9 is no harder than solving problem @xmath8 ( up to polynomial factors in time ) , and we say `` @xmath9 reduces to @xmath8 , '' or @xmath367 . since we can use the polynomial - time routine @xmath368 to produce @xmath369
, we conclude that doubling an integer reduces to adding integers , as expected . in complexity theory ,
problems are categorized into complexity classes according to the amount of resources required to solve them .
for example , the complexity class @xmath370 contains all problems which can be solved in polynomial time , while problems in @xmath371 may require as much as exponential time .
problems in @xmath1 have the defining quality that solutions can be verified in polynomial time given a certificate for the answer .
as an example , the graph isomorphism problem is in @xmath1 because , given an isomorphism between graphs ( a certificate ) , one can verify that the isomorphism is legit in polynomial time .
clearly , @xmath372 , since we can ignore the certificate and still solve the problem in polynomial time .
finally , a problem @xmath8 is called @xmath1-_hard _ if every problem @xmath9 in @xmath1 reduces to @xmath8 , and a problem is called @xmath1-_complete _ if it is both @xmath1-hard and in @xmath1 . in plain speak ,
@xmath1-hard problems are harder than every problem in @xmath1 , while @xmath1-complete problems are the hardest of problems in @xmath1 . at this point , it should be clear that @xmath1-hard problems are not merely problems that seem to require a lot of computation to solve .
certainly , @xmath1-hard problems have this quality , as an @xmath1-hard problem can be solved in polynomial time only if @xmath373 ; this is an open problem , but it is widely believed that @xmath59 . however , there are other problems which seem hard but are not known to be @xmath1-hard ( e.g. , the graph isomorphism problem ) .
rather , to determine whether a problem is @xmath1-hard , one must find a polynomial - time reduction that compares the problem to all problems in @xmath1 . to this end ,
notice that @xmath367 and @xmath374 together imply @xmath375 , and so to demonstrate that a problem @xmath376 is @xmath1-hard , it suffices to show that @xmath374 for some @xmath1-hard problem @xmath8 .
unfortunately , it can sometimes be difficult to find a deterministic reduction from one problem to another .
one example is reducing the satisfiability problem ( sat ) to the unique satisfiability problem ( unique sat ) . to be clear
, sat is an @xmath1-hard problem @xcite that asks whether there exists an input for which a given boolean function returns `` true , '' while unique sat asks the same question with an additional promise : that the given boolean function is satisfiable only if there is a _ unique _
input for which it returns `` true . '' intuitively , unique sat is easier than sat because we might be able to exploit the additional structure of uniquely satisfiable boolean functions ; thus , it could be difficult to find a reduction from sat to unique sat .
despite this intuition , there is a _
randomized _ polynomial - time reduction from sat to unique sat @xcite .
defined over all boolean functions of @xmath94 variables , the reduction maps functions that are not satisfiable to other functions that are not satisfiable , and with probability @xmath377 , it maps satisfiable functions to uniquely satisfiable functions . after applying this reduction to a given boolean function , if a unique sat oracle declares `` uniquely satisfiable , '' then we know for certain that the original boolean function was satisfiable .
but the reduction will only map a satisfiable problem to a uniquely satisfiable problem with probability @xmath377 , so what good is this reduction ?
the answer lies in something called _ amplification _ ; since the success probability is , at worst , polynomially small in @xmath94 ( i.e. , @xmath378 ) , we can repeat our oracle - based randomized algorithm a polynomial number of times @xmath379 and achieve an error probability @xmath380 which is exponentially small . in this section ,
we give a randomized polynomial - time reduction from a problem in matroid theory . before stating the problem , we first briefly review some definitions . to each bipartite graph with bipartition @xmath381 ,
we associate a _ transversal matroid _ @xmath382 , where @xmath319 is the collection of subsets of @xmath383 whose vertices form the ends of a matching in the bipartite graph ; subsets in @xmath319 are called @xmath384 .
hall s marriage theorem @xcite gives a remarkable characterization of the independent sets in a transversal matroid : @xmath385 if and only if every subset @xmath386 has @xmath387 neighbors in the bipartite graph .
next , just as spark is the size of the smallest linearly dependent set , the _ girth _ of a matroid is the size of the smallest subset of @xmath383 that is not in @xmath319 .
in fact , this analogy goes deeper : a matroid is _ representable over a field _
@xmath388 if , for some @xmath0 , there exists a mapping @xmath389 such that @xmath390 is linearly independent if and only if @xmath391 ; as such , the girth of @xmath382 is the spark of @xmath392 . in our reduction , we make use of the fact that every transversal matroid is representable over @xmath359 @xcite .
we are now ready to state the problem from which we will reduce full spark : before giving the reduction , we will show that problem [ prob.girth ] is @xmath1-hard .
the result comes from mccormick s thesis @xcite , which credits the proof to stockmeyer ; since @xcite is difficult to access and the proof is instructive , we include it below : we will reduce from the @xmath1-complete clique decision problem , which asks `` given a graph ,
does it contain a clique of @xmath35 vertices ? '' @xcite .
first , we may assume @xmath393 without loss of generality , since any such clique can otherwise be found in cubic time by an exhaustive search .
take a graph @xmath394 , and consider the bipartite graph @xmath395 between disjoint sets @xmath383 and @xmath396 , in which @xmath397 for every @xmath398 and @xmath399 , and @xmath400 for every @xmath398 and @xmath401 .
we claim that the girth of the transversal matroid of @xmath395 is @xmath402 precisely when there exists a @xmath35-clique in @xmath144 .
we start by analyzing the girth of a transversal matroid .
consider any dependent set @xmath403 with @xmath404 neighbors in @xmath395 .
then removing any member @xmath21 of @xmath376 will produce a smaller set @xmath405 with @xmath406 neighbors in @xmath395 , which is necessarily dependent by the pigeonhole principle .
now consider any dependent set @xmath403 with @xmath407 neighbors in @xmath395 . by the hall s marriage theorem
, there exists a proper subset @xmath408 with @xmath409 neighbors in @xmath395 , meaning @xmath410 is a smaller dependent set .
thus , the girth @xmath411 is the size of the smallest subset @xmath403 with @xmath412 total neighbors in @xmath395 .
suppose @xmath413 .
then since @xmath376 is adjacent to every vertex in @xmath414 , @xmath376 has @xmath415 neighbors in @xmath64 .
these are precisely the vertices @xmath133 in @xmath144 which are induced by the edges in @xmath376 , and so @xmath376 is contained in the set @xmath410 of edges induced by @xmath133 , of which there are @xmath416 , with equality only if @xmath133 induces a @xmath35-clique in @xmath144 .
since @xmath417 implies equality , there exists a @xmath35-clique in @xmath144 .
now suppose there exists a @xmath35-clique with edges @xmath376
. then @xmath376 has @xmath402 elements and @xmath418 neighbors in @xmath395 . to prove that @xmath413
, it suffices to show that there is no smaller subset @xmath419 with @xmath420 total neighbors in @xmath395 .
suppose , to the contrary , that @xmath421 for some @xmath422 .
then there exists @xmath419 with @xmath411 elements and @xmath423 neighbors in @xmath64 .
note that each @xmath398 contains two vertices in @xmath144 , and so by the definition of @xmath395 , @xmath410 necessarily has @xmath424 neighbors in @xmath64 . also , since the @xmath425 neighbors of @xmath410 in @xmath64 arise from the subgraph of @xmath144 induced by @xmath410 , and since those neighbors induce at most @xmath426 edges including @xmath410 , we have @xmath427 .
this inequality simplifies to @xmath428 , which combines with @xmath424 to contradict the fact that @xmath422 .
having established that problem [ prob.girth ] is @xmath1-hard , we reduce from it the main problem of this section . our proof is specifically geared toward the case where the matrix in question has integer entries ; this is stronger than manipulating real ( complex ) numbers exactly as well as with truncations and tolerances
. we will give a randomized polynomial - time reduction from problem [ prob.girth ] to full spark .
as such , suppose we are given a bipartite graph @xmath144 , in which every edge is between the disjoint sets @xmath9 and @xmath8 .
take @xmath429 and @xmath430 . using this graph
, we randomly draw an @xmath13 matrix @xmath14 using the following process : for each @xmath431 and @xmath432 , pick the entry @xmath433 randomly from @xmath434 if @xmath435 in @xmath144 ; otherwise set @xmath436 . in proposition 3.11 of @xcite
, it is shown that the columns of @xmath14 form a representation of the transversal matroid of @xmath144 with probability @xmath437 . for the moment , we assume that @xmath14 succeeds in representing the matroid . since the girth of the original matroid equals the spark of its representation , for each @xmath438 , we test whether @xmath439 . to do this ,
take @xmath167 to be some @xmath440 full spark frame .
we will determine an appropriate value for @xmath441 later , but for simplicity , we can take @xmath167 to be the vandermonde matrix formed from bases @xmath442 ; see lemma [ lem.vandermonde ] .
we claim we can randomly select @xmath35 indices @xmath443 and test whether @xmath444 is full spark to determine whether @xmath439 .
moreover , after performing this test for each @xmath438 , the probability of incorrectly determining @xmath445 is @xmath446 , provided @xmath441 is sufficiently large .
we want to test whether @xmath444 is full spark and use the result as a proxy for whether @xmath439 . for this to work ,
we need to have @xmath447 precisely when @xmath448 for every @xmath449 of size @xmath35 . to this end
, it suffices to have the nullspace @xmath450 of @xmath451 intersect trivially with the column space of @xmath452 for every @xmath453 . to be clear , it is always the case that @xmath454 , and so @xmath455 implies @xmath456 .
if we further assume that @xmath457 , then the converse also holds . to see this ,
suppose @xmath456 .
then by the rank - nullity theorem , there is a nontrivial @xmath458 . since @xmath459 , we must have @xmath460 , which in turn implies @xmath461 since @xmath457 by assumption .
thus , @xmath455 by the rank - nullity theorem .
now fix @xmath449 of size @xmath35 such that @xmath448 .
we will show that the vast majority of choices @xmath443 of size @xmath35 satisfy @xmath457 . to do this , we consider the columns @xmath462 of @xmath463 one at a time , and we make use of the fact that @xmath464 . in particular , since @xmath167 is full spark , there are at most @xmath170 columns of @xmath167 in the orthogonal complement of @xmath465 , and so there are at least @xmath466 choices of @xmath467 for which @xmath468 does not contain @xmath465 , i.e. , @xmath469 similarly , after selecting the first @xmath470 @xmath471 s , we have @xmath472 , where @xmath473 again , since @xmath167 is full spark , there are at most @xmath474 columns of @xmath167 in the orthogonal complement of @xmath475 , and so the remaining @xmath476 columns are candidates for @xmath477 that give @xmath478 overall , if we randomly pick @xmath443 of size @xmath35 , then @xmath479 where the final step is by bernoulli s inequality .
taking a union bound over all choices of @xmath449 and all values of @xmath438 then gives @xmath480 thus , to make the probability of failure @xmath446 , it suffices to have @xmath481 . in summary , we succeed in representing the original matroid with probability @xmath437 , and then we succeed in determining the spark of its representation with probability @xmath437 .
the probability of overall success is therefore @xmath482 .
since our success probability is , at worst , polynomially small , we can apply amplification to achieve an exponentially small error probability .
our use of random linear projections in the above reduction to full spark is similar in spirit to valiant and vazirani s use of random hash functions in their reduction to unique sat @xcite . since their randomized reduction is the canonical example thereof , we find our reduction to be particularly natural . as a final note
, we clarify that theorem [ thm.full spark hard ] is a statement about the amount of computation necessary in the _
worst case_. indeed , the hardness of full spark does not rule out the existence of smaller classes of matrices for which full spark is easily determined . as an example , theorem [ thm.uniformly distributed ] determines full spark in the special case where the matrix is formed by rows of a dft of prime - power order .
this illustrates the utility of applying additional structure to efficiently solve the full spark problem , and indeed , such classes of matrices are rather special for this reason .
davenport , m.a . ,
duarte , m.f . ,
eldar , y.c . ,
kutyniok , g. : introduction to compressed sensing . in : eldar , y.c .
, kutyniok , g. ( eds . ) , compressed sensing : theory and applications , cambridge university press ( 2011 ) | finite frame theory has a number of real - world applications . in applications like sparse signal processing , data transmission with robustness to erasures , and reconstruction without phase ,
there is a pressing need for deterministic constructions of frames with the following property : every size-@xmath0 subcollection of the @xmath0-dimensional frame elements is a spanning set .
such frames are called full spark frames , and this paper provides new constructions using the discrete fourier transform .
later , we prove that full spark parseval frames are dense in the entire set of parseval frames , meaning full spark frames are abundant , even if one imposes an additional tightness constraint . finally , we prove that testing whether a given matrix is full spark is hard for @xmath1 under randomized polynomial - time reductions , indicating that deterministic full spark constructions are particularly significant because they guarantee a property which is otherwise difficult to check . |
in time - independent domains the skorohod problem , in the form studied in this article , goes back to tanaka @xcite , who established existence and uniqueness of solutions to the skorohod problem in convex domains with normal reflection .
these results were subsequently generalized to wider classes of time - independent domains by , in particular , lions and sznitman @xcite and saisho @xcite . by imposing an admissibility condition on the domain , lions and sznitman @xcite proved existence and uniqueness of solutions to the skorohod problem in two different cases .
the first of the two cases considered normal reflection on domains satisfying a uniform exterior sphere condition , meaning that the domain is smooth except for `` convex corners . ''
moreover , the second case considered smoothly varying ( possibly oblique ) directions of reflection on smooth domains . in addition , for smoothly varying directions of reflection on domains satisfying a uniform exterior sphere condition , existence and uniqueness results were obtained in the special case when the oblique reflection cone can be transformed into the normal cone by multiplication by a smooth matrix function .
saisho saisho1987 later showed that in the first case considered in @xcite , that is , for normal reflection , the admissibility condition is not necessary and can be removed .
moreover , concerning oblique reflection , that is , when the cone of reflection differs from the cone of inward normals , we note that in the case of an orthant with constant directions of reflection on the sides , harrison and reiman @xcite found sufficient conditions for the existence and uniqueness of solutions to the skorohod problem as well as for continuity of the reflection map . in this context
we also mention that bernard and el kharroubi @xcite provided necessary and sufficient conditions for the existence of solutions to the skorohod problem in an orthant with constant directions of reflection on each face .
the most general results so far concerning the existence of solutions to the skorohod problem with oblique reflection in time - independent domains were derived by costantini @xcite .
costantini @xcite proved existence of solutions to the skorohod problem for domains satisfying a uniform exterior sphere condition with a nontangential reflection cone given as a continuous transformation of the normal cone .
note that this allowed for discontinuous directions of reflection at the corners .
the question of uniqueness of solutions to the skorohod problem with oblique reflection is , in general , still an open question and has been settled only in some specific cases . for example ,
dupuis and ishii @xcite obtained uniqueness for a convex polyhedron with constant directions of reflection on the faces assuming the existence of a certain convex set , defined in terms of the normal directions and the directions of reflection .
dupuis and ishii @xcite later extended this result to piecewise smooth domains with smoothly varying directions of reflection on each face .
in addition , we here also mention the work of dupuis and ramanan @xcite based on convex duality techniques .
in particular , in @xcite convex duality is used to transform the condition of dupuis and ishii @xcite into one that is much easier to verify . before we proceed , we here note that the outline above is an attempt to briefly discuss relevant previous developments concerning the skorohod problem in the form studied in this article . in particular , the study of reflected diffusion based on skorohod problems was first introduced by skorohod @xcite and this approach has , as briefly described , subsequently been developed in many articles , including @xcite and @xcite .
however , we emphasize that the literature devoted to skorohod problems , their extensions and applications is much larger than what is conveyed above and , in fact , many more researcher have contributed to this rich field . in particular , applied areas where skorohod problems occur include heavy traffic analysis of queueing networks ( see , e.g , @xcite ) , control theory , game theory and mathematical economics ( see , e.g. , @xcite ) , image processing ( see , e.g. , @xcite ) and molecular dynamics ( see , e.g. , @xcite ) . for further results concerning skorohod problems , as well as applications of skorohod problems , we also refer to @xcite and @xcite
. an important novelty of this article is that we conduct a thorough study of the skorohod problem , and the subsequent applications to stochastic differential equations reflected at the boundary , in the setting of time - dependent domains . to our knowledge
, the skorohod problem is indeed less developed in time - dependent domains . in particular , a first treatment of the skorohod problem in time - dependent domains was given by costantini , gobet and el karoui @xcite , who proved existence and uniqueness of solutions to the skorohod problem with normal reflection in smooth time - dependent domains .
moreover , existence and uniqueness for deterministic problems of skorohod type in time - dependent intervals have recently also been established by burdzy , chen and sylvester @xcite , burdzy , kang and ramanan @xcite .
the main contribution of this article is that we are able to generalize the results in @xcite , concerning cdlg solutions to the skorohod problem with oblique reflection , to time - dependent domains assuming less regularity on the domains compared to @xcite .
note also that in @xcite we use the results of this article to construct a numerical method for weak approximation of stochastic differential equations with oblique reflection in time - dependent domains .
finally , as in @xcite , we note , in particular , that reflecting brownian motions in time - dependent domains arise in queueing theory ( see , e.g. , @xcite ) , statistical physics , ( see , e.g. , @xcite ) , control theory ( see , e.g. , @xcite ) and finance ( see , e.g. , @xcite ) .
in particular , in future articles we hope to be able to further explore the results and techniques developed in this article in several applications . to properly formulate the multidimensional skorohod problem considered in this article , and our results , we in the following first have to introduce some notation . given @xmath5 , we let @xmath6 denote the standard inner product on @xmath7 and we let @xmath8 be the euclidean norm of @xmath9 whenever @xmath10 , we let @xmath11 and @xmath12 . moreover ,
given @xmath13 , @xmath14 , we let @xmath15 , @xmath16 be the closure of @xmath2 and @xmath17 , respectively , and we let @xmath18 denote the euclidean distance from @xmath19 to @xmath17 .
given @xmath5 , @xmath20 and an open , connected set @xmath21 , we will refer to @xmath22\times \mathbb{r } ^{d } ) , \label{timedep}\ ] ] as a time - dependent domain .
given @xmath2 and @xmath23 $ ] , we define the time sections of @xmath2 as @xmath24 , and we assume that@xmath25 . \label{timedep+}\ ] ] we let @xmath26 and @xmath27 , for @xmath23 $ ] , denote the boundaries of @xmath2 and @xmath28 , respectively .
a convex cone of vectors in @xmath7 is a subset @xmath29 such that @xmath30 for all @xmath31 and all @xmath32 .
we let @xmath33 be a function defined on @xmath34 such that @xmath35 is a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath23 $ ] . to give an example of a closed convex cone
, we consider the set @xmath37 , where @xmath38 is a closed , connected subset of @xmath39 satisfying @xmath40 for all @xmath41 .
given @xmath42 , we define @xmath43 .
then @xmath44 is an example of a closed convex cone and we note that @xmath45 , where @xmath46 can be viewed as the `` convex hull '' of @xmath38 on @xmath39 . given @xmath47 , we let @xmath48 .
given @xmath20 , we let @xmath49 , \mathbb{r } ^{d } ) $ ] denote the set of cdlg functions
@xmath50 \rightarrow \mathbb{r } ^{d}$ ] , that is , functions which are right continuous with left limits . for @xmath51 , \mathbb{r } ^{d } ) $ ] we introduce the norm @xmath52 for @xmath53 and , given @xmath54 , we let @xmath55 , \mathbb{r } ^{d } ) = \bigl\ { w\in{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \dvtx \sup_{t } \vert w_{t}-w_{t^{- } } \vert<\delta \bigr\ } \label{vvv1}\ ] ] denote the set of cdlg functions with jumps bounded by @xmath56 .
we denote the set of functions @xmath57
\rightarrow \mathbb{r } ^{d}$ ] with bounded variation by @xmath58 , \mathbb{r } ^{d } ) $ ] and we let @xmath59 denote the total variation of @xmath60 , \mathbb{r } ^{d } ) $ ] . in this article
we consider the skorohod problem in the following form .
[ skorohodprob ] let @xmath5 and @xmath20 .
let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and let @xmath47 be , for every @xmath36 , @xmath61 $ ] , a closed convex cone of vectors in @xmath7 .
given @xmath62 , \mathbb{r } ^{d } ) $ ] , with @xmath63 , we say that the pair @xmath64 is a solution to the skorohod problem for @xmath65 , on @xmath66 $ ] , if @xmath67 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] and if @xmath68 satisfies , for all @xmath69 $ ] , @xmath70 } \partial d_{s}\end{aligned}\ ] ] and @xmath71 \dvtx ( t , x_{t } ) \in d \ }
\bigr ) = 0 .
\label{sp4}\ ] ] the main results of this article will be proved for time - dependent domains @xmath13 satisfying ( [ timedep+ ] ) .
however , several additional restrictions will be imposed on @xmath2 , on the cones of reflection @xmath72 as well as on the interaction between @xmath2 and @xmath73 . in the following
we will outline these assumptions in order to be able to properly state our existence result concerning the skorohod problem with oblique reflection .
however , while these assumptions are introduced quite briefly here , the intuition behind the assumptions , as well as the implications of the assumptions , is explained in more detail in section [ geometry ] below .
we let @xmath74 denote the cone of inward normal vectors at @xmath36 , @xmath75 $ ] ; see ( [ conenorm ] ) below for a definition .
in particular , we assume that @xmath76 whenever @xmath36 , @xmath75 $ ] .
note that we allow for the possibility of several inward normal vectors at the same boundary point .
given @xmath77 , we let @xmath78 .
then the spatial domain @xmath28 is said to verify the uniform exterior sphere condition if there exists a radius @xmath79 such that@xmath80\times \mathbb{r } ^{d}\setminus d_t)\cap(\mathbb{r } ^{d+1}\setminus d ) , \label{extsphere-}\ ] ] whenever @xmath81 , @xmath82 .
note that @xmath83 is the open euclidean ball with center @xmath84 and radius @xmath85 .
we say that a time - dependent domain @xmath2 satisfies a uniform exterior sphere condition in time if the uniform exterior sphere condition in ( [ extsphere- ] ) holds , with the same radius @xmath85 , for all spatial domains @xmath28 , @xmath23 $ ] .
following @xcite , we let @xmath86 } } _ { \vert s - t \vert\leq r}\sup_{z\in\overline{d_{s}}}d ( z , d_{t } ) \label{defl}\ ] ] be the modulus of continuity of the variation of @xmath2 in time .
in particular , in several of our estimates related to the skorohod problem we will assume that @xmath87 following @xcite , we assume that @xmath88 \\[-8pt ] & & \qquad \mbox{for all } z\in\partial d_{t } , t\in [ 0,t ] .\nonumber\end{aligned}\ ] ] the assumption in ( [ cone1 ] ) eliminates the possibility of @xmath73 containing vectors in opposite directions .
we also assume that the set@xmath89 \ } \mbox { is closed . }
\label{cone2}\ ] ] the interpretation of the condition in ( [ cone2 ] ) is discussed in section [ geometry ] .
in addition , we need the following assumption concerning the variation of the cones @xmath90 .
let @xmath91 denote the hausdorff distance between the sets @xmath92 .
moreover , let @xmath93 be a sequence of points in @xmath34 , @xmath94 $ ] , @xmath95 , such that @xmath96 $ ] , @xmath97 .
we assume , for any such sequence of points @xmath93 , that @xmath98 for @xmath99 , @xmath100 $ ] , and @xmath101 we define@xmath102 and @xmath103 for technical reasons we also introduce the quantity@xmath104 in the proof of certain a priori estimates for the skorohod problem , established in the bulk of the article , we will consider time - dependent domains satisfying ( [ timedep+ ] ) and the uniform exterior sphere condition in time , with radius @xmath85 .
in addition , we will assume that there exist @xmath105 and @xmath106 , such that@xmath107 } \inf_{z\in\partial d_{s}}a_{s , z } ( \rho _ { 0},\eta_{0 } ) & = & a>0 , \\\label{crite } \sup_{s\in [ 0,t ] } \sup_{z\in\partial d_{s}}e_{s , z } ( \rho _ { 0},\eta_{0 } ) & = & e<1.\end{aligned}\ ] ] interpretations of ( [ adef ] ) , ( [ cdef ] ) , ( [ crita ] ) and ( [ crite ] ) are given in section [ geometry ] .
let @xmath108 , @xmath109 and let @xmath33 be given for all @xmath36 , @xmath75 $ ] .
we say that @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 if there exists , for any @xmath112 , @xmath75 $ ] , such that @xmath113 at least one projection of @xmath114 onto @xmath27 along @xmath115 , denoted @xmath116 , which satisfies@xmath117 concerning the existence and continuity of solutions to the skorohod problem , as defined in definition [ skorohodprob ] , we prove the following two theorems . [ theorem 1 ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) , ( [ limitzero ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
let @xmath118 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath23 $ ] , and assume that @xmath73 satisfies ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) .
assume that ( [ crita ] ) and ( [ crite ] ) hold for some @xmath119 , @xmath106 , @xmath120 and @xmath121 .
finally , assume that @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 , for some @xmath122 , @xmath109 , as defined in ( [ deltanolldef ] ) and ( [ rhonolldef ] ) .
then , given @xmath123 , \mathbb{r } ^{d } ) $ ] , with @xmath63 , there exists a solution @xmath64 to the skorohod problem for @xmath124 , in the sense of definition [ skorohodprob ] , with @xmath125,\mathbb{r } ) $ ] .
[ theorem 2]assume that the assumptions stated in theorem [ theorem 1 ] are satisfied and let @xmath126 be as in the statement of theorem theorem 1 .
let @xmath127 \rightarrow \mathbb{r } ^{d}$ ] be a continuous function and let @xmath128 be any solution to the skorohod problem for @xmath129 in the sense of definition [ skorohodprob ] .
if @xmath130 , \mathbb{r } ^{d } ) $ ] , then @xmath4 is continuous . in the following remarks
we have gathered comments concerning the importance of the assumptions imposed in theorems [ theorem 1 ] and theorem 2 , as well as comments concerning situations when these assumptions are fulfilled .
[ remarklex1 ] our proofs of theorems [ theorem 1 ] and theorem 2 rely , as outlined below , on certain a priori estimates proved in section [ sectcomp ] .
these estimates are proved assuming that @xmath13 is a time - dependent domain satisfying ( [ timedep+ ] ) , ( [ limitzero ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
furthermore , to derive these estimates , we also assume that ( [ crita ] ) and ( [ crite ] ) hold for some @xmath105 , @xmath106 , @xmath120 and @xmath121 and that @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 , for some @xmath122 , @xmath109 , as defined in ( [ deltanolldef ] ) and ( [ rhonolldef ] ) .
in particular , we do not have to assume that @xmath131 satisfies ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) in order to derive the results in section [ sectcomp ] .
[ remarklex1 + ] in section [ sectskorgen ] we proceed toward the final proof of theorem [ theorem 1 ] . in particular , we use the a priori estimates of section [ sectcomp ] to derive general results concerning the convergence of solutions to skorohod problems in time - dependent domains .
we note that our assumptions on @xmath2 do not exclude the possibility of holes in @xmath2 and @xmath28 , for some @xmath75 $ ] . nevertheless
, the assumptions on @xmath2 ensure that the number of holes in @xmath28 stays the same for all @xmath23 $ ] and that these holes can not shrink too much as time changes .
this observation , lemma [ llhatequivalence ] below and its proof allow us to conclude the validity of the conclusion in remark [ remsim ] , which , in turn , is used to complete the proofs in section [ sectskorgen ] .
simple examples show that the conclusion in remark [ remsim ] would not hold if we , for instance , allowed the number of holes in @xmath28 to change as a function of @xmath132 and if we , in particular , allowed the holes to vanish .
[ remarklex1++ ] the assumption that @xmath133 satisfies ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) is used to complete the proofs in section [ sectskorgen ] . in particular , focusing on theorem [ spconvergence+ ] , which is the convergence result actually used in the proof of theorem [ theorem 1 ] , we note that we need to assume ( [ gammanpre ] ) in order to be able derive ( [ bbbb2again ] ) .
we then use ( [ cone2 ] ) to complete the argument in the proof of theorem spconvergence+ .
note also the difference between ( [ cone2 ] ) and ( [ gammanpre ] ) .
assumption ( [ cone2 ] ) simply states that if @xmath134 is a sequence such that @xmath135 $ ] , and if @xmath136 in @xmath137 , for some @xmath138 , then @xmath139 $ ] .
assumption ( [ gammanpre ] ) , on the other hand , is a statement concerning the convergence , in the hausdorff distance sense , of the cones @xmath140 .
finally , to comment on assumption ( [ cone1 ] ) , which was also imposed in @xcite , we note that ( [ cone1 ] ) is only used in the proofs of theorems spconvergence and [ spconvergence+ ] and , in particular , in the verification of ( [ ddd5 ] ) and ( [ ddd3 ] ) .
assumption ( [ cone1 ] ) eliminates the possibility of @xmath73 containing vectors in opposite directions and we have not been able to complete our argument without this assumption . however , there are articles dealing with skorohod type lemmas and reflected brownian motion ; see @xcite , in particular , where this assumption is not required . as noted in @xcite , the inclusion of vectors in opposite directions can be viewed as a critical case and @xcite considers a related problem in a particular setting in the plane . in our general case
we leave this question as a subject for future research .
[ remarklex2 ] for examples of cases when the geometric assumptions imposed in theorems [ theorem 1 ] and [ theorem 2 ] are fulfilled , we refer to
. however , we here briefly discuss theorems theorem 1 and [ theorem 2 ] in the context of convex domains .
in particular , let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and ( [ limitzero ] ) .
assume , in addition , that @xmath28 is convex whenever @xmath75 $ ] .
let @xmath131 be as in the statement of theorem [ theorem 1 ] .
assume that @xmath141 } \inf_{z\in\partial d_{s}}a_{s , z } ( \rho,\eta ) & = & a>0 , \\\label{critecon } \lim_{\eta\rightarrow0}\lim_{\rho\rightarrow0}\sup_{s\in [ 0,t ] } \sup_{z\in\partial d_{s}}e_{s , z } ( \rho,\eta ) & = & e<1.\end{aligned}\ ] ] if @xmath28 is convex whenever @xmath75 $ ] , then there exists , for every @xmath142 given , @xmath109 such that @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 as defined in ( [ deltanolldef ] ) and ( [ rhonolldef ] ) . in this case
the conclusion of theorem [ theorem 1 ] is , as can be seen from the proofs below , that given @xmath143 , \mathbb{r } ^{d } ) $ ] , with @xmath63 , there exists a solution @xmath64 to the skorohod problem for @xmath124 , in the sense of definition [ skorohodprob ] , with @xmath144,\mathbb{r } ) $ ] .
moreover , if @xmath3 is a continuous function , then @xmath4 is continuous . in particular ,
if the time - slices @xmath145 are convex , then the restrictions , in theorems [ theorem 1 ] and [ theorem 2 ] , on the jump - sizes in terms of @xmath146 , @xmath126 can be removed . moreover ,
this is consistent with the results in @xcite valid in time - independent domains ; see theorem 4.1 and proposition 2.3 in @xcite .
we next formulate a subsequent application of theorems [ theorem 1 ] and [ theorem 2 ] to the problem of constructing weak solutions to stochastic differential equations in @xmath147 with reflection along @xmath115 on @xmath27 for all @xmath23 $ ] .
given @xmath20 , we let @xmath148 , \mathbb{r } ^{d } ) $ ] denote the class of continuous functions from @xmath149 $ ] to @xmath7 . in the following , we let @xmath150 be a positive integer and we let @xmath151 and @xmath152 be given functions which are bounded and continuous .
[ weaksol ] let @xmath5 and @xmath20 .
let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) , let @xmath153 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath75 $ ] , and let @xmath154 .
a weak solution to the stochastic differential equation in @xmath147 with coefficients @xmath155 and @xmath156 , reflection along @xmath115 on @xmath27 , @xmath23 $ ] , and with initial condition @xmath157 at @xmath158 , is a stochastic process @xmath159 with paths in @xmath148 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] , which is defined on a filtered probability space @xmath160 and satisfies , @xmath161-almost surely , whenever @xmath23 $ ] , @xmath162 \dvtx x_{t}^{0,\hat{z}}\in d_{t } \ } ) = 0.\end{aligned}\ ] ] here @xmath163 is a @xmath150-dimensional wiener process on @xmath164 and @xmath165 is @xmath166-adapted .
concerning weak solutions to stochastic differential equations in @xmath147 with oblique reflection along @xmath26 , we prove the following theorem .
[ theorem 3]let @xmath20 , @xmath13 and @xmath131 be as in the statement of theorem [ theorem 1 ] .
let @xmath167 and @xmath152 be given , bounded and continuous functions on @xmath147 and let @xmath168 .
then there exists a weak solution , in the sense of definition [ weaksol ] , to the stochastic differential equation in @xmath147 with coefficients @xmath155 and @xmath156 , reflection along @xmath115 on @xmath27 , @xmath23 $ ] , and with initial condition @xmath169 at @xmath158 .
we note that theorem [ theorem 3 ] generalizes the corresponding results in @xcite and @xcite .
furthermore , we note that there has recently been considerable activity in the study of reflected diffusions in time - dependent intervals .
in particular , in this context we mention @xcite and @xcite and we refer the interested reader to these articles for more information as well as for references to other related articles .
the rest of the article is organized as follows . in section [ contri ]
we first briefly outline two general and important themes present in the proofs of the results in this article .
the first theme concerns a priori estimates and compactness for solutions to skorohod problems and the second theme concerns convergence results for sequences of solutions to skorohod problems .
second , we discuss the proofs of theorems [ theorem 1 ] , [ theorem 2 ] and [ theorem 3 ] and we try to point out the new difficulties occurring due to the time - dependent character of the domain .
this section is included for further reference and , in particular , to convey some of the ideas to the reader . in section [ prel ]
we introduce additional notation , outline the restrictions imposed on @xmath2 and @xmath73 and collect a few notions and facts from the skorohod topology .
there is also an appendix attached to section [ prel ] , . in
we state sufficient conditions for the @xmath170-property of good projections along @xmath73 and we give examples of time - dependent domains satisfying the assumptions stated in theorems [ theorem 1 ] , [ theorem 2 ] and [ theorem 3 ] .
section [ sectcomp ] is devoted to estimates for solutions to the skorohod problem , with oblique reflection , which have bounded jumps and also to the corresponding estimates for certain approximations of the skorohod problem . in section [ sectskorgen ]
we first prove theorem spconvergence , containing a general result concerning convergence of solutions to skorohod problems in time - dependent domains .
furthermore , we establish the somewhat similar result for certain approximations of the skorohod problem .
the latter estimates are then used in the proof of theorem [ theorem 1 ] .
the final proofs of theorems [ theorem 1 ] , theorem 2 and [ theorem 3 ] are given in section [ sectskor ] .
the article ends with the , discussed above .
concerning proofs , we note that the arguments in this article follow two general and important themes which we here , to start with , briefly outline . to explain the a priori estimates , we let @xmath20 , @xmath13 and @xmath131 be as in the statement of theorem [ theorem 1 ] , and we let @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 .
assume that @xmath171 is a solution to the skorohod problem for @xmath65 such that @xmath172 , \mathbb{r } ^{d } ) $ ] .
under these assumptions , we prove ( see theorem compacttheorem below ) that there exist positive constants @xmath173 , @xmath174 , @xmath175 and @xmath176 such that @xmath177 \\[-8pt ] \vert w \vert _ { t_{1},t_{2}}+l_{4 } ( w , t )
l ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] whenever @xmath53 .
furthermore , we prove that if @xmath178 , \mathbb{r } ^{d } ) $ ] is relatively compact in the skorohod topology and @xmath179 , whenever @xmath180 , then there exist positive constants @xmath181 , @xmath182 , @xmath183 and @xmath184 , such that@xmath185 the a priori estimates and compactness result in ( [ apriori ] ) and ( [ apriori2 ] ) are useful for proving convergence of solutions to skorohod problems .
to explain this further , let @xmath186 be a sequence of time - dependent domains @xmath187 satisfying ( [ timedep+ ] ) and let @xmath188 be a sequence of closed convex cones of vectors in @xmath7 .
assume that @xmath189 and @xmath190 satisfy the conditions stated in theorem theorem 1 with constants that are `` uniform with respect to @xmath191 '' in a sense made precise in section [ sectskorgen ] .
let @xmath192 , with @xmath193 , be a sequence in @xmath194 , \mathbb{r } ^{d } ) $ ] which is relatively compact in the skorohod topology and which converges to @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 .
furthermore , let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) , let @xmath195 be , for every @xmath36 , @xmath196 $ ] , a closed convex cone of vectors in @xmath7 satisfying ( [ cone1 ] ) and ( [ cone2 ] ) .
assume that the sequences @xmath189 and @xmath197 converge to @xmath2 and @xmath73 , respectively , in a sense specified in theorem spconvergence .
if there exists , for all @xmath198 , a solution @xmath199 to the skorohod problem for @xmath200 such that @xmath201 , for all @xmath202 $ ] , and @xmath203 , \mathbb{r } ^{d } ) $ ] , then it follows , using ( [ apriori ] ) and ( [ apriori2 ] ) , that @xmath204 is relatively compact in @xmath194 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } _ { + } ) $ ] .
hence , we are able to conclude that @xmath205 converges to some @xmath206 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] with @xmath207 and we can , in addition , prove that @xmath208 is indeed a solution to the skorohod problem for @xmath65 .
this result , found in theorem [ spconvergence ] below , constitutes a general convergence result for sequences of solutions to skorohod problems based on the a priori estimates and compactness result in ( [ apriori ] ) and ( [ apriori2 ] ) .
although theorem [ theorem 1 ] does not follow directly from the results outlined above , we claim that ( [ apriori ] ) , ( [ apriori2 ] ) and theorem [ spconvergence ] , stated below , are of independent interest and may be useful in other applications involving the skorohod problem . to start an outline of the actual proofs of theorems [ theorem 1 ] , theorem 2 and [ theorem 3 ] , we note that to prove theorem theorem 1 we use arguments similar to those outlined above , but in this case we have to construct , given @xmath1 , an approximating sequence @xmath209 such that a solution @xmath210 to the skorohod problem for @xmath211 can be found explicitly . to discuss the construction of @xmath212 and @xmath213 used in the proof of theorem [ theorem 1 ] , we consider @xmath214 , \mathbb{r } ^{d } ) $ ] , with @xmath63 and with jumps bounded by some constant , and we now let @xmath215 define a partition @xmath216 of the interval @xmath217 $ ] , that is , @xmath218 . given @xmath216 , we let @xmath219 and , given @xmath216 and @xmath3 , we define @xmath220 and @xmath221
. then @xmath222 , \mathbb{r } ^{d } ) $ ] is a step function approximation of @xmath3 .
furthermore , assume that @xmath216 and @xmath223 are such that @xmath224 whenever @xmath225 . recall that @xmath226 is the constant appearing in the notion of the @xmath111-property of good projections .
we next define @xmath227 \\[-8pt ] \gamma_{t}^{\delta}&=&\gamma_{\tau _ { k-1 } } \qquad\mbox{whenever } t\in [ \tau_{k-1},\tau_{k } ) , k\in \ { 1,\ldots , n \ } , \nonumber\end{aligned}\ ] ] and @xmath228 . given @xmath223 , @xmath229 and @xmath230 as above
, we define a pair of processes @xmath231 as follows .
we let@xmath232 if @xmath233 for some @xmath234 , then , by the triangle inequality and ( [ korlim ] ) , @xmath235 hence , by the @xmath236-property of good projections , it follows that if @xmath237 , then there exists a point @xmath238 which is the projection of @xmath239 onto @xmath240 along @xmath241 .
furthermore , if @xmath242 , then we let @xmath243 based on this argument , we define , whenever @xmath244 , @xmath245 , @xmath246 \\[-8pt ] \lambda_{t}^{\delta } & = & \lambda_{\tau_{k-1}}^{\delta } + \bigl(x_{t}^{\delta } -(x_{\tau_{k-1}}^{\delta}+w_{\tau_{k}}^{\delta}-w_{\tau _ { k-1}}^{\delta } ) \bigr).\nonumber\end{aligned}\ ] ] finally , we define @xmath247 and @xmath248 as in ( [ approxw+++ ] ) by putting @xmath249 in ( [ approxw+++ ] ) . by construction ,
the pair @xmath250 is a solution to the skorohod problem for @xmath251 .
moreover , using the assumption on the size of the jumps of @xmath3 stated in theorem [ theorem 1 ] , we will be able to make the construction so that we can conclude that @xmath252 , \mathbb{r } ^{d } ) $ ] .
as the next step we then apply theorem compacttheoremapprox stated below , showing the existence of positive constants @xmath253 , @xmath254 , @xmath255 and @xmath256 such that @xmath257 \\[-8pt]\qquad { l}_{3 } ( w , t ) \vert w \vert_{t_{1},t_{2}}+\hat{l}_{4 } ( w , t ) \bigl(l ( t_{2}-t_{1 } ) + l(\delta^{\ast})\bigr),\nonumber\end{aligned}\ ] ] whenever @xmath53 . provided with the estimates in ( [ lll1 ] ) , we are then able to prove theorem [ theorem 1 ] by means of compactness arguments similar to those outlined above .
indeed , we construct an appropriate sequence of partitions @xmath258 , based on @xmath3 , such that @xmath259 , \mathbb{r } ^{d } ) $ ] for @xmath191 larger than some @xmath260 and such that @xmath261 is a solution to the skorohod problem for @xmath262 for @xmath263 .
then , using ( [ lll1 ] ) , we conclude that @xmath264 is a relatively compact sequence in the skorohod topology and that @xmath265 converges in the sense of the skorohod topology to a pair of functions @xmath128 .
note that an important difference here , compared to the situation outlined above , is that @xmath229 and @xmath230 as defined in ( [ approxw+ ] ) are discontinuous in time . to be able to handle this situation ,
we employ some additional arguments , similar to the ones in the proof of theorem spconvergence , in order to prove that @xmath0 is a solution to the skorohod problem for @xmath129 on @xmath66 $ ] .
this completes the proof of theorem [ theorem 1 ] . concerning theorem theorem 2
, we see that this theorem follows immediately from the continuity of @xmath3 and ( [ limitzero ] ) using the estimates in ( [ apriori ] ) . to prove theorem [ theorem 3 ] , we argue somewhat similarly as in the proof of theorem [ theorem 1 ] and we refer to the bulk of the article for details . to conclude
, we note that the proof of theorem [ theorem 1 ] is more involved compared to the proof of the corresponding result for time - independent domains established in @xcite and that new difficulties occur , naturally , due to the fact that we are considering time - dependent domains . in the time - independent case a solution @xmath266 to the skorohod problem for @xmath1
is constructed as the limit of a sequence @xmath267 , where @xmath268 is a solution to a skorohod problem based on @xmath269 . in this case
@xmath270 is a solution to a skorohod problem for @xmath271 , while in our case @xmath272 is a solution to a skorohod problem for @xmath273 .
hence , in the time - dependent case we , at each step , also have to discretize and approximate @xmath2 and @xmath73 due to the time - dependent character of the domain .
in particular , the fact that @xmath274 and @xmath275 , as defined in ( [ approxw+ ] ) , are discontinuous in time induces several new difficulties which we have to overcome in order to complete the proof of theorem [ theorem 1 ] .
in this section we introduce notation , collect a number of preliminary results concerning the geometry of time - dependent domains and recall a few notions and facts from the skorohod topology .
points in euclidean @xmath276-space @xmath34 are denoted by @xmath277 . given a differentiable function @xmath278 defined on @xmath279 ,
we let @xmath280 denote the partial derivative of @xmath281 at @xmath282 with respect to @xmath283 and we let @xmath284 denote the gradient @xmath285 .
higher order derivatives of @xmath281 with respect to the space variables will often be denoted by @xmath286 , @xmath287 and so on . furthermore , given a multi - index @xmath288 , @xmath289
, we define @xmath290 and we let @xmath291 denote the associated partial derivative of @xmath292 with respect to the space variables .
time derivatives of @xmath281 will be denoted by @xmath293 where @xmath294 .
as in the , we let @xmath6 denote the standard inner product on @xmath7 and we let @xmath8 be the euclidean norm of @xmath295 .
whenever @xmath10 , we let @xmath11 and @xmath12 . in addition , @xmath296 denotes the lebesgue @xmath297-measure on @xmath7 .
moreover , given @xmath14 , we let @xmath16 and @xmath298 be the closure and boundary of @xmath17 , respectively , and we let @xmath299 denote the euclidean distance from @xmath300 to @xmath17 . given @xmath301 ,
we let @xmath302 denote the parabolic distance between @xmath303 and @xmath304 and for @xmath305 , we let @xmath306 denote the parabolic distance from @xmath307 to @xmath308 . moreover , for @xmath307 and @xmath309 , we introduce the parabolic cylinder @xmath310 . given two real numbers @xmath120 and @xmath155 , we let @xmath311 and @xmath312 . finally , given a borel set @xmath313 , we let @xmath314 denote the characteristic function associated to @xmath17 . given a time - dependent domain @xmath315 , a function @xmath281 defined on @xmath316 and a constant @xmath317 $ ]
, we adopt the definition on page 46 in @xcite and introduce @xmath318 \\[-8pt ] & & { } + \sum _ { \vert\beta \vert=1}\sup _ { ( t , z ) \in d^{\prime}}\sup _ { ( s , y ) \in d^{\prime}\setminus \ { ( t , z ) \ } } \frac { \vert\partial_{z}^{\beta}f ( t , z ) -\partial_{z}^{\beta}f ( s , y ) \vert } { [ d_{p } ( ( t , z ) , ( s , y ) ) ] ^{\alpha}}.\nonumber\end{aligned}\ ] ] the third term on the right - hand side of @xmath319 is superfluous for our purposes , but we include it here for agreement with the theory of partial differential equations in time - dependent domains ( see @xcite ) . using the norm @xmath320 , we let @xmath321 denote the banach space of functions @xmath281 on @xmath315 with finite @xmath319-norm .
we here outline the geometric restrictions which we impose on the time - dependent domains and cones of reflections .
concerning @xmath2 , we first prove the following auxiliary lemma .
[ llhatequivalence ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and ( [ limitzero ] ) and assume that @xmath2 satisfies a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
let@xmath322 } } _ { \vert s - t \vert\leq r}\sup_{z\in\partial d_{s}}d ( z,\partial d_{t } ) .
\label{lhat}\ ] ] then , @xmath323 for all @xmath309 such that @xmath324 . in the following
we let @xmath325 be arbitrary and we consider @xmath326 small enough to ensure that @xmath327 .
with @xmath328 and @xmath326 fixed , we let @xmath329 $ ] , @xmath330 , be such that @xmath331 naturally , @xmath332 assume @xmath333 .
then we immediately obtain @xmath334 assume , on the contrary , that @xmath335 . in this case , as @xmath324 and @xmath336 satisfies the uniform exterior sphere condition with radius @xmath85 , we can conclude that there exists at least one point @xmath337 and obviously @xmath338 .
furthermore , again applying the uniform exterior sphere condition , we see that @xmath295 minimizes the distance from @xmath339 to @xmath336 .
\\[-8pt ] & \leq&\sup_{y_{z}\in\partial d_{t}}d ( y_{z},d_{s } ) \leq \sup_{y\in\overline{d_{t}}}d ( y , d_{s } ) \leq l ( r ) .\nonumber\end{aligned}\ ] ] combining ( [ lhat1])([lhat3 ] ) , we conclude that @xmath341 .
using ( [ tildel ] ) , it is clear that @xmath342 and then , as @xmath343 is arbitrary , @xmath344 . we next consider the opposite inequality . with @xmath343 and @xmath326 fixed , we let @xmath345 $ ] , @xmath346 , be such that @xmath347 in this case we have @xmath348 and in the following we can assume , without loss of generality , that @xmath349 . then , by the uniform exterior sphere condition , and the fact that @xmath324 , we see that every point @xmath350 can be written as @xmath351 for some @xmath339 and some @xmath352 , @xmath353 .
furthermore , there exists a point @xmath354 , with @xmath355 . once again applying the uniform exterior sphere condition
, we see that @xmath356 minimizes the distance from @xmath357 to @xmath27 and we obtain @xmath358 \\[-8pt]\qquad & \leq&\sup_{z\in\overline{d_{s}}\cap(\mathbb{r } ^{d}\setminus d_{t } ) } \vert\tilde{z}-y_{z } \vert\leq\sup_{\tilde{z}\in\partial d_{s}}d ( \tilde{z},\partial d_{t } ) \leq\hat{l } ( r ) .\nonumber\end{aligned}\ ] ] combining ( [ lhat1+])([lhat5 ] ) , we conclude that @xmath359 . using ( [ tildel+ ] ) , it is clear that @xmath360 and then , as @xmath343 is arbitrary , @xmath361 .
this completes the proof of the lemma .
[ remsim ] note also that the prerequisites of lemma llhatequivalence ensure that the number of holes in @xmath28 stays the same for all @xmath23 $ ] and , in particular , that these holes can not shrink too much as time changes .
furthermore , lemma llhatequivalence and its proof allow us to conclude that @xmath362 whenever @xmath363 $ ] , @xmath364 , @xmath365 .
concerning @xmath73 , we let @xmath33 be a function defined on @xmath34 such that @xmath35 is a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath75 $ ] and we assume that @xmath73 satisfies ( [ cone1 ] ) and ( [ cone2 ] ) . to understand the condition in ( [ cone2 ] ) ,
that is , the assumption that the graph @xmath366 is closed , we observe that one motivation for using a cone of reflection , rather than a single - valued direction of reflection , is to be able to deal with discontinuities in the direction of reflection .
such discontinuities arise , for instance , in the normal direction for a convex polygon . at a point of discontinuity of the direction of reflection
one can use the cone generated by all the limit vectors ( if they exist ) of the direction of reflection . for a cone of reflection , the assumption that the graph @xmath367 is closed provides a form of continuity of the cone .
in fact , for a cone of the form @xmath368 , for some @xmath7-valued function @xmath369 , the assumption that @xmath366 is closed is equivalent to the assumption that the function @xmath370 is continuous as a function of @xmath282 .
the cone @xmath77 of inward normal vectors at @xmath371 , @xmath75 $ ] , is defined as being equal to the set consisting of the union of the set @xmath372 and the set @xmath373\times \mathbb{r } ^{d})\setminus d \ } .
\label{conenorm}\ ] ] note that this definition does not rule out the possibility of several unit inward normal vectors at the same boundary point . given @xmath74
, we let @xmath78 , so that @xmath374 contains the set of vectors in @xmath375 with unit length . moreover , based on @xmath77 , we introduce the set @xmath376 \ } .
\label{gnormal}\ ] ] the spatial domain @xmath28 is said to verify the uniform exterior sphere condition if there exists a radius @xmath79 such that ( [ extsphere- ] ) holds .
it is easy to see that ( [ extsphere- ] ) is equivalent to the statement that @xmath377 for all @xmath378 , @xmath82 and @xmath379 .
moreover , as deduced from remark 2.1 in @xcite , the uniform exterior sphere condition in time asserts that @xmath77 is a closed convex cone for all @xmath36 , @xmath75 $ ] and that @xmath380 is closed .
for @xmath99 , @xmath100 $ ] , and @xmath381 , recall the definition of the quantity @xmath382 introduced in ( [ adef ] ) , @xmath383 the vector @xmath384 that maximizes the minimum of @xmath385 over all vectors @xmath386 in a time space neighborhood of a point @xmath387 , @xmath99 , @xmath100 $ ] , can be regarded as the best approximation of the @xmath388-vectors in that neighborhood . with this interpretation @xmath389
represents the cosine of the largest angle between the best approximation and a @xmath388-vector in the neighborhood .
hence , in a sense , @xmath389 quantifies the variation of @xmath73 in a space time neighborhood of @xmath390 . for @xmath391 ,
@xmath100 $ ] and @xmath381 , recall the definition of the quantity @xmath392 introduced in ( [ cdef ] ) , @xmath393 this quantity is close to one if the vectors @xmath394 , in a time space neighborhood , deviate much from the normal vectors and/or the domain is very concave .
hence , in a sense , @xmath395 quantifies the skewness of @xmath73 and the concavity of @xmath2 .
note that ( [ cdef ] ) implies@xmath396 for all @xmath397 , @xmath398 , @xmath399 and @xmath400 with @xmath379 , @xmath401 \subset [ 0,t ] $ ] .
this condition exhibits some similarity with the uniform exterior sphere property ( [ extsphere ] ) . finally , recall the definition of the quantity @xmath402 introduced in ( [ edef ] ) ,
@xmath403 furthermore , as stated in the , in the subsequent section we prove estimates related to the skorohod problem in time - dependent domains satisfying ( [ timedep+ ] ) and the uniform exterior sphere condition in time , with radius @xmath85 .
moreover , to derive these estimates , we also assume that there exist @xmath105 and @xmath106 , such that the assumptions in ( [ crita ] ) and ( [ crite ] ) hold , that is , @xmath404 } \inf_{z\in\partial d_{s}}a_{s , z } ( \rho _ { 0},\eta_{0 } ) & = & a>0 , \\ \sup_{s\in [ 0,t ] } \sup_{z\in\partial d_{s}}e_{s , z } ( \rho _ { 0},\eta_{0 } ) & = & e<1.\end{aligned}\ ] ] the function @xmath389 is a straightforward generalization of the function @xmath405 introduced by tanaka @xcite in his treatment of the skorohod problem . here
@xmath406 is a bounded spatial domain and @xmath407 is the set of unit inward normals at @xmath408 . in @xcite and @xcite the condition@xmath409 is used to rule out the case of tangential normal directions ( see costantini1992 for equivalent characterizations of domains satisfying this criterion ) . the functions @xmath395 and @xmath410 , introduced in ( [ cdef ] ) and ( [ edef ] ) , are straightforward generalizations of the functions @xmath411 and @xmath412 , respectively , introduced in @xcite .
moreover , the related functions @xmath413 and @xmath121 , also introduced in @xcite , are useful only in the context of convex domains .
hence , as we here consider general ( possibly nonconvex ) domains , only generalizations of the functions @xmath411 and @xmath412 are useful . for notational simplicity
, we have removed the tilde in our definition of the generalized versions of @xmath411 and @xmath412 .
given @xmath20 , let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
given a point @xmath414\times \mathbb{r } ^{d})\setminus\overline{d}$ ] in a neighborhood of @xmath2 , in this article we heavily use the projection of @xmath282 onto @xmath26 along the vectors in the cone @xmath73 .
while such projections can be defined in several ways , in this article we here only consider projections in space along vectors @xmath415 , @xmath416 , onto @xmath27 . with this restriction
, the analysis of section [ sectcomp ] in @xcite can be used to derive sufficient conditions for the existence of a projection of a point @xmath417 , onto @xmath27 , along @xmath115 . in other words
, we can determine whether or not there exist , for a given point @xmath417 , a point @xmath416 and a vector @xmath415 such that @xmath418 . in particular , it can be understood when , for @xmath419 , @xmath109 and @xmath33 given
, @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 in the sense defined in the introduction ; see ( [ deltanolldef ] ) and ( [ rhonolldef ] ) .
we refer to the , for more on this , as well as for a discussion of examples of time - dependent domains satisfying the restrictions imposed in theorems theorem 1 , [ theorem 2 ] and [ theorem 3 ] .
let @xmath20 and let @xmath420 , \mathbb{r } ^{d})$ ] .
given a bounded set @xmath421 $ ] , we let@xmath422 then , using lemma 1 on page 122 in @xcite , we see that there exists , for @xmath325 given , a sequence of points @xmath423 , such that@xmath424 in particular , there can only be finitely many points @xmath23 $ ] at which the jump @xmath425 exceeds a given positive number . to proceed , in the following we use the notation and exposition of chapter 3 in @xcite .
we let @xmath426 whenever @xmath427 and we let @xmath428,x , y)$ ] be the metric on the space @xmath194 , \mathbb{r } ^{d})$ ] introduced , for the interval @xmath217 $ ] , as in display ( 5.2 ) in @xcite . then , by theorem 5.6 in @xcite , we see that @xmath429 , \mathbb{r } ^{d}),d_{{\mathcaligr}{d}}([0,t],\cdot,\cdot))$ ] is a complete metric space and the topology on @xmath194 , \mathbb{r } ^{d})$ ] , induced by the metric @xmath428,\cdot,\cdot)$ ] , is known as the skorohod topology on @xmath194 , \mathbb{r } ^{d})$ ] . recall that if @xmath430 , \mathbb{r } ^{d})$ ] , then @xmath428,x , y)=0 $ ] implies that @xmath431 for every @xmath132 .
furthermore , if @xmath432 is a sequence in @xmath433 , \mathbb{r } ^{d})$ ] and @xmath420 , \mathbb{r } ^{d})$ ] , then the statement that @xmath428,x^{n},x)\rightarrow0 $ ] as @xmath434 is equivalent to the statement that there exists @xmath435 ( see @xcite for the definition of the space @xmath436 ) such that ( 5.6 ) in @xcite holds and such that @xmath437 for a proof of this result we refer to proposition 5.3 in @xcite . furthermore , to understand the relatively compact sets in @xmath194 , \mathbb{r } ^{d})$ ] , we introduce and use a modulus of continuity .
in particular , for @xmath420 , \mathbb{r } ^{d})$ ] and @xmath54 we define the quantity @xmath438 where the infimum is taken with respect to all partitions of the form @xmath439 , with @xmath440 . furthermore , given @xmath441 , \mathbb{r } ^{d } )
$ ] , we let @xmath442 using this notation , we first quote theorem 6.3 in @xcite which states that @xmath178 , \mathbb{r } ^{d } ) $ ] is relatively compact in the skorohod topology if and only if for every rational @xmath75 $ ] there exists a relatively compact set @xmath443 such that @xmath444 for all @xmath180 and such that @xmath445 finally , we also note the following . given @xmath446 , let@xmath447 where the infimum is taken with respect to all partitions as above but with the additional restriction that @xmath448 .
furthermore , given @xmath449 , \mathbb{r } ^{d } ) $ ] , we let @xmath450 then @xmath451
in this section we first prove certain estimates for solutions to the skorohod problem for @xmath1 , assuming that @xmath2 satisfies the assumptions stated in theorem [ theorem 1 ] and that @xmath452 , \mathbb{r } ^{d } ) $ ] with @xmath63 .
in particular , we prove that the modulus of continuity of cdlg solutions to the skorohod problem for @xmath1 , with bounded jumps , can be estimated from above by the modulus of continuity of @xmath3 and the modulus of continuity @xmath453 .
this result is derived in two steps . in the first step
we prove ( see lemma compactlemma below ) a local compactness result which is valid in a spatial neighborhood of a given boundary point and on a constructed time interval . in the second step
we then prove that corresponding global estimates ( see theorem [ compacttheorem ] below ) can be derived based on the local compactness result .
in particular , theorem [ compacttheorem ] is the main result we establish in this context . in section [ spest ]
we derive these estimates for solutions to the skorohod problem and in section [ spappest ] we establish the corresponding results for approximations to the skorohod problem .
in particular , given @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 and a partition @xmath454 , which we denote by @xmath216 , of the interval @xmath217 $ ] , we define @xmath223 , @xmath229 , @xmath230 , @xmath455 and @xmath456 as in ( [ approxw ] ) , ( [ approxw+ ] ) , ( [ approxw++ ] ) and ( [ approxw+++ ] ) . then , by construction , the pair @xmath250 is a solution to the skorohod problem for @xmath457 . in lemma [ compactlemmaapprox ] and
theorem [ compacttheoremapprox ] we prove estimates for solutions to the skorohod problem for @xmath458 , which are similar to the ones established in lemma [ compactlemma ] and theorem [ compacttheorem ] for the skorohod problem for @xmath129 .
we note that the reason for this twofold approach is that since we are considering time - dependent domains , the condition in ( [ limitzero ] ) will in general not hold for @xmath229 .
given @xmath459 and @xmath460 , we define the positive functions @xmath461 and @xmath462 as follows : @xmath463 \\[-8pt ] k_{3 } ( a , e ) & = & \frac{1+k_{1 } ( a , e ) } { a},\qquad k_{4 } ( a , e ) = \frac{1+k_{2 } ( a , e ) } { a}.\nonumber\end{aligned}\ ] ] in this section we first prove the following two general results for solutions to the skorohod problem . [ compactlemma ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) , ( [ limitzero ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
let @xmath118 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath23 $ ] .
assume that ( [ crita ] ) and ( [ crite ] ) hold for some @xmath119 , @xmath106 , @xmath120 and @xmath121 .
finally , assume that @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73 , for some @xmath108 , @xmath109 as defined in ( [ deltanolldef ] ) and ( [ rhonolldef ] ) .
let @xmath464 , \mathbb{r } ^{d } ) $ ] with @xmath63 and let @xmath465 be a solution to the skorohod problem for @xmath124 .
consider a fixed but arbitrary @xmath100 $ ] , such that @xmath466 , and note that it follows from ( [ crita ] ) and ( [ crite ] ) that there exist @xmath467 and @xmath468 such that@xmath469 then , for @xmath470 , @xmath471 where @xmath472 , @xmath473 . here
@xmath474 is defined as follows .
if there exists some @xmath132 such that @xmath475 and @xmath476 , then @xmath477 whereas if @xmath478 for all @xmath475 , then @xmath479 [ compacttheorem ] let @xmath20 , @xmath13 , @xmath85 , @xmath131 , @xmath105 , @xmath106 , @xmath120 , @xmath121 , @xmath146 and @xmath480 be as in the statement of lemma [ compactlemma ] .
let @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 and let @xmath465 be a solution to the skorohod problem for @xmath124 .
moreover , assume in addition that @xmath130 , \mathbb{r } ^{d } ) $ ] .
then there exist positive constants @xmath481 , @xmath174 , @xmath175 and @xmath482 such that @xmath483 whenever @xmath53 . furthermore , if @xmath441 , \mathbb{r } ^{d } ) $ ] is relatively compact in the skorohod topology and satisfies @xmath63 , whenever @xmath180 , then there exist positive constants @xmath181 , @xmath182 ,
@xmath183 and @xmath484 , such that@xmath485 versions of lemma [ compactlemma ] and theorem [ compacttheorem ] , valid only in the setting of time - independent domains , are proved in lemma 2.1 , theorems 2.2 and 2.4 in @xcite .
our contribution is that we are able to establish similar results when @xmath13 is a time - dependent domain .
furthermore , concerning related results in the setting of time - dependent domains , we note that if @xmath2 is an @xmath486-domain and if @xmath368 , for some @xmath487-valued continuous function @xmath488 such that@xmath489}\langle\gamma _ { t}(z),n_{t}(z)\rangle>\frac{\sqrt{3}}{2},\ ] ] then a version of theorem [ compacttheorem ] is proved in theorem c.3 in @xcite .
note also that if @xmath2 is an @xmath486-domain , then there exists a unique unit inward normal , @xmath490 , at @xmath36 , @xmath75 $ ] .
[ conenotres ] unlike in the statements of theorems [ theorem 1 ] , theorem 2 and [ theorem 3 ] , we need not assume that @xmath73 satisfies ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) in the prerequisites of lemma [ compactlemma ] and theorem [ compacttheorem ] .
this remark also applies to lemma [ compactlemmaapprox ] and theorem compacttheoremapprox stated below .
proof of lemma [ compactlemma ] to simplify the notation , we in the following let @xmath491 , @xmath492 and @xmath493 . moreover , we let @xmath384 be a unit vector such that @xmath494 for all @xmath495 , @xmath496 and @xmath497 \subset [ s,(s+\eta)\wedge t ] $ ] .
the existence of such a vector follows from the definition of @xmath498 . using properties ( [ sp1])([sp2 ] ) in definition [ skorohodprob ]
, we see that @xmath499 for any @xmath470 .
based on ( [ hahaha ] ) , we deduce that @xmath500 furthermore , again using properties ( [ sp1])([sp2 ] ) in definition [ skorohodprob ] , we also see that @xmath501 \\[-8pt]\qquad & = & \vert w_{t_{2}}-w_{t_{1 } } \vert^{2}+ \biggl ( \int_{t_{1}^{+}}^{t_{2}^{+}}\gamma_{r}\,d \vert\lambda \vert _ { r } \biggr ) ^{2}+2\int_{t_{1}}^{t_{2}^{+ } } \langle w_{t_{2}}-w_{t_{1}},\gamma_{r } \rangle\ , d \vert\lambda
\vert _ { r } , \nonumber\end{aligned}\ ] ] whenever @xmath470 .
note that the integrand in the last term in this display can be rewritten as @xmath502 \\[-8pt ] \hspace*{34pt } & = & \langle w_{t_{2}}-w_{r},\gamma_{r } \rangle+ \langle x_{r}-x_{t_{1}},\gamma_{r } \rangle- \biggl\langle \biggl ( \int_{t_{1}^{+}}^{r^{+}}\gamma_{u}\,d \vert\lambda \vert _ { u } \biggr ) , \gamma_{r } \biggr\rangle .
\nonumber\hspace*{-24pt}\end{aligned}\ ] ] in particular , combining ( [ haha1 ] ) and ( [ haha2 ] ) , we see that @xmath503 we now intend to derive bounds from above for all integrals in ( [ lemma15 ] ) .
to do this , we first note that the first integral on the right - hand side of ( [ lemma15 ] ) is bounded from above by@xmath504 to find an upper bound of the second integral in ( [ lemma15 ] ) , we must take into account that , due to the fact that our domain is time - dependent , @xmath505 might not belong to @xmath506 .
recall that we are assuming that @xmath507 for all @xmath508 , @xmath509 $ ] , and , given @xmath510 $ ] , @xmath511 , in the following we denote a projection of @xmath114 onto @xmath512 along @xmath513 by @xmath514 .
furthermore , whenever @xmath515 we let @xmath516 . using this notation , and the definition of @xmath517 , we see that @xmath518 equation ( [ haha3 ] ) implies that @xmath519 .
next , writing@xmath520 and using the fact that @xmath521 a.e .
when @xmath522 , together with a version of ( [ cdef2 ] ) , we deduce that @xmath523 furthermore , @xmath524 using the estimates derived above , we conclude that the second integral in ( [ lemma15 ] ) has the upper bound @xmath525 \\[-8pt ] & & { } + 2 ( c+1 ) l ( t_{2}-t_{1 } ) ( \vert\lambda \vert_{t_{2}}- \vert\lambda \vert_{t_{1 } } ) .\nonumber\end{aligned}\ ] ] next , we use lemma 2.1(ii ) in @xcite and rewrite the third integral in ( [ lemma15 ] ) as @xmath526 \\[-8pt ] & & { } -\sum_{t_{1}<r\leq t_{2 } } \underbrace { \vert\gamma _ { r } \vert^{2}}_{=1 } ( \vert\lambda \vert_{r}- \vert \lambda \vert_{r^{- } } ) ^{2}.\nonumber\end{aligned}\ ] ] based on the last display , it is clear that and the third and fourth integral in ( [ lemma15 ] ) reduce to the term@xmath527 putting the relations ( [ lemma15])([lemma21 ] ) together , we obtain@xmath528 \\[-8pt]\qquad & & { } -\sum_{t_{1}<r\leq t_{2 } } ( \vert\lambda \vert _ { r}- \vert\lambda \vert_{r^{- } } ) ^{2}+2 ( c+1 ) l ( t_{2}-t_{1 } ) ( \vert\lambda \vert _ { t_{2}}- \vert\lambda \vert_{t_{1 } } ) .\nonumber\end{aligned}\ ] ] if we now combine ( [ lemma12 ] ) and the properties ( [ sp1])([sp2 ] ) in definition [ skorohodprob ] , we first get@xmath529 and then , again using lemma 2.1(ii ) in @xcite as well as the fact that @xmath530 , we conclude that @xmath531 \\[-8pt]\quad & & { } + c ( \vert\lambda \vert_{t_{2}}- \vert\lambda \vert_{t_{1 } } ) ^{2}+2 ( c+1 ) l ( t_{2}-t_{1 } ) ( \vert\lambda \vert_{t_{2}}- \vert \lambda \vert_{t_{1 } } ) .\nonumber\end{aligned}\ ] ] relation ( [ lemma11 ] ) and the inequality in the last display yield@xmath532 in addition , combining ( [ lemma11 ] ) and ( [ lemma12 ] ) , we obtain @xmath533 \\[-8pt]\quad & & { } + \frac{2 ( c+1 ) } { a}l ( t_{2}-t_{1 } ) \vert x \vert_{t_{1},t_{2}}+\frac{2 ( c+1 ) } { a}l ( t_{2}-t_{1 } ) \vert w \vert_{t_{1},t_{2}}.\nonumber\end{aligned}\ ] ] the inequalities ( [ lemma14 ] ) and ( [ lemma13 ] ) can both be written on the form@xmath534 where the positive constants @xmath535 , @xmath536 and @xmath42 are easily shown to satisfy the condition @xmath537 in both cases .
we claim that @xmath538 indeed , suppose , on the contrary , that @xmath539 then , by ( [ haha12 ] ) , @xmath540 \\[-8pt ] & & \qquad\quad { } -d \vert x \vert_{t_{1},t_{2}}+d \biggl ( -\frac{a}{c } \vert x \vert_{t_{1},t_{2}}+\frac{d}{c } \biggr ) \nonumber\\ & & \qquad = a \vert x \vert_{t_{1},t_{2}}^{2 } \underbrace { \biggl ( 1-\frac{b}{c}-\frac{a}{c } \biggr ) } _ { = \frac{c - b - a}{c}=0}{}+ d \vert x \vert_{t_{1},t_{2}}\underbrace { \biggl ( \frac{b}{c}+\frac{2a}{c}-\frac{a}{c}-1 \biggr ) } _ { = \frac{a+b - c}{c}=0 } = 0.\nonumber\end{aligned}\ ] ] obviously ( [ haha13 ] ) contradicts ( [ lemma16 ] ) and , hence , the claim in ( [ haha11 ] ) is proved . to complete the proof of lemma [ compactlemma ]
, we first note that ( [ lemma14 ] ) implies @xmath541 and that ( [ lemma13 ] ) implies@xmath542 \\[-8pt]\qquad & = & \frac{a+2}{a-2c } \vert w \vert_{t_{1},t_{2}}+\frac{2 ( c+1 ) } { a-2c}l ( t_{2}-t_{1 } ) .\nonumber\end{aligned}\ ] ] from the definition @xmath543 we know that if @xmath544 , then we can set @xmath545 in ( [ lemma18 ] ) and obtain@xmath546 \\[-8pt ] & = & \frac{a+2a^{2}e+2+ae}{a ( 1-e ) } \vert w \vert _ { t_{1},t_{2}}+\frac{2a^{2}e+2}{a ( 1-e ) } l ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] whereas if @xmath547 , then we can set @xmath548 in ( lemma17 ) and obtain@xmath549 hence , in either case , we arrive at @xmath550 and the proof of estimate ( [ lemma1r1 ] ) is complete . finally , we note that estimate ( [ lemma1r2 ] ) now follows directly from ( [ lemma1r1 ] ) and ( [ lemma11 ] ) .
this completes the proof of lemma [ compactlemma ] .
proof of theorem [ compacttheorem ] we first note that the assumptions stated in theorem [ compacttheorem ] ensure that there exist some @xmath459 and @xmath551 such that @xmath552 and @xmath553 for all @xmath466 , @xmath100 $ ] .
next we recursively define two sets of time - points @xmath554 and @xmath555 .
in particular , we let @xmath556 and define , for @xmath557 , @xmath558 if @xmath559 for all @xmath75 $ ] and @xmath560 otherwise .
similarly , for @xmath561 , we let @xmath562 , if @xmath563 for all @xmath132 such that @xmath564 , and @xmath565 otherwise . using ( [ limitzero ] ) and the fact that @xmath4 is a right continuous function , it follows that @xmath566 for all @xmath567 .
moreover , using ( [ tdef])([taudef ] ) , we can apply lemma compactlemma to any pair of time points @xmath568 such that @xmath569 and obtain@xmath570 \\[-8pt ] k_{3 } ( a , e ) \vert w \vert_{t_{1},t_{2}}+k_{4 } ( a ,
e ) l ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] whenever @xmath569 where @xmath571 , @xmath572 , @xmath573 and @xmath462 are defined as in lemma [ compactlemma ] based on @xmath120 and @xmath121 introduced above .
next , we want to find a similar estimate whenever @xmath574 .
if @xmath575 , we are done and , hence , we assume that @xmath576 . in that case @xmath577 for all @xmath578 and , as a consequence ,
the changes in @xmath4 and @xmath3 coincide on this time interval . finally ,
considering the case @xmath579 , we have@xmath580 \\[-8pt ] { t}_{i}^{-}}-|\lambda @xmath581 and @xmath582 in ( [ hahaha2 ] ) can be handled using ( [ hahaha1 ] ) .
regarding the terms @xmath583 and @xmath584 in ( [ hahaha2 ] ) , we can , since @xmath585 , use ( [ cdef2 ] ) and the definition of the skorohod problem to first conclude that@xmath586 \\[-8pt ] & \geq & \vert x_{\hat{t}_{i}}-x_{\hat{t}_{i}^{- } } \vert^{2}+ ( \vert\lambda \vert_{\hat{t}_{i}}- \vert\lambda \vert _ { \hat{t}_{i}^{- } } ) ^{2 } \nonumber\\ & & { } -2c_{x_{\hat{t}_{i}},\hat{t}_{i } } ( \rho_{0},\eta_{0 } ) \vert x_{\hat{t}_{i}}-x_{\hat{t}_{i}^{- } } \vert ( \vert \lambda \vert_{\hat{t}_{i}}- \vert\lambda \vert_{\hat{t}_{i}^{- } } ) .\nonumber\end{aligned}\ ] ] then @xmath587 \\[-8pt ] & & { } + \bigl ( 1-c_{x_{\hat{t}_{i}},\hat{t}_{i } } ( \rho_{0},\eta_{0 } ) \bigr ) ( \vert\lambda \vert_{\hat{t}_{i}}- \vert \lambda \vert_{\hat{t}_{i}^{- } } ) ^{2},\nonumber\end{aligned}\ ] ] and , as @xmath588 , for all @xmath589 , @xmath100 $ ] , we obtain@xmath590 \\[-8pt ] & = & e_{x_{\hat{t}_{i}},\hat{t}_{i } } ( \rho_{0},\eta_{0 } ) \leq e.\nonumber\end{aligned}\ ] ] combining the estimates in ( [ hahaha4 ] ) and ( [ hahaha5 ] ) , we arrive at @xmath591 \\[-8pt]\qquad \vert\lambda \vert_{\hat{t}_{i}}- \vert\lambda \vert _
{ \hat{t}_{i}^{- } } & \leq&\frac{1}{\sqrt{1-e } } \vert w_{\hat { t}_{i}}-w_{\hat{t}_{i}^{- } } \vert.\nonumber\end{aligned}\ ] ] introducing the notation @xmath592 \\[-8pt ] k_{3 } & = & k_{3 } ( a , e ) + \frac{1}{\sqrt{1-e}},\qquad k_{4}=k_{4 } ( a , e ) , \nonumber\end{aligned}\ ] ] we can use the deductions in ( [ hahaha1])([theorem14 ] ) to conclude that @xmath593 \\[-8pt ] _ { t_{1},t_{2}}+k_{4}l ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] whenever @xmath594 .
we now intend to make use of the estimates in ( [ hahaha1again ] ) to complete the proof of theorem compacttheorem .
note that above we have constructed a set of time - points @xmath595 , where @xmath596 is so far undetermined , and @xmath597 if @xmath598 , let @xmath599 then , using ( [ hahaha1again ] ) , we have @xmath600 \\[-8pt]\qquad k_{3 } \vert w \vert_{t_{1},t_{2}}+k_{4}l ( t_{2}-t_{1 } ) \bigr ) + \sum_{i = h}^{v } \vert\lambda \vert_{t_{i}}- \vert \lambda \vert_{t_{i}^{-}}.\nonumber\end{aligned}\ ] ] moreover , arguing exactly as in the deduction of ( [ theorem14 ] ) , we obtain @xmath601 \\[-8pt ] \vert\lambda \vert_{t_{i}}- \vert\lambda \vert _ { t_{i}^{- } } & \leq&\frac{1}{\sqrt{1-e } } \vert w_{t_{i}}-w_{t_{i}^{- } } \vert,\nonumber\end{aligned}\ ] ] whenever @xmath602 .
hence , to complete the proof of theorem compacttheorem , we have to estimate @xmath596 .
to do this , we consider @xmath178 , \mathbb{r } ^{d } ) $ ] , which is assumed to be relatively compact in the skorohod topology and for which @xmath63 whenever @xmath603 .
we shall prove that the @xmath596 introduced above is bounded for every such set @xmath604 . to do this
, we use the notation introduced in section skorohodtop concerning the skorohod topology . in the following
let @xmath605 be a fixed number such that @xmath606 note that the existence of @xmath607 follows immediately from ( [ limitzero ] ) . using the definition of @xmath608 and the fact that @xmath178 , \mathbb{r } ^{d } ) $ ] is relatively compact in the skorohod topology
, we see , by ( [ e11 ] ) and ( [ d11 + ] ) , that @xmath609 in particular , using ( [ gaga ] ) , we can find a @xmath610 such that for every @xmath180 there exists a partition @xmath611 , in general depending on @xmath3 , such that @xmath612 and @xmath613 we claim that none of the intervals @xmath614 in this partition can contain more than one point from the sequence @xmath615 . to prove this , we suppose , on the contrary , that there exist @xmath616 and @xmath617 such that @xmath618 .
then , by construction , @xmath619 we intend to estimate @xmath620 .
we first note that if @xmath621 for all @xmath132 such that @xmath622 , then @xmath623 .
however , using ( [ ahah1 ] ) and ( [ h11- ] ) , it is clear that neither @xmath624 nor @xmath625 can occur .
hence , we can assume that @xmath626 is given by ( [ taudef ] ) and , as a consequence , that @xmath627 but on the other hand , using ( [ hahaha1again ] ) , we first see that @xmath628 \\[-8pt]&\leq & k_{1 } \vert w \vert_{{t_{i}},{\hat{t}_{i}}}+ ( k_{2}+1 ) l(\hat{t}_{i}-t_{i}),\nonumber\end{aligned}\ ] ] and then , using ( [ ahah1 ] ) , ( [ h11- ] ) and ( [ h11 ] ) , we deduce @xmath629 which contradicts the assumption @xmath618 . hence , none of the intervals @xmath614 in the partition can contain more than one point from the sequence @xmath615 and , in particular , we conclude that @xmath630 combining ( [ theorem12 ] ) , ( [ theorem14 + ] ) and ( [ theorem15 ] ) , we see that@xmath631 \\[-8pt ] & \leq & \biggl ( k_{1 } \biggl ( \frac{t}{\delta}+2 \biggr ) + \frac{1}{\sqrt{1-e}}\biggl ( \frac{t}{\delta}+1 \biggr ) \biggr ) \vert w \vert _ { t_{1},t_{2 } } \nonumber\\ & & { } + k_{2 } \biggl ( \frac{t}{\delta}+2 \biggr )
l ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] and , similarly , that @xmath632 \\[-8pt ] & & { } + k_{4 } \biggl ( \frac{t}{\delta}+2 \biggr ) l ( t_{2}-t_{1 } ) .\nonumber\end{aligned}\ ] ] the deductions in the last two displays complete the proof of theorem compacttheorem .
let @xmath20 , @xmath13 and @xmath131 satisfy the assumptions stated in theorem [ compacttheorem ] . in this section
we derive estimates for approximations to the skorohod problem for @xmath1 .
in particular , given @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 and a partition @xmath454 of the interval @xmath217 $ ] , which we denote by @xmath633 , we define @xmath223 as in ( [ approxw ] ) . recall that @xmath634 was defined in ( [ korlim+ ] ) .
furthermore , in the following , we will assume that ( [ korlim ] ) holds whenever @xmath225 .
based on the assumption in ( [ korlim ] ) , we define @xmath229 , @xmath635 , @xmath455 and @xmath456 as in ( [ approxw+ ] ) , ( [ approxw++ ] ) and ( [ approxw+++ ] ) . then , by construction , the pair @xmath250 is a solution to the skorohod problem for @xmath457 . in this section
we prove the following results . [ compactlemmaapprox ]
let @xmath20 , @xmath13 , @xmath85 , @xmath131 , @xmath105 , @xmath106 , @xmath120 , @xmath121 , @xmath146 and @xmath480 be as in the statement of theorem [ compacttheorem ] . given @xmath459 and @xmath460 ,
let the functions @xmath571 , @xmath572 , @xmath573 and @xmath462 be defined as in ( [ functions ] ) and let @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 .
let @xmath636 be a partition of the interval @xmath217 $ ] , let @xmath223 be defined as in ( [ approxw ] ) and assume that ( [ korlim ] ) holds . given @xmath216 and @xmath223 ,
let @xmath229 , @xmath230 , @xmath455 and @xmath456 be defined as in ( [ approxw+ ] ) , ( [ approxw++ ] ) and ( [ approxw+++ ] ) . consider a fixed but arbitrary @xmath637 $ ] , such that @xmath638 .
then , for @xmath639 , @xmath640 satisfies the estimates @xmath641 here @xmath642 is defined as follows . if there exists some @xmath132 such that @xmath643 and @xmath644 , then @xmath645 whereas if @xmath646 for all @xmath647 , then @xmath648 [ compacttheoremapprox ] let @xmath20 , @xmath13 , @xmath85 , @xmath131 , @xmath105 , @xmath106 , @xmath120 , @xmath121 , @xmath146 and @xmath480 be as in the statement of theorem [ compacttheorem ] . given @xmath459 and @xmath460 ,
let the functions @xmath571 , @xmath572 , @xmath573 and @xmath462 be defined as in ( [ functions ] ) and let @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 .
let @xmath636 be a partition of the interval @xmath217 $ ] , let @xmath223 be defined as in ( [ approxw ] ) and assume that ( [ korlim ] ) holds .
let @xmath216 be such that @xmath649 and let @xmath650 \\[-8pt ] l ( \hat{\delta}{}^{\prime } ) + l(\delta^{\ast})&\leq&\rho_{0}/\bigl(2\bigl(k_{2}(a , e)+1\bigr)\bigr).\nonumber\end{aligned}\ ] ] given @xmath216 and @xmath223 , let @xmath229 , @xmath651 , @xmath455 and @xmath456 be defined as in ( [ approxw+ ] ) , ( [ approxw++ ] ) and ( [ approxw+++ ] ) .
moreover , assume that @xmath652 , \mathbb{r } ^{d } ) $ ] .
then there exist positive constants @xmath653 , @xmath254 , @xmath654 and @xmath256 , independent of @xmath216 , such that @xmath655 \\[-8pt]\quad { l}_{3 } ( w , t ) \vert w \vert_{t_{1},t_{2}}+\hat{l}_{4 } ( w , t ) \bigl(l ( t_{2}-t_{1 } ) + l(\delta^{\ast})\bigr),\nonumber\end{aligned}\ ] ] whenever @xmath53 .
proof of lemma [ compactlemmaapprox ] naturally , the proof of this lemma is similar to the proof of lemma [ compactlemma ] and , thus , we only describe the main differences compared to the proof of lemma [ compactlemma ] .
first we note , by the assumptions on @xmath2 and the construction of @xmath229 based on @xmath2 , that there exists a unit vector @xmath384 such that @xmath656 for all @xmath657 , @xmath658 and @xmath659\subset [ s,(s+\eta_{0})\wedge t ] $ ] .
we also note that if @xmath660 and @xmath661 , for some @xmath662 , then @xmath663 if @xmath664 and otherwise @xmath665 .
now , using the fact that @xmath666 solves the skorohod problem for @xmath667 , we conclude , in analogy with ( [ hahaha ] ) , that @xmath668 for any @xmath639 , where @xmath657 for some @xmath669 . based on ( [ haha+ ] ) , we obtain @xmath670 furthermore , arguing as in the proof of lemma [ compactlemma ] , we derive @xmath671 as in the proof of lemma [ compactlemma ] , we have to find upper bounds of all integrals in ( [ lemma15 + ] ) and , naturally , particular attention has to be paid to the second integral , as @xmath672 might not belong to @xmath673 . for @xmath674 ,
@xmath675 $ ] , we let @xmath676 denote the set of inward normals at @xmath674 and given @xmath510 $ ] , @xmath677 in the following we denote a projection of @xmath114 onto @xmath678 along @xmath679 by @xmath680 . furthermore ,
if @xmath681 , then we let @xmath682 . using this notation , and the definition of @xmath683 , we see that @xmath684 equation ( [ haha3 + ] ) implies that @xmath685 . arguing as in ( [ haha4])([haha7 ] ) , we then deduce that @xmath686 \\[-8pt ] & \leq&c \vert x_{r}^{\delta}-x_{t_{1}}^{\delta } \vert+cl ( r - t_{1 } ) + cl(\delta^{\ast}),\nonumber\end{aligned}\ ] ] and that @xmath687 using the estimates derived above , we conclude that the second integral in ( [ lemma15 + ] ) has the upper bound @xmath688 equipped with ( [ lemma20 + ] ) , the proof of lemma [ compactlemmaapprox ] can now be completed following the lines of the proof of lemma compactlemma .
proof of theorem [ compacttheoremapprox ] proceeding as in the proof of theorem [ compacttheorem ] , we first recursively define two sets of time - points @xmath689 and @xmath690 in order to use lemma [ compactlemmaapprox ] . in particular , we let @xmath691 and define , for @xmath561 , @xmath692 if @xmath693 for all @xmath675 $ ] and @xmath694 otherwise . similarly , for @xmath561 we let @xmath695 , if @xmath696 for all @xmath132 such that @xmath697 . moreover ,
if the latter is not the case , we then define @xmath698 to equal @xmath699 we can then repeat the argument in ( [ hahaha1])([hahaha1again ] ) to conclude that @xmath700 \\[-8pt ] & k_{3 } \vert w \vert_{t_{1},t_{2}}+k_{4}\bigl(l ( t_{2}-t_{1 } ) + l(\delta^{\ast})\bigr),\nonumber\end{aligned}\ ] ] whenever @xmath701 .
furthermore , we note that the @xmath596 in @xmath595 is so far undetermined and , as in ( [ theorem12])([theorem14 + ] ) , we derive @xmath702 \\[-8pt ] m+1 ) \bigl ( k_{3 } \vert w \vert_{t_{1},t_{2}}+k_{4}\bigl(l ( t_{2}-t_{1 } ) + l(\delta^{\ast})\bigr ) \bigr ) \nonumber\\ & & { } + \sqrt{\frac{1}{1-e}}\sum_{i=1}^{m } \vert w_{t_{i}^{\delta } } -w_{t_{i}^{\delta,- } } \vert,\nonumber\end{aligned}\ ] ] whenever @xmath53 . to complete the proof of theorem compacttheoremapprox
, we can now proceed as in the proof of theorem compacttheorem and conclude that @xmath703 , where @xmath56 is given in the proof of theorem [ compacttheorem ] .
this completes the proof of theorem [ compacttheoremapprox ] .
in the first subsection of this section we prove the general convergence result for sequences of skorohod problems ( see theorem [ spconvergence ] stated below ) referred to in section [ contri ] .
then , in the second subsection we explicitly construct , given @xmath1 , an approximating sequence @xmath209 and , for each @xmath191 , an explicit solution @xmath210 to the skorohod problem for @xmath704 .
we then prove that the constructed sequence @xmath213 of solutions converges to a solution to the skorohod problem for @xmath1 .
let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) .
let @xmath153 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath75 $ ] and assume that @xmath73 satisfies ( [ cone1 ] ) and ( [ cone2 ] ) .
let @xmath186 be a sequence of time - dependent domains @xmath187 and let @xmath705 be a sequence of closed convex cones of vectors in @xmath7 .
let @xmath62 , \mathbb{r } ^{d } ) $ ] with @xmath63 and let @xmath706 with @xmath193 be a sequence of cdlg functions converging to @xmath3 in the skorohod topology .
assume that there exists a solution @xmath707 to the skorohod problem for @xmath708 .
then in theorem [ spconvergence ] we prove , by making appropriate assumptions on @xmath709 , @xmath73 , @xmath189 and @xmath190 , that if @xmath710 and @xmath711 in the sense defined in theorem [ spconvergence ] , then @xmath707 converges to @xmath465 and @xmath0 is a solution to the skorohod problem for @xmath712 .
however , to state theorem [ spconvergence ] , we need to introduce some additional notions and notation
. in particular , in the following we let @xmath713 and @xmath714 be defined as in ( [ adef ] ) and ( [ edef ] ) but with respect to @xmath715 .
we assume that @xmath716 , for @xmath198 , satisfies the uniform exterior sphere condition in time with radius @xmath85 , independent of @xmath191 , and that there exist @xmath105 and @xmath106 such that , for all @xmath198 , @xmath717 } \inf_{z\in\partial d_{s}^{n}}a_{s , z}^{n } ( \rho_{0},\eta_{0 } ) & = & a_{n}>0 , \\ \label{crite+ } \sup_{s\in [ 0,t ] } \sup_{z\in\partial d_{s}^{n}}e_{s , z}^{n } ( \rho_{0},\eta_{0 } ) & = & e_{n}<1.\end{aligned}\ ] ] furthermore , we let @xmath718 } } _ { \vert s - t \vert\leq r}\sup_{z\in\overline{d_{s}^{n}}}d ( z , d_{t}^{n } ) , \label{defl+}\ ] ] and we assume that @xmath719 note also that if we define @xmath720 as in ( [ lhat ] ) but with @xmath2 replaced by @xmath716 , then lemma [ llhatequivalence ] and ( [ limitzero+ ] ) imply that @xmath721 moreover , we assume that there exists @xmath722 such that @xmath723 and @xmath724 , for all @xmath198 and @xmath75 $ ] , and we let @xmath725 } \sup_{n}\max\{\operatorname { diam}(d_{t}^{n}),\operatorname{diam } ( d_{t } ) \ } , \label{defr}\ ] ] where diam@xmath726 , diam@xmath727 are the euclidean diameters of the spatial regions @xmath728 and @xmath28 , respectively . recalling that the set @xmath367 was introduced in ( [ cone2 ] ) ,
we here also introduce , for @xmath198 , @xmath729 \ } , \label{cone2+}\ ] ] and we let , whenever @xmath75 $ ] , @xmath730 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) \bigr ) , \nonumber \\[-8pt ] \\[-8pt ] g_{t}^{n } & = & g^{\gamma^{n}}\cap \bigl ( [ 0,t ] \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) \bigr ) .\nonumber\end{aligned}\ ] ] in the following we need to measure the distance between the sets @xmath731 and @xmath732 and , hence , we introduce an appropriate hausdorff distance for subsets of @xmath217\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] . in particular , we let , given @xmath733\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] and @xmath734\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] , @xmath735 denote the ( euclidean ) distance between @xmath736 and @xmath737 .
furthermore , based on @xmath17 , we define , given @xmath738\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] and @xmath733\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] , the distances @xmath739 , @xmath740 and@xmath741 in the natural way . furthermore , for @xmath742 and @xmath743 as above
, we introduce a hausdorff distance between @xmath742 and @xmath743 as @xmath744 in the following we say that @xmath732 converges to @xmath731 if @xmath745 imposing the assumptions on @xmath2 , @xmath73 stated above and assuming ( [ gdefhaus+ ] ) , we can , for example , ensure that if @xmath746 is a sequence of points in @xmath34 , @xmath94 $ ] , @xmath747 , @xmath96 $ ] , @xmath97 , then @xmath748 to see this , we consider , for @xmath93 and @xmath390 given as above , @xmath749 and @xmath750 . given @xmath751 ,
we let @xmath752 be a point on @xmath731 which minimizes the distance , as defined in ( [ gdefhaus- ] ) , from @xmath753 to @xmath731 . then , @xmath754
hence , @xmath755 where @xmath756 as , by assumption , @xmath731 is closed , we can now first conclude that @xmath757 as @xmath434 , and then we find , using ( [ gdefhaus+ ] ) , that @xmath758 as @xmath434 .
this completes the proof of ( [ gamman ] ) .
we are now ready to formulate our convergence result .
[ spconvergence ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) .
let @xmath153 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath23 $ ] , and assume that @xmath73 satisfies ( [ cone1 ] ) and ( [ cone2 ] ) .
let @xmath189 be a sequence of time - dependent domains @xmath187 satisfying ( [ timedep+ ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
let @xmath759 be a sequence of closed convex cones @xmath760 of vectors in @xmath7 for every @xmath761 , @xmath75 $ ] .
for all @xmath198 , @xmath716 and @xmath762 satisfy ( [ crita+ ] ) and ( [ crite+ ] ) for some @xmath105 , @xmath106 , @xmath763 , @xmath764 and , moreover , @xmath765 has the @xmath170-property of good projections along @xmath762 , for some @xmath766 , @xmath109 .
assume that @xmath767 , @xmath768 and ( [ limitzero+ ] ) hold .
regarding the convergence @xmath769 and @xmath770 , assume that @xmath771 } h ( { d_{t}^{n}},{d_{t } } ) & = & 0 , \\ \label{ddn } \lim_{n\rightarrow\infty}\sup_{t\in [ 0,t ] } h ( \partial{d_{t}^{n}},\partial{d_{t } } ) & = & 0,\end{aligned}\ ] ] and , with @xmath731 and @xmath732 defined as in ( [ gtdef ] ) , that@xmath772 let @xmath773 , \mathbb{r } ^{d } ) $ ] with @xmath193 and assume that there exists a solution @xmath707 to the skorohod problem for @xmath708 such that @xmath774 , \mathbb{r } ^{d } ) $ ] for all @xmath198 .
assume that @xmath192 is relatively compact in the skorohod topology and that @xmath192 converges to @xmath62 , \mathbb{r } ^{d } ) $ ] .
then @xmath775 converges to @xmath776 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] and @xmath0 is a solution to the skorohod problem for @xmath712 with @xmath777 , \mathbb{r } ^{d } ) $ ] .
[ subtel]we note that the formulation of theorem [ spconvergence ] contains several subtle points .
first , we do not have to assume that the elements in the sequence @xmath197 satisfy ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) .
the reason for this ( see remark [ conenotres ] ) is that theorem [ compacttheorem ] holds , with constants independent of @xmath191 , for each element in the sequence @xmath778 even without these assumptions .
second , we only have to impose very modest restrictions on @xmath2 but , as can be seen in the proof below , we have to assume that @xmath153 satisfies ( [ cone1 ] ) and ( [ cone2 ] ) .
proof of theorem [ spconvergence ] as @xmath706 is relatively compact in the skorohod topology , we first note that theorem [ compacttheorem ] can be used to conclude the existence of positive constants @xmath779 , @xmath780 , @xmath781 and @xmath782 , independent of @xmath191 , such that@xmath783 \\[-8pt ] \vert\lambda^{n } \vert_{t_{2}}- \vert\lambda ^{n } \vert_{t_{1 } } & \leq&l_{3 } \vert w^{n } \vert _ { t_{1},t_{2}}+l_{4}l_{n } ( t_{2}-t_{1 } ) , \nonumber\end{aligned}\ ] ] whenever @xmath53 . as @xmath192 converges to @xmath62 , \mathbb{r } ^{d } ) $ ] , we also see , using ( [ limitzero+ ] ) and ( [ gg1 ] ) , that @xmath784 is relatively compact in @xmath785 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } _ { + } ) $ ] .
furthermore , we know that @xmath786 for all @xmath75 $ ] , @xmath198 . hence , @xmath787 converges to some @xmath776 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] .
we intend to prove that @xmath788 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] solves the skorohod problem for @xmath129 and to do this , we have to prove that @xmath789 , \mathbb{r } ^{d } ) , \label{bv}\ ] ] and we have to verify that @xmath790 we begin by verifying ( [ sp1 ] ) in definition [ skorohodprob ] . to do this , we first note , using the convergence properties of the skorohod topology and the fact that @xmath707 solves the skorohod problem for @xmath708 , that @xmath791 for all points of continuity and , hence , since @xmath3 , @xmath4 and @xmath792 are cdlg functions , that ( [ hra1 ] ) holds for all @xmath793 $ ] .
hence , to verify ( [ sp1 ] ) in definition [ skorohodprob ] , it only remains to ensure that @xmath794 for all @xmath795 $ ] . to do this , we first note , using proposition 5.3 and remark 5.4 in chapter 3 of @xcite , that there exists a sequence @xmath796 such that@xmath797 \\[-8pt ] \lambda_{\tilde{t}_{n}}^{n}&\rightarrow&\lambda_{t},\qquad \lambda_{\tilde{t}_{n}^{-}}^{n}\rightarrow\lambda_{t^{-}}\qquad\mbox{as } n\rightarrow \infty.\nonumber\end{aligned}\ ] ] furthermore , using the triangle inequality , ( [ limitzero ] ) , ( [ dn ] ) and ( [ convth3 ] ) , we obtain @xmath798 \\[-8pt]\qquad & \leq & \vert x_{t}-x_{\tilde{t}_{n}}^{n } \vert+h ( \overline{d_{\tilde{t}_{n}}^{n}},\overline{d_{\tilde{t}_{n } } } ) + l(|\tilde{t}_{n}-t|)\rightarrow0 , \qquad\mbox{as } n\rightarrow\infty.\nonumber\end{aligned}\ ] ] this proves that @xmath794 for all @xmath23 $ ] and , hence , we have verified that @xmath799 satisfies ( [ sp1 ] ) .
we next prove ( [ bv ] ) , that is , that @xmath800 , \mathbb{r } ^{d } ) $ ] .
to do this , we use an argument similar to the proof of theorem 3.1 in @xcite , but , as described below , our argument is more subtle due to the fact that we consider sequences @xmath211 where , in particular , @xmath716 is time - dependent . recall that with @xmath801 as introduced in ( [ defr ] ) , we have @xmath802 } \vert x_{t}^{n } \vert < r , \qquad\sup_{t\in [ 0,t ] } \vert x_{t } \vert < r . \label{defr+}\ ] ] let @xmath803 , @xmath804 and @xmath805 be as in ( [ cone2 + ] ) and ( [ gtdef ] ) . by the prerequisites of theorem [ spconvergence ] ( see remark [ subtel ] ) , we have that @xmath367 is closed . we next define a positive measure @xmath806 on @xmath66 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] by setting , for every borel set @xmath807 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] , @xmath808 where @xmath809 is as in ( [ sp2])([sp4 ] ) for the solution @xmath707 to the skorohod problem for @xmath708 and @xmath810 is the characteristic functions for the set @xmath811 .
we then first note that@xmath812 . \label{4.22}\ ] ] we also note that the support of @xmath806 is contained in @xmath732 in the sense that @xmath813 whenever @xmath807 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] is such that @xmath814 .
using this , and the fact that ( [ sp2 ] ) holds for @xmath815 , we see that ( [ 4.22 ] ) implies that @xmath816 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) } \gamma\,d\mu^{n } ( s , z,\gamma ) \qquad\mbox{whenever } t\in [ 0,t ] .
\label{lambdamurel}\ ] ] next , using ( [ limitzero+ ] ) , ( [ gg1 ] ) , ( [ 4.22 ] ) and the fact that @xmath817 converges to @xmath818 , \mathbb{r } ^{d } ) $ ] , we conclude that @xmath819 which implies that @xmath820 is a compact set of measures , on @xmath66 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] , in the sense of the weak@xmath821-topology .
therefore , by the banach alaoglu theorem , we can conclude that @xmath820 converges in the weak@xmath821-topology to a measure @xmath822 such that @xmath823 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) \bigr)<\infty .
\label{boundmeas}\ ] ] moreover , since @xmath707 converges to @xmath171 in the sense of the skorohod topology , we obtain , using ( [ lambdamurel ] ) , that @xmath824 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) } \gamma\,d\mu ( s , z,\gamma ) \label{lambdamurel2}\ ] ] for all @xmath23 $ ] such that @xmath825 .
however , as both sides of ( [ lambdamurel2 ] ) are right continuous , ( lambdamurel2 ) holds for all @xmath23 $ ] . having proved ( [ lambdamurel2 ] )
, we see , also using ( [ boundmeas ] ) , that @xmath792 is of bounded variation and , hence , ( [ bv ] ) is proved .
we next claim that @xmath826 that is , we claim that the support of the measure @xmath822 is the set @xmath731 in the sense that if @xmath807 \times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] is such that @xmath827 , then @xmath828 . to see this ,
we let @xmath829\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) \setminus g_{t}$ ] and we see , as @xmath731 is closed , that if we define , for @xmath468 , @xmath830\times{b_{r } ( 0 ) } \times s_{1 } ( 0 ) $ ] , then there exists @xmath106 such that @xmath831 . recall that @xmath17 is the distance function introduced in ( [ gdefhaus- ] ) .
furthermore , the above setup implies that @xmath832 and since @xmath833 we can use the assumption in ( [ gamman+ ] ) to conclude that there exists @xmath834 such that @xmath835 for all @xmath263 .
in particular , @xmath836 for all @xmath263 .
hence , @xmath837 , for all @xmath838 , and @xmath839 by the weak@xmath821-convergence of @xmath806 to @xmath822 .
this completes the proof of ( [ lambdamurel2 + ] ) . having proved ( [ bv ] ) and ( [ lambdamurel2 + ] )
, we see that @xmath840 \label{laba2}\ ] ] for some @xmath487-valued borel measurable function @xmath841 and , to prove ( [ sp2 ] ) , we have to prove that @xmath842 for all @xmath843 $ ] . to prove this and to verify ( [ sp4 ] ) , we consider the following two cases : @xmath844 \mbox { is such that } x_{t}-x_{t^{-}}&\neq&0 , \nonumber \\[-8pt ] \\[-8pt ] \mbox{case 2.}\quad t\in [ 0,t ] \mbox { is such that } x_{t}-x_{t^{-}}&=&0.\nonumber\end{aligned}\ ] ] _ case _ 1 .
note that case 1 occurs for an at most countable set of jump times of @xmath4 .
moreover , in case 1 it is enough to prove that @xmath845 we first note , as we are assuming @xmath846 , that @xmath847 for @xmath191 sufficiently large ; see ( [ convth3 ] ) .
furthermore , since @xmath707 solves the skorohod problem for @xmath708 , we have that@xmath848 combining ( [ limitzero+aa ] ) , ( [ ddn ] ) , ( [ convth3 ] ) and ( convth2 ) , we obtain@xmath849 \\[-8pt]\qquad & \leq&|x_{t}-x_{\tilde{t}_{n}}^{n}|+h(\partial d_{\tilde{t}_{n}}^{n},\partial d_{\tilde{t}_{n}})+\hat{l}(|\tilde { t}_{n}-t|)\rightarrow0 \qquad\mbox{as } n\rightarrow\infty.\nonumber\end{aligned}\ ] ] hence , using ( [ bbb1 ] ) , we can , since @xmath27 is closed , conclude that @xmath850 we next recall that the set @xmath367 , defined in ( [ cone2 ] ) , is , by assumption , closed .
furthermore , arguing as in ( [ bbb1 ] ) , we first see that @xmath851 and then , letting @xmath434 , it follows , using ( [ gamman ] ) , that @xmath852 . applying the fact that the set @xmath367 is closed , we can therefore conclude that @xmath853 this concludes the proof of ( [ hra2 ] ) and , hence , we have verified ( sp2)([sp4 ] ) in case 1 . _ case _ 2 . to verify ( [ sp2])([sp4 ] ) in case 2 , we first see , by combining ( [ lambdamurel2 + ] ) and ( [ laba2 ] ) , that@xmath854 .
\label{lambdamurel3}\ ] ] we next introduce a measure @xmath855 on @xmath217 $ ] by setting @xmath856)=\mu ( g_{t } ) \qquad\mbox{whenever } t\in [ 0,t ] .
\label{nudef}\ ] ] combining ( [ lambdamurel3 ] ) and ( [ nudef ] ) , it is clear that @xmath857)=0 $ ] implies @xmath858 , showing that @xmath859 is absolutely continuous with respect to @xmath855 . to simplify the notation , in the following
we let , for @xmath860 , @xmath861 } \bigl ( \vert t - s \vert+ ( \vert z - x_{s } \vert\wedge \vert z - x_{s^{- } } \vert ) \bigr ) > \frac{1}{k } \biggr\ } .\ ] ] then , using theorem 1.2.1(iii ) in @xcite , the fact that @xmath862@xmath863 for all open sets @xmath864 and the fact that @xmath865 converges either to @xmath866 or @xmath867 , we can conclude that @xmath868 if @xmath869 , then , since @xmath36 for all @xmath870 , we deduce that @xmath871 and @xmath872 . hence , using ( [ nudef ] ) and ( [ muzero ] ) , we first see that @xmath873 \dvtx x_{t}=x_{t^{-}},x_{t}\in d_{t } \ } ) = 0,\ ] ] and then , by the absolute continuity of @xmath59 with respect to @xmath855 , we can conclude that @xmath874 \dvtx x_{t}=x_{t^{-}},x_{t}\in d_{t } \ } ) = 0 .
\label{milan1}\ ] ] in particular , ( [ milan1 ] ) proves ( [ sp4 ] ) .
hence , it only remains to prove that @xmath875 , as defined in ( [ laba2 ] ) , satisfies @xmath876 for all @xmath877 $ ] and @xmath795 $ ] . from ( [ lambdamurel3 ] ) and the fact that ( [ muzero ] ) implies that@xmath878 we deduce that @xmath879 \dvtx x_{s}=x_{s^{- } } \ } } \gamma _ { s}\,d \vert\lambda \vert_{s } & = & \int _ { \ { ( s , z,\gamma ) \in g_{t}\dvtx z = x_{s}=x_{s^{- } } \ } } \gamma\,d\mu ( s , z,\gamma ) \nonumber\\ & = & \int _ { \ { s\in [ 0,t ] \dvtx x_{s}=x_{s^{- } } \ } } \int_{\gamma_{s}^{1 } ( x_{s } ) } \gamma p ( s , x_{s},d\gamma ) \,d\nu,\end{aligned}\ ] ] whenever @xmath23 $ ] .
note that the last equality in ( [ ddd ] ) follows from the definition of @xmath855 in ( [ nudef ] ) . here
@xmath880 is a measure on the borel @xmath156-algebra of @xmath487 , concentrated on @xmath881 for @xmath882-almost all @xmath100 $ ] such that @xmath883 , and @xmath884 is a nonnegative borel measurable function for every borel set @xmath535 .
then , since @xmath885 is absolutely continuous with respect to @xmath855 , the radon nikodym theorem asserts the existence of a nonnegative borel measurable function @xmath281 such that@xmath886 \\[-8pt ] \eqntext{d\nu\mbox{-a.e . for all
} s\in \ { s\in [ 0,t ] \dvtx x_{s}=x_{s^{- } } \ } .}\end{aligned}\ ] ] from the assumption in ( [ cone1 ] ) we deduce that @xmath281 is strictly positive .
thus , by the convexity of @xmath887 and the absolute continuity of @xmath59 with respect to @xmath855 , we conclude that@xmath888 \dvtx x_{s}=x_{s^{- } } \ } .
\label{ddd3}\ ] ] in particular , ( [ ddd3 ] ) verifies the second part in ( [ sp2 ] ) and , hence , the proof in case 2 is also complete .
having completed the proof of cases 1 and 2 , we conclude that the proofs of ( [ bv ] ) and ( [ prop ] ) are complete .
hence , to complete the proof of theorem [ spconvergence ] , we only have to ensure that @xmath889,\mathbb{r } ) $ ] .
however , this follows from the assumption that @xmath890 , \mathbb{r } ) $ ] for all @xmath198 and from the fact that @xmath891 in the skorohod topology .
let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) , ( [ limitzero ] ) and a uniform exterior sphere condition in time with radium @xmath85 in the sense of ( [ extsphere- ] ) .
let @xmath118 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath23 $ ] , and assume that @xmath73 satisfies ( [ cone1 ] ) , ( [ cone2 ] ) and ( [ gammanpre ] ) .
assume that ( [ crita ] ) and ( [ crite ] ) hold for some @xmath119 , @xmath106 , @xmath120 and @xmath121 .
finally , assume that @xmath892 \times \mathbb{r } ^{d } ) \setminus\overline{d}$ ] has the @xmath893-property of good projections along @xmath73 , for some @xmath894 , @xmath109 and let @xmath895,\mathbb{r } ^{d } ) $ ] with @xmath63 .
the purpose of this subsection is to construct a sequence of solutions to skorohod problems which approximate the skorohod problem for @xmath129 .
based on this sequence , in the next section we conclude the existence of a solution @xmath128 to the skorohod problem for @xmath65 , in the sense of definition [ skorohodprob ] , with @xmath889,\mathbb{r } ^{d } ) $ ] .
this will then complete the proof of theorem [ theorem 1 ] .
to proceed , we let @xmath896 , @xmath897 , and we let @xmath898 be a sequence of real numbers which tends to @xmath899 as @xmath434 . then , for each @xmath191 , we can find a partition @xmath900 of the interval @xmath66 $ ] , that is , @xmath901 , such that ( [ korlim++])([korlim+++ ] ) stated below hold . in particular,@xmath902 and , for some @xmath903,@xmath904 \\[-8pt ]
\eqntext{\mbox{whenever } n\geq n_{0},k\in \ { 0,\ldots , n_{n}-1 \ } .}\end{aligned}\ ] ] furthermore , we define @xmath905 , @xmath61 $ ] , as @xmath906 and @xmath907 , so that @xmath908 , \mathbb{r } ^{d } ) , \nonumber \\[-8pt ] \\[-8pt ] w_{0}^{\delta_{n}}&\in&\overline{d_{0}}\quad \mbox{and}\quad d_{{\mathcaligr}{d}}([0,t],w^{\delta_{n}},w)\leq\epsilon_{n}.\nonumber\end{aligned}\ ] ] in particular , @xmath909 , \mathbb{r } ^{d } ) $ ] is a step function which approximates @xmath3 in the skorohod topology . given @xmath910 and @xmath269 , we define @xmath911 and @xmath275 as in ( [ approxw+ ] ) . furthermore , to obtain a more simple notation , from now on
we write @xmath912 , @xmath716 and @xmath762 for @xmath269 , @xmath274 and @xmath913 .
then , given @xmath912 , @xmath716 and @xmath762 as above , we next define a pair of processes @xmath707 as follows .
let@xmath914 if @xmath915 for some @xmath916 , then , by the triangle inequality and ( [ korlim lex ] ) , @xmath917 hence , by the @xmath236-property of good projections , it follows that if @xmath918 , then there exists a point @xmath919 which is the projection of @xmath920 onto @xmath921 along @xmath922
. furthermore , if @xmath923 , then we let @xmath924 based on this argument , we define , whenever @xmath925 , @xmath926 , @xmath927 \\[-8pt ] \lambda_{t}^{n } & = & \lambda_{\tau_{k-1}^{n}}^{n}+\bigl(x_{t}^{n}-(x_{\tau _ { k-1}^{n}}^{n}+w_{\tau_{k}^{n}}^{n}-w_{\tau_{k-1}^{n}}^{n})\bigr),\nonumber\end{aligned}\ ] ] and , finally , we define @xmath928 and @xmath929 using ( [ approxw+++b ] ) by simply setting @xmath930 in ( [ approxw+++b ] ) .
note that in this way we have @xmath931 for all @xmath916 .
next , again using the @xmath236-property of good projections , we see that@xmath932 hence , @xmath933 , \mathbb{r } ^{d } ) $ ] . using this notation , we next prove the following theorem .
[ spconvergence+]let @xmath20 , @xmath13 , @xmath85 , @xmath131 , @xmath105 , @xmath106 , @xmath120 , @xmath121 , @xmath146 and @xmath480 be as in the statement of theorem [ theorem 1 ] .
let @xmath3 be as in the statement of theorem theorem 1 and let @xmath912 , @xmath716 , @xmath762 , @xmath934 and @xmath935 be defined as above for @xmath198 .
then @xmath936 is a solution to the skorohod problem for @xmath200 and @xmath937 , \mathbb{r } ^{d } ) $ ] for all @xmath263 for some @xmath834 .
moreover , @xmath775 converges to @xmath938 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] and @xmath128 is a solution to the skorohod problem for @xmath129 .
furthermore , @xmath939 , \mathbb{r } ^{d } ) $ ] .
[ subtel2 ] note that for theorem [ spconvergence+ ] we , in contrast to in theorem [ spconvergence ] , also need to assume ( [ gammanpre ] ) in order to be able complete the proof [ see ( [ bbbb2again ] ) below ] .
proof of theorem [ spconvergence+ ] @xmath707 is , by construction , a solution to the skorohod problem for @xmath200 and the statement that @xmath940 , \mathbb{r } ^{d } ) $ ] for all @xmath263 , for some @xmath834 , is proved in ( [ bound3 ] ) .
next , using theorem compacttheoremapprox , we can conclude the existence of some positive constants @xmath253 , @xmath254 , @xmath255 and @xmath256 such that@xmath941 \\[-8pt ] \quad \vert\lambda^{n } \vert_{t_{2}}- \vert\lambda ^{n } \vert_{t_{1 } } & \leq&\hat{l}_{3 } ( w , t ) \vert w \vert_{t_{1},t_{2}}+\hat{l}_{4 } ( w , t ) \bigl ( l ( t_{2}-t_{1 } ) + l ( \delta_{n}^{\ast } ) \bigr ) , \nonumber\end{aligned}\ ] ] whenever @xmath53 . in particular , note that by choosing @xmath260 sufficiently large we can ensure , using ( [ korlim++ ] ) , that @xmath942 and that ( [ ahah1 + ] ) holds for all @xmath838 .
hence , theorem [ compacttheoremapprox ] is applicable .
based on ( [ lll1 + ] ) , we can now argue as in the proof of theorem [ spconvergence ] .
in particular , as @xmath192 converges to @xmath452 , \mathbb{r } ^{d } ) $ ] , we see , using ( [ lll1 + ] ) , that @xmath943 is relatively compact in @xmath785 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } _ { + } ) $ ] .
furthermore , we know that @xmath944 for all @xmath23 $ ] , @xmath198 . hence , @xmath205 converges to some @xmath945 , \mathbb{r } ^{d } ) \times{\mathcaligr}{d } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] .
we intend to prove that @xmath788 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] solves the skorohod problem for @xmath129 and , to do this , we have to prove that @xmath789 , \mathbb{r } ^{d } ) , \label{bv+}\ ] ] and we have to verify @xmath946 the proof of ( [ bv+ ] ) and ( [ prop+ ] ) follows along the lines of the proof of ( [ bv ] ) and ( [ prop ] ) in theorem [ spconvergence ] and we shall only outline the main differences between the proofs . to start with , the statements in ( [ hra1 ] ) and ( [ convth3 ] ) remain true . however , the argument in ( [ bbbb1- ] ) has to be changed .
in particular , in this case we see , using ( [ convth3 ] ) and ( [ korlim++ ] ) , that@xmath947 \\[-8pt ] & & { } + l(|\tilde{t}_{n}-t|)+l(\delta_{n}^{\ast})\rightarrow0 \qquad\mbox{as } n\rightarrow \infty,\nonumber\end{aligned}\ ] ] which , since @xmath948 is closed , proves that @xmath949 , for all @xmath23 $ ] .
hence , we have verified that @xmath950 satisfies ( [ sp1 ] ) .
next , arguing as in ( [ defr+])([laba2 ] ) , using ( [ lll1 + ] ) to conclude ( [ laba1 ] ) , we can conclude that ( [ bv+ ] ) holds and that @xmath951\ ] ] for some @xmath487-valued borel measurable function @xmath841 .
hence , to prove ( [ sp2 ] ) , we again have to prove that @xmath952 for all @xmath23 $ ] . as in the proof of theorem [ spconvergence ] , we consider case 1 and case 2 .
in fact , case 2 can be handled exactly as in the proof of theorem [ spconvergence ] and hence shall only discuss the proof of case 1 . to prove case 1
, we first see that the statements in ( [ hra2 ] ) and ( [ convth2 ] ) can be repeated and , arguing as in ( [ bbbb1-+ ] ) , we obtain@xmath953 \\[-8pt ] & & { } + \hat{l}(|\tilde{t}_{n}-t|)+\hat{l}(\delta_{n}^{\ast})\rightarrow0 \qquad\mbox { as } n\rightarrow\infty.\nonumber\end{aligned}\ ] ] hence , since @xmath27 is closed , we can conclude that @xmath954 .
to proceed , we deduce as in ( [ bbbb2 ] ) that@xmath955 obviously the first two terms on the right - hand side in ( [ bbbb2again ] ) tend to zero as @xmath434 . concerning the third term
, we first note that there exists , for @xmath191 large enough , some integer @xmath956
such that @xmath957 hence , as @xmath958 , we can conclude , using ( [ convth3 ] ) , that @xmath959 as @xmath434 .
moreover , as @xmath960 , we can also use ( [ convth3 ] ) to conclude that @xmath961 as @xmath434 .
in particular , based on these conclusions , it follows from ( [ gammanpre ] ) that also the third term on the right - hand side in ( [ bbbb2again ] ) tends to zero as @xmath434 .
hence , having proved that @xmath962 , the proof of case 1 can now be completed as in theorem [ spconvergence ] .
having completed the proof of cases 1 and 2 , we can conclude that the proofs of ( [ bv+ ] ) and ( [ prop+ ] ) are complete .
hence , to complete the proof of theorem [ spconvergence+ ] , we only have to ensure that @xmath939 , \mathbb{r } ^{d } ) $ ] .
however , again this follows from the fact that @xmath774 , \mathbb{r } ^{d } ) $ ] for all @xmath263 and from the fact that @xmath963 in the skorohod topology .
proof of theorem [ theorem 1 ] theorem [ theorem 1 ] now follows immediately from theorem [ spconvergence+ ] .
proof of theorem [ theorem 2 ] using theorem compacttheorem and ( [ limitzero ] ) , we see that @xmath964 this proves that @xmath4 is continuous .
proof of theorem [ theorem 3 ] let @xmath163 be a @xmath150-dimensional wiener process on a filtered probability space @xmath965 and in the following let @xmath966 , @xmath75 $ ] , be a continuous path of @xmath163 .
we define , for @xmath896 , @xmath897 , @xmath967 , @xmath968 \\[-8pt ] \gamma_{t}^{n}&=&\gamma_{kt/2^{n } } \qquad \mbox{whenever } t\in [ kt/2^{n } , ( k+1 ) t/2^{n } ) , \nonumber\end{aligned}\ ] ] and @xmath969 .
furthermore , we recursively define three processes @xmath970 , @xmath971 and @xmath972 , for @xmath75 $ ] , in the following way .
let @xmath973 and let , for @xmath967 , @xmath974 we here have to make sure that @xmath975 is well defined . to do this ,
we note that , either @xmath976 or @xmath977 . in the first case
we identify the projection with the point itself , whereas , in the latter case , we have to assert the existence of appropriate projections onto @xmath978 .
however , assuming @xmath979 , we see that @xmath980 hence , since @xmath966 is a continuous path , there must exist some @xmath981 such that @xmath982 whenever @xmath263 and @xmath967 . by ( [ dsds5 ] ) and the @xmath236-property of good projections
, it follows that the projection @xmath983 is well defined , for @xmath263 , whenever @xmath984 .
furthermore , using the definition of @xmath985 in ( [ dsds2 ] ) , we also see that @xmath986 \\[-8pt ] & & { } + \bigl(\sup_{\overline{d } } \vert\sigma \vert\bigr)\bigl|w _ { ( k+1 ) t/2^{n}}-w_{kt/2^{n}}\bigr|,\nonumber\end{aligned}\ ] ] and , hence , once more using that @xmath966 is a continuous path , we can ensure that @xmath987 , \mathbb{r } ^{d } ) , \nonumber \\[-8pt ] \\[-8pt ] \mathrm{(ii ) } & & \hspace*{5pt}h_{0 } \bigl ( \vert z^{n}\vert _ { kt/2^{n},(k+1)t/2^{n}}+l(\delta _ { n}^{\ast } ) \bigr ) + \vert z^{n}\vert _ { kt/2^{n},(k+1)t/2^{n}}\leq\rho_{0},\nonumber\end{aligned}\ ] ] whenever @xmath263 and @xmath967 .
we next let , for @xmath967 , @xmath988 finally , we define , for @xmath989 , @xmath990 , @xmath991 then , by arguing as in the proof of ( [ bound3 ] ) , using ( i ) and ( ii ) in ( [ dsds4ha1 ] ) , we can conclude that@xmath992 , \mathbb{r } ^{d } ) \qquad \mbox{whenever } n\geq n_{0}. \label{boliden}\ ] ] furthermore , using the definitions above , it is clear that @xmath993 where@xmath994 } \vert\varepsilon^{n } ( t ) \vert \leq\bigl(\sup_{\overline{d } } \vert b \vert\bigr)\frac{t}{2^{n}}+\bigl(\sup_{\overline{d } } \vert\sigma \vert\bigr)\mathop{\mathop{\sup}_{0\leq s\leq t\leq t , } } _ { \vert s - t \vert\leq { t}/{2^{n } } } \vert w_{t}-w_{s } \vert . \label{rsde1}\ ] ] by construction , @xmath995 solves the skorohod problem for @xmath996 and using theorem compacttheoremapprox , we can conclude that there exist positive constants @xmath997 , @xmath998 , @xmath999 and @xmath1000 , independent of @xmath191 , for @xmath263 , such that @xmath1001 \\[-8pt]\qquad z , t ) \vert z \vert_{t_{1},t_{2}}+\hat{l}_{4 } ( z , t ) \bigl(l ( t_{2}-t_{1 } ) + l(t/2^{n})\bigr),\nonumber\end{aligned}\ ] ] whenever @xmath53 and @xmath263 . hence , the sequence @xmath1002 is relatively compact in the skorohod topology and we can conclude , by the construction of @xmath1003 and ( [ rsde2])([rsde1 ] ) , that also the sequence @xmath1004 is relatively compact in the skorohod topology .
in fact , as this argument can be repeated for each continuous path of @xmath966 , it follows , by considering convergent subsequences if necessary , that the sequence of vector valued processes @xmath1005 defined on @xmath1006 converges in law to a stochastic process @xmath1007 .
furthermore , using the skorohod representation theorem ( see , e.g. , @xcite and @xcite ) , there exists a complete probability space @xmath1008 and versions @xmath1009 and @xmath1010 of @xmath1011 and @xmath1012 on @xmath1008 , such that @xmath1013 converges to @xmath1014 @xmath1015-almost surely .
moreover , using theorem spconvergence+ and the fact that @xmath1016 solves , @xmath1015-almost surely , the skorohod problem for @xmath1017 , it follows that @xmath1018 solves , @xmath1015-almost surely , the skorohod problem for @xmath1019 .
in particular , @xmath1020 , \mathbb{r } ^{d } ) \times{\mathcaligr}{bv } ( [ 0,t ] , \mathbb{r } ^{d } ) $ ] and @xmath1021 \dvtx \tilde{x}_{t}\in d_{t}\})=0\end{aligned}\ ] ] holds @xmath1015-almost surely whenever @xmath23 $ ] .
we next want to verify that @xmath1022 holds @xmath1015-almost surely whenever @xmath75 $ ] . indeed , using ( [ rsde2 ] ) and ( [ rsde1 ] ) , we can , following @xcite , simply quote theorem 2.2 in @xcite , which in our case states that since @xmath1023 converges to @xmath1024 @xmath1015-almost surely whenever @xmath843 $ ] , @xmath1025 \\[-8pt ] \eqntext{\tilde{p}\mbox{-almost surely,}}\end{aligned}\ ] ] whenever @xmath23 $ ] , as @xmath434 .
this proves ( [ hahaga ] ) .
now let @xmath1026 and @xmath1027 be the @xmath156-algebras generated by @xmath1028 and @xmath1029 , respectively .
we next prove that @xmath1030 is a @xmath150-dimensional wiener process on the filtered probability space @xmath1031 . to obtain this
, we first note that the @xmath156-algebra generated by @xmath1032 , for @xmath675 $ ] , is independent of @xmath1027 . furthermore , since @xmath1033 @xmath1015-almost surely , it follows that the @xmath156-algebra generated by @xmath1034 , for @xmath75 $ ] , is independent of @xmath1026 .
in particular , @xmath1035\}$ ] is a martingale with respect to @xmath1036\}$ ] and @xmath1015 . now let @xmath1037 and @xmath1038 . then , using essentially the same argument as in ( [ hahaga+++ ] ) , we also see that @xmath1039 , @xmath1015-almost surely , for all @xmath1040 , and , hence , as above , we can conclude that @xmath1041\}$ ] , with @xmath1042 being the kronecker delta , is a martingale with respect to @xmath1043\}$ ] and @xmath1015 . by the lvy characterization of wiener processes ( see , e.g. , theorem ii.6.1 in @xcite ) , we can thus conclude that @xmath1044 is a @xmath150-dimensional wiener process on @xmath1045 . to finally conclude that @xmath1018 is a weak solution in the sense of definition [ weaksol ] , and hence to complete the proof of theorem [ theorem 3 ]
, it only remains to prove the existence of a version of @xmath1046 on @xmath1047 , denoted @xmath1048 , such that @xmath1049 , \mathbb{r } ^{d})$ ] , @xmath1015-almost surely .
however , using standard arguments , we first note that there exists a version of @xmath1050 on @xmath1051 , denoted @xmath1052 , such that @xmath1053 , \mathbb{r } ^{d})$ ] , @xmath1015-almost surely , and such that @xmath1054 solves , @xmath1015-almost surely , the skorohod problem for @xmath1055 .
furthermore , by ( [ boliden ] ) , it is clear that @xmath1056 , \mathbb{r } ^{d } ) $ ] .
hence , by theorem [ theorem 2 ] , @xmath1046 is continuous @xmath1015-almost surely .
this completes the proof of theorem [ theorem 3 ] .
concerning the @xmath111-property of good projections along @xmath73 , the following result follows immediately from theorem 4.1 in @xcite .
[ costprop1 ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
let @xmath131 be a closed convex cone of vectors in @xmath7 for every @xmath36 , @xmath75 $ ] and assume that @xmath73 satisfies ( [ cone1 ] ) and ( [ cone2 ] ) .
assume that there exists a continuous map @xmath1057 such that @xmath1058 , \nonumber \\ q ( t , z,\lambda v ) & = & \lambda q ( t , z , v ) \\
\eqntext{\mbox{for all } \lambda\geq0 , v\in n_{t } ( z ) , z\in\partial d_{t } , t\in [ 0,t ] . } \ ] ] moreover , assume that @xmath1059 , z\in\partial d_{t}}\max_{v\in n_{t}^{1 } ( z ) } \vert q ( t , z , v ) \vert&:= & \vert q \vert<\infty,\nonumber \\[-8pt ] \\[-8pt ] \inf_{t\in\lbrack0,t ] , z\in\partial d_{t}}\min_{v\in n_{t}^{1 } ( z ) } v\cdot q ( t , z , v ) & : = & q>0.\nonumber\end{aligned}\ ] ] let @xmath1060 \\[-8pt ] h_{0}&:=&\frac{q/ \vert q \vert}{1-\sqrt{1- ( q/ \vert q \vert ) ^{2}}}.\nonumber\end{aligned}\ ] ] then @xmath110\times \mathbb{r } ^{d})\setminus\overline{d}$ ] has the @xmath111-property of good projections along @xmath73
. note that in lemma [ costprop1 ] we have that @xmath1061 , @xmath108 and @xmath109 by construction . to continue , given @xmath20 and @xmath2 as above
, we say that @xmath2 is a @xmath1062-domain if we can find a @xmath1063 such that , for all @xmath1064 , @xmath1065 $ ] , there exists a function @xmath1066 , @xmath1067 , with the properties@xmath1068 for all @xmath1069 .
[ compactprop ] let @xmath20 and let @xmath13 be a time - dependent domain satisfying ( [ timedep+ ] ) and a uniform exterior sphere condition in time with radius @xmath85 in the sense of ( [ extsphere- ] ) .
assume , in addition , that @xmath2 is a @xmath1070-domain for some @xmath317 $ ] .
let @xmath1071 be , for every @xmath36 , @xmath75 $ ] , a closed convex cone of vectors in @xmath7 with the specific form @xmath1072 , for some @xmath487-valued function @xmath488 which is uniformly continuous , in both space and time , and satisfies @xmath1073 } \inf_{z\in\partial d_{t } } \langle \gamma_{t } ( z ) , n_{t } ( z ) \rangle>0 . \label{condbeta}\ ] ] then @xmath2 satisfies ( [ crita ] ) and ( [ crite ] ) and @xmath1074 \label{holder}\ ] ] for some @xmath1075 and with @xmath1076 $ ] . by the uniform continuity of @xmath1077 in space and time ,
it is clear that the variation of @xmath488 can be made arbitrarily small on temporal neighborhoods .
following proposition 2.5 in @xcite , it therefore immediately follows that criteria ( [ crita ] ) and ( [ crite ] ) are satisfied for some @xmath105 and @xmath106 .
hence , it remains to prove ( [ holder ] ) and we note that it suffices to prove ( holder ) for small values of @xmath326 .
let @xmath300 be arbitrary and let @xmath1078 be such that @xmath1079 .
we now claim that @xmath1080 or , in other words , that @xmath1081 and @xmath1082 are parallel .
to prove this claim , we can assume , without loss of generality , that @xmath1083 . as @xmath2 is a time - dependent domain of class @xmath1084 , we can assume the existence of a function @xmath1085 , with property ( [ h1alpha ] ) , such that @xmath1086 , where @xmath1087 is a ( spatial ) neighborhood of the origin with the radius @xmath1088 as given in the definition of @xmath1062-domains .
we consider the minimization problem @xmath1089 then , as the minimum in ( [ minprob ] ) is realized at the origin , we see that @xmath1090 for some lagrange multiplier @xmath792 .
obviously , this proves ( [ 2 ] ) .
next , by lemma [ llhatequivalence ] , we see that , for @xmath1091 , @xmath1092 } } _ { \vert s - t \vert\leq r}\sup_{z\in\overline{d_{s}}}d ( z , d_{t } ) = \mathop{\mathop{\sup } _ { 0\leq s\leq t\leq t } } _ { \vert s - t \vert \leq r}\vert z_{t}-z_{s } \vert \label{lrep}\ ] ] for some @xmath1078 , @xmath1093 such that @xmath1094 and @xmath1095 .
furthermore , employing once more the fact that @xmath2 is a time - dependent domain of class @xmath1062 , we conclude the existence of a function @xmath1085 , with the property @xmath1096 , which is continuously differentiable in space .
taylor expanding @xmath1097 up to the first order in spatial coordinates , we obtain , by the @xmath1062-regularity of @xmath1085,@xmath1098 as @xmath1099 , it follows from ( [ lrep])([taylor ] ) that @xmath1100 \\[-8pt ] & \leq&\mathop{\mathop{\sup } _ { 0\leq s\leq t\leq t } } _ { \vert s - t \vert \leq r}\frac { \vert\psi ( t , z_{t } ) -\psi ( s , z_{t } ) \vert } { \vert\nabla_{z}\psi ( s , z_{s } ) \vert}+\mathop{\mathop{\sup } _ { 0\leq s\leq t\leq t } } _ { \vert s - t \vert\leq r}\frac{{\mathcaligr}{o } ( \vert z_{t}-z_{s } \vert^{1+\alpha } ) } { \vert\nabla_{z}\psi ( s , z_{s } ) \vert}.\nonumber\end{aligned}\ ] ] furthermore , for small @xmath326 , we can assume , without loss of generality , that for some @xmath54 independent of @xmath1101 we have @xmath1102 .
hence , by the @xmath1062-regularity of @xmath1085 and the definition of @xmath453 , we see that @xmath1103 provided @xmath1091 .
finally , since @xmath1104 as @xmath1105 , there exists @xmath325 such that if @xmath1106 , then @xmath1107 . combining these facts
, we conclude that @xmath1108 for some constant @xmath1109 .
hence , ( [ holder ] ) holds with @xmath1110 . in the following
let @xmath1111 and @xmath1112 be spaces containing all functions with bounded derivatives up to orders one and two , respectively .
consider a bounded spatial domain @xmath1113 which is @xmath1111-smooth and satisfies a uniform exterior sphere condition .
moreover , assume that the cone of directions of reflection @xmath1114 has the specific form @xmath1115 , for some @xmath487-valued function @xmath1116 which is continuous and satisfies@xmath1117 then proposition 2.5 in @xcite states that the time - independent counterparts of criteria ( [ crita ] ) and ( [ crite ] ) are satisfied .
furthermore , theorem 4.5 in @xcite states that the time - independent counterparts of ( [ crita ] ) and ( [ crite ] ) are also satisfied for piecewise @xmath1111-smooth domains @xmath38 if the function @xmath1116 is uniformly continuous on each face of @xmath1118 and satisfies some nondegeneracy and consistency criteria .
finally , we also mention theorem 4.6 in @xcite which states that unique projections may be found if @xmath38 is a piecewise @xmath1112-smooth domain and if @xmath1119 is lipschitz continuous on each face of @xmath1118 and satisfies some other nondegeneracy and consistency criteria .
the authors would like to express their sincere gratitude to several anonymous referees for their valuable comments and suggestions , which have resulted in considerable improvement of the results and presentation of this article . | the deterministic skorohod problem plays an important role in the construction and analysis of diffusion processes with reflection . in the form studied here ,
the multidimensional skorohod problem was introduced , in time - independent domains , by h. tanaka @xcite and further investigated by p .-
l .
lions and a .- s .
sznitman @xcite in their celebrated article .
subsequent results of several researchers have resulted in a large literature on the skorohod problem in time - independent domains . in this article
we conduct a thorough study of the multidimensional skorohod problem in time - dependent domains .
in particular , we prove the existence of cdlg solutions @xmath0 to the skorohod problem , with oblique reflection , for @xmath1 assuming , in particular , that @xmath2 is a time - dependent domain ( theorem [ theorem 1 ] ) .
in addition , we prove that if @xmath3 is continuous , then @xmath4 is continuous as well ( theorem [ theorem 2 ] ) .
subsequently , we use the established existence results to construct solutions to stochastic differential equations with oblique reflection ( theorem [ theorem 3 ] ) in time - dependent domains . in the process of proving these results
we establish a number of estimates for solutions to the skorohod problem with bounded jumps and , in addition , several results concerning the convergence of sequences of solutions to skorohod problems in the setting of time - dependent domains . . |
three dimensional gravity has often been used as a toy model for the four dimensional theory . here
we study point particles in 2 + 1 dimensions using spatial geometries analogous to those in @xcite .
these geometries are isomorphic to a gauge - reduced holonomy - flux phase space which descends from the @xmath0 limit of a loop quantum gravity hilbert space .
we consider a hamiltonian written in terms of frame - field and connection variables . the spacetime signature is taken to be euclidean since the relevant gauge group is then @xmath1 , as in 3 + 1 dimensions with lorentzian signature .
our first step is to specify the variables over an entire spacelike slice using a triangulation where particles sit on the vertices .
after gauge fixing , the hamiltonian is reduced to a boundary term which equals the total particle mass . in this gauge
the triangulation evolves according to particle momenta and undergoes discrete changes when a particle meets an edge , collapsing one of the triangles .
spacetime is taken to be @xmath2 where @xmath3 is a spacelike surface homeomorphic to a disc .
the first order formalism of general relativity parameterizes the gravitational field in terms of a connection @xmath4 and a frame - field @xmath5 , both being one - forms on @xmath3 taking values in the @xmath6 algebra .
we use @xmath6 basis elements @xmath7 ( for @xmath8 ) which are given by @xmath9 times the pauli matrices .
our notation is such that elements of @xmath6 are written as @xmath10 , for example , and all internal indices are written as superscripts . in units where @xmath11 a hamiltonian for pure gravity is given by @xcite : [ hamiltonian ] h & = & - _ ( n^i f^i + ^i g^i ) + _ n^i a^i , + f^i & = & a^i + ^ijk a^j a^k , + g^i & = & e^i + ^ijk a^j e^k . @xmath12 and @xmath13 are lagrange multipliers for the flatness and gauss constraints respectively .
we normalize the time coordinate by choosing @xmath14 . the boundary term and the condition
@xmath15 ensure that the variational principal is well - defined and the constraints are first class .
the poisson algebra is : \{a_a^i(x ) , e_b^j(y ) } & = & _ ab ^ij ^2(x - y ) , where @xmath16 labels the space coordinates .
it is well - known that in 2 + 1 dimensions point particles represent conical singularities @xcite .
we introduce particles by replacing the flatness constraint with @xmath17 so that each particle @xmath18 gives a contribution to the curvature proportional to its momentum @xmath19 at the location @xmath20 @xcite . in this treatment the particles
are defined by the fields @xmath21 and do not have independent parameters of their own .
we consider the total mass to be less than @xmath22 so that @xmath3 is open @xcite .
in order to solve the constraints and reduce the hamiltonian , we triangulate @xmath3 according to the particle locations so that we may give a piecewise definition of @xmath21 on @xmath23
. see @xcite for other approaches along these lines . for clarity , we take the boundary @xmath24 to be triangular ; the generalization to arbitrary polygons follows simply .
we take each particle position to be an internal vertex and connect all vertices with straight edges to obtain a triangulation .
there is ambiguity in this procedure , but the number of triangles @xmath25 is fixed by the number of particles @xmath26 to be @xmath27 via the euler characteristic for planar graphs .
different triangulations generally lead to different gauge choices . within each triangle @xmath28 , we must choose @xmath29 such that @xmath30 .
the general solution @xcite is given by an @xmath1 group element @xmath31 and a closed one - form @xmath32 : [ solutions ] _
= a_a_^-1 , _
= a__a_^-1 .
consider two adjacent triangles labeled @xmath33 and @xmath34 .
we ensure that the fields are continuous across the shared edge @xmath35 by requiring that there exists an @xmath36 on the edge such that : [ gluing ] .
a_2 |_e = .a_1 h_e |_e , .
_ 2|_e = . h_e^-1
_ 1 h_e |_e .
now , if we integrate the curvature over a region containing a single vertex @xmath18 we must obtain @xmath37 . in order to specify the group elements
@xmath31 we subdivide each triangle into three regions defined by edges joining the centroid to the vertices .
consider one such region @xmath38 where points @xmath39 are the endpoints of an edge , and @xmath40 is the centroid of the triangle .
we define : u = ^-1 ( v_+ ) + ^-1 ( v_- ) , where @xmath41 .
note that the line @xmath42 coincides with the edge , and @xmath43 consists of lines between the centroid of the triangle and the endpoints of the edge .
we set the group element in a region equal to : [ gena ] a__r(u)= ( _ 0^u f ( ) ) , where we have introduced a bump function @xmath44 which satisfies @xmath45 . for a given @xmath46
, this group element defines a rotation about a direction @xmath47 by an angle proportional to @xmath48 . with these ingredients
it is now possible to choose a gauge , i.e. to specify the frame - field and connection on @xmath23 . given a set of particles @xmath49 , one first defines a boundary and introduces a triangulation .
then we arbitrarily choose a triangle ( labeled 1 ) and make a choice for @xmath50 . moving to a neighbouring triangle
, we specify @xmath51 here and use ( [ gluing ] ) to find @xmath52 .
we continue in this manner until @xmath53 ( and thereby @xmath54 are defined in each triangle , taking care to ensure the connection yields proper curvature at each vertex by choosing the parameter @xmath55 in each @xmath56 to be consistent with particle momenta .
the aforementioned relationship between the number of triangles and the number of particles ensures that this procedure will work for arbitrarily many particles .
see @xcite for a more detailed description of this process .
preserving the gauge dynamically leads to a condition on the lagrange multipliers : [ condition ] _ a n^i + ^ijke^j^k = 0 .
this fixes three of the six degrees of freedom in @xmath12 and @xmath13 .
we eliminate the remaining ambiguity by choosing @xmath57 and setting @xmath58 .
this choice ensures that particles move along with the triangulation and remain at the vertices .
after specifying @xmath21 , the hamiltonian is reduced to a boundary term which evaluates to : h & = & _
n^i a^i = _ n^i _ a a^i + & = & _ n^i(q _ ) p_^i = _ m _ , where we used stokes theorem , the conditions on @xmath59 , and the flatness constraint .
dynamics are implicitly determined by the direction of momentum @xmath60 for each particle via the equation of motion : _
^i(t ) = q_^i(0 ) + t n^i(q _ ) , and the triangulation evolves accordingly .
non - trivial , discrete transitions occur when a particle meets an edge
. consider triangles @xmath33 and @xmath34 depicted in fig .
[ transition ] . before and after a discrete change in triangulation .
@xmath61 label the three regions of each triangle .
grey regions ( bounded by constant-@xmath46 lines ) indicate where the connection is generally non - zero . ]
if the particle at the top of the left hand side of the figure moves downward until it reaches the shared edge , then @xmath33 closes and @xmath34 splits into @xmath62 and @xmath63 ( shown on the right hand side ) preserving the total number of triangles .
the fields are defined in the new triangles by specifying the @xmath1 rotation parameters : & & _ 1^ _ = _ 1 _ + _ 2 _ + _ 1 _ , _ 1^ _ = _ 2 _ , + & & _ 2_^
= _ 1 _ + _ 2 _ + _ 1 _ , _
2_^ = _ 2 _ , + & & _ 1_^ = _ 2_^ = 0 .
notice @xmath64 along the new edge shared by @xmath62 and @xmath63 so that @xmath65 .
the fields in all other triangles are unaffected by this transition .
we have described a system of point particles in 2 + 1 dimensional gravity in terms of evolving triangulations .
a gauge choice for @xmath66 was specified by choosing @xmath31 and @xmath32 in each triangle .
the triangulation evolves according to particle dynamics so that particles remain at vertices at all times .
the discrete change in triangulation that occurs when a vertex meets an edge is well - defined
. this construction yields spatial geometries that are the two - dimensional analog of those in @xcite .
having @xmath53 in each triangle allows us to immediately write this data in terms of holonomy - flux variables .
moreover , knowing how @xmath53 behave under a discrete change in triangulation tells us how the holonomies and fluxes will change .
this sets the stage for a semiclassical loop gravity description of the system @xcite . | we study point particles in 2 + 1 dimensional first order gravity using a triangulation to fix the connection and frame - field .
the hamiltonian is reduced to a boundary term which yields the total mass .
the triangulation is dynamical with non - trivial transitions occurring when a particle meets an edge .
this framework facilitates a description in terms of the loop gravity phase space . |
we perform the 2d fourier transform of eq .
( [ bi ] ) @xmath48 where the vector function @xmath49 is defined by eq .
( [ y ] ) .
inverting eq .
( [ res ] ) in the reciprocal space , we can write @xmath50 d z'.\ ] ] then we can write for the gradient of the effective potential @xmath51 \int \left\ { \epsilon^{-1}_{\gv\gv'}(\qv , z',\omega ) \phi^\textit{ext}(\gv'+\qv , z',\omega ) + \frac{2\pi } { s_0 } \times \right . \\ & \left .
\sum\limits_{\beta } z_\beta e^{-i(\gv ' + \qv ) \cdot \bv_\beta } \ ! \ ! \int \
! \left [ \epsilon^{-1}_{\gv\gv'}(\qv , z ' , \omega ) -\delta_{\gv\gv ' } \delta(z - z')\right ] \ev_\beta \cdot \yv(z',\gv'+\qv ) \right\}_{z=0 } \ ! \ ! \ ! \ ! \ ! \ ! \ ! d z ' \left . \ ! - \ ! \ ! \ ! \ ! \ ! \ !
\sum\limits_{(\beta\rv)\ne(\alpha\0v ) } \ ! \ ! \ ! \ ! \ ! \ ! \ ! \ ! e^{i\qv\cdot\rv } \nabla ( \ev_\beta \ ! \cdot \ !
\nabla ) \frac{z_\beta}{|\rv \ ! - \
! \bv_\beta \ ! - \ !
\rv| } \right|_{\rv=\bv_\alpha } .
\end{split}\ ] ] noting that within the static approximation ( @xmath37 ) our theory reduces to the conventional density - functional perturbation theory ( dfpt ) @xcite , we can conveniently rewrite eq .
( [ force ] ) as eq .
( [ forced ] ) . | the existence of an acoustic plasmon in extrinsic ( doped or gated ) monolayer graphene was found recently in an _ ab initio _ calculation with the frozen lattice [ m. pisarra _
et al . _ , arxiv:1306.6273 , 2013 ] . by the _
fully dynamic _ density - functional perturbation theory approach , we demonstrate a strong coupling of the acoustic plasmonic mode to lattice vibrations . thereby , the acoustic plasmon in graphene does not exist as an isolated excitation , but it is rather bound into a combined plasmon - phonon mode . we show that the coupling provides a mechanism for the _ bidirectional _ energy exchange between the electronic and the ionic subsystems with fundamentally , as well as practically , important implications for the lattice cooling and heating by electrons in graphene . known for its extraordinary properties and vast potential applications @xcite , graphene a two - dimensional crystal comprised of a honeycomb lattice of carbon atoms continues to receive much attention as it reveals new remarkable features @xcite . for one of the recent findings , an acoustic plasmon ( apl )
( plasmon with _ linear _ wave - vector dispersion ) has been predicted theoretically in an extrinsic free - standing monolayer graphene @xcite .
this finding is extraordinary considering that apl generation conventionally involves a surface state immersed in the bulk of a metal @xcite . exhibiting linear wave - vector dispersion , acoustic apl persists down to low frequencies , where it can be expected to interact with phonon oscillations .
the possibility of coupling these two types of elementary excitations motivates questions of fundamental physics as well as of potential applications . in this letter
we show that the apl - phonon coupling indeed occurs in the electron - doped graphene and it provides a mechanism for the bidirectional energy exchange between the electronic and ionic subsystems .
the conventional treatment of lattice vibrations by frequency - independent density - functional perturbation theory ( dfpt ) @xcite is inadequate for capturing the essentially dynamic nature of the coupled plasmon - phonon modes , and we therefore implement a fully _ dynamic _ approach treating the electron - hole , plasmon , and phonon elementary excitations on the equal footing @xcite . our _ ab initio _ calculations for monolayer graphene employ the full - potential linear augmented plane - wave ( fp - lapw ) code elk @xcite .
the super - cell geometry is utilized with a separation of the layers in the @xmath0 direction of 40 bohr , which effectively ensures the non - interaction between the layers .
the local - density approximation to the exchange - correlation potential @xcite is used .
_ acoustic - plasmon and phonons in graphene. _ we start by reproducing the apl and phonon spectra of graphene without the coupling of the two excitations . in fig .
[ plph ] , left panel , the energy - loss function of graphene is plotted for a number of equidistant values of the wave - vector .
the calculation with the carbon atoms fixed at their equilibrium positions has been used .
the apl can be easily recognized by the linear dispersion of the peak with the wave - vector , which is in agreement with the recent findings in extrinsic graphene obtained with the use of the pseudopotential method @xcite . in fig .
[ plph ] , right panel , we plot the phonon dispersion spectra in graphene together with the apl dispersion derived from the energy - loss function .
acoustic plasmonic and optical phononic ( oph ) dispersion curves intersect , which fact suggests their interaction and constitutes the main motivation of the subsequent study of the coupled modes .
[ h ! ] _ coupled plasmon - phonon modes. _ we treat the problem of coupled plasmon - phonon oscillations within the _ dynamic _ ( frequency - dependent ) linear - response theory : self - consistently , ions are driven by an externally applied ac electric field and by the coulomb field of moving electrons , and in turn , electrons move under the action of the external field and the field of moving ions .
we consider an infinite two - dimensional ( 2d ) crystal lying in the @xmath1 plane .
the 2d lattice vectors are denoted by @xmath2 while the position of the @xmath3-th atom within the unit cell is @xmath4 .
a weak external potential of the form @xmath5 is applied to the system , where @xmath6 is the 3d position coordinate vector and @xmath7 is the 2d wave - vector .
we seek the response _ including the ionic oscillations _ around their equilibrium positions with the displacements given by @xmath8 with 3d vectors @xmath9 .
the total coulomb potential in the system is @xmath10 where @xmath11 is the ground - state coulomb potential and @xmath12 is its first - order perturbation .
the force experienced by the @xmath3-th ion in the @xmath2-th unit cell is @xmath13 where @xmath14 is the charge of the @xmath3-th ion within the unit cell and @xmath15 is the total coulomb potential minus the self - interaction of the @xmath16-th ion @xmath17 expansion of eq .
( [ eff ] ) to the first order in the perturbation gives @xmath18 since @xmath19 where @xmath20 is the ground - state electron particle - density and @xmath21 is the equilibrium ions potential @xmath22 we can write for the force acting on the @xmath3-th ion in the @xmath23-th cell @xmath24 + \nabla \delta \phi^\textit{eff}_{\alpha\0v}(\rv,\omega ) \right\}_{\rv=\bv_\alpha } , \end{split } \label{fff}\ ] ] where the corresponding 0-th order term has been set to zero because ions are in their equilibrium positions in the crystal s ground - state .
the electronic response is governed by the equation @xmath25 where @xmath26 is the ionic displacement bare potential and @xmath27 is the nonlocal dielectric function of the ideal crystal .
based on eqs .
( [ f ] ) and ( [ b1 ] ) - ( [ bi ] ) , a rather lengthy algebra , which we have included in the appendix , leads to the following expression for the force @xmath28 e_{\beta k } , \end{split } \label{forced}\ ] ] where @xmath29 are the so called dynamic matrices of the conventional dfpt @xcite and @xmath30 \phi^\textit{ext}(\gv+\qv , z,\omega ) \right|_{z=0 } , \end{split}\ ] ] @xmath31 @xmath32 where @xmath33 is the interacting - particles density - response function of the ideal crystal , @xmath34 is the area of the unit cell , @xmath35 , \label{y}\ ] ] and @xmath36 is the unit vector in the @xmath0 direction . in eq .
( [ forced ] ) , the first two terms are due to the dynamically screened external force in the ideal crystal and the third term is the statically screened restoring force of the displacement of the ions .
the fourth term contains all the effects responsible for dynamic electron - phonon interaction . obviously , with the neglect of the latter ( @xmath37 in the fourth term ) , eq .
( [ forced ] ) reduces to the conventional static dfpt case @xcite . with the use of eq .
( [ disp ] ) , newton s second law gives for the @xmath3-th nucleus at the @xmath23-th unit cell @xmath38 equations ( [ forced ] ) - ( [ nl ] ) form a @xmath39 system of linear equations for @xmath40 unknowns @xmath41 , @xmath42 , where @xmath43 is the number of atoms in an elementary unit cell . energy absorbed by the unit cell of the lattice per unit time is @xmath44 _ calculations and results. _ we excite the system with the external potential @xmath45 and solve for the amplitudes of the oscillations @xmath9 using the above formalism with the random - phase approximation to the density - response matrix @xmath33 . in fig .
[ exyz ] , the amplitude of ionic oscillations is plotted as a function of the frequency for three values of the wave - vector .
the @xmath46-vector dependence of the @xmath1-polarized coupled excitation is strongly influenced by that of the apl and @xmath1-polarized oph , while the former is very different from the both latter .
two coupled modes originated from the @xmath1-polarized oph and apl are prominent .
first of them has a strong @xmath46-dispersion , while the second is bound in the vicinity of @xmath1-polarized oph .
both modes are blue - shifted compared with the apl and oph , respectively . for larger @xmath46 ( upper panel of fig .
[ exyz ] ) , the frequency of apl becomes too high for ions to follow the oscillations , leading to the coupled mode convergence to the apl .
the @xmath0-polarized mode remains practically non - dispersive and , acquiring a finite but small line - width , is pinned at the position of the corresponding oph . in fig .
[ w ] , we plot the energy absorbed by the unit cell of the lattice per unit time .
the remarkable feature in this figure is that , depending on the frequency range , the lattice either receives the energy ( positive @xmath47 ) or gives it away to the electronic subsystem ( negative @xmath47 ) . we anticipate that this phenomenon will be experimentally observable in two - terminal suspended graphene experiments .
for example , yien _ et al .
_ @xcite recently demonstrated the ability to distinguish electronic from phononic heat conduction in a self - heated suspended device .
the associated analysis of electron - phonon scattering did not , however , include plasmonic effects , which would be observable at moderate temperatures and under ac fields near the resonances predicted here .
it must be also noted that thorough understanding of apl - phonons interactions is particularly important in the field of superconductivity @xcite .
figure [ cpldisp ] , right panel , shows the @xmath46-vector dispersion derived from the ions oscillations amplitude dependence on the frequency ( left panel ) .
we conclude , that at smaller @xmath46-vectors , the acoustic - like linear dispersion of the coupled mode is lost , indicating that in this regime apl in graphene is an artefact of the frozen - lattice approximation .
the second branch of the coupled mode remains close and above the @xmath1-polarized oph , varying non - monotonically and eventually converging to the latter . in conclusion , we have implemented a fully dynamic ( frequency - dependent density - functional perturbation theory ) calculation of coupled electron - lattice oscillations in graphene .
the coupled mode behaves quite differently from the individual phonon and acoustic plasmon modes , previously known in graphene , and the former replaces the two latter , as acoustic plasmons and phonons do not exist in graphene by themselves , but they constitute a unified excitation of the electronic and ionic subsystems .
the coupling provides a mechanism for the transfer of energy between the electronic subsystem and the lattice , which is shown to go in both directions depending on the frequency range . from this
, promising pathways of tunable heating and cooling of the lattice by the electronic subsystem can be clearly previewed .
v.u.n . acknowledges the support from national science council , taiwan , grant no .
100 - 2112-m-001 - 025-my3 . t.s.f . acknowledges the support of the us office of naval research ( award # n000141211006 , pm : dr .
mark spector ) .
v.u.n .
and t.s.f .
are grateful for the hospitality of qatar environment and energy research institute , qatar foundation , qatar .
17ifxundefined [ 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 10.1021/nl803395z [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.87.041405 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.045113 [ * * , ( ) ] @noop `` , '' ( ) , link:\doibase 10.1103/physrevb.70.205403 [ * * , ( ) ] in @noop _ _ , ( , ) pp .
link:\doibase 10.1038/nphys477 [ * * , ( ) ] @noop link:\doibase 10.1017/s0305004100016108 [ * * , ( ) ] link:\doibase 10.1103/physrevb.45.13244 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.241411 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.28.681 [ * * , ( ) ] link:\doibase 10.1103/physrevb.71.121106 [ * * , ( ) ] |
we would like to thank ulrich hansmann and thomas neuhaus for numerous useful discussions .
further , we are indebted to philippe backouche , norbert schultka , and unix_guys for valuable help .
t.c . was supported by tubitak of turkey and likes to thank joe lannutti and scri for the warm hospitality extended to him .
our simulations were performed on the scri cluster of fast risc workstations . | we present a recursive procedure to calculate the parameters of the recently introduced multicanonical ensemble and explore the approach for spin glasses .
temperature dependence of the energy , the entropy and other physical quantities are easily calculable and we report results for the zero temperature limit . our data provide evidence that the large @xmath0 increase of the ergodicity time is greatly improved .
the multicanonical ensemble seems to open new horizons for simulations of spin glasses and other systems which have to cope with conflicting constraints .
the theoretical understanding of spin glasses , for a review see @xcite , has remained a great challenge . in particular the low temperature limit leaves many open questions about the effects of disorder and frustration .
for instance , it has remained controversial whether parisi s @xcite mean field theory provides the appropriate description for 3d spin glasses .
the attractive alternative is the droplet model @xcite , which in turn is equivalent to a one parameter scaling picture @xcite .
the simplest spin glass system to study such questions numerically is the edwards - anderson model . in its ising version
it is described by the hamiltonian @xmath1 where the sum goes over nearest neighbours and the exchange interactions @xmath2 between the spins @xmath3 are quenched random variables . in our investigation
we impose the constraint @xmath4 for each realization .
recent simulations @xcite of the 3d model in a magnetic field support the mean field picture .
however , one may argue that sufficiently low temperatures on sufficiently large systems have not been reached . for previous simulations of the edwards - anderson model without magnetic field
see @xcite .
low temperature simulations of spin glasses suffer from a slowing down due to energy barriers . to illustrate the problem , let us consider a simple ferromagnet : the 2d ising model on an @xmath5 lattice .
in figure 1 we give its magnetic probability density versus @xmath6 .
the two distinct branches below the curie temperature are associated with free energy valleys in configuration space , each of which defines a ( pure ) thermodynamic state . at temperatures below the curie point
the ergodicity
time@xmath7 @xmath8 increases exponentially fast with lattice size , asymptotically like @xmath9 $ ] , where @xmath10 is the surface free energy . therefore ,
on large lattices at sufficiently low temperature the simulation of the system will , given a reasonable finite amount of computer time , never tunnel from one phase to the other .
besides for particular problems ( like the determination of the order - order surface free energy ) this lack of tunneling does not pose a major handicap for ising model simulations .
the reason is that the two configuration space valleys are related by the exact symmetry @xmath11 of the hamiltonian .
exploring one valley by means of a simulation yields also all the properties of the other one and , hence , allows to overlook the entire system .
the situation is much more involved for spin glasses . for low enough temperature
the system is supposed to split of into many thermodynamic states , separated by similar tunneling barriers as the two pure states of the ising model .
however , unlike in the ising model the states are not related to each other by a symmetry of the hamiltonian .
rather they appear because of accidental degeneracy which in turn occurs because of randomness and frustration of the system . for computer simulations
this means that one would like to explore many independent configuration space valleys while keeping track of their relative weights .
the groundstate energies associated with these valleys may or may not be degenerate , but it should be noted that even if they are not degenerate , tunneling between the valleys would still be governed by the energy barriers .
the physics of these barriers is far less well understood as in the ferromagnetic case . as detailed finite size
scaling ( fss ) studies do not exist , it is unclear to us to what extent these barriers depend on the system size , whereas the temperature dependence has been investigated @xcite .
we use the notation bifurcation temperature ( bifurcation point ) for the temperature at which the spin glass configuration space ( phase transition or not ) begins to split of into a number of valleys which are well separated by energy barriers . in the present paper
we suggest that the increase of @xmath12 can be reduced to a fairly decent power law by performing a simulation which covers in a single ensemble a whole temperature range from well above to far below the bifurcation point .
the appropriate formulation is provided by a generalization of the multicanonical ensemble @xcite . to test the approach we have performed multicanonical simulations of the 2d ising ferromagnet and then of the 2d edward - anderson ising spin glass model .
we concentrate on ground state properties what is a kind of worst case scenario for the performance of the multicanonical simulation .
it should be noted that figure 1 does rely on a multicanonical simulation , what explains the slight asymmetry between the two branches . ever since the pioneering paper by metropolis et al.@xcite most mc simulations concentrated on importance sampling for the canonical gibbs ensemble .
it has always been well - known , for instance @xcite , that one is allowed to choose phase - space points according to any other probability distribution , if it is convenient .
however , a systematic reasoning for a better than canonical choice has rarely been put forward@xmath7 .
it is our suggestion that in a large class of situations , in particular those where canonical simulations face severe ergodicity problems , it is more efficient to reconstruct the gibbs ensemble from a simulation of a multicanonical ensemble @xcite than simulating it directly . in canonical simulations configurations
are weighted with the boltzmann factor @xmath13 . here
@xmath14 is the energy of the system under consideration .
the resulting canonical probability density is @xmath15 where @xmath16 is the spectral density . in order of increasing severity problems with canonical spin glass simulations
are : _ i _ ) simulations at many temperatures are needed to get an overview of the system . _ ii _ ) the normalization in equation ( 2 ) is lost .
it is tedious to calculate important physical quantities like the free energy and the entropy .
_ iii _ ) the low temperature ergodicity time @xmath8 diverges fast with lattice size ( either exponentially or with a high power law ) .
the relative weights of pure states can only be estimated for small systems .
let us choose an energy range @xmath17 and define for a given function @xmath18 the function @xmath19 by the recursion relation ( with the hamiltonian ( 1 ) the energy changes in steps of 4 ) @xmath20 e,\ \alpha(e_{\max } ) = 0 .
\eqno(3)\ ] ] the multicanonical ensemble @xcite is then defined by weight factors @xmath21 , \eqno(4)\ ] ] where @xmath18 is determined such that for the chosen energy range the resulting multicanonical probability density is approximately flat : @xmath22 in the present study we take @xmath23 ( @xmath24 for @xmath25 ) and @xmath26 the ground state energy of the considered spin glass realization . a multicanonical function @xmath18 can be obtained via recursive mc calculations .
one performs simulations @xmath27 , @xmath28 , which yield probability densities @xmath29 with medians @xmath30 . for @xmath31
the probability density @xmath29 becomes unreliable due to insufficient statistics , caused by the exponentially fast fall - off for decreasing @xmath14 .
we start off with @xmath32 and @xmath33 .
the recursion from @xmath34 to @xmath35 reads @xmath36\ { \rm for}\ e^n_{\rm median } > e \ge e^n_{\min } ; \cr \beta^{n+1 } ( e^n_{\min})\ { \rm for}\ e < e^n_{\min } .
\cr } \eqno(6)\ ] ] here the @xmath37 simulation may be constrained to @xmath38 by rejecting all proposal with energy @xmath39 , but one has to be careful with such bounds in order to maintain ergodicity .
the recursion is stopped for @xmath40 with @xmath41 being groundstate . starting with this simple approach
we have explored several more sophisticated variants .
considerable speed - ups and gains in stability could be achieved .
the cpu time spent to estimate the multicanonical parameters was 10% to 30% of the cpu time spent for simulations with the final set .
once the functions @xmath18 and @xmath19 are fixed , the multicanonical simulation exhibits a number of desireable features : _ i _ ) by reweighting with @xmath42 $ ] the canonical expectation values @xmath43 where @xmath44 is the partition function , can be reconstructed for all @xmath45 in an entire range @xmath46 . here
@xmath47 and @xmath48 follow from the requirement @xmath49 , and @xmath50 follows from ( 7 ) with @xmath51 . with our choice @xmath52 and @xmath26 groundstate , @xmath53 and
@xmath54 follows .
_ ii _ ) the normalization constant @xmath55 in equation ( 5 ) follows from @xmath56 , where @xmath57 is the total number of spin variables .
this gives the spectral density and allows to calculate the free energy as well as the entropy .
_ iii _ ) we conjecture that the slowing down of canonical low temperature spin glass simulations becomes greatly reduced . for the multicanonical ensemble it can be argued @xcite that single spin updates cause a 1d random walk behaviour of the energy @xmath14 .
as @xmath58 , one needs @xmath59 updating steps to cover the entire ensemble . for first order phase transition
the observed slowing down was only slightly worse than this optimal behaviour .
our present mc data show more drastic modifications for spin glass simulations .
to quantify our discussion of the slowing down , we have to define the ergodicity time @xmath12 . roughly speaking it is the cpu time needed to collect independent configurations of the system , for instance groundstates when this is the main interest .
regarding the definition of @xmath12 , an @xmath0-independent over - all factor is free .
we define @xmath12 as the average number of sweeps needed to move the energy from @xmath60 to @xmath61 and back .
a sweep is defined by updating each spin on the lattice once ( in the average ) .
it should be noted that the number of `` tunneling '' events with respect to the ergodicity time gives a lower bound on the number of independent groundstates samples .
this follows from another trivial , but remarkable property of the multicanonical ensemble : each time a sweep is spent at @xmath62 , the memory of the previous markov chain is lost entirely , and a truly independent new series of configurations follows .
the condition @xmath63 is appropriate to substitute for the somewhat too strict constraint of a sweep at @xmath64 . as a corollary : with a disordered starting configuration the multicanonical ensemble is immediately in equilibrium . as an exercise and to check our code on exact results , we performed a multicanonical simulation of the 2d ising model with @xmath65 .
we kept the time series of two million sweeps and measurements on a @xmath66 lattice and verified that the finite lattice specific heat results of ferdinand and fisher @xcite are well reproduced .
no difficulties are encountered with the multicanonical ensemble when crossing the phase transition point . to explore the possibility of zero temperature entropy calculations
, we used @xmath67 as input and obtained @xmath68 for the total groundstate entropy .
this corresponds to an estimate of @xmath69 groundstates , _
i.e. _ within statistical errors we are in agreement with two . using @xmath70 and a time series of four million sweeps on a @xmath5 lattice we obtained @xmath71 for the number of groundstates .
figure 1 is obtained from the simulation of this lattice .
after this test we turned to the 2d edwards - anderson spin glass . on lattices of size
@xmath72 , 12 , 24 and 48 we performed multicanonical simulations . up
to @xmath73 we investigated ten different realizations per lattice and , due to cpu time constraints , we considered only five realizations for the @xmath74 lattice .
the multicanonical energy distribution for one of our @xmath74 realizations is depicted in figure 2 .
the fall - off for @xmath75 is like that of the canonical distribution at @xmath76 . for @xmath77 an impressive flatness
( about 800 energy entries on the lattice under consideration ) is quickly achieved by the recursion ( 6 ) .
close to the groundstate some fluctuations are encountered on which we comment elsewhere @xcite .
as it is not obvious from the scale of the figure , we like to remark that the groundstate is not the state with the lowest number of entries , but a state close to it .
further , it should be understood that deviations from the desired constant behaviour ( 5 ) do only influence the statistical error bars , but not the estimates themselves .
therefore , such deviations do not pose problems as long as they can be kept within reasonable limits of approximately one order of magnitude .
the table gives an overview of some of our numerical results .
the ergodicity time @xmath12 is as defined above , and for @xmath78 we take @xmath79 , where it should be noted that due to our computational procedure @xmath18 is a noisy function .
the reported values and their error bars are obtained by combining the results from the different realizations , which enter with equal weights . in figure 3
we plot the ergodicity time versus lattice size @xmath0 on a log log scale .
the data are consistent with a straight line fit ( @xmath80 denotes the goodness of fit ) , which gives the finite size behaviour @xmath81 in cpu time this corresponds to a slowing down @xmath82 .
it should be remarked that a fit of form @xmath83 results in a completely unacceptable goodness of fit @xmath84 .
still , the behaviour ( 8) is by an extra volume factor worse than the close to optimal performance we hoped for . the reasons will be considered elsewhere @xcite .
._overview of some results . for the data points marked with
* the statistics for different realizations varies somewhat and average values are given .
_ [ cols="^,^,^,^",options="header " , ] we estimate the infinite volume groundstate energy and entropy from fss fits of the form @xmath85 .
the entropy fit is depicted in figure 4 , and the energy fit looks similar .
our energy estimate is @xmath86 , consistent with the previous mc estimate @xcite @xmath87 as well as with the transfer matrix result @xcite @xmath88 .
our entropy estimate @xmath89 is also consistent with the mc estimate @xcite @xmath90 , but barely consistent with the more accurate transfer matrix result @xcite @xmath91 .
still , a larger statistical sample and a careful study of systematic error sources would be needed to claim that there is a significant discrepancy .
our results could be improved by exploiting the high temperature expansion of the entropy , as it was done in @xcite . however
, this would be against the spirit of this paper which is to explore the possibilities and limits of multicanonical spin glass simulations .
it is not entirely straightforward to compare multicanonical and standard simulations .
for instance autocorrelation times of multicanonical simulations come out short due to the triviality that the simulation spends most of its time at rather small effective @xmath92 values .
therefore , autocorrelations are not well suited .
our ergodicity time , the average number of sweeps to find truly independent groundstates , is a more useful quantity . when relying on it , _
i.e _ concentrating on groundstate properties , we should remember that this pushes the multicanonical simulation to its extreme .
this way of calculating groundstate properties gives for free all properties in - between , from @xmath76 on .
if a phase transition occurs , its study is also included ( we noticed no particular difficulties in the 2d ising model ) .
for instance in the 3d ising spin glass the canonical slowing down at @xmath93 @xcite is already worse than the one given by our equation ( 8) . although the slowing down ( 8) is severe , it seems to provide an important improvement when compared with the slowing down which canonical simulations encounter for temperatures below the bifurcation temperature . for @xmath94 canonical simulations @xcite are unable to equilibrate the systems at the @xmath78 values reported in our table since the relaxation time is by far too long .
presumably due to this fact the literature focused on other questions and does not provide detailed fss investigation of relaxation times . at the present stage we did not want to spent cpu time on extensive canonical simulations
. therefore , detailed comparisons are not possible . however , a rough estimate of the canonical ergodicity time may be @xcite @xmath95 the scaling with @xmath0 may then be hidden in the @xmath0 dependence of our @xmath78 values , which is argued to be divergent like @xmath96 , and this line of reasoning gives a slowing down of the canonical algorithm like @xmath97 . with @xmath98 ( our @xmath73 case )
one gets a canonical ergodicity time of order @xmath99 when the missing constant is assumed to be of order one , whereas our @xmath100 is about @xmath101 .
already for moderately large @xmath45-values , a doubtful procedure has emerged in the literature @xcite : realizations which ( according to a well defined criterium ) can not be equilibrated are simply omitted .
as these realizations are as legitimate members of the statistical ensemble as any other choice of the quenched random variables @xmath102 , an uncontrollable bias is inevitably introduced .
in contrast to this , we were able to equilibrate every single realization encountered .
when one is only interested in groundstate properties , minimization algorithms have to be considered . as a method simulated annealing @xcite stands out because of its generality , although there are more efficient algorithms for special cases , which should be used when appropriate . in simulated annealing
the results depend on the cooling rate @xmath103 . for our model
the behaviour @xmath104 with @xmath105 ( @xmath106 fixed ) is indicated @xcite . to find a true groundstate
, one has to reduce @xmath107 $ ] to the order @xmath108 .
assuming that the constant @xmath109 is volume independent ( only the lattice size @xmath110 was considered in @xcite ) , this translates to # @xmath111 , far worse than our equation ( 8) .
this result is kind of amazing as the multicanonical ensemble has eliminated directed cooling and is nevertheless more efficient . if one does not insist on true groundstates one
can then relax the condition @xmath112
\sim 1/v$ ] .
for instance , any behaviour @xmath112 \to 0 $ ] with @xmath113 would still give the correct density , and simulated annealing would slow down less dramatic . on the other hand , this would also imply a less stringent multicanonical simulation .
a comparison with the cluster - replica mc algorithm @xcite is even less clear cut .
the obtained estimates of the groundstate energy and entropy are in accuracy similar to ours .
as one has to simulate many replica at many @xmath45-values a direct comparison is impossible . clearly , the results reported on slowing down are more promising than ours for the ergodicity time . on the other hand , to our knowledge the replica mc approach has never been applied to the 3d edwards - anderson model , and one may well encounter difficulties . in contrast , 3d multicanonical simulations are straightforward .
in fact the dimension is just a parameter in all our computer programs and we have already carried out various 3d test runs .
let us address the question of algorithms from a more general perspective . with a metropolis type implementation
our optimum performance will be bounded from below by a slowing down @xmath114 in cpu time , and we are outperformed by any algorithm which can do better than this . for a number of important , but often highly specialized , applications such better algorithms exist and should be used .
the main advantage of the multicanonical ensemble is its generality . with this respect
our method resembles simulated annealing @xcite , while clearly avoiding some of its disadvantages .
most notably , the relationship to the equilibrium canonical ensemble remains exactly controlled .
finally , it should be stressed that the multicanonical ensemble is an ensemble and not an algorithm .
one may find better algorithms than conventional metropolis updating to simulate this new ensemble .
to try a combination with cluster algorithms is an attractive idea .
our results make clear that the multicanonical approach is certainly a relevant enrichment of the options one has with respect to spin glass simulations .
the similarities of spin glasses to other problems with conflicting constraints @xcite suggest that multicanonical simulations may be of value for a wide range of investigations : optimization problems like the travelling salesman , neural networks , protein folding , and others .
multicanonical simulations of the 3d edwards - anderson model may eventually shed new light on the question whether the model exhibits mean field like behaviour or some kind of droplet picture applies @xcite .
it seems that previous numerical work remained somewhat inconclusive as sufficiently low temperatures could not be reached without destroying the thermodynamic equilibrium . |
nowadays , there exists a growing interest to astrophysical objects , where the strong magnetic fields with the strength @xmath8 can be generated @xmath9 gs , @xmath10 is the electron mass , @xmath11 is the elementary charge . ]
is the so - called critical field value ) .
the influence of a strong external field on quantum processes is interesting because it catalyses the processes , it changes the kinematics and it induces new interactions .
it is especially important to investigate the influence of external field on the loop quantum processes where only electrically neutral particles in the initial and the final states are presented , such as neutrinos , photons and hypotetical axions , familons and so on .
the external field influence on these loop processes is provided by the sensitivity of the charged virtual fermion to the field and by the change of the photon dispersion properties and , therefore , the photon kinematics .
the research of the loop processes of this type has a rather long history .
the two - vertex loop processes ( the photon polarization operator in an external field , the decays @xmath12 , @xmath13 and so on ) were studed in the papers @xcite .
the general expression for the two - vertex loop amplitude @xmath14 in the homogeneous magnetic and in the crossed field was obtaned in the paper @xcite , where all combinations of scalar , pseudoscalar , vector and axial - vector interactions of the generalized currents @xmath15 with fermions were considered . a loop process with three vertices is also intresting for theoreticians .
for example , the photon splitting in a magnetic field @xmath16 is forbidden in vacuum .
the review @xcite and the recent papers @xcite were devoted to this process .
one more three - vertex loop process is the conversion of the photon pair into the neutrino pair , @xmath17 .
this process is interesting as a possible channel of stellar cooling .
a detailed list of references on this process can be found in our paper @xcite .
it is well - known ( the so - called gell - mann theorem @xcite ) , that for massless neutrinos , for both photons on - shell and in the local limit of the standard - model weak interaction , the process @xmath17 is forbidden .
because of this , the four - vertex loop process with an additional photon @xmath18 was considered by some authors . in spite of the extra factor @xmath19
, this process has the probability larger than the two - photon process .
the process @xmath18 was studied both in vacuum ( from the first paper @xcite to the recent refs .
@xcite ) , and under the stimulating influence of a strong magnetic field @xcite .
so , the calculation of the amplitude of the @xmath0-vertex loop quantum processes ( @xmath5 , @xmath20 , the axion and familon processes @xmath21 , @xmath22 and so on ) in a strong magnetic field is important , because these results can be useful for astrophysical applications .
the paper is organized as follows .
a general analysis of the @xmath0-vertex one - loop process amplitude in a strong magnetic field is performed in section 2 .
the amplitude , in which the one vertex is of a general type ( scalar @xmath23 , pseudoscalar @xmath24 , vector @xmath25 or axial - vector @xmath26 ) , and the other vertices are of the vector type ( contracted with photons ) , is calculated in section 3 .
this amplitude is the main result of the paper .
the analitical expressions for the amplitudes of the processes @xmath5 and @xmath7 are presented in sections 4 and 5 .
we use the effective lagrangian for the interaction of the generalized currents @xmath27 with electrons in the form : @xmath28 j_i , \label{eq : l}\end{aligned}\ ] ] where the generic index @xmath29 numbers the matrices @xmath30 , e.g. @xmath31 , @xmath27 is the corresponding quantum object ( the current or the photon polarisation vector ) , @xmath32 are the coupling constants . in particular , for the electron - photon interaction we have @xmath33 .
a general amplitude of the process , corresponding to the effective lagrangian ( [ eq : l ] ) , is described by fig .
[ fig : loop1 ] . in the strong field limit , after integration over the coordinates , the amplitude takes the form @xmath34 where @xmath35 , @xmath36 is the asymptotic form of the electron propagator in the limit @xmath37 , @xmath38 @xmath39,\ ] ] @xmath40 @xmath41 , @xmath42 , @xmath43 is the dimensionless field tensor , @xmath44 is the dual tensor , and the indices of the four - vectors and tensors standing inside the parentheses are contracted consecutively , e.g. @xmath45 . as is seen from eq .
( [ eq : m2 ] ) , the amplitude depends only on the longitudinal components of the momenta , if the magnetic field strength is the maximal physical parameter @xmath46 .
let the vertices @xmath47 are of the vector type , and the vertex @xmath48 is arbitrary . it can be shown that in the limit @xmath49 , for odd numbers of vertices , only the @xmath50 amplitude grows linearly with the magnetic field strength , while for even numbers of vertices the linear growth takes place only in the amplitudes @xmath2 , @xmath3 and @xmath4 . it should be noted , that in the amplitude ( [ eq : m2 ] ) the projecting operators @xmath51 separate out the photons of only one polarization @xmath52 of the two possible ( in adler s notation @xcite ) @xmath53 as can be deduced from the corresponding analysis , the calculation of any type of the amplitude can be reduced to the evaluation of the scalar integrals @xmath54 notice that the use of the standart method of feynman parametrization in calculation of the integrals ( [ eq : s1 ] ) can be non - optimal , because the number of integrations grows . for example , if @xmath55 , the double integral ( [ eq : s1 ] ) is transformed into the integral over the two feynman variables . if @xmath56 , the double integral ( [ eq : s1 ] ) is transformed into the integral over the three feynman variables and so on . here
we suggest another way . by integrating ( [ eq : s1 ] ) over @xmath57
, we obtain @xmath58 re \ ,
\bigg \ { \underset{k \ne i , l } { \prod \limits^{n}_{k=1 } } \frac{1}{y_{ilk } } \bigg \ } , \label { eq : s4}\end{aligned}\ ] ] where @xmath59 the function @xmath60 is defined by the expressions @xmath61 and it has the asymptotics @xmath62 @xmath63
let us apply the results obtained to the calculation of some quantum processes .
for the amplitude of the process @xmath5 in the framework of the model with the effective @xmath6 coupling of a scalar type we obtain from eqs .
( [ eq : m2 ] ) , ( [ eq : s1 ] ) , ( [ eq : s4 ] ) @xmath64 + q^2_{1\mprl } q^2_{2\mprl } q^2_{3\mprl } } \times \nonumber \nonumber \\ & \times & \ , \left \ { \left [ q^2_{1\mprl } q^2_{3\mprl } - 2m^2 ( q^2_{3\mprl } + q^2_{1\mprl } - q^2_{2\mprl})\right ] h \left ( \frac{q_{1\mprl}^2}{4m^2 } \right ) \right . \ , + \nonumber \nonumber \\ & + & \ , \left [ q^2_{2\mprl } q^2_{3\mprl } - 2m^2 ( q^2_{3\mprl } + q^2_{2\mprl } - q^2_{1\mprl})\right ] h \left ( \frac{q_{2\mprl}^2}{4m^2 } \right ) \ , + \nonumber \nonumber \\ & + & \ , \left .
q^2_{3\mprl}(4m^2 -q^2_{3\mprl } ) h \left ( \frac{q_{3\mprl}^2}{4m^2 } \right ) - 2q^2_{3\mprl}(q_1q_2)_{\mprl } \ , \right \ } , \label{eq : ms3}\end{aligned}\ ] ] where @xmath65 is the effective @xmath6 coupling constant in the left - right - symmetric extension of the standard model , @xmath66 is the small mixing angle of the charged @xmath67 bosons , @xmath68 $ ] is fourier transform of the scalar neutrino current , @xmath69 is the neutrino pair momentum . substituting the photon polarization vector @xmath70 from eq .
( [ eq : vect ] ) into ( [ eq : ms3 ] ) and using ( [ eq : s5 ] ) and ( [ eq : s6 ] ) , we obtain the asymptotics : * at low photon energies , @xmath71 @xmath72\ , \sqrt{q_{1\mprl}^2 q_{2\mprl}^2 } ; \label{eq : m<}\end{aligned}\ ] ] * at high photon energies , @xmath73 , in the leading log approximation : @xmath74\ , \frac{1}{\sqrt{q_{1\mprl}^2 q_{2\mprl}^2}}\ , \ln \frac{\sqrt{q_{1\mprl}^2 q_{2\mprl}^2}}{m^2}. \label{eq : m>}\end{aligned}\ ] ] these expressions coincide with the results obtained in the paper @xcite .
the process of this type , where one initial photon is virtual , namely , the photon conversion into neutrino pair on nucleus was considered , in the framework of the standard model , in the papers @xcite
. this process can be studied by using the amplitude of the transition @xmath75 , which can be obtained from eq .
( [ eq : m2 ] ) in the form : @xmath76 \times \nonumber \nonumber \\ & \times & \frac{1}{d } \left \ { i_4(q_{1\mprl},q_{2\mprl},q_{3\mprl } ) + i_4(q_{2\mprl},q_{1\mprl},q_{3\mprl } ) + i_4(q_{1\mprl},q_{3\mprl},q_{2\mprl } ) \right \ } , \label{eq : mva4}\end{aligned}\ ] ] where @xmath77 are the vector and axial - vector constants of the effective @xmath6 lagrangian of the standard model , @xmath78 , here the upper signs correspond to the electron neutrino , and lower signs correspond to the muon and tau neutrinos ; @xmath79 $ ] is the fourier transform of the neutrino current ; @xmath80 is the neutrino pair momentum ; @xmath81 the formfactor @xmath82 is expressed in terms of the integrals ( [ eq : s1 ] ) , ( [ eq : s4 ] ) @xmath83 s_4 ( q_{1 \mprl } , q_{1 \mprl}+q_{2 \mprl } , q_{4 \mprl},0 ) .
\end{aligned}\ ] ] using the asymptotics of the functions @xmath60 , we obtain * at low photon energies , @xmath84 @xmath85 , \label{eq : i_4_low}\end{aligned}\ ] ] + which is in agreement with the result of the paper @xcite ; * at high photon energies , @xmath86 , in the leading log approximation we obtain : @xmath87 \times \nonumber \nonumber \\ & & \times \frac{1}{q^2_{1 \mprl}q^2_{2 \mprl}q^2_{3 \mprl}q^2_{4 \mprl } } \ln \frac{\sqrt{q_{1 \mprl}^2 q_{2 \mprl}^2 q_{3 \mprl}^2}}{m^3}. \label{eq : i_4_hig}\end{aligned}\ ] ] to the best of our knowledge , this result is obtained for the first time .
we have obtained the general expressions ( [ eq : ms3 ] ) and ( [ eq : mva4 ] ) for the amplitudes of the processes @xmath5 ( in the framework of the model with the effective @xmath6 coupling of a scalar type ) and @xmath7 ( in the framework of the standard model ) for arbitrary energies of particles .
a comparison with the existing results has been performed .
99 w .- y .
tsai , phys . rev . *
d 10 * , 2699 ( 1974 ) .
a. e. shabad , tr .
nauk sssr * 192 * , 5 ( 1988 ) . v. v. skobelev , zh .
fiz . * 108 * , 3 ( 1995 ) [ jetp * 81 * , 1 ( 1995 ) ] .
a. a. gvozdev , n. v. mikheev , l. a. vassilevskaya , phys . rev . *
d 54 * , 5674 ( 1996 ) .
a. n. ioannisian , g. g. raffelt , phys . rev . *
d 55 * , 7038 ( 1997 ) .
borovkov , a. v. kuznetsov , n. v. mikheev , yad . fiz .
* 62 * , 1714 ( 1999 ) [ phys . at
. nucl . * 62 * , 1601 ( 1999 ) ] .
v. o. papanian , v. i. ritus , tr . fiz
nauk sssr * 168 * , 120 ( 1986 ) .
s. l. adler , c. schubert , phys .
lett . * 77 * , 1695 ( 1996 ) . v. n. baier ,
a. i. milstein and r. zh .
shaisultanov , phys .
. lett . * 77 * , 1691 ( 1996 ) . v. n. baier , a. i. milstein and r. zh .
shaisultanov , zh .
fiz . * 111 * , 52 ( 1997 ) m. v. chistyakov , a. v. kuznetsov , n. v. mikheev , phys . lett .
* b 434 * , 67 ( 1998 ) .
a. v. kuznetsov , n. v. mikheev , m. v. chistyakov , yad . fiz .
* 62 * , 1638 ( 1999 ) [ phys . at
. nucl . * 62 * , 1535 ( 1999 ) ] .
a. v. kuznetsov , n. v. mikheev , d. a. rumyantsev , yad . fiz . * 66 * , ( 2003 ) ( in press ) .
m. gell - mann , phys .
* 6 * , 70 ( 1961 ) .
nguen van hieu , e. p. shabalin , zh .
* 44 * , 1003 ( 1963 ) [ jetp * 17 * , 681 ( 1963 ) ] .
d. a. dicus , w. w. repko , phys .
lett . * 79 * , 569 ( 1997 ) .
d. a. dicus , c. kao , w. w. repko , phys . rev .
* d 59 * , 013005 ( 1999 ) .
a. abada , j. matias , r. pittau , phys . rev . * d 59 * , 013008 ( 1999 ) .
a. abada , j. matias , r. pittau , nucl . phys . * b 543 * , 255 ( 1999 ) .
a. abada , j. matias , r. pittau , phys . lett . * b 450 * , 173 ( 1999 ) .
m. loskutov , v. v. skobelev , theor .
* 70 * , 303 ( 1987 ) . v. v. skobelev , zh .
. fiz . * 120 * , 786 ( 2001 ) [ jetp * 93 * , 685 ( 2001 ) ] .
a. v. kuznetsov and n. v. mikheev , pisma v zhetf * 75 * , 531 ( 2002 ) [ jetp letters * 75 * , 441 ( 2002 ) ] . s. l. adler , _ ann
_ * 67 * ( 1971 ) 599 . | a general analysis of the @xmath0-vertex loop amplitude in a strong magnetic field is performed , based on the asymptotic form of the electron propagator in the field . as an example , the photon - neutrino processes are considered , where one vertex in the amplitude is of a general type , and the other vertices are of the vector type .
it is shown , that for odd numbers of vertices only the @xmath1 amplitude grows linearly with the magnetic field strength , while for even numbers of vertices the linear growth takes place only in the amplitudes @xmath2 , @xmath3 and @xmath4 .
the general expressions for the amplitudes of the processes @xmath5 ( in the framework of the model with the effective @xmath6 coupling of a scalar type ) and @xmath7 ( in the framework of the standard model ) for arbitrary energies of particles are obtained . a comparison with existing results
is performed .
_ talk presented at the 12th international seminar + `` quarks-2002 '' , + valday and novgorod , russia , june 1 - 7 , 2002 _ |
multiple orthogonal polynomial ( mop ) ensembles @xcite form an extension of the more familiar orthogonal polynomial ( op ) ensembles @xcite .
the latter appear as the eigenvalue distributions of unitary random matrix ensembles . a major topic of interest in random matrix theory deals with the universality conjecture .
this conjecture claims that , as the size of the matrices tends to infinity , the local eigenvalue statistics of a random matrix ensemble converge to universal limits , i.e. , limits that are independent of the precise probability distribution on the matrices .
the limits only depend on macroscopic eigenvalue characteristics and on the symmetry class of the matrix ensemble . for the class of unitary random matrix ensembles
the problem concerning universality can be translated into a problem concerning the associated sequence of orthogonal polynomials ( op ) . by the gaudin - mehta method @xcite
the eigenvalue density function can be written as a determinantal point process .
the relevant correlation kernel is the op kernel , which is also known as the christoffel - darboux kernel .
universality is then expressed as the convergence of this op kernel to a certain universal limiting kernel .
this convergence can be derived by considering the large degree asymptotics of the op .
common limiting kernels include the sine , airy and bessel kernels . for other random matrix ensembles a connection exists with multiple orthogonal polynomials ( mop ) :
the eigenvalues are a determinantal point process with the so - called mop kernel as correlation kernel .
a prime example of such a connection occurs in the unitary random matrix model with external source . by analysis of the mop kernel in this model universality
has been shown and the sine and airy kernels appeared @xcite . however , in a critical case this model also exhibits behavior that does not appear in op ensembles : under certain scaling the mop kernel converges to the so - called pearcey kernel @xcite .
mop ensembles also gave rise to new critical behavior in the papers @xcite .
this is a recurring feature : mop ensembles can exhibit a wider variety of critical phenomena than op ensembles . in this paper
we consider another type of mop ensembles , the so - called angelesco ensembles .
these ensembles do not appear in a natural way as the eigenvalue process of a random matrix ensemble , but the notion of universality still makes sense .
we show that a new kind of critical behavior occurs in angelesco systems , determined by a new kind of universal kernel , that we call the angelesco kernel @xmath8 .
an angelesco system of weights @xcite is a system of @xmath9 weights @xmath10 on the real line such that the supports of the weights are contained in intervals with pairwise disjoint interiors . thus there
exist bounded real intervals @xmath11 such that @xmath12 and @xmath13 is non - negative and integrable on the real line with @xmath14 angelesco ensembles are mop ensembles associated with angelesco systems of weights .
let @xmath15 be a multi - index @xmath16 and define @xmath17 .
the angelesco ensemble with respect to @xmath18 and @xmath15 is the probability measure on @xmath19 with probability density function @xmath20_{j , k=1}^{|\vec{n}| } \cdot \det \left [ f_{j}(x_{k } ) \right]_{j , k=1}^{|\vec{n}|},\ ] ] where @xmath21 is a normalization constant and @xmath22 the angelesco properties - imply that is non - negative for every @xmath23 , and can be non - zero only if @xmath24 of the particles @xmath25 are in @xmath26 for every @xmath27 .
thus in an angelesco ensemble , with probability @xmath28 , @xmath24 particles are located in @xmath26 for every @xmath29 .
the probability density function is a biorthogonal ensemble @xcite , which is a special case of a determinantal point process .
the correlation kernel is @xmath30 where @xmath31 is the @xmath32-th entry of the inverse of the matrix @xmath33 that is , see @xcite : @xmath34_{j , k = 1}^{|\vec{n}|},\ ] ] and for each @xmath35 , @xmath36_{j , k=1}^m\ ] ] where @xmath37 is the @xmath38-point correlation function .
another representation for @xmath39 is by means of a determinant @xmath40 where @xmath41 is the moment matrix . in this paper
we consider angelesco ensembles with respect to two weights @xmath42 and @xmath43 .
also we restrict ourselves to diagonal multi - indices @xmath44 .
our interest is in the local asymptotics of the mop kernel @xmath45 associated to the weights @xmath42 and @xmath43 . by rescaling
if necessary we can assume that @xmath46 $ ] and @xmath47 $ ] for some @xmath48 .
the type of local behavior around a certain point is suggested by the behavior of the limiting mean particle density in that point .
assume that @xmath49 a.e .
on @xmath26 .
then this limiting density only depends on the endpoints of the two intervals @xmath50 and @xmath51 .
depending on the values of @xmath5 and @xmath52 we get a qualitatively different picture .
we can distinguish 3 cases and a number of phase transitions , see figure [ fig:121 ] .
( 0,0 ) ( -.38,-3.8 )
( .5,-3.8 ) ( .5 , .5 )
( -4.8 , .5 ) ( -4.8,0 ) ( 0,0 ) ; ( -5,0 ) ( 1,0 ) ; ( 1,0 ) node[above right ] @xmath5 ; ( 0,-4 ) ( 0,1 ) ; ( 0,1 ) node[above left ] @xmath52 ; ( -.38 , -3.8 )
( 0,0 ) ; ( 0,0 ) circle ( 0.03 ) ; ( -2,0 ) circle ( 0.03 ) ; ( -.25 , -2.5 ) circle ( 0.03 ) ; plot[smooth ] coordinates ( -2.000000,0.000000)(-2.100000,-0.000088)(-2.200000,-0.000671)(-2.300000,-0.002160)(-2.400000,-0.004884)(-2.500000,-0.009107)(-2.600000,-0.015038)(-2.700000,-0.022835)(-2.800000,-0.032620)(-2.900000,-0.044481)(-3.000000,-0.058480)(-3.100000,-0.074654)(-3.200000,-0.093023)(-3.300000,-0.113593)(-3.400000,-0.136355)(-3.500000,-0.161290)(-3.600000,-0.188374)(-3.700000,-0.217572)(-3.800000,-0.248848)(-3.900000,-0.282159)(-4.000000,-0.317460)(-4.100000,-0.354705)(-4.200000,-0.393845)(-4.300000,-0.434831)(-4.400000,-0.477612)(-4.500000,-0.522139)(-4.600000,-0.568361)(-4.700000,-0.616230)(-4.800000,-0.665696)(-4.900000,-0.716712)(-5.000000,-0.769231 ) ; plot [ smooth ] coordinates(-2.000000 , 0.000000)(-1.900019 , -0.000097)(-1.800156 , -0.000820)(-1.700539 , -0.002916)(-1.601312 , -0.007289)(-1.502630 , -0.015026)(-1.404664 , -0.027435)(-1.307603 , -0.046078)(-1.211652 , -0.072827)(-1.117036 , -0.109912)(-1.024000 , -0.160000)(-0.932809 , -0.226266)(-0.843750 , -0.312500)(-0.757135 , -0.423220)(-0.673297 , -0.563820)(-0.592593 , -0.740741)(-0.515402 , -0.961683)(-0.442125 , -1.235866)(-0.373178 , -1.574344)(-0.308991 , -1.990395)(-0.250000 , -2.500000 ) ; ( -2.5,-2 ) node[left]iii ; ( -.4,-.5 ) node i ; ( -4.8,-.3 ) node ii ; ( -2,0 ) node[below ] @xmath53 ; ( -.25 , -2.5 ) node[left ] @xmath54 ; ( .05,.1 ) ( .4,.8 ) ; ( .6 , .7 ) node[above ] @xmath55 ; case iii corresponds to relatively large separation between the two intervals .
as a consequence there are hard edges in all four endpoints @xmath5 , @xmath52 , @xmath4 and @xmath28 , meaning that the limiting particle distribution blows up like an inverse square root .
after rescaling around one of these endpoints the mop kernels converge to limiting kernels that depend on the behavior of the weight at that endpoint . in the typical case of jacobi - type weight functions @xmath56}(x ) , \\
w_2(x ) & = x^{\gamma } ( 1-x)^{\delta } h_2(x ) \chi_{[0,1]}(x ) , \end{aligned}\ ] ] this leads to bessel kernels . in the cases i and ii the separation is small .
a gap will emerge on the larger of the two intervals @xmath50 and @xmath51 , due to the _ pushing of the zeros _ effect .
there will then be three hard edges , and one soft edge , where the limiting particle density vanishes like a square root . around the soft edge
the rescaled mop kernel will converge to the airy kernel ( in the case of weights ) .
the transition between case iii and cases i and ii is probably related to the painlev ii equation as in @xcite .
the bessel , airy and painlev ii kernels already appear as scaling limits for op ensembles @xcite .
this paper deals with the critical point @xmath57 , where the two intervals @xmath50 and @xmath51 are touching and of equal size . in this case
the limiting particle density behaves like @xmath58 as @xmath59 , see @xcite .
we study this critical case in a double scaling limit , where we put @xmath60 and let @xmath5 tend to @xmath6 as @xmath61 : @xmath62 the parameter @xmath63 is known as the double scaling parameter .
this scaling regime corresponds to the phase transition between cases i and ii with @xmath52 fixed at @xmath4 .
for @xmath64 and @xmath65 we have a gap in the spectrum between @xmath4 and @xmath66 where @xmath67 is given by @xmath68 see @xcite . in @xmath66 a soft edge appears , while in @xmath4 we have a hard edge .
plots of the limiting particle densities in the cases @xmath69 , @xmath70 and @xmath71 are given in figure [ fig:187 ] .
as @xmath5 tends to @xmath6 the gap closes and the soft edge meets with the hard edge .
this soft - to - hard - edge collision will give rise to new critical behavior in @xmath4 .
( -4.5,-1 ) rectangle ( 3.7,1 ) ; ( 2.85,-0.7 ) node[text = blue ] @xmath71 ; ( 2.85 , -1 ) node @xmath72 ; ( 0,-0.7 ) node[text = blue ] @xmath73 ; ( -3.25,-0.7 ) node[text = blue ] @xmath69 ; ( -3.25,-1 ) node @xmath74 ; ( -6,-1 ) rectangle ( 4.5,1 ) ; ( -1,0 ) ( 1,0 ) ; ( -1,0 ) circle ( .02 ) node[below ] @xmath6 ; ( 0,0 ) circle ( .02 ) node[below ] @xmath4 ; ( 1,0 ) circle ( .02 ) node[below ] @xmath28 ; plot[smooth ] coordinates ( -0.980000 , 1.090445 ) ( -0.960000 , 0.776795 ) ( -0.940000 , 0.639050 ) ( -0.920000 , 0.557697 ) ( -0.900000 , 0.502734 ) ( -0.880000 , 0.462600 ) ( -0.860000 , 0.431774 ) ( -0.840000 , 0.407244 ) ( -0.820000 , 0.387209 ) ( -0.800000 , 0.370518 ) ( -0.780000 , 0.356399 ) ( -0.760000 , 0.344309 ) ( -0.740000 , 0.333857 ) ( -0.720000 , 0.324754 ) ( -0.700000 , 0.316779 ) ( -0.680000 , 0.309763 ) ( -0.660000 , 0.303570 ) ( -0.640000 , 0.298097 ) ( -0.620000 , 0.293257 ) ( -0.600000 , 0.288983 ) ( -0.580000 , 0.285219 ) ( -0.560000 , 0.281919 ) ( -0.540000 , 0.279048 ) ( -0.520000 , 0.276575 ) ( -0.500000 , 0.274478 ) ( -0.480000 , 0.272739 ) ( -0.460000 , 0.271346 ) ( -0.440000 , 0.270291 ) ( -0.420000 , 0.269570 ) ( -0.400000 , 0.269184 ) ( -0.380000 , 0.269139 ) ( -0.360000 , 0.269447 ) ( -0.340000 , 0.270124 ) ( -0.320000 , 0.271194 ) ( -0.300000 , 0.272691 ) ( -0.280000 , 0.274657 ) ( -0.260000 , 0.277151 ) ( -0.240000 , 0.280247 ) ( -0.220000 , 0.284046 ) ( -0.200000 , 0.288686 ) ( -0.180000 , 0.294355 ) ( -0.160000 , 0.301320 ) ( -0.140000 , 0.309977 ) ( -0.120000 , 0.320935 ) ( -0.100000 , 0.335193 ) ( -0.080000 , 0.354543 ) ( -0.060000 , 0.382623 ) ( -0.040000 , 0.428486 ) ( -0.020000 , 0.525878 ) ( -0.018000 , 0.543019 ) ( -0.016000 , 0.562977 ) ( -0.014000 , 0.586659 ) ( -0.012000 , 0.615448 ) ( -0.010000 , 0.651607 ) ( -0.008000 , 0.699155 ) ( -0.006000 , 0.766210 ) ( -0.004000 , 0.872851 ) ( -0.002000 , 1.093359 ) ( 0.002000 , 1.093359 ) ( 0.004000 , 0.872851 ) ( 0.006000 , 0.766210 ) ( 0.008000 , 0.699155 ) ( 0.010000 , 0.651607 ) ( 0.012000 , 0.615448 ) ( 0.014000 , 0.586659 ) ( 0.016000 , 0.562977 ) ( 0.018000 , 0.543019 ) ( 0.040000 , 0.428486 ) ( 0.060000 , 0.382623 ) ( 0.080000 , 0.354543 ) ( 0.100000 , 0.335193 ) ( 0.120000 , 0.320935 ) ( 0.140000 , 0.309977 ) ( 0.160000 , 0.301320 ) ( 0.180000 , 0.294355 ) ( 0.200000 , 0.288686 ) ( 0.220000 , 0.284046 ) ( 0.240000 , 0.280247 ) ( 0.260000 , 0.277151 ) ( 0.280000 , 0.274657 ) ( 0.300000 , 0.272691 ) ( 0.320000 , 0.271194 ) ( 0.340000 , 0.270124 ) ( 0.360000 , 0.269447 ) ( 0.380000 , 0.269139 ) ( 0.400000 , 0.269184 ) ( 0.420000 , 0.269570 ) ( 0.440000 , 0.270291 ) ( 0.460000 , 0.271346 ) ( 0.480000 , 0.272739 ) ( 0.500000 , 0.274478 ) ( 0.520000 , 0.276575 ) ( 0.540000 , 0.279048 ) ( 0.560000 , 0.281919 ) ( 0.580000 , 0.285219 ) ( 0.600000 , 0.288983 ) ( 0.620000 , 0.293257 ) ( 0.640000 , 0.298097 ) ( 0.660000 , 0.303570 ) ( 0.680000 , 0.309763 ) ( 0.700000 , 0.316779 ) ( 0.720000 , 0.324754 ) ( 0.740000 , 0.333857 ) ( 0.760000 , 0.344309 ) ( 0.780000 , 0.356399 ) ( 0.800000 , 0.370518 ) ( 0.820000 , 0.387209 ) ( 0.840000 , 0.407244 ) ( 0.860000 , 0.431774 ) ( 0.880000 , 0.462600 ) ( 0.900000 , 0.502734 ) ( 0.920000 , 0.557697 ) ( 0.940000 , 0.639050 ) ( 0.960000 , 0.776795 ) ( 0.980000 , 1.090445 ) ; ( -4.5,0 ) ( -2,0 ) ; ( -4.5,0 ) circle ( .02 ) node[below ] @xmath5 ; ( -3,0 ) circle ( .02 ) node[below ] @xmath4 ; ( -2,0 ) circle ( .02 ) node[below ] @xmath28 ; plot[smooth ] coordinates ( -4.486500 , 1.058009 ) ( -4.473000 , 0.750927 ) ( -4.459500 , 0.615444 ) ( -4.446000 , 0.535020 ) ( -4.432500 , 0.480373 ) ( -4.419000 , 0.440216 ) ( -4.405500 , 0.409152 ) ( -4.392000 , 0.384234 ) ( -4.378500 , 0.363698 ) ( -4.365000 , 0.346417 ) ( -4.351500 , 0.331631 ) ( -4.338000 , 0.318807 ) ( -4.324500 , 0.307560 ) ( -4.311000 , 0.297603 ) ( -4.297500 , 0.288717 ) ( -4.284000 , 0.280732 ) ( -4.270500 , 0.273513 ) ( -4.257000 , 0.266953 ) ( -4.243500 , 0.260964 ) ( -4.230000 , 0.255474 ) ( -4.216500 , 0.250424 ) ( -4.203000 , 0.245764 ) ( -4.189500 , 0.241450 ) ( -4.176000 , 0.237448 ) ( -4.162500 , 0.233726 ) ( -4.149000 , 0.230258 ) ( -4.135500 , 0.227021 ) ( -4.122000 , 0.223994 ) ( -4.108500 , 0.221160 ) ( -4.095000 , 0.218504 ) ( -4.081500 , 0.216011 ) ( -4.068000 , 0.213670 ) ( -4.054500 , 0.211469 ) ( -4.041000 , 0.209400 ) ( -4.027500 , 0.207454 ) ( -4.014000 , 0.205622 ) ( -4.000500 , 0.203898 ) ( -3.987000 , 0.202276 ) ( -3.973500 , 0.200749 ) ( -3.960000 , 0.199314 ) ( -3.946500 , 0.197965 ) ( -3.933000 , 0.196697 ) ( -3.919500 , 0.195509 ) ( -3.906000 , 0.194395 ) ( -3.892500 , 0.193353 ) ( -3.879000 , 0.192380 ) ( -3.865500 , 0.191474 ) ( -3.852000 , 0.190633 ) ( -3.838500 , 0.189853 ) ( -3.825000 , 0.189135 ) ( -3.811500 , 0.188475 ) ( -3.798000 , 0.187873 ) ( -3.784500 , 0.187327 ) ( -3.771000 , 0.186837 ) ( -3.757500 , 0.186401 ) ( -3.744000 , 0.186020 ) ( -3.730500 , 0.185691 ) ( -3.717000 , 0.185416 ) ( -3.703500 , 0.185194 ) ( -3.690000 , 0.185025 ) ( -3.676500 , 0.184908 ) ( -3.663000 , 0.184846 ) ( -3.649500 , 0.184837 ) ( -3.636000 , 0.184883 ) ( -3.622500 , 0.184984 ) ( -3.609000 , 0.185142 ) ( -3.595500 , 0.185359 ) ( -3.582000 , 0.185635 ) ( -3.568500 , 0.185973 ) ( -3.555000 , 0.186375 ) ( -3.541500 , 0.186844 ) ( -3.528000 , 0.187383 ) ( -3.514500 , 0.187994 ) ( -3.501000 , 0.188683 ) ( -3.487500 , 0.189453 ) ( -3.474000 , 0.190310 ) ( -3.460500 , 0.191260 ) ( -3.447000 , 0.192310 ) ( -3.433500 , 0.193468 ) ( -3.420000 , 0.194743 ) ( -3.406500 , 0.196147 ) ( -3.393000 , 0.197692 ) ( -3.379500 , 0.199395 ) ( -3.366000 , 0.201272 ) ( -3.352500 , 0.203348 ) ( -3.339000 , 0.205648 ) ( -3.325500 , 0.208206 ) ( -3.312000 , 0.211063 ) ( -3.298500 , 0.214274 ) ( -3.285000 , 0.217905 ) ( -3.271500 , 0.222047 ) ( -3.258000 , 0.226821 ) ( -3.244500 , 0.232397 ) ( -3.231000 , 0.239020 ) ( -3.217500 , 0.247056 ) ( -3.204000 , 0.257090 ) ( -3.202200 , 0.258625 ) ( -3.200400 , 0.260214 ) ( -3.198600 , 0.261862 ) ( -3.196800 , 0.263572 ) ( -3.195000 , 0.265348 ) ( -3.193200 , 0.267193 ) ( -3.191400 , 0.269112 ) ( -3.189600 , 0.271111 ) ( -3.187800 , 0.273193 ) ( -3.186000 , 0.275364 ) ( -3.184200 , 0.277630 ) ( -3.182400 , 0.279997 ) ( -3.180600 , 0.282470 ) ( -3.178800 , 0.285055 ) ( -3.177000 , 0.287755 ) ( -3.175200 , 0.290574 ) ( -3.173400 , 0.293510 ) ( -3.171600 , 0.296559 ) ( -3.169800 , 0.299702 ) ( -3.168000 , 0.302906 ) ( -3.166200 , 0.306103 ) ( -3.164400 , 0.309162 ) ( -3.162600 , 0.311833 ) ( -3.160800 , 0.313607 ) ( -3.159000 , 0.313378 ) ( -3.157200 , 0.308452 ) ( -3.155400 , 0.290833 ) ( -3.153600 , 0.224388 ) ( -3.151800 , 0.000000 ) ( -3.150000 , 0.000000 ) ; plot[smooth ] coordinates ( -2.998000 , 1.093359 ) ( -2.996000 , 0.872851 ) ( -2.994000 , 0.766210 ) ( -2.992000 , 0.699155 ) ( -2.990000 , 0.651607 ) ( -2.988000 , 0.615448 ) ( -2.986000 , 0.586659 ) ( -2.984000 , 0.562977 ) ( -2.982000 , 0.543019 ) ( -2.980000 , 0.525878 ) ( -2.978000 , 0.510935 ) ( -2.950000 , 0.402307 ) ( -2.925000 , 0.360532 ) ( -2.900000 , 0.335193 ) ( -2.875000 , 0.317934 ) ( -2.850000 , 0.305406 ) ( -2.825000 , 0.295962 ) ( -2.800000 , 0.288686 ) ( -2.775000 , 0.283024 ) ( -2.750000 , 0.278618 ) ( -2.725000 , 0.275229 ) ( -2.700000 , 0.272691 ) ( -2.675000 , 0.270888 ) ( -2.650000 , 0.269738 ) ( -2.625000 , 0.269182 ) ( -2.600000 , 0.269184 ) ( -2.575000 , 0.269719 ) ( -2.550000 , 0.270777 ) ( -2.525000 , 0.272359 ) ( -2.500000 , 0.274478 ) ( -2.475000 , 0.277157 ) ( -2.450000 , 0.280432 ) ( -2.425000 , 0.284352 ) ( -2.400000 , 0.288983 ) ( -2.375000 , 0.294412 ) ( -2.350000 , 0.300750 ) ( -2.325000 , 0.308142 ) ( -2.300000 , 0.316779 ) ( -2.275000 , 0.326917 ) ( -2.250000 , 0.338898 ) ( -2.225000 , 0.353203 ) ( -2.200000 , 0.370518 ) ( -2.175000 , 0.391867 ) ( -2.150000 , 0.418855 ) ( -2.125000 , 0.454167 ) ( -2.100000 , 0.502734 ) ( -2.075000 , 0.574876 ) ( -2.050000 , 0.697399 ) ( -2.025000 , 0.977120 ) ( -2.020000 , 1.090445 ) ( -2.010000 , 1.536493 ) ; ( -3.15,0 ) circle ( .02 ) ; ( -3.20,0 ) node[below]@xmath66 ; ( 2,0 ) ( 3.7,0 ) ; ( 2,0 ) circle ( .02 ) node[below ] @xmath5 ; ( 2.7,0 ) circle ( .02 ) node[below ] @xmath4 ; ( 3.7,0 ) circle ( .02 ) node[below ] @xmath28 ; plot[smooth ] coordinates ( 2.007000 , 2.129405 ) ( 2.014000 , 1.511201 ) ( 2.021000 , 1.238426 ) ( 2.028000 , 1.076487 ) ( 2.035000 , 0.966445 ) ( 2.042000 , 0.885574 ) ( 2.049000 , 0.823011 ) ( 2.056000 , 0.772822 ) ( 2.063000 , 0.731458 ) ( 2.070000 , 0.696648 ) ( 2.077000 , 0.666863 ) ( 2.084000 , 0.641032 ) ( 2.091000 , 0.618377 ) ( 2.098000 , 0.598321 ) ( 2.105000 , 0.580423 ) ( 2.112000 , 0.564340 ) ( 2.119000 , 0.549803 ) ( 2.126000 , 0.536593 ) ( 2.133000 , 0.524536 ) ( 2.140000 , 0.513487 ) ( 2.147000 , 0.503324 ) ( 2.154000 , 0.493948 ) ( 2.161000 , 0.485273 ) ( 2.168000 , 0.477228 ) ( 2.175000 , 0.469750 ) ( 2.182000 , 0.462785 ) ( 2.189000 , 0.456288 ) ( 2.196000 , 0.450217 ) ( 2.203000 , 0.444538 ) ( 2.210000 , 0.439220 ) ( 2.217000 , 0.434234 ) ( 2.224000 , 0.429558 ) ( 2.231000 , 0.425168 ) ( 2.238000 , 0.421047 ) ( 2.245000 , 0.417177 ) ( 2.252000 , 0.413542 ) ( 2.259000 , 0.410129 ) ( 2.266000 , 0.406925 ) ( 2.273000 , 0.403920 ) ( 2.280000 , 0.401103 ) ( 2.287000 , 0.398466 ) ( 2.294000 , 0.396000 ) ( 2.301000 , 0.393699 ) ( 2.308000 , 0.391555 ) ( 2.315000 , 0.389564 ) ( 2.322000 , 0.387719 ) ( 2.329000 , 0.386017 ) ( 2.336000 , 0.384454 ) ( 2.343000 , 0.383025 ) ( 2.350000 , 0.381729 ) ( 2.357000 , 0.380562 ) ( 2.364000 , 0.379523 ) ( 2.371000 , 0.378610 ) ( 2.378000 , 0.377822 ) ( 2.385000 , 0.377158 ) ( 2.392000 , 0.376618 ) ( 2.399000 , 0.376202 ) ( 2.406000 , 0.375911 ) ( 2.413000 , 0.375745 ) ( 2.420000 , 0.375707 ) ( 2.427000 , 0.375798 ) ( 2.434000 , 0.376021 ) ( 2.441000 , 0.376379 ) ( 2.448000 , 0.376876 ) ( 2.455000 , 0.377516 ) ( 2.462000 , 0.378304 ) ( 2.469000 , 0.379247 ) ( 2.476000 , 0.380352 ) ( 2.483000 , 0.381625 ) ( 2.490000 , 0.383078 ) ( 2.497000 , 0.384719 ) ( 2.504000 , 0.386561 ) ( 2.511000 , 0.388617 ) ( 2.518000 , 0.390904 ) ( 2.525000 , 0.393439 ) ( 2.532000 , 0.396244 ) ( 2.539000 , 0.399341 ) ( 2.546000 , 0.402761 ) ( 2.553000 , 0.406536 ) ( 2.560000 , 0.410704 ) ( 2.567000 , 0.415312 ) ( 2.574000 , 0.420414 ) ( 2.581000 , 0.426078 ) ( 2.588000 , 0.432383 ) ( 2.595000 , 0.439430 ) ( 2.602000 , 0.447340 ) ( 2.609000 , 0.456271 ) ( 2.616000 , 0.466423 ) ( 2.623000 , 0.478059 ) ( 2.630000 , 0.491530 ) ( 2.637000 , 0.507320 ) ( 2.644000 , 0.526117 ) ( 2.651000 , 0.548941 ) ( 2.658000 , 0.577377 ) ( 2.665000 , 0.614069 ) ( 2.672000 , 0.663833 ) ( 2.679000 , 0.736706 ) ( 2.686000 , 0.858626 ) ( 2.693000 , 1.132654 ) ; plot[smooth ] coordinates ( 2.800900 , 0.000000 ) ( 2.801800 , 0.396577 ) ( 2.802700 , 0.488039 ) ( 2.803600 , 0.511062 ) ( 2.804500 , 0.516025 ) ( 2.805400 , 0.514424 ) ( 2.806300 , 0.510129 ) ( 2.807200 , 0.504713 ) ( 2.808100 , 0.498885 ) ( 2.809000 , 0.492987 ) ( 2.818000 , 0.444607 ) ( 2.827000 , 0.413284 ) ( 2.836000 , 0.391274 ) ( 2.845000 , 0.374695 ) ( 2.854000 , 0.361608 ) ( 2.863000 , 0.350930 ) ( 2.872000 , 0.342006 ) ( 2.881000 , 0.334409 ) ( 2.890000 , 0.327848 ) ( 2.899000 , 0.322120 ) ( 2.908000 , 0.317071 ) ( 2.917000 , 0.312589 ) ( 2.926000 , 0.308587 ) ( 2.935000 , 0.304996 ) ( 2.944000 , 0.301762 ) ( 2.953000 , 0.298840 ) ( 2.962000 , 0.296195 ) ( 2.971000 , 0.293797 ) ( 2.980000 , 0.291620 ) ( 2.989000 , 0.289644 ) ( 2.998000 , 0.287852 ) ( 3.007000 , 0.286227 ) ( 3.016000 , 0.284758 ) ( 3.025000 , 0.283432 ) ( 3.034000 , 0.282241 ) ( 3.043000 , 0.281176 ) ( 3.052000 , 0.280230 ) ( 3.061000 , 0.279395 ) ( 3.070000 , 0.278668 ) ( 3.079000 , 0.278043 ) ( 3.088000 , 0.277516 ) ( 3.097000 , 0.277083 ) ( 3.106000 , 0.276741 ) ( 3.115000 , 0.276488 ) ( 3.124000 , 0.276321 ) ( 3.133000 , 0.276239 ) ( 3.142000 , 0.276239 ) ( 3.151000 , 0.276322 ) ( 3.160000 , 0.276485 ) ( 3.169000 , 0.276729 ) ( 3.178000 , 0.277053 ) ( 3.187000 , 0.277456 ) ( 3.196000 , 0.277940 ) ( 3.205000 , 0.278504 ) ( 3.214000 , 0.279150 ) ( 3.223000 , 0.279877 ) ( 3.232000 , 0.280688 ) ( 3.241000 , 0.281584 ) ( 3.250000 , 0.282566 ) ( 3.259000 , 0.283637 ) ( 3.268000 , 0.284799 ) ( 3.277000 , 0.286054 ) ( 3.286000 , 0.287406 ) ( 3.295000 , 0.288858 ) ( 3.304000 , 0.290414 ) ( 3.313000 , 0.292077 ) ( 3.322000 , 0.293853 ) ( 3.331000 , 0.295745 ) ( 3.340000 , 0.297761 ) ( 3.349000 , 0.299906 ) ( 3.358000 , 0.302187 ) ( 3.367000 , 0.304612 ) ( 3.376000 , 0.307188 ) ( 3.385000 , 0.309926 ) ( 3.394000 , 0.312836 ) ( 3.403000 , 0.315929 ) ( 3.412000 , 0.319218 ) ( 3.421000 , 0.322718 ) ( 3.430000 , 0.326444 ) ( 3.439000 , 0.330415 ) ( 3.448000 , 0.334652 ) ( 3.457000 , 0.339177 ) ( 3.466000 , 0.344016 ) ( 3.475000 , 0.349201 ) ( 3.484000 , 0.354765 ) ( 3.493000 , 0.360748 ) ( 3.502000 , 0.367196 ) ( 3.511000 , 0.374163 ) ( 3.520000 , 0.381712 ) ( 3.529000 , 0.389919 ) ( 3.538000 , 0.398871 ) ( 3.547000 , 0.408677 ) ( 3.556000 , 0.419468 ) ( 3.565000 , 0.431405 ) ( 3.574000 , 0.444687 ) ( 3.583000 , 0.459570 ) ( 3.592000 , 0.476381 ) ( 3.601000 , 0.495548 ) ( 3.610000 , 0.517649 ) ( 3.619000 , 0.543479 ) ( 3.628000 , 0.574173 ) ( 3.637000 , 0.611416 ) ( 3.646000 , 0.657844 ) ( 3.655000 , 0.717862 ) ( 3.664000 , 0.799535 ) ( 3.673000 , 0.919733 ) ( 3.682000 , 1.122214 ) ( 3.691000 , 1.581148 ) ; ( 2.8009,0 ) circle ( .02 ) ; ( 2.85 , 0 ) node[below ] @xmath66 ; so for the remainder of the paper we take for @xmath2 @xmath75 , \qquad \delta_{2 } : = [ 0,1],\ ] ] and as in @xcite we consider the following modified jacobi weights : @xmath76}(x ) , \\
w_{2}(x ) & : = x^{\beta}(1-x)^{\gamma } h_{2}(x ) \chi_{[0,1]}(x ) , \end{aligned}\ ] ] with @xmath77 and functions @xmath78 that are positive on @xmath79 and analytic in a neighborhood of @xmath80 for @xmath81 .
note that we use the same exponent @xmath82 in both @xmath83 and @xmath84 .
in @xcite we considered the weights and obtained a mehler - heine asymptotic formula for the mop associated to @xmath42 and @xmath43 and @xmath44 as @xmath85 .
see also @xcite .
we use the notation @xmath86 to denote the dependence of the correlation kernel on the parameter @xmath5 .
the main result is then as follows : [ thm:145 ] let @xmath87 be the correlation kernel of the mop ensemble with weights and diagonal multi - index @xmath44 .
let @xmath88 with @xmath89 .
then as @xmath90 we have : @xmath91 for some limiting kernel @xmath92 that will be described below in propositions [ prop:169 ] and [ prop:183 ] .
the @xmath93-term in is uniform for @xmath94 in a bounded set and @xmath95 in a compact subset in @xmath96 .
the limiting kernel @xmath8 is referred to as the _ angelesco kernel_. it has an expression in terms of the angelesco model parametrix @xmath97 which we will describe in section [ sec:356 ] .
more explicit expressions exist , one involving a certain pairing of two analytic functions @xmath98 and one in terms of a double contour integral .
[ prop:169 ] define functions @xmath99 and @xmath100 by @xmath101 here the contours @xmath102 and @xmath103 are shown in figure [ fig:197 ] , and we choose the principal branches for the fractional powers @xmath104 and @xmath105
. then we can write @xmath92 as a pairing of @xmath106 and @xmath107 in the following way : @xmath108.\ ] ] ( 0,0 ) circle ( .03 ) node[above left]@xmath4 ; ( 0,0 ) ( -1.5,0 ) ; ( .4,0 ) ellipse ( .4 and .6 ) ; ( .5,.8 ) arc ( 90:-90:.5 and .8 ) ; ( .5,.8 ) .. controls ( 0,.8 ) and ( -.5,.5 ) .. ( -1,.5 ) ; ( .5,-.8 ) .. controls ( 0,-.8 ) and ( -.5,-.5 ) .. ( -1,-.5 ) ; ( .9,-.5 ) node [ right ] @xmath102 ; ( -.1,-.1 ) node[below ] @xmath103 ; ( -1,.5 ) ( -1.3,.5 ) ; ( -1,-.5 ) ( -1.3,-.5 ) ; from the expressions for @xmath99 and @xmath100 it is clear that we have @xmath109 hence the angelesco kernel satisfies the following ( expected ) symmetry : @xmath110 [ prop:183 ] let the contours @xmath102 and @xmath103 be again as in figure [ fig:197 ] , and take @xmath111 , @xmath112 .
then we have the following double integral formula for @xmath92 : @xmath113 again main branches are used for the fractional powers @xmath114 and @xmath105 . in @xcite , @xcite the authors consider a model of @xmath115 non - intersecting squared bessel paths , conditioned to start at time @xmath116 in the point @xmath117 , and to end at time @xmath118 at @xmath119 , see figure [ fig:246 ] .
for each fixed time @xmath120 the particles form a mop ensemble with respect to orthogonality weights involving the modified bessel functions @xmath121 and @xmath122 for some @xmath123 .
associated to each fixed time @xmath120 one has a limiting mean particle distribution as @xmath85 . for small values of @xmath120 the particles are supported on an interval in @xmath124 , away from @xmath4 . on the other hand , for values of @xmath120 near @xmath28 the point @xmath4 is contained in the support , meaning that particles stick to the hard wall in @xmath4 .
there is then a critical time @xmath125 at which the gap between the particles and the wall in @xmath4 closes .
this phase transition is similar to the phase transition in angelesco ensembles . by taking a double scaling limit around @xmath119 and @xmath126 a new kind of universal limiting kernel , @xmath127 , was obtained ( * ? ? ?
* equation ( 1.18 ) ) .
the function @xmath127 depends on two positive position variables @xmath94 and @xmath95 , a double scaling parameter @xmath63 , and the parameter @xmath82 appearing in the modified bessel functions .
it turns out that this critical kernel is almost the same as the angelesco kernel @xmath8 . by applying a simple substitution to the double integral formula for @xmath127 ( *
* equation ( 1.19 ) ) one obtains the following identity : @xmath128 where we used the notation @xmath129 to indicate the dependence of the kernel on the parameter @xmath82 . although the kernels are different this identity implies an equality at the level of the correlation functions .
hence , after proper identification of the parameters , the critical behavior in angelesco ensembles coincides with the critical behavior in the model of non - intersecting squared bessel paths .
a significant difference between the two models is that in the squared bessel paths model the particles remain on the positive half - line , while in our model they are located on both sides of @xmath4 and there is interaction between the two groups .
the proof of theorem [ thm:145 ] follows from the application of the deift / zhou steepest descent method on the riemann - hilbert ( rh ) problem for mops . in section [ sec:301 ] we recall the rh formulation for the weights @xmath42 and @xmath43 and we give a formula for the mop kernel in terms of the rh problem for mops .
also we recall the rh problem for the angelesco local model parametrix @xmath97 which is taken from @xcite .
the deift / zhou method consists of a sequence of invertible transformations reducing the original rh problem into a normalized rh problem , for which uniform estimates can be made .
the analysis is the same as in @xcite . instead of giving full details we will , in section [ sec:260 ] ,
give a quick overview of the transformations and describe their effect on the expression for the mop kernel . in section [ sec:288 ]
we shall use the series of transformations to show that the limit in theorem [ thm:145 ] indeed holds , and we give expressions for the angelesco kernel @xmath8 in terms of the angelesco parametrix @xmath97 .
finally , section [ sec:785 ] contains the proofs of the two explicit expressions for @xmath8 stated in proposition [ prop:169 ] and proposition [ prop:183 ] .
this section recalls the riemann - hilbert ( rh ) characterization of the relevant multiple orthogonal polynomials and the angelesco local model parametrix .
we state here the rh problem for mops @xcite with respect to the system of weights @xmath130 given in and the multi - index @xmath131 .
the endpoint conditions are as in @xcite .
we look for a @xmath132 matrix - valued function @xmath133 such that : * @xmath133 is defined and analytic on @xmath134 $ ] . * @xmath133 has continuous boundary values @xmath135 on @xmath136 and @xmath137 and they satisfy the jump relation @xmath138 with @xmath139 given by @xmath140 recall that @xmath141 outside of @xmath0 $ ] and @xmath142 outside of @xmath1 $ ] . * as @xmath143 we have
@xmath144 * @xmath133 has the following behavior at the endpoints of the intervals : @xmath145 @xmath146 @xmath147 the @xmath93-symbol is to be taken entry - wise .
this rh problem has a unique solution in terms of mop with respect to the weights @xmath42 and @xmath43 .
in particular we have that @xmath148 is a monic polynomial of degree @xmath149 that satisfies the multiple orthogonality conditions @xmath150 for the full expression of @xmath133 we refer to @xcite .
most important for the present paper is the following formula for the mop correlation kernel in terms of @xmath133 , see @xcite : @xmath151 this formula is valid for @xmath152 with @xmath153 . as before it is understood that @xmath154 for @xmath95 outside @xmath136 , and @xmath155 for @xmath95 outside @xmath137 .
the formula can be extended to the case @xmath156 by lhpital s rule . at a crucial step in the steepest descent analysis we need to do a local analysis at the point @xmath4 .
this step involves certain special functions , which are combined into a @xmath157 matrix - valued function , called the angelesco model parametrix @xmath97 and which was introduced in ( * ? ? ?
* section 2.2 ) .
the function @xmath97 depends on two parameters @xmath123 and @xmath112 .
since @xmath82 is considered fixed we do not emphasize the dependence on @xmath82 .
we may write @xmath158 to emphasize the dependence on @xmath63 . the function @xmath97 has the following rh characterization : ( -1,0 ) ( 0,0 ) ; ( 0,0 ) ( 1,0 ) ; ( -.707106781,-.707106781 ) ( 0,0 ) ; ( 0,0 ) ( .707106781,.707106781 ) ; ( -.707106781,.707106781 ) ( 0,0 ) ; ( 0,0 ) ( .707106781,-.707106781 ) ; ( 0,0 ) node[above]@xmath4 ; ( .2,0 ) arc ( 0:45:.2 ) ; ( .2,0 ) node[above right]@xmath159 ; ( -1,0 ) node[left , font= ] @xmath160 ; ( -.707106781,.707106781 ) node[left , font=]@xmath161 ; ( .707106781,.707106781 ) node[right , font=]@xmath162 ; ( 1,0 ) node[right , font= ] @xmath163 ; ( .707106781,-.707106781 ) node[right , font=]@xmath162 ; ( -.707106781,-.707106781 ) node[left , font=]@xmath164 ; * @xmath97 is defined and analytic on @xmath165 where @xmath166 is a contour consisting of six oriented rays through the origin , as shown in figure [ figurejumpspsi ] .
* @xmath97 has continuous boundary values on @xmath167 that satisfy the jump condition @xmath168 where the jump matrices @xmath169 are also given in figure [ figurejumpspsi ] . *
denote @xmath170 .
as @xmath171 with @xmath172 , we have @xmath173 where @xmath174 , @xmath175 and @xmath176 are defined by @xmath177 @xmath178 and the @xmath179 are defined by @xmath180 the expansion for @xmath181 as @xmath171 is valid uniformly for @xmath63 in a bounded set . *
denote by @xmath182 the sectors @xmath183 + around @xmath4 we then have the following estimate : @xmath184 note that the parameter @xmath82 appears in the jump condition and in the behavior around @xmath4 .
the dependence on @xmath63 is only in the asymptotic condition as @xmath143 .
the rh problem for @xmath97 has a unique solution .
it is constructed out of solutions of the third order linear differential equation @xmath185 this differential equation has solutions in the form of contour integrals @xmath186 where @xmath187 is an appropriate contour so that the integrand vanishes at the endpoints of the contour @xmath187 . define three contours @xmath188 and @xmath189 as in figure [ figurecontoursgamma ] , and define for @xmath190 with @xmath191 @xmath192 where we choose the branch of @xmath193 with a cut on the positive real axis , i.e. , @xmath194 the integrals only converge for @xmath190 with @xmath191 , but the functions @xmath195 can be continued analytically using contour deformations .
branch points for the @xmath196-functions are @xmath4 and @xmath197 and we take the analytic continuation to @xmath198 $ ] , thus with a branch cut on the negative real axis .
( 0,0 ) circle ( .02 ) node[above]@xmath4 ; ( 0,0 ) ( 1,0 ) ; ( -1,0 ) ( 0,0 ) ; ( -.5,0 ) node[above]@xmath189 ; ( -1.2,0 ) ( -1,0 ) ; ( 0,0 ) arc ( -90:90:.5 and .3 ) ; ( -1,.6 ) ( 0,.6 ) ; ( -1.2,.6 ) ( -1,.6 ) ; ( -.3,.6 ) node[above]@xmath199 ; ( 0,0 ) arc ( 90:-90:.5 and .3 ) ; ( -1,-.6 )
( 0,-.6 ) ; ( -1.2,-.6 )
( -1,-.6 ) ; ( -.3,-.6 ) node[above ] @xmath200 ; in the upper half - plane the unique solution for @xmath97 is then given by , see @xcite for more details , @xmath201 and in the lower half - plane by @xmath202
in this section we apply the deift / zhou steepest descent method to the rh problem for @xmath133 . for details of the various transformations
we refer to @xcite .
the deift / zhou method consists of a number of invertible transformations reducing the original matrix - valued function @xmath133 to a function @xmath203 that is uniformly close to the identity matrix . of main importance
is the effect of these transformations on the expression for the mop kernel @xmath204 , see , with diagonal multi - indices : @xmath205 the first transformation @xmath206 normalizes the rh problem for @xmath133 at infinity : the new matrix - valued function @xmath207 tends to the identity matrix at infinity .
the transformation uses functions @xmath208 and @xmath209 derived from a modified equilibrium problem with logarithmic potentials .
these @xmath210-functions are analytic on @xmath211 $ ] .
see ( * ? ? ?
* section 4.1 ) for precise details on the transformation @xmath206 . for the mop kernel
we obtain from and formula ( 4.4 ) in @xcite @xmath212 the constants @xmath213 and @xmath214 appear in the euler - lagrange variational conditions related to the equilibrium problem .
the second transformation @xmath215 is the opening of the lenses , see ( * ? ? ?
* section 4.2 ) . in this transformation
the rapidly oscillating jump matrices along @xmath136 and @xmath137 that appeared after the transformation @xmath206 are turned into constant jump matrices , with the side effect of creating exponentially small jump matrices on the lips of the lenses .
the expression for this transformation depends on the location of @xmath190 . from now on we restrict ourselves to positive @xmath94 and @xmath95 , since in the cases @xmath216 or @xmath217 the expressions for @xmath218 are different but similar .
remark that we have @xmath219 for @xmath220 . by and formulas ( 4.14 ) and ( 4.5 ) in @xcite ,
we then get for @xmath221 , ( note that @xmath222 ) : @xmath223 the next two transformations @xmath224 and @xmath225 are applied after constructing local and global parametrices .
these parametrices are built as approximations to the rh problem for @xmath226 , and one can think of @xmath227 as the approximation error .
the approximation around @xmath4 is not good enough for the further analysis , whence the need for a very last transformation @xmath225 .
we refer to ( * ? ? ?
* section 4.4 ) for precise details and formulas .
the local parametrix @xmath228 around @xmath4 is constructed on a disk @xmath229 centered in zero with shrinking radius @xmath230 , see formulas ( 4.68 ) and ( 4.70 ) of @xcite .
it involves the angelesco model parametrix @xmath97 described in section [ sec:356 ] .
assume that @xmath94 and @xmath95 are positive and belong to the disk @xmath229 .
then @xmath231 and inserting this in we obtain after some calculations @xmath232 for @xmath216 or @xmath217 the formula also holds , but with different row and column vectors , as given in the following table : @xmath233 the functions @xmath234 and @xmath63 appearing in were defined in @xcite . since we do not need to know their full expressions we will restrict ourselves to recalling their most important properties .
these functions are both analytic in a neighborhood of @xmath4 , with @xmath235 .
they depend on @xmath5 , and as @xmath236 one has : @xmath237 see ( * ? ? ?
* formula ( 4.76 ) ) . from this
it easily follows that : let @xmath238 and @xmath239 . put @xmath240 then as @xmath61 we have @xmath241 the @xmath93-term is uniform for @xmath94 in a bounded set .
this follows easily from and the fact that @xmath242 and @xmath243 converge uniformly as @xmath236 to @xmath244 and @xmath245 .
also we have the following technical estimate : [ prop:338 ] let @xmath246 , @xmath247 and put @xmath248 then as @xmath61 we have @xmath249 the @xmath93-term is uniform for @xmath94 and @xmath95 in bounded sets . assuming this proposition we can finish the proof of theorem [ thm:145 ] as follows .
let @xmath250 and let @xmath251 be as in .
then for @xmath115 large enough we have that @xmath252 .
using , and in we then obtain : @xmath253 this proves theorem [ thm:145 ] for @xmath250 .
the right - hand side of also provides us with an expression for @xmath254 in terms of the angelesco parametrix @xmath97 . for @xmath255 or
@xmath217 the proof is completely similar and holds , but with different row and column vectors , as given in .
the proof of proposition [ prop:338 ] is rather technical and requires deeper knowledge of the structure of the matrix functions @xmath227 and @xmath256 . from (
* section 5.2 ) we have @xmath257 where , see formula ( 5.13 ) of @xcite , @xmath258 uniformly in @xmath190 .
the function @xmath259 in is given by @xmath260 where @xmath261 and @xmath262 play a role in the analysis of @xcite .
the function @xmath263 is analytic on @xmath264 with a simple pole in @xmath4 , whose residue is @xmath262 . hence @xmath265 is analytic on @xmath229 . additionally , from ( * ? ? ?
* equation ( 4.84 ) ) it follows that @xmath265 is of the form @xmath266 for a certain constant diagonal matrix @xmath267 . as a consequence we have the following basic property of @xmath259 : @xmath268 additionally , @xmath269 is uniformly bounded on the disk @xmath270 .
since @xmath270 has radius @xmath230 this easily implies that @xmath271 if @xmath272 , and therefore @xmath273 from and the fact that @xmath274 , @xmath275 , it follows that @xmath276 combining , , , and we obtain @xmath277 the matrix - valued function @xmath256 defined in formula ( 4.80 ) of @xcite is analytic and invertible on a fixed neighborhood of @xmath4 .
it depends on both @xmath5 and @xmath115 .
it takes the form @xmath278 where @xmath279 is independent of @xmath115 , and @xmath279 converges uniformly as @xmath236 .
then @xmath280 and combining with yields @xmath281 it also follows from that @xmath282 a straightforward combination of , and , would yield an @xmath283 bound for the left - hand side of .
we need to go a little deeper into the structure of @xmath284 in order to improve the bound
. from the computations in section 6 of @xcite it follows that @xmath285 has the following behavior as @xmath286 : @xmath287 where @xmath288 is some non - zero constant , see ( * ? ? ? * equation ( 6.12 ) ) . from similar computations
one finds for some non - zero @xmath289 : @xmath290 from these expansions it follows that the leading terms of @xmath291 and @xmath285 cancel out with @xmath265 , see .
in particular it holds by , , and that @xmath292 e_{n}(x_{n } ) = \mathcal{o}(n^{-1 } ) \textrm { as } n \rightarrow \infty.\ ] ] taking into account the extra estimate we can perform the following computation ( we do not write @xmath293 , and we use , , , , and the fact that @xmath294 and @xmath295 are uniformly bounded ) @xmath296 ( i- v_{n}(x_{n } ) ) e_{n}(x_{n } ) \\ & = e_{n}(y_{n})^{-1 } \left [ i + v_{n}(y_{n } ) - v_{n}(x_{n } ) + \mathcal{o}\left(n^{-\frac{7}{6 } } \right)\right ] e_{n}(x_{n } ) \\ & = i + \mathcal{o}\left(n^{-\frac{1}{6}}\right ) , \end{aligned}\end{gathered}\ ] ] which completes the proof of proposition [ prop:338 ] .
in the previous section we derived expressions for @xmath8 in terms of @xmath97 , see right - hand side of . in the present section
these expressions will be used to obtain the explicit formulas stated in proposition [ prop:169 ] and proposition [ prop:183 ] .
we follow here the computations in ( * ? ? ?
* section 10.2 ) , see also ( * ? ? ?
* section 8.2 ) .
as stated in section [ sec:356 ] the entries of @xmath97 are solutions and derivatives of solutions to the third order differential equation @xmath297 the solution functions @xmath298 , @xmath299 have explicit expressions @xmath300 with contours @xmath301 as shown in figure [ figurecontoursgamma ] .
a first step towards the derivation of the double integral formula consists of finding similar explicit expressions for the entries of the inverse of @xmath97 .
this involves an _ adjoint differential equation _ to the one for the @xmath196-functions , and an associated _ bilinear concomitant _ , see e.g. @xcite , @xcite . the solutions of this adjoint differential equation have contour integral representations , and it takes some work to identify the appropriate contours .
when the explicit expressions for the entries of @xmath302 are known we will be in a position to prove the double integral formula for @xmath8 .
the adjoint differential equation reads @xmath303 the associated bilinear concomitant @xmath304 is a differential operator defined on general twice differentiable functions @xmath196 and @xmath305 : @xmath306(x , y ) & : = yr(y ) q''(x ) - [ ( \beta + 1 ) r(y ) + yr'(y ) ] q'(x ) \notag \\ & \qquad + [ yr''(y ) + ( \beta + 2 ) r'(y ) - \tau r(y ) ] q(x).\end{aligned}\ ] ] in order to ease the notation we introduce 3 operators @xmath307 , @xmath308 and @xmath309 @xmath310 so we can write the bilinear concomitant as @xmath311(x , y ) = \mathcal{b}_{2}r(y ) q''(x ) + \mathcal{b}_{1}r(y ) q'(x ) + \mathcal{b}_{0}r(y ) q(x ) .\ ] ] the central property of this bilinear concomitant is that if @xmath196 solves the differential equation for @xmath196 and @xmath305 solves the differential equation for @xmath305 , then the bilinear concomitant evaluated on the diagonal @xmath312(x , x)$ ] is a constant .
this can be checked by taking the derivative of @xmath312(x , x)$ ] with respect to @xmath94 and plugging in the differential identities for @xmath196 and @xmath305 .
the differential equation for @xmath305 allows solutions in the form of a contour integral : for an appropriate contour @xmath313 the function @xmath314 solves the differential equation . by an appropriate contour
is meant a contour such that the integrand vanishes exponentially fast at the endpoints . in this case
this implies that @xmath313 must end and start in @xmath4 or at @xmath197 .
now we construct solutions @xmath315 to such that @xmath316(z , z ) = \delta_{jk } \quad j , k = 1,2,3 , \label{concomitant01}\ ] ] where we put the cuts of the functions @xmath317 , if any , on the negative real axis . using these relations
we can find expressions for @xmath302 . in each sector of the complex plane @xmath181 is of the form @xmath318 for some invertible matrix @xmath319 .
the inverse of @xmath97 is then given by @xmath320 indeed , the product @xmath321 then equals @xmath322(z , z ) & \mathcal{b } [ q_{2},r_{1}](z , z ) & \mathcal{b}[q_{3},r_{1}](z , z ) \\ { } \mathcal{b}[q_{1},r_{2}](z , z ) & \mathcal{b}[q_{2},r_{2}](z , z ) & \mathcal{b}[q_{3},r_{2}](z , z ) \\ { } \mathcal{b}[q_{1},r_{3}](z , z ) & \mathcal{b}[q_{2 } , r_{3}](z , z ) & \mathcal{b}[q_{3 } , r_{3}](z , z ) \end{pmatrix } \cdot c,\ ] ] which by is the identity matrix .
[ prop:537 ] let @xmath326 be a non - empty open set .
assume that the following two contour integrals converge for @xmath327 : @xmath328 also assume that the cuts of @xmath193 and @xmath329 coincide , such that @xmath330 for @xmath331 . if @xmath187 and @xmath313 do not intersect , or only intersect in @xmath4 , then @xmath312(z , z ) = 0 $ ] for @xmath327 .
if they however intersect once transversally at a non - zero point , and if @xmath187 meets @xmath313 on the negative side of @xmath313 , then @xmath332(z , z ) = 2 \pi i , \qquad
z \in \omega.\ ] ] using the definition of the bilinear concomitant and the definitions of @xmath337 and @xmath338 we can write the concomitant @xmath339(z , z)$ ] of @xmath196 and @xmath305 as a double integral : @xmath340 the fraction in the above integral can be split up as follows : @xmath341 suppose that the contours @xmath187 and @xmath313 do not intersect , or only intersect in @xmath4 .
then the three terms on the right - hand side of remain bounded and we can write : @xmath342(z , z ) \notag \\ & \quad = \int_{t \in
\gamma } \int_{s \in \widehat{\gamma } } \frac { 1}{t^{3 } ( t - s ) } \frac { \partial \theta_{r}(s)}{\partial s } e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt \notag \\ & \qquad + \int_{t \in \gamma } \int_{s \in \widehat{\gamma } } \frac { 1}{s^{3}(t - s ) } \frac{\partial \theta_{q}(t)}{\partial t } e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt \notag \\ & \qquad - \int_{t \in \gamma } \int_{s \in \widehat{\gamma } } \frac{s^{2 } + st + t^{2}}{s^{3}t^{3 } ( t - s)}e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt.\end{aligned}\ ] ] we can apply integration by parts to the first two integrals in .
the boundary terms vanish .
what remains is : @xmath343(z , z ) & = -\int_{t \in \gamma } \int_{s \in \widehat{\gamma } } \frac { 1}{t^{3 } ( t - s)^{2 } } e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt \notag \\ & \quad + \int_{t \in \gamma } \int_{s \in \widehat{\gamma } } \frac { 1}{s^{3}(t - s)^{2 } } e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt \notag \\ & \quad - \int_{t \in \gamma } \int_{s \in \widehat{\gamma } } \frac{t^{3 } - s^{3}}{s^{3}t^{3 } ( t - s)^{2}}e^{\theta_{q}(t ) } e^{\theta_{r}(s ) } \ , ds dt \notag \\ & = 0.\end{aligned}\ ] ] now suppose that @xmath187 and @xmath313 intersect in a point @xmath344 , with @xmath187 going from the right side of @xmath313 to the left side .
then the contours can be deformed such that @xmath187 contains the interval @xmath345 $ ] while @xmath313 contains @xmath346 $ ] for some @xmath347 , see figure [ fig:611 ] .
define a new contour @xmath348 by @xmath349 $ ] with @xmath350 .
then we have using : @xmath332(z , z ) = \lim_{\varepsilon \rightarrow 0^{+ } } \int_{t \in \gamma } \int_{s \in \widehat{\gamma}^{\varepsilon } } \left [ \begin{array}{c } z(s^{2 } + st + t^{2 } ) - \tau \\ + ( \beta + 2)s - ( \beta + 1)t \end{array } \right ] \frac { e^{\theta_{q}(t ) } e^{\theta_{r}(s)}}{s^{3 } t^{3 } } \,dsdt.\ ] ] since @xmath348 and @xmath187 do not intersect we can use the expression in with @xmath313 replaced by @xmath348 and apply integration by parts .
however , now the boundary term in @xmath66 remains : @xmath332(z , z ) = \lim_{\varepsilon \rightarrow 0^{+ } } \int_{\gamma } \frac{1}{t^{3 } } \left [ \frac{1}{t - s } e^{\theta_{r}(s ) } \right]^{s = c - \varepsilon}_{s = c + \varepsilon } e^{\theta_{q}(t ) } \ , dt .\ ] ] by continuity we have @xmath351 as @xmath352 .
hence we obtain @xmath353(z , z ) = \lim_{\varepsilon \rightarrow 0 } \left ( e^{\theta_{r}(c ) } + \mathcal{o}(\varepsilon)\right ) \int_{\gamma } \frac{1 } { t^{3 } } \left[\frac{1}{t - c+ \varepsilon } - \frac{1}{t - c-\varepsilon } \right ] e^{\theta_{q}(t ) } \ , dt.\ ] ] ( -2 , 0 ) ( 2,0 ) ; ( -2,0 ) circle ( .03 ) node[above]@xmath354 ; ( 2,0 ) circle ( .03 ) node[above]@xmath355 ; ( 0,-2 ) ( 0,2 ) ; ( 0 , -2 ) circle ( .03 ) node[right]@xmath356 ; ( 0 , 2 ) circle ( .03 ) node[right]@xmath357 ; ( 0,0 ) circle ( .03 ) node[above right ] @xmath358 ; ( 2,0 ) .. controls ( 2.5,0 ) and ( 3,.3) .. (3.5,1 ) ; ( 0,2 ) .. controls ( 0,2.5 ) and ( .2 , 3 ) .. ( .6 , 3.5 ) ; ( 3.5,1 ) node [ right ] @xmath313 ; ( 0,-2 ) .. controls ( 0,-2.5 ) and ( .2 , -3 ) .. ( .6 , -3.5 ) ; ( -2,0 ) .. controls ( -2.5,0 ) and ( -3,-.1) .. (-3.5,-.3 ) ; ( .6,-3.5 ) node [ right ] @xmath187 ; finally we deform the segment @xmath359 \subset \gamma$ ] into the union of a small circle around @xmath360 and a semicircle with radius @xmath361 around @xmath358 , see figure [ fig:650 ] .
the small circle around @xmath360 will give us a residue contribution , while the contribution of the semicircle and the remaining part of @xmath187 tends to zero as @xmath362 .
remark that the small circle around @xmath360 has clockwise orientation .
then we have as @xmath362 : @xmath363 e^{\theta_{q}(t ) } \ , dt = 2 \pi i \frac{1}{(c + \varepsilon)^{3 } } e^{\theta_{q}(c + \varepsilon ) } + \mathcal{o}(\varepsilon ) .\ ] ] ( 0,-2 ) ( 0,2 ) ; ( .7,0 ) circle ( .03 ) ; ( .72,-.1 ) ( .9,-.7 ) ; ( .9,-.7 ) node[below ] @xmath364 ; ( 2,0 ) circle ( .03 ) node[right]@xmath355 ; ( .7,-.45 ) arc ( 270:-90:.45 ) ; ( 0,-2 ) arc ( -90:90:2 ) ; ( 0,-2 ) .. controls ( 0,-2.5 ) and ( .2 , -3 ) .. ( .6 , -3.5 ) ; ( 0,2 ) .. controls ( 0,2.5 ) and ( .2 , 3 ) .. ( .6 , 3.5 ) ; ( 0,2 ) circle ( .03 ) node[left ] @xmath365 ; ( 0,-2 ) circle ( .03 ) node[left ] @xmath366 ; choose contours @xmath325 as in figure [ fig:656 ] and let the functions @xmath317 be defined by @xmath367 where @xmath329 has a cut on the positive real axis and @xmath368 .
then it follows from proposition [ prop:537 ] that the concomitant conditions are satisfied .
the integrals define analytic functions on @xmath369 with possible branch points in @xmath4 and @xmath197 and we put the branch cut on the negative real axis .
( 0,0 ) circle ( .02 ) node[above right]@xmath4 ; ( 0,0 ) ( 1.5,0 ) ; ( -.3,0 ) ellipse ( .3 and .5 ) ; ( 0,0 ) .. controls ( 0,1 ) and ( .6,.4 ) .. ( 1.2,.4 ) ; ( 0,0 ) .. controls ( 0,-1 ) and ( .6,-.4 ) .. ( 1.2,-.4 ) ; ( 1.2,.4 ) ( 1.4,.4 ) ; ( 1.2,-.4 ) ( 1.4,-.4 ) ; ( -.3,.5 ) node[above]@xmath370 ; ( .6,.4 ) node[left ] @xmath371 ; ( .6,-.4 ) node[left]@xmath372 ; having found explicit expressions for the entries of @xmath302 we can return to the expression for @xmath8 given in .
the @xmath373 and @xmath374 matrices pick out certain linear combinations of the @xmath195 and @xmath317 , and we obtain @xmath375(x , y),\ ] ] with @xmath99 and @xmath100 given by @xmath376 @xmath377 the functions @xmath378 and @xmath379 have branch cuts on the negative real axis . hence we have to specify that we need the positive boundary value in in the case @xmath216 . using contour deformation and a simple substitution in the integral
one can see that @xmath99 and @xmath100 are given by the contour integrals and .
one can check that the derivatives of @xmath99 and @xmath100 have the following expressions : @xmath380 @xmath381 for @xmath382 and @xmath383 .
the contours @xmath102 and @xmath103 are given in figure [ fig:197 ] . from and the definition of the bilinear concomitant it follows that we can write @xmath384 \frac { e^{\frac { \tau x}{t } - \frac { x^{2}}{2t^{2 } } + t } } { e^{\frac { \tau y}{s } - \frac{y^{2}}{2s^{2 } } + s } } \,ds dt .
\label{eq:906}\end{aligned}\ ] ] we have the following straightforward identity for the expression inside the square brackets : @xmath390 the terms involving derivatives of the phase functions @xmath385 and @xmath386 can be simplified using integration by parts : @xmath391 by plugging and together with into we find @xmath392 e^{\eta_{q}(t ) } e^{\eta_{r}(s ) } \,ds dt \notag \\ & \qquad = ( x - y ) \int_{t \in \gamma_{0 } } \int_{s \in \widehat{\gamma}_{0 } } \frac{y^{3}}{s^{3}(xs - yt ) } e^{\eta_{q}(t ) } e^{\eta_{r}(s ) } \,ds dt.\end{aligned}\ ] ] from we then conclude that @xmath92 is given by @xmath393 t. claeys and a. kuijlaars , _ universality in unitary random matrix ensembles when the soft edge meets the hard edge _ , in `` integrable systems and random matrices : in honor of percy deift '' , contemporary mathematics * 458 * , amer . math . soc . , providence ri , 2008 , pp . 265280 .
p. deift , t. kriecherbauer , k.t .- r .
mclaughlin , s. venakides , and x. zhou , _ uniform asymptotics for polynomials orthogonal with respect to varying exponential weights and applications to universality questions in random matrix theory _ , comm .
pure appl . math .
* 52 * ( 1999 ) , 13351425 .
k. deschout and a.b.j .
kuijlaars , _
double scaling limit for modified jacobi - angelesco polynomials _ , in `` notions of positivity and the geometry of polynomials '' ( p. brndn , m. passare , and m. putinar , eds . ) , trends in mathematics , springer , basel 2011 , pp .
115161 .
m. duits and d. geudens , _ a critical phenomenon in the two - matrix model in the quartic / quadratic case _ , preprint arxiv:1111.2162 p.j .
forrester , _ log - gases and random matrices _ , princeton university press , princeton , nj , 2010 .
kuijlaars , _ multiple orthogonal polynomial ensembles _ , in : `` recent trends in orthogonal polynomials and approximation theory '' ( j. arves , et al . , eds . ) , contemp .
math . * 507 * , amer .
2010 , pp . 155176 .
kuijlaars , a. martnez - finkelshtein , and f. wielonsky , _ non - intersecting squared bessel paths and multiple orthogonal polynomials for modified bessel weights _ , comm .
* 286 * ( 2009 ) , 217275 .
w. van assche , j.s .
geronimo , and a.b.j .
kuijlaars , _ riemann - hilbert problems for multiple orthogonal polynomials _ , in : special functions 2000 : current perspective and future directions ( j. bustoz et al . , eds . ) , kluwer , dordrecht , 2001 , pp . | we consider angelesco ensembles with respect to two modified jacobi weights on touching intervals @xmath0 $ ] and @xmath1 $ ] , for @xmath2 . as @xmath3 the particles around @xmath4 experience a phase transition .
this transition is studied in a double scaling limit , where we let the number of particles of the ensemble tend to infinity while the parameter @xmath5 tends to @xmath6 at a rate of @xmath7 .
the correlation kernel converges , in this regime , to a new kind of universal kernel , the angelesco kernel @xmath8 .
the result follows from the deift / zhou steepest descent analysis , applied to the riemann - hilbert problem for multiple orthogonal polynomials .
department of mathematics , katholieke universiteit leuven , + celestijnenlaan 200b , 3001 leuven , belgium , + klaas.deschoutgmail.com , arno.kuijlaarswis.kuleuven.be |
the darrieus - landau instability , first described in the context of combustion , is a hydrodynamic instability that is caused by the thermal expansion of the burning gas @xcite .
it is characterized by the fact that the growth rate of the instability at the flame front is positive for perturbations of any wavelength , and is responsible for the curving of initially planar flames .
in addition to combustion , the darrieus - landau instability has been observed in different types of plasmas , from the interstellar medium to inertial confinement fusion , see , e.g. , refs .
@xcite .
another system in which combustion - like processes have been observed are crystals of molecular ( nano ) magnets .
these molecular magnets have large spin ( @xmath0 ) , and their crystals present an anisotropy , with an `` easy '' axis along which the spin will align . in the presence of an external magnetic field along the easy axis
, the two different orientations will not have the same energy , resulting in an effective skewed double - well potential ( see @xcite and references therein ) .
a crystal prepared in the metastable magnetic orientation , after local heating to overcome the activation energy , will see a propagation of the spin reversal , as the energy released by the spin flip will propagate to neighboring molecules , in a process dubbed a spin avalanche or magnetic deflagration @xcite .
the spin reversal can also occur without the activation energy being attained , through spin tunneling @xcite .
this phenomenon leads to the presence , for certain values of the magnetic field strength , of tunneling resonances that greatly increase the speed of propagation of the spin reversal front @xcite .
in previous work @xcite , we demonstrated that the propagation of this magnetic deflagration front is unstable , due to the fact that any distortion in the front increases the local magnetic field , creating a positive feedback . in this paper , we take a closer look at the stability of the front , and derive an analytical expression for the instability growth rate , in the limit of an infinitely thin front .
we also study the instability numerically , accounting for a finite magnetic front thickness .
our results are also compared to experimental data @xcite , taking into account the presence of tunneling resonances .
we consider a crystal of mn@xmath1-acetate , which has an effective spin number @xmath2 @xcite , placed in an external magnetic field @xmath3 aligned along the @xmath4-axis , which corresponds also to the easy axis .
the energy levels of molecular magnet can be described by the simplified spin hamiltonian @xcite @xmath5
where @xmath6 @xcite , @xmath7 is the gyromagnetic factor @xcite , and @xmath8 is the bohr magneton . the first term is due to the anisotropy of the crystal , while the second term describes the dipole interaction between the external magnetic field and the spin of the molecule .
we consider a crystal with all molecules initially in the @xmath9 metastable state , which is then locally heated at one extremity , and study the propagation of the spin reversal to the stable @xmath10 state . using hamiltonian ( [ eq : hamilton ] ) , we find the zeeman energy release @xmath11 @xmath12 and the energy barrier or activation energy @xmath13 , @xmath14 expressed in temperature units per molecule . for the particular external field @xmath15 , these quantities are depicted in fig .
[ fig : states ] together with the energy levels of mn@xmath1-acetate .
( color online ) energy levels of a molecular magnet mn@xmath1-acetate in an external magnetic field @xmath15 . a molecule initially in the metastable state @xmath9 ( on the left ) must overcome an energy threshold @xmath13 in order to relax to the stable state @xmath10 ( on the right ) .
after relaxation , the energy difference ( zeeman energy ) @xmath11 is released as heat.,width=316 ] the evolution of the system is governed by the heat transfer and the dynamics of the molecules in the metastable state .
the zeeman energy release is transformed into phonon thermal energy and is described as @xmath16 where @xmath17 is the phonon energy and @xmath18 is the thermal diffusion constant , which depends on temperature as @xmath19 .
the number of molecules in the metastable state evolves according to @xmath20 where @xmath21^{-1}$ ] is the thermal equilibrium concentration @xcite .
the prefactor in eq .
( [ eq : dndt ] ) stands for the thermal relaxation rate over the potential barrier @xmath13 , shown fig .
[ fig : states ] . in the simplest form
, it may be written as the arrhenius law @xmath22 generally speaking , the @xmath23 factor is not a constant , but depends on both longitudinal and perpendicular components of the magnetic field .
in addition , the presence of quantum tunneling resonances can increase the @xmath23 factor by several orders of magnitude for certain values of the magnetic field @xcite . in our analysis , as for experimental measurements , it is more convenient to work with the temperature variable @xmath24 rather than the phonon energy @xmath17 .
the molecular magnets must be kept at cryogenic temperatures in order to observe the spin reversal phenomenon .
the typical locking temperature for mn@xmath1-acetate lies in a region of a few degrees above absolute zero . under such conditions , the phonon energy is a strong function of temperature @xcite , @xmath25 where @xmath26 is a constant for this particular crystal type , @xmath27 is the debye temperature , and @xmath28 is the dimensionality of space .
we start by considering a stationary one - dimensional magnetic deflagration front , which propagates in the negative @xmath4 direction with a velocity @xmath29 . the internal front structure , consisting of the temperature , energy release , and molecular concentration ,
is shown in fig .
[ fig : profile ] .
( color online ) profile of the stationary deflagration front in a crystal of nanomagnets of mn@xmath1-acetate .
the front moves from right to left .
the external magnetic field is @xmath30 and the final temperature is @xmath31 , with @xmath32.,width=316 ] the final temperature @xmath33 behind the front is found from the energy conservation , @xmath34 where index @xmath35 corresponds to initially `` cold '' matter ( left side in fig .
[ fig : profile ] ) and index @xmath36 corresponds to the final `` hot '' ( right side ) of the front . in the case of incomplete burning ( i.e. , @xmath37 )
, this equation is a transcendental one and must be solved numerically .
we assume that the temperature behind the front @xmath33 is constant because the heat escape to an external heat sink for this particular configuration can be neglected .
the time for the spin reversal of the entire sample at the slowest deflagration rate approximately is @xmath38 .
the characteristic time of cooling is @xmath39 ( for this assumption we have used the thermal diffusion constant from ref .
@xcite , @xmath40 , and surface area @xmath41 ) .
note also that in ref .
@xcite , the time for the sample the return to the temperature of the bath was measured as @xmath42 .
therefore , since @xmath43 , we will neglect cooling effects in this study . to simplify further derivations we introduce dimensionless variables for the coordinate and temperature together with scaled activation and zeeman energies , @xmath44 and
define @xmath45 here , @xmath46 is a characteristic length of the problem .
then in the reference frame of the moving front the equations ( [ eq : heat])([eq : dndt ] ) form the dimensionless system @xmath47 where @xmath48 stands for the heat flux and @xmath49 is an eigenvalue of the stationary front .
the new variable @xmath48 allows us to write down the governing system as a set of first - order differential equations , which is important for stability analysis described in sec .
[ sec : stability ] . in order to calculate the stationary profiles depicted in fig .
[ fig : profile ] , we integrate the system ( [ eq : sys_stat ] ) from the left , `` cold '' side towards the right , `` hot '' side , and the eigenvalue @xmath50 is found by this shooting method @xcite , matching the results of numerical integration to the analytical solution given by eq .
( [ eq : t_f ] ) for the final temperature .
it should be noted that the assumption of a stationary front is valid when the front thickness is much smaller than the length of the sample .
the characteristic front width can be determined as the half - width of the energy release peak ; from fig .
[ fig : profile ] we find that @xmath51 .
the typical crystal size used in experiments is 1 @xcite , which is almost two orders of magnitude larger than the front width .
consequently , there is enough room to form a steady propagating front of magnetic deflagration .
another issue to mention is the ambiguity in determining the front velocity @xmath29 . resolving the stationary profiles ,
we compute the front eigenvalue @xmath50 , however in order to find @xmath29 we need to know the value of @xmath52 according to expressions ( [ denotes ] ) , as @xmath53 so far , the actual parameters @xmath54 and @xmath23 have not been measured , and we estimate the relation @xmath52 by fitting the velocity to experimental data @xcite . the dependence of @xmath23 on the magnetic field has been the subject of many studies , see , e.g. , refs .
@xcite . in this paper ,
we interpolate @xmath55 by fitting experimental data @xcite ( see fig . [
fig : res_vel ] ) using a gaussian function to model tunneling resonances as @xcite @xmath56\right\ } , \label{gammar}\ ] ] where @xmath57 is the resonance magnetic field , @xmath58 is a constant , and parameters @xmath59 , @xmath60 are the amplitude and the width of the resonance , respectively . according to experimental data @xcite shown in fig .
[ fig : res_vel ] , these parameters are calculated as @xmath61 with the estimate @xmath62{4e5}{\meter\squared\per\second\squared}$ ] .
( color online ) magnetic deflagration velocity @xmath29 for the planar front as a function of the longitudinal magnetic field @xmath3 .
the two peaks occur due to the tunneling resonance .
the markers correspond to experimental data extracted from ref .
@xcite and the line represents the fitted theoretical dependence.,width=316 ]
the propagation of the magnetic deflagration front is unstable @xcite as any distortion of the front increases the magnetic field where the front bends , as shown in fig . [ fig : instability ] .
this , in turn , leads to an increase of the front velocity , resulting in positive feedback . in this section
, we will take a closer look at the front stability properties .
( color online ) magnitude of the magnetic field in a 2d simulation for magnetic deflagration with a corrugated front .
the external magnetic field is parallel to the @xmath4-axis , @xmath30 .
the front moves along the @xmath4-axis in the positive direction .
the magnetization of the medium flips from @xmath63 in region 1 to @xmath64 in region 2 , with @xmath65 . the dipole field produced by the crystal results in an increase of the field at the tip of the humps.,width=316 ] in order to perform this analysis we need to make two assumptions .
first , we assume that the magnetization of the particular nanomagnet is produced by the spins of the molecules , with the behavior of the spins described by hamiltonian ( [ eq : hamilton ] ) .
the external magnetic field will significantly affect the eigenstates of hamiltonian only when the second term in eq .
( [ eq : hamilton ] ) becomes comparable to the first term , i.e. , for an anisotropy field @xmath66 .
for mn@xmath1-acetate , @xmath67 is of the order of @xcite . in our study
, we focus on fields strengths much lower than the anisotropy field , where the dependence of @xmath68 on @xmath69 is negligible , see fig . 2 in ref .
therefore , we suppose that the amplitude of the crystal magnetization @xmath68 does not depend on the strength of the external field .
second , we assume that front width is infinitely thin , so that the profiles presented in fig .
[ fig : profile ] reduce to step functions , separating the cold and the hot regions of the crystal .
we will remove these restrictions in the next section and consider the instability properties for a continuous front structure .
with respect to the front propagation and easy axis , multiple mutual orientations are possible .
we consider the two principal cases : the front propagation is aligned with , fig .
[ fig : linear_geometry](a ) , or perpendicular to , fig .
[ fig : linear_geometry](b ) , the easy axis of the crystal .
( color online ) possible orientations of the front propagation vs. the crystal easy axis : ( a ) most common geometry used in experiments , where the easy axis is parallel to the direction of the front propagation ; ( b ) hypothetical orientation of the crystal , where the easy axis is perpendicular to the direction of the front propagation .
the external magnetic field @xmath70 is oriented along the easy axis of the crystal.,width=316 ] the position of a planar front , propagating with a constant velocity @xmath29 , is given by @xmath71 . the front
is then perturbed with a superposition of fourier modes written as @xmath72 where @xmath73 is the instability growth rate , @xmath74 is the wave number , and @xmath75 is the wavelength of the perturbations . if @xmath76 the perturbations grow in time and the front becomes unstable ; in the opposite case , @xmath77 , the front is remains stable .
the imaginary part of @xmath73 leads to oscillations and pulsations of the front @xcite ; in this paper we consider the case when @xmath78 , such that @xmath73 takes only purely real values .
we start with the case when the front propagates parallel to the easy axis and the magnetic field , fig .
[ fig : linear_geometry](a ) .
this is a common crystal orientation in experimental studies @xcite .
magnetization flips from @xmath63 in the cold region before the front to @xmath64 in the hot region behind the front , see fig .
[ fig : linear_geometry](a ) .
the deformation of the front induces perturbations in the magnetic field , @xmath79 the magnetic field inside the crystal is governed by the stationary maxwell equations for nonconducting media , @xmath80 with the relation @xmath81 , where @xmath82 is the magnetic constant .
far from the front all the perturbations must vanish , such that the magnetic field perturbations along @xmath4 can be written as @xmath83 .
boundary conditions for the magnetic field on the front interface are @xmath84 & = 0 , & \widehat{\mathbf{a}}_{n}\times \left[\mathbf{h}\right ] & = 0,\end{aligned}\ ] ] where @xmath85 \equiv f_{2}-f_{1}$ ] designates the difference of any value @xmath86 across the front and the normal vector to the perturbed front is @xmath87 . resolving maxwell s equations ( [ eq : maxwell ] ) together with the boundary conditions ,
we find the relation between the magnetic fields ahead and behind the front .
taking @xmath88 we find that @xmath89 which leads to an increase of the magnetic field at the tip of the hump . within the linear stability problem ,
the perturbation of the front velocity is given by @xmath90 , where @xmath91 @xcite .
it yields the dispersion relation in a very concise form as @xmath92 this result means that an infinitely thin magnetic deflagration front is unstable with respect to perturbations of all wavelength , since @xmath93 for any @xmath94 .
mathematically , this relation @xmath95 is similar to the darrieus - landau instability @xcite .
next , we consider the crystal configuration where the front propagation is perpendicular to the easy axis of the crystal , fig .
[ fig : linear_geometry](b ) . in this case
, the magnetization varies as @xmath96 and the external magnetic field is given as @xmath97 . using the same approach as above
, we obtain the dispersion relation @xmath98 hence such a configuration results in a stable propagation of the magnetic deflagration wave .
characteristic values of the relative strength of the instability @xmath99 may vary significantly for different materials , depending on the magnetization @xmath100 and front velocity sensitivity @xmath101 .
we therefore expect noticeable magnetic instabilities in two cases , when either @xmath100 or @xmath101 are high .
strong magnetization can be found in ferromagnetic materials , so this instability might affect propagation of the domain walls . in magnetic nanomagnets ,
the magnetization is relatively weak , @xmath102 @xcite , but the velocity slope @xmath101 theoretically can reach infinite values at the tunneling resonances @xcite .
in sec . [ sec : analytical ] , we found that the magnetic deflagration front is unstable in the infinitely thin front limit .
however , such a method does not provide any characteristic length scale for the instability nor the strength of the instability or its dependence on the external magnetic field . here
, we investigate the instability properties taking into account a finite front width and the continuous structure of the deflagration front obtained in sec .
[ sec : deflagration ] . for a finite front thickness
, the magnetic field @xmath69 , together with the all other variables , changes continuously within the front .
we introduce the magnetic vector potential @xmath103 defined from @xmath104 , such that the first maxwell equation @xmath105 is satisfied automatically . for a planar front
, the vector potential has only one component @xmath106 and for uniform field @xmath70 it reduces to @xmath107 .
consequently , the magnetic field components are @xmath108 the second maxwell equation @xmath109 can be rewritten as @xmath110 where we assume that the magnetization changes proportionally to the ratio of the metastable molecules @xmath111 .
we introduce the dimensionless magnetic potential defined as @xmath112 and a new variable @xmath113 in order to have differential equations of the first order only .
next , we apply a small perturbation so that every variable is written in as @xmath114 . after straightforward calculations ,
the linearized equations ( [ eq : sys_stat ] ) and ( [ eq : magvec2 ] ) can be written in matrix form as @xmath115 where @xmath116 is the vector of perturbations .
@xmath117 is a @xmath118 matrix of the coefficients of the system of differential equations , @xmath119 where the matrix components are @xmath120 , \\
d_{31 } & \equiv \beta\psi\theta^{\beta-1}-jw\frac{\theta}{\theta^2 } \left(n - n_{\mathrm{eq}}-n^2_{\mathrm{eq}}\frac{\delta}{\theta } e^{\delta/\theta}\right ) \\ &
\quad + s\theta^\alpha+k^2\theta^{\alpha-\beta } , \\
d_{34 } & \equiv \frac{k}{b_z } \left [ w(n - n_{\mathrm{eq } } ) \left ( 1+b_{0z}\frac{c_1}{\theta } \right ) -c_2 \right ] , \end{aligned}\ ] ] with @xmath121 introduced for brevity . in the equations above , @xmath122 is a dimensionless external magnetic field , and @xmath123 and @xmath124 are the scaled instability growth rate and perturbation wave number , respectively . some of the matrix coefficients in ( [ eq : d_matrix ] ) are known from the stationary profile , while the others depend on @xmath125 and @xmath126 as parameters . in order to find the instability dispersion relation @xmath127
, we apply the same method as in similar studies of instabilities @xcite .
first , we search for solutions to the system ( [ eq : fullsys ] ) in the uniform regions , where all the coefficients in matrix @xmath128 are constant . in this case the perturbations decay exponentially as @xmath129 where the @xmath130 are the so - called system modes and @xmath131 are constant perturbation amplitudes . substituting eq .
( [ eq : prtrdec ] ) into eqs .
( [ eq : fullsys ] ) , we obtain @xmath132 we compute the eigenvalues @xmath130 and the corresponding eigenvectors @xmath131 .
we consequently obtain five modes for the cold and the hot sides of the front , although not all of them are physical . in order to pick out the physical eigenvectors
, we use the condition that perturbations must vanish far from the front as @xmath133 . in other words
, we consider eigenvectors @xmath131 for which @xmath134 at @xmath135 or @xmath136 at @xmath137 .
if the problem is self - consistent , there will be exactly five modes @xmath130 satisfying these conditions , usually three modes on one side and two on the other . after that we integrate eqs .
( [ eq : fullsys ] ) from the front boundaries using @xmath131 as boundary conditions .
we match the results of the integration at the point of maximal energy release @xmath138 ( shown in fig .
[ fig : profile ] ) . generally speaking this matching point can be chosen in a different way without affecting the final results , however the current choice minimizes the numerical integration errors @xcite . at this point
the integrated amplitudes constitute a matrix and the dispersion relation @xmath139 is obtained when the matrix determinant becomes equal to zero .
( color online ) dimensionless instability growth rate versus the dimensionless wave number for different magnetic fields .
values @xmath140 , @xmath141 , and @xmath142 are shown only for the dispersion relation for @xmath143.,width=316 ] the dispersion relation @xmath139 is shown in fig .
[ fig : dispersion ] , for several magnitudes of external magnetic field .
it has a parabola - like shape similar to the one obtained for the darrieus - landau instability in combustion and laser ablation @xcite . in the region of small wave numbers
the instability displays a strong increase of the growth rate against the variation of the wave number .
then , at a certain wave number @xmath141 , the instability growth rate reaches its maximum @xmath142 .
after that , the instability becomes weaker until it vanishes at @xmath140 .
as in the case of the darrieus - landau instability , the stabilization is attributed to the final front width due to thermal conduction .
another important outcome from fig .
[ fig : dispersion ] is that the instability is stronger for relatively weak fields .
this can be qualitatively explained in the following way .
the instability is caused by the dipole field created by the crystal magnetization . in our model
this magnetization does not depend on the external magnetic field .
hence for weak fields ( @xmath144 ) the magnetization at the curved front creates a relatively strong dipole field as compared to the external field .
this , in turn , increases the front velocity driving the instability to grow further . on the other hand , for high external fields , the increase due to
crystal magnetization is relatively weak , leading to a much smaller increase of the front velocity .
the influence of the magnetic field is better shown in figs .
[ fig : smax ] and [ fig : cutoff ] , where we present the maximum instability growth rate as well as the cutoff wavelength and wavelength at the maximum as a function of the external magnetic field . in both these figures we use dimensional quantities , allowing a simpler comparison to experiments . as
discussed above , we observe strong decrease of the instability growth rate with respect to the magnetic field , fig .
[ fig : smax ] .
the two peaks correspond to quantum tunneling at resonant magnetic fields . as follows from fig .
[ fig : smax ] , the instability is the strongest and can be observed in the region of very small fields . on the other hand , in fig .
[ fig : cutoff ] we see that the wavelength @xmath145 corresponding to the maximal growth rate can be rather high in that range of magnetic field .
( color online ) maximum of the instability growth rate versus magnetic field for mn@xmath1-acetate.,width=316 ] ( color online ) cutoff wavelength and wavelength at the maximum , for mn@xmath1-acetate .
the horizontal dashed gray line shows the typical sample width in experiments @xcite.,width=316 ] it is important to emphasize that the scaling parameters @xmath54 and @xmath58 are difficult to measure directly .
it is also difficult to estimate these quantities using theoretical models .
for instance , values of @xmath146 , @xmath147 , and @xmath148 , in units of , were found using different heat transfer models @xcite . here , we used @xmath149 and @xmath150{0.04}{\meter\squared\per\second}$ ] in figs .
[ fig : smax ] and [ fig : cutoff ] . our analysis could be used to measure the product @xmath151 , as the instability of the magnetic deflagration can be detected in two ways : ( i ) direct observation , using magneto - optical imagining @xcite , where the planar front becomes corrugated with a parabola - like shape as predicted by numerical simulations @xcite ; ( ii ) the instability can be detected by measuring deflagration speed , as a curved front propagates faster then a planar one @xcite . a stationary propagation of the curved front was predicted and explained within non - linear theory in ref .
similar results were obtained using direct numerical simulation in ref .
@xcite and ref .
importantly , such an instability occurs regardless of the presence of resonances
. meanwhile , turbulent propagation of the front observed in ref .
@xcite occurs when the field is resonant ( where the theoretical model predicts nearly infinitely fast relaxation ) .
propagation of the front at resonant field strengths ( when the relaxation rate is extremely high ) must be taken as a separate problem and is not considered here .
perturbations with different @xmath94 have different growth rates , see eq .
( [ eq : disp1 ] ) .
therefore , the perturbation with highest growth rate @xmath73 develops faster and leads to a stationary , curved front ( see the numerical simulations in ref .
@xcite ) .
the growth rate stands for the characteristic time needed for the instability to develop from a planar front to a stationary , curved front . for a possible observation of the instability , this time should be much smaller than the time of propagation of the magnetic deflagration front within the crystal .
in particular , recent experiments @xcite were performed with a sample size of @xmath152 with @xmath153 , corresponding to a typical time of about .
[ fig : smax ] predicts @xmath154 for such a field magnitude , hence the instability might have enough time to develop .
however , the wavelength where the growth rate is maximal is larger than the sample width , depicted as a dashed line in fig .
[ fig : cutoff ] , which leads to a decrease of the instability growth rate , since the wavelength of the perturbation can not exceed the width of the sample .
in addition , this means that the front may be only slightly curved . under restrictions of the size of the sample
, the resonant magnetic field makes the observation of the magnetic instability more plausible .
we see that the resonance enhances the instability growth rate in two ways .
first it increases the growth rate @xmath155 , as demonstrated by the peaks in fig .
[ fig : smax ] .
second , it decreases the cutoff wavelength and the wavelength at the maximum ( fig .
[ fig : cutoff ] ) , such that the estimated values are within the range of the dimensions of the crystal .
in addition to that , the amplitude of the resonance can be increased significantly by applying a transverse magnetic field , effectively increasing @xmath58 .
if the instability becomes strong enough , acceleration of the deflagration front can create a weak shock wave ahead of the front , which might lead to the deflagration - to - detonation transition @xcite . in the weak detonation regime , the front propagates at the speed of sound , and such a spin reversal phenomena
has already been observed experimentally near a resonance @xcite .
we have studied the front instability in magnetic deflagration and found that it behaves in a similar fashion to the darrieus - landau instability in combustion . that is , in the limit of an infinitely thin front , the instability has a positive growth rate at all wavelengths .
the dispersion relation of the growth rate as a function of the instability wave number is also similar to that of the darrieus - landau instability .
the case of a finite - width front was explored numerically , and we found that the instability should be observable in current experimental setup , in particular close to a tunneling resonance , the latter resulting in a smaller value of the wavelength at which the instability will have the highest growth rate . analysing the effect of the direction of propagation of the front with respect to the easy axis , we showed that the instability would not grow for a perpendicular front .
we suggest two different experimental setups ( a ) and ( b ) , see fig .
[ fig : linear_geometry ] .
theory predicts an unstable front in case ( a ) , resulting in a faster propagation of the front as result , and a stable front in case ( b ) . by comparing velocities of the magnetic deflagration of these two different geometries one can verify the presence of the instability .
signatures of the presence of the instability might also explain some previous experimental results .
for instance , for strong longitudinal fields , the velocities recorded in the experiments are higher than the theoretical predictions @xcite , which could be explained by the effect of the instability on front speed .
( note that while the front would be curved due to the instability , the propagation speed of the curved front will also be steady .
@xcite ) likewise , the front broadening also observed in ref .
@xcite could be explained by the front instability .
finally , we showed that there is a relation between the instability and the front velocity @xmath29 , diffusion constant @xmath54 , and the thermal relaxation rate @xmath23 .
experimental studies of the instability could help measure the values of the latter two parameters .
35ifxundefined [ 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/s0370 - 1573(99)00081 - 2 [ * * , ( ) ] link:\doibase 10.1146/annurev.fluid.38.050304.092153 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1634969 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2356846 [ * * , ( ) ] http://stacks.iop.org/0004-637x/623/i=1/a=337 [ * * , ( ) ] http://stacks.iop.org/0004-637x/652/i=2/a=1331 [ * * , ( ) ] http://stacks.iop.org/0741-3335/49/i=12b/a=s49 [ * * , ( ) ]
\doibase http://dx.doi.org/10.1063/1.2898402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.016103 [ * * , ( ) ] link:\doibase 10.1002/anie.200390099 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.147201 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.217205 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.054410 [ * * , ( ) ] http://stacks.iop.org/0295-5075/84/i=6/a=67010 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.027203 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1007/s10909 - 005 - 6016 - 3 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.172410 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.092403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.144408 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.217206 [ * * , ( ) ] link:\doibase 10.1021/ja00015a057 [ * * , ( ) ] link:\doibase 10.1021/ja00058a027 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.214417 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.094403 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1103/physrevb.78.020405 [ * * , ( ) ] link:\doibase 10.1103/physreve.80.046403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.61.12200 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.180401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.56.11102 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.014409 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.064413 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.207208 [ * * , ( ) ] | the magnetic instability at the front of the spin avalanche in a crystal of molecular magnets is considered .
this phenomenon reveals similar features with the darrieus - landau instability , inherent to classical combustion flame fronts .
the instability growth rate and the cut - off wavelength are investigated with respect to the strength of the external magnetic field , both analytically in the limit of an infinitely thin front and numerically for finite - width fronts .
the presence of quantum tunneling resonances is shown to increase the growth rate significantly , which may lead to a possible transition from deflagration to detonation regimes .
different orientations of the crystal easy axis are shown to exhibit opposite stability properties .
in addition , we suggest experimental conditions that could evidence the instability and its influence on the magnetic deflagration velocity . |
an unprecedented volume of observational data and information regarding the distribution and properties of optical sources in the universe is becoming available from ongoing and future sky surveys , both imaging and spectroscopic . these include boss ( baryon oscillation spectroscopic survey ; @xcite ) , des ( dark energy survey ; @xcite ) , kids ( kilo - degree survey ; @xcite ) , sumire , desi ( dark energy spectroscopic instrument ; @xcite ) , 4most ( 4-meter multi - object spectroscopic telescope ; @xcite ) , j - pas ( javalambre - physics of the accelerated universe astrophysical survey ; @xcite ) , lsst ( large synoptic survey telescope ; @xcite ) , euclid @xcite , and wfirst ( wide - field infrared survey telescope ; @xcite ) .
combined with microwave background observations from the planck satellite and ground - based telescopes , such as act ( atacama cosmology telescope ) and spt ( south pole telescope ) , the mining of these and other datasets , such as resulting from new radio surveys , e.g. , vlass ( very large array sky survey ) and x - ray surveys , are expected to yield a host of cosmological and astrophysical insights .
there are several foundational cosmological questions that the datasets will directly address .
perhaps the most pressing is the mysterious cause of the accelerated expansion of the universe whether it is due to dark energy or a modification of general relativity .
in addition , the observations will also bear on the ultimate nature of dark matter , provide information on the physics of the early universe by probing primordial fluctuations , and enhance our knowledge of the neutrino family , the lightest known massive particles in the universe . aside from these basic cosmological questions ,
the survey data provides an enormous resource for attacking a very large number of astrophysical problems related primarily to understanding the formation of complex structure in the universe . in order to extract the full extent of information from these remarkable surveys , a similar level of effort
must be undertaken in the realm of theory and modeling . to attain the necessary realism and accuracy , sophisticated , large scale simulations of structure formation
must be carried out .
these simulations address a large variety of tasks : providing predictions for many different cosmological models to solve the inverse problem related to determining cosmological parameters , investigating astrophysical and observational systematics that could mimic physics effects of the dark sector , enabling careful calibration of errors ( by providing precision covariance estimates ) , testing , optimizing , and validating observational strategies with synthetic catalogs , and finally , exploring new and exciting ideas that could either explain puzzling aspects of the observations ( such as cosmic acceleration ) or help to motivate and design the implementation of new types of cosmological probes .
the simulations have to be large enough in volume to cover the observed universe ( or at least a large part of it ) and at the same time possess sufficient mass and force resolution ( a spatial dynamic range of roughly a million ) to resolve objects down to the smallest relevant scales .
as one example , a recently constructed synthetic galaxy and quasar catalog for desi is shown in figure [ syn_cat ] .
this catalog was based on a 1.1 trillion - particle hacc simulation , with a box - size of @xmath0 gpc ( @xcite ) .
the diverse uses of simulations outlined above also demand fast turn - around times , as not one such simulation is required but , eventually , many hundreds to hundreds of thousands .
the simulation codes , therefore , have to be capable of exploiting the largest supercomputing platforms available today and into the future .
viewed as a single entity , the field of ` modern ' computational cosmology ( @xcite , @xcite ) has largely kept pace with the growth of computational power , but new challenges will need to be faced over the next decade .
this is due to the failure of dennard scaling ( @xcite ) , which underlay the success of moore s law for about two decades . as a consequence of the fact that single - core clock rate and performance have stalled since 2004/5
, the design of future microprocessors is branching into several new directions , to overcome the related performance bottlenecks ( @xcite ) the key constraint being set by electrical power requirements . the resulting impact on large supercomputers
is already being seen ; aside from the familiar large multi - core processor clusters , two major approaches can be easily identified . the first approach is the large homogeneous system , built around a ` system on chip ' ( soc ) design or around many - core nodes , with concurrencies in the multi - million range the ibm blue gene / q is an excellent example of such an approach ; specific examples include sequoia ( 20 pflops peak ) at lawrence livermore national laboratory and mira ( 10 pflops peak ) at argonne national laboratory , supporting up to @xmath1 million - way concurrency ( sequoia ; half of that on mira ) . the second route is to take a conventional cluster but to attach computational accelerators to the cpu nodes .
the accelerators can range from the ibm cell processor ( as on los alamos national laboratory s roadrunner , first to break the petaflop barrier ) , to gpus as on titan ( 27 pflops peak ) at oak ridge national laboratory , to the intel xeon phi coprocessor as on stampede ( 10 pflops peak ) at the texas advanced computing center .
future evolution of supercomputer systems will almost certainly involve further branching in the space of possible architectures . because development of the supercomputing software environment is likely to lag significantly behind the pace of hardware evolution , it appears prudent , if not essential , to develop a code design philosophy and implementation plan for cosmological simulations that can be reconfigured relatively quickly for new architectures , has support for multiple programming paradigms , and at the same time , can extract high levels of sustained performance . with this challenge in mind ,
we have designed hacc ( hardware / hybrid accelerated cosmology codes ) , an n - body cosmology code framework that takes full advantage of all available architectures .
this paper presents an overview of hacc ` theory and practice ' by covering the methods employed , as well as describing important components of the implementation strategy .
an example of hacc s capabilities is shown in figure [ zoom ] , a recent large cosmological run on mira , evolving more than 1 trillion particles ; this is the same simulation on which the results of figure [ syn_cat ] are based . box - size simulation with hacc on 32 blue gene / q racks .
the force resolution is @xmath2 kpc and the particle mass , @xmath3 m@xmath4 .
the image , taken during a late stage of the evolution , illustrates the global spatial dynamic range covered by the simulation , @xmath5 , although the finer details are not resolved in this visualization.,width=278 ] cosmological simulations can be broadly divided into two classes : gravity - only n - body simulations and ` hydrodynamic ' simulations that incorporate gasdynamics and models of astrophysical processes .
since gravity dominates at large scales , and dark matter outweighs baryons by roughly a factor of five , n - body simulations are an essential component in modeling the formation of structure .
several post - processing strategies can be used to incorporate missing physics in gravity - only codes , such as the halo occupation distribution ( hod ) approach ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , subhalo / halo abundance matching ( s / ham ) ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) or semi - analytic modeling ( sam ) ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) for adding galaxies to the simulations .
whenever a more detailed understanding of the dynamics of baryons is required , gasdynamic , thermal , and radiative processes ( among others ) must be modeled , as well as processes such as star formation and local feedback mechanisms ( outflows , agn / sn feedback ) .
a compact review of cosmological simulation techniques and applications can be found in @xcite .
a concise review of phenomenological galaxy modeling is given in @xcite .
carrying out a fully realistic first principles simulation program for all aspects of modeling cosmological surveys will not be possible for quite some time .
the required gasdynamic simulations are very expensive and there is considerable uncertainty in the modeling and physics inputs .
progress is nevertheless possible by combining high - resolution , large - volume n - body simulations and post - processing inputs from simulations that include gas physics to build robust phenomenological models .
the parameters of these models would be determined by a set of observational constraints ; yet other observations would then function as validation tests .
the hacc approach assumes this starting point as an initial design constraint , but one that can be relaxed in the future .
the overall structure of hacc is based on the realization that a large - scale computational framework must not only meet the challenges of spatial dynamic range , mass resolution , accuracy , and throughput , but , as already discussed , be cognizant of disruptive changes in computational architectures .
as an early validation of its design philosophy , hacc was among the pioneering applications proven on the heterogeneous architecture of roadrunner ( @xcite ) , the first computer to break the petaflop barrier . with its multi - algorithmic structure , hacc allows the coupling of mpi with a variety of local programming models mpi+`x ' to readily adapt to different platforms .
currently , hacc is implemented on conventional and cell / gpu - accelerated clusters , on the blue gene / q architecture ( @xcite ) , and has been run on prototype intel xeon phi hardware .
hacc is the first , and currently the only large - scale cosmology code suite world - wide , that has been demonstrated to run at full scale on _ all _ available supercomputer architectures .
another key aspect of the hacc code suite is an inbuilt capability for fast `` on the fly '' or _ in situ _ data analysis . because the raw data from each run can easily be at the petabyte ( pb ) scale or larger , it is essential that data compression and data analysis be maximized to the extent possible , before the code output is dumped to the file system . in order to comply with storage and data transfer bandwidth limitations ,
the amount of data reduction required is roughly two orders of magnitude .
hacc s _ in situ _ data analysis system is designed to incorporate a number of parallel tools such as tools for generating clustering statistics ( power spectra , correlation functions ) , tessellation - based density estimators , a fast halo finder @xcite with an associated merger tree capability , real - time visualization , etc .
the hacc framework has been used to generate a number of large simulations to carry out a variety of scientific projects .
these include a suite of 64 billion particle runs for predicting the baryon acoustic oscillation signal in the quasar ly-@xmath6 forest , as observed by boss ( @xcite ) , a high - statistics study of galaxy group and cluster profiles , to better establish the halo concentration - mass relation ( @xcite ) , tests of a new matter power spectrum emulator ( @xcite ) , and a study of the effect of neutrino mass and dynamical dark energy on the matter power spectrum ( @xcite ) .
results from other simulation runs will be available shortly .
these range from simulation campaigns in the @xmath7 billion particle range per simulation , e.g. , simulations for determining sampling covariance for boss ( @xcite ) , to individual simulations in the @xmath8 billion particle class . as an example of the latter , a recently completed simulation on titan used 550 billion particles to evolve a 1.3 gpc periodic box , with mass resolution , @xmath9 m@xmath10 , and force smoothing scale of @xmath11 kpc ( @xcite ) .
the paper is organized as follows . in section [ sec : feat ] we introduce the basic hacc design and algorithms , focusing on how portability and scaling on very large supercomputers is achieved , including discussions of practical matters such as memory management and parallel i / o . a significant aspect of supercomputer architecture is diversity at the level of the computational nodes .
as mentioned above , the hacc design adopts different short - range solvers for different nodal architectures , and this feature is discussed separately in section [ sec : specs ] .
section [ sec : verif ] presents some results from our extensive code verification program , showcasing a comparison test with gadget-2 @xcite , one of the most widely used cosmology codes today .
the _ in situ _ analysis tool suite is covered in section [ sec : tools ] , and selected performance results are given in section [ sec : perf ] .
we conclude in section [ sec : conc ] with a recap and a discussion of future evolution paths for hacc .
in the standard model of cosmology , structure formation at large scales is described by the gravitational vlasov - poisson equation ( @xcite ) , a 6-d partial differential equation for the liouville flow ( [ le ] ) of the one - particle phase space distribution , arising from the non - relativistic limit of the vlasov - einstein set of equations , @xmath12 where the poisson equation encodes the self - consistency of the evolution : @xmath13 the expansion history of the universe is given by the time - dependence of the scale factor @xmath14 governed by the specifics of the cosmological model , the hubble parameter , @xmath15 , @xmath16 is newton s constant , @xmath17 is the critical density , @xmath18 , the average mass density as a fraction of @xmath17 , @xmath19 is the local mass density , and @xmath20 is the dimensionless density contrast , @xmath21 @xmath22 in general , the vlasov - poisson equation is computationally very difficult to solve as a partial differential equation because of its high dimensionality and the development of nonlinear structure including complex multistreaming on ever finer scales , driven by the gravitational jeans instability .
consequently , n - body methods , using tracer particles to sample @xmath23 are used ; the particles follow newton s equations , with the forces given by the gradient of the scalar potential @xmath24 ( cf .
@xcite ; for an early comparison of direct and particle methods in a non - gravitational context , see @xcite ) .
the cosmological n - body problem is characterized by a very large value of the number of interacting particles and a very large spatial dynamic range . if one wishes to track billions of galaxy - hosting halos at a reasonable mass resolution , then hundreds of billions to trillions of tracer particles
are required .
since gravity can not be shielded , this obviously precludes the use of brute - force direct particle - particle algorithms for the particle force computation .
popular alternatives include pure particle - based methods ( tree codes ) or multi - scale grid - based methods ( amr codes ) , or hybrids of the two ( treepm , particle - particle particle - mesh , p@xmath25 m ) .
it is not our purpose here to go into many details of the algorithms and their implementations ; good coverage of the background material can be found in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
the hacc design approach acknowledges that , as a general rule , particle and grid - based methods both have their limitations . for physics , algorithmic , and data structure reasons ,
grid - based techniques are better suited to larger ( ` smooth ' ) lengthscales , with particle methods possessing the opposite property .
this suggests that higher levels of code organization should be grid - based , interacting with particle information at a lower level of the computational hierarchy . following this central idea
, hacc uses a hybrid parallel algorithmic structure , splitting the gravitational force calculation into a specially designed grid - based long / medium range spectral particle - mesh ( pm ) component that is retained on all computational architectures , and an architecture - tunable particle - based short / close - range solver .
the spectral pm component can be viewed as an upper layer that is implemented using c++/c / mpi , essentially independent of the target architecture , whereas , the bottom or node level is where the short / close - range solvers reside .
these are chosen and optimized depending on the target architecture and use different local programming models as appropriate .
of the 6 orders of magnitude required for the spatial dynamic range for the force solver , the grid is responsible for 4 orders of magnitude , while the particle methods handle the critical 2 orders of magnitude at the shortest scales where particle clustering is maximal and the bulk of the time - stepping computation takes place .
the short - range solvers can employ direct particle - particle interactions , i.e. , a p@xmath25 m algorithm @xcite , as on some accelerated systems , or use both tree and particle - particle methods as on the ibm blue gene / q ( ` pptreepm ' with a recursive coordinate bisection ( rcb ) tree ) . as two extreme cases , in non - accelerated systems , the tree solver provides very good performance but has some complexity in the data structure , whereas for accelerated systems , the local @xmath26 approach is more compute - intensive but has a very simple data structure , better - suited for computational accelerators such as cells and gpus .
the availability of multiple algorithms within the hacc framework also allows us to carry out careful error analyses , for example , the p@xmath25 m and the pptreepm versions agree to within @xmath27 for the nonlinear power spectrum test in the code comparison suite of @xcite ( see section [ sec : verif ] for more details ) .
this level of error control easily meets the minimal requirements set by the increased statistical power of next - generation survey observations . in the following
we first describe the long / medium - range force solver employed by hacc .
as mentioned above , this solver remains unchanged across all architectures . after doing this , we provide details of the architecture - specific short - range solvers , and the sub - cycled time - stepping scheme used by the code suite .
an important aspect of large - volume cosmological simulations is that the density distribution is very highly clustered , with an overall topology descriptively referred to as the `` cosmic web '' .
the clustering is such that the maximum distance moved by a particle is roughly 30 mpc , very much smaller than the overall scale of the simulation box ( @xmath28gpc ) . with a 3-d domain decomposition , each ( non - cubic ) nodal volume ( mpi rank ) is roughly of linear size 50 - 500 mpc , depending on the simulation run size .
the idea behind particle overloading is to ` overload ' the node with particles belonging also to a zone of size roughly 3 - 10 mpc extending out from the nominal spatial boundary for that node ( so - called `` passive '' particles ) .
note that copies of these particles essentially a replicated particle cache , roughly analogous to ghost zones in pde solvers will also be held by other processors , in only one of which will they be `` active '' , hence the use of the term ` overloading ' . because more than one copy of these particles is held by neighboring domains ,
overloading is not the same as the guard zone conventionally used to reduce communication in particle codes .
the point of having this particle cache is two - fold .
first , for a number of time steps no particle communication across nodes is required .
additionally , the cached particle ` skin ' allows particle deposition and force interpolation for the spectral particle - mesh method to be done using information entirely local to a node , thus grid communication is also reduced .
the particle cache is refreshed ( replacement of passive particles in each computational domain by active particles from neighboring domains ) at some given number of time steps .
this involves only nearest neighbor communication and the penalty is a trivial fraction of the time spent in a single global time step .
the second advantage of overloading is that when a short - range solver is added to each computational domain , no communication infrastructure is associated with this step .
thus , in principle , short - range solvers can be developed completely independently at the node level , and then inserted into the code as desired .
consequently , hacc s weak scaling is purely a function of the properties of the long - range solver . the overloaded pm solver is formally exact
each node sends its local density field ( computed with active particles only ) to the global spectral poisson solver , which then returns the force for both active and passive particles to each node . the short - range force calculation for passive particles
is computed in the same way as for the active particles , except that passive particles , which are closer to the outer domain boundary than the short - range force - matching scale , @xmath29 ( defined in the following section ) , do not have their short - range forces computed , and are subject to only long - range forces .
this avoids force anisotropy near the overloaded zone boundary , at the expense of a force error on the passive particles that are close to the edge of the boundary . since the role of the passive particles is primarily to provide a buffer
` boundary condition ' zone for the active particles near the nodal domain s active zone boundary , consequences of this error are easy to control .
overloading has two associated costs : ( i ) loss of memory efficiency because of the overloading zone , and ( ii ) the just - discussed numerical error for the short - range force that slowly leaks in from the edge of the outer ( passive particle ) domain boundary . in cosmology applications ,
the memory inefficiency can be tolerated to the point where the memory in passive and active particles is roughly equal , but this is not a limitation in the majority of the cases of interest .
the second problem is easily mitigated by balancing the boundary thickness against the frequency of the particle cache refresh , `` recycling '' the passive particles after some finite number of time - steps , chosen to be such that the refresh time is smaller than the error diffusion time : each domain gets rid of all of its passive particles and then refreshes the passive zone with ( high accuracy ) active particles exchanged from nearest - neighbor domains .
conventional particle - based codes employ a combination of spatial and spectral techniques .
the cloud - in - cell ( cic ) scheme used for particle deposition is an example of a real space operation , whereas the fast fourier transform ( fft)-based poisson solver is a @xmath30-space or spectral operation .
the spatial operations in a typical particle code are the particle deposition , the force interpolation , and the finite - differences for computing field derivatives .
the spectral operations include the influence ( or pseudo - green ) function fft solve , digital filtering , and spectral differentiation techniques .
spatial techniques are often less flexible and more tedious to implement than their spectral counterparts . also , higher - order spatial operations can be complicated and lead to messy and dense communication patterns ( e.g. , indirection ) .
accurate p@xmath25 m codes are usually run with triangle - shaped - cloud ( tsc ) , a high - order spatial deposition / filtering scheme , as well as with high - order spatial differencing templates . in terms of grid units ,
the cic deposition kernel filters roughly at the level of two grid cells or less with a large amount of associated `` anisotropy noise '' ; tsc filters at about three grid cells with much reduced noise .
the resulting long - range / short - range force matching is usually done at four / five grid cells or so ( but can be as small as three grid cells ) .
treepm codes can move the matching point to a larger number of grid cells because of the inherent speed of the tree algorithm , so they can continue to use cic deposition ( with some additional gaussian filtering ) .
hacc allows the use of both p@xmath25 m and treepm algorithms , using a shorter matching scale than most treepm codes .
one advantage of the shorter matching scale is that a low - order polynomial expression can be used in the force representation , which greatly speeds up evaluation of the force kernel .
behavior is tuned to occur at a separation of three grid cells , which sets the force - matching scale , @xmath29 .
the low level of the force anisotropy noise is shown by the bracketing red lines representing the 1-@xmath31 deviation .
the solid curve is the fitting formula of eq .
[ fitform].,width=321 ] the hacc long - range force solver uses only cic mass deposition and force interpolation with a gauss - sinc spectral filter to mimic tsc - like behavior .
in addition , a fourth - order super - lanczos spectral differentiator ( @xcite ) is used , along with a sixth - order influence function .
this approach allows the data motion to be simplified as no complicated spatial differentiation is needed .
also , by moving more of the poisson - solve to the spectral domain , the inherent flexibility of fourier space can be exploited .
for example , the filtering is flexible and tunable , allowing careful force matching at only three grid cells .
figure [ ppforce ] shows the force - matching with the spectral techniques using a pair of test particles . in this test
, multiple realizations of particle pairs were taken at fixed distances , but with random orientations of the separation vector in order to sample the anisotropy imposed by the grid - based calculation of the force .
the solution of the poisson equation in hacc s long range force - solver is carried out using large ffts ; the corresponding spectral representation of the inverse ( discrete ) laplacian is the influence function .
hacc employs the following three - dimensional , sixth - order , periodic , influence function : @xmath32^{-1 } , \label{greenf}\end{aligned}\ ] ] where the sum is over the three spatial dimensions , @xmath33 is the grid spacing , and @xmath34 the physical box size .
as mentioned earlier , the cic - deposited density field is spectrally filtered using a sinc - gaussian filter : @xmath35^{n_s}. \label{filter}\ ] ] the aim of the filter is to reduce the anisotropy noise as well as control the matching scale where the short - range and long - range forces are matched ; the nominal choices in hacc are @xmath36 and @xmath37 .
this filtering reduces the anisotropy `` noise '' of the basic cic scheme by better than an order of magnitude , and allows for the use of the higher - order spectral differencing scheme . instead of solving for the scalar potential , and then using a spatial stencil - based differentiation
, hacc uses spectral differentiation within the poisson - solve itself , using fourth - order spectral lanczos derivatives , as previously mentioned .
the ( one - dimensional ) fourth - order super - lanczos spectral differentiation for a function , @xmath38 , given at a discrete set of points is @xmath39 where @xmath40 are coefficients in the fourier expansion of @xmath38 .
the `` poisson - solve '' in the hacc code is the composition of all the kernels above in one single fourier transform .
note that each component of the field gradient requires an independent fft .
this entails some extra work , but is a very small fraction of the total force computation , the bulk of which is dominated by the short - range solver .
an efficient and scalable parallel fast fourier transform ( fft ) is an essential component of hacc s design , and determines its weak scaling properties .
although parallel fft libraries are available , hacc uses its own portable parallel fft implementation optimized for low memory overhead and high performance .
since slab - decomposed parallel ffts are not scalable beyond a certain point ( restricted to @xmath41 ) , the hacc fft implementation uses data partitioning across a two - dimensional subgrid , allowing @xmath42 , where @xmath43 is the number of mpi ranks and @xmath44 is the linear size of the 3-d array .
the resulting scalable performance is sufficient for use in any supercomputer in the foreseeable future ( @xcite ) .
the implementation consists of a data partitioning algorithm which allows an fft to be taken in each dimension separately .
the data structure of the computing nodes prior to the fft is such as to divide the total space into regular three - dimensional domains .
therefore , to employ a two - dimensionally decomposed fft , the distribution code reallocates the data from small ` cubes ' , where each cube represents the data of one mpi rank , to thin two - dimensional ` pencil ' shapes , as depicted schematically in figure [ pencil ] .
once the distribution code has formed the pencil data decomposition , a one - dimensional fft can be taken along the long dimension of the pencil .
moreover , the same distribution algorithm is employed to carry out the remaining two transforms by redistributing the domain into pencils along those respective dimensions .
the transposition and fft steps are overlapped and pipelined , with a reduction in communication hotspots in the interconnect .
lastly , the dataset is returned to the three - dimensional decomposition , but now in the spectral domain .
pairwise communication is employed to redistribute the data , and has proven to scale well in our larger simulations .
a demonstration of this is provided by the blue gene / q sytems , where we have run on up to @xmath45 million mpi ranks ( @xcite ) . as the grid size
is increased on a given number of processors , the communication efficiency ( i.e. , the fraction of time spent communicating data between processors ) , remains unchanged .
this is an important validation of our implementation design , as the communication cost of the algorithm must not outpace the increase in local computation performance when scaling up in size .
further details of the parallel fft implementation will be presented elsewhere .
the total force on a particle is given by the vector sum of two components : the long - range force and the short - range force .
at distances greater than the force - matching scale , only the long - range force is needed ( at these scales , the ( filtered ) pm calculation is an excellent approximation to the desired newtonian limit , see figure [ ppforce ] ) . at distances less than the force - matching scale , @xmath29 ,
the short - range force is given by subtracting the residual filtered grid force from the exact newtonian force . to find the residual filtered pm force , we compute it numerically using a pair of test particles ( since in our case no analytic expression is available ) , evaluating the force at many different distances at a large number of random orientations .
the results are fit to an expression that has the correct asymptotic behaviors at small and large separation distances ( cf .
we used the particular form : @xmath46 which , with @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , and @xmath54 , provides an excellent match to the data from the test code ( figure [ ppforce ] ) , with errors much below @xmath27 . at very small distance scales ,
the gravitational force must be softened , and this can be implemented using plummer or spline kernels ( see , e.g. , @xcite ) . the force expression , eq .
( [ fitform ] ) , is complex and to implement faster force evaluations one can either employ look - ups based on interpolation or a simpler polynomial expression
. the communication penalty of look - ups can be quite high , whereas an extended dynamic range is difficult to fit with sufficiently low - order polynomials . in our case ,
the choice of a short matching scale , @xmath29 , enables the use of a fifth - order polynomial approximation , which can be vectorized for high performance . depending on the target architecture , hacc uses two different short - range solvers , one based on a tree algorithm , the other based on a direct particle - particle interaction ( p@xmath25 m ) .
tree methods are employed on non - accelerated systems , while both p@xmath25 m and tree methods can be used on accelerated systems .
hacc uses an rcb tree in conjunction with a highly - tuned short - range polynomial force kernel .
( an oct - tree implementation also exists , but is not the current production version . ) the implementation of the rcb tree , although not the force evaluation scheme , generally follows the discussion in @xcite .
( multiple rcb trees are used per rank to enhance parallelism , as described in section [ rcb ] . )
two core principles underlie the high performance of the rcb tree s design ( @xcite ) . _ spatial locality . _
the rcb tree is built by recursively dividing particles into two groups .
the dividing line is placed at the center of mass coordinate perpendicular to the longest side of the box ( figure [ rcb ] ) .
once the line of division is chosen , the particles are partitioned such that particles in each group occupy disjoint memory buffers .
local forces are then computed one leaf node at a time .
the net result is that the particle data exhibits a high degree of spatial locality after the tree build ; because the computation of the short - range force on the particles in any given leaf node , by construction , deals with particles only in nearby leaf nodes , the cache miss rate during the force computation is very low . _
walk minimization .
_ in a traditional tree code , an interaction list is built and evaluated for each particle . while the interaction list size scales only logarithmically with the total number of particles ( hence the overall @xmath55 complexity ) ,
the tree walk necessary to build the interaction list is a relatively slow operation .
this is because it involves the evaluation of complex conditional statements and requires `` pointer chasing '' operations .
a direct @xmath26 force calculation scales poorly as @xmath56 grows , but for a small number of particles , a thoughtfully - constructed kernel can still finish the computation in a small number of cycles .
the rcb tree exploits our highly - tuned short - range force kernels to decrease the overall force evaluation time by shifting workload away from the slow tree - walking and into the force kernel .
up to a point , doing this actually speeds up the overall calculation : the time spent in the force kernel goes up but the walk time decreases faster .
obviously , at some point this breaks down , but on many systems , tens or hundreds of particles can be in each leaf node before the crossover is reached .
we point out that the force kernel is generally more efficient as the size of the interaction list grows : the relative loop overhead is smaller , and more of the computation can be done using unrolled vectorized code .
in addition to the performance benefits of grouping multiple particles in each leaf node , doing so also increases the accuracy of the resulting force calculation : the local force is dominated by nearby particles , and as more particles are retained in each leaf node , more of the force from those nearby particles is calculated exactly . in highly - clustered regions ( with very many nearby particles ) , the accuracy can increase by several orders of magnitude when keeping over 100 particles per leaf node .
the p@xmath25 m implementation within hacc follows the standard method of building a chaining mesh to control the number of particle - particle interactions ( @xcite ) .
this algorithm is used within hacc when working with accelerated systems .
we defer further details regarding the architecture - specific implementation of the short - range force to section [ sec : specs ] , where the different alternatives are covered separately .
the time - stepping in hacc is based on the widely employed symplectic scheme , as used , e.g. , in the impact code ( @xcite ) , the forerunner of mc@xmath57 ( mesh - based cosmology code , see , e.g. , @xcite ) , in turn the pm precursor of hacc .
the basic idea here is not to finite - difference the equations of motion , but to view evolution as a symplectic map on phase space .
symplectic integration in hacc approximates the full evolution to second order in the time - step by composing elementary maps using the campbell - baker - hausdorff series expansion ( @xcite ) . in pm mode ,
the elementary maps are the ` stream ' and ` kick ' maps @xmath58 and @xmath59 corresponding to the free particle ( kinetic ) piece and the one - particle effective potential in the hamiltonian , respectively . in the stream map ,
the particle position is drifted using its known velocity , which remains unchanged ; in the kick map , the velocity is updated using the force evaluation , while the position remains unchanged .
a symmetric ` split - operator ' symplectic step @xmath60 is termed sks ( stream - kick - stream ) ; a ksk step is another way to implement a second - order symplectic integrator .
( in the presence of explicitly time - dependent hamiltonan pieces , the map evaluations have to be implemented at the correct times to maintain second - order accuracy . ) in the presence of both short and long - range forces , we split the hamiltonian into two parts , @xmath61 where @xmath62 contains the kinetic and particle - particle force interaction ( with an associated map @xmath63 ) , whereas , @xmath64 is just the long range force , corresponding to the map @xmath65 . since the long range force varies relatively slowly , we construct a single time - step map by subcycling @xmath63 : @xmath66 the total map @xmath63 being a usual second - order symplectic integrator .
this corresponds to a ksk step , where the s is not an exact stream step as in the pm case , but has enough @xmath63 steps composed together to obtain the required accuracy . for typical problems
the number , @xmath67 , of short time steps for each long time step will range between 3 - 10 , depending on accuracy requirements .
because the late - time distribution of particles is highly clustered , there can be a substantial advantage in using different ( synchronized ) local time steps down to the single - particle level . although hacc is currently designed for a regime where extreme dynamic range
is not needed , as in treating the innermost part of galaxy halos or in tracking orbits around black holes where this advantage is most felt ( e.g. , @xcite ) the automatic density information available in the short - range force solvers is used to enable multi - level time - stepping , resulting in speed - ups by a factor of 2 - 3 , with only small effects on the accuracy .
more on this topic can be found in section [ sec : specs ] .
hacc uses comoving coordinates for positions and velocities .
the actual internal representation of all variables is in the dimensionless form : @xmath68 where the fundamental scaling length , @xmath69 , is the length of a single grid cell of the long - range force pm - solver , @xmath70 , where @xmath34 is the box - size and @xmath71 is the number of grid points in a single dimension , @xmath72 is the current value of the hubble parameter , and the background mass density , @xmath73 .
hacc uses powers of the scale factor , @xmath74 , as the actual evolution variable , with a nominal default value of @xmath75 ; time - stepping is performed using the variable @xmath76 , @xmath77 .
besides easy portability between different architectures , another very important feature of hacc is its highly optimized memory footprint .
pushing the simulation limits in large - scale structure formation problems means running simulations with as many particles as possible , and this often implies running as close as possible to the memory limit of the machine . as a result , memory fragmentation becomes a serious problem .
to make matters worse , hacc is required to allocate and free different data structures during different parts of each time step because there is not enough available memory to hold all such structures at the same time .
furthermore , many of these data structures , such as the rcb tree used for the short - range force calculation , have sizes that change dynamically with each new time step .
this , combined with other allocations from the mpi implementation , message printing , file i / o , etc . with lifetimes that might outlast a time - step phase ( e.g. long - range force computation , short - range force computation , _ in situ _
analysis ) , is a recipe for fatal memory fragmentation problems problems that we actually encountered on the blue gene / q systems . to mitigate this difficulty
we have implemented a specialized pool allocator called bigchunk .
this allocator grabs a large chunk of memory , and then distributes it to various other subsystems .
during the first time step , bigchunk acts only as a wrapper of the system s memory allocator , except that it keeps track of the total amount of memory used during each phase of the time step . before the second time step begins
, bigchunk allocates an amount of memory equal to the maximum used during any phase of the previous time step plus some safety factor .
subsequent allocations are satisfied using memory from the ` big chunk ' , and all such memory is internally marked as free after the completion of each time - step phase .
this en - masse deallocation , the only kind of deallocation supported by bigchunk , allows for an implementation that has minimal overhead , and the time it takes to allocate memory from bigchunk is very small compared to the speed of the system allocator . because the bigchunk memory is not released back to the system , memory fragmentation no longer fatally affects the ability of the time - step phases to allocate their necessarily - large data structures .
the persistent state information in the simulation is carried by the particle attributes .
while the number of particles in each mpi rank s ( overloaded ) spatial sub - volume is similar , structure formation implies some variance . once the overload cache is filled , the ( total ) number of particles on a rank is fixed until the cache is emptied and refreshed , at which point the number of particles on each rank can change . in order to avoid memory fragmentation with persistent information
we must anticipate the maximum number of particles any rank will need during the course of a simulation run and allocate that amount of memory for particle information at the start , before any other significant memory allocation occurs .
we estimate the maximum number of particles by choosing a maximum representative volume from which bulk motion could move all of the particles into a rank s ( overloaded ) sub - volume and multiply that volume by the average particle density .
the actual memory allocations are monitored during runtime , allowing for adjustments to be made while the code is running . in practice for the sub - volumes with side lengths that are at least several tens of mpc
this results in allocating memory for an additional skin of particles that is 6 - 10 mpc thick .
hacc prints a memory diagnostic at each time step to indicate the extrema of particle memory usage across all ranks , and the amount of extra memory can be adjusted when restarting the code if the initial estimate appears to be insufficient for later times .
in addition to the space required for the actual particle information , we also allocate an array of integers to represent a permuted ordering of the particles and a scratch array large enough to hold any single particle attribute .
these enable out - of - place re - ordering of the particle information one attribute at a time without additional memory allocation . a plurality of i / o strategies are needed for different use cases , machine architectures , and data sizes because no single i / o strategy works well under all of these conditions .
our three main approaches are described below .
_ one file per process .
_ using one output file per process ( i.e. mpi rank ) is the simplest i / o strategy , and continues to provide the best write bandwidth compared to any other strategy .
because every process writes into a separate file , after file creation , there is no locking or synchronization needed in between processes .
unfortunately , while simple and portable , one file per process only works for a modest number of processes ( typically less than 10,000 ) .
no file system can manage hundreds of thousands of files that would result from checkpointing a large - scale run with one file per process .
additionally , as a practical matter , managing hundreds of thousands of files is cumbersome and error - prone . finally , even when the number of files is reasonable , reading the stored data back into a different number of processes than used to write the data requires redistribution .
this happens when the output is used for analysis on a smaller cluster ( or machine partition ) , or for visualization . in such cases ,
the required reshuffling of all of the data in memory is equivalent to the aggregation done by more complex collective i / o strategies and cancels out the simplicity of the one file per process approach . for improved scalability and flexibility , hacc supports the following additional i / o strategies . _ many processes per file . _
the default i / o strategy used by hacc , called genericio , partitions the processes in a system - specific manner , and each partition writes data into a custom self - describing file format .
each process writes its data into a distinct region of the file in order to reduce contention for file - system - level page locks . within the region assigned to each process ,
each variable is written contiguously .
on modern supercomputers , such as ibm blue gene and cray x series machines , dedicated i / o nodes execute special i / o forwarding system software on behalf of a set of compute nodes .
for example , on ibm blue gene / q systems , one i / o node is assigned to 128 compute nodes . by writing one file per i / o node ,
the total number of files is reduced by at least the ratio of compute nodes to i / o nodes , and a high percentage of the peak available bandwidth can be captured by the reading and writing processes . by partitioning the processes by i / o - node assignment and providing each process with a disjoint data region
, we are taking advantage of the technique successfully used by the glean i / o library ( @xcite ) on the blue gene / p and blue gene / q systems .
the i / o implementation can use mpi i / o , in either collective or non - collective mode , or ( non - collective ) posix - level i / o routines .
importantly , 64-bit cyclic - redundancy - check ( crc ) codes are computed for each variable for each rank , and this provides a way to validate data integrity .
this detects corruption that occurs while the data is stored on disk , while files are being transferred in between systems , and while being transmitted within the storage subsystem(s ) . during a recent run which generated 100 tb of checkpoint files , a single error in a single variable from one writing process
was detected during one checkpoint - reading process , and we were able to roll - back to a previous checkpoint and continue the simulation with valid data .
furthermore , over the past two years , crc validation has detected corrupted data from malfunctioning memory hardware on storage servers , misconfigured raid controllers and bugs in ( or miscompiled ) compression libraries .
the probability of seeing corrupt data from any of these conditions is small ( even while they exist ) , and overall , modern storage subsystems are highly reliable , but when writing and reading many petabytes of data at many facilities the probability of experiencing faults is still significant .
the crc64 code is in the process of being transformed into an open - source project .
an example of genericio performance under production conditions on the blue gene / q system mira is given in table [ table : genio - perf ] . in tests , i / o performance very close to the peak achievable has been recorded . under production conditions ,
we still achieve about two - thirds of the peak performance . _ a single file using parallel netcdf .
_ when peak i / o performance is not required , and the system s mpi i / o implementation can deliver acceptable performance , we can make use of a parallel - netcdf - based i / o implementation ( @xcite ) . the netcdf format is used by simulation codes from many different science domains , and readers for netcdf have been integrated with many visualization and analysis tools .
because netcdf has an established user community , and will likely be supported into the foreseable future , writing data into a netcdf - based format should make distributing data generated by hacc to other outside groups easier than if only custom file formats were supported .
the file schema that we developed for the parallel netcdf format is shown in figure [ fig : pnetcdf ] .
as the figure shows , particles are organized and indexed according to the spatial subdomains ( blocks ) of the simulation , one block per process .
the spatial extents of each block are also indexed .
currently , three modes of reading the parallel netcdf files are supported .
first , given some number of processes that need not be the same as the number of blocks , particles can be redistributed uniformly among the new number of processes while being read collectively .
second , the particles in a single block can be retrieved given the block i d .
third , particles in a desired bounding box can be retrieved .
the queried bounding box need not match the extents of any one block and particles overlapping several blocks may be retrieved in this manner .
the performance for writing the parallel netcdf output is shown in table [ table : pnetcdf - perf ] .
these tests were run on hopper ( cray xe6 ) at the national energy research scientific computing center ( nersc ) with a lustre file system . because the file system is shared by all running jobs
, performance results will vary due to resource contention ; the values in table [ table : pnetcdf - perf ] are the means of four runs for each configuration . to put these results in perspective , consider the last row of table [ table : pnetcdf - perf ] .
the peak performance expected for the number of object storage targets ( osts , 128 in this case ) is approximately 26.7 gib / s ( 1 gib = @xmath78 = 1,073,741,824 bytes ) .
this value represents ideal conditions of writing large amounts of data to one file per ost .
in contrast , our strategy uses one shared file with collective i / o aggregation and a high - level format with index data in addition to raw particle arrays .
even so , we achieve 56% of the peak ideal bandwidth .
b0.5 in b0.5 in b0.5 in b0.5 in b0.3 in no .
particles & no .
processes & file size ( gib ) & write time ( s ) & write bandwidth ( gib / s ) + @xmath79 & 512 & 43.8 & 22.0 & 1.90 + @xmath80 & 16384 & 1332.4 & 99.0 & 12.88 + @xmath81 & 262144 & 43821.6 & 380.5 & 109.9 + [ table : genio - perf ] b0.5 in b0.5 in b0.5 in b0.5 in b0.3 in no .
particles & no .
processes & file size ( gib ) & write time ( s ) & write bandwidth ( gib / s ) + @xmath82 & 64 & 4.6 & 3.54 & 1.30 + @xmath79 & 512 & 36 & 14.34 & 2.51 + @xmath83 & 4096 & 288 & 19.10 & 15.1 + [ table : pnetcdf - perf ]
in this section we go over the choice of algorithms deployed as a function of nodal architecture , as well as the corresponding optimizations implemented so far performance optimization is a continuous process . in order to evaluate the short - range force on non - accelerated systems , such as the blue gene / q
, hacc uses an rcb tree in conjunction with a highly - tuned short - range polynomial force kernel , as has been discussed in section [ srf ] .
an important consideration in this implementation is the tree - node partitioning step , which is the most expensive part of the tree build .
the particle data is stored as a collection of arrays the so - called structure - of - arrays format .
there are three arrays for the three spatial coordinates , three arrays for the velocity components , in addition to arrays for mass , a particle identifier , etc .
our implementation in hacc divides the partitioning operation into three phases .
the first phase loops over the coordinate being used to divide the particles , recording which particles will need to be swapped .
next , these prerecorded swapping operations are performed on six of the arrays .
the remaining arrays are identically handled in the third phase .
dividing the work in this way allows the blue gene / q hardware prefetcher to effectively hide the memory transfer latency during the particle partitioning operation and reduces expensive read - after - write dependencies .
the premise underlying the multilevel timestepping scheme ( section [ time - stepper ] ) is that particles in higher density regions will require finer short - range time steps .
a local density estimate can be trivially extracted from the same rcb tree constructed for evaluating the short - range interparticle forces .
each `` leaf node '' in the rcb tree holds some number of particles , the bounding box for those particles has already been computed , and a constant density estimate is used for all particles within the leaf node s bounding box . because the bounding box , and thus the density estimate , changes as the particles are moved , the timestepping level assigned to each leaf node is fixed when the tree is constructed .
this implies that , after particles have been moved , the leaf - node bounding boxes may overlap .
the force - evaluation algorithm is insensitive to these small overlaps , and the effect on the efficiency of the force calculation is apparently negligible .
[ see eq .
( [ kicks ] ) ] versus 5 sub - cycle steps for each particle .
shown is the ratio of the final power spectra for a small test problem ( 256 @xmath84mpc , 256@xmath25 particles ) , going out to the particle nyquist wave number .
the multi - level time step approach leads to a speed - up of the full simulation by a factor of @xmath282 , with only a negligible change in the error.,width=302 ] in between consecutive long - range force calculations , each particle is operated on by the kick ( velocity update ) and stream ( position update ) operators of the short - range force .
if a particle , based on the density of its leaf node at the beginning of the subcycle , is evolved using @xmath85 kicks , then it needs to be acted on by @xmath86 stream operators each evolving the particle by @xmath87 . to ensure time synchronization , these stream operators are further split such that all particles are acted on by the same number of stream operators
the number of kicks used for particles in a leaf node is determined by : @xmath88 where @xmath89 is an adjustable linear scaling parameter . in addition , the maximum level is capped by an additional user - provided parameter .
we show examples of accuracy control in the multi - level time stepping scheme in figures [ time1 ] [ time3_dens ] . in figure
[ time1 ] , we compare the power spectrum obtained from a simulation with 5 sub - cycle steps for each particle with a result that was obtained in the following way : 5 sub - cycles per step per particle are used until @xmath90 ; since the clustering at that point is still modest , this point is reached relatively quickly .
after @xmath90 we evolve each particle with at least 2 sub - cycles and allow
depending on the density two more levels of sub - cycling . in this test , setting the scaling parameter @xmath89 to 20 or 10 leads to accurate results , better than 0.2% out to the particle nyquist wavenumber , @xmath91 , where @xmath33 is the initial inter - particle separation ( the test case used 256@xmath25 particles in a 256 @xmath84mpc volume with 500 pm steps . ) in precision cosmology applications , one desires better than @xmath92 accuracy , and this is attained at wavenumbers less than @xmath93 ( @xcite ) .
consequently , the current error limits are comfortably wihtin the required bounds , the error at @xmath93 being only @xmath94 .
using the multi - level stepping can speed up the simulation by a factor of two ( changing @xmath89 in the range shown leaves the performance unaffected ) .
we have carried out a suite of convergence tests , concluding that the setting with @xmath95 satisfies our accuracy requirements . at much smaller length scales ,
the power spectrum test above can be augmented by checking the stability of small scale structures in the halos as the adaptive time - stepping parameters are varied . as typical examples thereof ,
we show results for the largest and second - largest halos ( identified using a ` friends - of - friends ' or fof algorithm ) in the same simulation discussed above in figures [ time2 ] [ time3_dens ] .
the halo density field is computed via a tessellation - based method in three dimensions and then projected onto a two - dimensional grid ; details of this implementation will be presented elsewhere ( @xcite ) .
the angle - averaged ( spherical ) density profiles are also shown in figures [ time2_dens ] and [ time3_dens ] . as can be seen from these results , aside from trivial differences due to the fof linking noise , the halo substructure depends relatively mildly on the choice of the values of the scaling parameter , @xmath89 , over the chosen ranges used . )
halo at @xmath96 for the same run as in figure [ time1 ] , with different values for @xmath89 .
minor variations in the outskirts of the halos are due to fof linking ` noise'.,width=264 ] at corresponding values of the scaling parameter , @xmath89.,width=302 ] for the second largest halo.,width=283 ] for the second - largest halo.,width=302 ] aside from optimizing the number of force evaluations , one also has to minimize the time spent in evaluating the force kernel .
this is a function of the design of the compute nodes .
here we provide a description of the blue gene / q - specific short - range force kernel and how it is optimized .
( very similar implementations were carried out for cray xe6 and xc30 systems . ) as mentioned earlier , the compactness of the short - range interaction ( cf .
section [ sec : lrange ] ) , allows the kernel to be represented as @xmath97 where @xmath98 , @xmath99 is a 5-th order polynomial in @xmath100 , and @xmath101 is a short - distance cutoff ( plummer softening ) .
this computation must be vectorized to attain high performance ; we do this by computing the force for every neighbor of each particle at once .
the list of neighbors is generated such that each coordinate and the mass of each neighboring particle is pre - generated into a contiguous array .
this guarantees that 1 ) every particle has an independent list of particles and can be processed within a separate thread ; and 2 ) every neighbor list can be accessed with vector memory operations , because contiguity and alignment restrictions are taken care of in advance .
every particle on a leaf node shares the interaction list , therefore all particles have lists of the same size , and the computational threads are automatically balanced .
the filtering of @xmath100 , i.e. , checking the short - range condition , can be processed during the generation of the neighbor list or during the force evaluation itself ; since the condition is likely violated only in a number of `` corner '' cases , it is advantageous to include it into the force evaluation in a form where ternary operators can be combined to remove the need of storing a value during the force computation .
each ternary operator can be implemented with the help of the instruction , which also has a vector equivalent .
even though these alterations introduce an ( insignificant ) increase in instruction count , the entire force evaluation routine becomes fully vectorizable .
there is significant flexibility in choosing the number of mpi ranks versus the number of threads on an individual blue gene / q node . because of the excellent performance of the memory sub - system , a large number of openmp threads up to 32 per node
can be run to optimize performance .
concurrency in the short - range force evaluation is exposed by , first , building a work queue of leaf - node - vs - tree interactions , and second , executing those interactions in parallel using openmp s dynamic scheduling capability .
each work item incorporates both the interaction - list building and the force calculation itself for each leaf node s particles . to further increase the amount of parallel work
, hacc builds multiple rcb trees per rank .
first , the particles are sorted into fixed bins , where the linear size of each bin is roughly the scale of the short - range force .
an rcb tree is then constructed within each bin , and because this process in each bin is independent of all other bins , this is done in parallel .
this parallelization of the tree - building process provides a significant performance boost to the overall force computation . when the force on the particles in each leaf node is computed , not only must the parent tree be searched , but so must the other 26 neighboring trees . because of the limited range of the short - range force , only nearest neighbors need to be considered . while searching many neighboring trees
adds extra expense , the trees are individually not as deep , and so the resulting walks are less expensive . also , because we distribute ` ( leaf node , neighboring tree ) ' pairs among the threads , this scheme also increases the amount of available parallelism post - build ( which helps with thread - level load balancing ) .
all told , using multiple trees in this fashion provides a significant performance advantage over using one large tree for the entire domain .
the first version of hacc was originally written for the ibm powerxcell 8i - accelerated hardware of roadrunner , the first machine to break the petaflop barrier . this architecture posed three critical challenges , all of which continue to be faced in one way or the other on all accelerated systems .
a more detailed description of the cell implementation and the roadrunner architecture is given in @xcite .
( see also @xcite . )
the three challenges for a roadrunner - style architecture are as follows .
( i ) _ memory balance . _
the machine architecture ( figure [ rr_arch ] ) has a top layer of conventional multi - core processors ( in this case , two dual - core amd opterons ) to which are attached ibm powerxcell 8i cell broadband engines ( cell bes ) via an eight - lane pci - e bus .
the relative performance of the opterons is small compared to that of the cell bes , by roughly a factor of 1:20 , but they carry half the memory and possess access to a communication fabric that is balanced to their level of computational performance . for large - scale n - body codes
, memory is a key limiting factor , therefore the code design must make the best use possible of the cpu layer ( this situation continues in current and future accelerated systems , as discussed further below ) .
( ii ) _ communication balance .
_ the cell bes dominate the computational resource , but are starved for communication , due to the relatively slow pci - e link to the host cpu ( figure [ rr_arch ] ) . from the point of view of the compute / communication ratio , such a machine is 50 - 100 times out of balance .
we note that this situation also continues to hold in the current generation of accelerated systems such as cpu / gpu or cpu / xeon phi machines : the computational approach taken must therefore maximize computation for a given amount of data motion .
( iii ) _ multiple programming models . _ accelerated systems have a multi - layer programming model . on the cpu level ,
standard languages and interfaces can be used ( hacc uses c / c++/mpi / openmp ) but the accelerator units often have a hardware - specific programming paradigm .
( although attempts to overcome this gap now exist , the results in actual practice are not yet compelling . ) for this reason , it is desirable to keep the code on the accelerator ( the cell be in this case ) as simple as possible and avoid elaborate coding .
in addition , it also proved advantageous to keep the data structures and communication patterns on the cell be as straightforward as possible to optimize computational efficiency . with these challenges in mind ,
hacc is matched to the machine architecture as follows : at the first level of code organization , the medium / long range force is handled by the fft - based method that operates at the opteron layer , as for all other architectures . at this layer ,
only grid information is stored and manipulated ( except when carrying out analysis steps ) .
particles live only at the cell layer .
there is a rough memory balance between the grid and particle data , matching well to the memory organization on the machine , and combating the first challenge mentioned above .
the particle - grid deposition and grid - particle interpolation steps are performed at the cell layer with only grid information passing between the two layers .
this compensates for the limited bandwidth available between the cell bes and the opterons .
the local force calculations reside at the cell level .
this addresses the second challenge , as aided by the particle overloading discussed in section [ sec : overload ] . because implementing complicated data structures at the cell level is difficult , and conditionals are best avoided , our choice for the local force solve is a direct particle - particle interaction . to make this interaction more efficient , we use a chaining mesh to control the number of interactions ( see , e.g. , @xcite )
this leads to an efficient , hardware - accelerated p@xmath25 m algorithm , thereby overcoming the third challenge ( @xcite ) .
as already discussed above , some of the challenges for gpu - accelerated systems are very similar in spirit to those for cell - accelerated systems .
the low - level gpu programming model ( opencl or cuda ) adds another layer of complexity and the compute to communication balance is heavily skewed towards computing .
one major difference between the two architectures is the memory balance : while on the cell - accelerated systems the cell layer has the same amount of memory as the cpu layer , this is generally not the case for cpu / gpu systems .
for example , the cray xk7 system , titan , at oak ridge , has 32 gb of host - side memory on a single node , with only 6 gb of gpu memory .
this adds yet another challenge , that of memory imbalance .
we have overcome this by partitioning the local data into smaller overlapping blocks , which can fit on device memory . very similar in spirit to particle overloading ,
the boundaries of the partitions are duplicated , such that each block can be evolved independently .
we emphasize again that the long / medium range calculations on this architecture remain unchanged , and only the short - range force kernel needs to be optimized .
the data partitioning is illustrated in figure [ data_part ] .
we utilize a two - dimensional decomposition of data blocks , which are in turn composed of slabs that are spatially separated by the extent of the short - range force roughly 3 grid cells ( see figure [ ppforce ] ) . in complete analogy to particle overloading , the data blocks are composed of ` active ' particles ( green ) that are updated utilizing the ` passive ' particles ( yellow and red ) from the boundary .
the ( red ) edge slabs are solely streamed , as opposed to performing the full sks time stepping described in section [ time - stepper ] .
this mediates the error inflow on these passive particles , as they can only `` see '' the interior particles within the domain .
we note that since the data has been decomposed into smaller independent work items , these blocks can now be communicated to any nodes that have the available resources to handle the extra workload .
hence , this scheme provides for a straightforward load - balancing algorithm by construction .
details of the error analysis and load balancing schemes will be described in an upcoming paper devoted to the gpu implementation of hacc @xcite .
m version ( blue short - dashed ) .
we also show the comparison with a gadget-2 run .
the agreement is very good the treepm runs agree with the gpu version of hacc to better than 0.1% up to the particle nyquist wavenumber .
the level of agreement with gadget-2 is noteworthy because it is a completely independent code .
, width=307 ]
[ cols="^,^,^,^,^",options="header " , ] . the solid curve is the prediction from the extended coyote emulator of @xcite .
the agreement across the two runs is at the fraction of a percent level , while the agreement with the emulator is at the 2% level , which is the estimated emulator accuracy .
, width=302 ] hacc has been subjected to a large number of standard convergence tests ( second - order time - stepping , halo profiles , power spectrum measurements , etc . ) . in this section
we focus mostly on a subset of hacc test results using the setup of the code comparison project , originally carried out in @xcite . in that work , a set of initial conditions was created for different problems ( mainly different volumes ) and a number of cosmological codes were run on those , all at their nominal default settings .
the final outputs were compared by measuring a variety of statistics , including matter fluctuation power spectra , halo positions and profiles , and halo mass functions .
the initial conditions and final results from the tests are publicly available and have been used subsequently by other groups for code verification , e.g for gadget-2 @xcite , and most recently for nyx @xcite .
in addition , we will also show some results from recently carried out large - scale simulations .
we first restrict attention to the larger volume simulation ( 256 @xmath84mpc ) and compare hacc results with those found for gadget-2 , as published in @xcite . while the simulation is only modest in size ( 256@xmath25 particles ) it does present a relatively sensitive challenge and is capable of detecting subtle errors in the code under test .
not only are statistical measures such as the power spectrum robust indicators of code accuracy , but visual inspection of the particle data itself presents a quick qualitative check on code behavior and correctness ; it is particularly valuable in identifying problems at early stages of code development .
we use paraview for this purpose ( @xcite ) .
the code comparison test was run with a force resolution of @xmath102kpc , very similar to what was used in the gadget-2 simulation .
we compare results from different hacc versions ( pptreepm on the blue gene / q and hopper , p@xmath25 m on a cell - accelerated and a gpu - accelerated system ) with those from gadget-2 .
the result for the matter power spectrum is shown in figure [ comp_pk ] , where , as in figure [ time1 ] , we show results up to @xmath103 .
all the code results are very close to each other ( note the scale on the y - axis ) .
the agreement over the full @xmath30-range is better than 0.5% , including gadget-2 . considering the different implementations of force solvers and time steppers ,
this closeness is very reassuring , particularly as no effort was made to fine - tune the level of agreement .
the treepm versions of hacc and the cpu / gpu version agree to better than a tenth of a percent .
next we present results from a more qualitative , but nevertheless , very detailed comparison , shown in figure [ comp_vis ] . in this test , we identify all particles that belong to halos with at least 100 particles per halo .
this is done within paraview , using an fof halo finder with linking length @xmath104 .
the particles are colored with respect to halo mass .
the most massive halo in the image ( colored in red ) has a mass of @xmath105m@xmath10 .
while there are differences in the images as is to be expected the overall agreement is striking .
almost all small halos exist in all images ( the ones that are missing are just below the cut of 100 particles , but they do actually exist ) and many of the fine details within the halo structures of the larger halos are well - matched .
the mass profiles of the three largest halos in the simulation are shown in figure [ comp_prof ] .
the binning shot noise dominates the comparison at small radii , but beyond that the agreement is very good .
other quantitative halo comparison statistics are given in table [ tab1 ] , to further illustrate the close match of the results from all of the different algorithms .
we illustrate the dynamic range and accuracy of the hacc approach by comparing results for the matter power spectrum from the ` outer rim ' and ` q continuum ' runs in figure [ big_pk ] .
the q continuum run on titan had @xmath106 billion particles , a 1.3 gpc box , with a mass resolution , @xmath107 m@xmath10 , and the outer rim run on mira had @xmath108 trillion particles , a 4.225 gpc box , with a mass resolution , @xmath109 m@xmath10 .
the numerical results ( run with different short - range force algorithms ) agree to fractions of a percent , and agree at the 2% level with the ( extended ) coyote emulator predictions of @xcite , which is at the level of accuracy expected from the emulator .
finally , we show results for the halo mass function at very large simulation scales illustrating excellent agreement across different sized simulations carried out using different short - range force implementations .
figure [ mf_big ] shows the ( fof , @xmath104 ) halo mass function at @xmath90 resulting from three different simulations with the same cosmology , ( i ) a run from the mira universe suite ( @xmath110 billion particles , 2.1 gpc box , mass resolution , @xmath111 m@xmath10 ) , ( ii ) the q continuum run on titan , and ( iii ) the outer rim run on mira . )
halo mass function for three large simulations run with hacc , measured at @xmath90 .
the box sizes range from @xmath112 gpc and the particle number in each simulation ranges from @xmath110 billion to over one trillion ( mass resolutions range from @xmath113 m@xmath10 to @xmath114 m@xmath10 ; for details , see text).,width=321 ]
an entire suite of _ in situ _ analysis tools ( cosmotools ) for hacc is under continuous development , driven by evolving science goals ; cosmotools exists in both _ in situ _ and stand - alone versions ( to be used for off - line processing ) . the overall structure of the _ in situ _
framework is shown in figure [ fig : framework ] .
output from the analysis is available either at run time or for post - processing . in the former case ,
a paraview server can be launched and connected to the simulation through a tool called catalyst . in the latter case ,
results of the _ in situ _ analysis are written to a parallel file system and later retrieved .
several standard tools for the analysis of cosmological simulations are part of the _ in situ _
framework , such as halo finders ( @xcite ) and merger tree constructors .
the first tool to be part of this framework that works on the full particle data to produce field information is a parallel voronoi tessellation that computes a polyhedral mesh whose cell volume is inversely proportional to the distance between particles .
such a mesh representation acts as a continuous density field that affords accurate sampling of both high- and low - density regions .
connected components of cells above or below a certain density can also approximate large - scale structures .
two important criteria for _ in situ _ analysis filters are that they should scale similarly as the simulation and have minimal memory overhead .
the parallel tessellation approach meets these criteria ; full details are given in @xcite .
the various tools can be turned on through the configuration file for hacc , and the frequency of their execution is also adjustable .
hacc runs on a variety of supercomputing platforms and has scaled to the maximum size of some of the fastest machines in the world , including roadrunner at los alamos , sequoia at livermore , titan at oak ridge , mira at argonne , and hopper at nersc .
we have carried out detailed scaling and performance studies on the blue gene / q systems ( @xcite ) and on titan ; a sample of our results is presented below . for both systems
, we carried out weak and strong scaling tests .
for the weak scaling tests we fix a physical volume and number of particles per node . when scaling up to more nodes , the volume and particle loading therefore increases , while the mass resolution stays constant .
the wall - clock time for a run should hence stay constant if the code scales or , equivalently , the time to solution per particle per step should decrease . the absolute performance measured in tflops per seconds will rise while the percentage per peak will stay constant . for our weak scaling tests ,
the particle mass is @xmath115m@xmath10 and the force resolution , 6 kpc .
all simulations are for a @xmath116cdm model .
simulations of cosmological surveys focus on large problem sizes , therefore the weak scaling properties are of primary interest . for the weak scaling test on the blue gene / q systems , we ran with 2 million particles in a @xmath28(100 @xmath84mpc)@xmath25 volume per core , using a typical particle loading in actual large - scale simulations on these systems . tests with 4 million particles per core produce very similar results . as demonstrated in figure [ comp ] ( right panel ) , weak scaling is ideal up to 1,572,864 cores ( 96 racks , all of sequioa ) , where hacc attains a peak performance of 13.94 pflops and parallel efficiency of @xmath117 . the largest test simulation on sequoia evolved @xmath283.6 trillion particles and a ( very high accuracy ) particle substep took @xmath118 ns for the full high - resolution code .
the scaling results were obtained by averaging over 50 sub - cycles . on titan we ran with 32 million particles per node in a fixed ( nodal ) physical volume of @xmath119 , representative of the particle loading in actual large - scale runs ( the gpu version was run with one mpi rank per node ) .
the results are shown in the left panel of figure [ comp ] .
in addition ( not shown ) we have timing results for a 1.1 trillion particle run , where we have kept the volume per node the same but increased the number of particles per node by a factor of two to 64.5 million . as for the blue gene / q systems , hacc weak - scales essentially perfectly up to the full machine .
for the strong scaling tests , we chose the same problem size on both systems , a ( 1000 @xmath84mpc)@xmath25 volume with 1024@xmath25 particles .
this is a rather small problem and strong scaling is expected to break down at some point .
the results for both systems are shown in figure [ comp ] the strong scaling regime is remarkably large .
on the blue gene / q system , we demonstrate strong scaling between 512 and 16384 cores ( with somewhat degraded performance at the largest scale ) . for titan
, we increase the number of nodes for this problem from 32 to 8192 , almost half of the machine .
as can be seen in the left panel in figure [ comp ] , strong scaling only starts degrading after 2048 nodes .
( the results for titan have been improved since these earlier tests , they will be reported in @xcite ) .
the significance of the strong scaling tests is in showing how well a code can perform as the effective memory per core reduces hacc can run very effectively at values as low as 100mb / core , which is roughly equivalent to the memory per core available in next - generation many - core systems .
the impressive scale and quality of data from sky survey observations requires a correspondingly strong response in theory and modeling .
it is widely recognized that large - scale computing must play an essential role in shaping this response , not only in interpreting the data , but also in optimizing survey strategies and validating data and analysis pipelines .
the hacc simulation framework is designed to produce synthetic catalogs and to run large campaigns for precision predictions of cosmological observables .
the evolution of hacc continues to proceed in two broad directions : ( i ) the further development of algorithms ( and their optimization ) for future - looking supercomputing architectures , including areas such as power management , fault - tolerance , exploitation of local non - volatile random - access memory ( nvram ) and investigation of alternative programming models , and ( ii ) addition of new physics capabilities in both modeling and simulation ( e.g. , gas physics , feedback processes , etc . ) and in the analysis part of the framework ( e.g. , increased sophistication in semi - analytic galaxy modeling , and an associated validation program ) .
the use of hacc for large - scale simulation campaigns covers applications such as those required to construct cosmological emulators ( @xcite ) , to determine covariance matrices ( @xcite ) , to help optimize survey design and test associated pipelines with synthetic catalogs ( @xcite ) , and , finally , to carry out mcmc - based parameter estimation across multiple cosmological probes ( @xcite ) .
a major component of the future use of hacc is therefore related to the production and exploitation of large simulation databases , including public access and analysis ; work in this area is in progress with a number of collaborators .
for useful interactions and discussions over time , the authors thank jim ahrens , viktor decyk , nehal desai , wu feng , chung - hsing hsu , rob latham , pat mccormick , rob ross , robert ryne , paul sathre , sergei shandarin , volker springel , joachim stadel , and martin white .
we acknowledge early contributions to hacc by nehal desai and paul sathre . running on a number of supercomputers ,
a large fraction of which were in their acceptance phase , required the generous assistance and support of many people .
we record our indebtedness to susan coghlan , kalyan kumaran , joe insley , ray loy , paul messina , mike papka , paul rich , adam scovel , tisha stacey , william scullin , rick stevens , and tim williams ( argonne national laboratory ) , brian carnes , kim cupps , david fox , and michel mccoy ( lawrence livermore national laboratory ) , lee ankeny , sam gutierrez , scott pakin , sriram swaminarayan , andy white , and cornell wright ( los alamos national laboratory ) , and bronson messer and jack wells ( oak ridge national laboratory ) .
argonne national laboratory s work was supported under u.s .
department of energy contract de - ac02 - 06ch11357 .
part of this research was supported by the doe under contract w-7405-eng-36 .
partial support for this work was provided by the scientific discovery through advanced computing ( scidac ) program funded by the u.s .
department of energy , office of science , jointly by advanced scientific computing research and high energy physics .
this research used resources of the alcf , which is supported by doe / sc under contract de - ac02 - 06ch11357 and resources of the olcf , which is supported by doe / sc under contract de - ac05 - 00or22725 . some of the results presented here result from awards of computer time provided by the innovative and novel computational impact on theory and experiment ( incite ) and ascr leadership computing challenge ( alcc ) programs at alcf and olcf . | current and future surveys of large - scale cosmic structure are associated with a massive and complex datastream to study , characterize , and ultimately understand the physics behind the two major components of the ` dark universe ' , dark energy and dark matter .
in addition , the surveys also probe primordial perturbations and carry out fundamental measurements , such as determining the sum of neutrino masses .
large - scale simulations of structure formation in the universe play a critical role in the interpretation of the data and extraction of the physics of interest . just as survey instruments continue to grow in size and complexity , so do the supercomputers that enable these simulations .
here we report on hacc ( hardware / hybrid accelerated cosmology code ) , a recently developed and evolving cosmology n - body code framework , designed to run efficiently on diverse computing architectures and to scale to millions of cores and beyond .
hacc can run on all current supercomputer architectures and supports a variety of programming models and algorithms .
it has been demonstrated at scale on cell- and gpu - accelerated systems , standard multi - core node clusters , and blue gene systems .
hacc s design allows for ease of portability , and at the same time , high levels of sustained performance on the fastest supercomputers available .
we present a description of the design philosophy of hacc , the underlying algorithms and code structure , and outline implementation details for several specific architectures .
we show selected accuracy and performance results from some of the largest high resolution cosmological simulations so far performed , including benchmarks evolving more than 3.6 trillion particles .
cosmology large scale structure of the universe ; n - body simulations |
fatty acids and other lipids are essential to every living organism . because of their amphiphilic nature
, they spontaneously self - assemble into bilayer membranes that define the limits of cells and serve as permeability barrier to prevent proteins , ions and metabolites from leaking out of the cell and unwanted toxins leaking in @xcite .
in euokaryotic cells , membranes also surround the organelles allowing for organization of biological processes through compartmentalization . in addition
, biological membranes host numerous proteins that are crucial for the mechanical stability of the cell , and which carry out a variety of functions such as energy and signal transduction , communication , and cellular homeostasis @xcite .
an important aspect of biological membranes is that they are typically not free but rather confined by other surrounding membranes , adhere to other membranes , and attach to elastic networks like the cytoskeleton and the extracellular matrix .
several model systems with reduced compositional complexity have been designed to mimic biological membranes .
these biomimetic systems include phospholipid bilayers deposited onto solid substrates ( solid - supported membranes ) @xcite , or on ultra - thin polymer supports ( polymer - supported membranes ) @xcite .
placing a membrane on a flat substrate allows for the application of several different surface sensitive techniques , including atomic force microscopy , x - ray and neutron diffraction , ellipsometry , nuclear magnetic resonance , and others @xcite . with the aid of biochemical tools and generic engineering
, supported membranes can be functionalized with various membrane - associated proteins @xcite .
synthetic supported membranes with reconstituted proteins are increasingly used as controlled idealized models for studying key properties of cellular membranes @xcite .
they provide a natural environment for the immobilization of proteins under nondenaturating conditions and in well - defined orientations @xcite .
another attractive application of supported membranes is the design of phantom cells exhibiting well defined adhesive properties and receptor densities @xcite . using advanced imaging techniques ,
detailed information can be obtained about the structure of the adhesion zone between the receptor - functionalized supported membrane and ligand - containing vesicles that can bind to the supported membrane @xcite .
these studies provide insight into the dynamics of adhesion processes and the molecular interactions involved in cell adhesion @xcite .
understanding these interactions is crucial for the development of drug delivery systems that depend on efficient adhesion between a liposome and the plasma membrane of the target cell .
adhesion between two membranes or between a membrane and another surface can , in principle , be facilitated by non - specific attractive interactions ( e.g. , coulomb and van der waals interactions ) @xcite .
cell adhesion , however , is usually caused by highly specific receptor molecules located at the outside of the plasma membrane of the cell , that can bind to specific ligands on the opposite surface @xcite .
typically , the area density of the receptor molecules located at the outside of the plasma membrane is rather low which does not lead to efficient adhesion .
however , when facing a surface with enough ligands , the receptors may cluster into highly concentrated adhesion domains to establish much stronger binding @xcite .
formation of adhesion clusters occurs in many biological processes @xcite , including the binding of white blood cells to pathogens @xcite , cadherin - mediated adhesion of neighboring cells @xcite , and focal adhesion of cells to the extracellular matrix @xcite .
many biophysical aspects of specific adhesion processes , ranging from the cooperativity in adhesion cluster formation to the influence of stochastic processes such as the ligand - receptor reaction kinetics , have been and continue to be studied theoretically using various models @xcite .
adhesion induced domain formation requires some attractive intermolecular interactions between the receptor - ligand pairs .
these interactions include both _
direct _ and _ membrane - mediated _ contributions .
the former are typically described by pairwise potentials which are infinitely repulsive at very small molecular separations and attractive at somewhat longer ( but still finite ) distances @xcite .
their effect can , therefore , be studied in the framework of the thoroughly researched lattice - gas model @xcite . in contrast , much less is known about the membrane - mediated mechanism , which has been proposed by braun _
_ to explain to formation of gap junctional plaque at cell - cell interfaces @xcite , and whose origin can be understood as follows : consider two adhesion bonds between two membranes or between a membrane and a surface ( fig . [ fig:1](a ) ) .
the adhesion points restrict the thermal height fluctuations of the membrane in their vicinity .
this entropy loss can be minimized if the two adhesion bonds are brought to the same place ( fig .
[ fig:1](b ) ) , in which case the membrane becomes pinned at only one place rather than two .
the membrane fluctuations , thus , induce an attractive potential of mean force between the adhesion bonds .
this effect is often named after casimir who predicted the existence of an attractive force between two conducting plates , due to quantum fluctuations of the electromagnetic field in the intervening space @xcite .
later , fisher and de gennes generalized this concept to classical interactions induced by thermal fluctuations in soft matter systems @xcite . for bilayer membranes
, there is a great body of theoretical work on the casimir effect between trans - membrane proteins ( see review in @xcite , and refs .
therein ) .
just like adhesion bonds , membrane inclusions represent a `` constraint '' on the shape of the membrane and , therefore , one should expects that they also interact with each other through casimir - like interactions .
in addition to the fluctuation - induced forces , the inclusions also experience other membrane - mediated interactions which arise from the membrane curvature elasticity and from the packing of the lipids near the inclusions surfaces ( see review in @xcite , and refs . therein ) .
these other types of membrane - mediated interactions are also expected to exist between membrane adhesion bonds .
the fundamental difficulty in attempting to provide a statistical - mechanical analysis of the aggregation behavior of the adhesion bonds is the need to integrate out the membrane degrees of freedom and write down the potential of mean force as a function of the coordinates of the adhesion sites @xmath1 .
this is a non - trivial problem since the membrane - mediated potential @xmath1 is a many - body potential which can not be expressed as the sum of two body terms .
the many - body nature of @xmath1 is best illustrated by the following example : consider the configuration shown in fig .
[ fig:2](a ) with two adhesion bonds at located at @xmath2 and @xmath3 and , in comparison , the one shown in fig .
[ fig:2](b ) with a single bond at @xmath2 and a cluster of three bonds around @xmath3 .
clearly , the spectrum of membrane thermal fluctuations in both cases is quite the same and , therefore , the adhesion bond located at @xmath2 is attracted to the three - point cluster in [ fig:2](b ) by the same force to which it is attracted to the single adhesion point in [ fig:2](a ) . if @xmath1 was the sum of pair interactions , the force in fig .
[ fig:2](b ) would be three times larger than the force in [ fig:2](a ) .
what is the difference between the aggregation of adhesion bonds in supported membranes and the traditional process of gas to liquid condensation ?
condensation phase transitions are usually associated with a competition between the mixing entropy @xmath4 which is higher in the dilute gas phase , and the interaction energy @xmath5 which is lower in the condensed liquid state .
the equilibrium phase corresponds to the minimum of the free energy @xmath6 , where @xmath7 is the temperature of the system . at high @xmath7 ,
the free energy @xmath8 is `` entropy - dominated '' and equilibrium is attained in the gas phase .
conversely , at low @xmath7 , the free energy is `` energy - dominated '' and , therefore , the condensed phase becomes thermodynamically more favorable .
the liquid - gas phase transition can be analyzed in the framework of an ising - like model of identical particles that populate a lattice .
excluded volume interactions between the particles are represented by the fact each lattice site can be occupied by no more than one particle .
when two particles occupy nearest - neighbor sites , they interact in a pairwise fashion with an attractive energy @xmath9 . denoting the occupancy of a lattice site by @xmath10 , with @xmath11 for an empty site and @xmath12 for an occupied site , the hamiltonian of the lattice - gas model is given by @xmath13 where the sum runs over all the pairs of lattice nearest neighbor sites .
the phase diagram of the lattice - gas model is well known .
there exists a critical value @xmath14 such that if the interaction energy @xmath15 , the particles will be distributed uniformly within the lattice . above this critical value , @xmath16 ,
a uniform distribution of the particles is observed only at low concentrations of particles ( `` gas phase '' ) , but upon increasing the concentration of particles , the system undergoes a first order phase transition and a second coexisting phase appears with a considerably larger concentration ( `` condensed phase '' ) . as discussed in the previous section ,
the aggregation process of adhesion domains involves an additional attractive potential of mean force resulting from the membrane thermal fluctuations . a lattice model where each lattice particle represents an adhesion bond in a supported membrane
can , therefore , be introduced by supplementing eq.([eq : latticegas ] ) with an energy term corresponding to the fluctuation - induced interactions .
since the functional form of this many - body potential of is yet unknown , we would , at this moment , introduce it via a general potential function @xmath17 that depends on the coordinates of the lattice particles : @xmath18 our first task must be to derive an expression for @xmath19 .
once this is accomplished , one can attempt to analyze the statistical mechanical properties of the model and address the question appearing at the beginning of section [ sec : lattice - gas ] .
one particular issue that we would like to address is whether the fluctuation induced attractive potential ( which is of entropic origin ) can win the competition against the repulsive force originating from the mixing entropy ?
in other words , can adhesion clusters form for purely entropic grounds , i.e. for @xmath20 in eq.([eq : latticegas2 ] ) ?
gas to liquid condensation transitions are generally believed to involve energy vs. entropy competition @xcite , but purely entropic phase transitions from a fluid ( disordered ) phase into a crystalline ( ordered ) phase are known to exist .
hard sphere systems , for instance , undergo a first order phase transition from a low density fluid phase into a high density solid phase @xcite .
this transition results from the competition between two entropies - the configurational mixing entropy which is higher in the disordered phase , and the entropy associated with the free volume available for each sphere , which is higher in the ordered crystal .
we start our analysis by considering the system shown schematically in fig .
[ fig:3 ] , consisting of a membrane with bending rigidity @xmath21 that fluctuates above a flat impenetrable surface @xcite .
let @xmath22 be the height of the membrane with respect to the surface , and assume that the membrane is pinned to the surface at one fixed point located at @xmath23 ( @xmath24 ) .
the elastic curvature energy of the membrane is given by the helfrich effective hamiltonian @xcite @xmath25\phi\left(h\right ) \delta\left[h\left(\vec{r_0}\right)\right]d^2\vec{r } , \label{eq : helfrich}\ ] ] where @xmath26 represents the hard wall constraint due to the surface ( @xmath27 for @xmath28 , and @xmath29 for @xmath30 ) , @xmath31 is the dirac delta - function , and the integration is taken over the cross sectional ( projected ) area of the membranes of size @xmath32 . one can calculate the partition function @xmath33 corresponding to hamiltonian ( [ eq : helfrich ] ) , by considering the helfrich effective hamiltonian of a _ freely fluctuating _
membrane @xmath34d^2\vec{r}. \label{eq : helfrich2}\ ] ] in this case , the associated partition function @xmath35 is readily calculated by introducing the fourier transformation of @xmath36 : @xmath37 , which decompose hamiltonian ( [ eq : helfrich2 ] ) into the sum of independent harmonic oscillators @xmath38 where @xmath39 is a microscopic length scale of the order of the bilayer thickness . hamiltonian ( [ eq : helfrich ] ) which also includes the functions @xmath26 and @xmath31 can not be diagonalized in the same manner .
however , one can relate the partition function @xmath33 of hamiltonian ( [ eq : helfrich ] ) with the partition @xmath35 of the free membrane hamiltonian ( [ eq : helfrich2 ] ) , by using the following simple argument .
the energy of a freely fluctuating membrane is invariant with respect to rigid - body transformations such as a vertical translation ( @xmath40 ) of the membrane s center of mass .
therefore , one can draw a flat surface and translate the free membrane such that the global minimum of its height function coincides with the surface ( see fig . [ fig:4](a ) ) .
the vertically translated free membrane looks very similar to the pinned membrane shown in fig .
[ fig:3](a ) .
the only difference between them is that the former can also glide over the surface ( see fig .
[ fig:4](b ) ) , while the latter is pinned at a fixed position on the surface .
this suggests that the pinning point effectively eliminates the membrane horizontal translational degree of freedom . in a statistical mechanical language
, the configurational phase space of the pinned membranes is smaller than , yet _
similar _ to , the phase space of a free membrane .
each subspace of identical free membrane configurations , like the ones shown in fig .
[ fig:4](b ) , includes one pinned membrane configuration - the configuration where the minimum of @xmath36 is at the pinning point @xmath23 ( or , more precisely , within a microscopic area of size @xmath41 around the pinning point , where @xmath39 is the spatial resolution of the continuum model ) .
this pinned membrane configuration occupies a fraction @xmath42 of the corresponding larger free membrane configurational subspace , which implies that the partition functions of the two systems are related by @xmath43 .
the free energy is obtained from @xmath44 the first term on the right hand side is the free energy of the free membrane whose elastic energy is given by helfrich hamiltonian ( [ eq : helfrich ] ) .
the second term , @xmath45 is the free energy cost of attaching the membrane to the surface at one point . following the above argument leads to a very interesting conclusion .
because of the similarity mapping that exists between the configurational phase spaces of the two problems , the statistical properties of the pinned and the free membrane must be identical to each other . this surprising result can be demonstrated by using an implicit - solvent coarse - grained ( iscg ) bilayer model which enables molecular simulations of mesoscopically large bilayer membranes over relatively large time - scales @xcite . towards this end
, we ran two independent monte carlo ( mc ) simulations - one of a free membrane ( without a surface ) and one of a membrane supported by a flat impenetrable surface . a snapshot from the supported membrane simulations is shown in fig . [ fig:5 ] .
each lipid molecule is represented in the model by a short string of three spherical beads , where one of the beads ( depicted as a dark gray sphere in fig .
[ fig:5 ] ) represents the hydrophilic head group and two beads ( light gray spheres in fig .
[ fig:5 ] ) represent the hydrophobic tail of the lipid . in the supported membrane simulations ,
the head bead of one of the lipids ( appearing in the corner at the front of the figure and indicated by an arrow ) was fixed to a flat surface which the lipids were not allowed to cross .
we measured the fourier spectrum of the membrane height function .
for the free membrane , the application of the equipartition theorem to the fourier - space representation of the helfrich hamiltonian ( [ eq : helfrich3 ] ) yields the following relationship between the mean squared amplitude of the fourier modes ( `` spectral intensity '' ) and the wave - vector @xmath46 : @xmath47 fig .
[ fig:6 ] depicts the results of our mc simulations for the spectral intensity vs. the wavenumber @xmath48 .
the figure show that , in agreement with our predictions : ( i ) the free ( open circles ) and pinned ( solid circles ) membranes exhibit the same statistics of thermal height fluctuations , and ( ii ) the spectral intensities of both membranes follow the @xmath49 power - law dependence anticipated by eq.([eq : eqpartition ] ) ( dashed line ) .
one can reverse the argument and derive eq.([eq : attachmentfenergy ] ) starting from the assumption that the spectral intensity of the supported membrane is identical to that of a free membrane and , therefore , can be described by eq.([eq : eqpartition ] ) .
the derivation proceeds as follows : first , from eq.([eq : eqpartition ] ) , one can straightforwardly show that the typical height at which the membrane undulates above the surface at a distance @xmath50 away from the pinning point scales linearly with @xmath50 @xcite ( see fig .
[ fig:7 ] ) : @xmath51 there is a repulsive force acting between the fluctuating membrane and the underlying surface , caused by their mutual steric hindrance .
helfrich @xcite showed that the associated repulsive interaction free energy density ( per unit area ) has the following scaling behavior @xmath52 which , together with eq.([eq : linear ] ) , yields @xmath53 by integrating this energy density over the projected area of the membrane , one derives eq.([eq : attachmentfenergy ] ) up to a numerical prefactor @xmath54 to set @xmath55 , as in eq.([eq : attachmentfenergy ] ) , one needs to replace the scaling relation eq.([eq : fenergydensity ] ) with the equality @xmath56
as noted by helfrich @xcite , the free energy density eq.([eq : fenergydensity2 ] ) due to the steric hindrance between the two surfaces ( i.e. , the fluctuating membrane and the underlying supporting surface ) is directly related to the rate of collisions between them . in other words ,
the probability density that the membrane hits the supporting surface at a distance @xmath50 from the pinning point exhibit the same dependence on @xmath50 as @xmath57 : @xmath58\sim \frac{1}{r^2}. \label{eq : prob1}\ ] ] this relationship provides the information needed for calculating the fluctuation induced attractive potential between two adhesion points .
this is done by regarding the point of collision between the membrane and the surface as a second pinning point which can diffuse across the surface . in this context , the probability density @xmath59 $ ] is identified with the pair correlation function between the adhesion points which , therefore , also follows the scaling form @xmath60 by definition , the pair potential of mean force is given by @xmath61 = 2k_bt\ln\left(r\right ) , \label{eq : pairpotential}\ ] ] which is an infinitely long range attractive potential that does not depend of the bending rigidity of the membrane , @xmath21 . the validity of eq.([eq : pair1 ] ) can be tested by using mc simulations of the iscg model shown in fig .
[ fig:5 ] with two lipid heads attached to surface - one fixed at the origin and the other allowed to diffuse on the flat surface . the pair correlation function is then directly computed by sampling the position of the mobile adhesion point .
our results @xcite , which are shown in fig .
[ fig:8 ] , agree very well with eq.([eq : pair1 ] ) .
the slope of the straight line on the log - log plot is equal to @xmath62 .
the deviations from the power law behavior @xmath63 at small values of @xmath50 ( @xmath64 ) are related to the breakdown of the continuum description of the helfrich hamiltonian at small spatial scales . at small separations , the molecular nature of
the lipids becomes important and the radial pair distribution function is dominated by the depletion shells around the lipids .
what if , in addition to the excluded volume repulsion , the membrane and the surface also interact via an attractive potential of somewhat longer range ? let us consider , for instance , the case when a harmonic confining potential is added to the helfrich hamiltonian : @xmath65d^2\vec{r}. \label{eq : helfrich4}\ ] ] for the harmonically confined membrane , one can define the length scale @xmath66 which marks the transition between two scaling regimes . for @xmath67 ,
the thermal fluctuations are governed by the bending rigidity term in hamiltonian ( [ eq : helfrich4 ] ) , while for @xmath68 the harmonic confinement term becomes dominant .
the latter term is a local one , which implies that the influence of the adhesion point becomes screened at large distances . in the case of a single adhesion point ,
the height of the fluctuations is now given by ( compare with eq.([eq : linear ] ) ) @xmath69 as illustrated schematically in fig .
[ fig:9](a ) .
the correlation function of a pair of adhesion points is given by ( compare with eq.([eq : pair1 ] ) ) @xmath70 the results of mc simulations of an iscg molecular model of a harmonically confined membrane verify this crossover between the two scaling regimes of @xmath71 ( see fig .
[ fig:9](b ) ) .
the energy of a membrane subjected to lateral surface tension @xmath72 is given by the following hamiltonian @xmath73d^2\vec{r}. \label{eq : helfrich5}\ ] ] scaling arguments @xcite show that , in this case , the pair correlation function exhibits behavior intermediate between eqs.([eq : pair1 ] ) and ( [ eq : pair2 ] ) : @xmath74 where the crossover length @xmath75 .
this scaling form is also confirmed by mc simulations ( see fig . [ fig:10 ] ) .
one of the main questions we aim to explore is whether the fluctuation - induced interactions are sufficiently strong to allow the formation of adhesion clusters . in the case of two adhesion
points the answer is no . despite of the attractive force between the adhesion points , they remains unbound and their mean pair separation grows linearly with the system size @xmath76 . more generally , if the pair correlation function decays algebraically at large distance , @xmath77 , the mean pair separation is given by @xmath78 the physically relevant cases in eq.([eq : pair4 ] ) are @xmath79 which , respectively , correspond to pinned , pinned - stressed , and pinned - confined membranes . in all of these cases
, @xmath80 grows with the size of the system .
another quantity of interest is the mean number @xmath81 of contacts between the surface and a membrane with one adhesion point . as discussed in section [ sec : twopoints ] , the probability density that membrane comes into contact with the surface at a distance @xmath50 from the pinning point has the same scaling form as the pair correlation function @xmath82 .
thus , @xmath83 we can use this last result to generalize and recalculate the attachment free energy of one adhesion point , eq.([eq : attachmentfenergy ] ) .
our original derivation of eq.([eq : attachmentfenergy ] ) was based on the argument that the configuration phase space of a pinned membrane comprises a small subspace within the configuration phase space of a free membrane .
more precisely , we argued that this subspace includes the free membrane configurations in which the global minimum of the height function occurs at the pinning point of the corresponding supported membrane .
we further argued that the relative size of the subspace is @xmath42 , which was based on the assumption that typically there is only one contact point with the surface and , therefore , this contact point must be the adhesion site .
however , as we see from eq.([eq : contacts ] ) , a typical configuration makes @xmath81 contacts with the surface .
therefore , the partition functions of the two problems ( free vs. pinned membranes ) are actually related by @xmath84z_{\rm free}$ ] . the attachment free energy is given by @xmath85 & { \rm for\ c=2 } \end{array } \right . .
\label{eq : attachmentfenergy3}\ ] ] notice that for sufficiently large @xmath76 , @xmath86 .
indeed , the attachment of a free membrane to a surface is likely to be more costly than the attachment of stressed and harmonically confined membranes that exhibit reduced fluctuations and , thus , remain close to the surface anyway .
let us look back at fig .
[ fig:8 ] which shows the pair correlation function between two adhesion points .
the figure demonstrates that the scaling form eq.([eq : pair1 ] ) holds over almost the entire range of pair separations considered ( @xmath87 ) .
the deviations from the power law at small pair distances arising from the short range depletion forces between lipids have already been discussed in section [ sec : twopoints ] .
what is quite surprising , though , is the pretty good agreement between the mc results and eq.([eq : pair1 ] ) at large pair distances .
eq.([eq : pair1 ] ) has been derived for two adhesion points in a very large membrane , neglecting boundary effects . in the simulations the conditions are different
- the membrane has a finite size and periodic boundary conditions are employed to reduce the finite size effects .
thus , each adhesion point interacts not only with the other adhesion point but also with its infinite array of periodic images . nevertheless , the existence of periodic images seems to have a very small impact on the results .
this observation is particularly unexpected for @xmath88 corresponding to situations where one of the adhesion points is equally close to two images of the other adhesion point . the only possible way to explain this surprising observation is to assume that the periodic images of the adhesion points are largely screened .
this assumption is consistent with the following physical picture : the membrane mediated interactions originate from the entropic cost due to the suppression of the membrane thermal undulations .
thus , the presence of each adhesion point is felt only in the region where it affects the fluctuations and cause their reduction , while outside of this region , the adhesion point is effectively screened . in this perspective , the idea that distant adhesion points are screened seems logical .
the fluctuations vanish at each adhesion point , irrespective of the distribution of the others .
moreover , in the immediate vicinity of each adhesion point , one expects the amplitude of the fluctuations to depend only on the distance from that adhesion point .
if the membrane is neither stressed nor experiencing a confining surface potential , the amplitude of the fluctuations in this region grows linearly with the distance @xmath50 from the adhesion point , as given by eq.([eq : linear ] ) .
we now wish to introduce a more general expression that holds over the entire area of the membrane and coincides with eq.([eq : linear ] ) close to every adhesion point .
our suggestion is as follows @xcite : in each unit area of the membrane , the mean height of the membrane above the surface is given by ( compare with eq.([eq : linear ] ) ) @xmath89 where @xmath90 is the distance of the unit area from the _ nearest _ adhesion point .
we also replace @xmath50 with @xmath90 in eq.([eq : fenergydensity2 ] ) for the attachment free energy density , which now reads @xmath91 the total attachment free energy of a given distribution of adhesion points is obtained by integrating the attachment free energy density eq.([eq : fenergydensity - many ] ) over the entire membrane area . this calculation
is done by constructing the voronoi diagram of the distribution of adhesion points , integrating the free energy density with each cell ( where in each cell the distance is measured from the adhesion point located in the cell , and a small region of microscopic size @xmath39 around the point is excluded from the integral ) , and summing the contributions of the different cells : @xmath92 in a lattice - gas model , the discrete analog of this expression applies @xmath93 where the sum run over all the empty lattice sites ( @xmath11 ) and @xmath41 is the area per lattice site . as discussed in section [ sec : lattice - gas ] ,
our main goal is to develop and use a lattice - gas model for the aggregation problem of adhesion points . in the model
, each lattice point represents an adhesion point between the membrane and surface .
the energy of a given configuration of lattice points is given by eq.([eq : latticegas2 ] ) , where the first term represents the short - range attraction between adhesion points and the second term is a many - body fluctuation - induced potential @xmath19 .
our journey to derive an expression for @xmath19 started in section [ sec : onepoint ] , and has finally reached the end .
@xmath19 is a potential of mean force which , for a given distribution of adhesion points , is determined by tracing over all the relevant membrane configurations and calculating the free energy penalty associated with the reduced thermal fluctuations .
eq.([eq : fenergy - lattice ] ) provides this expression by assigning a free energy cost with each empty lattice site that represents a fluctuating unit area of the supported membrane .
introducing eq.([eq : fenergy - lattice ] ) into eq.([eq : latticegas2 ] ) , yields the energy function of our lattice - model of adhesion points @xmath94 let us see how one can re - derive eq.([eq : pairpotential ] ) for the pair potential of mean force by calculating the attachment free energy eq.([eq : fenergy - voronoi ] ) . towards this end , consider the membrane shown schematically in fig .
[ fig:11 ] with two adhesion points , each of which located a distance @xmath95 from the center of the membrane .
the dashed line shows the boarder between the voronoi cells of the adhesion points , where each cell extends over half of the area of the membrane . for the configuration shown in fig .
[ fig:11 ] , the attachment free energy eq.([eq : fenergy - voronoi ] ) reads : @xmath96}\right.\nonumber \\ & + & \left.\int_{(r+l)/2}^{l/2}dx\,\frac{k_bt}{\pi\left[y^2 + \left(x - r/2\right)^2\right]}\right ] .
\label{eq : pairrevised1}\end{aligned}\ ] ] integrating over @xmath97 yields , @xmath98 .
\label{eq : pairrevised2}\end{aligned}\ ] ] assuming that @xmath99 , the inverse tangent function in eq.([eq : pairrevised2 ] ) can be approximated by the constant value of @xmath100 over most of the integration range . with this approximation
, one gets @xmath101 the first term in eq.([eq : pairrevised3 ] ) is the free energy cost of a single adhesion site [ eq.([eq : attachmentfenergy ] ) ] , which is the expected value when the two adhesion points coincide ( @xmath102 ) to form a single cluster .
the second term , which represents the additional free energy cost associated with the separation of the adhesion points , is identified as the fluctuation induced pair potential , in agreement with eq.([eq : pairpotential ] ) .
we now come back to the many - body problem and start with a mean field analysis of our lattice model hamiltonian ( [ eq : latticegasadhesion ] ) .
let us consider a lattice of @xmath103 sites of which @xmath104 sites are occupied by adhesion points .
let us further assume that the adhesion points form @xmath105 adhesion clusters .
the free energy of system includes three contributions : ( i ) the mixing entropy of the adhesion clusters , @xmath106 , ( ii ) the lattice - gas energy , @xmath107 , of the direct interactions between the adhesion points [ first term in eq.([eq : latticegasadhesion ] ) ] , and ( iii ) the attachment free energy , @xmath108 [ second term in eq.([eq : latticegasadhesion ] ) ] . the first free energy contribution is given by @xmath109 + \frac{1}{2}c\left(\frac{n_c^2}{n_s}\right),\ ] ] where @xmath110 is the second virial coefficient .
on average , each cluster consists of @xmath111 adhesion points ; and if we assume that it has a roughly circular shape than @xmath112 . denoting the number densities of the adhesion points by @xmath113 , and of the clusters by @xmath114 , the free energy of mixing per lattice site
is given by @xmath115 + 2\rho\rho^*. \label{eq : free1}\ ] ] the second contribution to the free energy is due to the direct interactions between the adhesion points .
the ground state of the interaction energy @xmath107 is achieved when a single circular adhesion domain with minimal surface is formed .
if we set the ground state as the reference energy , the energy of an ensemble of clusters can be estimated as being proportional to the total length of the domain boundaries .
for @xmath116 circular clusters of size @xmath111 we have @xmath117 where @xmath118 , the associated dimensionless line tension , is proportional to the interaction energy @xmath119 @xmath120 and @xmath121 is the mean number of nearest - neighbor vacant sites per occupied site on the boundary of a cluster ( @xmath122 for very large clusters ) .
the sum of free energy contributions ( [ eq : free1 ] ) and ( [ eq : free2 ] ) constitutes the total free energy density ( per lattice site ) of a 2d lattice gas of clusters : @xmath123 the third contribution to the attachment free energy can be estimated as follows .
the clusters form @xmath116 voronoi cells , each of which has on average an area of @xmath124 .
the attachment free energy of each voronoi cell is given by an equation similar to eq.([eq : attachmentfenergy ] ) for the attachment free energy of one adhesion point , but with @xmath125 instead of the total membrane area @xmath32 .
thus @xmath126,\ ] ] and the attachment free energy density is given by @xmath127 which eliminates the first term in the lattice - gas free energy density [ eq.([eq : freelg ] ) ] , yielding @xmath128 we consider a low density of adhesion sites @xmath129 , which also implies a low number density of adhesion clusters since @xmath130 . by minimizing the free energy density
we obtain the equilibrium value of the @xmath131 for the standard lattice - gas model [ eq.([eq : freelg ] ) ] and for the adhesion points of a fluctuating supported membrane [ eq.([eq : free ] ) ] . in both cases ,
the system undergoes a first order phase transition at @xmath132 from the gas phase ( @xmath133 ) to a condensed phase consisting of only a few clusters ( @xmath134 ) .
also , in both cases , the free energy reaches a maximum at intermediate densities ( @xmath135 ) .
this free energy barrier for condensation disappears at the spinodal point @xmath136 .
for the lattice - gas problem we find @xmath137 while for the adhesion points of fluctuating membranes we have @xmath138 the results of eqs.([eq : lambdalg ] ) and ( [ eq : lambda ] ) are summarized in fig .
[ fig:12](a ) and [ fig:12](b ) , respectively .
the important points in the results are that : ( i ) @xmath139 , which means that the fluctuation induced interactions alone are _ not _ sufficient to induce aggregation of adhesion domains , but ( ii ) they greatly reduce the strength of the direct interactions required to facilitate cluster formation since @xmath140 ( and also @xmath141 ) .
below , we support these conclusions with mc simulations and show that for adhesion points of fluctuating membranes , the site - site cohesive energy @xmath119 for the onset of aggregation falls below the thermal energy @xmath0 .
to further investigate the aggregation behavior in supported membranes , we performed mc simulations of both our lattice model of adhesion points and of the standard 2d lattice - gas model @xcite .
we simulated the system at two different densities @xmath142 and @xmath143 , and for various values of @xmath119 ranging from 0 to 3 @xmath0 .
snapshots taken from simulations for @xmath144 and @xmath143 are shown in fig .
[ fig:13 ] .
[ fig:13](a ) shows the initial configuration where the points are randomly distributed on the lattice .
[ fig:13](b ) and [ fig:13](c ) show , respectively , typical equilibrium configurations of the standard lattice - gas model and of our model of adhesion points .
one clearly sees that for the same strength of the interaction energy @xmath144 , the standard lattice gas model remains in the gas phase , while the adhesion points ( that , in addition to the direct interactions , also attract each other via the fluctuation - induced mechanism ) condense into a large cluster containing almost all the adhesion points . in order to determine the onset of the gas to liquid transition
, we measured the average number of clusters in the system ( where a cluster is defined as a set of neighboring occupied sites ) , and the mean value of the energy of direct interactions between sites , @xmath145 [ first term in eq.([eq : latticegasadhesion ] ) ] .
our results are summarized in fig .
[ fig:14](a ) ( for @xmath146 ) and [ fig:14](b ) ( for @xmath143 ) . for each @xmath147 , we measured these quantities both for the standard lattice - gas model ( open symbols and dash - dotted lines in fig .
[ fig:14 ] ) and for the adhesion points model ( solid symbols and solid lines in fig .
[ fig:14 ] ) . the number of clusters is denoted by squares ( values on the right @xmath97-axis of the figures ) , while @xmath148 is represented by circles ( values on the left @xmath97-axis ) .
the gas phase is characterized by a large number of small clusters , some of which may be of the size of a single site . furthermore , since each occupied site has a relatively small number of neighboring occupied sites , the mean configurational energy @xmath149 is relatively low .
conversely , when the sites form large clusters in the condensed phase , @xmath150 is high , and the total number of clusters decreases ( and in many cases , especially for large values of @xmath119 , we simply observe only a single cluster in our system ) .
[ fig:14 ] exhibits an abrupt , clearly first - order , transition from a gas phase with a large number of clusters and small @xmath149 to a condensed state with a small number of clusters and large @xmath149 .
the estimated values of @xmath119 at the transition are ( see vertical lines in fig .
[ fig:14 ] ) : @xmath151 ( @xmath146 ) and @xmath152 ( @xmath143 ) . in comparison
( see also fig .
[ fig:14 ] ) , for the standard lattice - gas model , the transition values are roughly twice larger than these values : @xmath153 ( @xmath146 ) and @xmath154 ( @xmath143 ) . our computational results which show that the fluctuation mediated interactions reduce the strength of @xmath155 , are in a qualitative agreement with the mean field theory prediction .
to make a quantitative comparison between the theory and the simulations , one needs to estimate the parameter @xmath121 appearing in eq.([eq : b_parameter ] ) .
several reasons make such an estimation difficult and inaccurate : first , our non - standard mean field theory is based on the assumption that the clusters are circular and roughly have the same size , which is quite a crude approximation .
second , tracing the precise location of @xmath155 in fig . [ fig:14 ] is largely inaccurate because of the finite size of the system that makes the transitions look like crossovers . to reduce the large uncertainties associated with the determination of @xmath155
, one can look at the difference between the value of this quantity in our model of adhesion points and for the standard lattice - gas model .
using @xmath156 for @xmath143 , we find @xmath157 , as indeed expected for large clusters .
in this review we presented a statistical thermodynamics analysis of the aggregation behavior of adhesion points between a fluctuating membrane and a supporting surface .
our analysis focused on the contribution of the membrane thermal fluctuations to this process , via the attractive interactions that they mediate between the adhesion points .
the origin of the fluctuation - induced ( casimir - like ) interactions are the restrictions imposed on the membrane thermal fluctuations by the adhesion points , and the associated free energy cost which is minimized when the adhesion points localize in a cluster .
we investigated both the two- and many - body fluctuation - induced interactions .
for the two - body problem , our analysis reveals that the fluctuation induced pair potential is infinitely long - range with a logarithmic dependence on the pair distance .
if , in addition to the excluded volume interactions , the membrane and the surface also interact via an attractive confining potential , the fluctuation - induced pair potential becomes screened at large distances .
the screening of the pair potential is due to the fact that far away from each adhesion point , the amplitude of the fluctuations is governed by the strength of the external potential rather than by the presence of the other adhesion point . in the many - body problem ,
the fluctuation - induced interactions are self - screened .
the amplitude of the thermal fluctuations at each unit area of the membrane is governed by the distance to the closest adhesion points , which implies that each point interacts only with a few nearby points .
this justifies our mapping of the problem into the 2d lattice - gas model with an effectively larger ( `` renormalized '' ) interaction energy . depending on the strength of the renormalized interactions , the system may be either in a `` gas '' ( uniform distribution ) or a `` condensed '' ( adhesion cluster ) phase .
the interesting question which arises is whether the fluctuation - induced contribution to the attraction is sufficiently strong to allow cluster formation .
our analysis finds that the answer to this question is no .
the fluctuation - induced interactions alone are too weak to induce the condensation transition .
they do , however , greatly reduce ( to below the thermal energy @xmath0 ) the strength of the direct interactions at which the transition takes place .
i wish to express my deep gratitude to prof .
phil pincus for inspiring me to work on this fascinating problem .
i have benefited enormously from many insightful discussions with him .
i also wish to thank my student , noam weil , who participated in the final part of the research with such enthusiasm and creativity .
h. a. rinia , g. w. h. wurpel , m. mller , visualization and characterization of domains in model membranes , in : a. ottova - tien ( ed . ) , advances in planar lipid bilayers and liposomes vol . 3 , elsevier , amsterdam , 2006 , pp.85 - 123 .
t. r. weikl , m. asfaw , h. krobath , b. rycki , r. lipowsky , adhesion of membranes via receptor - ligand complexes : domain formation , binding cooperativity , and active processes , soft matter 5 ( 2009 ) 3213 - 3224 .
h. krobath , g. j. schtz , r. lipowsky , t. r. weikl , lateral diffusion of receptor - ligand bonds in membrane adhesion zones : effect of thermal membrane roughness , europhys . lett . 78 ( 2007 ) 38003 - 1 - 38003 - 6 .
m. lamblet , b. delord , l. johannes , d. van effenterre , p. bassereau , key role of receptor density in colloid / cell specific interaction : a quantitative biomimetic study on giant vesicles , eur .
j. e 26 ( 2008 ) 205 - 216 .
e. reister - gottfried , k. sengupta , b. lorz , e. sackmann , u. seifert , a. suncna smith , dynamics of specific vesicle - substrate adhesion : from local events to global dynamics , phys .
( 2008 ) 208103 - 1 - 208103 - 4 . | supported lipid membranes are useful and important model systems for studying cell membrane properties and membrane mediated processes
. one attractive application of supported membranes is the design of phantom cells exhibiting well defined adhesive properties and receptor densities .
adhesion of membranes may be achieved by specific and non - specific interactions , and typically requires the clustering of many adhesion bonds into `` adhesion domains '' .
one potential mediator of the early stages of the aggregation process is the casimir - type forces between adhesion sites induced by the membrane thermal fluctuations . in this review
, i will present a theoretical analysis of fluctuation induced aggregation of adhesion sites in supported membranes .
i will first discuss the influence of a single attachment point on the spectrum of membrane thermal fluctuations , from which the free energy cost of the attachment point will be deduced .
i will then analyze the problem of a supported membrane with two adhesion points . using scaling arguments and monte carlo simulations
, i will demonstrate that two adhesion points attract each other via an infinitely long range effective potential that grows logarithmically with the pair distance .
finally , i will discuss the many - body nature of the fluctuation induced interactions .
i will show that while these interactions alone are not sufficient to allow the formation of aggregation clusters , they greatly reduce the strength of the residual interactions required to facilitate cluster formation . specifically , for adhesion molecules interacting via a short range attractive potential , the strength of the direct interactions required for aggregation is reduced by about a factor of two to below the thermal energy @xmath0 . |
we are grateful to v.a .
petrov for the interesting discussions on the impact parameter dependence of the mean multiplicity .
99 a. de rjula , nucl .
151 , 23 , 2006 , [ hep - ph/0412094 , astro - ph/0411763 ] .
r. engel , t.k .
gaisser , p. lipari , t. stanev , phys . rev .
d 58 , 014019 , 1998 ; + r. engel , astro - ph/0504358 .
t. stanev , astro - ph/0411113 ; + s.i .
nikolsky , v.g .
sinitsina , phys . atom .
67 , 1900 , 2004 , ; + a.a .
petrukhin , proc . of the 28th international cosmic ray conferences
icrc 2003 , tsukuba , japan , 31 jul - 7 aug 2003 , 275 .
hrandel , talk at 19th european cosmic ray symposium , florence , italy , 30 aug - 3 sep 2004 ; astro - ph/0501251 .
t. antoni et al . , phys . rev .
d 71 , 072002 , 2005 .
s. m. troshin , n. e. tyurin , phys . lett .
b 316 , 175 , 1993 . j.r .
hrandel , j. phys .
g 29 , 2439 , 2003 .
troshin , n.e .
tyurin , phys .
35 , 555 , 2004 .
troshin , n.e .
tyurin , eur .
j. c 21 , 679 , 2001 ; + v.a .
petrov , a.v .
prokudin , s.m .
troshin , n.e .
tyurin , j. phys .
g 27 , 2225 , 2001 .
t. arisawa , et al .
b 424 , 241 , 1994 . s.m .
troshin , phys .
b 597 , 391 , 2004 .
troshin , n.e .
tyurin , eur .
j. c 39 , 435 , 2005 .
shabelski , r.m .
weiner , g. wilk , z. wlodarczyk , j. phys .
g 18 , 1281 , 1992 .
j. dias de deus , phys . rev .
d 32 , 2334 , 1985 ; + s. barshay , y. ciba , phys . lett .
b 167 , 449 , 1985 . l.w .
jones , nucl .
b75a , 54 , 1999 .
gaisser et al . , phys . rev .
d 36 , 1350 , 1993 .
troshin , n.e .
tyurin , j. phys .
g 29 , 1061 , 2003 . | we note that the new unitarity regime when scattering amplitude goes beyond the black disc limit ( antishadowing ) could help in the explanation of the regularities such as knee in the energy spectrum , existence of penetrating and long - flying particles and other features observed in the measurements of the extensive air showers which originate from cosmic particles interactions with the atmosphere .
+ the experimental and theoretical studies of cosmic rays are the important source of astrophysical information ( cf . e.g. @xcite ) and they simultaneously provide a window to the future results of accelerator studies of hadron interaction mechanism at the lhc .
[ ts : fig1 ] it can happen that the investigations of cosmic rays will give us a clue that the hadron interaction and mechanism of particle generation is changing in the region of @xmath0 tev@xcite . indeed , the energy spectrum which follows simple power - like law @xmath1 changes its slope in this energy region and becomes steeper : index @xmath2 increases from @xmath3 to @xmath4 .
it is important that the knee in the energy spectrum appears in the same energy region where the penetrating and long flying particles also start to appear in the extensive air showers ( eas ) : the absorbtion length is also changing from @xmath5 @xmath6 to @xmath7 @xmath6 ( cf .
@xcite ) .
there is also specific feature of the events at the energies beyond knee such as alignment cf . @xcite and the references for the earlier papers therein .
the above phenomena were interpreted as a result of appearance of the new particles which have a small inelastic cross
section and/or small inelasticity .
these new particles can be associated with a manifestation of the supersymmetry , quark
gluon plasma formation and other new mechanisms .
however , there is another possibility to treat those cosmic rays phenomena observed in eas as the manifestations of the new unitarity regime ( antishadow scattering mode ) at such energies @xcite .
unitarity of the scattering matrix @xmath8 implies , in principle , an existence at high energies @xmath9 , where @xmath10 is a threshold tev @xcite , which corresponds to the energy in the laboratory system @xmath11 @xmath12 . ] of the new scattering mode antishadow one .
it has been revealed in @xcite and described in some detail ( cf . @xcite and references therein ) and
the most important feature of this mode is the self - damping of the inelastic channels contributions at small values of impact parameter antishadowing .
the antishadowing leads to @xmath13 at @xmath14 , where @xmath15 is a probability of the absence of the inelastic interactions , @xmath16 , where @xmath17 is the elastic scattering @xmath17matrix .
self - damping of the inelastic channels leads to asymptotically dominating role of elastic scattering .
the cross section of inelastic processes rises with energy as @xmath18 , while elastic and total cross sections behave asymptotically as @xmath19 .
the antishadow scattering mode could definitely be observed at the lhc energies and studies of the extensive air showers originated from the cosmic particles interactions with the atmosphere provide evidence for it as we will argue in what follows .
starting at some threshold energy @xmath10 ( where amplitude reaches the black disk limit at @xmath20 ) , antishadowing can occur at higher energies in the limited region of impact parameters @xmath21 ( while at large impact parameters only shadow scattering mode can be realized ) .
the inelastic overlap function @xmath22 becomes peripheral when energy goes beyond @xmath23 ( fig.2 ) . at such energies the inelastic overlap function reaches its maximum value at @xmath24 , where @xmath25 is the interaction radius , while the elastic scattering occurs at smaller values of impact parameter , i.e. @xmath26 .
note that @xmath27 where @xmath28 and @xmath29 and unitarity condition in the impact parameter space is the following @xmath30 where @xmath31 is the elastic scattering amplitude .
the quantity @xmath32 is a measure of the reaction peripherality . despite that the asymptotics for @xmath33 and @xmath34 are different , the quantities @xmath35 and @xmath36 have the same asymptotical energy dependence , proportional to @xmath19 .
so , beyond the transition energy range there are two regions in impact parameter space : the central region where self - damping of inelastic channels occurs ( antishadow scattering at @xmath37 ) and the peripheral region of shadow scattering at @xmath38 . and maximum at @xmath24 with the energy increase .
in the region of @xmath24 the complete absorbtion takes place , i.e. @xmath39.,width=340 ] at the energies @xmath40 small impact parameter scattering is almost elastic one . thus head on colliding particles will provide appearance of penetrating long - flying component in the eas and such particles will spend only small part of their energy for the production of secondaries .
the head - on collisions will lead to smaller number of secondary particle and it will provide faster decrease of the energy spectrum of cosmic rays , i.e. it will result in the appearance of the knee .
this qualitative picture will be explained in more detail in what follows .
it should be noted that this effect has a threshold in the energy dependence .
it is also important to note that due to small probability of the sequential head - on collisions the number of events with penetrating particles also should be small .
nontheless , such events have been observed in the experiments pamir @xcite .
antishadowing leads to suppression of particle production at small impact parameters : @xmath41 i.e. multiplicity distribution @xmath42 and mean multiplicity @xmath43 in the impact parameter representation have no absorptive corrections , but peripherality of @xmath44 leads to suppression of particle production at small impact parameters and the main contribution to the integral multiplicity @xmath45 comes from the region of @xmath46 ( eq . ( [ mm ] ) ) .
this would lead to the events with alignment observed in eas and also to the imbalance between orbital angular momentum in the initial and final states since particles in the final state will carry out large orbital angular momentum . to compensate this orbital momentum spins of secondary particles should become lined up , i.e. the spins of the produced particles should demonstrate significant correlations when the antishadow scattering mode appears @xcite .
thus , the observed phenomena of alignment in eas @xcite and predicted spin correlations of final particles should have a common origin .
the model estimate for the primary energy when these phenomena should appear is @xmath47 @xmath12 @xmath48 is the energy when the new unitarity regime starts to develop at small impact parameters .
the detected particle composition of the eas is closely related to the quantity known as gap survival probability .
antishadowing leads to the nonmonotonous energy dependence of this quantity @xcite .
the gap survival probability , namely the probability to keep away inelastic interactions which can result in filling up by hadrons the large rapidity gaps , reaches its minimal values at the tevatron highest energy and this is due to the fact that the scattering at this energy is very close to the black disk limit at @xmath20 ( fig .
3 ) . [ fig4 ]
it is clear that its higher value means higher fraction for diffractive component and consequently the increasing of this component would result in the enhancement of the relative fraction of protons in the observed cosmic rays spectrum .
otherwise , decreasing of this quantity will lead to increase of pionization component and consequently to the increasing number of muons observed as multi - muon events .
experiment reveals that relative fraction of protons in cosmic rays also shows nonmonotonous energy dependence ( cf .
fig .
4 ) . to explain such dependence an additional component
is introduced _
ad hoc _ at the energies above @xmath49 gev .
it was shown that account of the antishadowing makes an introduction of this _ ad hoc _ component unnecessary .
the inelasticity parameter @xmath50 , which is defined as ratio of the energy going to inelastic processes to the total energy , is important for the interpretation of the eas cascades developments .
its energy dependence is not clear and number of models predict the decreasing energy dependence while other models insist on the increasing energy behaviour at high energies ( cf . e.g. @xcite ) .
adopting simple ansatz of geometrical models
where parameter of inelasticity is related to inelastic overlap function we can use the following equation for @xmath51 @xcite @xmath52 to get a qualitative knowledge on the inelasticity energy dependence .
the estimation of inelasticity based on the particular model with antishadowing @xcite leads to increasing dependence of inelasticity with energy till @xmath53 gev . in this region inelasticity reaches maximum value @xmath54 , since @xmath55 and then starts to decrease at the energies where this ratio goes beyond the black disk limit @xmath56
. such qualitative nonmonotonous energy dependence of inelasticity is the result of transition to the antishadowing scattering regime .
the distribution on the inelasticity is related to the distribution on the effective mass number , i.e. changes of @xmath57 are equivalent to changes of @xmath51 , and , for example , high - inelasticity primary proton interaction produces the same result at the ground level as the low - inelasticity primary interaction of the heavy nuclei @xcite .
the available experimental data on the average logarithm of the effective nuclear mass number , extracted from the energy dependence of the depth of eas maximum , have large error bars , but they also indicate a nonmonotonous energy dependence with the maximum in the region @xmath58 gev @xcite .
it is also worth to note that the maximum in inelasticity energy dependence , when the pionization component is maximal , is correlated with the minimum of the relative component of protons in the eas , the following simple relation can be supposed @xmath59 i.e. the relative proton component in the detected eas should have a non - monotonic energy dependence and this is in agreement with the experimental analysis represented in fig . 4 . [ pfrac ] it should be noted that the behaviour of the ratio @xmath60 when it goes to unity at @xmath14 does not imply decreasing energy dependence of @xmath34 .
the inelastic cross
section @xmath34 increases monotonically and it grows as @xmath18 at @xmath14 .
such a dependence of @xmath34 is in good agreement with the experimental data and , in particular , with the observed falling slope of the depth of shower maximum distribution @xcite .
the predicted numerical value of the inelastic cross - section is @xmath61 @xmath62 at the lhc energy @xmath63 @xmath64 .
this value is also in a good agreement with the value for this quantity extracted from the proton - air inelastic cross - section @xcite .
this approach provides a reasonable description @xcite of the energy dependence of mean multiplicity and leads to its power - like growth with a small exponent .
the relation of the knee and other effects observed in the eas measurements with the modification of particle generation mechanism is under discussion since the time when they were discovered .
we propose here one particular realization of this idea an approach where the corresponding particle generation mechanism in eas is strongly affected by the unitarity effects and the energy region between the knee and the ankle coincides with the transition region to the scattering mode where antishadowing develops at small and then at moderate values of impact parameter , i.e. the energy spectrum of the primary cosmic particles @xmath65 is modulated by the significant variation of the scattering matrix @xmath17 in the energy region starting from about @xmath66 gev and finishing at about @xmath67 gev and this resulting in the regularities in the observed spectrum @xmath68 measured in the eas studies . below the energy @xmath69 and beyond the energy @xmath70 variation of scattering matrix is slow and the primary energy spectrum @xmath71 is almost not affected .
it seems to be a rather natural explanation of the observed regularities in the eas measurements and has a close interrelation with the nonmonotonous energy dependence of gap survival probability and inelasticity .
this hypothesis is based on the saturation of the unitarity and can be experimentally checked at the lhc @xcite .
the studies of the proton scattering in the forward region at the lhc will be very helpful for improving the interpretation of the results of the cosmic rays experiments . |
we have detected resonances in n - body simulations with evolving disks in live halos using only trajectories of the particles .
we find that all resonances are trapping .
no gaps or scattering resonances are found . in our models , there are little evolution after particles get trapped .
the corotation resonance captures particles in a ring outside the bar .
another important result is that the inner lindblad resonance is not localized at a given radius .
we have also detected trapping resonances in the halo . | the dynamics of the inner disk of many galaxies is dominated by bars . as a result
, resonances between the bar and the disk become an important factor . in order to detect resonances , we measure angular and radial frequencies of individual orbits in n - body models of barred galaxies . in our simulations with several million particles , stellar disks are immersed in live halos of dark matter . a bar forms spontaneously and interacts with the disk through resonances .
we detect 9 of them .
the corotation ( cr ) and the inner lindblad resonance ( ilr ) are the most significant .
the spatial distributions of particles trapped at these resonances show that ilr is not localized at a given radius .
particles trapped at ilr lie along the bar spanning a wide range of radii . on the other hand , particles at the corotation resonance form a ring around the corotation radius .
we study how orbits evolve near corotation and how they are trapped .
when an orbit is getting trapped , it evolves strongly , but after that there is little evolution .
those trapped orbits circulate along the corotation ring or librate around a stable lagragian point .
the same technique is applied to study resonances in the halo .
the bar interacts strongly with the halo particles that stay close to the disk .
we find the corotation resonance and the inner lindblad resonance for halo particles .
this paper contains the figures presented in the conference : " dynamics of galaxies : baryons and dark matter ( las vegas , nv , march 2005 ) . the full talk is available in http://astronomy.nmsu.edu/dept/html/directory.grads.danielcv.shtml |
spin - gap antiferromagnets have been actively studied during the last decades . because of particular geometry of the spin - spin interactions ( frequently featuring dimer motives ) these systems of strongly exchange coupled spins do not order antiferromagnetically , but remain in disordered spin - singlet ( @xmath5 ) state down to the lowest temperatures .
stability of the singlet state is ensured by an energy gap @xmath6 of exchange origin separating the ground state from the triplet ( @xmath7 ) excitations . among examples of such systems
are : a spin - peierls magnet cugeo@xmath8 @xcite , a dimer magnet tlcucl@xmath8@xcite , various haldane magnets ( e.g. , pbni@xmath2v@xmath2o@xmath9 @xcite or organo - metallic compound nenp @xcite ) and various spin - ladder systems @xcite .
application of an external magnetic field leads to the zeeman splitting of the triplet states and to lowering of the energy of one of its substates .
this leads to a quantum critical point at a certain critical field @xmath10 .
this quantum critical point has been actively discussed recently since it is formally quivalent to bose - einstein condensation @xcite .
if a spin subsystem is three - dimensional , then a field induced antiferromagnetically ordered phase is formed above the critical field @xcite .
controlled introduction of defects is another way to affect stability of the ground state .
bond - doping , i.e. introduction of defects that affect interspin couplings but do not affect magnetic sublattices , is of particular interest as a way to realize different random coupling models .
theoretical models considering such systems make interesting predictions , including formation of a glass - like state@xcite or a random singlet state @xcite .
this mechanism was already exploited in the study of random - bond systems ipa - cucl@xmath8 @xcite , sul - cu@xmath2cl@xmath0 @xcite , and dtn@xcite
. recently found organo - metallic compound ( c@xmath0h@xmath1n@xmath2)(cu@xmath2cl@xmath3)(abbreviated as phcc for piperazinium hexachlorodicuprate ) is a good test model to study quantum critical behaviour and the effect of impurities . the pure compound is well characterised by a variety of techniques including bulk measurements , elastic and inelastic neutron scattering @xcite and electron spin resonance @xcite .
these measurements have confirmed the existence of an energy gap @xmath11 mev , triplet nature of magnetic excitations and closing of the energy gap by a magnetic field above approximately 8 t .
inelastic neutron scattering experiments revealed the presence of at least six relevant exchange couplings@xcite , the exchange couplings geometry can be envisioned as a 2d set of moderately coupled spin - ladders with the dominant interaction on the rung of the ladder and some frustrating couplings .
this system allows to perform a substitution of chlorine ions mediating the interspin interactions by bromine with the nominal br concentration above 10% . as it was shown by neutronographic and calorimetric experiments @xcite doping results in an increase of the gap and the critical field , as well as in a broadening of the excitation spectrum and in a change of the crossover critical exponent .
the magnon damping in the doped samples turned out to be strongly dependent on the wavevector with damping rate increasing away from the zone center @xcite . here
we present the results of an electron spin resonance ( esr ) study of the bromine doped phcc single crystals .
high energy resolution of esr technique and its ability to distinguish between different paramagnetic centers allows us to demonstrate that particular kind of s=1 paramagnetic centers are formed with bond doping : low temperature esr response is dominated by resonance absorption of gapless triplets . at the same time , increase of esr linewidth indicates reduction of lifetime of thermally excited triplet excitations .
samples of pure and bromine substituted ( c@xmath0h@xmath1n@xmath2)(cu@xmath2cl@xmath12br@xmath13)(phxc for short for the @xmath14 samples ) were grown from solution as described in @xcite . the saturated solution for the growth of pure phcc
was prepared by adding of piperazine c@xmath0h@xmath15n@xmath2 ( from sigma aldrich ) , dissolved in a minimal amount of concentrated hcl , to cucl@xmath16h@xmath2o ( 99.99% purity from sigma aldrich ) , dissolved in a minimal amount of concentrated hcl , at a 4:1 molar ratio of cucl@xmath2 to c@xmath0h@xmath15n@xmath2 .
saturated solution for the diluted crystals was prepared by introducing a proportional amount of cubr@xmath16h@xmath2o and hbr .
we will use nominal bromine concentration ( determined by the br to cl ratio in the initial solution components ) for the sample identification . as grown samples have a typical size of @xmath17 mm@xmath18 with a well developed plane normal to the @xmath19-axis and the longest edge of the sample parallel to the @xmath20-axis .
details of the lattice structure ( which is triclinic ) can be found in @xcite .
the quality of the samples was checked by x - ray diffraction on a bruker apex ii diffractometer .
it was found that up to the nominal bromine content of 12.5% lattice symmetry remains the same ( triclinic ) and lattice constants slowly increase @xcite . above 12.5%
a phase separation was found to occur .
detailed structural analysis @xcite has demonstrated that the average bromine concentration @xmath21 in ( c@xmath0h@xmath1n@xmath2)(cu@xmath2cl@xmath12br@xmath13 ) is slightly below the nominal value @xmath22 , @xmath23 , and the occupation of different halogen positions varies : bromine ions enter at almost nominal concentration on the positions responsible for the strongest couplings ( rung and leg of the ladder ) , while bromine occupation of the positions responsible for the interladder couplings is by a factor of three smaller .
the static magnetization of all samples was measured with a quantum design mpms - xl system .
the m(t ) curves are shown on fig.[fig : magn ] .
all curves demonstrate a broad maximum around 15k .
below this temperature magnetization decreases due to the diminishing population of the gapped magnetic excitations . since the exchange bonds structure of phcc is complicated ( at least six relevant couplings are suggested in ref.@xcite ) we do not attempt to fit these magnetization curves with any compact simple model .
inelastic neutron scattering study of ref.@xcite demonstrated that the gap in the spectrum increases with doping and is above 10k for both pure and doped samples .
thus , contribution of the thermally excited triplets to the magnetic susceptibility at 2k is negligible ( see reference line in figure [ fig : magn ] ) .
the increase of magnetization at lowest temperatures is due to the presence of some paramagnetic centers .
a crude estimation of the amount of paramagnetic centers can be performed by a direct comparison of the 2k magnetization data with the curie law .
the amplitude of the magnetization at low temperatures corresponds to the concentration of paramagnetic centers ( assuming @xmath24 @xmath25 ) of about 1% per copper ion for the nominally 5% br - doped sample and about 1.5% per copper ion for the nominally 10% br - doped sample ( see dashed lines in figure [ fig : magn ] ) .
electron spin resonance experiments were carried out with the help of home - made transmission type esr spectrometers at the frequencies 10 - 70 ghz and at temperatures down to 450 mk .
limited sensitivity of our spectrometer required usage of the quite big samples ( about @xmath17 mm@xmath18 ) which in turn limited possibilities for the sample mounting . for this reason esr experiments for all samples were carried out for the magnetic field @xmath26 orientation .
for the same reason , as the sample size is comparable to the microwave wavelength , the field distribution inside the sample is unknown a priori ( in particular , different polarizations of microwave magnetic field with respect to the static field are present ) , but this distribution remains the same for each series of experiments ( with the same microwave frequency and the same sample mounting ) .
here we briefly recall some basics of esr technique that will be necessary to interpret experimental data @xcite .
first , esr technique allows to discern different paramagnetic centers by their resonance fields .
this allows to determine both @xmath27-factor and spin of the paramagnetic center .
characteristic fingerprint of a @xmath28 center is splitting of the absorption line due to zero - field splitting of its spin sublevels ( so called crystal field splitting which is described in the simplest case by axial term in the spin hamiltonian @xmath29 ) . magnitude and sign of the crystal field parameter @xmath30 can be deduced from the positions and relative intensities of the split absorption components . in particular , for a @xmath7 center a direct transition between @xmath31 sublevels ( so called `` two - quantum '' transition since it corresponds to @xmath32 ) results in the characteristic absorption signal at one half of the usual paramagnetic resonance field .
zero - field splitting of the collective triplet excitations of a spin - gap magnet can be interpreted as effective crystal field microscopically originating from different anisotropic spin - spin interactions .
unravelling the exact microscopic origin of this splitting is a separate task , this problem has been solved by different approaches for various spin - gap magnets in literature @xcite .
second , esr absorption can be interpreted in absolute units . in our experiment
the microwave power transmitted through the cavity with the sample is recorded as a function of slowly swept magnetic field at a fixed microwave frequency . for a weakly absorbing point sample the decrease of the transmitted microwave power is proportional to the imaginary part of the high - frequency susceptibility @xmath33 for the appropriate polarization of the microwave magnetic field . for a finite size sample
the absorption has to be integrated over the sample volume taking into account the variation of the microwave field inside the sample .
the proportionality coefficient between @xmath33 and the decrease of detector output @xmath34 is a priory unknown , it depends on the microwave fields distribution , cavity q - factor and other technical details .
however , it remains the same for each series of experiments .
thus , we have to scale our absorption with a reference data provided by static susceptibility measurements .
if the resonance absorption spectrum consists of a single narrow paramagnetic resonance line kramers - kroenig relations can be used to scale the integrated intensity of the paramagnetic absorption due to the microwave field perpendicular to the static field with the static susceptibility @xmath35 : @xmath36 this allows to establish the scaling factor for the given experiment and to calculate the imaginary part of the susceptibility in absolute units . which , in turn , allows to estimate the amount of paramagnetic centers .
note , that for the case of @xmath7 such a scaling is applicable only to the absorption components excited by the transversely polarised microwaves and not to the `` two - quantum '' absorption . finally , the intensity of the esr absorption due to transitions between spin sublevels @xmath37 and @xmath38 ( @xmath39 ) depends on the sublevels populations and matrix element of the transition .
@xmath40 here the integration is performed over the sample volume to take into account the distribution of microwave field polarizations @xmath41 through the sample .
temperature dependence of a selected component intensity is determined by the combination of the boltzmann exponents .
the results of esr investigation of the nominally pure ( @xmath42% ) phcc are reported in detail in @xcite : at high temperatures ( above 20k ) a single resonance absorption component is observed with an anisotropic @xmath27-factor slightly above 2.0 , as it is typical for cu@xmath43 ions . below 20k
the esr absorption loses intensity as the magnetic excitations freeze out .
as the population of magnetic excitations drops down , their interaction vanishes and the zero - field splitting of the triplet @xmath7 excitations appears . in figure
[ fig : low-17ghz - scans ] we present reference data for the @xmath42% sample taken at the same frequency and orientation as the data for the bromine - diluted samples . as was described above , there is a single component esr absorption line above 5k .
the esr absorption line splits around 4k into three components : two main components ( fig.[fig : low-17ghz - scans ] ) corresponding to @xmath44 transitions and the weaker component at approximately half of the paramagnetic resonance field corresponding to the @xmath45 transition .
we mark these components in the order of increasing resonance field as `` a '' , `` b '' and `` c '' corresponding to the `` two - quantum '' transition and the two main components .
splitting of the main components and their relative intensities are related to the parameters of the effective crystal field and to the orientation of the anisotropy axes with respect to the field .
intensity of the `` c '' component slightly exceeds the intensity of the `` b '' component .
split components continue to lose intensity on cooling and below 2.0k the esr absorption spectrum mainly consists of the paramagnetic absorption due to the defects and impurities ( marked as `` pm '' ) .
temperature evolution of esr absorption for the sample with low nominal bromine content ( @xmath46% ) is qualitatively the same as for the pure sample ( fig.[fig : low-17ghz - scans ] ) .
on cooling below 10k the esr absorption loses intensity , the esr line broadens and splits at approximately 4k into three components ( again labelled as `` a '' , `` b '' and `` c '' on fig.[fig : low-17ghz - scans ] ) .
these components freeze out on further cooling and at the lowest temperature the esr absorption spectrum consists of single line with @xmath27-factor close to 2.0 , which is related to defects and impurities .
thus , the observed esr absorption in the low - doped sample is mostly due to the triplet excitations , as in the pure sample . as in the case of pure compound , intensity of the `` c '' component for the 1% doped sample
slightly exceeds that of the `` b '' component , which indicates that the effective anisotropy constant is also with the same sign .
temperature evolution of esr absorption in the samples of phxc with higher nominal bromine concentration @xmath475% and 10% ( fig.[fig : high-17ghz - scans ] ) is qualitatively different from that of the lightly doped sample . at temperatures above 4 k
the intensity of absorption decreases with cooling and the splitting of the esr line into similar components occurs .
but the split components remain clearly visible even at the lowest temperatures .
below we will focus our consideration on the sample with @xmath48% , absorption spectra for the sample with @xmath49% is qualitatively similar to the @xmath48% case . to check whether this absorption signal corresponds to the @xmath7 objects or not and to follow its temperature evolution , we performed measurements at temperatures down to 450mk on the @xmath48% sample and measured the frequency - field dependence of the observed esr absorption signal .
these data demonstrate ( fig .
[ fig:10-he3-scans ] ) that `` a '' and `` b '' components of the split absorption spectrum survive down to the lowest temperatures and gain intensity on cooling , while `` c '' component loses intensity below 1k and almost disappears at 450 mk .
the frequency - field dependences ( fig .
[ fig:10-f(h ) ] ) for all three modes are in quantitative agreement with known @xcite frequency - field dependences for a @xmath7 object in an axial crystal field with @xmath27-factor @xmath50 and effective anisotropy constant @xmath51 ghz .
contrary to the case of pure and low - doped samples , the integrated intensity of `` b '' component at @xmath52k ( fig .
[ fig : high-17ghz - scans ] ) slightly exceeds the intensity of the `` c '' component , this difference becomes obvious on cooling below 1k ( fig .
[ fig:10-he3-scans ] ) .
thus , the effective anisotropy constant in @xmath26 orientation for the high - doped samples has a different sign from that for the pure or low - doped samples .
at temperatures above 6k esr absorption line for all the samples consists of a single component and demonstrates an increase of the intensity on heating .
this increase of intensity is in an agreement with the increase of static susceptibility on heating in the same temperature range .
thus , esr absorption at these temperatures is dominated by the triplet excitations of the spin - gap magnet .
comparison of esr absorption lines measured at the same temperature @xmath53k ( figure [ fig:10k - width ] ) demonstrates a monotonous increase of linewidth with doping . as esr linewidth
is directly related to the lifetime of the spin precession , i.e. to the lifetime of the triplet excitations , this observation is in qualitative agreement with increase of magnon damping rate observed in neutron scattering experiments .
@xcite the `` pm '' mode observed in all samples with @xmath27-factor close to 2 increases in intensity on cooling and has an irregular lineshape .
this allows to interpret it as being related to some uncontrolled paramagnetic defects ( e.g. , off - site copper ions , paramagnetic impurities etc . ) . to estimate the concentration of the paramagnetic centers we scale the integral intensity of the `` pm '' mode measured at 17.5ghz ( figures [ fig : low-17ghz - scans ] , [ fig : high-17ghz - scans ] ) at low temperatures
( between 1.5k and approximately 3k ) with the total intensity of the esr signal at high temperature ( 10k ) .
the high temperature total intensity can then be scaled to the measured static magnetization ( see eqn.([eqn : chi - scaling ] ) ) .
then the concentration of the paramagnetic centers can be estimated from the curie law .
we neglect here anisotropy of the phcc @xmath27-factor which would yield insignificant corrections @xcite . we have found that the intensity of the `` pm '' mode in all the samples corresponds to approximately 0.1% @xmath25 @xmath54 ( measured @xmath27-factor ) paramagnetic centers per copper ion ( table [ tab : pm ] ) .
.estimated amount of paramagnetic defects ( assuming @xmath25 @xmath54 ) in different samples .
[ cols="^,^ " , ] apparent increase of amplitude of the `` pm '' mode with increasing bromine concentration in figures [ fig : low-17ghz - scans],[fig : high-17ghz - scans ] is purely visual effect due to broadening of the main absorption line .
note that the low - temperature magnetization of the doped samples ( figure [ fig : magn ] ) corresponds to ten - fold higher amount of paramagnetic centers .
this discrepancy with esr observations rules out trivial creation of paramagnetic defects on doping .
additionally , the estimated amount of the paramagnetic defects is much smaller than the nominal br concentration .
this implies that br substitution indeed affects the magnetic bonds instead of creating additional structural or magnetic defects in agreement with x - ray analysis of ref.@xcite .
the esr absorption spectra measured at the lowest temperatures ( fig .
[ fig:10-he3-scans ] ) prove that the low - temperature absorption in the high - doped samples is dominated by some other @xmath7 objects affected by the effective crystal field @xmath55 ( here we assume @xmath56).the magnitude of the constant @xmath57 ghz and the @xmath27-factor value @xmath50 are determined from the frequency - field dependence @xmath58 , see figure [ fig:10-f(h ) ] .
the negative sign of the @xmath59 is evidenced by the higher relative intensity of the `` b '' component as this transition corresponds to the excitation from the lowest sublevel ( see insert of figure [ fig:10-f(h ) ] ) .
temperature dependence of the esr absorption intensity ( figure [ fig : he3-i(t ) ] ) was determined for components `` a '' , `` b '' and `` c '' by fitting them with a lorentzian lineshape .
determination of intensity of mode `` a '' is the most reliable since this mode is well separated from other absorption components . but
mode `` a '' can not be used for the determination of the @xmath7 objects concentration because of the unknown excitation conditions for this `` two - quantum '' transition .
the modes `` b '' and `` c '' are excited by the conventional transverse polarization and their intensity can be scaled with the static susceptibility data .
closeness of the modes `` b '' and `` c '' with the defects contributed mode `` pm '' complicates fit procedure , leading in particular to the systematic overestimation of the intensity of the weaker mode `` c '' located on the shoulder of the `` pm '' mode .
however , as the scaling with the static magnetization is done at 5k where the `` pm '' mode intensity is relatively weak and the low - temperature ( below 0.6k ) absorption is dominated by the mode `` b '' , the absolute value of esr absorption at low temperatures ( which is determined by the concentration of the @xmath7 objects ) can be determined reliably . when scaling the esr absorption with the static susceptibility we neglect the @xmath27-factor anisotropy which would yield only small ( about 5% ) correction in our case @xcite .
to model temperature dependence of the esr intensity we assumed that these @xmath7 objects are gapless with energy of the field - independent @xmath60 substate being zero .
temperature dependences of esr intensities can be then deduced from eqn.[eqn : esr - absorption ] ( see appendix for details ) with known microwave frequency , @xmath27-factor and anisotropy constant , leaving the scaling factors as the only fitting parameters .
experimentally measured intensities of the esr modes follow modelled curves well .
best fit is obtained with the concentration of the @xmath7 objects @xmath61 per molecule of phxc .
the results of this fit are shown on figure [ fig : he3-i(t ) ] . at higher temperatures ( above 4k )
the observed esr absorption is dominated by thermally excited triplets .
temperature dependence of the total esr absorption intensity in the temperature range from 0.45k to 6k can be fitted by a sum of contributions from the @xmath7 objects and a gapped contribution @xmath62 with the gap value fixed to @xmath63k .
the gap value was obtained by extrapolation of the neutronographic data @xcite ( see dashed line at figure [ fig : he3-i(t ) ] ) .
also , note that as temperature increases , the dominating esr component slowly switches from the low - field `` b '' component to the higher - field `` c '' component .
this redistribution of absorption intensities follows crossover from the absorption dominated by gapless triplets at low temperatures to the absorption dominated by triplet excitations at higher temperatures .
the origin of these @xmath7 gapless triplets can not be unambiguously recovered from our data .
however , the finding that concentration of these objects is close to 1% per phxc molecule for the 10% nominal doping suggests the following explanation .
the building block of phcc magnetic subsystem is a cu@xmath2cl@xmath3 dimer .
two of the chlorine ions are responsible for the intradimer coupling and the four remaining chlorines participate in the formation of the inter - dimer couplings . at the bromine doping level of 10% ( assuming uniform doping ) the probability to find a dimer with both intra - dimer bond - forming halogens being bromine ions is exactly 1% .
a detailed structural analysis of @xcite suggested that the occupation of the inter - dimer halogen positions ( position cl@xmath64 in terms of @xcite ) is slightly less then the nominal value and is about 7.5% , which makes the probability of the double substitution 0.56% even closer to our estimation of the @xmath7 objects concentration .
thus , dimers with two of the intra - dimer exchange bonds modified by bromine substitution are the possible source of the gapless @xmath7 triplets . these triplets could be interpreted as a localized triplet mode , created by a potential well centered on such a doubly substituted dimer . in the limit of strong localization
this localized triplet mode transforms into a ferromagnetically coupled pair of copper spins . finally , we would like to note that the contribution of the @xmath7 spin to the susceptibility exceeds that of the @xmath25 spin by the factor of 8/3 ( spin enters curie law as @xmath65 ) .
thus , the determined concentration of 0.4% of @xmath7 objects per phcc molecule ( or 0.2% per copper ion ) corresponds to the concentration of about 0.5% of @xmath25 centers per copper ion which is reasonably close to the amount estimated from the magnetization data ( figure [ fig : magn ] ) .
additional contribution to the static magnetization of the doped phxc should also arise from a van - vleck - like mechanism suggested in @xcite .
we have studied the effect of bond - doping on the spin - gap magnet ( c@xmath0h@xmath1n@xmath2)(cu@xmath2cl@xmath3 ) .
our experiments show that substitution of chlorine ions by bromine up to 10% nominal doping does not lead to the destruction of the spin gap state .
we have shown , that introduction of bromine ions does not lead to formation of structural defects the concentration of paramagnetic @xmath25 centers remains practically the same ( about or below 0.1% ) for the full set of samples ranging from pure compound to 10% nominally doped sample .
however , bond - doping does affect thermally excited triplet excitations by shortening their lifetime , as evidenced by increase of esr linewidth .
finally , we have established that the low - temperature esr absorption is dominated by gapless @xmath7 triplets formed with doping .
the work was supported by the russian foundation for basic research ( rfbr ) projects no.12 - 02 - 31220 , no.12 - 02 - 00557 and no.15 - 02 - 05918 , russian presidential grant for the support of the leading scientific schools no.4889.2012.2 and no.5517.2014.2 .
this work was partially supported by the swiss national fund ( division 2 ) .
this project was partially supported by the estonian ministry of education and research under grant iut23 - 03 and estonian research council grant put451 .
if we assume that the @xmath7 objects responsible for the low - temperature esr absorption have no energy gap , then , taking energy of the field independent @xmath60 sublevel for zero , energies of the @xmath31 sublevels are @xmath66 .
the temperature dependence of the intensity for each of the resonance absorption components is described by the combination of boltzmann exponents ( eqn.([eqn : esr - absorption ] ) ) .
the main absorption components `` b '' and `` c '' correspond to the ordinary @xmath68 transitions and are excited by the microwave magnetic field perpendicular to the static field .
this allows to analyse the intensities of these modes quantitatively and to estimate the concentration of these gapless @xmath7 objects .
the scaling factor between the observed absorption intensity and the static susceptibility was calculated at temperatures from 5 to 6 k , which are the highest temperatures available in our he-3 experimental setup . at these temperatures
the intensity of the absorption signal related to the paramagnetic defects is negligible compared to the total intensity .
this allows to express the intensities of `` b '' and `` c '' components in terms of concentration of the @xmath7 objects : yu r , yin l , sullivan n s , xia j s , huan c , paduan - filho a , oliveira n f , haas s , steppke a , miclea c f , weickert f , movshovich r , mun e - d , zapf v s and roscilde t , nature * 489 * , 379 ( 2012 ) ( arxiv:1109.4403v2 ) static susceptibility was measured for @xmath74 , while the esr absorption is recorded for @xmath75 .
the absorption by thermally activated triplets , `` b '' , `` c '' and `` pm '' modes are excited by the transversely polarized microwaves with the microwave magnetic fields distributed in the @xmath76-plane . due to the big sample size ,
various polarizations are present .
value of the @xmath27-factor varies in the @xmath76-plane from 2.05 to 2.14 @xcite .
accounting for this @xmath27-factor variation would give a correction of about 5% for the scaling between the static susceptibility and the integrated esr absorption , which is much less then the accuracy of the concentration determination . | we report results of an electron spin resonance ( esr ) study of a spin - gap antiferromagnet ( c@xmath0h@xmath1n@xmath2)(cu@xmath2cl@xmath3)(nicknamed phcc ) with chlorine ions partially substituted by bromine .
we found that up to 10% of nominal doping the contribution of the random defects to the absorption spectra remains at about 0.1% per copper ion , almost the same as in the pure system . instead
, a particular kind of esr absorption corresponding to gapless s=1 triplets is observed at low temperatures in samples with high nominal bromine content @xmath4% .
increase of bromine concentration also leads to the systematic broadening of esr absorption line indicating reduction of the quasi - particles lifetime . |
although the standard cosmological model has been succesful in explaining the homogenous expansion of the universe and the 2.7 k cosmic microwave background radiation , it has some shortcomings like the initial singularity , horizon ( or causality ) , flatness , homogeneity and isotropy problems.@xcite among these problems the initial singularity problem may be the one which weakens the model much more than the others , in the sense that it causes infinities in the physical quantities such as the density , pressure and temperature . since these infinities can not be accepted by any model which claims to be physical , during the past two decades , several authors have considered the possibility of describing the universe with a singularity - free cosmological model . @xcite recent observations point out that the universe has an @xmath7 , @xmath8 value lower than unity which means that the universe is spatially open .
@xcite in the light of these , we propose an open singularity - free cosmological model with the same motivations as in ref .
however , initially static condition @xmath9 used by the closed universe models can no longer be used in the open universe case since it leads to negative energy densities . for this reason in our model universe
starts expanding with an initial velocity .
hence , we have a two - parameter universe model in which one of the parameters characterizes the equation of state used to describe the prematter era , and the other one corresponds to the initial expansion rate of the universe .
our model describes an open , spatially homogenous and isotropic universe with a space - time geometry given by the robertson - walker ( rw ) line element : @xmath10 , \ ] ] where @xmath11 are the comoving coordinates , @xmath12 is the scale factor which represents the size of the universe . for the rw metric , einstein field equations lead us to the following differential equations : @xmath13 @xmath14 where @xmath15 and @xmath16 are the energy density and pressure in the universe respectively , where we have used perfect fluid energy momentum tensor in comoving coordinates .
here a dot denotes differentiation with respect to the cosmic time @xmath17 and we use the units so that @xmath18 where @xmath19 is the planck mass . combining eqs .
( 2 ) and ( 3 ) and using an equation of state given as @xmath20 one obtains the following relation : @xmath21 where , @xmath22 is assumed to be a constant parameter during each era in the history of the universe .
we can eliminate @xmath15 between eqs .
( 2 ) and ( 5 ) and obtain an equation involving only the scale factor @xmath12 : @xmath23 to solve this equation for any @xmath22 , it is advantageous to work in conformal time @xmath24 which is defined as @xmath25 using this transformation , eq .
( 6 ) is solved to give two linearly independent solutions given as @xmath26 ^{1/c},\ ] ] @xmath27 ^{1/c},\ ] ] where @xmath28 and @xmath29 and @xmath30 are the integration constants .
these solutions correspond to different signs of the energy density @xmath31 . assuming that the universe is filled with positive energy , we eliminate eq .
( 9 ) and consider only eq .
( 8) as the solution for the scale factor of the universe . in this work , we model the universe as a series of perfect fluid eras connected by first order phase transitions .
according to the considerations of this scenario , the universe starts expanding with a period of rapid expansion called inflationary era .
this period is characterized by an equation of state in the form @xmath32 and inflation arises due to this vacuum - like characteristic of the equation of state .
since matter would be under extreme conditions during this period and behaves very differently from the ordinary matter , it is called `` prematter '' .
@xcite during this period due to the unusual characteristic of the equation of state , temperature increases although the universe expands enormously .
we assume that inflation continues until the temperature reaches the maximum allowed temperature i.e. the planck temperature @xmath33 @xmath34 .
this behavior , which does not necessitate a `` re - heating mechanism '' as in the other models of the universe , follows from the vacuum like characteristic of the equation of state used to describe the universe in this era .
initially , we assume that the universe is in a vacuum - like state and has a limiting density called the planck density @xmath35 which is first formulated by markov as a universal law of nature .
@xcite since our model describes a universe starting from a finite size and density , the initial expansion rate must be taken as positive .
hence we take the initial expansion rate as @xmath36 where @xmath37 is some positive constant .
solutions for the scale factor in different eras are:@xmath38^{1/c_{p}}\ & \ \ 0\leq
\eta \leq \eta _ { r } , \\ a_{0}^{(r)}[\sinh ( \eta + \delta _ { r } ) ] & \eta
_ { r}\leq \eta \leq \eta _ { m } , \\ a_{0}^{(m)}[\sinh ( \eta /2+\delta _ { m})]^{2}\ & \ \eta _ { m}\leq \eta . \end{array } \right.\ ] ] where _ p _ , _ r _ , and _
m _ denote the prematter , radiation and matter eras respectively .
we next impose the boundary condition that the scale factor and its derivative are continuous at points @xmath39 and @xmath40 where the phase transitions take place .
this determines the integration constants as , @xmath41 ^{\frac{1}{c_{p}}+1}d^{-1/2},\ ] ] @xmath42 @xmath43^{\frac{1}{c_{p}}% -1},\ ] ] @xmath44 @xmath45 @xmath46 as seen from the solutions for the scale factor , the model that we propose is a two - parameter model .
the parameters are the @xmath47 value and the initial value of the hubble constant @xmath48 .
the former determines the amount of inflation that the universe has experienced during the prematter era and the latter is related to the initial expansion rate of the universe
. we can find the comoving times corresponding to the conformal times @xmath49 @xmath40 and @xmath50 by using the definition given in eq .
assuming that @xmath51 at @xmath52 we get from eq .
( 8) @xmath53 ^{1/c}d\eta ^{\prime } .\ ] ] this integral depends on the values of @xmath54 and has to be computed numerically in the prematter era .
whereas , for the radiation @xmath55 and matter @xmath56 eras , eq .
( 18 ) could be integrated to yield analytical expressions .
the expressions for the comoving times corresponding to @xmath57 @xmath58 and @xmath50 are @xmath59^{1/c_{p}}d\eta , \ ] ] @xmath60 , \ ] ] @xmath61 .\ ] ] hubble constants at @xmath62 and @xmath50 could now be obtained by using @xmath63 as @xmath64^{\frac{1}{c_{p}}+1}}~km\cdot s^{-1}\cdot mpc^{-1},\ ] ] @xmath65 @xmath66 the evolution of the energy density is described by @xmath67 eq .
( 26 ) could then be solved to yield @xmath68 where @xmath69 and @xmath70 are the initial and final instants of any conformal time interval in a given era .
choosing @xmath71 we obtain respectively @xmath72 @xmath73 @xmath74
in order to see this dependence and the development of the universe in this model , we changed one of the parameters while fixing the other and provided some numerical results .
this allowed us to identify the character of the dependency of the model to each parameter .
first , we fix @xmath75 to @xmath76 @xmath77 , and changed @xmath2 in such a way that it took values between @xmath78 and @xmath79 it is to be noted that in this broad range , the value of hubble constant has a wide spectrum between @xmath80 @xmath77 and @xmath81 @xmath82 next , we set @xmath2 to @xmath83 and varied @xmath75 in the range between @xmath84 @xmath77 and @xmath85 @xmath77 .
the corresponding values for the hubble constant vary between @xmath86 @xmath77 and @xmath87 @xmath82
we construct an open singularity - free cosmological model with the assumptions that the universe is initially in a vacuum like state and the physical quantities are limited by their planck values .
evolution of the temperature of the universe is governed by its expansion . during the prematter era ,
temperature increases due to inflation and the universe is characterized by a vacuum like equation of state in the form @xmath88 where @xmath89 is a parameter which determines the vacuum dominance of the early universe .
hence , the model we construct is a two - parameter universe model , the parameters being @xmath75 and @xmath90 .
this era ends when the temperature reaches the maximum allowed temperature @xmath91 then the universe enters the radiation era and starts cooling down .
this cooling continues during the matter era described by the standard model .
the so - called flatness problem is not present in this model since the initial value of @xmath92 is not set to unity . in the standard model ,
such a precise initial condition has to be assumed without explanation to produce a universe resembling the actual one . in this model ,
the universe starts its journey with an @xmath92 value no matter how close to unity and during the prematter era @xmath92 is driven toward unity . at the end of this era ,
the universe is nearly flat since its @xmath92 value is very close to one .
this fact could be attributed to the inflation mechanism which causes the universe to expand enormously in a very small time interval ( planck time ) and thus become flatter than at the beginning . during the radiation and matter eras , @xmath92 is driven away unity which apparently displays the open character of the space - time geometry of the model . in this work ,
we have not been interested in the microphysics of the inflationary era .
this includes quantum mechanical investigation of the material content of the early universe and the form of the scalar field potential responsible for the inflation .
however , with this simple form , the scenario proposed in this work might provide a guidance for the future more complete versions of the open inflation . | in the light of recent observations which point to an open universe @xmath0 , we construct an open singularity - free cosmological model by reconsidering a model originally constructed for a closed universe .
our model starts from a nonsingular state called prematter , governed by an inflationary equation of state @xmath1 where @xmath2 @xmath3 is a small positive parameter representing the initial vacuum dominance of the universe . unlike the closed models universe
can not be initially static hence , starts with an initial expansion rate represented by the initial value of the hubble constant h(0 ) .
therefore , our model is a two - parameter universe model @xmath4 . comparing the predictions of this model for the present properties of the universe with the recent observational results
, we argue that the model constructed in this work could be used as a realistic universe model .
we started the evolution of a flat universe from a nonsingular state called prematter which is governed by an inflationary equation of state @xmath1 where @xmath2 represents the initial vacuum dominance of the universe .
the evolution of the universe - except in the prematter era - is affected neither by the initial vacuum dominance nor the initial expansion rate of the universe . on the other hand , present properties of the universe such as hubble constant ,
age and density are sensitive to the temperature at the decoupling over a range between value between 50 and 80 @xmath5 for the present value of the hubble constant @xmath6 . assuming that the thermal history of the universe is independent from its geometry , the above range could be considered as transition range for the decoupling temperature |
in the past , divergent series have played a central role in mathematics .
mathematicians like for instance lacroix , fourier , euler , laplace , etc .
have used them extensively .
nowadays they play a marginal role in mathematics and are often not mentioned in the standard curriculum .
a number of mathematicians and most students are not aware that they can be of any use . where was the turn ?
according to @xcite , the turn occured in the time of cauchy and abel , when the need was felt to construct analysis on absolute rigor .
so let us go back a little and see what cauchy and abel have said of divergent series .
cauchy ( preface of analyse mathmatique `` , 1821 ) : _ jai t forc dadmettre diverses propositions qui paratront peut - tre un peu dures .
par exemple quune srie divergente na pas de somme " _
( i have been forced to admit some propositions which will seem , perhaps , hard to accept .
for instance that a divergent series has no sum '' ) cauchy made one exception : he justified rigorously the use of the divergent stirling series to calculate the @xmath0-function .
we will explain below the kind of argument he made when looking at the example of the euler differential equation .
abel ( letter to holmboe , january 16 1826 ) : _ les sries divergentes sont en gnral quelque chose de bien fatal et cest une honte quon ose y fonder aucune dmonstration .
on peut dmontrer tout ce quon veut en les employant , et ce sont elles qui ont fait tant de malheurs et qui ont enfant tant de paradoxes .
enfin mes yeux se sont dessills dune manire frappante , car lexception des cas les plus simples , par exemple les sries gomtriques , il ne se trouve dans les mathmatiques presque aucune srie infinie do nt la somme soit dtermine dune manire rigoureuse , cest - - dire que la partie la plus essentielle des mathmatiques est sans fondement .
pour la plus grande partie les rsultats sont justes il est vrai , mais cest l une chose bien trange .
je moccupe en chercher la raison , problme trs intressant . " _
( divergent series are , in general , something terrible and it is a shame to base any proof on them . we can prove anything by using them and they have caused so much misery and created so many paradoxes . . finally my eyes were suddenly opened since , with the exception of the simplest cases , for instance the geometric series , we hardly find , in mathematics , any infinite series whose sum may be determined in a rigorous fashion , which means the most essential part of mathematics has no foundation . for the most part , it is true that the results are correct , which is very strange .
i am working to find out why , a very interesting problem . " ) the author was struck by this citation when she first read it 27 years ago and this citation has followed her for her whole career . in her point of view , this citation contains the past , present and future of divergent series in mathematics . * the past : * as remarked by abel , divergent series occur very often in many natural problems of mathematics and physics . their use has permitted to do successfully a lot of numerical calculations .
one example of this is the computation by laplace of the secular perturbation of the orbit of the earth around the sun due to the attraction of jupiter .
the calculations of laplace are verified experimentally , although the series he used were divergent . *
the present : * in the 20-th century divergent series have occupied a marginal place .
however it is during the same period that we have learnt to justify rigorously their use , answering at least partially abel s question . in the context of differential equations the borel summation , generalized by calle and others to multi - summability ,
give good results ( see for instance @xcite , @xcite , @xcite ) .
* the future : * why do so many important problems of mathematics lead to divergent series ( see for instance @xcite ) ? what is the meaning of a series being divergent ? finding an answer to
this question is a fascinating research field .
we will illustrate all this on the example of the euler differential equation : @xmath1 since this is a short paper , the list of references is by no means exhaustive .
the euler differential equation has the formal solution @xmath2 which is divergent for all nonzero values of @xmath3 .
on the other hand , it is a linear differential equation whose solution can be found by quadrature : @xmath4 the integral is convergent for @xmath5 and hence yields a solution of .
we can rewrite this solution as @xmath6 in which we make the change of coordinate @xmath7 .
this yields @xmath8 the integral is convergent for @xmath9 , so the solution @xmath10 is well defined for @xmath5 and moreover @xmath11 . what is now the link between the divergent power series @xmath12 and the function @xmath10 ?
we will show that the difference between @xmath10 and a partial sum @xmath13 is smaller than the first neglected term ( this part has been inspired by @xcite ) .
this is exactly as in the leibniz criterion for alternating series .
for any @xmath9 @xmath14 proof .
we use the following formula which is easily checked @xmath15 then @xmath16 using the following formula @xmath17 which implies @xmath18 this yields @xmath19 where @xmath20 , the remainder , is the difference between @xmath10 and the partial sum @xmath21 of the power series .
the result follows since @xmath22 @xmath23 the argument given here , which justifies rigorously the use of the divergent series in numerical calculations , is very similar to the argument made by cauchy for the use of the stirling series .
in particular , cauchy used the formula . *
if we use the power series to approximate the function @xmath10 , how good is the approximation ?
* we encounter here the main difference between convergent and divergent series . with convergent series , the more terms we take in the partial sum , the better the approximation .
this is not the case with divergent series . indeed ,
if we take @xmath3 fixed in the general term tends to @xmath24 .
so we are better to take the partial sum for which the first neglected term is minimum .
@xmath25 is minimum when @xmath26 . in that case @xmath27 .
we use stirling formula to approximate @xmath28 for @xmath29 large : @xmath30 this gives us @xmath31 which is exponentially small for @xmath3 small .
not only have we bounded the error made when approximating the function by the partial sum of the power series , but this error is exponentially small for small @xmath3 , i.e. very satisfactory from the numerical point of view
. the phenomenon we have described here is not isolated and explains the successes encountered when using divergent series in numerical approximations .
looking at what we have done with the euler equation , someone can have the impression we have cheated .
indeed we have chosen a linear differential equation , thus allowing us to construct by quadrature a function which is solution of the differential equation .
but what about the general case ? in general
, once we have a formal solution by means of a power series , we use a _ theory of resummation _ for finding a function which is a solution . in this paper
, we will focus on the _ borel method of resummation _ , also called 1-summability .
we start with the properties that an adequate theory of summability must satisfy : * properties of a good method of resummation ( see for instance @xcite ) : * we consider a series @xmath32 , to which we want to associate a number @xmath33 , called its sum : \(1 ) if @xmath32 is convergent , then @xmath33 should be the usual sum of the series .
\(2 ) if @xmath32 and @xmath34 are summable with respective sums @xmath33 and @xmath35 then @xmath36 is summable with sum @xmath37 .
\(3 ) a series @xmath32 is _ absolutely summable _ if the series @xmath38 is summable .
\(4 ) if @xmath32 and @xmath34 are absolutely summable with respective sums @xmath33 and @xmath35 , then the product of the series , @xmath39 , where @xmath40 , is absolutely summable with sum @xmath41 .
\(5 ) if @xmath32 is summable with sum @xmath33 , then @xmath42 is summable with sum @xmath43 .
\(6 ) ( in the context of differential equations ) if @xmath44 is summable with sum @xmath10 , then @xmath45 is summable with sum @xmath46 . *
the borel method of resummation for a series : * we present it in a way which proves at the same time the property ( 1 ) : the idea is to take a convergent series @xmath47 and to write its sum @xmath48 in a different way . for this purpose ,
we use and we write @xmath49 changing the order of the summation and the integral in the last line requires a proof , which we leave as an exercise . a series @xmath32 is _ borel - summable _ if the series @xmath50 has a nonzero radius of convergence , if it can be extended along the positive real axis and if the integral @xmath51 is convergent with value @xmath33 .
we call @xmath33 the _ borel sum _ of the series
. * example : * in the case of the solution of the euler differential equation we have @xmath52 .
then @xmath53 hence @xmath54 which is .
we adapt the definition of borel - summability to power series in a manner that will allow to use it for analytic extension . 1
. a power series @xmath55 is 1-summable ( borel - summable ) in the direction @xmath56 , where @xmath56 is a half - line from the origin in the complex plane if * the series @xmath57 has a nonzero radius of convergence , and the sum of the series is an analytic function @xmath58 on the disk of convergence , * the function @xmath58 can be extended along the half line @xmath56 , * and the integral @xmath59 is convergent with value @xmath60 .
+ we call @xmath33 the sum of the series .
a power series @xmath55 is 1-summable if it is 1-summable in all directions @xmath56 , except a finite number of exceptional directions .
* example : * the solution of the euler differential equation is 1-summable in all directions except the direction @xmath61 .
the problem in the direction @xmath61 comes from the singularity at @xmath62 in or .
a theory of resummation is useful if there are theorems associated to it .
for instance , for the borel summation , let us cite this theorem of borel : @xcite we consider an algebraic differential equation @xmath63 where @xmath64 is a multivariate polynomial . if @xmath65 is a formal solution of and @xmath66 is absolutely borel - summable with borel sum @xmath10 , then @xmath10 is solution of the differential equation and has the asymptotic expansion @xmath67 .
\(1 ) the sums of a 1-summable series in the different directions @xmath56 give functions which are analytic extensions one of the other as long as we move the line @xmath56 continuously through directions in which the series is 1-summable .
this yields a function defined on a sector with vertex at the origin .
more details can be found for instance in @xcite , @xcite and @xcite .
\(2 ) the borel sum of a divergent power series can never be uniform in a punctured neighborhood of the origin .
it is necessarily ramified .
this is what is known in the literature as the _ stokes phenomenon_. \(3 ) if a series @xmath68 has a radius of convergence equal to @xmath69 and its sum is a function @xmath10 for @xmath70 , then a theorem of complex analysis states that the function @xmath10 has a least one singularity on the circle @xmath71 .
the idea of borel is that a divergent series is a series with radius of convergence @xmath72 .
hence , we have at least one singularity hidden in some direction : for the euler differential equation , this is the direction @xmath61 .
the operation @xmath73 sends the singularity at a finite distance , as if we had blown the disk of convergence to bring it from radius @xmath74 to radius @xmath75 .
@xmath76 is called the _ borel transform _ of @xmath66 .
the 1-summability and its extensions have been extensively studied during the 20-th century .
extensions of the notion of 1-summability have been obtained by allowing ramifications @xmath77 .
then 1-summability in @xmath3 corresponds to @xmath78-summability in @xmath79 .
the notion of multi - summability has also been introduced : a series @xmath66 is multi - summable if it is a finite sum @xmath80 , each @xmath81 being @xmath82-summable .
explicit criteria allow to decide a priori that some systems of differential equations have multi - summable solutions in the neighborhood of certain singular points ( see for instance @xcite ) .
let us now look at a generalized euler differential equation @xmath83 for almost all functions @xmath84 analytic in the neighborhood of the origin and such that @xmath85 the formal solution of vanishing at the origin is given by a divergent series .
only in very special cases the equation has an analytic solution at the origin .
* example : * @xmath86 is the analytic solution of @xmath87 what is the difference between the equations and
? to understand we will apply successively the two steps : * complexify : we will allow @xmath88 ; * unfold : @xmath89 is a double singular point of each equation .
hence , we will introduce a parameter so as to split the double singular point into two simple singular points .
* complexification : * let us consider a function @xmath10 which is the borel sum of a solution of , and its analytic extension when we turn around the origin .
the function @xmath10 is ( figure [ fig1 ] ) : * uniform for ; * ramified for , and generically ramified for a solution of .
the two branches differ by a multiple of @xmath90 ( which is a multiple of a solution of the homogeneous equation ) . * how to understand that generically we should have ramification ? * to answer we unfold and embed into @xmath91 with @xmath92 .
we will limit here our discussion to @xmath93 , although all complex values of @xmath94 are of interest .
since solutions of the homogeneous equation associated to are given by @xmath95 , the local model of solutions at @xmath96 is given by @xmath97 with @xmath98 analytic .
hence , the solution @xmath98 ( corresponding to @xmath99 ) is the unique solution which is analytic and bounded at @xmath100 .
similarly the local model for solutions at @xmath101 is given by @xmath102 we now have two cases : \(1 ) if @xmath103 , then @xmath104 is analytic .
since @xmath105 is ramified at @xmath106 for @xmath107 , all solutions but @xmath104 are ramified .
it is of course exceptional that the extension of @xmath98 at @xmath108 be exactly the solution @xmath104 and , generically , we should expect that the analytic extension of @xmath109 is @xmath110 with @xmath107 .
hence the extension of @xmath98 should be ramified at @xmath106 . if this ramification holds till the limit @xmath111 we get figure 1(a ) .
\(2 ) let us now consider the case @xmath112 .
we must again divide the discussion in two cases : \(i ) in the generic case , @xmath104 is ramified : it contains one term of the form @xmath113 .
\(ii ) in the exceptional case , @xmath104 is analytic , and so are all solutions through @xmath106 .
hence , it is impossible for @xmath98 to be ramified at @xmath106 .
this case is excluded in the unfolding of an equation whose solution is ramified .
let us now summarize the very interesting phenomenon we have discovered : if the formal solution of is divergent , then in the unfolding : * necessarily @xmath98 is ramified at @xmath108 : the divergence means a form of _ incompatibility _ between the local solutions at two singular points @xmath114 , which remains until the limit at @xmath111 .
* * parametric resurgence phenomenon : * for sequences of values of the parameter @xmath94 converging to @xmath111 ( here @xmath115 ) , the pathology of the system is located exactly at one of the singular points . indeed , here
, the only way for the system to have an incompatibility is that one of the singular points be pathologic .
we have understood why divergence is the rule , and convergence the exception .
the phenomena described above are much more general than the context of described here , and are explored by the author and her collaborators in different contexts .
the common thread is that @xmath75-summability occurs when two equilibrium points of a dynamical system coallesce in a double point .
there are some rigid dynamics attached to each simple equilibrium , and these rigid dynamics do not match well until the limit when the two equilibria coallesce .
divergent series also occur in the phenomena involving small denominators .
a source of divergence in this case is the accumulation of particular solutions .
for instance , in the case of a fixed point of a germ of analytic diffeomorphism @xmath116 , @xmath117 , with @xmath118 , the divergence of the linearizing series could come from the accumulation of periodic points around the fixed point at the origin .
the study of this phenomenon is part of the work for which jean - christophe yoccoz received the fields medal in 1994 .
divergent series occur generically in many situations with differential equations and dynamical systems . the divergence of the series carries a lot of geometric information on the solutions .
for instance , if the formal power series solution of the euler equation had been convergent , its sum could not have been ramified .
y. ilyashenko , in the theory of normal forms of analytic differential equations , divergence is the rule and convergence the exception when the bryuno conditions are violated , _
moscow university mathematics bulletin _ , * 36 * ( 1981 ) 1118 . | the present paper presents some reflections of the author on divergent series and their role and place in mathematics over the centuries .
the point of view presented here is limited to differential equations and dynamical systems . |
while temperature is usually computed from the kinetic energy , it can also be measured from derivatives of the potential energy @xmath0 with respect to coordinates @xmath1 @xcite . in terms of the inverse temperature @xmath2 ,
with @xmath3 being the boltzmann constant , we have @xmath4 where @xmath5 denotes an ensemble average , and @xmath6 is the logarithmic derivative of the density of states @xmath7 in the microcanonical ensemble : @xmath8 @xcite , or the parameter in the canonical ensemble , etc . except in the canonical ensemble @xcite , the last expression in eq .
( [ eq : ctemp ] ) has an @xmath9 error in a system of @xmath10 degrees of freedom @xcite ( see also appendix [ apd : ctemp ] ) .
( [ eq : ctemp ] ) defines the so - called _ configuration temperature _
, for it depends only on coordinates .
it has been used in verifying a monte carlo simulation @xcite , constructing equilibrium @xcite and non - equilibrium @xcite microcanonical sampling , and building a thermostat @xcite , etc . as eq .
( [ eq : ctemp ] ) requires a continuous system with up to the second derivatives of the potential energy , it is inapplicable if the molecular potential is discontinuous @xcite , or if the second derivatives are difficult to compute .
here we show that one can avoid the derivatives by computing the potential - energy change caused by a virtual perturbation , and obtain formulas of the configuration temperature that remain valid for irregular potential energy functions .
we begin with a mechanical translation of eq .
( [ eq : ctemp ] ) by a random perturbation . given a configuration @xmath1 ,
a small perturbation @xmath11 in coordinates changes the potential energy by @xmath12 .
now consider a random @xmath13 that satisfies ( a ) the distribution of @xmath13 is symmetric with equal probabilities for any @xmath13 and @xmath14 ; ( b ) the components of @xmath13 are independent : @xmath15 , where the overline denotes an average over the perturbation @xmath13 , and @xmath16 is the standard deviation of any @xmath17 .
on average , we get @xmath18 and @xmath19 to the leading order . further averaging over @xmath20 ( denoted by @xmath21 )
yields @xmath22 the above perturbation need not to cover all coordinates , since eq .
( [ eq : mechtrans ] ) holds for a single coordinate : @xmath23 .
the second step follows from a generalized formula @xmath24
@xcite with @xmath25 .
( [ eq : mechtrans ] ) shows that the temperature can be obtained from the statistical moments , or , equivalently , the distribution , of the energy change @xmath26 .
note that a positive temperature @xmath6 demands an asymmetry @xmath27 from the symmetric perturbation .
this reflects the fact that the entropy increases with the energy , which induces , on average , a convex potential - energy surface @xcite that favors a positive @xmath26 . as eq .
( [ eq : mechtrans ] ) requires no derivative of the potential energy function @xmath28 , we expect it to be applicable to a discontinuous @xmath28 .
nonetheless , the perturbation still needs to be rather small , @xmath29 .
below we derive more practical integral relations for an @xmath26 as large as @xmath30 .
consider a canonical ensemble .
if each configuration @xmath20 is shifted by @xmath13 to @xmath31 , then the potential - energy change @xmath32 satisfies @xcite @xmath33 where @xmath34 is the partition function .
thus , the equation @xmath35 has a nontrivial root at @xmath36 .
( [ eq : canon ] ) yields the same @xmath6 value as eq .
( [ eq : mechtrans ] ) in the limit of small perturbation .
first , since @xmath37 , where @xmath38 , we have @xmath39 which is reduced to eq .
( [ eq : mechtrans ] ) after averaging over perturbations : @xmath40 is @xmath41 , while @xmath26 is @xmath42 , so @xmath43 . as eq .
( [ eq : gaussian ] ) is also exact for a gaussian @xmath26 distribution , a perturbation involving many similar degrees of freedom can assume a larger @xmath26 .
we can replace the above continuous configuration @xmath20 by a discrete one @xcite , and/or the perturbation @xmath44 by an invertible transformation from @xmath20 to @xmath45 with a unit jacobian : @xmath46 @xcite .
some generalizations of eq .
( [ eq : canon ] ) are discussed in appendix [ apd : fluc ] , and verified on a harmonic oscillator ( appendix [ apd : ho ] ) .
the requirement for a unit jacobian should , however , be relaxed in the non - equilibrium case ( appendix [ apd : langevin ] ) .
we now adapt eq .
( [ eq : canon ] ) to a constant potential - energy ensemble , or a @xmath47 ensemble below , which collects configurations with the same potential energy @xmath47 .
the ensemble is commonly used to build a multicanonical ensemble of a flat potential - energy distribution @xcite .
configurations of the @xmath47 ensemble sum to the density of potential - energy states @xmath48\ , d{\mathbf{q}}$ ] , and the temperature of potential energy is defined as @xmath49 .
we will show @xmath50 for a nonnegative integer @xmath51 , where @xmath52 denotes an average over the surface @xmath53 .
particularly , @xmath54 . if the perturbation @xmath13 is symmetric ( i.e. , @xmath44 and @xmath14 are equally likely ) , then for each @xmath13 that carries @xmath20 to @xmath31 , the inverse @xmath14 that carries @xmath45 back to @xmath20 shares the same probability . without any _ a priori _ bias of the start - point configuration , the overall flux @xmath55 , or the total number of @xmath13 that reach the @xmath56 ensemble from the @xmath47 ensemble , therefore , equals the reverse flux @xmath57 , or @xmath58 . but
@xmath55 is the product of the density of states @xmath59 and the @xmath26 distribution @xmath60 for perturbations that start from the @xmath47 ensemble ; so @xmath61 if @xmath60 changes slowly with @xmath47 : @xmath62 , then @xmath63 and @xmath64 for any @xmath65 . by taking derivatives with respect to @xmath65 , and then setting @xmath66 , we get eq .
( [ eq : expu ] ) .
alternatively , we can integrate eq .
( [ eq : probrev ] ) with the metropolis acceptance probability as @xmath67 for a nonnegative @xmath51 .
the @xmath68 version shows that the energy change after a metropolis step from the @xmath47 ensemble averages to zero if the parameter @xmath6 in the acceptance probability @xmath69 roughly matches @xmath70 . besides , the function @xmath71 is roughly even at @xmath72 .
thus , for a nonnegative @xmath51 @xmath73 eq .
( [ eq : expu ] ) has an @xmath9 error , and can be corrected as @xmath74 where @xmath75 is the @xmath76 value determined from eq .
( [ eq : expu ] ) ( as an equality ) .
the correction reduces the error of @xmath76 to @xmath77 if @xmath26 is @xmath30 , such that the error of @xmath78 , integrated over an @xmath79 domain , is @xmath9 .
( [ eq : expuc ] ) is also exact for any @xmath10 in the limit of small @xmath26 ( appendix [ apd : series ] ) .
the derivatives with respect to @xmath47 are readily computed in a usual simulation that allows the potential energy to fluctuate . the corrections for eqs .
( [ eq : symu ] ) and ( [ eq : halfu ] ) can be found in appendix [ apd : series ] .
the temperature @xmath80 in the microcanonical ensemble can be obtained by substituting the total energy @xmath81 for the potential energy @xmath47 in the formulas for the @xmath47 ensemble ; e.g. , eq .
( [ eq : expu ] ) becomes @xmath82 .
the perturbations can still be limited to a subset of degrees of freedom , which can be made of only coordinates , but not momenta .
the limit @xmath83 , however , eliminates the @xmath9 corrections in , e.g. , eq .
( [ eq : expuc ] ) .
thus , eqs .
( [ eq : expu ] ) , ( [ eq : probrev])-([eq : halfu ] ) are exact if the interaction is weak compared to the kinetic energy , or if the system is embedded in a much larger isolated reservoir , which constitutes a canonical ensemble ( see also appendix [ apd : fluc ] ) .
we may therefore interpret the relations as virtual `` thermometers '' gauged in the canonical ensemble , e.g. , the average @xmath84 invariably reads 1.0 in the canonical ensemble of the correct @xmath6 for any uniform perturbation , but not so in other ensembles .
it is , however , possible to construct an exact thermometer gauged in the microcanonical ensemble .
the momenta - averaged weight for a configuration , whose potential energy is @xmath47 , is @xmath85 for a system of @xmath10 degrees of freedom @xcite .
thus , the total energy @xmath81 can be estimated from the root of @xmath86^{n/2 - 1 } { \rangle}\approx 1 $ ] ; then @xmath87 @xcite .
the extension to pressure @xmath88 is straightforward . for a virtual volume move
@xcite induced by the scaling of coordinates @xmath89 ( the dimension @xmath90 ) , we have @xcite @xmath91 { \rangle}= 1 , \label{eq : pressure}\ ] ] where @xmath92{v'/v}{\mathbf{q } } ) - u({\mathbf{q}})$ ] , and @xmath93 .
the @xmath94 is computed from either the jacobian of the coordinate scaling as @xmath95 , or from the kinetic energy change as @xmath96 \ , \beta \ , { \mathbf{p}}^2/(2m)$ ] of the conjugate momentum scaling @xmath97 @xcite . here , @xmath10 is the number of degrees of freedom , not the number of particles . for an infinitesimal volume change , we have @xmath98 , where @xmath99/(dv)$ ] or @xmath100 with @xmath101 . eq . ( [ eq : pressure ] ) holds for a fixed @xcite or variable @xmath102 . for a hard - core system , only negative @xmath102
should be used @xcite ( similar to widom s method @xcite for estimating the chemical potential , in which particles are inserted , but not removed ) .
if @xmath103 is changed on an exponential scale as @xmath104 @xcite , then an additional factor @xmath105 should be inserted into the brackets @xmath106 .
the advantage of using a variable volume change is that the virtual volume move is identical to a generalized ( or possibly optimized @xcite ) mc volume trial in an isothermal - isobaric @xcite or gibbs - ensemble @xcite mc simulation , and thus can be realized simultaneously .
the formulas are exact in the isothermal - isobaric ensemble , but approximate in general .
the formulas for temperature can be generalized to compute the logarithmic derivative , or the mean force , of a distribution @xmath107 of an extensive quantity @xmath108 @xcite : @xmath109 \ , w({\mathbf{q } } ) \ ,
d{\mathbf{q } } , \label{eq : distr1d}\ ] ] where @xmath110 is the ensemble weight for a configuration @xmath20 , e.g. , @xmath111 $ ] in the canonical ensemble .
we wish to find @xmath112 , whose integral yields the free energy @xcite @xmath113 .
we define an _ adjusted _ perturbation @xmath114 as a symmetric perturbation @xmath13 adjusted by the metropolis acceptance probability : @xmath115 thus , @xmath114 is equal to @xmath13 if the unadjusted perturbation @xmath13 is accepted , or * 0 * if rejected . like
an mc move @xcite , the @xmath116 satisfies detailed balance : @xmath117 , where @xmath118 .
summing over @xmath20 and @xmath45 under @xmath119 and @xmath120 leads to @xmath121 here , @xmath107 replaces @xmath59 in eq .
( [ eq : dbu ] ) .
the temperature @xmath122 is then mapped to the mean force @xmath123 , and @xmath124 { \rangle}_x \approx 1 , \label{eq : expx}\ ] ] where @xmath125 denotes the @xmath110 weighted configuration average at a fixed @xmath126 .
note that @xmath26 is the change of @xmath126 after the adjustment , which is 0 for a rejected perturbation .
similarly , the counterparts of eqs .
( [ eq : symu ] ) and ( [ eq : halfu ] ) [ and the @xmath127 corrections ( [ eq : expuc ] ) , ( [ eq : symuc ] ) and ( [ eq : halfuc ] ) ] can be obtained by @xmath128 and @xmath129 .
the above mc - type adjusted perturbation can be replaced by a reversible time evolution @xcite , as long as the latter also satisfies eq .
( [ eq : dbx ] ) .
we can therefore treat a short segment of such a trajectory as an adjusted perturbation , and use eq .
( [ eq : expx ] ) with @xmath130 ( @xmath131 is the segment length ) .
further , since eq .
( [ eq : expx ] ) does not require the potential energy , it can be readily applied to trajectories of colloidal particles monitored by microscopy experiments @xcite .
we now consider a multidimensional distribution @xmath132 \
, w({\mathbf{q } } ) \ , d{\mathbf{q } } \label{eq : distrmd}\ ] ] of @xmath133 extensive quantities @xmath134 , @xmath135 . to find all @xmath136 , we can extend , e.g. , eq .
( [ eq : expu ] ) with @xmath68 , as @xmath137 where @xmath138 is the change of @xmath139 by the adjusted perturbation , and @xmath140 denotes an average in the ensemble at a fixed set of @xmath141 .
( [ eq : expxmd ] ) offers @xmath133 equations , all components @xmath142 can be determined .
thus , @xmath143 to the first order , and the @xmath9 correction [ cf .
( [ eq : expuc ] ) ] is @xmath144 where @xmath145 is the inverse of the matrix @xmath146 .
the small - perturbation limit of eqs .
( [ eq : expxmd ] ) and ( [ eq : expxmdc ] ) is free from the @xmath9 error , and identical to the exact relation given by eqs .
( [ eq : mfmd ] ) and ( [ eq : mfmd2 ] ) . first , according to the definition of the adjusted perturbation eq .
( [ eq : adjpert ] ) , we have @xmath147 ( cf . sec . [ sec : mechtrans ] ) .
thus , @xmath148 ( the acceptance probability only affects half of the perturbations that point against the gradient of @xmath149 ) and @xmath150 to the leading order [ we have averaged over symmetric perturbations and used eq .
( [ eq : mfmd2 ] ) ] . for small @xmath151 ,
( [ eq : expxmd ] ) becomes @xmath152 , or @xmath153 .
( [ eq : expxmdc ] ) is then reduced to @xmath154 , which is eq .
( [ eq : mfmd ] ) [ the @xmath155 term is @xmath156 , hence negligible ] .
we show some results for the temperature formulas .
we use eqs .
( [ eq : expu ] ) , ( [ eq : symu ] ) or ( [ eq : halfu ] ) , etc . , in an mc / md simulation in this way : once every few steps along the trajectory , we perturb the current configuration @xmath20 by a random @xmath13 ( which is conducted as a virtual displacement so as not to disturb the real trajectory ) , and register the resulting change @xmath157 in the potential energy ; the formulas then estimate the temperature @xmath6 from the accumulated distribution @xmath158 . for the ising model
, @xmath20 is a spin configuration , and @xmath44 means to flip a random spin .
four @xmath26-distributions @xmath158 from simulations on the 108-particle lennard - jones ( lj ) fluid at @xmath159 and @xmath160 are shown in fig .
[ fig : basic](a ) .
the first simulation used a regular md ( in the microcanonical - like ensemble ) , while the others used the metropolis mc ( in the canonical ensemble ) .
the pair potential was switched smoothly from the standard form @xmath161 $ ] ( @xmath162 ) at @xmath163 to a 7th order polynomial @xcite that vanishes at @xmath164 to avoid artifacts . in the first three cases ,
each coordinate of a random particle was displaced by a random number in @xmath165 , with @xmath166 ( the first two cases ) or 0.1 ( the third ) . in the last case ,
the perturbation was applied to all particles with @xmath167 .
the number of perturbations was @xmath168 in each case .
( [ eq : symu ] ) and ( [ eq : halfu ] ) were adapted to the md or canonical ensemble in the sense of @xmath169 .
a comparison of the first two cases shows that the distribution was not very sensitive to the ensemble type in this case .
the distributions from the single - particle perturbations peaked at @xmath170 [ cf .
( [ eq : hope ] ) ] .
the distribution from the all - particle perturbation , however , was gaussian - like [ cf .
( [ eq : hogaussian ] ) ] .
the integral relations applied to all cases , e.g. , the @xmath6 given by eq .
( [ eq : canon ] ) were 1.0077 , 0.9990 , 1.0160 , and 0.9997 , respectively . the solution process of eq .
( [ eq : canon ] ) is illustrated in fig .
[ fig : basic](b ) : if we plot @xmath171 against @xmath6 , its intersection with 1.0 then gives the desired @xmath6 .
three systems , the lj fluid ( with both single- and all - particle perturbations ) , square - well @xcite fluid and @xmath172 ising model were simulated in the canonical ensemble at @xmath160 using the metropolis algorithm .
the square - well potential of two particles is infinity if their distance @xmath173 , or @xmath174 if @xmath175 , or 0 otherwise ( @xmath176 , @xmath177 , and @xmath178 ) .
perturbations that produced clashes , hence infinite @xmath26 , were excluded from entering eq .
( [ eq : canon ] ) . in fig .
[ fig : basic](c ) , the temperatures computed from eqs .
( [ eq : ctemp])-([eq : gaussian ] ) , ( [ eq : symu ] ) and ( [ eq : halfu ] ) , were plotted against @xmath179 , the size of the single - particle perturbation .
all formulas worked well with a small @xmath179 , but with a large @xmath179 , eq .
( [ eq : mechtrans ] ) and ( [ eq : gaussian ] ) quickly lost accuracy , while the integral formulas were little affected .
the effect of the potential truncation @xcite was studied using a set of mc simulations ( @xmath166 ) with the lj potential deliberately truncated at small @xmath180 . the discontinuity rendered eq .
( [ eq : ctemp ] ) approximate @xcite . as shown in fig .
[ fig : basic](d ) , a small @xmath180 indeed affected eq .
( [ eq : ctemp ] ) , but not eqs .
( [ eq : canon ] ) , ( [ eq : symu ] ) , and ( [ eq : halfu ] ) .
the temperature profile @xmath70 along the potential energy was computed on the @xmath172 ising model . to cover the entire energy range , we ran a multicanonical simulation @xcite using the exact density of states @xmath181 @xcite in the sampling weight for @xmath168 mc steps per site .
as shown in fig .
[ fig : isent](a ) , the temperature from eqs .
( [ eq : mechtrans ] ) and ( [ eq : gaussian ] ) behaved badly except around @xmath182 or @xmath183 , while the integral relations ( [ eq : expu ] ) , ( [ eq : symu ] ) and ( [ eq : halfu ] ) agreed with the reference @xmath184 .
this is expected , as the smallest energy change is 4 in the system , @xmath185 rarely holds , while @xmath26 is still @xmath186 to justify the use of the integral relations .
but even in the latter case , there was an @xmath186 difference between the integral @xmath187 and the exact @xmath188 [ fig .
[ fig : isent](b ) ] , showing that the temperature from eqs .
( [ eq : expu ] ) , ( [ eq : symu ] ) and ( [ eq : halfu ] ) had an @xmath9 systematic error without the corrections , as the entire potential - energy range was @xmath79 .
the corrections ( [ eq : expuc ] ) , ( [ eq : symuc ] ) and ( [ eq : halfuc ] ) , however , effectively removed the remaining errors ( except around the ground states , where the density of states was intrinsically irregular ) .
as an application , eq .
( [ eq : canon ] ) can be used to help a simplified potential - energy function emulate a more complex or realistic one @xcite . in the following example , the simplified function is the hard - sphere potential @xmath189 , and the more realistic one is the lj potential @xmath190 .
we first run a simulation using @xmath191 as the potential energy . from the trajectory , we evaluate the effective temperature @xmath192 , by seeking the solution of @xmath193 , with @xmath194 being the change caused by a virtual perturbation .
we repeat the process with a modified @xmath191 until the @xmath192 matches the simulation temperature @xmath6 : @xmath195 thus , eq .
( [ eq : tmatchlj ] ) calibrates the hard - sphere system by the virtual thermometer gauged in the lj system ( cf .
appendix [ apd : relentropy ] ) . similar to eq .
( [ eq : tmatchlj ] ) , one can also match the potential energy as @xmath196 or pressure as @xmath197 .
another alternative is to minimize the relative entropy @xcite @xmath198 w_{{\mathrm{hs}}}({\mathbf{q } } ) \
, d{\mathbf{q } } , \label{eq : relentlj}\ ] ] which measures the difference between the distributions @xmath199 and @xmath200 .
although eq .
( [ eq : tmatchlj ] ) and the conditions of matching the potential energy and pressure can all be derived from minimizing @xmath201 regarding virtual variations @xcite of @xmath202 ( appendix [ apd : relentropy ] ) , they do not necessarily find the exact optimal @xmath191 that minimizes @xmath201 .
the methods were tested on a fluid system of 108 particles .
the diameter @xmath203 of the hard - sphere system was varied to match the lj system under three conditions .
the ratio of the partition functions @xmath204 , required for computing @xmath201 , were obtained by bennett s acceptance ratio method @xcite from independent simulations on the two systems .
table [ tab : tmatch ] shows that the obtained @xmath203 from the above methods were generally close in the gaseous phase .
it was , however , not always possible to find solutions for all methods , e.g. , matching the potential energy failed in the second case , while the relative entropy was hard to compute in the third case .
thus , the methods can be complementary . like the force - matching method @xcite
, the temperature matching requires simulation in only one ( hard - sphere ) system .
one may also treat the configuration temperature as a special thermal force , and thus add a restraint to the force - matching method .
also note that eq .
( [ eq : tmatchlj ] ) allows a local perturbation with cheap local energy calculations .
* 7y @xmath205 & @xmath206 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 + + 0.900 & 15.2 & 0.24 & @xmath212 & @xmath213 & 1.620 & 0.728 + 0.952 & * 7.1 * & 0.62 & @xmath214 & @xmath213 & * 0.729 * & * 0.729 * + 0.970 & 8.4 & 0.86 & @xmath215 & @xmath215 & 0.532 & 0.728 + 0.978 & 9.4 & * 1.00 * & @xmath216 & @xmath213 & 0.459 & 0.727 + 1.000 & 13.6 & 1.52 & @xmath217 & @xmath213 & 0.289 & 0.728 + + 0.900 & 13.4 & 0.23 & @xmath218 & @xmath219 & 0.065 & 0.036 + 0.981 & @xmath220 & @xmath221 & @xmath222 & @xmath219 & 0.038 & 0.036 + 0.993 & 8.8 & 1.27 & @xmath222 & @xmath219 & * 0.036 * & * 0.036 * + 1.000 & 9.0 & 1.40 & @xmath222 & @xmath219 & 0.035 & 0.036 + + 0.900 & 180 & 0.28 & @xmath223 & @xmath224 & 9.88 & @xmath225 + 0.970 & 37 & * 1.00 * & @xmath226 & @xmath224 & 1.89 & @xmath227 + 0.999 & - & 1.74 & @xmath228 & @xmath229 & @xmath230 & @xmath231 + 1.000 & - & 1.77 & @xmath232 & @xmath232 & @xmath233 & @xmath225 + + + + + + [ tab : tmatch ]
to sum up , temperature can be extracted from the distribution @xmath158 of the potential - energy change @xmath26 caused by configuration perturbations , as the nontrivial root of eqs .
( [ eq : expu ] ) , ( [ eq : symu ] ) or ( [ eq : halfu ] ) .
the formulas can be understood as virtual thermometers gauged in a corresponding canonical ensemble . when used in the constant potential - energy ensemble ,
the formulas have an @xmath9 error , but can be corrected systematically .
the approach can be extended to the mean force of a multidimensional distribution @xmath234 by an ensemble - adjusted perturbation .
the adjusted perturbation is equivalent to a short trajectory of a reversible dynamics , making the mean force formulas ( [ eq : expx ] ) and ( [ eq : expxmd ] ) readily usable in simulations and experiments @xcite .
the computer code of the examples can be found in ref .
i thank dr .
m. w. deem , dr . b. m. pettitt , and the referees for helpful discussions and comments .
i am especially indebted to the second referee for various insightful suggestions on , among others , the non - equilibrium extensions , connections with coarse - graining methods , and verifications on experiments and exact models .
following refs .
@xcite , we show that the temperature @xmath235 in the constant potential - energy ensemble ( cf . sec .
[ sec : uensemble ] ) satisfies @xmath236 where @xmath237 is a vector field that satisfies @xmath238 @xcite .
the second term on the right is @xmath9 .
we first define @xcite @xmath239 \ , d{\mathbf{q } } = { \langle}{\mathbf{b } } \cdot \nabla u { \rangle}_u \ , g(u).\ ] ] then @xmath240 on the other hand , integration by parts yields @xmath241 d { \mathbf{q } } \notag \\ & = \int ( \nabla \cdot { \mathbf{b } } ) \delta[u({\mathbf{q } } ) - u ] d { \mathbf{q } } = { \langle}\nabla \cdot { \mathbf{b } } { \rangle}_u g(u ) .
\label{eq : dgdub}\end{aligned}\ ] ] equating eqs .
( [ eq : dgdua ] ) and ( [ eq : dgdub ] ) gives ( [ eq : ctempu ] ) .
( [ eq : ctemp ] ) represents two special cases . with @xmath242 , @xmath243 , and @xmath244 where the second term is @xmath9 . with @xmath245 , we get @xmath246 we can show eq .
( [ eq : ctemp ] ) from the equation of motion @xcite : @xmath247 where @xmath131 is the length of a long trajectory .
we have assumed a thermostat @xcite that makes momenta @xmath248 independent : @xmath249 , and uncorrelated with @xmath250 and @xmath251 .
this derivation has the advantage of replacing the strong assumption of global ergodicity in the ensemble theory by a weaker condition of sufficiently randomized momenta ( which requires only local equilibration ) .
. ( [ eq : canon ] ) can be generalized by several fluctuation theorems @xcite . for any @xmath65
, we have @xmath252 { \rangle}_\beta = { \langle}\exp(-\beta ' { \varepsilon}^ * ) { \rangle}_\beta,\ ] ] where @xmath26 and @xmath253 are the energy changes caused by a pair of uniform perturbations @xmath44 and @xmath14 , respectively @xcite , because @xmath254 } \ , e^{-\beta u({\mathbf{q}})}/z(\beta ) \ , d{\mathbf{q } }
\\ = \int e^ { -\beta ' \ , [ u({\mathbf{q}}'-{\mathbf{u } } ) - u({\mathbf{q } } ' ) ] } \ , e^{-\beta u({\mathbf{q}}')}/z(\beta ) \ , d{\mathbf{q}}',\end{aligned}\ ] ] where @xmath255 . for a symmetrically randomized perturbation ,
the distributions of @xmath26 and @xmath253 are identical , and @xmath252 { \rangle}_\beta = { \langle}\exp(-\beta ' { \varepsilon } ) { \rangle}_\beta .
\label{eq : canonbp}\ ] ] the @xmath26 distribution can be found from the exponential average @xmath256 @xcite as @xmath257 . by taking the fourier transform of eq .
( [ eq : canonbp ] ) , we get @xmath258 assuming that @xmath259 has no singularity in the strip @xmath260 of the complex plane .
note that while eqs .
( [ eq : canonbp ] ) and ( [ eq : canondb ] ) require a symmetric perturbation , eq .
( [ eq : canon ] ) does not .
( [ eq : canon ] ) has a few other generalizations . for a hamiltonian @xmath261 parameterized by @xmath262 , we can treat a circular switch of @xmath263 starting from the canonical equilibrium state at @xmath264 as an elaborate perturbation
. then the jarzynski equality states @xmath265 for the non - equilibrium work
@xmath266 over a period @xmath131 @xcite .
similarly , we have @xmath267 for the work @xmath268 derived from a time - dependent driving force @xmath269 ( excluding the component from the conservative potential ) @xcite .
these relations can also be used to extract the equilibrium temperature @xmath6 of the initial equilibrium state .
we verify a few formulas for the canonical ensemble on a @xmath270-dimensional harmonic oscillator with the potential energy @xmath271 . for a gaussian perturbation @xmath272 applied to a fixed @xmath20 , the average @xmath273 } p_{\mathbf{u } } ( { \mathbf{u } } ) \ , d{\mathbf{u } } \\ & = ( 1+\alpha b)^{-d/2 } \exp[\alpha b^2 u /(1+\alpha
b)],\end{aligned}\ ] ] where @xmath274 . averaging @xmath275 over @xmath20 yields @xmath276^{-d/2}. \label{eq : hogenf}\end{aligned}\ ] ] eq .
( [ eq : hogenf ] ) satisfies eqs .
( [ eq : canon ] ) and ( [ eq : canonbp ] ) , i.e. , @xmath277 .
since @xmath278 gives the moment generating function @xcite , the @xmath26 distribution @xmath279 is @xmath280^{(d-1)/2}}{(\alpha/\beta)^{d/2 } } \dfrac { k_{(d-1)/2}(\gamma |{\varepsilon}| ) } { \sqrt\pi \gamma(d/2 ) } e^{\beta { \varepsilon}/2 } , \label{eq : hope}\ ] ] where @xmath281 , @xmath282 is the modified bessel function of the second kind , and @xmath283 is the gamma function .
( [ eq : hope ] ) satisfies eq .
( [ eq : canondb ] ) , and shows that @xmath158 at @xmath170 depends critically on the dimension @xmath270 : it diverges if @xmath284 ; it has a finite cusp if @xmath285 ; and it is differentiable if @xmath286 . in the limit of @xmath287 and @xmath288 ,
the distribution is gaussian : @xmath289 , \label{eq : hogaussian}\ ] ] where @xmath290 .
we derive the corrections for eqs .
( [ eq : expu ] ) , ( [ eq : symu ] ) and ( [ eq : halfu ] ) from @xmath291 \sum_{m = 0}^{+\infty } \dfrac{(-{\varepsilon})^m}{m ! } \dfrac{d^m p_u({\varepsilon } ) } { du^m } , \label{eq : series}\end{aligned}\ ] ] where @xmath292 .
( [ eq : expu ] ) seeks the root @xmath293 of @xmath294 \ , { \varepsilon}^k \,d { \varepsilon}.\ ] ] using eq .
( [ eq : series ] ) for @xmath295 yields : @xmath296 { \varepsilon}^{k+m } \right\rangle_u , \label{eq : expuav}\ ] ] where @xmath297 denotes a differentiation that applies only to the moments of @xmath26 , but not to the @xmath75 or @xmath298 .
( [ eq : expuav ] ) can be solved by successive approximations : @xmath299 with @xmath300 . to the first order , we use @xmath301 in eq .
( [ eq : expuav ] ) , and @xmath302 which yields @xmath303 .
next , we set @xmath304 , and @xmath305 which yields @xmath306/ { { \langle}{\varepsilon}^{k+1 } { \rangle}_u}$ ]
. we reach eq .
( [ eq : expuc ] ) by @xmath307 .
( [ eq : expuc ] ) is exact for any @xmath10 in the limit of small @xmath26 . for an even @xmath51 in this limit , it becomes ( cf .
[ sec : mechtrans ] ) @xmath308 with @xmath309 . to show this
, we use @xmath310 partial integration yields @xmath311 = 2 { \langle}{\varepsilon}^{k+1 } { \rangle}_u \ , g(u)$ ] , which is eq .
( [ eq : expucsmalle ] ) . for an odd @xmath51 ,
the @xmath76 value is identical to that from the @xmath312 case in the small @xmath26 limit .
the correction for eq .
( [ eq : symu ] ) can be constructed from @xmath313\ , d{\varepsilon}.\ ] ] using eq .
( [ eq : series ] ) for @xmath60 yields @xmath314 where latexmath:[$m_l \equiv \int_{-\infty}^{+\infty } to correct the @xmath75 value determined from eq .
( [ eq : halfu ] ) , we use @xmath316\ , d{\varepsilon},\ ] ] which yields @xmath317 eq .
( [ eq : halfuc ] ) lacks the @xmath318 term in eqs .
( [ eq : expuc ] ) and ( [ eq : symuc ] ) , as it has been included in differentiating the @xmath293 in the @xmath319 .
for the @xmath133-dimensional distribution @xcite @xmath320 defined by eq .
( [ eq : distrmd ] ) , we have @xmath321 where @xmath322 is the @xmath133 by @xmath133 matrix with @xmath323 ( the @xmath324 are vector fields such that the matrix has an inverse @xmath325 ) , and @xmath326 particularly , if @xmath327 , @xmath328 which is similar to eq .
( [ eq : ctempu ] ) . to show eq .
( [ eq : mfmd ] ) , we observe @xmath329 w({\mathbf{q}})\,d{\mathbf{q}}.\ ] ] then partial integration gives @xmath330 \
, \textstyle\prod_{\gamma } \delta[x_\gamma({\mathbf{q } } ) - x_\gamma ] \ , d { \mathbf{q } } \\ & = \textstyle\sum_\gamma \partial{(w_{\alpha\gamma}\,\rho)}/{\partial x_\gamma } = \textstyle\sum_\gamma \left ( w_{\alpha \gamma } f_\gamma + \partial{w_{\alpha\gamma}}/ { \partial x_\gamma } \right)\ , \rho.\end{aligned}\ ] ] multiplying @xmath331 to both sides yields eq .
( [ eq : mfmd ] ) . if @xmath324 is constructed by the gram - schmidt orthonormalization such that @xmath332 @xcite , then @xmath333 the resulting @xmath334 can be quite complex @xcite , e.g. , for @xmath335 , @xmath336 we can alternatively set @xmath337 , and @xmath338 which limits the second derivatives to the laplacians , but requires the corrections @xmath339 from numerical differentiation .
we show below that the temperature matching condition eq .
( [ eq : tmatchlj ] ) locally minimizes the relative entropy @xcite @xmath340 \ , w_b({\mathbf{q } } ) \ , d{\mathbf{q}},\ ] ] of two distributions @xmath341 and @xmath342 ( @xmath343 ) along the force .
if @xmath344 , the minimum is located at @xmath345 [ use @xmath346 with @xmath347 @xcite ] . if , however , @xmath348 is minimized along an infinitesimal coordinate transformation @xmath349 , then @xmath350 , or @xmath351 . in the canonical ensemble ,
@xmath352 for @xmath353 $ ] . particularly , with @xmath354 , we have @xmath355 , which is eq .
( [ eq : tmatchlj ] ) in the small - perturbation limit .
generally , eq . ( [ eq : beteff ] ) defines a distinct effective temperature @xmath356 for each vector field @xmath250 , reflecting the fact that one can define different effective temperatures @xcite in a non - equilibrium state ( @xmath357 : equilibrium , @xmath358 : non - equilibrium ) .
the conditions for matching the potential energy and pressure can be similarly obtained by varying @xmath357 with respect to the temperature and volume , respectively .
normally , we set @xmath357 as the model distribution , and optimize its parameters to match the reference @xmath358 @xcite ; thus , the averages should be performed in the reference @xmath359 system .
this is , however , inconvenient , for the example in sec .
[ sec : tmatch ] , since @xmath360 , when averaged over configurations produced by the reference lj potential , can be infinite , due to the hard - sphere potential @xmath361 .
we therefore set @xmath362 and @xmath363 .
we use a driven system in constant contact with a heat bath as a model to study the temperature in a non - equilibrium steady state .
we will show below that , to correctly extract the presumably constant heat bath temperature , eq .
( [ eq : canon ] ) should be applied to a nonuniform perturbation .
consider the overdamped langevin equation @xcite : @xmath364 , where the total force @xmath365 is the conservative component @xmath366 plus a driving @xmath367 , @xmath368 is the viscosity , and @xmath369 is a gaussian white noise satisfying @xmath370 .
the steady - state distribution @xmath149 satisfies @xmath371 where @xmath372 is the constant current , and the ratio @xmath373 is the average local velocity @xcite .
the equilibrium case @xmath374 is recovered if @xmath375 .
consider a perturbation @xmath376 generated by a flow from the vector field @xmath377 , @xmath378 $ ] , over a period of `` virtual time '' @xmath379 , such that @xmath380 and @xmath381 . we then have , by eq .
( [ eq : fp ] ) , @xmath382 where @xmath383 \cdot { \mathbf{u}}[{\mathbf{q}}(\tau ) ] \ , d\tau$ ] is the change of the local effective potential @xcite or the absorbed heat , @xmath384 \cdot { \mathbf{v}}[{\mathbf{q}}(\tau ) ] \ , d\tau$ ] is the housekeeping heat @xcite , and @xmath385 \ , d\tau \right ) = eq .
( [ eq : expnej ] ) can be made to locally resemble the equilibrium version [ eq . ( [ eq : canon ] ) ] as @xmath386 by a constraint @xmath387 .
the constraint can be satisfied by the vector field @xmath388 , where @xmath389 is a constant and @xmath390 satisfies @xmath391 .
the @xmath392 induces a nonuniform perturbation in the steady state with nonzero current . in this way
, we get an energy - based reading of the temperature in the non - equilibrium steady state , just as in the equilibrium state , such that the same @xmath6 value can be obtained for perturbations of different sizes and directions .
similarly , if the perturbation is a short trajectory that follows the langevin equation , or one that satisfies @xmath393 , then @xmath394 @xcite yields the same @xmath6 .
b. a. berg and t. neuhaus , phys .
lett . * 68 * , 9 ( 1992 ) ; j. lee , phys . rev . lett .
* 71 * , 211 ( 1993 ) ; f. g. wang and d. p. landau , phys .
lett . * 86 * , 2050 ( 2001 ) ; j. kim , j. e. straub , and t. keyes , phys .
lett . * 97 * , 050601 ( 2006 ) .
g. m. wang , e. m. sevick , e. mittag , d. j. searles , and d. j. evans , phys .
lett . * 89 * , 050601 ( 2002 ) ; y. han and d. g. grier , phys .
lett . * 92 * , 148301 ( 2004 ) ; v. blickle , t. speck , l. helden , u. seifert , and c. bechinger , phys .
* 96 * , 070603 ( 2006 ) ; d. andrieux , p. gaspard , s. ciliberto , n. garnier , s. joubaud , and a. petrosyan , phys .
* 98 * , 150601 ( 2007 ) .
a. p. lyubartsev and a. laaksonen , phys .
e * 52 * , 3730 ( 1995 ) ; w. g. noid , j .- w .
chu , g. s. ayton , v. krishna , s. izvekov , g. a. voth , a. das , and h. c. andersen , j. chem .
* 128 * , 244114 ( 2008 ) .
f. ercolessi and j. b. adams , europhys . lett .
* 26 * , 583 ( 1994 ) ; s. izvekov , m. parrinello , c. j. burnham , and g. a. voth , j. chem . phys .
* 120 * , 10896 ( 2004 ) ; s. izvekov and g. a. voth , j. phys .
b * 109 * , 2469 ( 2005 ) .
r. p. feynman , _ statistical mechanics : a set of lectures _ ( w. a. benjamin , reading , ma , 1972 ) ; h. b. callen , _ thermodynamics and an introduction to thermostatistics _ ( wiley , new york , 1985 ) ; p. m. chaikin and t. c. lubensky , _ principles of condensed matter physics _ ( cambridge university press , cambridge , uk , 1995 ) .
g. j. martyna , m. l. klein , and m. tuckerman , j. chem . phys . *
97 * , 2635 ( 1992 ) ; g. bussi , d. donadio , and m. parrinello , j. chem .
126 * , 014101 ( 2007 ) ; d. j. evans , w. g. hoover , b. h. failor , b. moran , and a. j. c. ladd , phys .
a * 28 * , 1016 ( 1983 ) ; h. mori , prog .
phys . * 34 * , 399 ( 1965 ) ; r. zwanzig , j. stat . phys . * 9 * , 215 ( 1973 ) . | the notion of configuration temperature is extended to discontinuous systems by identifying the temperature as the nontrivial root of several integral equations regarding the distribution of the energy change upon configuration perturbations .
the relations are generalized to pressure and a distribution mean force . |
the cohomology of a flag manifold @xmath7 has an integral basis of schubert classes @xmath8 indexed by elements @xmath9 of the weyl group of @xmath10 .
the algebraic structure of these rings is known @xcite with respect to a monomial basis , and there are methods ( schubert polynomials ) for expressing the @xmath8 in terms of this basis @xcite .
moreover , their multiplicative structure with respect to the schubert basis is determined by chevalley s formula @xcite . despite this
, it remains an open problem to give a closed or bijective formula for the integral structure constants @xmath11 defined by the identity @xmath12 these @xmath11 are non - negative as they count the flags in a suitable triple intersection of schubert varieties .
they are expected to be related to the enumeration of chains in the bruhat order of the weyl group ( see @xcite and the references therein ) . of particular interest are pieri - type formulas which describe the constants @xmath11 when @xmath13 is a special schubert class pulled back from a grassmannian projection ( @xmath14 , @xmath2 maximal parabolic ) , as these determine the ring structure for the cohomology of @xmath14 when @xmath2 is any parabolic subgroup .
when @xmath10 is @xmath15 , a pieri - type formula for multiplication by a special schubert class was described @xcite in terms of the weyl group element @xmath16 .
a formula in terms of chains in the bruhat order was conjectured @xcite and given a geometric proof @xcite .
our main results are the analogous formulas when @xmath10 is @xmath17 or @xmath18 and @xmath13 is a special schubert class pulled back from a grassmannian of maximal isotropic subspaces .
these are common generalizations of the pieri - type formulas for @xmath15 , chevalley s formula , and pieri - type formulas for grassmannians of maximal isotropic subspaces @xcite .
our proof uses results on the bruhat order , identities of these structure constants , a decomposition of intersections of schubert varieties , and formulas in the cohomology of the @xmath15-flag manifold to explicitly determine a triple intersection of schubert varieties .
this shows the coefficients in the pieri - type formula are the intersection number of a linear space with a collection of quadrics .
some intermediate results , including a fundamental identity and some additional ` hidden symmetries ' of the structure constants , are deduced from constructions on @xmath15-flag manifolds @xcite .
this analysis leads to other results , including a new partial order on the infinite coxeter group @xmath6 and a monoid for chains in this order as in @xcite .
we show how the pieri - type formula implies our fundamental identity , use the identities to express many structure constants in terms of the littlewood - richardson coefficients for the multiplication of schur @xmath2 ( or @xmath4 ) functions @xcite , and apply the hidden symmetries to the enumeration of chains in the bruhat order .
schubert classes in the cohomology of the flag manifolds @xmath19 and @xmath20 form integral bases indexed by elements of the weyl group @xmath21 .
we represent @xmath21 as the group of permutations @xmath9 of @xmath22 satisfying @xmath23 for @xmath24 .
let @xmath25 denote the schubert class indexed by @xmath26 in @xmath27 and @xmath28 that in @xmath29 .
the degree of these classes is @xmath30 , where the _ length _ @xmath31 of @xmath9 is @xmath32 for an integer @xmath33 , let @xmath34 denote @xmath35 .
for each @xmath36 , define @xmath37 by @xmath38 this indexes a ( maximal isotropic ) special schubert class in either cohomology ring , written as @xmath39 and @xmath40 .
we state the pieri - type formula for the products @xmath41 and @xmath42 in terms of chains in the bruhat order on @xmath21 . for this
, we need a definition .
[ def:0-bruhat ] the _ @xmath43-bruhat order
_ @xmath44 on @xmath21 is defined recursively as follows : @xmath45 is a cover in the @xmath43-bruhat order if and only if 1 .
@xmath46 , and 2 .
@xmath47 is a reflection of the form @xmath48 or @xmath49 for some @xmath50 .
chevalley s formula @xcite may be stated as follows : @xmath51 where @xmath52 is the number of transpositions in the reflection @xmath47 .
we enrich the 0-bruhat order in two complementary ways .
write the two types of covers in the 0-bruhat order as @xmath53 and @xmath54 where @xmath55 .
the _ labeled @xmath43-bruhat rseau _ is a labeled directed multigraph with vertex set @xmath21 and labeled edges between covers in the 0-bruhat order given by the following rule : if @xmath53 , then a single edge is drawn with label @xmath56 .
if @xmath54 , then two edges are drawn with respective labels @xmath57 and @xmath56 . thus if @xmath58 , then @xmath52 counts the edges from @xmath59 to @xmath9 in this 0-bruhat rseau
. the _ labeled @xmath43-bruhat order _ is obtained from this rseau by removing edges with negative integer labels . given a ( saturated ) chain @xmath60 in either of these structures ,
let @xmath61 denote the endpoint of @xmath60 .
peak _ in a chain @xmath60 is an index @xmath62 with @xmath63 , where @xmath64 is the sequence of edge labels in @xmath60 .
a _ descent _ is an index @xmath65 with @xmath66 and an _ ascent _ is an index @xmath65 with @xmath67 .
* theorem a. * ( pieri - type formula ) _ let @xmath68 and @xmath69 . then _ 1 .
( odd - orthogonal pieri - type formula ) @xmath70 the sum over all chains @xmath60 in the labeled @xmath43-bruhat order of @xmath21 which begin at @xmath59 , have length @xmath71 , and no peaks .
( symplectic pieri - type formula ) @xmath72 1 .
the sum over all chains @xmath60 in the labeled @xmath43-bruhat rseau of @xmath21 which begin at @xmath59 , have length @xmath71 , and no descents .
2 . the same sum , except with no ascents . this generalizes chevalley s formula and the pieri - type formulas for @xmath73 , which are expressed in @xcite as a sum of certain labeled chains in the bruhat order on the symmetric group @xmath74 with no ascents / no descents .
the duality of these two formulas , one in terms of peaks for an order , and the other in terms of descents / ascents for an enriched structure on that order has connections with other dualities in combinatorics .
these include fomin s duality of graded graphs @xcite and stembridge s theory of enriched @xmath2-partitions @xcite , where peak and descent sets play a complementary role .
these relations are explored in @xcite , which extends the theory developed in @xcite to the ordered structures of this manuscript .
[ ex : pieri ] represent permutations @xmath75 by their values @xmath76 .
consider the products @xmath77 and @xmath78 .
figure [ fig : height2 ] shows the part of the 0-bruhat rseau of @xmath79 height 2 above @xmath80 in @xmath81 .
( erase edges with negative labels to obtain its analog in the 0-bruhat order . )
chains of length 2 are peakless , so by theorem a i , we have @xmath82 every chain in figure [ fig : height2 ] with increasing labels may be paired with a chain with decreasing labels having the _ same _ underlying permutations , and this pairing exhausts all chains .
thus , by theorem a ii , we have @xmath83 if @xmath84 is a _ strict partition _ ( decreasing integral sequence @xmath85 )
, then @xmath84 determines a unique _ grassmannian permutation _
@xmath86 where @xmath87 for @xmath88 and @xmath89 .
if @xmath90 and @xmath91 , then @xmath92 . the schubert classes @xmath93 and @xmath94 are pullbacks of schubert classes from the grassmannians of maximal isotropic subspaces @xmath95 and @xmath96 , where @xmath97 is the maximal parabolic associated to the simple root of exceptional length .
formulas for products of these @xmath2- and @xmath4-classes are known @xcite as these classes are specializations of schur @xmath2- and @xmath4-functions @xcite .
our proof of theorem a uses identities among the structure constants @xmath98 and @xmath99 defined by the following formulas .
@xmath100 if @xmath101 , then these constants do not depend upon @xmath102 . iterating chevalley s formula ( [ eq : chevalley ] ) shows that if either of @xmath98 or @xmath99 is non - zero , then @xmath103 and @xmath104 equals @xmath105 , the sum of the parts of @xmath84 .
in fact the constant @xmath98 determines and is determined by the constant @xmath99 : let @xmath106 count the sign changes ( @xmath107 ) in @xmath9
. then the map @xmath108 embeds @xmath29 into @xmath27 and induces an isomorphism of their rational cohomology rings .
thus it suffices to work in @xmath29 .
this is fortunate , as a key geometric result , theorem [ thm : geom_facts](2 ) , holds only for @xmath20 .
let @xmath109 count the saturated chains in the interval @xmath110_0 $ ] and @xmath111 count the saturated chains in the rseau @xmath110_0 $ ] .
iterating chevalley s formula ( [ eq : chevalley ] ) with @xmath112 , the identity permutation , we obtain the following expressions . @xmath113
multiplying the first expression by @xmath114 and collecting the coefficients of @xmath25 in the resulting expansion ( likewise for the second expression ) gives the following proposition .
[ prop : chains ] let @xmath115 . then @xmath116 theorem a and proposition [ prop : chains ] show a close connection between chains in the 0-bruhat order / rseau and the structure constants @xmath98 and @xmath99 .
this justifies an elucidation of the basic properties of the @xmath43-bruhat order and rseau , which we do in sections 2 and 6 .
these structures have a remarkable property and there are related fundamental identities among the structure constants .
* theorem b. * _ suppose @xmath103 and @xmath117 in @xmath21 with @xmath118 . then _ 1 .
the map @xmath119 induces an isomorphism of labeled intervals in the 0-bruhat order and 0-bruhat rseau @xmath110_0\stackrel{\sim}{\longrightarrow}[x , z]_0 $ ] .
2 . for any strict partition @xmath84 , @xmath120 we prove theorem b(1 ) in section 2.1 using combinatorial methods .
theorem b(2 ) is a consequence of a geometric result ( theorem [ thm : geometric_shapes ] ) proven in section 4 .
both parts of theorem b are key to our proof of the pieri - type formula .
interestingly , the pieri - type formula and theorem b(1 ) together imply theorem b(2 ) : for any composition @xmath121 with each @xmath122 , let @xmath123 , @xmath124 , and @xmath125 .
the _ peak set _ of a ( maximal ) chain in a labeled order is the set of indices of peaks in the chain . given a chain in a labeled rseau , its _ descent set _
( respectively _ ascent set _ ) is the set of indices of descents ( respectively ascents ) in the chain .
[ cor : peak - set ] let @xmath126 . 1 .
let @xmath127 be any composition .
then the coefficient of @xmath25 in the product @xmath128 is the number of chains in the interval @xmath110_0 $ ] in the labeled @xmath43-bruhat order with peak set contained in @xmath129 .
2 . let @xmath127 be any composition .
then the coefficient of @xmath28 in the product @xmath130 is the number of chains in the interval @xmath110_0 $ ] in the labeled @xmath43-bruhat rseau with descent set contained in @xmath129 .
this is also the number with ascent set contained in @xmath129 .
3 . suppose the pieri - type formula ( theorem a ) holds .
then the intervals @xmath110_0 $ ] and @xmath131_0 $ ] have the some number of chains with peak set @xmath129 for every composition @xmath127 if and only if for every strict partition @xmath84 , @xmath132 . the same statement holds for ascent / descent sets for chain in the rseaux and the coefficients @xmath133 .
in particular , theorem b(1 ) implies theorem b(2 ) .
moreover , the numbers in 1 and 2 depend only upon the multiset @xmath134 .
parts 1 and 2 follow from theorem a. for 3 , note that the schur @xmath2-polynomials ( respectively @xmath4-polynomials ) are linear combinations of the @xmath135 ( respectively the @xmath136 ) ( * ? ? ?
* iii.8.6 ) .
this linear combination gives a formula for @xmath98 ( respectively @xmath99 ) in terms of chains with given peak sets ( respectively , given ascent / descent sets ) .
let @xmath137 . by theorem b(1 )
, we may define @xmath138 if there is a @xmath139 with @xmath140 and @xmath141 whenever @xmath142 .
then @xmath143 is a graded partial order with rank function @xmath144 . by the identity of theorem b(2 ) , we may define @xmath145 and @xmath146 for any @xmath139 with @xmath147 and @xmath148 .
these coefficients satisfy one obvious identity , @xmath149 , as @xmath150 where @xmath151 is the longest element .
they also satisfy two others , which we call hidden symmetries .
let @xmath152 be the permutation defined by @xmath153 for @xmath154 .
then @xmath155 is the element with largest rank in @xmath143 .
let @xmath156 be defined by @xmath157 , so that @xmath158 . *
theorem c. * _ for any @xmath137 , _ 1 .
@xmath159 and for any strict partition @xmath84 , we have @xmath160 and @xmath161 .
2 . if @xmath162 for all @xmath163 , then @xmath164 and for any strict partition @xmath84 , we have @xmath165 and @xmath166 .
we prove a strengthening of theorem b , relaxing the condition of equality of @xmath16 and @xmath167 to that of _ shape equivalence_. permutations @xmath168 are shape equivalent if there exist sets @xmath169 and @xmath170 such that @xmath171 acts as the identity on @xmath172 , @xmath173 acts as the identity on @xmath174 , and @xmath175 if and only if @xmath176 .
theorems b ( the stronger version ) and c allow us to determine many of the constants @xmath98 and @xmath99 , showing they equal certain littlewood - richardson coefficients @xmath177 and @xmath178for schur @xmath2- and @xmath4-functions .
these are defined by the identities @xmath179 a combinatorial formula for these coefficients was given by stembridge @xcite .
let @xmath180 be strict partitions with @xmath181 .
we say that a permutation @xmath137 has _ skew shape _
@xmath182 if 1 .
either @xmath173 or @xmath183 is shape equivalent to @xmath184 , or 2 .
if @xmath162 for all @xmath163 , and one of @xmath185 is shape equivalent to @xmath184 .
if @xmath186 are permutations in @xmath21 and @xmath16 has a skew shape @xmath182 , then for any strict partition @xmath84 we have @xmath187 we call the partial order @xmath188 the _ lagrangian order _ and transfer the labeling from the @xmath43-bruhat order to obtain the labeled lagrangian order . in the same fashion , we may transfer the labeling and multiple edges of the 0-bruhat rseau to @xmath143 , obtaining the ( labeled ) _
lagrangian rseau_. by corollary [ cor : peak - set](3 ) , theorem c has a purely enumerative corollary . [ cor : hidden - peaks ] for any @xmath137 , 1 .
for any subset @xmath189 of @xmath190 , the intervals @xmath191_\prec$ ] and @xmath192_\prec$ ] in the lagrangian order have the same number of chains with peak set @xmath189 .
2 . for any subset @xmath189 of @xmath193 , the intervals @xmath191_\prec$ ] and @xmath192_\prec$ ] in the lagrangian rseau order
have the same number of chains with descent set @xmath189 and the same number of chains with ascent set @xmath189 , and these two numbers are equal .
if @xmath162 for all @xmath163 , then the same is true for @xmath191_\prec$ ] and @xmath194_\prec$ ] . in general , @xmath191_\prec\not\simeq [ e,\rho\zeta\rho]_\prec$ ] ,
( see figure [ fig : cyclic ] in example [ ex : cyclic ] ) and if @xmath162 for all @xmath163 , then in general , @xmath191_\prec\not\simeq [ e,\gamma\zeta\gamma^{-1}]_\prec$ ] .
( see figure [ fig : hidden ] in example [ ex : no - iso ] . )
let @xmath195 , the _ support _ of @xmath173 .
a permutation @xmath196 is _ reducible _ if it has a non - trivial factorization @xmath197 with @xmath198 where @xmath171 and @xmath199 have disjoint supports ( @xmath200 ) .
if @xmath200 with @xmath201 , then @xmath202 is a _
disjoint product_. permutations @xmath203 have unique factorizations into irreducibles .
we represented @xmath204}$ ] , the group of permutations of @xmath205 .
we also have @xmath206 , and the image consists of those @xmath173 with @xmath207 for every @xmath163 : for @xmath208 , let @xmath209}$ ] be the permutation such that @xmath210
. then @xmath211 . for @xmath137 , define @xmath212 if @xmath173 is in the image of @xmath74 , and @xmath213 otherwise . in section [ sec : minimal ]
we establish the following result .
let @xmath196 and suppose @xmath173 is irreducible . then @xmath214 .
if @xmath215 , then either 1 .
we have @xmath212 and there exists a cycle @xmath208 with @xmath216 , or 2 .
we have @xmath213 and @xmath173 is a single cycle in @xmath217}$ ] .
if every irreducible factor @xmath171 of @xmath173 satisfies @xmath218 , then we say that @xmath173 is _
minimal_. if @xmath137 is minimal , then set @xmath219 if @xmath173 is not minimal , then set @xmath220 .
we state the pieri - type formula in terms of the permutation @xmath16 .
* theorem d. * _ let @xmath221 and @xmath222
. then @xmath223 the sum over all @xmath203 with @xmath186 and @xmath224 .
_ this is similar to the form of the pieri - type formula for @xmath73 in @xcite , which is in terms of the cycle structure of the permutation @xmath16 .
this also generalizes the form of the the pieri formula for grassmannians of maximal isotropic subspaces @xcite .
our proof for @xmath225 shows these multiplicities to arise from the intersection of a linear subspace of @xmath226 with a collection of quadrics , one for each irreducible factor @xmath171 of @xmath16 with @xmath227 , similar to the proof of the pieri - type formula for maximal isotropic grassmannians in @xcite .
we relate the two formulations ( theorems a and d ) of the pieri - type formula in sections [ sec : lagr - or ] and [ sec : lagr - re ] .
we return to example [ ex : pieri ] . for @xmath228 and @xmath9
equal to each of @xmath229 , @xmath230 , @xmath231 , and @xmath232 in turn , @xmath16 is the permutation in @xmath233}$ ] : @xmath234 as permutations in @xmath81 , the first has 2 irreducible factors , for these we have @xmath235 and @xmath236 .
the other permutations are irreducible with @xmath237 and @xmath238 .
thus the values of @xmath239 are @xmath240 and of @xmath241 are @xmath242 , which shows the two forms of the pieri - type formula agree on this example .
we remark that the last two permutations , @xmath243 and @xmath244 , are shape equivalent . while we use the cohomology rings of complex varieties , our results and methods are valid for the chow rings @xcite and @xmath245-adic ( tale ) cohomology @xcite of these same varieties over any field not of characteristic 2 .
this paper is organized as follows : section 2 contains basic combinatorial definitions and properties of the bruhat order on @xmath6 analogous to those of the symmetric group established in @xcite .
section 3 contains the basic geometric definitions . in section 4
, we use geometry to establish the main identity , theorem b(2 ) . in section 5 , we establish additional geometric results and prove theorem c. in section 6 , we establish further combinatorial properties of the lagrangian order and rseau needed for the proof of the pieri - type formula , which is given in section 7 .
we derive the basic properties of the 0-bruhat order on @xmath6 analogous to properties of the @xmath246-bruhat order on @xmath247 .
further properties are developed in section 6 .
let @xmath248 be the cardinality of a finite set @xmath189 .
for an integer @xmath249 , its absolute value is @xmath250 and let @xmath251 . likewise , for a set @xmath2 of integers , define @xmath252 and @xmath253 .
set @xmath254:=\{1,\ldots , n\}$ ] and let @xmath217}$ ] be the group of permutations of @xmath255 $ ] .
let @xmath256 be the identity permutation in @xmath217}$ ] and @xmath257 the longest element in @xmath217}$ ] : @xmath258 .
then @xmath21 is the subgroup of @xmath217}$ ] for which @xmath259 and @xmath260 .
we also have @xmath261}$ ] , the symmetric group on @xmath262:=\pm[n]\cup\{0\}$ ] .
we refer to elements of these groups as permutations .
permutations @xmath26 are often represented by their values @xmath263 .
for example , @xmath264 .
the _ length _ @xmath31 of @xmath203 is @xmath265 thus @xmath266 .
note that @xmath257 is the longest element in @xmath21 .
an important class of permutations are the _ grassmannian permutations _ , those @xmath267 for which @xmath268 .
such a permutation is determined by its initial negative values . if @xmath269 , define @xmath270 to be the decreasing sequence @xmath271 .
note that @xmath272 .
likewise , given a decreasing sequence @xmath273 of positive integers ( a _ strict partition _ ) with @xmath274 , let @xmath275 be the grassmannian permutation with @xmath276 .
we write @xmath277 for strict partitions @xmath278 if @xmath279 for all @xmath33 , equivalently , if @xmath280 . the inclusion @xmath255\hookrightarrow\pm[n{+}1]$ ] induces inclusions @xmath281 and @xmath217}\hookrightarrow{\mathcal s}_{\pm[n+1]}$ ] .
define @xmath282 and @xmath283}$ ] . in this representation
, @xmath6 has three types of reflections , which are , as elements of @xmath284 : @xmath285 these reflections act on positions on the right and on values on the left .
the _ bruhat order _ on @xmath6 is defined by its covers : @xmath286 if @xmath287 and @xmath47 is a reflection . for each @xmath288 , define the _ @xmath246-bruhat order _ ( on @xmath21 or @xmath6 ) by its covers : set @xmath289 if @xmath286 and @xmath290 for example , figure [ fig : covers ] shows all covers @xmath291 of @xmath292 , the reflection @xmath47 , and for which @xmath246 this is a cover in the @xmath246-bruhat order .
@xmath293 while these orders are analogous to the @xmath246-bruhat orders on @xmath247 @xcite , only the 0-bruhat order on @xmath294 enjoys most properties of the @xmath246-bruhat orders on @xmath247 .
this is because the 0-bruhat order is an induced suborder of the @xmath43-bruhat order on @xmath284 .
the _ length _ @xmath295 of a permutation @xmath296 counts the inversions of @xmath9 : @xmath297 the bruhat order ( @xmath298 ) on @xmath284 is defined by its covers : @xmath299 if and only if @xmath16 is a transposition and @xmath300 . if @xmath301 , this is a cover ( written @xmath302 ) in the k - bruhat order ( @xmath303 ) on @xmath284 ( or @xmath217}$ ] ) if @xmath304 with @xmath305 .
the k - bruhat order has a non - recursive characterization , needed below : [ prop : one ] let @xmath306 and @xmath301 . then @xmath307 if and only if 1 .
@xmath308 implies @xmath309 and @xmath310 .
2 . if @xmath311 , @xmath312 , and @xmath313 , then @xmath305 . for the remainder of this section
, we will be concerned with the case @xmath314 .
[ thm : induced_order ] the 0-bruhat order on @xmath6 is the order induced from the 0-bruhat order on @xmath284 by the inclusion @xmath315 .
* proof . * for @xmath316 ,
it is straightforward to verify @xmath317 and @xmath318 thus @xmath319 , and so @xmath320 is a suborder of @xmath321 . to show this suborder
is induced , suppose that @xmath322 with @xmath316 , and argue by induction on @xmath323 .
suppose @xmath324 . if @xmath325 , then @xmath326 and we are done by induction .
suppose now that @xmath327 . by the involution @xmath328 of @xmath329 , @xmath330 also satisfies @xmath331 .
the criteria of proposition [ prop : one ] show that either @xmath332 and @xmath333 so that @xmath334 , or else @xmath335 and so @xmath336 . in the first case , @xmath337 , and in the second , @xmath338 , which completes the proof .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ rem : covers ] if @xmath339 , then either @xmath340 or else both @xmath341 and @xmath342 .
we illustrate theorem [ thm : induced_order ] in figure [ fig : obtain ] .
there , the elements of @xmath81 are boxed .
@xmath343 this relation between the two partial orders @xmath344 and @xmath329 makes many properties of @xmath344 easy corollaries of the analogous results for @xmath329 ( established in @xcite ) .
we discuss these properties in the remainder of this section , omitting proofs .
the 0-bruhat order has a non - recursive characterization : [ prop : thma ] let @xmath316 .
then @xmath186 if and only if 1 .
@xmath345 , and 2 .
@xmath346 and @xmath347 . for @xmath348 ,
let @xmath349 be the cardinality of @xmath2 .
then the inclusion @xmath350 induces compatible inclusions @xmath351 } \put(30,47){\vector(1,0){65 } } \put(30,49){\oval(4,4)[l ] } \put(5,37){\vector(0,-1){24 } } \put(7,37){\oval(4,4)[t ] } \put(105,37){\vector(0,-1){24 } } \put(107,37){\oval(4,4)[t ] } \put(60,8){$\varepsilon_p$ } \put(60,52){$\varepsilon_p$ } \put(0,45){${\mathcal b}_{\#p}$ } \put(-5,1){${\mathcal s}_{\pm\#p}$ } \put(100,45){${\mathcal b}_\infty$ } \put(98,1){${\mathcal s}_{\pm\infty}$ } \end{picture}\ ] ] _ shape equivalence _ is the equivalence relation on @xmath6 induced by @xmath352 for @xmath353 and @xmath354 .
let @xmath110_0:=\{v\mid u\leq_0v\leq_0w\}$ ] denote the interval in the 0-bruhat order between @xmath59 and @xmath9 , a finite graded poset .
a corollary of theorems [ thm : induced_order ] and theorem e(i ) of @xcite is the following fundamental result about the 0-bruhat order on @xmath6 .
* theorem b(1 ) * _ suppose @xmath355 with @xmath16 shape equivalent to @xmath167
. then @xmath110_0\simeq [ x , z]_0 $ ] .
if @xmath356 , then this isomorphism is given by @xmath357_0\ni v\ \longmapsto\ \varepsilon_p(vu^{-1})x \in [ x , z]_0.\ ] ] _
this property is not shared by the @xmath246-bruhat order on @xmath6 , for any @xmath358 .
[ ex : not_nice ] consider the following two intervals in the 1-bruhat order on @xmath359 : @xmath360 note that @xmath361 and the two intervals are not isomorphic .
[ rem : x ] for any @xmath362 , there is a @xmath363 with @xmath147 : suppose @xmath364 with @xmath365 .
further let @xmath366 .
if we define @xmath367 , then proposition [ prop : thma ] implies that @xmath147 . note that
if @xmath368=\{a>0\mid \zeta(a)\neq a\}$ ] , then @xmath369 is a grassmannian permutation . by theorem b(1 )
, we may define a new partial order on @xmath6 , which we call the _ lagrangian order : _ for @xmath370 , set @xmath371 if there is a @xmath363 with @xmath140 . by remark [ rem : x ]
, it has a unique minimal element @xmath256 .
this order is graded by the rank , @xmath372 , where @xmath373 whenever @xmath147 .
these notions have definitions independent of @xmath44 : [ defth:2.6 ] let @xmath370 . 1
. then @xmath138 if and only if 1 .
@xmath374 with @xmath375 , and 2 .
@xmath376 with @xmath311 , @xmath377 , and @xmath378 .
@xmath379 + @xmath380 .
* let @xmath59 be the permutation with @xmath142 constructed from @xmath173 in remark [ rem : x ] .
if @xmath381 , then @xmath171 satisfies the conditions in ( _ i _ ) , and conversely . for ( _ ii _ ) , consider the difference @xmath382 .
the length of @xmath369 is the first sum , plus the number of inversions of the form @xmath383 with @xmath384 . in the construction of @xmath59 ,
each of these is also an inversion in @xmath59 involving positions @xmath383 , and so are canceled in the difference .
the second term counts the remaining inversions of this type in @xmath59 , the third term counts the inversions with @xmath50 in @xmath59 , and the fourth term is @xmath385 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' the lagrangian order is the @xmath6-counterpart of the grassmann - bruhat order @xmath386 on @xmath247 of @xcite .
this is defined as follows : let @xmath387 .
then @xmath388 if and only if there is a @xmath389 with @xmath390 .
the grassmann - bruhat order is ranked with @xmath391 whenever @xmath392 .
let @xmath393 count the sign changes @xmath394 in @xmath173 .
we have the following relation between these two orders .
[ cor : compare_orders ] 1 .
@xmath395 is an induced suborder of @xmath396 .
2 . for @xmath397
, we have @xmath398 .
* the first statement is a consequence of theorem [ thm : induced_order ] . for the second statement , consider any maximal chain in @xmath191_\prec$ ] ( in @xmath6 ) . by theorem [ thm : induced_order
] , this gives a maximal chain in @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] ( in @xmath284 ) , where covers of the form @xmath399 are replaced by @xmath400 .
thus @xmath401 , where @xmath402 counts the covers in that chain if the form @xmath399 .
since only covers of the form @xmath403 contribute to @xmath393 , we have @xmath404 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath370 .
if @xmath405 with @xmath406 , and neither of @xmath173 or @xmath171 is the identity , then @xmath407 is the _ disjoint product _ of @xmath171 and @xmath173 .
( in general @xmath408 . )
if a permutation can not be factored in this way , it is _ irreducible_. permutations @xmath362 factor uniquely into irreducibles .
this is described in terms of non - crossing partitions @xcite : ( a non - crossing partition of @xmath409 is a set partition such that if @xmath410 with @xmath411 in a part @xmath412 and @xmath413 in a part @xmath414 , then @xmath415 , as otherwise @xmath416 are crossing . ) first , consider @xmath173 as an element of @xmath284 .
let @xmath417 be the finest non - crossing partition of @xmath409 which is refined by the partition given by the cycles of @xmath173 .
for each non - singleton part @xmath412 of @xmath417 , let @xmath418 be the product of the cycles of @xmath173 which partition @xmath412 .
( these @xmath418 are the irreducible factors of @xmath173 , as an element of @xmath284 . ) since @xmath362 , for each such part @xmath412 of @xmath417 , either @xmath419 or else one of @xmath420 consists solely of positive integers . in the first case
, @xmath418 is an irreducible factor of @xmath173 ( as an element of @xmath6 ) , and in the second , @xmath421 is an irreducible factor of @xmath173 . the main result concerning this disjointness is the following : [ thm : disj - prod ] suppose @xmath422 is the factorization of @xmath362 into irreducibles .
then the map @xmath423 induces an isomorphism @xmath424_\prec\times\cdots\times[e,\zeta_s]_\prec\ \stackrel{\sim}{\relbar\joinrel\longrightarrow}\ [ e,\zeta]_\prec.\ ] ] this factorization into irreducibles suggests defining a type @xmath0 non - crossing partition to be a non - crossing partition @xmath417 whose blocks @xmath412 are either stable under @xmath257 ( @xmath425 ) , or else @xmath426 are distinct , with one consisting solely of positive integers .
these differ from the non - crossing partitions of type @xmath0 introduced by reiner @xcite , which form a graded lattice .
figure [ fig : three ] shows the partitions of @xmath427 $ ] defined here .
@xmath428 we summarize some properties of @xmath395 .
[ thm : prpoerties ] 1 .
@xmath395 is a graded poset with minimal element @xmath256 and rank function @xmath144 .
the map @xmath429 exhibits the lattice of strict partitions as an induced suborder of @xmath395 .
if @xmath147 , then @xmath430 induces an isomorphism @xmath191_\prec\stackrel{\sim}{\longrightarrow}[u,\zeta u]_0 $ ] .
if @xmath371 , then @xmath431 induces an isomorphism @xmath432_\prec \stackrel{\sim}{\longrightarrow}[e,\zeta\eta^{-1}]_\prec$ ] .
5 . for every infinite set @xmath353 ,
the map @xmath433 is an injection of graded posets .
thus if @xmath370 are shape equivalent , then @xmath191_\prec\simeq [ e,\eta]_\prec$ ] .
the map @xmath434 induces an order - reversing isomorphism between @xmath191_\prec$ ] and @xmath435_\prec$ ] . if @xmath362 is factored into into disjoint cycles in @xmath284 , the resulting cycles have one of two forms : @xmath436 with @xmath437 distinct
furthermore , every cycle @xmath438 of the first type is paired with another , @xmath439 , also of the first type .
this motivates a ` cycle notation ' for permutations @xmath362 .
write @xmath440 for the product @xmath441 and @xmath442 $ ] for cycles @xmath443 of the second type .
call either of these _ cycles _ in @xmath6 . in some examples and figures , the commas may be omitted . with this notation , figure [ fig : new_order ]
shows the lagrangian order on @xmath81 .
the thickened lines are between skew grassmannian permutations @xmath444 for @xmath277 .
@xmath445 the ` schubert vs. schur ' structure constants @xmath98 and @xmath99 are related to the enumeration of ( saturated ) chains in the 0-bruhat order on @xmath6 , and hence to the enumeration of chains in the lagrangian order on @xmath6 .
we develop the elementary theory of chains in these orders along the lines of @xcite . a chain in either @xmath446_0 $ ] or @xmath191_\prec$ ] is a particular factorization of @xmath173 into transpositions @xmath447 and @xmath448 .
we give an algorithm for finding a chain in @xmath191_\prec$ ] .
for this , set @xmath449 .
[ alg : chain ] input : a permutation @xmath450 .
output : permutations @xmath451 such that @xmath452 is a saturated chain in the lagrangian order .
output @xmath173 . while @xmath453 , do 1 .
choose @xmath454 maximal subject to @xmath455 .
2 . choose @xmath163 minimal subject to @xmath456 .
@xmath457 , output @xmath173 . before every execution of 3 , @xmath458 . moreover , this algorithm terminates in @xmath372 iterations and the reverse of the sequence produced is a chain in @xmath191_\prec$ ] .
the lagrangian order has a theory of reduced decompositions , analogous to the usual theory for the coxeter group @xmath294 with respect to the weak order .
( also analogous to that for the grassmann - bruhat order on @xmath247 @xcite . )
we express this in the context of monoids .
define a monoid @xmath459 with 0 and generators @xmath460 for integers @xmath461 , one for each reflection of the weyl group @xmath6 . to simplify the list of relations these satisfy ,
set @xmath462 .
also , if @xmath463 is a relation between words @xmath464 and @xmath465 in these generators , then @xmath466 , where @xmath467 is the word @xmath464 read backwards . that said , these generators are subject to the following relations : @xmath468 these hold because @xmath459 is a sub monoid of the monoid for the grassmann - bruhat order @xcite in the same way the lagrangian order is an induced suborder of the grassmann - bruhat order on @xmath247 .
the relation between @xmath459 and the lagrangian order on @xmath6 is obtained via a faithful representation of @xmath459 as linear operators on @xmath469 : define linear operators @xmath470 and @xmath471 on @xmath469 by @xmath472 to simplify the following statement , let the index @xmath127 represent either of @xmath473 or @xmath474 . 1 . the operators @xmath475 satisfy the relations ( 1.1 ) , and a composition of operators is characterized by its value at the identity .
that is , @xmath476 if and only if @xmath477 .
2 . for @xmath478 , the map @xmath479 is a faithful representation of @xmath459 .
3 . the following map is a well defined bijection : @xmath480 4 . for @xmath481 , @xmath371 if and only if there exists @xmath482 such that @xmath483 .
the set @xmath484 corresponds to the set of all maximal chains in @xmath191_\prec$ ] .
we call the elements of @xmath485 the _ @xmath188-reduced decompositions _ of @xmath173 .
let @xmath486 denote either @xmath487 equipped with non - degenerate symmetric bilinear form or @xmath488 equipped with a non - degenerate alternating bilinear form . in the first case ,
@xmath486 an _ odd orthogonal ( vector ) space _ , and in the second , a _
symplectic ( vector ) space_. a linear subspace @xmath489 of @xmath486 is _ isotropic _ if the restriction of the form to @xmath489 is identically zero .
isotropic subspaces have dimension at most @xmath102 .
an _ isotropic flag _ in @xmath486 is a sequence @xmath490 of isotropic subspaces : @xmath491 where @xmath492 .
let @xmath493 be the annihilator of a subset @xmath489 of @xmath486 .
given an isotropic flag @xmath490 in @xmath486 , we obtain a canonical complete flag in @xmath486 ( also written @xmath490 ) by defining @xmath494 for @xmath495 , and in the odd orthogonal case , @xmath496 . henceforth , flags will always be complete , although we may only specify @xmath497 . the group @xmath10 of linear transformations of @xmath486 which preserve the given form acts transitively on the set of isotropic flags in @xmath486 .
since the stabilizer of an isotropic flag is a borel subgroup @xmath0 of @xmath10 , this exhibits the set of isotropic flags as the homogeneous space @xmath7 . here , @xmath10 is either @xmath18 ( odd orthogonal ) or @xmath17 ( symplectic ) .
similarly , @xmath10 acts transitively on the set of maximal isotropic subspaces of @xmath486 , exhibiting it as the homogeneous space @xmath498 . here
@xmath97 is the stabilizer of a maximal isotropic subspace , a maximal parabolic subgroup associated to the simple root of exceptional length .
let @xmath499 be the projection map .
the rational cohomology rings @xcite of @xmath7 for both the symplectic and odd - orthogonal flag manifolds are isomorphic to @xmath500/ \langle e_i(x_1 ^ 2,\ldots , x_n^2 ) , i=1,\ldots , n\rangle,\ ] ] where @xmath501 is the @xmath33th elementary symmetric polynomial in @xmath502 . however , their integral cohomology rings differ @xcite : @xmath503/ \left\langle e_i(x_1 ^ 2,\ldots , x_n^2 ) \right\rangle.\\ h^*so_{2n+1}{\mathbb c}/b&\simeq & { \mathbb z}[x_1,\ldots , x_n , c_1,\ldots , c_n]/
i,\\ i & = & \left\langle e_i(x_1 ^ 2,\ldots , x_n^2),\ , 2c_i - e_i(x_1,\ldots , x_n ) , \ , c_{2i}=(-1)^ic_i^2\right\rangle.\end{aligned}\ ] ] these rings have another description in terms of schubert classes .
since an isotropic flag @xmath504 is also a complete flag in @xmath486 , we have a canonical embedding @xmath505 , the manifold of complete flags in @xmath486 .
similarly , there is an embedding @xmath506 , the grassmannian of @xmath102-dimensional subspaces of @xmath486 .
we use these maps to understand some structures of @xmath7 . given @xmath26 and an isotropic ( complete ) flag @xmath504 , the _
schubert variety _
@xmath507 of @xmath7 is the collection of all flags @xmath508 satisfying @xmath509 for each @xmath510 $ ] and @xmath511 ( @xmath512 in the symplectic case ) .
this has codimension @xmath31 in @xmath7 .
also , @xmath513 if and only if @xmath514 in the bruhat order .
the schubert cell @xmath515 is the set of flags @xmath516 for which equality holds in ( [ eq : flag_def ] ) .
these are the flags in @xmath517 which are not in any sub - schubert variety ( @xmath507 with @xmath514 ) . if now @xmath518 and @xmath519}$ ] ( @xmath217}$ ] in the symplectic case ) , then the schubert variety @xmath520 of @xmath521 is the collection of flags @xmath522 satisfying [ eq : flag_def ] ) for all @xmath523 ( @xmath524 in the symplectic case ) . furthermore , if @xmath26 and @xmath504 , then @xmath525 and this is a scheme - theoretic equality @xcite .
the schubert cells constitute a cellular decomposition of @xmath7 . thus _
schubert classes _ , the cohomology classes poincar dual to the fundamental cycles of schubert varieties , form @xmath526-bases for these cohomology rings .
write @xmath25 for the class @xmath527 $ ] in @xmath27 poincar dual to the fundamental cycle of @xmath507 of @xmath19 and @xmath28 for the corresponding class in @xmath29 .
since these are bases , there are integral structure constants @xmath528 and @xmath11 for @xmath529 defined by the identities : @xmath530 let @xmath106 count the number of sign changes in the permutation @xmath9 . then the isomorphism of rational rings is induced by the map @xcite : @xmath531 thus @xmath532 it suffices to establish identities and formulas for @xmath20 .
we do this , because a crucial geometric result ( theorem [ thm : geom_facts](2 ) ) does not hold for @xmath19 .
two flags @xmath533 are _ opposite _ if @xmath534 for all @xmath33 . in
what follows , @xmath535 and @xmath536 , will always denote schubert varieties defined by fixed , but arbitrary , opposite isotropic flags .
a consequence of kleiman s theorem on the transversality of a general translate @xcite , results in @xcite , and some combinatorics , is the following proposition .
[ prop : transverse ] let @xmath115
. then @xmath537 if and only if @xmath514 in the bruhat order .
if @xmath514 , then @xmath538 meet generically transversally , and the intersection cycle is irreducible of dimension @xmath104 .
the top - dimensional component of @xmath539 is generated by the class of a point @xmath540={\mathfrak b}_{\omega_0}$ ] or @xmath541 .
the map @xmath542 selects the coefficient of [ pt ] in a cohomology class .
the _ intersection pairing _ on @xmath543 is the composition @xmath544 by proposition [ prop : transverse ] , the product @xmath545\cdot [ y_v]$ ] is the cohomology class @xmath546 $ ] . in particular , when @xmath547 , these intersections are single , reduced points , so that @xmath545 $ ] and @xmath548 $ ] are dual under the intersection pairing .
thus @xmath549 which is also the number of points in the intersection @xmath550 where @xmath551 is defined by a flag @xmath552 opposite to both @xmath490 and @xmath553 ( which define @xmath535 and @xmath554 ) .
we derive a useful description of flags in @xmath555 when @xmath186 . for @xmath556 , let @xmath557 be the linear span of @xmath189 . [
lem : param ] suppose that @xmath186 and @xmath533 are opposite isotropic flags in @xmath486 .
then there are algebraic functions @xmath558 for @xmath559 such that for each flag @xmath560 and each @xmath559 , 1 .
@xmath561 , and 2 .
. * proof . * the representation of schubert cells via parameterized matrices give @xmath486-valued functions @xmath563 defined on the schubert cell @xmath515 such that if @xmath516 is a flag in that cell , then @xmath564 , and @xmath565 .
construct the functions @xmath566 inductively .
first , set @xmath567 for @xmath560 . since @xmath568 , conditions 1 and 2 are satisfied for @xmath569 .
suppose we have constructed @xmath570 for @xmath571 .
let @xmath572 be the intersection of @xmath573 with the affine space @xmath574 there is a unique point of intersection : since @xmath575 , @xmath576 since @xmath186 , if @xmath577 and @xmath578 , then necessarily @xmath579 , by condition 2 of proposition [ prop : thma ] .
hence @xmath580 and so @xmath581 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' schubert varieties @xmath582 of @xmath498 are indexed by strict partitions @xmath84 : decreasing sequences @xmath583 of positive integers .
the projection map @xmath499 maps schubert varieties to schubert varieties , with @xmath584 , where @xmath84 consists of the positive numbers among @xmath585 arranged in decreasing order . for a strict partition @xmath84 ,
let @xmath586 be the grassmannian permutation whose ( initial ) negative values are @xmath587 .
let @xmath588 be the decreasing sequence obtained from the integers in @xmath254 $ ] which do not appear in @xmath84 .
then @xmath589 and @xmath590 is birational .
one may see this by considering typical elements of their schubert cells
. set @xmath591 $ ] in @xmath592 ( equivalently @xmath593 $ ] in @xmath27 ) and let @xmath594 be the corresponding class in the symplectic case . for @xmath36 , the _ special schubert variety _
@xmath595 is the collection of all maximal isotropic subspaces which meet a fixed @xmath596-dimensional isotropic subspace .
let @xmath3 ( respectively @xmath5 ) denote the class @xmath597 in either @xmath27 or @xmath592 ( respectively the class @xmath598 in either @xmath29 or @xmath599 ) .
we are particularly interested in the constants @xmath600 and @xmath601 which give the structure of the cohomology of @xmath7 as a module over the cohomology of @xmath498 . using the intersection pairing and the projection formula ( _ cf .
_ @xcite ) , we have @xmath602 and a similar formula for @xmath98 . our main technique will be to find formulas for @xmath603 by studying the effect of the map @xmath412 on the cycle @xmath604 . to that end
, define @xmath605 .
these cycles @xmath606 are , like schubert varieties , defined up to translation by @xmath10 . in the theorems below , write @xmath607 to mean that the cycles may be carried onto each other by an element of @xmath10 .
( the proofs are more explicit . )
the main result of section [ sec : identity ] concerns these cycles .
[ thm : geometric_shapes ] let @xmath115 with @xmath608
. then 1 .
the map @xmath609 has degree 1 . 2 .
if we have @xmath610 with @xmath611 and @xmath16 shape equivalent to @xmath167 , then @xmath607 .
this implies the identity of theorem b(2 ) . as a consequence of theorem [ thm : geometric_shapes](2 ) ,
define @xmath612 for any @xmath613 with @xmath147 .
these cycles satisfy more identities .
let @xmath614 be a @xmath615-dimensional symplectic space .
we study the manifolds of maximal isotropic subspaces of @xmath616 , and @xmath617 , together with a map @xmath618 defined by @xmath619 . also define @xmath152 by @xmath620 for @xmath154 and @xmath156 by @xmath621 for @xmath622 and @xmath623 .
[ thm : geom_facts ] 1 . for any @xmath137 , @xmath624 .
2 . if @xmath407 is a disjoint product in @xmath625 with @xmath626 shape equivalent to @xmath171 and @xmath627 shape equivalent to @xmath173 , then @xmath628 3 . for any @xmath137 , @xmath629 4 . for any @xmath137 with @xmath207 for every @xmath163
, @xmath630 the first statement follows from the observation that @xmath631 . since @xmath632 , theorem c(1 ) follows from part ( 3 ) , by the projection formula
. similarly , theorem c(2 ) follows from part ( 4 ) .
we remark that part ( 2 ) is true only for the symplectic case , while ( 1 ) , ( 3 ) , and ( 4 ) hold also for the odd orthogonal flag manifold . statements ( 2 ) ,
( 3 ) , and ( 4 ) are proven in section [ sec : more_identities ] .
we establish theorem [ thm : geometric_shapes ] which implies theorem b(2 ) . as in section
[ sec : orders ] , many results and methods are similar to those of @xcite for analogous results about @xmath73 . our discussions are therefore brief .
the results here hold for both @xmath19 and @xmath20 , with nearly identical proofs . we only provide justification for @xmath20 .
let @xmath633 be a symplectic space of dimension 2 .
then the orthogonal direct sum @xmath634 is a symplectic space of dimension @xmath635 . for each @xmath636 , define embeddings @xmath637 , the space of isotropic flags in @xmath634 , by @xmath638 define @xmath639 by replacing @xmath640 with @xmath641 in the definition above .
we find the effect of these maps on cohomology by determining the image of a schubert variety under @xmath642 .
first , define two maps between @xmath21 and @xmath643 . for every @xmath636 and @xmath644 $ ] , define the injection @xmath645 by : @xmath646
let @xmath647 be the left inverse of @xmath648 , defined by @xmath649 .
if we represent permutations as permutation matrices , then the effect of @xmath650 on @xmath651 is to erase the @xmath652th and @xmath653th columns and the @xmath654th and @xmath655th rows .
the effect of @xmath648 on @xmath9 is to expand its permutation matrix with new @xmath652th , @xmath653th columns and @xmath656th , @xmath657th rows filled with zeroes , except for 1s at positions @xmath658 and @xmath659 . for example : @xmath660 these definitions imply the following proposition ( _ cf . _
@xcite , lemma 12 ) .
[ prop : subsets ] let @xmath26 , @xmath661 , and @xmath490 any isotropic flag .
then @xmath662 recall that @xmath256 is the identity permutation and @xmath663 is the flag manifold @xmath7 .
[ cor : intersection ] let @xmath664 be opposite isotropic flags .
then for any @xmath644 $ ] , @xmath665 are opposite flags , and for any @xmath636 , @xmath666 equals either of the following cycles @xmath667 * proof . *
it is straightforward to check that the flags are opposite .
moreover , by proposition [ prop : subsets ] , @xmath666 is a subset of either intersection , as @xmath668 . since @xmath669 and @xmath670 , proposition [ prop : transverse ] implies that all three cycles are irreducible with the same dimension , proving their equality .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ cor : push - forward ] for any @xmath26 and @xmath671 , we have @xmath672 as in section 4.2 of @xcite , the schubert classes @xmath673 and @xmath674 are certain special schubert classes from grassmannian projections . were the corresponding pieri - type formulas known , we could deduce formulas for @xmath675 .
only one of these classes is a special schubert class from @xmath498 : @xmath676 we deduce formulas for @xmath677 and @xmath678 from this .
recall that @xmath29 is generated by @xmath679 .
these classes are chern classes of certain line bundles on @xmath20 : let @xmath680 be the flag of bundles whose fibre at @xmath490 is @xmath490 .
then @xmath681 .
thus @xmath682 let @xmath683
. then 1 . in @xmath27 , @xmath684 , the sum over all @xmath685 with @xmath686 .
2 . in @xmath29 , @xmath687 , the sum over all @xmath685 with @xmath686 . *
* these are consequences of the projection formula and corollary [ cor : push - forward ] .
for the first , @xmath688 but @xmath689 since @xmath690 and @xmath691 , the result follows by the pieri - type formula ( theorem d ) .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ lem : reduction ] suppose @xmath692 in @xmath643 and @xmath693 for some @xmath636
. then 1 .
@xmath694 and @xmath695 .
2 . for any opposite isotropic flags
@xmath664 in @xmath486 , @xmath696 * proof . *
since @xmath103 and @xmath697 , proposition [ prop : thma ] implies that @xmath694 .
moreover , @xmath16 is shape equivalent to @xmath698 , so the first statement follows from theorem b(1 ) . for ( 2 ) , proposition [ prop : subsets ]
gives the inclusion @xmath699 . by corollary [ cor : intersection ] , @xmath700 and @xmath701 are opposite flags .
thus , by proposition [ prop : transverse ] and ( 1 ) , both sides are irreducible and have the same dimension , proving their equality .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ thm : elem_equiv ] suppose @xmath692 in @xmath643 and @xmath693 for some @xmath636 .
then for any strict partition @xmath84 , we have @xmath702 * proof .
* we first study the map @xmath703 , defined by @xmath704 .
here , @xmath705 is the grassmannian of maximal isotropic subspaces of @xmath634 .
if @xmath664 are opposite isotropic flags in @xmath486 , the analog of corollary [ cor : intersection ] is @xmath706 where @xmath707 is a decreasing sequence of length 1 .
we leave this to the reader . as this intersection is generically transverse , @xmath708 .
thus @xmath709 to see this , note that @xmath710 , where @xmath711 the kronecker delta , by the pieri formula for isotropic grassmannians @xcite .
consider the commutative diagram : @xmath712 thus @xmath713=\psi_*[{\mathcal y}^{w/_p}_{u/_p}]$ ] and the maps @xmath412 have the same degree @xmath714 as the horizontal maps are isomorphisms .
let @xmath84 be a strict partition .
then @xmath715)$ ] which is @xmath716 ) = \delta\cdot \deg ( \psi^*(q_{\lambda})\cdot [ { \mathcal y}^{w/_p}_{u/_p } ] ) = \delta\cdot\deg(q_{\lambda}\cdot[{\mathcal y}^{w/_p}_{u/_p } ] ) = c^{w/_p}_{u/_p\,\lambda}.\qquad { \setlength{\unitlength}{1.0pt}\begin{picture}(10,7.5 ) \put(5,-2.5){\rule{2.5pt}{2.5pt } } \put(2.5,0){\rule{5pt}{2.5pt } } \put(0,2.5){\rule{10pt}{2.5pt } } \end{picture}\vspace{10pt}}\ ] ] [ lem : identity ] suppose @xmath103 and @xmath717 in @xmath21 with @xmath718 and @xmath719 for all @xmath720 $ ] . then , if @xmath721 , ( respectively @xmath722 ) are defined with the same flags , there is commutative diagram @xmath723 where @xmath724 is an isomorpism between zariski open subsets of @xmath725 and @xmath726 , and the maps @xmath412 have degree 1 .
* proof of theorem b(2 ) .
* let @xmath126 with @xmath727 shape equivalent to @xmath728 and @xmath84 a strict partition . if @xmath697 , then @xmath729 with @xmath698 shape equivalent to @xmath727 .
thus we may assume that the hypotheses of lemma [ lem : identity ] hold , by theorem [ thm : elem_equiv ] .
but then @xmath730= \pi_*[y_x\bigcap y'_{\omega_0 z}]$ ] , hence @xmath731 , showing @xmath732 . ( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' * proof of lemma [ lem : identity ] . *
let @xmath566 for @xmath559 be the functions of lemma [ lem : param ] , defined for all @xmath733 .
for such @xmath516 , let @xmath734 be the flag whose @xmath735-th subspace is @xmath736 then @xmath724 is an isomorphism between the intersections of schubert cells , which are zariski dense in the intersections of schubert varieties , and the diagram is commutative .
since we may assume @xmath9 is a grassmannian permutation , and in this case , the map @xmath737 has degree 1 , it follows that the maps @xmath412 ( which have the same degree ) have degree 1 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' ''
suppose that @xmath614 is a @xmath615-dimensional complex symplectic vector space and consider the map @xmath738 defined by @xmath739 where @xmath740 and @xmath741 are maximal isotropic ( lagrangian ) subspaces .
let @xmath742 with @xmath407 a disjoint product and @xmath743 , @xmath744 .
then for any @xmath745 and @xmath746 with @xmath747 and @xmath748 ( @xmath749 is shape equivalence ) , there is an element @xmath750 of @xmath751 such that @xmath752 * proof .
* this is a consequence of lemma 5.2.1 of @xcite , the analogous fact for the classical flag manifold and grassmannian .
restricting that result to the symplectic flag manifold and grassmannian of maximal isotropic subspaces proves the theorem .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' * remark .
* this does not hold for the odd orthogonal case .
in fact , even the map @xmath753 can not be defined : if @xmath754 are odd - orthogonal spaces , then @xmath617 is an even - dimensional space .
define the permutation @xmath755 by @xmath756 .
for example , in @xmath359 , we have @xmath757 , in the cycle notation of section [ sec : orders ] . in general , @xmath758 . for a permutation @xmath137 ,
let @xmath759 denote the conjugation of @xmath173 by @xmath155 .
a ( saturated ) chain in either the lagrangian order or lagrangian rseau gives a sequence @xmath760 of labels .
the _ peak set _ of that chain consists of those @xmath761 with @xmath762 , its _ descent set _ of those @xmath763 with @xmath66 , and its _ ascent set _ those @xmath33 with @xmath67 .
* theorem c(1 ) . *
_ let @xmath137 . then @xmath764 and if @xmath84 is a strict partition with @xmath148 , then @xmath765 _ * proof . * the first statement follows from corollary [ cor : compare_orders ] as @xmath766 in @xmath767},{{\,\prec\hspace{-5.2pt}\raisebox{1.26pt}{\parenfont ) } \,}})$ ] and both @xmath173 and @xmath183 have the same number of sign changes .
the last statement is a consequence of lemma [ lem : cyclic ] below , which shows @xmath768 whenever @xmath769 and @xmath770 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ ex : cyclic ] in @xmath359 , let @xmath771}}$ ] .
then @xmath772}}$ ] .
consider the intervals @xmath191_\prec$ ] and @xmath773_\prec$ ] in the labeled lagrangian rseau displayed in figure [ fig : cyclic ] .
@xmath774 while they are not isomorphic , they have the same rank , the same number of maximal chains , @xmath775 , and the underlying orders each have 5 chains .
moreover , they each have 2 chains with peak set @xmath776 , and one each with peak sets @xmath777 , and @xmath778 .
the rseaux have the same number of chains with fixed descent sets .
the @xmath249th component of the following vector records the number of chains with descent set equal to the position of the 1 s in the binary representation of @xmath779 : @xmath780 [ def : etdot ] let @xmath664 be opposite isotropic flags in @xmath486 . define a flag @xmath781 by :
@xmath782 define @xmath783 the same way , but with the roles of @xmath490 and @xmath553 reversed .
this gives opposite flags @xmath784 , and since @xmath785 , they are isotropic .
[ lem : cyclic ] suppose @xmath126 with @xmath786 , and @xmath787 for @xmath559 .
then , for any opposite isotropic flags @xmath664 in @xmath486 , there is a commutative diagram @xmath788 with @xmath724 an isomorpism of zariski open subsets of @xmath789 and @xmath790 .
* let @xmath791 be oppsite ( not necessarily isotropic ) flags in @xmath486 .
define @xmath792 to be @xmath793 define @xmath794 to be @xmath795 for @xmath796}$ ] , let @xmath797 be the conjugation of @xmath173 by the cycle @xmath798 . in section 5.3 of @xcite , the following proposition is proven : [ prop : cyclic ] let @xmath799}$ ] with @xmath322 , @xmath800 , @xmath801 , and @xmath9 is a grassmannian permutation with descent @xmath43 , ( @xmath802 and @xmath803 ) . if @xmath804 is the projection , then there is a commutative diagram : @xmath805 with @xmath724 an isomorpism of zariski open subsets of @xmath806 and @xmath807 .
it suffices to prove lemma [ lem : cyclic ] with @xmath9 a grassmannian permutation , by lemma [ lem : identity ] .
observe that @xmath808 is the result of @xmath102 applications of the map @xmath809 .
similarly , @xmath810 .
thus , iterating proposition [ prop : cyclic ] @xmath102 times gives the commutative diagram in @xmath811 and @xmath812 : @xmath813 restricting this to the subset of isotropic flags gives the diagram of the lemma .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' these same arguments prove the analog of lemma [ lem : cyclic ] for @xmath814 . until now
, we have deduced identities in @xmath29 by rectricting constructions involving schubert subvarieties of @xmath521 to those in @xmath20 via the embedding @xmath815 . this is the geometric counterpart of the embedding @xmath816}$ ] studied in section 2 . here
, we explore the geometry of the map @xmath817 , where a permutation @xmath818 is extended to act on @xmath819 $ ] by @xmath820 .
an immediate consequence of definition - theorem [ defth:2.6 ] and its analog for @xmath386 is the following lemma . [ lemma : iota ] the map @xmath821 is an embedding of bruhat orders @xmath822 and it respects the length functions in each order .
furthermore , @xmath823 consists of those permutations @xmath137 with @xmath162 for all @xmath163 .
let @xmath824 be complementary lagrangian subspaces in @xmath486 .
the pairing @xmath825 , where @xmath56 is the alternating form , identifies them as linear duals .
given a subspace @xmath826 of @xmath827 , let @xmath828 denote its annihilator in @xmath829 .
then @xmath830 is a lagrangian subspace of @xmath486 .
let @xmath831 be the space of complete flags @xmath832 in @xmath827 .
note that here @xmath833 .
for each @xmath834 , define an injective map @xmath835 by @xmath836 then @xmath837 is lagrangian , showing that @xmath838 is an isotropic flag . for @xmath839
the schubert variety @xmath520 of @xmath831 consists of those flags @xmath840 satisfying @xmath841 we determine the image of schubert varieties of @xmath831 under these maps @xmath842 .
define @xmath843 by @xmath844 note that @xmath845 .
[ lem : epsilon_k ] let @xmath846 with @xmath307 .
then @xmath847 induces an isomorphism of graded posets @xmath110_{\vartriangleleft_k } \stackrel{\sim}{\longrightarrow}[u , w]_0 $ ] and @xmath848 . * proof . *
a consequence of the definitions is that , for @xmath846 , @xmath849 and @xmath848 .
the lemma follows immediately from these observations .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' the map @xmath821 is an embedding of ranked orders @xmath850 .
let @xmath851 be defined by @xmath852 .
then @xmath520 and @xmath853 are dual under the intersection pairing , where @xmath664 are opposite flags .
[ lem : hidden_subset ] with these definitions , @xmath854 is a subset of either @xmath855 * proof .
* let @xmath856 .
we show @xmath857 , that is , for each @xmath511 and @xmath559 ( @xmath512 ) , @xmath858 suppose that @xmath859
. then @xmath860 .
if @xmath861 , then @xmath862 .
thus if @xmath863 , ( [ eq : key - identity ] ) holds as both sides equal @xmath864 .
suppose @xmath865 .
then @xmath866 and so the left side of [ eq : key - identity ] ) is @xmath867 now suppose that @xmath868 .
. thus the left side of ( [ eq : key - identity ] ) is @xmath870 if @xmath865 , then @xmath871 and only the first term of ( [ eq : new - dim ] ) contributes . by the previous paragraph , this is @xmath872 if @xmath873 , then @xmath874 , showing this equals the right side of ( [ eq : key - identity ] ) .
if now @xmath863 , then @xmath875 .
thus ( [ eq : new - dim ] ) is @xmath876 but this is @xmath877 , which is at least @xmath878 @xmath879 this equals the right side of ( [ eq : key - identity ] ) since @xmath880 implies @xmath881 .
similar arguments show @xmath882 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath846 with @xmath307 and @xmath883 be opposite flags
. then @xmath884 are opposite isotropic flags , and @xmath885 * proof .
* lemma [ lem : hidden_subset ] gives the inclusion @xmath699 and it is easy to see that @xmath884 are opposite . by lemma [ lem : epsilon_k ] , both sides have the same dimension , proving equality .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' for each @xmath886 , let @xmath887 be the projection induced by @xmath888 .
as in lemma [ lem : identity ] , if @xmath307 in @xmath74 and @xmath664 are opposite flags in @xmath831 , then the intersection @xmath889 is mapped birationally onto its image @xmath890 in @xmath891 .
furthermore , the image cycle depends only upon @xmath892 .
denote it by @xmath893 .
define @xmath894 by @xmath895 .
then @xmath896 and we have the following corollary . [ cor : hidden - identity ] for any @xmath897 , @xmath898 , where @xmath899 .
recall that @xmath900 .
* theorem c(2 ) . *
_ let @xmath901 .
then @xmath902 and if @xmath84 is a strict partition with @xmath148 , then @xmath903 _ * proof . * the first statement follows from lemma [ lemma : iota ] as @xmath904 for @xmath897 . the last statement is a consequence of the identity @xmath905 ( proposition [ prop : cyclic ] ) and corollary [ cor : hidden - identity ] .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ ex : no - iso ] let @xmath906
. then @xmath907 and @xmath908 .
the labeled intervals @xmath909_{{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}}$ ] and @xmath191_\prec$ ] are isomorphic .
consider the intervals @xmath191_\prec$ ] and @xmath910_\prec$ ] in the labeled lagrangian rseau displayed in figure [ fig : hidden ] . while they are not isomorphic , they have the same rank , the same number of maximal chains , 16 , and the underlying orders each have 2 maximal chains .
moreover , they each have a peakless chain and one with peak set @xmath911 .
the rseaux each have 2 increasing chains , 2 decreasing chains , 6 with descent set @xmath912 , and 6 with descent set @xmath911 .
for a cycle @xmath362 , let @xmath212 if @xmath173 has the form @xmath914 for @xmath915 and @xmath213 otherwise .
note that @xmath212 if and only if @xmath916 implies @xmath917 .
[ lem : minimal_cycle ] let @xmath362 be a cycle
. then @xmath918 .
recall that @xmath393 counts the number of sign changes in @xmath173 .
* proof . * a saturated chain in @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] ( in @xmath284 ) gives a factorization of @xmath173 into transpositions .
if @xmath173 consists of two cycles in @xmath284 , then @xmath919 and by corollary [ cor : compare_orders](2 ) , @xmath920 with equality only if @xmath921 , that is , only if @xmath212 .
similarly , if @xmath173 is a single cycle in @xmath284 , then @xmath922 . since @xmath213 and @xmath923 , corollary [ cor : compare_orders](2 ) gives @xmath924 if @xmath362 is irreducible and @xmath925 , then @xmath173 is a single cycle .
* recall that if @xmath926 have disjoint supports , then @xmath927 , with equality only when @xmath202 is a disjoint product . thus , by lemma [ lem : minimal_cycle ] , @xmath928 with equality only when @xmath173 is a single cycle .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' a _ minimal cycle _ is a cycle @xmath362 for which @xmath925 . for minimal cycles , @xmath929 .
a permutation @xmath362 is _ minimal _ if each of its irreducible factors are minimal cycles .
if @xmath370 with @xmath930 and @xmath173 is minimal , then so is @xmath171 * proof . * by theorem [ thm : disj - prod ] , we may assume @xmath173 is irreducible .
then @xmath173 is a single cycle and the result follows by induction on @xmath372 , similar to the proof of lemma [ lem : minimal_cycle ] .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ cor : minimal_perm ] let @xmath362 .
if @xmath422 is the factorization of @xmath173 into irreducibles , then @xmath931 with equality only if @xmath173 is minimal .
[ lem : sign - cover ] if @xmath362 is a minimal cycle with @xmath213 , then there is a unique @xmath932 with @xmath933 .
furthermore , @xmath934 * proof .
* let @xmath362 be a minimal cycle with @xmath213 .
then @xmath935 , so there is a unique @xmath916 with @xmath936 .
we prove the lemma when @xmath937 : if @xmath938 , then @xmath939 , as @xmath173 is irreducible and if @xmath940 , then replacing @xmath173 by @xmath941 , reduces to the case @xmath937 .
suppose @xmath937 .
we claim that if @xmath942 , then @xmath943 .
the lemma follows from this claim .
indeed , then condition ( ii ) of definition [ defth:2.6](1 ) is satisfied , and hence @xmath944 . since @xmath940
, we have @xmath945 . as @xmath946
, it follows that @xmath947 and thus @xmath948 .
let @xmath173 be a permutation for which the claim does not hold with @xmath949 minimal . by algorithm [ alg : chain ] ,
@xmath950 , where @xmath951 is maximal in @xmath952 and @xmath953 is minimal subject to @xmath954 .
note that @xmath955 . since @xmath956 , @xmath957 . as @xmath173 is minimal and @xmath958 , either @xmath171 is irreducible and @xmath959 or else @xmath171 is the disjoint product of two minimal cycles with @xmath951 in the support of one and @xmath960 in the support of the other . if @xmath961 , then @xmath962 as @xmath963 . then @xmath964 and @xmath965 , so @xmath966 and so @xmath171 is irreducible with @xmath951 in the support of one component and @xmath163 in the other .
but @xmath967 , contradicting disjointness .
suppose now that @xmath968 .
then @xmath969 , for otherwise @xmath970 .
thus @xmath971 .
since @xmath972 with @xmath973 , the component @xmath974 of @xmath171 whose support contains @xmath163 also contains @xmath473 and @xmath975 with @xmath976 , contradicting the minimality of @xmath949 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' we develop some additional combinatorics for the symmetric group @xmath247 . for @xmath977 ,
the @xmath246-bruhat order ( @xmath303 ) on @xmath978 is the analog of the @xmath246-bruhat order on @xmath979 .
the interval @xmath110_k$ ] in the @xmath246-bruhat order depends only upon @xmath16 , so we define the _ grassmann - bruhat order _ on @xmath978 ( the lagrangian order on @xmath6 is its analog ) by @xmath388 if there is a @xmath977 and @xmath389 with @xmath980 .
this order is ranked by @xmath981 for @xmath982 , and it has an independent description : * definition 3.2.2 of @xcite * _ let @xmath387 . then @xmath388 if and only if _ 1 .
@xmath983 2 .
@xmath984 3 .
if @xmath985 and @xmath986 ( respectively @xmath987 and @xmath455 ) with @xmath311 and @xmath988 , then @xmath989 . covers @xmath990 correspond to transpositions @xmath991 and we construct a labeled hasse diagram for @xmath992 , labeling such a cover with the greater of @xmath993 . by theorem 3.2.3 of @xcite , the map @xmath994 induces an order - reversing isomorphism between @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] and @xmath435_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] , preserving the edge labels .
also , if @xmath995 and @xmath996 is the map induced by the inclusion @xmath350 ( these maps induce shape equivalence ) , then @xmath997 induces an isomorphism @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}\stackrel{\sim}{\longrightarrow } [ e,\varepsilon_p(\zeta)]_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] , preserving the relative order of the edge labels .
specifically , an edge label @xmath33 of @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] is mapped to the label @xmath998 of @xmath999_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] .
lastly , we remark that algorithm [ alg : chain ] restricted to @xmath247 , and with @xmath448 replaced by the transposition @xmath1000 , gives a chain in the @xmath386-order on @xmath247 from @xmath256 to @xmath173 .
[ lem : prec_cover ] let @xmath1001 and suppose that @xmath951 is maximal subject to @xmath1002 .
then , for any @xmath127 1
. @xmath1003 .
2 . @xmath1004 .
* proof . * for 1 , let @xmath1005 and set @xmath1006 and @xmath1007 . note that @xmath1008 and @xmath1009 .
we claim that @xmath1010 and @xmath940 , which will establish 1 .
suppose @xmath1011 .
then , by the maximality of @xmath951 , @xmath1012 and so the definition of @xmath386 implies @xmath1013 .
since @xmath1014 , this implies @xmath1015 , a contradiction .
suppose now that @xmath1016 . by the definition of @xmath386 , this implies that @xmath1017 , and so @xmath1018 , contradicting the maximality of @xmath951 .
the second assertion follows from the first by applying the anti - isomorphism @xmath1019 between @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] and @xmath435_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] , @xmath1020 a cycle @xmath1001 is _ minimal _ if @xmath1021 .
a permutation is _ minimal _ if it is the disjoint product of minimal cycles
. a maximal chain in an interval @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] is _ peakless _ if we do not have @xmath63 for any @xmath1022 , where @xmath1023 is the sequence of labels in that chain .
[ lem : unique - vee ] suppose @xmath1001 is a minimal cycle .
then there is a unique peakless chain in the labeled interval @xmath1024_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] . if @xmath56 is the smallest label in such a chain , then the transposition of that cover is @xmath1025 where @xmath1026 are the two smallest elements of @xmath952 .
* we argue by induction on @xmath1027 , which we assume is at least 2 , as the case @xmath1028 is immediate . replacing @xmath173 by a shape equivalent permutation if necessary , we may assume that @xmath1029 $ ] , so that @xmath1030 . replacing @xmath173 by @xmath941
would only reverse such a chain , so we may assume that @xmath1031 .
we claim that @xmath1032 .
given this , the conclusion of the lemma follows .
indeed , let @xmath1033 . since @xmath1034 , this is an irreducible minimal permutation in @xmath1035 . by the inductive hypothesis , @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont )
} \,}}$ ] has a unique chain with labels @xmath1036 , and each @xmath1037 .
the unique extension of this to a chain in @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] has @xmath1038 .
this is the unique such chain in @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] as @xmath990 is the unique terminal cover in @xmath191_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] with edge label @xmath102 , by lemma [ lem : prec_cover ] . also note that unless @xmath1039 , @xmath1040 , which proves the second part of 1 . by algorithm [ alg : chain ] , if @xmath953 is chosen minimal so that @xmath1041 , then @xmath1042 .
we show that @xmath1043 , which will establish the claim and complete the proof .
suppose @xmath955 .
since @xmath1044 , the minimality of @xmath953 implies that @xmath1045 .
but then @xmath1046 consists of two cycles @xmath171 and @xmath974 and we have @xmath1047 and @xmath1048 . since @xmath1049 , these cycles are not disjoint , so we have @xmath1050 a contradiction .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' the labeled lagrangian and 0-bruhat orders on @xmath6 are obtained from the hasse diagrams of the underlying orders by labeling each cover with the integer @xmath56 , where that cover is either @xmath1051 ( @xmath1052 ) or @xmath1053 ( @xmath1054 ) .
recall that the map @xmath1055 maps the labeled grassmann - bruhat order on @xmath247 isomorphically onto its image in the labeled lagrangian order , preserving edge labels .
the pieri - type formula for @xmath19 has two formulations ( theorems a and d ) , which we relate here .
we say that a chain in @xmath191_\prec$ ] is _ peakless _ if in its sequence @xmath1056 of labels , we do not have @xmath1057 , for any @xmath1058 .
[ lem : minimal - peak ] let @xmath362 be a minimal cycle .
then there is a unique peakless chain in the labeled interval @xmath191_\prec$ ] . if @xmath213 , then the minimal label corresponds to the cover whose reflection is of the form @xmath1059 .
* proof . * if @xmath212 this is an immediate consequence of lemma [ lem : unique - vee ] .
suppose that @xmath213 .
replacing @xmath173 by a shape equivalent permutation if necessary , we may assume that @xmath1029 $ ] and @xmath1060 .
replacing @xmath173 by @xmath941 if necessary , we may assume that @xmath1061 .
as in the proof of lemma [ lem : unique - vee ] , @xmath1062 in the grassmann - bruhat order on @xmath284 . by remark [ rem : covers ] , either @xmath1063 or else we have both @xmath1064 and @xmath1065 .
the second case implies @xmath1066 , contradicting the minimality of @xmath173 .
thus @xmath1067 .
then @xmath171 is a minimal cycle with @xmath957 and @xmath1068 $ ] . appending the cover @xmath1069 to the unique peakless chain in @xmath909_\prec$ ]
gives a peakless chain in @xmath191_\prec$ ] .
moreover , @xmath171 is the unique permutation with @xmath958 and @xmath1069 , showing the uniqueness of this chain .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' for any @xmath362 , let @xmath1070 be the number of peakless chains in @xmath191_\prec$ ] .
[ lem : peak - prod ] if @xmath1071 are disjoint , then @xmath1072 * proof . * for @xmath1073 ,
let @xmath1074 be the multiset of words formed from labels of maximal chains in @xmath1075_\prec$ ] .
the alphabet of these words is a subset of @xmath1076 .
thus @xmath1077 and @xmath1078 have disjoint alphabets .
note that @xmath1079 consists of all pairs of words in @xmath1080 .
the lemma follows from lemma [ lem : combin ] , a combinatorial result concerning peakless words and shuffles . for a set @xmath1081 of words in
an ordered alphabet @xmath1082 , let peak@xmath1083 be the subset of peakless words from @xmath1081 .
suppose that @xmath1084 is another set of words with a different alphabet @xmath1085 and fix some total order on the disjoint union @xmath1086 which extends the given orders on each of @xmath1087 .
let sh@xmath1088 be all shuffles of pairs of words in @xmath1089 .
[ lem : combin ] the natural restriction map sh@xmath1090 induces a 2 to 1 map @xmath1091 * proof . *
it is clear that the restriction map takes a peakless word in sh@xmath1088 to a pair of peakless words in @xmath1089 . given a pair of peakless words @xmath1092 , there are exactly two shuffles of @xmath1093 which are peakless : suppose the minimal letter @xmath163 in @xmath1094 is greater than the minimal letter in @xmath1095 .
then these two shuffles differ only in their subwords consisting of @xmath163 and @xmath1096 , where @xmath1096 is that subword of @xmath1095 consisting of all letters less than @xmath163 .
then @xmath1096 is a segment of @xmath1095 , as @xmath1095 is peakless .
the two subwords of peakless shuffles are @xmath1097 and @xmath1098 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ lem : peakless - unique ] let @xmath362 and suppose there is a peakless chain in @xmath191_\prec$ ] .
then @xmath173 is minimal .
* suppose by way of contradiction that @xmath362 is irreducible and not minimal , but @xmath1099 .
we may further assume that among all such permutations , @xmath173 has minimal rank , and that @xmath1029 $ ] .
let @xmath1100 be the labels in a peakless chain in @xmath191_\prec$ ] .
replacing @xmath173 by @xmath941 if necessary ( which merely reverses the chain ) , we may assume that @xmath1101 and so @xmath1102 , by lemma [ lem : prec_cover ] and theorem [ thm : induced_order ] .
let @xmath171 be the penultimate member of this chain .
then @xmath1103 , as the initial segment of this chain gives a peakless chain in @xmath909_\prec$ ] .
thus @xmath171 is a minimal permutation , by our assumption on @xmath173 , and so @xmath1104 as @xmath173 is not minimal and @xmath958 so @xmath1105 .
therefore the weak inequalities must be equalities , so that @xmath1106 $ ] . since @xmath1107 are the labels of a chain in @xmath909_\prec$ ] and @xmath1108 , we must have @xmath1109 , as @xmath1106 $ ] .
but this contradicts our earlier observation about @xmath1110 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' ''
we relate the two formulations of the pieri formula in the odd - orthogonal case . for @xmath362 , define @xmath1111 [ cor : theta - peak ] for @xmath362 , @xmath1112 . * proof .
* this is clear if @xmath173 is minimal as both @xmath1070 and @xmath1113 satisfy the same recursion , by lemmas [ lem : minimal - peak ] and [ lem : peak - prod ] .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' the enumerative significance of the structure constants @xmath1114 is best expressed in terms of maximal chains in certain directed multigraphs associated to intervals in the bruhat order , which we call labeled rseaux . a cover
@xmath958 in the lagrangian order corresponds to a reflection @xmath1115 , which is either of the form @xmath448 or of the form @xmath1059 .
the _ lagrangian rseau _ on @xmath6 is the labeled directed multigraph where a cover @xmath958 in the lagrangian order with @xmath1116 is given a single edge @xmath1117 and a cover with @xmath1118 is given two edges @xmath1119 and @xmath1120 .
we obtain the labeled lagrangian order from this rseau by erasing those edges whose negative labels . in the grassmann - bruhat order on @xmath247 ,
there are two conventions for labeling a cover @xmath1121 : this cover gives a transposition @xmath1122 with @xmath1026 , and we may choose either @xmath127 or @xmath56 .
for want of a better term , we call the consistent choice of @xmath127 the lower convention , and the consistent choice of @xmath56 the upper convention . we make use of the following fact .
[ prop : unique - increasing ] let @xmath1123 . if there is a chain in @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] with decreasing labels in the lower convention , then there is a chain in @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] with decreasing labels in the upper convention , and these chains are unique .
the same is true for chains with increasing labels , and in either case @xmath171 is minimal .
a chain with increasing labels is an _ increasing chain _ and one with decreasing labels is a _
decreasing chain_. [ lem : inc - exists ] let @xmath362 be a minimal cycle .
then the rseau @xmath191_\prec$ ] has an increasing chain . if @xmath212 , then there are at least 2 increasing chains . * proof . * consider the peakless chain in the labeled order @xmath191_\prec$ ] : @xmath1124 let @xmath1125 be the minimal label in this chain .
then @xmath1126 for some @xmath1127 . to see this , if @xmath213 , then by lemma [ lem : minimal - peak ] , the label @xmath1125 corresponds to the only cover whose reflection is not in @xmath1128 , and so @xmath1129 .
the pullback of the initial segment of this chain to @xmath1130_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] gives a decreasing chain ( with labels @xmath1131 ) in the upper convention .
consider the unique decreasing chain @xmath1132 in the lower convention .
then @xmath1133 is an increasing chain in the rseau @xmath1134_\prec$ ] . concatenating the end of the peakless chain ( [ eq : pkls_chain ] ) onto this
gives an increasing chain @xmath1135 in the rseau @xmath1136_\prec$ ] .
suppose @xmath212 and consider the middle portion of this increasing chain : @xmath1137 let @xmath1138 be the label of the other edge between @xmath1139 and @xmath1140
. then we claim that @xmath1141 so that replacing @xmath1142 by @xmath1143 gives a second increasing chain in the rseau . to see this , first note that @xmath1144 is impossible as these are consecutive covers in the lagrangian order ( see relation ( iv ) of equation ( [ eq : relations ] ) ) .
define @xmath171 by @xmath1145 and pull this chain back to @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] .
it is the unique peakless chain in @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] and @xmath1125 is the minimal label . by lemma [ lem : unique - vee](1 ) ,
@xmath1146 and so @xmath1147 .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath362 and suppose there is an increasing chain in the rseau @xmath191_\prec$ ] .
then @xmath173 is minimal .
if @xmath173 is a minimal cycle , then there are precisely @xmath1148 such chains .
* let @xmath1149 be an increasing chain in @xmath191_\prec$ ] .
suppose that @xmath1150 .
then for @xmath1151 , @xmath1152 .
define @xmath1153 by @xmath1154 for @xmath1151 .
then @xmath1155 is a decreasing chain in @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] , with the lower labeling convention .
let @xmath1156 be the unique decreasing chain in the upper labeling convention .
concatenating the image of this chain in @xmath191_\prec$ ] with the end of the chain ( [ eq : inc_chain ] ) gives a peakless chain @xmath1157 in the interval @xmath191_\prec$ ] in the lagrangian order . by lemma [ lem : peakless - unique ]
, @xmath173 is necessarily minimal .
suppose now that @xmath173 is a minimal cycle , then the rseau @xmath191_\prec$ ] has an increasing chain and the peakless chain ( [ eq : peakless ] ) is unique .
consider another increasing chain @xmath1158 and form @xmath1159 , @xmath1160 , @xmath1161,and @xmath1162 as for the original chain [ eq : inc_chain ] ) .
if @xmath1163 , then the chains ( [ eq : inc_chain ] ) , ( [ eq : new_inc_chain ] ) coincide : the final segments agree , by the uniqueness of ( [ eq : peakless ] ) , as do their initial segments , by proposition [ prop : unique - increasing ] .
if @xmath213 , then the minimal label in the peakless chain ( [ eq : peakless ] ) ( either @xmath1164 or @xmath1165 ) corresponds to the cover whose reflection has the form @xmath1059 . as @xmath1129 , this must be @xmath1165 and so @xmath1163 and the chain ( [ eq : inc_chain ] ) is the unique increasing chain in the rseau @xmath191_\prec$ ] .
suppose now that @xmath212 and @xmath1166 .
since @xmath1167 for @xmath1151 , @xmath1168 for @xmath1169 , and @xmath1170 and @xmath1171 , we must have @xmath1172 .
but then @xmath1173 for @xmath1151 and also @xmath1174 for @xmath1175 , and so the two chains ( [ eq : inc_chain ] ) and ( [ eq : new_inc_chain ] ) agree except for the label of the cover @xmath1176 .
thus there are at most 2 increasing chains in the rseau @xmath191_\prec$ ] and their underlying permutations coincide .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' suppose @xmath1177 is a permutation in @xmath1178 .
consider @xmath1179 , a permutation in @xmath1180 .
figure [ fig : up - down ] shows the rseau @xmath191_\prec$ ] .
@xmath1181 in this rseau , there are two increasing chains @xmath1182 which correspond to the unique peakless chain in @xmath909_{{\prec\hspace{-1.5pt}\raisebox{.6pt}{\subfont ) } \,}}$ ] with @xmath1183 @xmath1184 let @xmath1185 count the increasing chains in the rseau @xmath191_\prec$ ] . if @xmath370 are disjoint , then @xmath1186 .
* proof . * as with lemma [ lem : peak - prod ] , this is a consequence of the analogous bijection concerning increasing words among shuffles of words with disjoint alphabets .
( 10,7.5 ) ( 5,-2.5 ) ' '' '' ( 2.5,0 ) ' '' '' ( 0,2.5 ) ' '' '' for @xmath362 , define @xmath1187 let @xmath1188 enumerate the decreasing chains in the rseau @xmath191_\prec$ ] . by theorem [ thm : prpoerties](6 ) , an increasing chain in @xmath173 becomes a decreasing chain in @xmath941 .
the following result is now immediate .
[ cor : chi - descent ] for @xmath362 , @xmath1189 .
recall that for @xmath137 , @xmath212 if @xmath916 implies @xmath917 and @xmath213 otherwise .
* we may suppose that @xmath1029 $ ] ; otherwise pull back the quadratic form along the map @xmath1192 ( @xmath1193 ) of section [ sec : identity ] . let @xmath664 be opposite isotropic flags . since @xmath533 are opposite flags , we have @xmath1194 .
let @xmath56 be the alternating form on @xmath486 and define a symmetric bilinear form @xmath656 on @xmath486 by @xmath1195 where @xmath1196 are the projections of @xmath1197 to the summands @xmath1198 and @xmath1199 .
this form is non - degenerate .
let @xmath724 be the associated quadratic form .
let @xmath68 with @xmath147 and @xmath369 a grassmannian permutation , as in remark [ rem : x ] .
then there is a @xmath1200 $ ] with @xmath1201 and @xmath1202 . since @xmath212 , @xmath1203 for @xmath88 and @xmath1204 for @xmath1205
let @xmath1206.then @xmath1207 and @xmath1208 thus @xmath1209 .
since @xmath1210 is isotropic for the alternating form @xmath56 , this decomposition shows it is isotropic for the symmetric form @xmath656 , proving the lemma .
[ cor : quad_forms ] suppose @xmath422 is the factorization of @xmath173 into irreducible permutations .
suppose exactly @xmath1211 of the @xmath1212 have @xmath1152 .
then there exist @xmath1211 linearly independent quadratic forms @xmath1213 on @xmath486 with @xmath1214 for every @xmath1191
. * proof .
* by theorem [ thm : geom_facts](1 ) , there exist symplectic spaces @xmath1215 with @xmath1216 such that @xmath1217 may be written as @xmath1218 , where , for @xmath1219 , @xmath1220 and @xmath1221 , with @xmath1222 shape equivalent to @xmath1212 .
( here , @xmath1223 . )
we may assume that the @xmath1212 are ordered so that @xmath1152 for @xmath1224 and @xmath1225 for @xmath1226 . by lemma [ lem :
dimension ] , for each @xmath1224 , there is a quadratic form @xmath1227 on @xmath1228 vanishing on @xmath1221 .
the pullback of @xmath1227 along the orthogonal projection @xmath1229 is a quadratic form on @xmath486 , also denoted @xmath1227 .
these are the desired forms .
the special schubert variety @xmath595 of @xmath96 consists of those lagrangian subspaces which meet a fixed @xmath1230-dimensional isotropic subspace @xmath1231 of @xmath486 . in what follows ,
write @xmath1232 for this schubert variety , and assume @xmath1231 is in general position in @xmath486 . for @xmath137 ,
define @xmath1233 for @xmath137 , define @xmath1234 to be the zero locus of the forms @xmath1213 of corollary [ cor : quad_forms ] , a complete intersection .
then @xmath1234 is the cone over a subvariety of @xmath1235 of degree @xmath1236 .
if @xmath173 is minimal , then this degree is @xmath1237 .
let @xmath137 with @xmath1238 and @xmath1029 $ ] .
let @xmath1231 be a general isotropic @xmath1230-dimensional subspace of @xmath486 .
then @xmath1239 unless @xmath173 is minimal , and in that case , @xmath1240 is @xmath1237 reduced lines .
* proof . *
since @xmath17 acts transitively on @xmath1235 , kleiman s theorem on the transversality of a general translate @xcite will imply the corollary if we show @xmath1241 with equality only if @xmath173 is minimal . by corollary [ cor : quad_forms ] , @xmath1242 , where @xmath1211 counts the irreducible factors @xmath171 of @xmath173 with @xmath227 . since @xmath1243 , corollary [ cor : minimal_perm ] implies that @xmath1244 , with equality only if each @xmath1212 is a minimal cycle , establishing ( [ eq : dim ] ) .
we deduce the pieri - type formula from theorem [ thm : singleton ] .
first , define @xmath1247 if @xmath173 is not minimal , and for @xmath173 minimal , set @xmath1248 recall that @xmath1249 was the structure constant corresponding to @xmath1250 for @xmath19 .
* let @xmath393 be the number of sign changes in a permutation @xmath362 .
since @xmath1253 and @xmath1254 if @xmath142 , equation ( [ eq : coeff_compare ] ) implies that @xmath1255 . since , for a minimal cycle @xmath173 , @xmath929 , we have @xmath1256 thus it suffices to show @xmath1251 . by theorem b(2 ) , replacing @xmath173 by a shape equivalent permutation if necessary , we may assume that @xmath1029 $ ] .
let @xmath1231 be a general isotropic @xmath596-dimensional subspace of @xmath486 . by the projection formula , @xmath1257 .
since @xmath1258 implies that @xmath489 meets @xmath1240 non - trivially , we see that @xmath1259 where @xmath1260 counts the @xmath1217 which contain a general line of @xmath1234 , and the degree is taken in @xmath1235 . by theorem [ thm : singleton ] , @xmath1261 , which completes the proof .
let @xmath422 be the irreducible factorization of @xmath173 . in the notation of the proof of corollary [ cor :
quad_forms ] , a general @xmath1262 has the form @xmath1263 , where @xmath1264 for @xmath1219 .
moreover @xmath1265 if and only if @xmath1266 . thus is suffices to prove theorem [ thm : singleton ] for @xmath173 a minimal cycle .
we do this in the following sections
. * proof . * define @xmath1123 by @xmath1145 . set @xmath1269 .
recall the notation of section [ sec : hidden ] : let @xmath1270 be complementary lagrangian subspaces of @xmath486 , which are identified as linear duals .
define the map @xmath1271 by @xmath1272 , where @xmath828 is the annihilator of @xmath826 .
define @xmath887 by @xmath888 . by corollary [ cor : hidden - identity ] ,
@xmath1273 where @xmath1274 .
schubert varieties @xmath1275 of the grassmannian @xmath891 are indexed by ordinary partitions ( _ weakly _ decreasing sequences ) @xmath1276 @xcite .
we show @xmath1277 where @xmath1278 is the hook - shaped partition with first row @xmath1279 and first column @xmath246 .
it follows from the projection formula that @xmath1280 which is @xmath1281 , the product in @xmath1282 . by (
* theorem 8) , this counts the chains in the @xmath246-bruhat order @xmath1283 with @xmath1284 .
the conclusion follows by lemma [ lem : unique - vee ] . to show ( [ eq : pullback ] ) , suppose @xmath1285 and with @xmath1286 .
let @xmath1287 be the projection of @xmath1288 into @xmath827 and @xmath1289 its projection to @xmath829
. then @xmath1290 and @xmath1291 and @xmath1292 so that @xmath1293 .
thus @xmath1294 , which is just the schubert variety @xmath1295 .
this completes the proof of theorem [ thm : singleton ] , as @xmath1297 counts the @xmath1191 which meet a generic line in @xmath486 .
we use some geometric constructions to reduce the computation of @xmath1298 to the cohomology of the classical flag manifold .
let @xmath824 be complementary lagrangian subspaces of @xmath486 with @xmath829 identified with the linear dual of @xmath827 as in section [ sec : hidden ]
. for @xmath1299 , set @xmath1300 a variety of partial flags in @xmath827 .
let @xmath1301 be the projection .
then the projections @xmath1302 of @xmath831 to @xmath1303 , @xmath891 factor through @xmath1304 .
let @xmath1302 also denote the projections of @xmath1305 to @xmath1303 , @xmath891 .
let @xmath1306 denote @xmath96 , the grassmannian of lagrangian subspaces of @xmath486 .
consider the incidence variety @xmath1307 : @xmath1308 then @xmath1307 is a @xmath1309-bundle over @xmath1305 and @xmath724 is genericaly 1 - 1 : the image of @xmath724 consists of those @xmath489 with @xmath1310 and @xmath1311 , which is an intersection of schbert varieties .
thus a generic @xmath489 in this intersection determines @xmath1312 uniquely .
let @xmath137 be minimal with @xmath213 , @xmath1317 , and @xmath1029 $ ] .
we construct a minimal permutation @xmath897 with @xmath1318 , a @xmath1319 $ ] , and @xmath839 with @xmath1320 , and @xmath1321 .
let @xmath1322 be @xmath1323 .
then @xmath1324= \wp_*({\mathfrak s}_\eta w\cdot{\mathfrak s}_{w^\vee})$ ] .
we show theorem [ thm : last ] follows from these lemmas .
@xmath1326\cdot q_n ) & = & \deg ( f_*g^ * [ { \mathcal x}_{\eta^{-1}}]\cdot q_n)\\ & = & \deg ( [ { \mathcal x}_{\eta^{-1}}]\cdot g_*f^ * q_n),\end{aligned}\ ] ] by the projection formula and lemma [ lem : geom - pull - push ] . by lemma [ lem : push - pull
] , this is @xmath1327 since @xmath1321 , only the first term is non - zero . by
* theorem 8) and lemma [ lem : unique - vee ] , this degree is 1 . *
proof of lemma [ lem : push - pull ] * the class @xmath1328 is represented by the schubert variety @xmath1329 where @xmath1330 and @xmath56 is the alternating form
. then @xmath1331 is represented by @xmath1332 which is @xmath1333 since @xmath1334 , we may write a general @xmath1288 uniquely as @xmath1335 with @xmath1336 and @xmath1337 and so @xmath1332 is a subset of @xmath1338 this is an intersection of schubert varieties ( in general position if @xmath1288 is general ) of codimensions @xmath1279 and @xmath1339 , respectively .
these schubert varieties have classes @xmath1340 and @xmath1341 .
since @xmath1342 has codimension @xmath102 , @xmath1332 equals this intersection if the map @xmath1343 is finite .
thus @xmath1344 where @xmath1260 is the degree of the map @xmath1343 ( which is 0 if the map is not finite ) . to compute @xmath1260 ,
let @xmath1345 satisfy @xmath1346 and @xmath1347 with @xmath1348 and @xmath1349 .
then @xmath1350 , and so @xmath1351 , which shows @xmath1261 .
lastly , the pieri - type formula @xcite in @xmath1352 shows @xmath1353 * proof of lemma [ lem : geom - pull - push ] * we first make some definitions .
replacing @xmath173 by @xmath941 if necessary , we may assume that @xmath1354 is the unique number with @xmath948 , by lemma [ lem : sign - cover ] .
since @xmath946 , we define @xmath897 by @xmath1355 .
set @xmath1356 let @xmath139 satisfy @xmath147 with @xmath369 a grassmannian permutation ( _ cf .
_ remark [ rem : x ] ) . since @xmath213 , @xmath1357 .
define @xmath1358 by @xmath1359 .
then @xmath1360 . let @xmath818 be defined by @xmath1361 we claim that @xmath1320 . since @xmath1106 $ ] , we can not have both @xmath1320 and @xmath1362 . since @xmath1363 , condition ( 1 ) of proposition [ prop : one ] is satisfied . for condition
( 2 ) , as @xmath369 is a grassmannian permutation , @xmath1364 so we only need show @xmath1151 with @xmath1365 implies @xmath1366 .
let @xmath1367 .
then @xmath1368 , which implies @xmath1369 and hence @xmath1366 , as @xmath1363 .
let @xmath1370}}\in{\mathcal b}_7 $ ]
. then @xmath1371 and @xmath1372 so that @xmath1373 . here , @xmath1374 . if we set @xmath1375 , then @xmath1376 is a grassmannian permutation and @xmath1377 .
we see that @xmath1378 and @xmath1379 so that @xmath1380 and @xmath1381 . * proof . *
we first define an injective map @xmath1388 , then show the restriction of @xmath640 to @xmath1389 has image contained in @xmath1390 . since the maps @xmath412 and @xmath1304 are generically 1 - 1 , by the analog of theorem [ thm : geometric_shapes ] ( theorem 5.1.4 of @xcite ) for @xmath831 , the concluson follows .
let @xmath1391 be a complete flag and let @xmath1392 , so that @xmath1393 and @xmath1394 satisfy @xmath1395 define @xmath1396 by @xmath1397 we show this defines a flag in @xmath1398 .
note first that @xmath1399 for @xmath1400 . since @xmath1401 for @xmath1402 , lemma [ lem : hidden_subset ] shows that @xmath1403 for @xmath1400 .
we show this for @xmath1404 , which shows @xmath1405 .
since @xmath369 is a grassmannian permutation , and , for @xmath916 @xmath1406 , we need only show that @xmath1407 for @xmath1404 ( as @xmath1408 in this range ) .
it suffice to show this for the dense subset of @xmath1409 with @xmath1410 .
then @xmath1411 is the image of @xmath489 under the projection @xmath1412 .
thus , for @xmath916 @xmath1413 if @xmath1414 , then @xmath1415 and these exhaust the positive values of @xmath369 .
if @xmath1404 and we set @xmath1416 , we see that @xmath1417 is @xmath1418 . since @xmath1419 is a grassmannian permutation , @xmath1420 thus @xmath1421 and so @xmath1422 , which completes the proof that @xmath1405 for @xmath1423
we complete the proof of lemma [ lem : big - diagram ] , showing that if we further require @xmath1434 , then @xmath1435 .
let @xmath553 be a flag opposite to @xmath490 in @xmath827 .
we show @xmath1436 for @xmath1437 .
note first that @xmath516 satisfies @xmath1438 as before @xmath1439 for @xmath1440 since @xmath1441 and @xmath59 agree in this range , as do @xmath1442 and @xmath1443 . for @xmath1404
, it suffices to establish ( [ eq : last ? ] ) for the dense subset of those @xmath1409 with @xmath1444 .
suppose @xmath916 .
since @xmath1444 , @xmath1445 is the image of @xmath489 under the projection @xmath1412 .
similarly , @xmath1445 is the image of @xmath1446 under this projection . since the kernel of this is @xmath1425 and @xmath1447 , @xmath1448
since @xmath1449 are the positive values of @xmath59 and @xmath1404 , this is @xmath1450 which shows ( [ eq : last ? ] ) .
now suppose @xmath1451 .
then @xmath1452 and so @xmath1453 this has dimension @xmath1454 the values @xmath1455 for @xmath1456 are simply the negative values @xmath1457 for @xmath1458 .
thus this is @xmath1459 , which completes the proofs of lemma [ lem : big - diagram ] and the pieri - type formula .
, _ sur les dcompositions cellulaires des espaces @xmath7 _ , in algebraic groups and their generalizations : classical methods , american mathematical society , 1994 , pp . 123 .
proceedings and symposia in pure mathematics , vol .
56 , part 1 . | we give the formula for multiplying a schubert class on an odd orthogonal or symplectic flag manifold by a special schubert class pulled back from a grassmannian of maximal isotropic subspaces .
this is also the formula for multiplying a type @xmath0 ( respectively , type @xmath1 ) schubert polynomial by the schur @xmath2-polynomial @xmath3 ( respectively , the schur @xmath4-polynomial @xmath5 ) .
geometric constructions and intermediate results allow us to ultimately deduce this from formulas for the classical flag manifold .
these intermediate results are concerned with the bruhat order of the coxeter group @xmath6 , identities of the structure constants for the schubert basis of cohomology , and intersections of schubert varieties .
we show these identities follow from the pieri - type formula , except some ` hidden symmetries ' of the structure constants .
our analysis leads to a new partial order on the coxeter group @xmath6 and formulas for many of these structure constants . |
few properties are known to hold for all groups ; in the recent years , counterexamples have been found to numerous `` plausible conjectures '' , usually formulated as questions : is there an infinite , finitely generated group all of whose elements have finite order ?
is there an amenable group that can not be produced using extensions and filtered colimits of virtually abelian groups ?
is there a group whose word growth is strictly between polynomial and exponential ?
the `` farrell - jones conjectures '' , predicting how the algebraic k-/l - theory of the group ring @xmath0 may be expressed in terms of the algebraic k-/l - theory of @xmath1 and the group theory of @xmath2 , is one of the prominent remaining conjectures @xcite .
if it is satisfied by the group @xmath2 , numerous group - theoretical consequences for @xmath2 follow , in particular @xmath0 has no non - trivial idempotent if @xmath2 is torsion - free and @xmath1 is a domain of characteristic @xmath3 .
the farrell - jones conjectures are inherited under many group - theoretical operations ( finite direct and free products , filtered colimits ) , but possibly not under wreath products ; we say the farrell - jones conjectures _ hold with wreathing _ if they hold for all wreath products @xmath4 with a finite permutation group @xmath5 . in search of a possible counterexample to the farrell - jones conjectures
, it might have been speculated that the `` self - similar groups '' studied by alshin , grigorchuk , gupta and sidki since the 1970s would play an important role ; indeed , these groups have served to answer or illuminate all the questions in the first paragraph .
self - similar groups are groups acting in a recursive manner on a regular rooted tree @xmath6 .
if the recursion of every element involves only a linearly growing subtree of @xmath6 , the group is said to be _
we show in this note that considerable care will be required to construct a counterexample within the class of self - similar groups .
we prove ( see below for precise definitions ) : let @xmath2 be a bounded self - similar group . then @xmath2 satisfies the farrell - jones conjectures .
let @xmath2 be a contracting similar group .
then @xmath2 satisfies the farrell - jones conjectures if its universal contracting cover satisfies the farrell - jones conjectures with wreathing .
the alshin - grigorchuk , gupta - sidki , ggs , and generalized grigorchuk groups all satisfy the farrell - jones conjectures .
wolfgang lck encouraged me to write this short note , with the intent of narrowing the domains of group theory in which a counterexample is to be searched .
thomas schick and wolfgang lck generously provided valuable feedback on a preliminary version , and clarified for me the status of the farrell - jones conjectures with respect to wreath products .
we review very briefly the statement of the farrell - jones conjectures ; we include them for definiteness , but will never work directly with their definition . a model for the _ virtually cyclic classifying space _
@xmath7 is a topological @xmath2-space @xmath8 whose isotropy groups are all virtually cyclic , and such that for any topological @xmath2-space @xmath9 with virtually cyclic isotropy groups there exists up to @xmath2-homotopy a unique @xmath2-map @xmath10 .
the farrell - jones conjectures assert that the natural map @xmath11 induced by @xmath12 , is an farrell - jones for all @xmath13 . here
@xmath14 is either the k - theory spectrum @xmath15 or the l - theory spectrum @xmath16 over the orbit category associated with an additive @xmath2-category @xmath17 . for our purposes
, it suffices to note that the class of groups for which the conjectures are known to hold contains virtually abelian groups , hyperbolic groups @xcite for @xmath18 , cat(0 ) groups @xcite , cocompact lattices in virtually connected lie groups , threefold groups @xcite and arithmetic groups over algebraic number fields ( unpublished ) .
it is closed under taking subgroups , colimits @xcite*corollary 0.8 , and finite direct and free products .
( this is the advantage of using the more general version with cofficients in an additive category the inheritance properties come almost for free ) .
note that , in general , it is not known whether the conjectures are inherited under finite extensions .
since every finite extension is a subgroup of the wreath product with a finite group @xcite , the question reduces to whether the conjecture is inherited by finite wreath products .
this is known in some specific cases , in particular for cocompact lattices in virtually connected lie groups , threefold groups , arithmetic groups over algebraic number fields , and cat(0 ) groups , as we now explain .
cat(0 ) spaces are metric spaces in which triangles are at least as thin as in euclidean space ; see the classical reference @xcite .
cat(0 ) groups , also called _ non - positively curved groups _ , are groups acting properly , isometrically and cocompactly on a cat(0 ) space of finite topological dimension . that class contains virtually abelian groups , and is closed under direct , free and finite wreath products .
[ lem : cat0 ] if @xmath2 is cat(0 ) , then so is @xmath4 for any finite permutation group @xmath5 .
let @xmath2 act properly discontinuously on the cat(0 ) space @xmath8 , and let @xmath5 be a permutation group on @xmath13 points .
then @xmath19 acts properly discontinuously on @xmath20 , with @xmath21 acting cordinatewise and @xmath5 by permutation of the cordinates .
we summarize the notion of _ self - similar group _ , presenting it in a slightly more general and algebraic manner than is usual ; see @xcite or @xcite for classical references . by @xmath22
we denote the permutational wreath product @xmath23 .
a _ self - similar group _ is a group @xmath2 endowed with a homomorphism @xmath24 , called its _ self - similarity structure_. the integer @xmath25 is the _ degree _ of the self - similarity structure .
usually , the self - similarity is implicit , and one simply denotes by @xmath2 the self - similar group . the map @xmath26 can be applied diagonally to all entries in @xmath27 , yielding a map @xmath28 , and therefore a map @xmath29 ; more generally , we get maps @xmath30 which we all denote by @xmath26
. we may compose these maps , and write @xmath31 for the iterate @xmath32 . by projecting to the permutation part
, we then have homomorphisms @xmath33 and , assembling them together , a permutational action of @xmath2 on @xmath34 ; one may identify @xmath6 with the vertex set of a rooted @xmath25-regular tree , by connecting @xmath35 to @xmath36 for all @xmath37 , in such a way that @xmath2 acts by graph isometries .
this action need not be faithful ; if it is , then @xmath2 is called a _
faithful self - similar group_. a self - similar group is _ contracting _ if there exists a finite subset @xmath38 such that , for all @xmath39 and all @xmath13 large enough , @xmath40 .
the smallest such @xmath41 is called the _ nucleus _ of @xmath2 .
let @xmath42 denote the free group on @xmath41 . by definition ,
the nucleus satisfies the condition @xmath43 .
the restriction of @xmath26 to @xmath41 can therefore uniquely be extended to a homomorphism @xmath44 . set @xmath45 similarly ,
we have @xmath46 . set @xmath47 .
the homomorphism @xmath48 then induces a homomorphism , again written @xmath49 .
note that @xmath50 is a finitely presented group , and that the natural map @xmath51 defines a homomorphism @xmath52 .
we will see in lemma [ lem : contr ] that @xmath50 is contracting , with nucleus @xmath41
. however , the self - similarity structure of @xmath50 need not be faithful , even if that of @xmath2 was faithful .
we call @xmath50 the _ universal contracting cover _ of @xmath2 . note also that in general the homomorphism @xmath52 need not be onto , or equivalently @xmath41 need not generate @xmath2 .
this is , however , the case in all examples we present here . here
are some extreme examples ; more classical ones appear in [ s : examples ] .
the full group @xmath53 of isometries of @xmath6 is self - similar , but not contracting ; actually not even countable .
its subgroup @xmath54 is faithful , self - similar , and contracting with nucleus @xmath55 .
any group @xmath2 , with @xmath56 the diagonal embedding , defines a non - faithful self - similar structure on @xmath2 , which is contracting precisely when @xmath2 is finite .
consider finally @xmath57 a finite group , and @xmath2 the group of finitely - supported functions @xmath58 .
take @xmath59 , and set @xmath60 with @xmath61 and @xmath62 .
this defines a self - similarity structure on @xmath2 , which is not faithful , and contracting with nucleus @xmath63 .
our main result does not give any interesting information on such actions .
we now generalize the definitions above to more general groups .
a group @xmath2 is _ similar _ if there exists a sequence @xmath64 of groups , a sequence of integers @xmath65 , and a sequence of homomorphisms @xmath66 .
the similarity structure is faithful if the corresponding permutational action on @xmath67 is faithful . again abusing notation , the compositions of @xmath68 s
are written @xmath69 .
let @xmath70 be a sequence of finite sets , with @xmath71 for all @xmath13 .
we say that @xmath2 _ contracts to @xmath72 _ if for every @xmath73 and every @xmath74 large enough , @xmath75 . in that case , it is possible , up to enlarging the @xmath76 s , to assume @xmath77 , and we always make that additional assumption .
we call the sequence @xmath70 a _ nucleus _ of @xmath2 . note however that the sequence @xmath70 is not unique for example , it is always possible to replace finitely many of the initial terms by @xmath78 .
we say @xmath2 is _ generated by its nucleus _ if @xmath76 generates @xmath79 for all @xmath13 .
extending the previous definition , let @xmath80 be the finitely presented group @xmath81 we then have induced homomorphisms @xmath82 , defining a similarity structure for the group @xmath83 .
[ lem : contr ] the similar group @xmath50 contracts to @xmath72 .
we again call @xmath50 the _ universal contracting cover _ of @xmath2 ; note that it depends on the choice of @xmath72 .
consider @xmath84 .
for every @xmath85 , there exists @xmath86 such that @xmath87 , by the contraction condition .
since there are finitely many @xmath88 s under consideration , there exists @xmath89 such that @xmath90 on the other hand , consider @xmath91 a word of length @xmath92 in the alphabet @xmath76 , and denote by @xmath93 and @xmath94 respectively its image in @xmath79 and in @xmath80 .
the entries in @xmath95 have length precisely @xmath96 , by construction .
they are termwise equal , in @xmath97 , to the entries of @xmath98 .
since @xmath99 contains all relations of length @xmath100 , these entries are also termwise equal in @xmath99 .
it follows that , for every @xmath101 of length @xmath102 , all entries of @xmath103 all belong to @xmath104 .
consider now @xmath105 , of length @xmath106 in the alphabet @xmath76 .
set inductively @xmath107 and @xmath108 . by the previous paragraph ,
the entries of @xmath109 have length @xmath110 over @xmath111 , and more generally the entries of @xmath112 have length @xmath113 in @xmath114 , that is , they belong to @xmath114 .
we call a similar group _ contracting _ if it has been endowed with a sequence @xmath72 to which it contracts .
note that this fixes the choice of a contracting finitely presented cover .
similar contracting groups naturally include self - similar groups , by considering constant sequences @xmath2 , @xmath26 , @xmath41 and @xmath50 .
note that we explicitly allow the sequences @xmath115 to be constant while the @xmath76 s increase .
quite generally , if each @xmath79 is countable , then there exists a sequence of finite sets to which it contracts ; namely , enumerate @xmath116 , and let @xmath76 be the set of cordinates of @xmath117 for all @xmath118 .
understandably , our main result applies formally to such constructions , but does not yield any useful information .
[ prop : main ] let @xmath2 be a faithful contracting similar group , generated by its nucleus . if all terms @xmath80 of the universal contracting cover of @xmath2 satisfy the farrell - jones conjectures with wreathing , then @xmath2 satisfies the farrell - jones conjectures . in the self - similar case , set @xmath119 , and @xmath120 for all @xmath121 ; and finally @xmath122 . more generally , in the similar case , set @xmath123 and @xmath122 .
there is an natural homomorphism @xmath124 , which we prove to be an farrell - jones .
let @xmath125 be in the kernel of @xmath126 ; then , because @xmath50 is contracting , there is @xmath84 such that @xmath127 belongs to @xmath128 ; furthermore , the permutation is trivial because @xmath129 , and the entries in @xmath76 are trivial because @xmath80 contains relations of length @xmath78 in @xmath76 .
therefore @xmath130 so @xmath131 , as was to be shown
. we then have @xmath132 , and because the farrell - jones conjectures are stable under colimits it suffices to see that @xmath133 satisfies the farrell - jones conjectures . by the first farrell - jones theorem
, @xmath133 is a subgroup of @xmath134 , so it suffices to show that @xmath134 satisfies the farrell - jones conjectures . since @xmath80 satisfies the farrell - jones conjectures with wreathing , we are done .
as stated in the introduction , proposition [ prop : main ] applies in particular to contracting similar groups whose universal contracting cover are cat(0 ) groups , lattices in virtually connected lie groups , or arithmetic groups over algebraic function fields .
we now give some examples of contracting , similar groups , recall some of their basic properties , and show that they satisfy the farrell - jones conjectures .
we follow a slightly unorthodox path to define ( self-)similar groups : we first give their contracting covers , and then simply say that the group itself is the faithful quotient of the cover .
this , of course , defines uniquely the self - similar group @xmath2 in question : it is the quotient of its universal contracting cover @xmath50 by the normal subgroup @xmath135 .
we denote by @xmath136 an element of the wreath product @xmath22 , with @xmath137 written as a product of disjoint cycles .
the alshin - grigorchuk group is obtained as follows .
set @xmath138 and define @xmath139 by @xmath140 let @xmath2 be the faithful self - similar quotient of @xmath50 . this group ( up to finite index ) was first considered in @xcite , providing a `` tangible '' example of infinite , finitely generated , torsion group ( the first examples of groups with these properties are due to golod @xcite ) .
grigorchuk proved in @xcite that its word growth is strictly between polynomial and exponential , and in @xcite that it is amenable , but not elementary amenable .
it is contracting , with nucleus @xmath141 .
since @xmath50 is cat(0 ) , as a free product of finite groups , @xmath2 satisfies the farrell - jones conjectures by proposition [ prop : main ] .
more elaborate examples have also been constructed by grigorchuk @xcite .
fix an infinite sequence @xmath142 of epimorphisms @xmath143 , and assume that @xmath144 contains infinitely many of each of the three possible epimorphisms .
define homomorphisms @xmath145 for all @xmath121 by @xmath146 let @xmath147 be the faithful similar quotient of @xmath50 using this similarity structure .
again , @xmath147 is contracting with nucleus @xmath148 for all @xmath84 , so all such groups satisfy the farrell - jones conjectures .
there are uncountably many such groups , and they all are torsion @xmath149-groups of intermediate word growth .
the gupta - sidki groups are obtained as follows .
choose a prime @xmath150 , set @xmath151 and define @xmath152 by @xmath153 let @xmath2 be the faithful self - similar quotient of @xmath50 .
these groups are shown in @xcite to be infinite , finitely - generated torsion @xmath154-groups .
since @xmath50 is cat(0 ) , as a free product of finite groups , @xmath2 satisfies the farrell - jones conjectures by proposition [ prop : main ] .
assume that @xmath2 is a self - similar group , and that , for every @xmath39 , there exists a bound @xmath155 such that , for all @xmath84 , there are at most @xmath156 non - trivial entries in @xmath127 .
note that it suffices to check this property for the generators of @xmath2 ; and that it holds for the generators of the grigorchuk group with @xmath157 , and those of the gupta - sidki groups for @xmath158 .
it is then known ( see @xcite ) that @xmath2 is contracting .
more precisely , @xmath2 is isomorphic to a subgroup of a self - similar group of very special type ( see @xcite ) .
fix an integer @xmath159 , set @xmath160 and define @xmath49 by @xmath161 since @xmath50 is cat(0 ) , as a free product of finite groups , @xmath2 satisfies the farrell - jones conjectures by proposition [ prop : main ] .
note that the faithful quotient of @xmath50 is amenable ; this is how @xcite show that all bounded self - similar groups are amenable .
let @xmath162 be a branched covering of a topological space @xmath163 ; this means there is an open dense subset @xmath164 and a covering @xmath165 .
we assume @xmath162 has finite degree @xmath25 .
let @xmath166 denote the _ post - critical locus _ of @xmath162 : @xmath167 assume finally that @xmath168 is path - connected .
choose a basepoint @xmath169 , and for each @xmath170 choose an arc @xmath171 from @xmath169 to @xmath172 in @xmath173 .
number also @xmath174 as @xmath175 .
these data define a self - similar group as follows .
it is again defined via a cover , @xmath176 .
consider @xmath177 . for each @xmath178 ,
let @xmath179 denote the unique @xmath162-lift of @xmath180 that starts at @xmath181 , and let it end at @xmath182 .
define then @xmath49 by @xmath183 if @xmath163 is in fact a locally simply connected metric space , and @xmath162 is uniformly expanding ( meaning there exists @xmath184 such that @xmath185 whenever @xmath186 is sufficiently small ) , then @xmath50 is contracting .
this applies in particular to @xmath163 a complex manifold and @xmath162 a holomorphic map ( which is then expanding for the kobayashi metric ) . the special case @xmath187 and @xmath162 a degree-@xmath149 polynomial has been extensively studied in @xcite .
the cover @xmath50 is a free group , so this provides more examples of groups satisfying the farrell - jones conjectures .
one important such example , associated with the map @xmath188 , has been studied in @xcite and @xcite ; it is amenable , orderable , of exponential growth , and residually poly-@xmath189 .
other examples , on higher - dimensional manifolds , have been considered by koch et al .
@xcite . there
, the universal contracting cover is the sphere braid group .
we have shown that if a counter - example to the farrell - jones conjectures exists in the class of ( self-)similar groups , it will not be an easy matter to establish that fact .
for another , calculations in a contracting self - similar groups are usually reduced to calculations in a finitely presented group , in which one may manipulate words .
it would be surprising that the farrell - jones conjectures fail for self - similar group , yet be unsettled for its cover .
since the farrell - jones conjectures is not settled for the sphere braid group , we have , at the present , no argument to check the farrell - jones conjectures on the faithful self - similar quotient of the braid groups that arise in this manner .
let @xmath2 be a self - similar group , and let @xmath190 be given .
assume that , for every @xmath39 , there exists a bound @xmath155 such that , for all @xmath84 , there are at most @xmath191 non - trivial entries in @xmath127 .
then @xmath2 is said to be of _ polynomial activity growth _ of degree @xmath192 ; see @xcite , who proves that such groups do not contain free subgroups .
it is then known @xcite that @xmath2 embeds , possibly for larger @xmath25 , in a specific group @xmath193 of polynomial activity growth , defined by its cover as follows . set @xmath194 and @xmath195 for @xmath196 ; set @xmath197 and define @xmath49 by @xmath198 these are non - contracting self - similar groups if @xmath199 ; for @xmath200 , the faithful quotient is amenable @xcite , while amenability of the faithful quotient is open for larger @xmath192 .
the arguments in @xcite show that the nucleus @xmath41 of @xmath193 , while infinite , admits a partial well ordering , such that every @xmath201 has the form @xmath202 or @xmath203 with @xmath204 for all @xmath205 .
presumably this means that arguments similar to those given here show that @xmath193 , and therefore all its subgroups , satisfy the farrell - jones conjectures . at the other extreme of contracting self - similar groups
bireversible groups_. these are self - similar groups @xmath206 such that the map @xmath207 , given by @xmath208 if @xmath209 , is a bijection .
they are related to the infinite simple groups constructed in @xcite .
they would seem like a natural class in which to look at counterexamples , though all examples studied up to now are lattices in virtually connected lie groups . | we show that contracting self - similar groups satisfy the farrell - jones conjectures as soon as their universal contracting cover is non - positively curved .
this applies in particular to bounded self - similar groups .
we define , along the way , a general notion of contraction for groups acting on a rooted tree in a not necessarily self - similar manner . |
the chemostat model provides the foundation for much of current research in bio - engineering , ecology , and population biology @xcite . in the engineering literature ,
the chemostat is known as the continuously stirred tank reactor .
it has been used for modeling the dynamics of interacting organisms in waste water treatment plants , lakes and oceans . in its basic setting , it describes the dynamics of species competing for one or more limiting nutrients .
if there are @xmath0 species with concentrations @xmath1 for @xmath2 and just one limiting nutrient with concentration @xmath3 and dilution rate @xmath4 , then the model takes the form @xmath5 where @xmath6 denotes the per capita growth rate of species @xmath7 and @xmath8 is the time derivative of any variable @xmath9 .
( in much of the paper , we simplify our notation by omitting the arguments of the functions .
for instance , when no confusion can arise from the context , we denote @xmath10 simply by @xmath3 . ) the functions @xmath6 depend only on the nutrient concentration , and are zero at zero , continuously differentiable and strictly increasing , although non - monotone functions have been the subject of research as well .
the conversion of nutrient into new biomass for each species @xmath7 happens with a certain yield @xmath11 and the natural control variables in this model are the input nutrient concentration @xmath12 and the dilution rate @xmath13 .
the latter variable is defined as the ratio of the volumetric flow rate @xmath14 ( with units of volume over time ) and the reactor volume @xmath15 which is kept constant .
therefore it is proportional to the speed of the pump that supplies the reactor with fresh medium containing the nutrient .
the equations ( [ model - full ] ) are then straightforwardly obtained from writing the mass - balance equations for the total amounts of the nutrient and each of the species , assuming the reactor content is well - mixed .
the full model ( [ model - full ] ) is illustrated in figure [ chem ] . in the present work ,
we consider the case where there is just one species with concentration @xmath16 , in which case the equations ( [ model - full ] ) take the form @xmath17 ( but see theorem [ iss - track ] below for results on chemostats with disturbances , and section [ several ] for models involving several species ) .
we assume @xmath12 is a given positive constant , while the per capita growth rate @xmath18 is a monod function ( which is also known as a michaelis - menten function ) taking the form @xmath19 for certain positive constants @xmath20 and @xmath21 that we specify later .
the dilution rate is an appropriate continuous positive periodic function we also specify below . since @xmath22
when @xmath23 is near zero , one can readily check that ( [ model - reduced ] ) leaves the domain of interest @xmath24 positively invariant ( i.e. , trajectories for ( [ model - reduced ] ) starting in @xmath25 remain in @xmath25 for all future times ) ; see theorem [ iss - track ] for a more general invariance result for perturbed chemostats .
since we are taking @xmath12 to be a fixed positive constant , we rescale the variables to reduce the number of parameters . using the change of variables @xmath26 and dropping bars , we eliminate @xmath12 and @xmath27 and so obtain the new dynamics @xmath28 again evolving on the state space @xmath29 . . in the next section ,
we briefly review the literature focusing on what makes our approach different . in section
[ track ] , we fix the reference signal we wish to track . in section [ define ] , we precisely formulate the definitions and the stability problem we are solving .
we state our main stability theorem in section [ thm ] and we discuss the significance of our theorem in section [ discuss ] .
we prove our stability result in section [ main ] . in section
[ several ] , we show that the stability is maintained when there are additional species that are being driven to extinction .
we validate our results in section [ simulations ] using a numerical example .
we conclude in section [ concl ] by summarizing our findings .
the behavior of the system @xmath30 is well understood when @xmath12 and @xmath13 are positive constants , as well as cases where @xmath31 and either of these control variables is held fixed while the other is periodically time - varying .
see @xcite for periodic variation of @xmath12 and @xcite for periodic variation of @xmath13 and the general reference @xcite on chemostats . when both @xmath12 and @xmath13 are constants , the so - called `` competitive exclusion principle '' holds , meaning that at most one species survives .
mathematically this translates into the statement that system @xmath30 has a steady state with at most one nonzero species concentration , which attracts almost all solutions ; see @xcite .
this result has triggered much research to explain the discrepancy between the ( theoretical ) competitive exclusion principle and the observation that in real ecological systems , many species coexist .
the results on the periodically - varying chemostat mentioned above should be seen as attempts to explain this paradox .
they involve chemostats with @xmath31 species , and their purpose is to show that an appropriate periodic forcing for either @xmath32 or @xmath33 can make the species coexist , usually in the form of a ( positive ) periodic solution .
few results on coexistence of @xmath34 species are available .
an exception is @xcite , where a periodic function @xmath32 is designed ( with @xmath13 kept fixed ) so that the resulting system has a ( positive ) periodic solution with an arbitrary number of coexisting periodically varying species .
the stability properties of this solution are not known .
more recent work has explored the use of state - dependent but time invariant feedback control of the dilution rate @xmath13 to generate coexistence ; see @xcite for monotone growth rate functions in the @xmath31 species case , and @xcite for the @xmath35 species case .
the paper @xcite considers feedback control when the growth rate functions are non - monotone . in @xcite , @xcite , and @xcite ,
coexistence is proved for models taking into account intra - specific competition . in these models ,
the usual growth functions @xmath36 are replaced by functions @xmath37 which are decreasing with respect to the variable @xmath1 .
all the results discussed so far apply to a more general model than @xmath38 involving @xmath39 species .
this is because the main purpose of these papers is to investigate environmental conditions under which the competitive exclusion principle fails and several species can coexist . here
we will not consider any coexistence problems .
our main objective is to provide a proof of stability of a periodic solution based on a lyapunov - type analysis and to investigate the robustness properties of the periodic solution with respect to perturbations .
as an illustration we show that the stability of the periodic solution is robust with respect to additional species that are being driven to extinction , or to small disturbances on the initial nutrient concentration or dilution rate .
these features set our work apart from the known results on periodically forced chemostat models which do not rely on the construction of a lyapunov function .
proving stability in the chemostat usually relies on reduction and monotonicity arguments , and not so often on lyapunov functions ( but see for instance theorem @xmath40 in @xcite which uses a lyapunov function introduced in @xcite and more recently @xcite ) .
finally we point out that closely related to our results is @xcite where a single - species chemostat with a continuous and bounded ( but otherwise arbitrary ) function @xmath32 and constant dilution rate is investigated ; there it is shown that two positive solutions converge to each other . however , the proof is not based on a lyapunov function .
the advantage of having a lyapunov function is that it can be used to _ quantify _ the effect of additional noise terms on the stability of the unperturbed dynamics .
in fact , to our knowledge , our work provides the first input - to - state stability analysis of chemostats whose dilution rates and initial concentrations are perturbed by small noise ; see remark [ aboutiss ] for a discussion on the importance of input - to - state stability in control theory and engineering applications .
we first choose the dilution rate @xmath41 that will give a reference trajectory @xmath42 for ( [ model ] ) which we show to be stable .
we assume a growth rate with constants @xmath43 as @xmath44 @xmath45 which we refer to as a _ reference trajectory _
when we choose @xmath46 condition ( [ mbound ] ) then provides constants @xmath47 such that @xmath48 for all @xmath49 : @xmath50 see figure [ refff ] for the graph of @xmath33 for @xmath51 and @xmath52 .
we wish to solve the following stability problem : * * * given any trajectory @xmath53 for ( [ model ] ) corresponding to the dilution rate @xmath33 from ( [ chli ] ) and @xmath18 as in ( [ mbound ] ) ( i.e. for any initial value for @xmath54 ) , show that the corresponding deviation @xmath55 of @xmath54 from the reference trajectory ( [ reftraj ] ) asymptotically approaches @xmath56 as @xmath57 .
we will solve ( sp ) by proving a far more general tracking result for a single species chemostat acted on by a disturbance vector @xmath58 as follows : @xmath59(1+u_2(t)-s(t))-\mu(s(t))x(t)\\[.5em ] \dot x(t)&= & x(t)[\mu(s(t))-d(t)-u_1(t ) ] \end{array } \right .. \ ] ] we will quantify the extent to which the reference trajectory ( [ reftraj ] ) tracks the trajectories of ( [ mod1 ] ) . to this end , we need to introduce a priori bounds on @xmath60 and @xmath61 ; see remark [ bound - u ] .
our main theoretical tool will be the input - to - state stability ( iss ) property @xcite which is one of the central paradigms of current research in nonlinear stability analysis ; see remark [ aboutiss ] .
the relevant definitions are as follows .
we let @xmath62 denote the set of all continuous functions @xmath63 for which ( i ) @xmath64 and ( ii ) @xmath27 is strictly increasing and unbounded .
we let @xmath65 denote the class of all continuous functions @xmath66 for which ( i ) @xmath67 for each @xmath68 , ( ii ) @xmath69 is non - increasing for each @xmath70 , and ( iii ) @xmath71 as @xmath57 for each @xmath70 .
consider a general control - affine dynamic @xmath72 evolving on a given subset @xmath73 where @xmath74 is a given subset of euclidean space .
( later we specialize to dynamics for the chemostat . ) for each @xmath75 and @xmath76 , let @xmath77 denote the solution of ( [ gen ] ) satisfying @xmath78 for a given control function @xmath79 ; i.e. the solution of the initial value problem @xmath80 we always assume that such solutions are uniquely defined on all of @xmath81 ( i.e. , ( [ gen ] ) is forward complete and @xmath82 is positively invariant for this system ) and that there exists @xmath83 such that @xmath84 everywhere , where @xmath85 is the usual euclidean norm .
for example , @xmath86 is the solution of ( [ mod1 ] ) for the disturbance @xmath87 satisfying the initial condition @xmath88 .
[ issdef ] we call ( [ gen ] ) _ input - to - state stable ( iss ) _ provided there exist @xmath89 and @xmath90 such that @xmath91 for all @xmath92 , @xmath75 , @xmath76 , and @xmath93 .
here @xmath94 denotes the essential supremum of @xmath93 . by causality , the iss condition ( [ issest ] ) is unchanged if @xmath94 is replaced by the essential supremum @xmath95}$ ] of @xmath96 restricted to @xmath97 $ ] . in particular , ( [ issest ] )
says @xmath98 as @xmath57 for all initial values @xmath99 and initial times @xmath100 , where @xmath101 is the zero disturbance @xmath102 .
[ aboutiss ] the theory of iss systems originated in @xcite .
iss theory provides the foundation for much current research in robustness analysis and controller design for nonlinear systems , and has also been used extensively in engineering and other applications @xcite .
the iss approach can be viewed as a unification of the operator approach of zames ( e.g. @xcite ) and the lyapunov state space approach .
the operator approach involves studying the mapping @xmath103 of initial data and control functions into appropriate spaces of trajectories , and it has the advantages that it allows the use of hilbert or banach space techniques to generalize many properties of linear systems to nonlinear dynamics .
by contrast , the state space approach is well suited to nonlinear dynamics and lends itself to the use of topological or geometric ideas .
the iss framework has the advantages of both of these approaches including an equivalent characterization in terms of the existence of suitable lyapunov - like functions ; see remark [ pba ] below . for a comprehensive survey on many recent advances in iss theory including its extension to systems with outputs , see @xcite . to specify the bound @xmath104 on our disturbances @xmath105 ,
we use the following constants whose formulas will be justified by the proof of our main stability result : @xmath106
from now on , we assume the disturbance vector @xmath105 in ( [ mod1 ] ) takes all of its values in a fixed square control set of the form @xmath107 0 < \bar u\ ; < \ ; \min\left\{\dfrac{c_1}{\sqrt { 8(1 + 2c\kappa c_1 ) } } , \dfrac{d_o}{2}\right\ } \end{array}\ ] ] where @xmath108 , @xmath109 , and @xmath110 are in ( [ ck ] ) ( but see remark [ bound - u ] for related results under less stringent conditions on the disturbance values ) .
we will prove the following robustness result : [ iss - track ] choose @xmath33 , @xmath18 , and @xmath111 as in ( [ mbound])-([chli ] ) .
then the corresponding solutions of ( [ mod1 ] ) satisfy @xmath112\\[.25em]\ ; \ ; \ ; \rightarrow\ ; \ ; \ ; \left[\ ; s(t ; t_0 , ( s_0,x_0 ) , \alpha ) > 0 \ ; \ ; \ & \ ; \ ; x(t ; t_0 , ( s_0,x_0 ) , \alpha)>0\ ; \right]\ ; \ ; .\end{array}\ ] ] moreover , there exist @xmath89 and @xmath90 such that the corresponding transformed error vector @xmath113 \left(s(t ; t_0 , ( s_0,x_0 ) , \alpha ) - s_r(t ) , \ln(x(t ; t_0 , ( s_0,x_0 ) , \alpha ) ) - \ln(x_r(t))\right)\end{array}\ ] ] satisfies the iss estimate ( [ issest ] ) for all @xmath93 , @xmath114 , @xmath115 , @xmath116 , and @xmath117 , where @xmath118 .
before proving the theorem , we discuss the motivations for its assumptions , and we interpret its conclusions from both the control theoretic and biological viewpoints .
[ invariance ] condition ( [ c1ru1 ] ) says @xmath119 is positively invariant for ( [ mod1 ] ) .
one may also prove that @xmath120 is positively invariant for ( [ mod1 ] ) , as follows .
suppose the contrary .
fix @xmath75 , @xmath121 , @xmath122 , and @xmath123 for which the corresponding trajectory @xmath124 for ( [ mod1 ] ) satisfying @xmath125 exits @xmath120 in finite time .
this provides a finite constant @xmath126\}$ ] .
then @xmath127 , since otherwise @xmath128 and @xmath129 for all @xmath130 and then we could use the continuity of @xmath3 to contradict the maximality of @xmath131 . since @xmath132 , the continuity of @xmath3 and @xmath16 and the fact that @xmath127 provide a constant @xmath133 such that @xmath134 and @xmath135 for ( almost ) all @xmath136 $ ] , hence also @xmath137 for all @xmath138 $ ] ( since @xmath139 everywhere ) .
hence , @xmath140 for all @xmath141 $ ] .
since @xmath142 clearly stays in @xmath143 , this contradicts the maximality of @xmath131 .
the positive invariance of @xmath120 follows .
[ means ] theorem [ iss - track ] says that in terms of the error signals @xmath144 , any componentwise positive trajectory of the unperturbed chemostat dynamics ( [ mod1 ] ) converges to the nominal trajectory ( [ reftraj ] ) , uniformly with respect to initial conditions .
this corresponds to putting @xmath145 in ( [ issest ] ) .
it also provides the additional desirable robustness property that for an arbitrary @xmath146-valued control function @xmath123 , the trajectories of the _ perturbed _ chemostat dynamics ( [ mod1 ] ) are `` not far '' from ( [ reftraj ] ) for large values of time . in other words , they `` almost '' track ( [ reftraj ] ) with a small overflow @xmath147 from the iss inequality ( [ issest ] ) .
similar results can be shown for general choices of @xmath148 and @xmath33 .
for example , we can choose any @xmath149 that admits a constant @xmath150 such that @xmath151 for all @xmath68 and @xmath152 .
in this case , we take the dilution rate @xmath153 which is again uniformly bounded above and below by positive constants .
the proof of this more general result is similar to the proof of theorem [ iss - track ] we give below except with different choices of the constants @xmath108 and @xmath109 .
[ mfre1 ] the robustness result @xmath154\le \beta(|(s_0,x_0)|,t - t_0 ) + \gamma(|\alpha|_{[t_0,t]})\end{array}\ ] ] of theorem [ iss - track ] differs from the classical iss condition in the following ways : 1 . for biological reasons , negative values of the nutrient level @xmath3 and the species level @xmath16 do not make physical sense
hence , only componentwise positive solutions are of interest . therefore ,
( [ cru1 ] ) is not valid for all @xmath155 but rather only for @xmath156 .
2 . our condition ( [ cru1 ] )
provides an estimate on the _ transformed _ error component @xmath157 instead of the more standard error @xmath158 .
our reasons for using the transformed form of the error are as follows .
the function @xmath159 goes to @xmath160 when @xmath16 goes to zero .
this property is relevant from a biological point of view .
indeed , in the study of biological systems , it is important to know if the concentration of the species is above a strictly positive constant when the time is sufficiently large or if the concentration admits zero in its omega limit set . in the first case ,
the species is called _
persistent_. the persistency property is frequently desirable , and it is essential to know whether it is satisfied .
hence , the function @xmath161 has the desirable properties that ( a ) it goes to @xmath162 if @xmath163 does , ( b ) it is equal to zero when @xmath164 is equal at time @xmath165 to the value of @xmath148 , and ( c ) it goes to @xmath166 if @xmath164 goes to zero .
therefore , roughly speaking , if the species faces extinction , then it warns us .
[ pba ] our proof of theorem [ iss - track ] is based on a lyapunov type analysis .
recall that a @xmath167 function @xmath168 is called an _ iss lyapunov function ( iss - lf ) _ for ( [ gen ] ) provided there exist @xmath169 such that 1 .
@xmath170 and 2 . @xmath171\le -\gamma_3(|y|)+\gamma_4(|u|)$ ] hold for all @xmath172 , @xmath68 , and @xmath173 .
the function @xmath174 we will construct in the proof of theorem [ iss - track ] is not an iss - lf for the chemostat error dynamics because of the specificities of the state space , which preclude the existence of the necessary functions @xmath175 in condition 1 above .
hence , we can not directly apply the result that the existence of an iss lyapunov function implies that the system is iss e.g. ( * ? ? ? * theorem 1 ) to prove our theorem .
instead , we prove our theorem [ iss - track ] directly from the decay inequality satisfied by the time derivative of @xmath174 along the trajectories .
the proof that the decay inequality implies iss is very similar to that part of the proof of ( * ? ? ?
* theorem 1 ) , so we only sketch that part of our proof in the appendix .
[ bound - u ] our estimate ( [ cru1 ] ) would not hold if we had instead chosen the full control set @xmath176 .
in fact , taking the disturbance @xmath177 and any initial condition @xmath178 in ( [ mod1 ] ) would give @xmath179 and so also @xmath180 as @xmath181 ( since @xmath182 almost everywhere ) .
therefore extinction would occur and ( [ cru1 ] ) would not be satisfied .
on the other hand , if our set @xmath146 is replaced by @xmath183 ^ 2 $ ] for any fixed constant @xmath184 , then the chemostat error dynamics instead satisfies the less stringent _ integral _ iss property ; see remark [ iiss - rk ] below for details .
the proof of ( [ c1ru1 ] ) is immediate from the structure of the dynamics ( [ mod1 ] ) and the fact that @xmath185 ( which imply that @xmath22 when @xmath186 is sufficiently small ) ; see remark [ invariance ] for a similar argument . it remains to prove the iss estimate ( [ cru1 ] ) for suitable functions @xmath89 and @xmath90 . throughout the proof , all ( in)equalities should be understood to hold globally unless otherwise indicated .
also , we repeatedly use the simple `` ( generalized ) triangle inequality '' relation @xmath187 for various choices of @xmath188 , @xmath189 , and @xmath190 that we specify later .
fix @xmath75 , @xmath191 , @xmath192 , and @xmath123 , and let @xmath193 denote the corresponding solution of ( [ mod1 ] ) satisfying @xmath194 . for simplicity , we write @xmath195 as @xmath196 , omitting the time argument as before .
we first write the error equation for the variables @xmath197 where @xmath198 , @xmath199 , @xmath200 , and @xmath201 .
one easily checks that @xmath202[1+u_2(t)-z(t)]\\ \dot x(t)&= & x(t)[\mu(z(t)-x(t))-d(t)-u_1(t)]\ , .
\end{array}\ ] ] therefore , since @xmath203 ( which implies @xmath204[1-z_r(t)]$ ] ) , our formula ( [ mbound ] ) for @xmath18 immediately implies that the ( transformed ) error @xmath205 satisfies the _ chemostat error dynamics _ @xmath206 [ \tilde z - u_2(t ) ] \\[.5em ] \dot{\tilde{\xi } } & = & ma\dfrac{\tilde{z } - e^{\xi_r(t)}(e^{\tilde{\xi } } - 1)}{(a + z - e^{\xi})(a +
z_r(t ) - e^{\xi_r(t ) } ) } - u_1(t ) . \end{array } \right.\ ] ] we are going to show that ( [ dis ] ) has the lyapunov function @xmath207 and @xmath208 is the constant defined in ( [ ck ] ) . from the explicit expressions @xmath209 and @xmath210 , we deduce that the time derivative of @xmath211 along trajectories of ( [ dis ] ) satisfies @xmath212 where we also used the fact that @xmath213 . since @xmath214 everywhere , one readily checks that along the trajectories of ( [ dis ] ) , @xmath215 where @xmath216 , the constant @xmath108 is defined by ( [ ck ] ) , and the last inequality used ( [ triangle ] ) with the choices @xmath217 , @xmath218 , and @xmath219 .
the fact that @xmath220 easily follows because @xmath221 . along the trajectories of ( [ dis ] ) ,
@xmath222 we distinguish between two cases . _
: @xmath223_. then since @xmath224 , we get @xmath225 using the triangle inequality ( [ triangle ] ) with the choices @xmath226 we deduce from ( [ s2h ] ) that @xmath227 & & - \left[\dfrac{\kappa}{2 } - \dfrac{8(4a + 3)(a + 2)m}{a ( 4 a + 1)^2}\right ] \tilde z^2 - ( e^{\tilde{\xi } } - 1 ) u_1 + \kappa c u^2_2 .
\end{array}\ ] ] _ : @xmath228_. since @xmath229 , it follows that @xmath230 .
therefore , since @xmath231 , @xmath232 since @xmath228 and @xmath233 , we have @xmath234 . as @xmath235 , condition ( [ a4h ] ) gives @xmath236 from this last inequality and the inequality @xmath237 , we deduce from dropping the first term in ( [ s2mple ] ) that @xmath238 & \leq & - \dfrac{1}{200}\kappa ( e^{\tilde \xi } - 1)^2 - \left[\dfrac{1}{4 } \kappa - \dfrac{28 m}{4 a + 1}\right ] \tilde z^2 - ( e^{\tilde{\xi } } - 1 ) u_1 + \kappa c u^2_2 .
\end{array}\ ] ] we deduce from our choice ( [ ck ] ) of @xmath109 , ( [ s21 ] ) , and ( [ u2upue ] ) that in cases 1a-2a , @xmath239 - ( e^{\tilde{\xi } } - 1 ) u_1 + \kappa c u^2_2 \end{array}\ ] ] where @xmath110 is defined in ( [ ck ] ) . using ( [ triangle ] ) with @xmath240 , @xmath241 , and @xmath242 , and then the upper bounds of @xmath60 and @xmath61 , we deduce from ( [ a2upue ] ) that @xmath243 + \dfrac{1}{2 c_1 } \bar u |u_1| + \kappa c \bar u |u_2| \\ & \leq & - \dfrac{c_1}{2 } [ ( e^{\tilde \xi } - 1)^2 + \tilde z^2 ] + c_2 |u|,\ ; \ ; \ ; { \rm where } \ ; \ ; \ ; c_2 : = \left(\dfrac{1}{c_1 } + 2\kappa c\right ) \bar u \end{array}\ ] ] and where the last inequality used the relationship @xmath244 between the @xmath245-norm and the @xmath246-norm .
we consider two additional cases . :
then ( [ a2uiue ] ) gives @xmath248 next notice that ( [ constraint ] ) and our choice of @xmath249 $ ] give @xmath250 : @xmath251 .
then @xmath252 is in a suitable bounded set , so since @xmath253 are locally bounded when defined to be zero at @xmath254 , one can readily use ( [ a2uiue ] ) to compute constants @xmath255 and @xmath256 such that @xmath257 where @xmath258 was used to get @xmath259 and @xmath260 was used to get @xmath261 .
it follows from ( [ s2dodo])-([een ] ) that , in cases 1b-2b , @xmath262 with @xmath263 .
condition ( [ goal ] ) is the classical iss lyapunov function decay condition for the transformed error dynamics evolving on our restricted state space .
therefore , a slight variant of the classical iss arguments combined with ( [ goal ] ) give the iss estimate asserted by theorem [ iss - track ] . for details , see the appendix below .
this concludes the proof .
[ iiss - rk ] if our control set @xmath146 is replaced by the larger control set @xmath183 ^ 2 $ ] for any fixed constant @xmath184 , then the error dynamics ( [ dis ] ) instead satisfies the less stringent _ integral _ iss property .
the relevant definitions are as follows .
we say that ( [ gen ] ) is _ integral input - to - state stable ( iiss ) _ provided there exist @xmath264 and @xmath89 such that @xmath265 everywhere for all measurable essentially bounded functions @xmath266 .
this condition is less restrictive than iss since e.g. @xmath267 is iiss but not iss @xcite .
iiss - lf _ for ( [ gen ] ) ( with controls in @xmath268 ) is then defined to be a @xmath167 function @xmath168 for which there are @xmath269 and a positive definite function @xmath270 ( i.e. , @xmath271 is continuous and zero only at zero ) such that conditions 1 - 2 in remark [ pba ] hold everywhere .
this is less restrictive than the iss - lf condition since @xmath270 need not be of class @xmath272 . arguing as in the proof of theorem [ iss - track ] up through ( [ a2uiue ] ) and solving the appropriate constrained minimum problem to get @xmath270 shows that @xmath273 satisfies the iiss lyapunov function decay condition ( namely , condition 2 from remark [ pba ] ) for the error dynamics ( [ dis ] ) and the control set @xmath268 using @xmath274 .
therefore , this system is in fact iiss , by a slight variant of the proof of the iiss estimate in ( * ? ? ?
* theorem 1 ) .
we leave the details to the reader .
theorem [ iss - track ] shows that the stability of the reference trajectory ( [ reftraj ] ) is robust with respect to small perturbations of the dilution rate and initial concentration . to further demonstrate the robustness of our results
, we next show that the stability of ( [ reftraj ] ) is also maintained when the model ( [ model ] ) is augmented to include additional species that are being driven to extinction , in the following sense .
we assume for simplicity that @xmath275 .
consider the augmented system @xmath276 { \dot y_i } & = & y_i(\nu_i(s ) - d(t)),\;\ ; i=1,\dots , n , \end{array}\right.\ ] ] where @xmath18 is as in ( [ mbound ] ) and @xmath277 is continuous and increasing and satisfies @xmath278 for @xmath279 .
the variables @xmath280 represent the levels of @xmath0 additional species .
we choose @xmath13 and @xmath281 as in ( [ chli ] ) and ( [ aer1 ] ) , and we assume @xmath282 for @xmath279 .
( this assumption is , in a sense , natural because one can easily check that it ensures that each species concentration @xmath280 converges to zero . indeed ,
the fact that , for all @xmath283 , @xmath284 and @xmath285 ensures , in combination with the inequalities @xmath282 , that there exists an instant @xmath286 and a constant @xmath287 such that for all @xmath288 and for @xmath279 , @xmath289 .
this implies that @xmath280 s converge to @xmath290 exponentially . )
we show that the transformed error @xmath291 between any componentwise positive solution @xmath292 of ( [ nmodel ] ) and the reference trajectory @xmath293 converges exponentially to the zero vector as @xmath57 . to this
end , notice that in the coordinates ( [ nte ] ) , the system ( [ nmodel ] ) becomes @xmath294 by the same calculations that led to ( [ dis ] ) .
set @xmath295 where @xmath296 .
since @xmath297 , @xmath203 , @xmath298 , and @xmath299 , we deduce that the derivative of @xmath300 along the trajectories of ( [ dmodel ] ) satisfies @xmath301 & & -2\left[\dfrac{\sqrt{m}}{\sqrt{a}}{\tilde z}\right]\left[\dfrac{4\sqrt{m}}{\sqrt{a}{d_o } } \displaystyle \sum_{i=1}^n\nu_i(s)y_i\right ] \\[1em ] & \leq & - \dfrac{ma(e^{\tilde{\xi } } - 1)^2}{16(a+1 ) ( a + 2 + \tilde{z}^2 ) } - \dfrac{3m}{a}\tilde{z}^2 + \dfrac{16 m } { a d^2_o}\left[\displaystyle \sum_{i=1}^n\nu_i(s)y_i\right]^2 \end{array}\ ] ] where the second inequality is by ( [ triangle ] ) with @xmath302 , @xmath303 , and @xmath304 and the last inequality used the relation @xmath305 for real values @xmath306 and @xmath307 . on the other hand , since @xmath308 for each @xmath7 , the form of the dynamics for @xmath3 and the nonnegativity of @xmath18 and the @xmath277 s along our componentwise positive trajectories imply that there exist @xmath309 and @xmath310 such that ( i ) @xmath311 for all @xmath312 and ( ii ) @xmath313 for all @xmath279 .
we deduce that , for all @xmath279 and for all @xmath288 , @xmath314 where @xmath315 .
hence , each @xmath316 converges exponentially to zero .
next notice that along each pair @xmath252 , the function @xmath317 is positive if and only if @xmath318 . by ( [ dmo1 ] ) and ( [ exs1 ] ) , the time derivative of @xmath319 along the trajectories of ( [ dmodel ] )
satisfies @xmath320 ^ 2 - 2a\delta\displaystyle \sum_{i=1}^ny^2_i\\ & \le & - \delta(\tilde \xi,\tilde z ) - \dfrac{3m}{a}\tilde{z}^2 + \dfrac{16mn^2}{ad^2_o}\displaystyle \sum_{i=1}^n\nu^2_i(s)y^2_i-2a\delta\displaystyle \sum_{i=1}^ny^2_i\\&\le & - \delta(\tilde \xi,\tilde z ) -\dfrac{3m}{a}\tilde
z^2-\dfrac{16mn^2}{a}\displaystyle \sum_{i=1}^ny^2_i\ ; \ ; = : \ ; \ ; -m(\tilde \xi,\tilde z , y ) \end{array}\ ] ] provided @xmath321 where @xmath322 is chosen as above .
( the second inequality in ( [ surmise ] ) follows because for any nonnegative @xmath323 , we get @xmath324 which we sum and then square to get @xmath325 .
the last inequality used @xmath326 for all @xmath312 and our choice of @xmath327 . )
it is tempting to surmise from ( [ surmise ] ) and the structure of @xmath328 that @xmath328 is a lyapunov function for ( [ dmodel ] ) since then we could use standard lyapunov function theory to conclude that @xmath329 asymptotically converges to zero .
however , such an argument would not be technically correct , since the state space of ( [ dmodel ] ) is not @xmath330 ( because the original augmented chemostat model ( [ nmodel ] ) is only defined for componentwise nonnegative values of the state ) .
instead , we argue as follows ( in which we may assume for simplicity that the initial time for the augmented error dynamics is zero ) . for any @xmath283 , integrating the last inequality of ( [ surmise ] ) over
@xmath331 $ ] gives @xmath332 it follows that , for all @xmath283 , @xmath333 therefore @xmath334 is a bounded function .
similarly , @xmath335 and @xmath336 are bounded .
we deduce that @xmath337 , @xmath338 , and the components of @xmath144 are uniformly continuous , since their time derivatives ( [ dmodel ] ) are bounded . reapplying ( [ da2 ] ) therefore implies @xmath339 is finite .
it follows from barbalat s lemma @xcite and the structure of the function @xmath340 that @xmath341 as @xmath342 .
this establishes our stability condition for the multi - species model .
notice that @xmath343 is bounded below by a quadratic of the form @xmath344 along the trajectories of ( [ dmodel ] ) , and that @xmath328 is bounded above and below by such quadratics along the trajectories as well , since the trajectories are bounded . from this fact and ( [ surmise ] ) , one can deduce that the trajectories @xmath345 converge _ exponentially _ to zero .
to validate our convergence result , we simulated the dynamics ( [ mod1 ] ) with the initial values @xmath346 and @xmath347 and the reference trajectory @xmath149 , using the parameters @xmath51 and @xmath52 and @xmath348 . in this case ,
the lower bound on @xmath33 provided by ( [ aer1 ] ) is @xmath349 .
it follows from remark [ iiss - rk ] that the convergence of @xmath142 to @xmath149 is robust to disturbances that are valued in @xmath350 ^ 2 $ ] for any positive constant @xmath351 , in the sense of integral input - to - state stability .
moreover , using the estimate ( [ a2uiue ] ) , one easily checks that in this case , estimate ( iiss ) on p. above holds with @xmath352 ; cf .
the proof of ( * ? ? ? * theorem 1 ) . for our simulation
, we took the disturbance @xmath353 on the dilution rate and @xmath354 .
this gave the plot of @xmath142 and @xmath149 against time in figure [ xsol ] .
our simulation shows that the state trajectory @xmath142 closely tracks the reference trajectory @xmath149 even in the presence of small disturbances and so validates our findings .
the chemostat model is a useful framework for modeling species competing for nutrients . for the case of one species competing for one nutrient and a suitable time - varying dilution rate , we proved stability of an appropriate reference trajectory .
moreover , we found that the stability was maintained even if the model is augmented with other species that are being driven to extinction , or if there are disturbances of appropriately small magnitude acting on the dilution rate and input nutrient concentration .
for completeness , we provide the slight variant of the classical iss arguments needed to finish the proof of theorem [ iss - track ] . multiplying through ( [ goal ] ) by @xmath355 and applying the standard `` variation of parameters '' formula to @xmath97\ni l\mapsto v(\tilde \xi(l),\tilde z(l))$ ] ( by integrating between @xmath356 and @xmath357 ) gives @xmath358 } \ ; , \end{array}\ ] ] where we enlarged @xmath359 without relabeling .
we deduce that @xmath360}\right)\ ] ] where @xmath211 is defined in ( [ vchoice ] ) .
since @xmath361 and @xmath362 for all @xmath363 , we deduce from the formula for @xmath174 that @xmath364 in particular , @xmath365 . from ( [ 49 ] ) and the inequality @xmath366 , @xmath367 @xmath368 } } \end{array}\ ; .\ ] ] the relations @xmath369 and @xmath370 give @xmath371 } } \\ & & \hspace{-.5in}+ \frac{1}{2 } \left(e^{2\sqrt{2 e^{(t_0 - t ) c_5 } \omega\left(|(\tilde{\xi}(t_0),\tilde z(t_0))|\right ) } } - 1\right ) + \frac{1}{2 } \left(e^{2\sqrt{2 c_2 |u|_{[t_o , t ] } } } - 1\right ) .
\end{array}\ ] ] the desired iss estimate ( [ issest ] ) now follows immediately from this last inequality and ( [ tal ] ) with the choices @xmath372 this completes the proof of theorem [ iss - track ] .
part of this work was done while p. de leenheer and f. mazenc visited louisiana state university ( lsu ) .
they thank lsu for the kind hospitality they enjoyed during this period .
f. mazenc thanks claude lobry and alain rapaport for illuminating discussions .
malisoff was supported by nsf / dms grant 0424011 .
de leenheer was supported by nsf / dms grant 0500861 .
the authors thank the referees for their comments , and they thank hairui tu for helping with the graphics .
the second author thanks ilana aldor for stimulating discussions .
f. grognard , f. mazenc , and a. rapaport , polytopic lyapunov functions for the stability analysis of persistence of competing species .
proceedings of the 44th ieee confence on decision and control and european control conference ecc 2005 , seville , spain ( 2005 ) 3699 - 3704 .
[ also discrete and continuous dynamical systems - series b , to appear . ]
smith and p. waltman , the theory of the chemostat .
cambridge university press , cambridge , 1995 .
sontag , smooth stabilization implies coprime factorization .
ieee trans .
automatic control 34 ( 1989 ) 435 - 443 .
e. d. sontag , the iss philosophy as a unifying framework for stability - like behavior .
nonlinear control in the year 2000 , volume 2 , a. isidori , f. lamnabhi - lagarrigue , and w. respondek , eds . , lecture notes in control and information sciences vol .
259 , springer - verlag , berlin , 2000 , 443 - 468 .
g. zames , on the input - output stability of time - varying nonlinear feedback systems .
part i : conditions using concepts of loop gain , conicity , and positivity .
ieee trans .
automatic control 11 ( 1966 ) 228 - 238 .
g. zames , on the input - output stability of time - varying nonlinear feedback systems .
part ii : conditions involving circles in the frequency plane and sector nonlinearities .
. automatic control 11 ( 1966 ) 465 - 476 . | we study the chemostat model for one species competing for one nutrient using a lyapunov - type analysis .
we design the dilution rate function so that all solutions of the chemostat converge to a prescribed periodic solution . in terms of chemostat biology
, this means that no matter what positive initial levels for the species concentration and nutrient are selected , the long term species concentration and substrate levels closely approximate a prescribed oscillatory behavior .
this is significant because it reproduces the realistic ecological situation where the species and substrate concentrations oscillate .
we show that the stability is maintained when the model is augmented by additional species that are being driven to extinction .
we also give an input - to - state stability result for the chemostat tracking equations for cases where there are small perturbations acting on the dilution rate and initial concentration .
this means that the long term species concentration and substrate behavior enjoys a highly desirable robustness property , since it continues to approximate the prescribed oscillation up to a small error when there are small unexpected changes in the dilution rate function .
frdric mazenc projet mere inria - inra umr analyse des systmes et biomtrie inra 2 , pl .
viala , 34060 montpellier , france michael malisoff department of mathematics louisiana state university baton rouge , la 70803 - 4918 patrick de leenheer department of mathematics university of florida 411 little hall , po box 118105 gainesville , fl 326118105 ( communicated by ? ? ? ? ) |
models with radiative neutrino mass are of significant experimental interest .
the inherent loop - suppression in such models allows the new physics responsible for neutrino mass to be lighter than otherwise expected . such light new physics can be within experimental reach , either directly , through collider experiments , or indirectly , via e.g. searches for lepton flavor violating ( lfv ) effects .
the loop suppression becomes more severe as the number of loops increases .
thus , models with three - loop masses are particularly interesting , as they generically require new physics at or around the tev scale
. he we present a class of models with radiative neutrino mass at the three - loop level @xcite .
we focus primarily on the knt model @xcite and report recent analysis showing that the model satisfies lfv constraints , such as @xmath0 , and fits the neutrino oscillation data .
furthermore , the model contains a viable candidate for the dark matter ( dm ) in the universe , in the form of a light right - handed ( rh ) neutrino .
we also show that a strongly first order electroweak phase transition can be achieved with a higgs mass of @xmath1 gev , as measured at the lhc @xcite .
the model contains new charged scalars and we discuss their effect on the one - loop higgs decay to neutral gauge bosons .
afterwards , we show that the knt model belongs to a larger class of related three - loop models and briefly outline their features .
the knt model @xcite is an extension of the sm with three rh neutrinos , @xmath2 , and two electrically charged scalars , @xmath3 and @xmath4 , that are singlet under the @xmath5 gauge group .
in addition , a discrete @xmath6 symmetry is imposed on the model , under which @xmath7 , and all other fields are even .
this symmetry plays two key roles , preventing a tree - level coupling between @xmath8 and the sm higgs , which would otherwise induce tree - level neutrino masses , and ensuring that the lightest fermion @xmath9 is a stable dm candidate .
the lagrangian reads @xmath10 where @xmath11 is the left - handed lepton doublet , @xmath12 are yukawa couplings which are antisymmetric in the generation indices @xmath13 and @xmath14 , @xmath15 are the majorana rh neutrino masses , @xmath16 is the charge conjugation matrix , and @xmath17 is the tree - level scalar potential , which is given by @xmath18 here @xmath19 denotes the sm higgs doublet . the neutrino mass matrix elements , arising from the three - loop diagram in fig .
[ diag ] , are given by @xmath20 where @xmath21 are the charged leptons flavor indices , @xmath22 denotes the three rh neutrinos , and the function @xmath23 is a loop integral which is @xmath24 @xcite .
it is interesting to note that , unlike the conventional seesaw mechanism , the radiatively generated neutrino masses are directly proportional to the charged lepton and rh neutrino masses , as well as being loop - suppressed .
the lagrangian ( [ l ] ) induces flavor violating processes such as @xmath25 for @xmath26 , generated at one loop via the exchange of the charged scalars @xmath27 .
the branching ratio of such process is given by @xmath28 with @xmath29 , @xmath30 being the fine structure constant and @xmath31 .
for the case of @xmath32 , this leads to a new contribution to the muon anomalous magnetic moment , @xmath33 . + in our scan of the parameter space of the model we impose the experimental bound @xmath34 @xcite ; and use the allowed values for the neutrino mixing @xmath35 , @xmath36 , @xmath37 , and the mass squared difference @xmath38 @xmath39 and @xmath40 @xcite .
an immediate implication of the @xmath6 symmetry is that that lightest right handed neutrino , @xmath41 , is stable , and hence a candidate for dark matter ( dm ) .
the @xmath41 number density get depleted through the process @xmath42 via the @xmath43 and u channels exchange of @xmath4 . in the non - relativistic limit
, the total annihilation cross section is given by @xmath44 with @xmath45 is the relative velocity between the annihilation @xmath41 s .
the relic density after the decoupling of @xmath41 can be obtained by solving the boltzmann equation , and it is approximately given by @xmath46 where @xmath47 is the thermal average of the relative velocity squared of a pair of two @xmath41 particles , @xmath48 is planck mass ; and @xmath49 is the total number of effective massless degrees of freedom at the freeze - out temperature @xmath50**. * * in fig .
[ msmn ] , we plot the allowed mass range for @xmath51 that gives the observed dark matter relic density @xcite .
as seen in the figure , the neutrino experimental data , the bound on lfv process , combined with the relic density , seems to prefer @xmath52 for large regions of parameter space .
however , the masses of both the dm and the charged scalar @xmath4 are bounded from above , with @xmath53 and @xmath54 , respectively . * * [ om ] ( red ) and @xmath55 ( green ) versus the lightest rh neutrino mass , where the consistency with the neutrino data , lfv constraints and the dm relic density have been imposed.__,width=283,height=207 ] although the sm has all the qualitative ingredients for electroweak baryogenesis , the amount of matter - antimatter asymmetry generated is too small .
one of the reasons has to with the fact that the electroweak phase transition ( ewpt ) is not strongly first order , which is necessary to suppress the sphaleron processes in the broken phase .
the strength of the ewpt can be improved if there are new scalar degrees of freedom around the electroweak scale coupled to the sm higgs , which is the case in the knt model .
the investigation of the scalar effective potential reveals that , within the allowed parameter space of the model , the strength of the electroweak phase transition ( ewpt ) can be first order @xcite .
we find that if the one - loop corrections to the higgs mass are sizeable , then the condition @xmath56 for having a first order ewpt can be realized while keeping the higgs mass around @xmath57 gev .
the reason for this being that the extra charged singlets affect the dynamics of the sm scalar field vev around the critical temperature @xcite . in fig .
[ ct ] , we show the plot for @xmath58 versus the critical temperature and one observes that a strongly first order ewpt is possible while the critical temperature lies around 100 @xmath59 .
.__,width=283,height=207 ] since the rh neutrinos couple to the charged leptons , one excepts them to be produced at @xmath60 colliders , such as the ilc and clic , with a collision energy @xmath61 of few hundreds of up to a tev
. if the produced pairs are the form @xmath62 or @xmath63 , then @xmath64 will decay into a charged lepton and @xmath4 , and subsequently @xmath4 decays into @xmath41 and a charged lepton .
in addition , the sm neutrinos will be also produced via the decay of the charged scalar @xmath65 .
here we concentrate on he process @xmath66
@xcite , where the missing energy corresponds to any state in the set @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 .
the total expected cross section of the processes @xmath74 is represented by @xmath75 , while @xmath76 denotes the cross section of different subprocesses . as a benchmark ,
we consider @xmath77 we use lanhep and calchep to simulate our model and generate the differential cross section and the relevant kinematic variables for different cm energy : @xmath78250 , 350 , 500 and1 , with unpolarized beams at first ; and then we consider the polarized beams with @xmath79 $ ] and/or @xmath80 $ ] . imposing the appropriate cuts given in table-[t1 ] , we show in fig .
[ svsl ] the dependance of the significance on the accumulated luminosity with and without polarized beams for the considered cm energies .
we clearly see that for a polarized beam , the signal can be observed even with relatively low integrated luminosity . for example , at @xmath81 , the 5 @xmath82 required luminosity is 150 @xmath83 for polarized beam as compared to 700 @xmath83 without polarization .
.__relevant cuts for the process @xmath84 for different cm energies . here
@xmath85 and @xmath86 are the charged lepton energy in its pseudo - rapidity , @xmath87 is the electron - muon invariant mass and @xmath88 is the missing invariant mass . _ _ [ cols="^,^",options="header " , ] ; with ( left ) and without ( right ) polarized beams .
the two horizontal dashed lines represent @xmath89 and @xmath90 , respectively.__,title="fig:",width=283,height=207 ] ; with ( left ) and without ( right ) polarized beams .
the two horizontal dashed lines represent @xmath89 and @xmath90 , respectively.__,title="fig:",width=283,height=207 ]
the knt model is a simple theory that is capable of simultaneously addressing the neutrino mass and dm problems .
it predicts new physics that can be probed at colliders and precision experiments . within the knt model ,
the dm is part of a @xmath91-odd sector that propagates in the inner loop of the neutrino mass diagram ( see figure [ diag ] ) .
one might wonder whether generalizations of the knt model exist .
it is well known that the seesaw mechanism can be generalized to a type - iii variant with @xmath92 triplet fermions @xcite , and a further variant with quintuplet fermions @xcite .
it is perhaps not surprising , therefore , to learn that the knt model forms part of a larger class of theories that similarly achieve neutrino mass at the three - loop level , with dm propagating in the inner loop .
the generalized models require that the sm be extended to include the charged singlet scalar @xmath93 , a new scalar @xmath94 , and three real fermions @xmath8 .
however , instead of taking @xmath94 and @xmath8 as @xmath95 singlets , the variant models employ larger multiplets .
the resulting models share key similarities with the knt model , with the neutrino mass diagram retaining the exact form in figure [ diag ] , and the dm remaining as the lightest fermion @xmath8 . however , there are also some interesting differences , relative to the knt model .
firstly , the dm is now part a larger multiplet with non - trivial gauge interactions .
this modifies expectations for the dm mass and allows new dm interactions .
secondly , there are interesting consequences for the @xmath91 symmetry , which depend on the specific details of the model , as we now outline : * : here @xmath96 and @xmath97 are taken as @xmath95 triplets @xcite .
similar to the knt model , the triplet model requires a @xmath91 symmetry with action @xmath98 to prevent tree - level neutrino masses and ensure a stable dm candidate .
the dm is the neutral component of the lightest triplet fermion and has extra yukawa interactions mediated by @xmath99 in eq . .
consequently its mass increases to @xmath100 tev @xcite .
* : this model employs the multiplets @xmath101 and @xmath102 @xcite . unlike the knt model and the triplet model
, the quintuplet variant does not require a @xmath91 symmetry to prevent tree - level neutrino mass .
it is therefore a viable model of radiative neutrino mass , independent of dm considerations .
interestingly , the @xmath91 symmetry @xmath98 is broken by a single coupling @xmath103 in the model ; for technically - natural values of @xmath104 , one obtains a long - lived dm candidate , while in the limit @xmath105 the @xmath91 symmetry becomes exact and the dm is absolutely stable @xcite .
the quintuplet model is therefore a model of radiative neutrino mass , with or without dm , depending on the region of parameter space considered . due to the gauge interactions
, the case with dm requires @xmath106 tev . * : this case has @xmath107 and @xmath108 , and is of particular interest @xcite .
similar to the quintuplet model , the @xmath91 symmetry is not required to prevent tree - level neutrino mass .
however , different from the quintuplet model , the septuplet model _ automatically _ possesses an accidental @xmath91 symmetry with action @xmath109 , and therefore always contains a stable dm candidate .
this gives a common description for neutrino mass and dm without requiring any new symmetries .
the dm is relatively heavy , with @xmath110 tev , due to the sommerfeld enhancement induced by @xmath92 gauge boson exchange @xcite .
detailed studies show that the variant models have large regions of viable parameter space that fit the neutrino oscillation data , reproduce the observed dm relic density , and satisfy experimental constraints from , e.g , lfv searches @xcite . in each of the variant models ,
the dm is a heavy majorana fermion and the mass for @xmath94 should obey @xmath111 , placing both multiplets beyond the reach of current colliders .
however , the scalar @xmath65 can retain an @xmath112 gev mass in all cases , offering the best chance for testing the models at colliders .
the models with larger @xmath95 multiplets also generate sizable contributions to lfv processes like @xmath113 .
in fact , these are typically enhanced relative to the knt case , improving the prospects for testing the model in future lfv searches @xcite . due to the new gauge interactions for the dm
, direct - detection prospects also improve for the models with larger multiplets @xcite .
there are additional models of neutrino mass with dm that are related to the knt model @xcite .
for example , one can replace the internal sm leptons in figure [ diag ] with down - type quarks , @xmath114 , and take @xmath115 and @xmath116 as colored multiplets .
further colored variants are also possible @xcite .
alternatively , one may replace the real ( majorana ) fermion @xmath8 with a complex ( dirac ) fermion and consider a modified topology for the loop - diagram , with the dm being an inert scalar ( see figure [ general ] for an example ) .
a systematic description of these variant models appears in ref .
@xcite . , @xmath117 and @xmath118 .
the beyond - sm multiplets @xmath119 and @xmath23 are taken odd under a @xmath91 symmetry and the dark matter belongs to the inert doublet @xmath65.__,width=264,height=113 ]
models with radiative neutrino mass offer a promising way to experimentally discern the new physics responsible for the origin of neutrino mass .
three - loop models are particularly interesting as the new physics is generically expected at or around the tev scale . furthermore , such models can provide viable dm candidates , thereby solving both the neutrino mass and dm problems .
here we reported recent analysis of the knt model , which show the model to be a viable theory for neutrino mass and dm that can be tested at collider experiments .
we also showed that the knt model belongs to a larger class of three - loop models that can generate similarly interesting observable effects .
s. nasri would like to thank shinya kanemura and all the members of the organizing committee of hpnp2015 for the very pleasing meeting and their great hospitality in toyama .
the authors thank r. soualah , c. s. chen and t. toma for productive collaborations in this field .
99 l. m. krauss , s. nasri and m. trodden , phys .
d * 67 * , 085002 ( 2003 ) .
m. aoki , s. kanemura and o. seto , phys .
lett . * 102 * , 051805 ( 2009 ) .
m. gustafsson , j. m. no and m. a. rivera , phys . rev .
* 110 * , no .
21 , 211802 ( 2013 ) [ erratum - ibid .
25 , 259902 ( 2014 ) ] .
h. hatanaka , k. nishiwaki , h. okada and y. orikasa , nucl .
b * 894 * , 268 ( 2015 ) .
g. aad et al .
( atlas collaboration ) , phys .
b 716 , 1 - 29 ( 2012 ) .
s. chatrchyan et al .
( cms collaboration ) , phys .
b 716 , 30 - 61 ( 2012 ) . a. ahriche and s. nasri , jcap * 1307 * , 035 ( 2013 ) .
j. adam _ et al . _
[ meg collaboration ] , phys .
lett . * 110 * , 201801 ( 2013 ) [ arxiv:1303.0754 [ hep - ex ] ] .
forero , m. tortola and j.w.f .
valle , phys .
d 86 , 073012 ( 2012 ) . p.a.r .
ade et al .
[ planck collaboration ] , astron .
astrophys . * 571 * , a1 ( 2014 ) .
a. ahriche , phys .
d75 , 083522 ( 2007 ) ; a. ahriche and s. nasri , phys . rev .
d 83 , 045032 ( 2011 ) ; phys .
d 85 , 093007 ( 2012 ) .
r. foot , h. lew , x. g. he and g. c. joshi , z. phys .
c * 44 * , 441 ( 1989 ) .
k. kumericki , i. picek and b. radovcic , phys .
d * 86 * , 013006 ( 2012 ) [ arxiv:1204.6599 [ hep - ph ] ] ; y. liao , jhep * 1106 * , 098 ( 2011 ) [ arxiv:1011.3633 [ hep - ph ] ] .
a. ahriche , c. s. chen , k. l. mcdonald and s. nasri , phys .
d * 90 * , 015024 ( 2014 ) [ arxiv:1404.2696 [ hep - ph ] ] .
a. ahriche , k. l. mcdonald and s. nasri , jhep * 1410 * , 167 ( 2014 ) [ arxiv:1404.5917 [ hep - ph ] ] .
a. ahriche , k. l. mcdonald , s. nasri and t. toma , arxiv:1504.05755 [ hep - ph ] . c. s. chen , k. l. mcdonald and s. nasri , phys . lett .
b * 734 * , 388 ( 2014 ) [ arxiv:1404.6033 [ hep - ph ] ] . | in this work , we report on recent analysis of three - loop models of neutrino mass with dark matter .
we discuss in detail the model of krauss - nasri - trodden ( knt ) @xcite , showing that it offers a viable solution to the neutrino mass and dark matter problems , and describe observable experimental signals predicted by the model .
furthermore , we show that the knt model belongs to a larger class of three - loop models that can differ from the knt approach in interesting ways . |
in 1980 , heckman identified a class of galactic nuclei whose optical emission - line properties differ from those of classical active galactic nuclei ( agn ) and regions .
these objects have narrow emission lines whose relative intensities are indicative of a relatively low ionization state of the emitting gas .
heckman ( 1980 ) named these objects low - ionization nuclear emission - line regions ( liners ) and defined them using the ratios of the optical forbidden lines of oxygen : [ ] @xmath83727/[]@xmath85007@xmath91 and [ ] @xmath86300/[]@[email protected] . since the division is observationally motivated , any object satisfying one of the criteria and close to satisfying the other one is often called a liner . recognizing that this classification is somewhat arbitrary , heckman ( 1980 ) noted the existence of `` transition '' objects whose spectra are intermediate between those of pure liners and regions . the original idea proposed by heckman ( 1980 ) to explain liners was that the emission lines are produced by shocks in the nucleus of the host galaxy . in fact , the optical spectrum arising from a shock - heated gas closely resembles the spectrum of liners @xcite .
similar emission could be produced by turbulent mixing layers between cold and hot ism components @xcite .
another possibility is that the lines originate in dense gas photo - ionized by x - rays ( ferland & netzer 1983 ; halpern & steiner 1983 ) .
the ionizing source can be a low - luminosity agn ( llagn ) , an idea supported by the discovery of broad emission lines ( fwhm@xmath10 km s@xmath2 ) in the spectra of many liners ( e.g. , filippenko & halpern 1984 ; ho et al .
1997b ; stocrchi - bergmann , baldwin , & wilson 1993 ; bower et al .
1996 ; barth et al . 1999 ; shields et al .
2000 ; barth et al . 2001
; eracleous & halpern 2001 ) .
alternatively , the source of ionizing photons could be the wolf - rayet or ob stars in a compact , young star cluster in the nucleus of the host galaxy ( terlevich & melnick 1985 ; filippenko & terlevich 1992 ; shields 1992 ; barth & shields 2000 ) .
yet another possibility is that post - agb stars from an old stellar population provide the ionizing photons that power the liner - like emission lines ( binette et al .
1994 ) .
recent spectroscopic surveys have shown that liners are very common in local galaxies , found in at least 30% of all galaxies and 50% of early - type spirals ( see review by ho 1996 , and references therein ) .
since liners are so common , determining the nature of their power source is important and could have far - reaching implications . if most liners are genuine agns , they would be the most common type of agn in the local universe and would represent the faint end of the agn luminosity function . the study of liners could shed light on the evolution of agn and their host galaxies since they are likely to represent the end state of quasar evolution .
it would also set a lower limit to the frequency of accreting supermassive black holes ( smbhs ) in galaxies .
moreover , the spectral energy distribution of liners is different from that of classical agns implying that , if they are agns , their mode of accretion might also be different @xcite .
suggested accretion modes are advection dominated accretion flows ( adafs ; narayan & yi 1994 , 1995 ) , convection - dominated accretion flows ( cdaf ; quataert & narayan 1999 ) and adiabatic inflow - outflow solutions ( adios ; blandford , & begelman 1999 ) .
if liners are powered by hot stars or stellar remnants , one can ask why their spectra are different from disk regions .
moreover , the presence of a central star cluster influences the chemical and dynamical evolution of the bulge of the galaxy @xcite , so the frequency of compact nuclear star clusters is of significant interest . in view of the importance of determining the nature of the central engines of liners ,
we have undertaken the analysis of a large sample of archival _ chandra _ observations of galaxies hosting liners .
the x - ray morphology of the central kiloparsec of the nucleus at the angular resolution of _ chandra _ can give a first hint to the power source .
an unresolved hard x - ray source at the center of the galaxy is a good indication of an agn , while a cluster of point sources and knotty diffuse emission suggests stellar phenomena .
the high spatial resolution of _ chandra _ and its spectroscopic capabilities also make it possible to extract the spectra of the spatially - distinct components ( point sources and diffuse emission ) to confirm their nature and to evaluate their contribution to the energy budget .
this is an important issue because some liners may be powered by a combination of an agn and a starburst ( hereafter `` composites '' ; e.g. , veron et al .
1981 , moran et al .
1996 , fernandes et al .
2004 ) .
previous surveys with _
chandra _ have studied the x - ray properties of liners with different techniques , purposes or samples than our study .
terashima & wilson ( 2003 ) used snapshot observations of liners with high radio fluxes in order to search for agns and determine their properties .
because of their short exposure times , they were able to detect agns with x - ray luminosities greater than @xmath11 .
satyapal et al . (
2005 ) and dudik et al .
( 2005 ) used a combination of long and short _ chandra _ observations to assess the presence of an agn in liners and iras observations to determine the presence of star formation but they do not examine the energy bugdet of these sources .
they did , however , determine that at least 50% of liners harbor a central hard x - ray source consistent with an agn .
filho et al .
( 2004 ) find that a similar fraction of liners harbor agns by means of radio observations of 16 liners and _ chandra _ observations of 8 of them .
they focused on the properties of the nucleus and argued that liners are powered by an agn , which is often undetected in the x - rays because the flux is highly absorbed .
eracleous et al . (
2002 ) studied three liners addressing the same questions that we address here .
out of those three liners , one was powered by an agn alone and two by stellar processes alone suggesting a diversity in the nature of the ionization sources of liners . in this paper
we continue and extend the work of eracleous et al .
some of the objects in our sample were studied individually using the same _ chandra _ data that we use here .
however , we repeat the analysis for the sake of uniformity and we add two important elements in our study : a detailed examination of the circumnuclear regions , and a careful consideration of the energy budget .
the objects in our sample are unlike some of the notorious liners whose nature is clear - cut , but they are typical of the greater population of liners .
in this paper we present the results and conclusions of our study , organized as follows .
the sample and initial data reduction procedures are described in [ sec : targets ] .
model fits to the spectra of individual discrete sources and the average spectra of diffuse sources are described in
[ sec : analysis ] . in [ sec : populations ] we analyze the properties of the point source populations collectively and in [ sec : individual ] , we evaluate the power source of each galaxy from the sample individually .
finally , we summarize the results and discuss the implications in [ sec : discussion ] .
we examined observations of liners in the _ chandra _ archive up to 2002 december and selected a sample of 19 galaxies hosting a pure liner or a transition object ( ho et al .
1997a ) based on their distance ( @xmath12 mpc ) and their exposure time ( @xmath13 ks ) .
these selection criteria ensure good spatial resolution in the central kiloparsec of each galaxy , a high signal - to - noise ratio ( s / n ) and a low detection limit ( the limiting 0.510 kev point source luminosity is in the range @xmath14 to @xmath15 ) .
most ( 14 out of 19 ) of the galaxies were observed for reasons unrelated to the presence of a liner , thus the sample is largely unbiased with respect to the liner properties . in [ sec : discussion ] , we use the distribution of liner types in our sample , to argue that it is representative of the general population of liners .
the properties of the host galaxies and the exposure times are summarized in table 1 .
the liner type follows the classification of ho et al .
( 1997a ) where l1 and l2 are `` pure '' liners .
l1 have fairly strong narrow emission lines combined with weak broad h@xmath16 but no broad h@xmath17 while l2 have only narrow emission lines .
`` transition '' objects with line ratios intermediate between those of regions and liners are referred to as t2 .
the distance is taken from ho et al .
( 1997a ) where they assume @xmath18 km s@xmath2 mpc@xmath2 .
column 5 gives the galactic hydrogen column density ( @xmath19 ) in the direction of the center of each galaxy @xcite .
ngc 1553 was not included in the ho et al .
( 1997a ) sample , but was listed as a liner by phillips et al .
( 1986 ) . among the spiral galaxies ,
ngc 2681 , ngc 3507 , ngc 4314 , ngc 5055 and ngc 7331 are viewed approximately face - on while ngc 4111 is approximately edge - on .
table 1 also includes the h@xmath16 luminosities of the target galaxies , as well as the estimated black hole masses .
the h@xmath16 luminosities were taken from from ho et al .
( 1997a ) , who measured them from spectra obtained through a @xmath20 aperture .
ngc 1553 is the only exception ; its h@xmath16 luminosity was taken from phillips et al .
the black hole masses were estimated from the stellar velocity dispersion using the prescription of tremaine et al .
the derived black hole masses are between @xmath21 and @xmath22 , with an average of @xmath23 and typical uncertainties of a factor of 2 ( or @xmath24 dex ; see discussion in barth et al .
2002 ) .
all the target galaxies were observed with the advanced ccd imaging spectrometer ( acis ; garmire et al .
the nuclei of most galaxies were located near the s3 ccd aimpoint , which did not suffer significant radiation damage during the early stages of the mission . only ngc 3607 and ngc 3608 were on a different ccd : i2 and i0 respectively .
the initial data screening was carried out with the ciao software package v 3.2.1 .
it consisted of the selection of events with grades 0 , 2 , 3 , 4 and 6 and the exclusion of events occurring during times of poor aspect solution and background flares .
we also removed the 05 pixel randomization and checked for aspect - solution errors .
we also applied a cti correction to the data for the two acis - i sources .
we corrected for pointing offsets using the appropriate information on the _ chandra _ x - ray center s web - site in order to improve the astrometry .
after screening , we produced an event file of the region centered on the nucleus with a 1 kpc radius .
the coordinates of the stellar nucleus were determined from isophotal contours of infrared images obtained with 2mass .
when coordinates derived from radio measurements or dynamical measurements were available , they were compared with the 2mass coordinates and we found that they were always in agreement within errors ( except for the vlba coordinates of ngc 4552 , which were 1.25 @xmath25 away from the 2mass coordinates ) . for reference
we note that the typical astrometric uncertainty from 2mass isophotal fitting is 125 , while the typical astrometric uncertainties from radio observations are 25 mas ( vla ) , 5 mas ( vlba ) , and 1.3 mas ( vlbi ) . in comparison ,
the astrometric uncertainty of _ chandra _ observations is 05 .
the nucleus of all the galaxies is at most 7away from the acis - s aimpoint , except for ngc 3607 and ngc 3608 , whose nuclei are respectively 211 and 140 from the acis - i aimpoint .
we begin the analysis of the data by constructing images of the central region of each galaxy , which we present in [ sec : images ] , below .
this is followed by fitting models to the spectra of bright , discrete sources in the nuclei of the host galaxies ( discussed in [ sec : fits ] ) and to the average spectra of diffuse emission in spiral and elliptical galaxies ( presented in
[ sec : diffuse ] ) . from the processed event files , we produced images of a region of radius 1 kpc around the center of each galaxy ( hereafter , the `` central kiloparsec '' ) in two different bands : 0.52 kev and 210 kev .
we also produced hardness ratio maps by combining the soft ( @xmath26 ) and hard ( @xmath27 ) images in the following way : @xmath28 .
for pixels which had no counts in either the soft or the hard band , we adopted a hardness ratio equal to the average of the image .
figure 1 shows the image of the central kiloparsec of each target galaxy in the soft band ( 0.52 kev ) and the hard band ( 210 kev ) and the hardness ratio map .
the morphology of each galaxy is summarized in table 2 .
the first column gives the name of the galaxy , while the second column gives the number of point sources that were identified by the wavdetect algorithm @xcite in the 0.510 kev band .
to prevent wavdetect from confusing fluctuations in diffuse emission with true discrete sources , we ran the algorithm on a @xmath29 kpc region with a threshold of 10@xmath30 .
an s in the third column indicates that a point source is exactly at the center of the galaxy in the 0.52 kev energy band , while an h indicates that a point source in the 210 kev energy band is at the galaxy center .
the combination of an s and an h indicates that the central source is detected in both energy bands . to determine
if a source is coincident with the center of a galaxy , we checked whether the _ chandra _ source coordinates matched the nuclear coordinates ( from 2mass or radio observations ) within the 1-@xmath25 uncertainty region .
the fourth column gives the same information as the third column but for the diffuse emission .
the morphology of the nuclear region as a function of energy can give a first hint for the power source of the liners .
a hard central source suggests an agn and a cluster of point sources or a knotty diffuse emission suggests stellar processes .
when both are observed at the same time , this hints at a combination of a stellar processes and an agn .
we extracted the spectra of all the point sources using the acis extract software ( ae , * ? ? ? * ) .
ae uses polygonal extraction regions which approximate contours of the _ chandra _ -acis point spread function ( psf ) .
a psf contour is selected such that the fraction of the psf energy enclosed is approximately 95% .
the background was measured as close to the point sources as possible in order to decrease the contamination from diffuse emission , which is fairly common .
spectral fitting was carried out using xspec v.11.2.0x @xcite .
the spectra were truncated at 0.5 kev at the low - energy end and between 2 and 10 kev , depending on the s / n , at the high - energy end .
the spectral models were also modified by interstellar photoelectric absorption with cross sections as given by morrison & mccammon ( 1983 ) .
spectra of faint sources with 20200 counts were binned adaptively by ae .
these spectra were then fitted with an absorbed power - law model in order to determine the point source flux .
sources with less than 20 counts could not be fitted because of the small number of degrees of freedom . for these sources
the luminosity was estimated by scaling a power law model ( @xmath31 , close to the median @xmath32 for the individual fits to the point sources ) absorbed by the average galactic column density ( @xmath33 ) so that the predicted count rate equals the observed count rate .
we chose this nominal value of @xmath19 because it is an average galactic column for the sample and a small variation in the @xmath19 ( @xmath34 ) did not have an impact on the resulting x - ray flux . using this model
, we also estimated the lowest detectable 0.510 kev point source luminosity for each galaxy , which we list in table 2 .
the luminosities of all the discrete sources are used in our study of the luminosity functions in [ sec : populations ] .
we estimated the uncertainties in our _ faint _
( @xmath35 counts ) x - ray point source luminosities by varying @xmath32 by @xmath36 for a power - law absorbed by the galactic column .
the 0.510 kev luminosity then varied by 510% . when varying @xmath32 by @xmath37 , the luminosity changes by 30%40% .
when increasing the absorbing column density by one order of magnitude , the luminosities increased by 525% . increasing the absorbing column density by two orders of magnitude dramatically increased the luminosity by 100%200% .
a small variation in the absorbing column density ( 20% ) did not have an impact on the x - ray luminosity estimate .
thus , based on the above estimates , and assuming that an absorbed power - law model provides an acceptable description of the spectrum , the typical uncertainty in the x - ray luminosities we derive is 30% , unless the absorbing column density is grossly underestimated .
the spectra of sources with more than 200 counts were binned with a minimum of 20 counts per bin and different models were fitted to the spectra in order to determine the nature of the source and the luminosity .
there is never more than one source per galaxy with more than 200 counts and it is usually the central point source .
it is important to find the most apropriate model for the x - ray spectrum so as to determine the absorbing column reliably .
systematic errors in the absorbing column can lead to large systematic errors in the source luminosity .
thus , we have tested the following models : * a blackbody spectrum ( bb ) , with the temperature ( @xmath38 ) and the normalization as free parameters . * a comptonized blackbody spectrum ( compbb , after nishimura et al . 1986 ) with the seed blackbody photon temperature ( @xmath39 ) , the electron temperature of the hot plasma ( @xmath40 ) , the optical depth of the plasma ( @xmath41 ) , and the normalization as free parameters .
* a spectrum from an accretion disk consisting of multiple blackbody components ( diskbb , after mitsuda et al .
1984 ; makishima et al . 1986 ) with the temperature at the inner disk radius ( @xmath42 ) and the normalization as free parameters . * an emission spectrum from hot diffuse gas with up to three components at different temperatures ( mekal , mekal2 t , mekal3 t ; after mewe et al . 1985 , 1986 ; kaastra 1992 ; liedhal et al . 1995 ) with the plasma temperature ( @xmath43 ) , the metallicity ( @xmath44 , initially set to solar ) , and the normalization as free parameters .
the redshift was set to 0 since the galaxies are all nearby . *
a power - law spectrum ( pow ) with the photon index of the power - law ( @xmath32 ) , and the normalization as free parameters . * a spectrum from an accretion disk around a black hole in an x - ray binary compton scattered from in a converging , plasma flow ( bmc , after titarchuk et al .
1997 ; titarchuk & zannias 1998 ; laurent & titarchuk 1999 ; borozdin et al .
1999 ; schrader & titarchuk 1999 ) with the temperature of the thermal photon source ( @xmath45 ) , the index of the electron power - law energy distribution ( @xmath16 ) , the illumination parameter , and the normalization as free parameters .
we have also tested certain linear combinations of these models such as blackbody plus power law or mekal plus power law ( as well as multiple mekal components at different temperatures ) .
these models can describe the spectra of x - ray binaries , accreting supermassive black holes or thermal diffuse emission depending on their luminosity and fit parameters .
for low mass x - ray binaries , the shape of the spectrum depends on the spectral state @xcite . in the high soft state
, the spectrum can be modeled as a soft component ( bb with @xmath46 kev or diskbb with @xmath42 between 0.5 and 1.5 kev ) and a comptonized component ( compbb ) with @xmath47 , @xmath48 kev and @xmath45 between 0.3 and 1.5 kev and with @xmath49 between @xmath3 and @xmath50 . in the low hard hard state
, the spectra low - mass x - ray binaries can be modeled as a soft component and a power - law tail but with with lower relative luminosities ( and absolute luminosities below the detection limit in our sample galaxies ) .
high - mass x - ray binaries have a hard spectrum with @xmath51 that can be modeled as a power - law of @xmath52 @xcite . accreting supermassive black hole such as those found at the center of seyfert galaxies
are observed to have a power - law spectrum with @xmath53 .
their luminosities are typically greater than @xmath54 erg s@xmath2 ( see mushotzky , done , & pounds , 1993 , and references therein ) , but less luminous agns are well known to exist , especially in liners .
table 3 summarizes the two best - fitting models for the spectrum of the central object in each galaxy , or the brightest off - nuclear object . fits with an acceptable value of the reduced @xmath55 but with unphysical parameters ( @xmath19 lower than that listed in table 1 , unreasonable photon index or temperature )
were not included in table 3 .
column 2 of table 3 gives the models that provided the best two fits and the values of the model parameters .
column 3 gives the @xmath55 and the number of degrees of freedom , and column 4 the luminosity calculated for the model in the 0.510 kev range .
for all these models , we assumed a solar metal abundance . in figure
[ fig : spectra ] , we show the best - fitting model for each spectrum ( the first one listed for each object in table 3 ) superposed on the data along with the post - fit residuals .
when there were significant residuals , we tried to modify the metallicity using the values given by terlevich & forbes ( 2002 ) and letting it vary , but it never suppressed the residuals , nor did it improve the reduced @xmath55 , and was not supported by a f - test ; therefore we kept solar abundances .
we explored the effects of assuming solar abundance in objects with sub - solar or super - solar abundances by increasing and lowering the abundace and measuring the change in the other fit parameters .
we found that the other fit parameters ( column density , power - law index , temperature , luminosity ) varied very slightly ( @xmath56 ) when the abundance increased or decreased by one order of magnitude . in 70% of the galaxies of our sample we find diffuse emission in the central kiloparsec .
the frequent presence of diffuse emission was already noted by halderson et al .
( 2001 ) . in 87% of the galaxies that have diffuse emission
, this emission contributes over 50% of the observed x - ray luminosity .
the flux from the diffuse emission in each galaxy was not high enough to yield a high s / n spectrum individually except for ngc 3628 .
thus , we stacked the spectra from all the galaxies .
first , we extracted the spectrum from a 1 kpc radius around the center of each galaxy by carefully subtracting all the point sources and by choosing a background region free of diffuse emission and point sources . in the process of subtracting the point sources we also subtracted some diffuse emission , which makes our estimate of the diffuse emission luminosity a lower limit , strictly speaking .
since the number of point sources in each galaxy is small ( 3 on average ) , we estimate that this lower limit is not too far from the true luminosity . before combining the spectra
, we normalized them to a fiducial total count rate of 0.1 s@xmath2 so that the result is a good representation of the average spectrum of this type of source .
then we stacked the spectra using the ftool addspec and propagated the errors appropriately .
ngc 3607 and ngc 3608 were not included in the stacking procedure because they were observed with a different ccd and addspec requires that all the observations are made with the same ccd .
we stacked separately the diffuse emission from spiral and elliptical galaxies .
figure 2 shows the resulting spectra .
the diffuse emission spectra were not well fitted by a two - temperature component plasma , instead they required a plasma model plus a power - law , with parameters as given in table 3 .
the spectra themselves , with models superposed are shown in figure [ fig : stackedspec ] .
the model parameters are very similar between elliptical and spiral galaxies , @xmath57 , @xmath58 kev , @xmath591.31.5 , with solar abundance .
we also let the abundance vary as a free parameter , but the improvement in reduced @xmath60 was not significant according to the f - test . in [ sec : populations ] we estimate the average luminosity of unresolved x - ray binaries in the inner kiloparsec and we show that it is comparable to the luminosity of the power - law spectral component .
we also fitted the diffuse emission spectrum of ngc 3628 independently and we fount that the two best models were either a two - temperature plasma model or a power law with parameters as given in table 3 ( the spectrum is shown in fig .
[ fig : spectra ] ) .
the spectral parameters for this particular galaxy are in general agreement with what we find for the stacked spectra , above .
we also note that david et al .
( 2005 ) studied the diffuse emission spectrum of one of the galaxies in our sample , ngc 3379 , and found that it could be fitted with a similar model as our stacked spectra , above .
the plasma component of the diffuse emission has a similar temperature in both types of galaxies with a value that is slightly lower than what is observed in normal galaxies ; @xmath610.7 kev for spirals @xcite and @xmath62 kev for ellipticals @xcite . from the normalization constant of the xspec fit ( after rescaling it to the average _ observed _ count rate )
, we can calculate the emission measure and then the density .
we find that @xmath63 for the elliptical galaxies and @xmath64 for the spiral galaxies .
these densities are comparable to those found in normal galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . from the stacked spectra
, we can also deduce that the average x - ray luminosity of the diffuse emission ( integrated over the inner kiloparsec ) is @xmath65 for the ellipticals and @xmath66 for the spirals . within uncertainties , the luminosities of the diffuse emission in spiral and elliptical galaxies are comparable to each other . we calculated the luminosity of the diffuse emission for each galaxy individually by scaling the model luminosity to the count rate for each galaxy .
table 4 summarizes the 0.52 and 210 kev luminosity of the diffuse emission within the central kiloparsec of each galaxy ( columns 2 and 3 ) and the 0.52 and 210 kev luminosity of the diffuse emission in the central 25 region ( columns 4 and 5 ) .
the last quantity was computed by integrating the model spectrum after scaling it to the count rate of each individual ; we use this quantity in our evaluation of the power sources of individual galaxies in [ sec : individual ] . the sound speed of the hot gas is @xmath67 km s@xmath2 ( assuming solar composition , full ionization and @xmath68 kev ; @xmath69 is the ratio of specific heats and @xmath70 is the mean molecular weight ) , slightly larger than the stellar velocity dispersions in these galaxies .
the dynamical and cooling time scales of the hot gas are @xmath71 years and @xmath72 years , respectively ( assuming that the gas is contained in a sphere of @xmath73 pc ) . in view of these time scales and the fact that the sound speed is not considerably higher than the stellar velocity dispersion , it appears likely that the gas is bound to the host galaxy and its temperature is sustained by interaction with the stars .
the gas may , in fact , originate in stellar mass loss . as david et al .
( 2005 ) note , the mass of hot gas in ngc 3379 is approximately @xmath74 and it could have been supplied by stellar mass loss over a period of only @xmath21 years .
the liners in our sample have between zero and 12 point sources in their central kiloparsec ( including candidate agns ) with an average number of approximately 3 point sources and a most frequent value of two point sources .
the point source detection limit for our sample varies between @xmath75 and @xmath76 in the 0.510 kev band ( see table 2 ) .
the 0.510 kev luminosity of the detected point sources varies between @xmath77 and @xmath78 with an average of @xmath79 . in order to determine whether the point source populations of the central kiloparsec of liners differ from those of normal galaxies , we constructed luminosity functions ( including point sources from all the galaxies in the sample but excluding candidate agns ) and compared them to those of normal galaxies
the normal galaxy comparison sample comprises 6 galaxies with a wide range of hubble types that have weak optical emission lines and have been observed intensively in the x - ray band ( ngc 1637 , ngc 3556 , ngc 4365 , ngc 4382 , ngc 4631 and ngc 4697 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we constructed the average luminosity functions for the elliptical galaxies and spiral galaxies separately using point source luminosities measured as described in [ sec : fits ] .
we excluded candidate agns and truncated each luminosity function at the highest detection limit of galaxies in that category ( @xmath80 and @xmath81 for ellipticals and spiral respectively ) .
there are 21 sources in the elliptical galaxy luminosity function and seven sources in the spiral galaxy luminosity function .
the resulting cumulative luminosity functions are shown in figure 4 .
we fitted a simple power - law through the cumulative luminosity function of the sources in spiral galaxies , which we derived from the differential luminosity function as follows : @xmath82 @xmath83 where @xmath84 and @xmath85 . in this convention
@xmath86 is a fiducial luminosity and @xmath87 and @xmath88 are dimensionless constants ( whose values depend on the choice of @xmath86 ) .
the resulting index for spiral hosts , @xmath89 , is in agreement with that found by colbert et al .
( 2004 ) for normal spiral galaxies ( @xmath90 ) .
the cumulative luminosity function for the sources in elliptical galaxies is best fitted by a broken power - law with @xmath91 ( for @xmath92 ) , @xmath93 ( for @xmath94 ) and @xmath95 .
the value of @xmath96 is comparable to the value found by colbert et al ( 2004 ) for normal elliptical galaxies ( @xmath97 ) . but colbert et al .
( 2004 ) do not mention a break in the luminosity function , although some of their luminosity functions seem to flatten in their figure 3 .
other authors find a break in the point - source luminosity function of normal elliptical galaxies @xcite . on the other hand , sivakoff et al .
( 2003 ) do not find a break in the luminosity function of the two elliptical galaxies ngc 4365 and ngc 4382 .
the break luminosity that some authors find is comparable to that from our luminosity function .
this break could indicate that the nature of the xrb is different for sources above and below the break luminosity .
sarazin et al .
( 2001 ) suggested that the accretor of xrbs below the break luminosity is a neutron star , while the accretor of xrbs above the break luminosity is a black hole , based on the agreement of the break luminosity with the eddington luminosity of a 1.4 neutron star .
the value of @xmath98 does not match any value cited by colbert et al .
( 2004 ) even though it is closest to the slope of the cumulative luminosity function of merger galaxies ( @xmath99 ) . at any rate , the slope of the cumulative luminosity functions is well in the range of those for normal galaxies so liners do not appear to have an unusual xrb population .
ngc 3379 has 10 point sources in its inner 1 kpc region , so we were able to construct a luminosity function for this individual galaxy .
the cumulative luminosity function is well fitted by a power - law with @xmath100 .
this value is close to the value found by colbert et al ( 2004 ) for starburst galaxies . using the above properties of the point source populations
, we can investigate whether the power - law component in the diffuse emission spectra ( see [ sec : diffuse ] ) can be due to the presence of unresolved x - ray binaries . to test this hypothesis , we extrapolated the luminosity function of equation ( [ eq : lumfunc ] ) to very low luminosities and integrated it up to the minimum detectable luminosity of a resolved point source , i.e. , @xmath101 @xmath102\ ] ] @xmath103 where we have neglected the term in square brackets since @xmath104 and @xmath105 .
we apply equation ( [ eq : psource ] ) to elliptical and spiral galaxies using the lowest value from the luminosity functions of figure [ fig : lumfunc ] as the value of @xmath106 .
we find that for elliptical galaxies ( @xmath107 , @xmath108 , @xmath109 , @xmath110 ) , the estimated unresolved point source luminosity ( 0.510 kev ) is @xmath111 , while for spiral galaxies ( @xmath112 , @xmath113 , @xmath114 , @xmath115 ) , @xmath116 .
these luminosities are certainly high enough to explain the hard x - ray component in the average diffuse emission spectra also the spectra of the hard component have a power - law index of @xmath117 , which is consistent with low hard state x - ray binaries .
from figure 1 and table 2 , there is no obvious trend in the morphology of the central kiloparsec of the galaxies .
some galaxies have a central point source , some do not .
some galaxies have diffuse emission and some do not . from table 3 and figure 2 ,
the spectrum of the brightest sources in the galaxies are not all fitted by the same model and have a wide range of luminosity and model parameters .
an evaluation of the power source of each individual galaxy is then required . to aid in our interpretation
, we checked if the nuclei of the galaxies were observed at other wavelengths , especially in the radio and the uv .
in general terms , a high x - ray radio loudness parameter ) instead of the b luminosity ( @xmath118 ) was introduced by terashima & wilson ( 2003 ) and the cut - off value of @xmath119 was derived from the the optical cut - off @xmath120 and the observed correlation between @xmath121 and @xmath118 for a sample of llagns , seyferts and pg quasars : @xmath122 .
] , defined as @xmath123 > -4.5 \label{eq : rx}\ ] ] and a flat radio spectrum are indicative of a low luminosity agn ( hereafter llagn ) @xcite so the presence of a compact , high - brightness temperature radio core provides strong evidence of the presence of a llagn .
an unresolved uv point source with non - stellar spectrum is also strong evidence for a llagn @xcite , while an extended uv source is unlikely to be an agn .
if the flux of the uv point source varies over time , this indicates an agn @xcite .
when a broad h@xmath16 was detected ( i.e. , the liner is classified as l1 ) , this demonstratates the presence of a llagn .
finally , we checked whether our interpretation of the power source could be supported by the estimate of the age of the stellar population from population synthesis models , applied to specific galaxies . in cases where an agn is not a likely power source ,
the emission lines could be powered either by photoionization from massive stars from a young stellar population or from post - agb stars and the nuclei of young planetary nebulae associated with an old stellar population . for liners with composite power sources ( llagn plus stellar processes )
, we determined the contribution of the agn to the central x - ray luminosity using the the spectrum from a circle of radius 25 at the center of the galaxy .
we chose such an aperture because it has a similar projected area as the aperture used in the survey of ho et al .
( 1997a ) to measure the emission line luminosities .
a radius of 25 corresponds to 120 pc for a galaxy at 10 mpc and 300 pc for a galaxy at 25 mpc , i.e. , it represents the very core the galaxies .
we fitted this spectrum with a two - temperature plasma model plus a power - law .
the power - law component parameters were fixed to those obtained during the analysis of the central point source ; the two - temperature plasma models represents the diffuse emission .
then we calculated the 210 kev luminosity of the central point source by integrating the power - law component , the 210 kev luminosity of the full 25 central region from the two - temperature plasma plus power - law model , and thus the fraction of the x - ray luminosity in the central region contributed by the point source .
the various 210 kev luminosities are listed in table 4 .
we use this fractional contribution as a basis for classifying a liner as a composite .
when we use the 0.510 kev band to calculate the contribution from the point source to the total central x - ray luminosity , the numbers change slightly without changing the conclusion about the dominant source ( except in the case of ngc 4438 as explained in its individual description ) .
all the luminosities are corrected for absorption .
ngc 1553 : : ( l2/t2 , s0 ) has three point sources in its central kiloparsec .
the brightest one is exactly at the center of the galaxy and its spectrum is well fitted either by a two - temperature plasma model ( @xmath124 kev , @xmath125 kev ) or a power - law ( @xmath126 ) .
the preference of a the two - temperature over a one - temperature plasma model and a power - law model is justified by the f - test .
both of the above models give a luminosity of @xmath127 erg s@xmath2 .
with such a luminosity and spectral characteristics , this source is probably an agn .
this is supported by observations in the radio band : ngc 1553 is a weak radio source ( 10 mjy at 843 mhz ; harnett 1987 ) . assuming a radio spectral index of @xmath128 ( adopted as our default value hereafter ; where @xmath129 ) , we calculate the radio flux at 5 ghz to be @xmath130 mjy and @xmath131 , which means that this source is radio loud .
blanton , sarazin & irwin ( 2001 ) came to the same conclusion on the nature of the central source using the same data set but they estimated its luminosity to be slightly higher ( @xmath132 erg s@xmath2 ) .
we evaluated the luminosity of the central 25 region to be @xmath133 erg s@xmath2 .
the luminosity of the agn is evaluated by integrating the power - law that fits the data to be @xmath134 erg s@xmath2 , which means that the agn is the source of 92% of the x - ray luminosity from the central region .
this is consistent with estimates of the age of the stellar population that indicate that the stars in the nucleus of ngc 1553 are old ( @xmath9 17 gyr , terlevich & forbes 2002 ; 10 - 15 gyr , longhetti et al .
ngc 2681 : : ( l1 , sb0/a ) has three point sources within its central kiloparsec . the brightest point source is at the exact center and is detected in both the soft and hard bands .
this is reminiscent of ngc 1553 , except that ngc 2681 has more diffuse emission than ngc 1553 .
the spectrum of the central source is best fitted by a two - temperature plasma model or a power - law plus a single - temperature plasma model ( see table 3 ) .
the luminosity inferred from these models is @xmath135 erg s@xmath2 .
these fits do not exclude the possibility that the central source is a llagn since circumnuclear hot gas can give rise to the thermal component superposed on a power - law agn spectrum .
this source was detected in the radio waveband ( @xmath136=12.0 mjy at 1.49 ghz ; condon 1987 ) leading to @xmath137 ( radio loud ) .
a broad h@xmath16 emission line was detected in the optical @xcite suggesting that this source is a llagn .
we estimated the total luminosity from the 25 central region to be @xmath138 erg s@xmath2 and the luminosity of the agn to be @xmath139 erg s@xmath2 ( by integrating the power - law component ) , which means that the potential agn contributes only @xmath140% of the total x - ray luminosity of the central kiloparsec .
this indicates that the contribution from stellar processes is dominant .
the other 80% of the luminosity comes from three other point sources in the central 25 region and diffuse emission .
this is consistent with estimates that the solar metallicity stellar population of ngc 2681 was formed in a starburst episode about 1 gyr ago @xcite .
an intermediate age ( @xmath141@xmath142 years ) population has also been detected in the nucleus of this galaxy by @xcite .
ngc 3379 : : ( t2 , e1 ) harbors a cluster of 10 point sources in its nucleus .
one of these point sources ( not the brightest one ) is at the exact center of the galaxy .
the spectrum of this source had only 89 counts , thus its luminosity was estimated by fitting an absorbed power - law : @xmath143 erg s@xmath2 .
an unresolved radio source of @xmath136=0.7 mjy has been detected at 5 ghz at the center of this galaxy @xcite .
we calculate @xmath144 , which makes this source radio loud and suggests that it is a llagn .
the luminosity of the central 25 region was evaluated to be @xmath145 erg s@xmath2 and the llagn has @xmath146 erg s@xmath2 .
the candidate llagn contributes @xmath14720% to the luminosity of the central 25 region , making this liner a composite .
however , age estimates indicate a stellar population of intermediate age , in the range of 69 gyr @xcite .
we were able to fit the brightest source in the nucleus with a number of models involving a power law and a soft component ( diskbb , bb , mekal ) .
the addition of the soft component to the simple power - law is not strongly justified by the f - test ( chance probability between 14 and 44% depending on the model for the soft component ; a plasma model to the power - law gives the smallest chance probability ) .
the total luminosity of this source is @xmath148 erg s@xmath2 . from its spectrum , luminosity , and off - center position
, one can infer that this source is probably an x - ray binary ( xrb ) .
ngc 3507 : : ( l2 , sbb ) has two point sources in its central kiloparsec : one very faint ( 14 counts ) close to the edge of the region and a central extended source embedded in a bright patch of diffuse emission .
the central source had only 187 counts , which did not allow us to test a wide variety of spectral models .
instead we estimated its luminosity by fitting a simple mekal model ( the photon index of the best - fitting power - law model was unrealistic ) and we find @xmath149 after correcting for absorption .
a weak , _ extended _ radio source ( @xmath136=16 mjy ) is detected at 1.45 ghz @xcite , which suggests that this object is not a llagn .
it could be a collection of snrs producing a superbubble around it seen as the x - ray diffuse emission and radio extended source .
the fact that the data could not be fitted with a realistic power - law index strengthen this idea .
this indicates that the x - ray emission of this liner is powered by stellar processes .
this result is consistent with the estimate of the age of the stellar population , which indicates the presence of a young stellar component ( @xmath150 years ) @xcite .
ngc 3607 : : ( l2 , s0 ) has two point sources embedded in diffuse emission , but none of them is at the center .
the brightest point source did not have enough counts to justify fitting models to its spectrum ; its luminosity was estimated to be @xmath151 erg s@xmath2 by scaling the count rate to that expected from a power - law model .
the absence of a central point source is a strong case against an agn , in agreement with terashima et al .
( 2002 ) who did not find evidence for an agn from an x - ray spectrum obtained with asca .
the point sources and the diffuse emission suggest that the x - ray emission is powered by stellar processes .
the age of the stellar population is estimated to be between 3.6 and 5.6 gyr @xcite .
ngc 3608 : : ( l2 , e2 ) has two point sources with the brightest one at the center , embedded in diffuse emission .
the spectrum of the brightest source does not have enough counts to fit a variety of models to it but its luminosity was estimated to be @xmath152 erg s@xmath2 .
there is no detection of this source at 5 ghz but only an upper limit of 0.5 mjy @xcite .
we can then put a limit on the radio loudness : @xmath153 .
thus this source might be radio loud , hence a candidate llagn .
we note that this galaxy is the second most distant object in the sample and that the source is offset by 140from the aim point of the ccd and the 50% encircled energy radius at 1.49 kev at this off - axis location is 15 , i.e. a large fraction of the 25 radius region used to determine the contribution to the central x - ray luminosity . also such a large
psf increases the chance of source confusion and the chance that this is a multiple source .
the age of the stellar population in ngc 3608 has been estimated to be between 9 and 10 gyr @xcite .
we estimate that the luminosity of the central 25 is @xmath154 erg s@xmath2 and the luminosity of the central source is @xmath155 erg s@xmath2 so the hypothetical agn contributes 60% of the x - ray luminosity .
ngc 3628 : : ( t2 , sb ) has five point sources in its central kiloparsec including a bright point source 1 kpc away from the center .
there is no point source at the center of the galaxy , but a central patch of diffuse emission .
the spectrum of the diffuse emission is well fitted by a variety of models with a two - temperature plasma or a power - law giving the best fits .
the resulting luminosity is ( @xmath156 erg s@xmath2 ) .
an extended radio source was detected at the center of ngc 3628 @xcite , which could indicate a sn - driven superbubble .
this galaxy is believed to be a starburst galaxy due to the presence of numerous xrbs @xcite .
there is no study of the stellar population that gives an estimate of the age .
the absence of a llagn , the point sources and the diffuse emission suggest that the x - ray emission of this liner is powered by stellar processes ( perhaps a young starburst ) .
ngc 4111 : : ( l2 , s0 ) has only one central point source with a hard x - ray spectrum .
there is also a fair amount of diffuse emission .
the spectrum extracted from the central source does not have enough counts to justify fitting models to it but the luminosity can be estimated to be @xmath157 erg s@xmath2 and @xmath158 erg s@xmath2 .
there is a detection of a central source in the radio with @xmath159 mjy @xcite , which means that @xmath160 .
the radio loudness , the central position of the source and the hard x - ray spectrum suggest that this source is a llagn .
we evaluated the luminosity of the central 25 region to be @xmath161 erg s@xmath2 , meaning that the point source contributes to @xmath162% of the x - ray luminosity .
the age of the stellar population of ngc 4111 was estimated to less than 1 gyr @xcite .
ngc 4125 : : ( t2 , e6 ) has only one hard point source at its center surrounded by diffuse emission .
this point source did not have enough counts to justify fitting various models to its spectrum , but its luminosity was estimated to be @xmath163 erg s@xmath2 .
this source was not detected in the radio but a limit was set : 0.5 mjy at 5 ghz ( wrobel & heeschen 1991 ) .
this means that @xmath164 so this source could be radio loud .
due to its central position and possible radio loudness , this source could be a llagn .
we estimate that the luminosity of the central 25 region is @xmath165 erg s@xmath2 and the llagn has @xmath166 erg s@xmath2 , i.e. @xmath167% of the x - ray luminosity .
the remaining x - ray luminosity is contributed by the diffuse emission .
ngc 4314 : : ( l2 , sba ) has no point sources , just sparse soft diffuse emission .
ngc 4314 is one of the nearest examples of galaxies hosting a circumnuclear ring of star formation .
its diameter is 10 , which falls within the central kiloparsec .
the age of the stellar population in the nuclear ring has been estimated to be very young ( 115 myr , benedict et al .
so the x - ray emission of this liner is very likely powered by young stars .
ngc 4374 : : ( l2 , e1 ) has four point sources in its nucleus . the brightest one is at the center and has a hard x - ray spectrum . there is also significant soft , knotty , diffuse emission , although the knots could be embedded point sources .
there are many models that give good fits , in particular a power - law and a plasma model plus a power - law .
the f - test justifies the addition of a plasma component to the power - law ( chance probability = 4@xmath168 ) .
the luminosity is @xmath169 erg s@xmath2 .
the stellar population of ngc 4374 is quite old ( 13.7 gyr , proctor & sansom 2002 ; 11.8 gyr , terlevich & forbes 2002 ) .
radio images of ngc 4374 show structures resembling jets ( * ? ? ?
* ; * ? ? ?
* ) , which is a strong indication of an agn .
the flux of the core at 5 ghz was measured to @xmath170 mjy @xcite , giving @xmath171 , which means that this source is radio - loud .
the total x - ray luminosity in the central 25 radius region is @xmath172 erg s@xmath2 and the luminosity of the agn is @xmath173 erg s@xmath2 ( obtained by integrating the power - law spectral component ) .
so the contribution from the llagn to the x - ray luminosity is 90% .
ngc 4438 : : ( l1 , s0/a ) has only one point source , which is close to the center but not exactly centered ( within 2 @xmath25 ) .
it has a hard x - ray spectrum and is embedded in soft diffuse emission with an elongated morphology .
the spectrum of the central source is well fitted by a thermal plasma model or a plasma component plus a power - law with a total luminosity of @xmath174 erg s@xmath2 . since
the source is deeply embedded in diffuse emission , it is possible that it is an agn .
this source is weakly detected at 1.49 ghz with a flux of 116 mjy @xcite , which gives @xmath175 , i.e. it can be radio loud .
also a broad h@xmath16 line was detected @xcite , which supports the agn hypothesis .
an amorphous patch about 5 in size was detected in the uv @xcite , which was interpreted as a superbubble and might be associated with the extended patch of diffuse emission seen in the x - ray .
however , age estimates of the stellar population indicate that an old population dominates the optical starlight spectrum in the bulge @xcite .
we estimated the x - ray luminosity from the central 25 radius region to be @xmath176 erg s@xmath2 and the luminosity of the agn to be @xmath176 erg s@xmath2 ( obtained by integrating the power - law component ) .
thus the hard x - ray luminosity of this liner is primarily powered by the llagn .
we note that this is the only liner whose dominant x - ray source luminosity changes depending of what band we consider .
the agn dominates in the 210 kev band , while the plasma emission dominates in the 0.510 kev band .
ngc 4457 : : ( l2 , s0/a ) has one bright central source with a hard spectrum , embedded in soft diffuse emission . the spectrum of the central source is well fitted by a power - law plus plasma model or a two - temperature plasma model ( @xmath177 erg s@xmath2 ) .
this can be interpreted as the spectrum of an embedded agn and is confirmed by the detection of this source at 1.49 ghz @xcite . from this detection
we infer @xmath178 , which means that this source is radio loud .
the luminosity of the agn is @xmath179 erg s@xmath2 while the luminosity of the central 25 is @xmath180 erg s@xmath2 so the agn contributes @xmath147 55% of the x - ray luminosity .
this rest of the luminosity comes from the diffuse emission , so this liner is a composite .
ngc 4494 : : ( l2 , e1 - 2 ) has six point sources in its central kiloparsec and the coordinates of the brightest source agree with the center of the galaxy within 2 @xmath25 .
this source has a hard x - ray spectrum , which is best fitted by a two - temperature plasma model or a power - law .
the luminosity is @xmath181 erg s@xmath2 .
there is no detection of a source at 5 ghz , with an upper limit of 0.5 mjy @xcite , leading to @xmath182 .
thus , this source could be radio loud and could be an agn .
the luminosity of the central 25 is @xmath183 erg s@xmath2 and the luminosity of the power agn is @xmath184 erg s@xmath2 so the agn candidate contributes 90% of the luminosity .
the stellar population was estimated to be less than 5 gyr old @xcite .
ngc 4552 : : ( t2 , e1 ) has a bright , hard central source , plus four other point sources in its central kiloparsec and soft diffuse emission . the spectrum of the central source is well fitted by a power - law or a comptonized blackbody .
these models give a luminosity of @xmath185 erg s@xmath2 .
this source was clearly detected at 4.85 ghz with @xmath186mjy and a spectral index @xmath187 @xcite so we calculate @xmath188 , which means that this source is radio loud .
moreover a uv flare from this source was detected @xcite and the uv flux was observed to vary over a short period of time @xcite indicating a weak agn .
the stellar population was estimated to be old ( 9.6 gyr , terlevich & forbes 2002 ) .
the x - ray luminosity from the central 25 region is @xmath189 erg s@xmath2 and the luminosity of the agn is @xmath190 erg s@xmath2 .
thus the x - ray luminosity of this liner comes mostly from its agn ( @xmath191 ) .
ngc 4636 : : ( l1 , e1/s0 ) has bright knotty , soft , diffuse emission .
there are five point sources deeply embedded in the diffuse emission .
one of them is at the center but this is not the brightest source and its spectrum is soft .
this source did not have enough counts to justify fitting its spectrum with a variety of models , but its luminosity was evaluated to be @xmath192 erg s@xmath2 .
a broad h@xmath16 component was detected @xcite indicating a llagn , but the x - ray signature of an agn is not detected .
loewenstein et al .
( 2001 ) failed to detect an agn using the same data set , as did terashima et al .
( 2002 ) using asca observations .
if there is an agn ( as suggested by the detection of a broad h@xmath16line ) , it could be heavily absorbed .
ngc 5055 : : ( t2 , sb / c ) has a five point sources in its nucleus , clustered close to the center , and a bright hard source at the very center .
the spectrum of this source is well fitted by a two - temperature plasma model , but a power - law plus plasma model gives a very poor fit .
we did not attempt more complex models due to the small number of degrees of freedom .
maoz et al .
( 1995 ) resolved the central point source in the uv and obtained a radius of 7 pc ( at the distance assumed in this paper ) .
maoz et al .
( 2005 ) did not observe any variation of the uv flux of this source .
all these pieces of evidence indicate that this source is more likely a compact star cluster .
this source is not detected at various radio frequencies except at 57.5 mhz with @xmath136=2.1 jy @xcite , which means that this source is radio quiet ( @xmath193 ) and not likely to be an agn .
the stellar population of ngc 5055 has a range of ages from 1 myr to 10 gyr with most of the population of intermediate and old age ( * ? ? ?
* ; * ? ? ?
so the x - ray emission of this liner is powered by stellar processes .
ngc 5866 : : ( t2 , s0 ) has a central , hard x - ray source and one other point source far away from the center and embedded in soft diffuse emission .
the spectrum of the central source did not have enough counts to attempt a fit .
however its luminosity was estimated to be @xmath194 erg s@xmath2 with 98% of it emitted in the hard band .
this is also the luminosity from the central 2.5 region indicating that this source is the main contributor to the central x - ray luminosity .
this source has @xmath195 mjy at 5 ghz @xcite .
we calculated that @xmath196 , which means that this source is radio loud and possibly a llagn .
ngc 7331 : : ( t2 , sb ) has three point sources but none are at the center ( at the @xmath197 confidence level ) .
there is no hard x - ray emission close to the center of the galaxy .
no source had enough counts to justify spectral fits .
the stellar population does not have a young component and is mostly 1 - 10 gyr old @xcite .
so this liner might be powered by stellar processes , most likely associated with the old stellar population .
in order to generalize our results , the first issue we examine is whether the present sample is representative of the entire population of liners , in terms of the distribution of liner classes .
the ratio of transition objects to pure liners in the ho et al .
( 1997c ) survey is 0.75 while in our sample , it is 0.63 ( ignoring ngc 1553 , which was not classified by ho ) .
so the transition objects are slightly underrepresented in our sample . moreover in our sample , 27% of the pure liners are l1 , which is comparable to the 25% of l1 liners in the ho et al .
( 1997c ) survey .
we also note that the h@xmath16 luminosity distribution of the sample closely resembles that from the ho et al ( 1997c ) sample .
thus , one can generalize the results from this sample with reasonable confidence . soft , circumnuclear diffuse emission is common in our objects with a typical 0.52 kev luminosity on the order of @xmath198 , in agreement with the results of gonzlez - martin et al .
( 2005 , 2006 )
. in 12 out of 19 liners there are signs of an agn whether it dominated the x - ray luminosity or not . in 5/19 liners
we did not detect an agn , with upper limits to the luminosity as listed in table 4 . for 2/19 liners ( ngc 4636 and ngc 7331 )
, it was difficult to determine whether there was an agn because of the knotty structure of the diffuse emission .
thus , at least 63% ( and possibly up to 74% ) of the objects in the sample harbor an agn
. table 4 summarizes the 0.52 and 210 kev luminosity of the agn candidates in each galaxy ( columns 6 and 7 ) along with their eddington ratio ( column 8 ; see details below ) and x - ray radio loudness parameter ( column 9 ) . when comparing the 210 kev luminosity of the agn ( column 4 ) to the luminosity of the diffuse emission ( columns 2 and 3 )
, we notice that for 10 of the 12 galaxies with an agn candidate , the agn dominates the 210 kev luminosity in the inner 25 ; in 9 out of these 10 cases , the agn dominates the total 210 kev luminosity in the inner kiloparsec .
when we consider the luminosity in the entire 0.510 kev band , however , the contribution of the diffuse emission is often comparable to the contribution of the agn , even in the inner 25 .
figure 5 shows the distribution of luminosities of the agns found in our sample ( and upper limits when there is no detection of an agn ) , the distribution of @xmath199 and the distribution of the fractional contribution of the agn to the 0.510 kev luminosity of the central 25 region .
we note that the detected agns span a wide range in 210 kev luminosity , from @xmath200 to @xmath201 , with an average of @xmath202 .
this luminosity range is comparable to that found by terashima & wilson ( 2003 ) for llagn in liners with bright , compact radio sources .
we can express the eddington ratio of the agns we have detected ( the ratio of the bolometric luminosity of the agn and the eddington luminosity ) as follows : @xmath203 @xmath204 where @xmath205 is the black hole mass reported in table 1 , @xmath206 is the x - ray luminosity ( reported in table 4 ) and @xmath207 is a `` bolometric correction '' ( see ho 1999 ) .
the resulting values of @xmath208 are included in column 8 of table 4 ; they span the range @xmath209 , for _ detected _ agns in our sample .
these values are comparable to those found by filho et al .
( 2004 ) , for a sample of transition - type liners ( using the same method for black hole mass determination ) .
these results however are significantly different from those of satyapal et al .
( 2005 ) who find @xmath210 and @xmath211 for the candidate agns in their infrared - bright liner sample .
we note that they used a bolometric correction three times higher than the one we use , but this does not account for the whole difference .
their agns have higher x - ray luminosities so the very low eddington ratios agns in our sample are probably the very low luminosity agns missing from their sample . [ [ comparison - with - normal - galaxies - and - the - source - of - fuel - for - the - agn ] ] comparison with normal galaxies and the source of fuel for the agn ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the x - ray luminosities and eddington ratios of the accreting black holes in our sample of liners are very similar to those of black holes in the `` quiescent '' galaxies studied by soria et al .
( 2006a , b ) .
moreover , the properties of the diffuse emission ( gas temperature , density , and soft x - ray luminosity ; see [ sec : diffuse ] ) as well as the black hole masses in the two samples are quite similar .
therefore , we can infer that the bondi accretion rates should also be similar .
indeed , an inspection of figure 5 of soria et al .
( 2006b ) shows that several of the liners in our sample fall in the same region of the eddington ratio @xmath212 bondi rate diagram . as soria et al .
( 2006b ) demonstrate , the fuel necessary to power the observed low - level activity in their sample can easily be provided by a combination of mass loss from stars within the black hole s `` sphere of influence '' and from the flow of hot interstellar gas into that sphere from larger radii ( bondi accretion ) .
in fact , the hot interstellar gas itself is likely to have originated in stellar mass loss over a larger volume around the center of the galaxy ( see our earlier discussion in [ sec : diffuse ] ) .
the same interpretation applies to the liners that we study here , based on the close similarities between the nuclear properties of the two samples .
these similarities also contribute to the puzzle of why some galaxies host liners while others do not . to address this question
we examine the energy budget of the agn in the next section in the broader context of the power - source behind the optical emission - line spectra of liners . in this sample , stellar processes , as manifested by soft , diffuse emission , seem to frequently contribute to the energy budget of the central 25 region , whether alone or in combination with an agn .
the agn is often not the major power source . to illustrate this we plot in figure [ fig : balance ] the luminosity from the diffuse emission @xmath212 the luminosity from the agn in the 0.510 kev and 210 kev bands . in more than half of the cases
the diffuse emission dominates .
it would be interesting to determine whether the dominant contribution to the x - ray luminosity depends on the liner classification or the hubble type , but our sample is too small to allow any firm conclusions on these questions . to investigate the role of the agn in powering the observed optical emission lines , we looked for correlations between the different x - ray luminosities and the h@xmath16 luminosity ( taken from ho et al 1997a ) .
we find a correlation between @xmath213 and @xmath214 of the central 25 region ( spearman rank correlation coefficient @xmath215 ) .
our objects fall in the same region of the @xmath216@xmath213 diagram as the lowest - luminosity objects of ho et al .
( 2001 ) and also very close to the correlation between these two quantities followed by low - redshift quasars and seyfert 1 galaxies . this can be seen in figure [ fig : hoandmine ] where the squares represent the data from our sample and the stars represent the liners observed by ho et al .
( 2001 ) and @xmath217 @xmath218 to the observed count rate of the central sources to estimate the luminosity of the sources .
the objects from ho et al .
( 2001 ) that do not follow the correlation are mostly transition - type liners , which were not included in the calculation of the correlation . ] .
we also draw the best fit found by ho et al .
( 2001 ) as the solid line and the range of luminosity where a starburst would fall ( dotted lines ) .
the latter locus is obtained by combining relations between these two quantities and the star formation rate ( kennicutt 1998 ; ranalli , comastri , & cetti 2003 ; grimm , gilfanov , & sunyaev 2003 ; colbert et al .
2004 ; persic et al .
it is noteworthy that none of our objects are close to the starburst locus , which is consistent with an absence of young stellar populations in the majority of the galaxies in our sample . in figure
[ fig : lxvslha ] , we focus on the correlation between the 25 x - ray and h@xmath16 luminosities for the objects in our sample .
the trend persists regardless of whether we use the 210 kev or the 0.52 kev luminosity .
in fact , the correlation appears stronger when considering the soft x - ray band instead of the hard x - ray band : the spearman rank correlation coefficients are 0.48 and 0.33 in the soft and hard bands respectively .
this suggests that production mechanism of the h@xmath16 line is not strongly correlated to the source of the hard x - ray radiation .
figure [ fig : lxvslha ] gives a preliminary indication that photoionization by the agn is unlikely to be responsible for powering the optical emission lines .
we have carried out the following tests to investigate this issue further . 1 .
the first test is based on energy balance between the emission lines and the ionizing continuum .
assuming that the ionizing continuum is a power - law of photon index @xmath32 from 1 ry to 100 kev ( consistent with adaf models ) , its total luminosity can be obtained from the observed 210 kev luminosity via @xmath219 , where the function @xmath220 has values between 5 and 12 for @xmath32 in the range 1.11.9 .
thus for the ionizing continuum to power the h@xmath16 luminosity alone , we require that @xmath221 . out of the 19 objects in our sample ,
only four satisfy this condition , under the most optimistic assumption . in the case of ngc 1553 ,
ngc 4374 ngc 4494 , and ngc 4552 the ionizing continuum luminosity exceeds the h@xmath16 luminosity by a factor of a few in the best case .
additional power would be available , if the agn sed included a significant `` uv bump '' , but this appears rather unlikely , in view of the llagn sed shapes shown by ho ( 1999 ) .
it is also unlikely that we have underestimated the x - ray luminosity of the agn because of absorption .
this is because the spectra of 2/3 of our agn have a sufficiently high s / n that we were able to fit them , determine the absorbing column , and correct the luminosity ( of course , we can not rule out the possibility that the entire x - ray emission is scattered and that we do not have a direct view of the agn ) .
another test for the photoionization hypothesis relies on photon counting .
assuming the same continuum shape as above , we can express the ionizing photon rate produced by the agn in terms of the observed x - ray luminosity as @xmath222 .
the function @xmath223 has values between 2 and 30 for @xmath32 in the range 1.11.9 . by relating this to the h@xmath16 photon rate
, we can derive the following requirement for the h@xmath16 luminosity : @xmath224 , where @xmath225 is the fraction of ionizing photons absorbed by the emission - line gas and one in 2.2 recombinations leads to the emission of an h@xmath16 photon ( osterbrock 1989 ; we have assumed that there are no secondary ionizations or excitations by fast photoelectrons ) .
none of the objects in our sample satisfy this condition , even if @xmath226 .
the four objects identified in the previous paragraph could satisfy this condition if each ionizing photon leads to 25 ionizations , on average .
all other objects would require more than 10 secondary ionizations ; most require more than 100 .
thus the agns in the objects of our present sample are not energetically significant .
it is also noteworthy that in more than half the cases , the electromagnetic power of the agn is not enough to drive the diffuse x - ray emission . examining the samples of terashima & wilson ( 2003 ) and filho et al .
( 2004 ) , we notice a similar effect .
in particular , the former sample includes agns typically an order of magnitude more luminous than ours but the faintest 25% of their agns fail the energy budget test and can not power the h@xmath16 emission . in the latter sample the x - ray detected agns have comparable luminosities to ours and they all fail the energy budget test .
it appears , at least at first glance , that as the luminosity of the agn decreases , its important as the power source of the optical emission lines diminishes too .
therefore , we conclude our discussion by considering other possible sources of power for the optical emission lines below . the analysis presented above
shows that the properties of the liners in the present sample are very similar to those of normal galaxies .
more specifically , the hot interstellar gas in these liners has , on average , the same temperature , density , and luminosity as in normal galaxies .
moreover , the average properties of the stellar populations in these liners are similar to those of normal galaxies : the x - ray binary luminosity functions are the same and in most of our liners the optical spectra indicate old stellar populations . finally ,
even though an agn is often detected , its electromagnetic power does not appear to be significant .
thus we must look for an explanation for the liner optical spectra that involves either stellar processes or a mechanical interaction of the agn with its immediate environs . in three of the liners in our sample ( ngc 3507 , ngc 3628 , and ngc 5055 )
there is evidence for a young stellar population , while in one other object ( ngc 2681 ) there is evidence for an intermediate - age population . in more general terms , cid fernandes et al .
( 2005 ) find that 38% of liners and transition objects have intermediate - age stellar populations in their very nuclei .
hot stars from a young stellar population are obvious sources of ionizing photons that could power the optical emission lines ( see 1 ) . in old and especially in intermediate - age
stellar populations , ionizing photons can be provided by post - agb stars and the central stars of planetary nebulae @xcite .
an alternative explanation is that the agn powers the emission lines by depositing mechanical energy via outflows ( jets or winds ) into the emission line gas . at the low eddington ratios that we find here
, the kinetic power of jets from the agn can be significantly higher ( by up to 3 orders of magnitude ) than the electromagnetic luminosity ( see nagar , falcke , & wilson 2005 ) .
some of this power may be dissipated by shocks in the immediate vicinity of the agn , which would naturally lead to liner - like emission line spectra .
this hypothesis is bolstered by the observation that the emission - line regions of liners are rather amorphous or filamentary on scales of 10100 pc @xcite .
this type of feedback from the agn may also contribute to additional heating of the diffuse gas , which in turn would limit the accretion rate and lead to the very low eddington ratio that we infer ( see the discussion by soria et al .
2006b ) . in conclusion , we note that even though some liners are obviously powered by their agns ( m81 and ngc 4579 are among the best known examples ) , the electromagnetic luminosity of the agn is not always enough to account for the observed emission - line luminosity .
all of the physical processes discussed in the literature so far may contribute to powering the emission line spectra , making liners a heterogeneous class of objects and suggesting that agn photoionization may be the exception rather than the rule .
this work was supported by the national aeronautics and space administration through _ chandra _ award number ar4 - 5010a issued by the _
chandra _ x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics and space administration under contract nas8 - 03060 .
we would like to thank patrick broos and leisa townsley for their assistance with the use of the acis extract software .
we also thank the anonymous referee for helpful comments .
argyle , r.w . , clements , e.d .
1990 , obs , 110 , 93 arnaud , k.a .
1996 , in asp conf .
ser . 101 , astronomical data analysis software and systems v , ed .
jacoby , g. & barbes j. ( san francisco : asp)17 barret , d. 2001 , adspr , 28 , 307 barth , a.j . ,
filippenko , a.v . , & moran , e.c . 1999 ,
apj , 525 , 673 barth , a.j .
, ho , l.c . ,
filippenko , a.v . ,
rix , h .- w .
, & sargent , w.l.w .
2001 , apj , 546 , 205 barth , a.j . , & shields , j.c . 2000 , pasp , 112 , 753 barth , a. j. , ho , l. c. & sargent , w. l. w. 2002 , aj , 124 , 2607 benedict , g.f . , howell , d.a . ,
jrgensen , i. , kenney , j.d.p . , & smith , b.j .
2002 , , 123 , 1411 binette , l. , magris , c.g . , stasinska , g. , & bruzual , a.g .
1994 , a&a , 292 , 13 blandford , r.d . , & begelma , m.c . 1999 ,
mnras , 303 , 1 blanton , e.l . ,
sarazin , c.l . ,
irwin , j.a . 2001 , apj , 552 , 106 boker , t. , 2000 , in asp conf . ser .
211 , massive stellar clusters , ed .
a. lancon & c.m .
boily ( san francisco : asp ) , 227 bonatto , c. , bica , e. , pastoriza , m.g . ,
& alloin , d , 2000 , , 355 , 99 borozdin , k. , revnivtsev , m. , trudolyubov , s. , shrader , c. , & titarchuk , l. 1999 , apj , 517 , 367 botte , v. , ciroi , d. , di mille , f. , rafanelli , p. , & romano , a. 2005 , mnras , 356 , 789 bower , g.a . ,
wilson , a.s . , heckman , t.m .
, & richstone , d.o .
1996 , aj , 111 , 1901 broos , p.s . ,
townsley , l.k . , getman , k. , & bauer , f.e .
2002 , acis extract , an acis point source extraction package ( university park : pennsylvania state univ . ) burstein , d. , bertola , f. , buson , l.m . ,
faber , s.m . , & lauer , t.r .
1988 , apj , 328 , 440 cappellari , m. , bertola , f. , burstein , d. , buson , l.m . ,
greggio , l. , & renzini , a. 2001 , apj , 551 , 197 cappellari , m. , renzini , a. , greggio , l. , di serego alighieri , s. , buson , l.m . ,
burstein , d. , & bertola , f. 1999 , apj , 519 , 117 carollo , m. 1999 , apj , 523 , 566 colbert , e.j.m .
, heckman , t.m . ,
ptak , a.f . , strickland , d.k .
, & weaver , k.a .
2004 , apj , 602 , 231 combes , f. 2001 advanced lectures on the starburst - agn connection , eds , i. aretxaga , d. kunth , & r. mjica ( singapore : : world scientific ) , 223 condon , j.j .
1987 , apjss , 65 , 485 condon , j.j . ,
frayer , d.t . , & broderick , j.j .
1991 , aj , 101 , 362 david , l. p. , jones , c. , forman , w. , & murray , s. s. 2005 , apj , 635 , 1053 de young , d.s . , butcher , h. , & condon , j.j .
1980 , , 242 , 511 dickey , j.m .
, & lockman , f.j .
1990 , ara&a , 28 , 215 dopita , m.a , koratkar , a.p . ,
evans , i.n . , allen , m. , bicknell , g.v . ,
sutherland , r.s . ,
hawley , j.f . , & sadler , e. 1996 , in the physics of liners in view of recent observations .
asp conference series , vol .
eracleous , m. , koratkar , a. , leitherer , c. , & ho , l. , p.44 dopita , a.m. , & sutherland , r.s .
1995 , apj , 455 , 468 eracleous , m. , & halpern , j.p .
2001 , 554 , 240 ferland , g.j . , &
netzer , h. 1983 , apj , 264 , 105 cid fernandes , r.c . ,
gonzalez delgado , r.m .
, storchi - bergmann , t. , pires martins l. , & schmitt , h. 1005 , mnras , 356 , 270 cid fernandes , r.c .
, et al . 2004 , , 605 , 105 eracleous , m. , shields , j.c . , chartas , g. , & moran , e.c .
2002 , apj , 565 , 108 ferrarese , l. & merrit , d. 2000 , apj , 539 , l9 fey , a.l .
, ma , c. , arias , e.f . ,
charlot , p. , feissel - vernier , m. , gontier , a .- m . , jacobs , c.s .
, li , j. , & macmillan , d.s .
2004 , aj , 127 , 3587 filippenko , a.v .
1996 , in asp conf .
103 , the physics of liners in view of recent observations , ed .
eracleous , m. , koratkar , a. , leitherer , c. , & ho , l. ( san francisco : asp ) 17 filippenko , a.v . ,
& halpern , j.p .
1984 , apj , 285 , 458 filippenko , a.v . , & terlevich , r. 1992 , apj , 397 , l79 filho , m.e . ,
fraternali , f. , markoff , s. , nagar , n.m . , barthel , p.d .
, ho , l.c . , & yuan , f. 2004 , a&a , 418 , 429 forman , w. , jones , c. , & tucker , w. 1985 , apj , 293 , 102 freeman , p.e .
2002 , apjss , 138,185 garmire , g.p .
, bautz , f.w . ,
ford , p.g . ,
nousek , j.a . , & ricker , g.r .
2003 , , 4851 , 28 gebhardt , k. , et al .
2000 , apj , 539 , l13 gonzalez delgado , r.m . , cid fernandes , r. , perez , e. , martins , l.p .
, storchi - bergmann , t. , schmitt , h. , heckman , t. , & leitherer , c. 2004 gonzlez - martin , o. , masegosa , j. , mrquez , j. , dultzin - hacyan , d. , & guerrero , m. a. 2005 , mem.s.a.it .
, 76 , 91 gonzlez - martin , o. , masegosa , j. , mrquez , j. , dultzin - hacyan , d. , & guerrero , m. a. 2005 , in populations of high energy sources in galaxies ( proc . iau symp . 230 ) , in press gregg , m.d . 1989 , , 337 , 45 gregg , m.d . , ferguson , h.c . , minniti , d. , tanvir , n. , & catchpole , r. 2004 , aj , 127 , 1441 , h .- j . ,
gilfanov , m. , & sunyaev , r. 2002 , mnras , 339 , 793 halderson , e.l . , moran , e.c . , filippenko , a.v . , & ho , l.c .
2001 , apj , 122 , 637 halpern , j.p . , &
steiner , j.e .
1983 , apj , 269 , l37 harnett , j.i .
1987 , mnras , 227 , 887 heckman , t.m 1980 , a&a , 87 , 152 ho , l.c .
1996 , in asp conf .
103 , the physics of liners in view of recent observations , ed .
eracleous , m. , koratkar , a. , leitherer , c. , & ho , l. ( san francisco : asp ) 70 ho , l.c .
1999 , apj , 516 , 672 ho , l.c . , feigelson , e.d . ,
townsley , l.k . ,
sambruna , r.m . ,
garminre , g.p . ,
brandt , w.n .
, filippenko , a.v . , griffiths , r.e .
, ptak , a.f . , & sargent , w.l.w .
2001 , apj , 549 , l51 ho , l.c . , filippenko , a.v . , &
sargent , w.l.w .
1997a , apjs , 112 , 315 ho , l.c . , filippenko , a.v . , sargent , w.l.w . , & peng , c. 1997b , apjs , 112 , 391 ho , l.c . ,
filippenko , a.v . , & sargent , w.l.w .
1997c , apj , 487 , 568 ho , l.c , rudnick , g. , rix , h .- w . , shields , j.c . , mcintosh , d.h . , filippenko , a.v . , sargent , w.l.w . , &
eracleous , m. 2000 , apj , 541 , 120 hummel , e. , & saikia , d.j .
1991 , a&a , 249 , 43 immler , s. , wang , q.d . ,
leonard , d.c . , & schlegel , e.m . 2003 , apj , 595 , 727 israel , f.p . , & mahoney , m.j . 1990 , apj , 352 , 30 kaastra , j.s .
1992 , an x - ray spectral code for optically thin plasmas ( internal sron - leiden report , updated version 2.0 ) laing , r.a . , &
bridel , a.h .
1987 , , 228 , 557 kennicutt , r. c. 1998 , ara&a , 36 , 189 lasota , j.p .
, et al . 1996 , apj , 462 , 142 laurent , p. , & titarchuk , l. 1999 , apj , 511 , 289 liedahl , d.a . , osterheld , a.l . , and goldstein , w.h .
1995 , apjl , 438 , 115 loewenstein , m. , mushotzky , r.f . , angelini , l. , & arnaud , k.a .
2001 , apj , 555 , l21 longhetti , m. , bressan , a. , chiosi , c. , & rampazzo , r. 2000,a&a,353,917 makishima , k. , maejima , y. , mitsuda , k. , bradt , h.,v . ,
remillard , r.a .
, tuohy , i.r .
, hoshi , r. , & nakagawa , m. 1986 , apj 308 , 635 maoz , d. , filippenko , a.v .
, ho , l.c . , rix , h .-
w , bahcall , j.n . ,
schneider , d.p . , & macchetto , f.d .
1995 , apj , 440 , 91 maoz , d. , filippenko , a.v .
, ho , l.c . , macchetto , f.d . ,
rix , h .- w . , & schneider , d.p .
1996 , apjs , 107 , 215 maoz , d. , nagar , n.m . , falcke , h. , & wilson , a.s .
2005 , apj , 625,699 mewe , r. , gronenschild , e.h.b.m . , and van den oord , g.h.j .
1985 , a&as , 62 , 197 mewe , r. , lemen , j.r . , and van den oord , g.h.j .
1986 , a&as , 65 , 511 mitsuda , k. , koyama , k. , makishima , k. , matsuoka , m. , ogawara , y. , suzuki , k. , tanaka , y. , shibazaki , n. , & hirano , t. 1984 , pasj , 36 , 741 moran , e.c . ,
halpern , j.p . , & helfand , d.j 1996 , apjs , 106 , 341 morrison , r. , & mccammon , d. 1983 , apj , 270 , 119 mushotzky , r.f . , done , c. , & pounds , k. a. 1993 , ara&a , 31 , 717 nagar , n.m . , falcke , h. , & wilson , a.s .
2005 , a&a , 435 , 521 nagar , n.m . ,
falcke , h. , wilson , a.s . ,
& ho , l.c .
2000 , apj , 542 , 186 nagar , n.m . ,
falcke , h. , wilson , a.s . , & uvelstad , j.s .
2002 , a&a , 392 , 53 narayan , r. , & yi , i. 1994 , apj , 428 , l13 narayan , r. , & yi , i. 1995 , apj , 444 , 231 nishimura , j. , mitsuda , k. , & itoh , m. 1986 , pasj , 38 , 819 osterbrock , d. e. 1989 , astrophysics of gaseous nebulae and active galactic nuclei ( mill valley : university science books ) osullivan , e. , & ponman , t.j .
, 2004 , , 349 , 535 persic , m. , & rephaeli , y. 2001 , a&a , 382 , 843 persic , m. , rephaeli , y. , braito , v. , cappi , m. , della ceca , r.,franceschini , a. , & gruber , d. e. 2004 , , 419 , 849 phillips , m.m . ,
jenkins , c.r .
, dopita , m.a . ,
sadler , e.m .
, & binette , l. 1986 , aj , 92 , 1062 pogge , r.w . , maoz , d. , ho , l.c . , &
eracleous , m. 2000 , apj , 532 , 323 .
proctor , r.n . , &
sansom , a.e .
2002 , mnras , 333,517 quataert , e. , & narayan , r. 1999 , apj , 516 , 399 , p. , comastri , a. , & setti , g. 2003 , , 399 , 39 randall , s.w . ,
sarazin , c.l . , & irwin , j.a .
2004 , apj , 600 , 729 renzini , a. , greggio , l. , di serego - alighieri , s. , cappellari , m. , burnstein , d. , & bertola , f. 1995 , , 378,39 sarazin , c.l . , irwin , j.a . , & bregman , j.n .
2001 , apj , 556,533 schrader , c. , & titarchuk , l. 1999 , apj 521 , l21 shields , j.c .
1992 , apj , 399 , l27 shields , j.c . ,
rix , h .- w . ,
mcintosh , d.h .
, ho , l.c . , rudnick , g. , filippenko , a.v . ,
sargent , w.l.w . , & sarzi , m. 2000 , apj , 543 , l27 sivakoff , g.r . ,
sarazin , c.l . , & irwin , j.a .
2003 , apj , 599 , 218 sivakoff , g.r . ,
2004 , apj , 617 , 262 soria , r. , et al . 2006a , apj , in press ( astro - ph/0511293 ) soria , r. , et al . 2006b , apj , in press ( astro - ph/051134 ) storchi - bergmann , t. , baldwin , j.a . , & wilson , a.s .
1993 , apj , 410 , l11 strickland , d.k . ,
colbert , e.j.m . ,
heckman , t.m . , weaver , k.a . , dahlem , m. , & stevens , i.r .
2001 , apj , 560 , 707 terashima , y. , iyomoto , n. , ho , l.c . , &
ptak , a.f .
2002 , apjs , 139 , l36 terashima , y. & wilson , a.s .
2003 , apj , 583 , 145 terlevich , a.i . , & forbes , d.a .
2002 , mnras , 330 , 547 terlevich , r. , & melnick , j. 1985 , mnras , 213 , 841 titarchuk , l. , mastichiadis , a. , & kylafis , n.d .
1997 , apj , 487 , 834 titarchuk , l. , & zannias , t. 1998 , apj,493 , 863 tremaine , s. et al .
2002 , apj , 574 , 740 trinchieri , g. , fabbiano , g. , & canizares , c.r .
1986 , apj , 310 , 637 veron - cetty , m .- p . &
veron , p.1986 , a&as , 66 , 335 veron , p. , veron , m.p . ,
bergeron , j. , & zuidervijk , e.j .
1981 , a&a , 97 , 71 wang , q.d . ,
immler , s. , walterbos , r. , lauroesch , j.t . , & breitschwerdt , d. 2001 , apjl , 555 , 99 wang , q.d . , xhaves , t. , & irwin , j.a .
2003 , apj , 598 , 969 wrobel , j.m . &
heeschen , d.s . 1991 ,
aj , 101 , 148 cccrccccc 1553 & l2/t2 & s0 & 16.1 & 1.50 & @xmath227 & 8.0 & 34.17 & 2000/01/02 + 2681 & l1 & sb0/a & 13.3 & 2.42 & @xmath228 & 7.2 & 80.00 & 2001/01/30 + 3379 & t2 & e1 & 8.1 & 2.75 & @xmath229 & 8.3 & 31.92 & 2001/02/13 + 3507 & l2 & sbb & 19.8 & 1.63 & @xmath230 & & 39.76 & 2002/03/08 + 3607 & l2 & s0 & 19.9 & 1.48 & @xmath231 & 8.4 & 39.00 & 2001/06/12 + 3608 & l2 & e2 & 23.4 & 1.49 & @xmath232 & 8.1 & 39.00 & 2001/06/12 + 3628 & t2 & sb & 7.7 & 2.23 & @xmath233 & 7.9 & 58.70 & 2000/12/02 + 4111 & l2 & s0 & 17.0 & 1.40 & @xmath234 & 7.8 & 15.00 & 2001/04/03 + 4125 & t2 & e6 & 24.2 & 1.84 & @xmath235 & 8.4 & 65.08 & 2001/09/09 + 4314 & l2 & sba & 9.7 & 1.78 & @xmath236 & 7.2 & 16.28 & 2001/04/02 + 4374 & l2 & e1 & 16.8 & 2.60 & @xmath237 & 8.8 & 28.85 & 2000/05/19 + 4438 & l1 & s0/a & 16.8 & 2.66 & @xmath238 & & 25.40 & 2002/01/29 + 4457 & l2 & s0/a & 17.4 & 1.80 & @xmath239 & 7.0 & 39.38 & 2002/12/03 + 4494 & l2 & e1 - 2 & 9.7 & 1.52 & @xmath240 & 7.6 & 25.16 & 2001/08/05 + 4552 & t2 & e1 & 16.8 & 2.57 & @xmath241 & 8.5 & 55.14 & 2001/04/22 + 4636 & l1 & e1/s0 & 17.0 & 1.81 & @xmath242 & 8.2 & 53.05 & 2000/01/26 + 5055 & t2 & sb / c & 7.2 & 1.32 & @xmath243 & 7.1 & 28.36 & 2001/08/27 + 5866 & t2 & s0 & 15.3 & 1.47 & @xmath244 & 7.7 & 34.18 & 2002/11/14 + 7331 & t2 & sb & 14.3 & 8.61 & @xmath245 & 7.6 & 30.13 & 2001/01/27 + ccccc 1553 & 3 & hs & & @xmath246 + 2681 & 3 & hs & s & @xmath247 + 3379 & 12 & s & & @xmath248 + 3507 & 2 & s & s & @xmath249 + 3607 & 2 & & & @xmath249 + 3608 & 2 & s & s & @xmath250 + 3628 & 5 & & h & @xmath251 + 4111 & 1 & h & s & @xmath252 + 4125 & 1 & hs & & @xmath253 + 4314 & 0 & & s & @xmath254 + 4374 & 4 & hs & s & @xmath255 + 4438 & 1 & hs & s & @xmath256 + 4457 & 1 & hs & s & @xmath246 + 4494 & 6 & s & & @xmath257 + 4552 & 5 & hs & s & @xmath258 + 4636 & 5 & s & h & @xmath259 + 5055 & 5 & hs & & @xmath260 + 5866 & 2 & h & s & @xmath261 + 7331 & 3 & & & @xmath261 + llcc 1553 & mekal2 t , @xmath262 @xmath218 , @xmath263 kev , @xmath264 kev & 51.76/40 & @xmath265 + & pow , @xmath266 @xmath218 , @xmath126 & 63.6/43 & @xmath267 + 2681 & mekal2 t , @xmath268 @xmath218 , @xmath269 kev , @xmath270 kev & 8.3/14 & @xmath271 + & pow+mekal , @xmath272 @xmath218 , @xmath273 kev , @xmath32=2.0@xmath274 & 8.2/14 & @xmath275 + 4374 & mekal+pow , @xmath276 @xmath218 , @xmath277 kev , @xmath278 & 14.6/27 & @xmath279 + & pow , @xmath280 @xmath218 , @xmath281 & 18.4/29 & @xmath282 + 4438 & mekal+pow , @xmath283 @xmath218 , @xmath284 kev , @xmath285 & 9.3/12 & @xmath286 + & mekal , @xmath287 @xmath218 , @xmath288 kev & 8.4/14 & @xmath289 + 4457 & mekal+pow , @xmath290 @xmath218 , @xmath291 kev , @xmath292 & 6.1/9 & 2.1@xmath293 + & mekal2 t , @xmath294 @xmath218 , @xmath295 kev , @xmath296 kev & 8.8/9 & 1.8@xmath293 + 4494 & mekal2 t , @xmath297 , @xmath298 kev , @xmath299 kev & 13.2/14 & @xmath300 + & pow , @xmath301 @xmath218 , @xmath302 & 18.6/16 & @xmath303 + 4552 & pow , @xmath304 @xmath218 , @xmath305 & 43.7/43 & @xmath306 + & compbb , @xmath307 @xmath218 , @xmath308 kev , @xmath309 & 42.5/42 & @xmath306 + 5055 & mekal2 t , @xmath310 @xmath218 , @xmath311 kev , @xmath312 kev & 4.3/10 & @xmath313 + ellipticals & mekal+pow , @xmath314 , @xmath315 kev , @xmath316 & 208.4/177 & @xmath317 + spirals & mekal+pow , @xmath318 @xmath218 , @xmath319 kev , @xmath320 & 237.1/236 & @xmath321 + 3628 & mekal2 t , @xmath322 @xmath218 , @xmath323 kev , @xmath324 kev & 18.07/18 & @xmath325 + & pow , @xmath19=2@xmath326 @xmath218 , @xmath327 & 21.26/20 & @xmath325 + 3379 & pow , @xmath328 @xmath218 , @xmath329 & 31.8/27 & @xmath330 + & mekal+pow , @xmath331 @xmath218 , @xmath332 kev , @xmath333 & 27.2/25 & @xmath334 + cccccrrrrc 1553 & @xmath335 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & @xmath340 & @xmath341 & @xmath342 + 2681 & @xmath343 & @xmath344 & @xmath345 & @xmath346 & @xmath337 & @xmath347 & @xmath348 & @xmath349 + 3379 & @xmath350 & @xmath351 & @xmath352 & @xmath353 & @xmath354 & @xmath200 & @xmath355 & @xmath356 + 3507 & @xmath357 & @xmath358 & @xmath359 & @xmath335 & @xmath360 & @xmath361 & & + 3607 & @xmath362 & @xmath363 & @xmath364 & @xmath365 & @xmath366 & @xmath367 & @xmath368 & + 3608 & @xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath373 & @xmath374 & @xmath375 & @xmath376 + 3628 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & @xmath381 & @xmath382 & @xmath383 & + 4111 & @xmath343 & @xmath384 & @xmath385 & @xmath386 & @xmath384 & @xmath387 & @xmath388 & @xmath389 + 4125 & @xmath390 & @xmath391 & @xmath336 & @xmath392 & @xmath393 & @xmath394 & @xmath395 & @xmath376 + 4314 & @xmath396 & @xmath397 & @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & + 4374 & @xmath403 & @xmath343 & @xmath343 & @xmath404 & @xmath405 & @xmath406 & @xmath407 & @xmath408 + 4438 & @xmath387 & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath397 & & + 4457 & @xmath413 & @xmath414 & @xmath415 & @xmath351 & @xmath317 & @xmath416 & @xmath417 & @xmath418 + 4494 & @xmath419 & @xmath410 & @xmath420 & @xmath421 & @xmath422 & @xmath345 & @xmath423 & @xmath424 + 4552 & @xmath425 & @xmath426 & @xmath397 & @xmath427 & @xmath425 & @xmath428 & @xmath407 & @xmath429 + 4636 & @xmath430 & @xmath358 & @xmath394 & @xmath431 & @xmath432 & @xmath433 & @xmath434 & + 5055 & @xmath300 & @xmath435 & @xmath386 & @xmath398 & @xmath436 & @xmath437 & @xmath438 & + 5866 & @xmath439 & @xmath352 & @xmath440 & @xmath441 & @xmath442 & @xmath443 & @xmath444 & @xmath429 + 7331 & @xmath404 & @xmath393 & @xmath415 & @xmath351 & @xmath445 & @xmath446 & @xmath447 & + | using archival _ chandra _ observations of 19 liners we explore the x - ray properties of their inner kiloparsec to determine the origin of their nuclear x - ray emission , to investigate the presence of an agn , and to identify the power source of the optical emission lines .
the relative numbers of liner types in our sample are similar to those in optical spectroscopic surveys .
we find that diffuse , thermal emission is very common and is concentrated within the central few hundred parsec .
the average spectra of the hot gas in spirals and ellipticals are very similar to those of normal galaxies .
they can be fitted with a thermal plasma ( @xmath0 kev ) plus a power law ( photon index of 1.3 - 1.5 ) model .
there are on average 3 detected point sources in their inner kiloparsec with @xmath1 erg s@xmath2 .
the average cumulative luminosity functions for sources in spirals and ellipticals are identical to those of normal galaxies . in the innermost circle of 25 radius in each galaxy
we find an agn in 12 of the 19 galaxies .
the agns contribute a median of 60% of the 0.5 - 10 kev luminosity of the central 25 region , they have luminosities of @xmath3@xmath4 ( eddington ratios of @xmath5 to @xmath6 ) .
the ionizing luminosity of the agns is not enough to power the observed optical emission lines in this particular sample .
thus , we suggest that the lines are powered either by the mechanical interaction of an agn jet ( or wind ) with the circumnuclear gas , or by stellar processes , such as photoionization by post - agb stars from an old or intermediate - age population ( in most cases ) or by young stars ( in a few cases ) .
# 1#2#1@xmath7 |
quantum chromodynamics ( qcd ) is a gauge field theory that describes the strong interactions of quarks and gluons @xcite .
all experimental results to date are consistent with qcd predictions to within the experimental and theoretical errors . in this review ,
we discuss the current status of the extraction of the strong interaction coupling constant @xmath0 from the experimental data .
the qcd lagrangian describing the interactions of quarks and gluons is @xmath1 where @xmath2 is the gluon field strength tensor , @xmath3 is the gauge covariant derivative , and @xmath4 are the @xmath5 representation matrices normalized so that @xmath6 , and the sum on @xmath7 is over the six different flavors ( @xmath8 ) of quarks . at the classical level , the qcd lagrangian depends on the six quark masses @xmath9 , and the strong interaction coupling constant @xmath10 , or equivalently , the strong fine - structure constant @xmath11 .
the quantum theory contains an additional parameter , the @xmath12-angle , that violates cp .
the experimental limit on this parameter is @xmath13 @xcite , so we will set it to zero for the purposes of this article .
one can evaluate qcd scattering amplitudes in powers of @xmath0 using a feynman diagram expansion . as is typical in a quantum field theory , loop graphs are divergent and need to be treated using a renormalization scheme .
the most commonly used scheme is modified minimal subtraction ( ) @xcite , and we will use this scheme throughout .
an important consequence of renormalization is that the parameters @xmath0 and @xmath9 of the qcd lagrangian depend in a calculable manner on the subtraction - scale @xmath14 .
the @xmath14 dependence of @xmath0 is described by the @xmath15-function , @xmath16 in perturbation theory , @xmath17 where ( for @xmath18 flavors of quarks ) @xmath19 and the next two terms are also known @xcite . if @xmath0 is small , the renormalization group equation eq .
( [ 3 ] ) can be integrated using only the @xmath20 term to give @xmath21 since @xmath22 for @xmath23 , @xmath24 as @xmath25 .
the vanishing of the qcd coupling for large values of @xmath14 is referred to as asymptotic freedom .
one important consequence of asymptotic freedom , is that qcd processes at high energies can be reliably computed in a perturbation expansion in @xmath0 .
a measurable quantity , such as the total cross - section for @xmath26 at high energies can be computed as a function of the qcd coupling constant @xmath27 and the center of mass energy @xmath28 , @xmath29 where @xmath30 is a dimensionless function of its arguments .
the form of the cross - section given in eq .
( [ 2 ] ) follows from dimensional analysis : @xmath31 has dimensions of @xmath32 , and @xmath14 has dimensions of energy . in the sscheme ,
any dependence on @xmath14 is logarithmic , so @xmath30 can only depend on @xmath33 .
the cross section @xmath34 is a measurable quantity and can not depend on the subtraction scale @xmath14 , so the @xmath14 dependence on the right hand side of eq .
( [ 2 ] ) must cancel , @xmath35 and any value of @xmath14 can be used on the right hand side of eq .
( [ 2 ] ) . in practice
, one can only compute the right hand side of eq .
( [ 2 ] ) at some finite order in perturbation theory , and the approximate value of @xmath30 can depend on @xmath14 at higher order in perturbation theory .
typically , one finds that the perturbation expansion has terms of the form @xmath36^n,\ ] ] which are referred to as `` leading logarithms . ''
even if @xmath27 is small , the perturbation expansion can break down if @xmath37 is large .
for this reason , it is conventional to choose the subtraction scale of order the center of mass energy @xmath28 .
the exact choice of scale ( for example , whether @xmath38 or @xmath28 or @xmath39 ) is arbitrary , and differences in choice of scale are formally of higher order in @xmath0 .
many methods have been proposed to determine the optimum scale to use for a given calculation @xcite .
the only way to determine the `` best '' scale at a given order is to compute the cross - section at next order .
[ of course , in this case , one might as well use the more accurate formula to determine the cross - section . ]
the scale dependence of a given quantity can also be used to estimate the size of neglected higher order corrections .
scale dependence is a dominant source of error in many of the quantities that will be used to determine @xmath0 .
perturbation theory is valid if one chooses @xmath14 to be of order @xmath28 , so that the expansion parameter is @xmath40 , with no large logarithms .
this shows that at high - energies , the coupling constant is small because of asymptotic freedom , and qcd cross - sections are approximately those of free quarks and gluons . at low , energies ,
non - perturbative effects become important .
the value of @xmath0 is determined by computing a quantity in terms of @xmath0 , and comparing with its measured value .
one might think that it is better to use high - energy processes to determine @xmath0 , since perturbation theory is more reliable .
this is not necessarily the case .
high energy processes can be computed more reliably precisely because they do not depend very much on @xmath0 .
this means that errors in the experimental measurement or theoretical calculation get amplified when they are converted to an error on @xmath0 .
we will see in this article that low - energy extractions of @xmath0 have comparable errors to those at high energy .
in addition to scale dependence , the coupling constant @xmath0 is also subtraction scheme dependent .
the scheme dependence of the coupling constant is compensated for by the scheme dependence in the functional form for a measurable quantity , so that the value of an observable is scheme independent .
the scheme will be used in this article , but there is still some residual scheme dependence we need to consider . in the scheme , heavy quarks do not decouple in loop graphs at low - energy .
for example , the @xmath15-function coefficient @xmath41 , where @xmath18 is the number of quark flavors .
this expression is true for all energies , irrespective of the mass of the quark .
one might expect that at energies much smaller than the mass @xmath42 of a heavy quark , the quark does not contribute to the @xmath15-function .
this can not happen in the scheme , since is a mass - independent subtraction scheme , which results in a mass - independent @xmath15-function .
what happens is that there are large logarithms of the form @xmath43 that compensate for the `` incorrect '' @xmath15-function at low - energies . in practice ,
one deals with this problem using an effective field theory . at energies smaller than @xmath42 ,
one switches from a qcd lagrangian with @xmath18 flavors to a qcd lagrangian with @xmath44 flavors by integrating out the heavy quark flavor .
the effect of the heavy quark is taken into account by higher dimension operators in the qcd lagrangian , and by shifts in the parameters @xmath0 and @xmath9 .
thus it is necessary to specify the effective theory when quoting the value of @xmath0 .
we will use the notation @xmath45 to denote the value of @xmath0 in the @xmath18 flavor theory .
one can compute the relation between @xmath45 and @xmath46 at the scale @xmath47 of the heavy quark .
this relation is known to three loops @xcite , @xmath48.\label{7}\end{aligned}\ ] ] when quoting the value of @xmath27 , it is also necessary to specify the number of flavors in the effective theory . in most of the @xmath0 determinations we consider ,
the appropriate effective theory to use is one with @xmath49 , and @xmath0 will refer to @xmath50 unless otherwise specified .
classical qcd is a scale invariant theory , but this scale invariance is broken at the quantum level .
the quantum theory has a dimensionful parameter @xmath51 that characterizes the scale of the strong interactions .
the @xmath51 parameter is determined in terms of @xmath27 .
the solution of the renormalization group equation eq .
( [ 3 ] ) including the first three terms in the @xmath15-function is @xmath52 \over \ln \left ( \mu^2 / \lambda^2 \right ) } \right .
\nonumber \\ & & \left .
+ { 4 \beta_1 ^ 2 \over \beta_0 ^ 4 \ln^2 \left ( \mu^2 / \lambda^2 \right ) } \left ( \left ( \ln \left [ \ln \left ( \mu^2 / \lambda^2 \right ) \right]-{1\over 2 } \right)^2 + { \beta_2 \beta_0 \over \beta_1 ^ 2 } - { 5 \over 4 } \right ) \right].\end{aligned}\ ] ] this equation can be used to determine @xmath51 if @xmath0 is known at some scale @xmath14 .
the value of @xmath51 depends on the number of terms retained in eq .
( [ 6 ] ) .
the expansion parameter in eq .
( [ 6 ] ) is @xmath53 \over \ln \left ( \mu^2 / \lambda^2 \right ) } , \ ] ] which is small as long as @xmath54 .
in qcd , @xmath51 is of order 200 mev .
the last term in eq .
( [ 6 ] ) is often dropped in the definition of @xmath51 . for a fixed value of @xmath55 ,
the shift in @xmath51 is approximately 15 mev if the last term in eq .
( [ 6 ] ) is dropped .
the qcd @xmath15-function depends on @xmath18 , and so changes across quark thresholds .
this in turn implies that @xmath51 changes across quark thresholds , so that @xmath56 is the value of @xmath51 with @xmath18 dynamical quark flavors .
the matching conditions for @xmath57 can be computed using the matching condition eq .
( [ 7 ] ) for @xmath0 .
the differences between @xmath58 , @xmath59 , and @xmath60 are numerically very significant .
in addition to the perturbative effects discussed so far , non - perturbative effects play an important role in strong interaction processes .
the size of non - perturbative effects is governed by the ratio of the strong interaction scale @xmath51 to the typical energy @xmath28 of a given process . in many cases ,
non - perturbative effects are estimated using a model analysis , or by a phenomenological fit to the experimental data .
a few processes can be analyzed rigorously using the operator product expansion ( ope)@xcite ; in this case , one can calculate the size of non - perturbative effects in terms of matrix elements of gauge invariant local operators .
two classic examples of this type are deep inelastic scattering , and the total cross - section for @xmath61 .
it is convenient to calculate @xmath62 , the ratio of the total cross - sections for @xmath26 and @xmath63 at center of mass - energy @xmath64 .
one can show that @xmath65 where the first non - perturbative correction depends on the vacuum expectation value of the square of the gluon field - strength tensor . by dimensional analysis , this quantity is of order @xmath66 , so the non - perturbative corrections are of order @xmath67 .
the existence of an ope provides some crucial information on the size of non - perturbative corrections . for @xmath62 , we know that the corrections vanish at least as fast as @xmath67 for large values of @xmath68 , because @xmath69 is the lowest dimension operator that can contribute in the ope . similarly , it is known in deep inelastic scattering that the first non - perturbative corrections arise from twist - four operators , and are of order @xmath70 , where @xmath71 is the momentum transfer .
one can estimate the size of non - perturbative corrections for processes with an ope , by estimating the value of operator matrix elements . in some cases ,
one is fortunate enough that the relevant matrix element can actually be determined from some other measurement , or computed from first principles .
the size of non - perturbative corrections is much less certain if the process does not have an ope .
non - perturbative effects can fall off like a fractional power of @xmath72 , or could have some more complicated dependence on @xmath68 .
typically , one uses some model estimate of the non - perturbative corrections .
perturbative and non - perturbative corrections to scattering cross - sections are interrelated , because the qcd perturbation series is an asymptotic expansion , rather than a convergent expansion . a dimensionless quantity @xmath73 has an expansion of the form @xmath74
typically , the coefficients @xmath75 grow as @xmath76 , so that the series has zero radius of convergence .
the large - order behavior of the perturbation series can be computed in certain limiting cases @xcite .
if one studies qcd in the limit of a large number of flavors , @xmath77 , with @xmath78 fixed , one can sum all terms of the form @xmath79 .
this is sometimes referred to as the `` bubble chain '' approximation , because the graphs one sums are of the type show in fig .
[ fig : bubble ] .
qcd is not asymptotically free as @xmath77 .
nevertheless , one can try and apply the bubble chain results to qcd .
the bubble chain graphs contribute to the qcd @xmath15-function . in the large @xmath18 limit , the coefficient @xmath80 .
one therefore computes the bubble chain sum , makes the replacement @xmath81 , and uses the resultant expression for qcd with @xmath82 .
this seemingly unjustified procedure has provided some useful insights into the nature of the qcd perturbation series .
a detailed discussion of this method is beyond the scope of the present article .
it is typically found that the coefficients @xmath75 in the perturbation expansion have a factorial divergence in the bubble chain approximation .
one can try and sum a series of this type using a borel transformation .
one defines the borel transform of @xmath30 by @xmath83 then the original function can be obtained by the inverse borel transform , @xmath84 suppose the coefficients of @xmath73 have the form @xmath85 then @xmath86 and the inverse borel transform gives @xmath87 the behavior of the integral is governed by the singularities in the complex @xmath88 plane , which are referred to as renormalons . if @xmath89 , the integral is well - defined , since the singularity at @xmath90 is not along the path of integration . if @xmath91 , the singularity is along the contour of integration .
one can regulate the integral by deforming the contour around the singularity .
the integral depends on the precise prescription used .
the prescription dependence is related to the pole at @xmath90 , which gives a contribution to the integral of the form @xmath92=
\exp\left[{-\beta_0 \ln(\mu/\lambda ) \over 2 \pi a } \right ] = \left ( { \lambda \over \mu } \right)^{\beta_0 / 2 \pi a}.\ ] ] the value of @xmath14 is chosen to be the typical energy in the process , such as the momentum transfer @xmath71 for deep inelastic scattering . the perturbative series then has the same structure as a typical non - perturbative correction , a power law correction of the form @xmath93 where @xmath94 is the renormalon singularity in the variable @xmath95 .
[ it is conventional to refer to the location of the renormalon singularity in @xmath96 rather than in @xmath88 .
] contributions of the form eq .
( [ 8 ] ) are called renormalon ambiguities , since their value depends on the way in which one performs the inverse borel transform .
renormalon ambiguities have the same structure as non - perturbative corrections .
it has been suggested in the literature @xcite that renormalons can be used as a guide to the size of non - perturbative corrections .
there is one non - trivial check to this idea .
a renormalon singularity at @xmath97 corresponds to a non - perturbative ambiguity of the form eq .
( [ 8 ] ) . in processes that have an ope ,
all non - perturbative effects should be given by the matrix elements of gauge invariant local operators . to every renormalon ambiguity in the perturbation expansion
, there should be a corresponding ambiguity in the operator matrix element , such that the sum is well - defined @xcite .
this can only happen provided that there is a gauge invariant local operator corresponding to every renormalon singularity .
for example , the first gauge invariant operator corrections to deep inelastic scattering are of the form @xmath70 , and it is known that the first renormalon ambiguity is at @xmath98 .
similarly , for @xmath99 , the first non - perturbative corrections are of the form @xmath100 , and the first renormalon ambiguity is at @xmath101 .
the matching between renormalon singularities and the ope occurs in all examples that have been computed so far .
for this reason , renormalon singularities have also been taken as an indication of the size of non - perturbative effects in processes without an ope .
non - perturbative effects are expected to fall off faster if the renormalons are at larger values of @xmath96 .
infrared sensitivity is also used to estimate the size of non - perturbative corrections to a measurable quantity @xcite .
one computes the quantity in the presence of an infrared cutoff momentum @xmath102 .
for example , one can imagine working in a box of size @xmath103 , or using a gluon mass of order @xmath102 .
cross - sections can have infrared divergences of the form @xmath104 .
if one computes measurable cross - sections for color singlet states to scatter into color singlet states , one finds that the @xmath105 terms cancel , and the cross - section is infrared finite , a result known as the kln @xcite theorem .
it is important to include finite detector resolution to get a finite cross - section , as for qed . for example , the bhabha scattering cross - section for @xmath106 has an infrared divergence at one - loop order .
however , it is impossible to distinguish @xmath106 from @xmath107 if the photon energy @xmath108 is smaller than the detector resolution @xmath109 .
the measurable quantity is the sum of the @xmath110 cross - section and the @xmath111 cross - section for @xmath112 , which is free of infrared singularities . while the @xmath104 term must cancel , terms of order @xmath102 , @xmath113 , etc . which vanish as @xmath114 need not cancel .
the first non - vanishing term is an indication of the infrared - sensitivity of a given quantity . in qcd
, one can imagine that the scale @xmath102 represents the confinement scale @xmath51 . a process that is infrared sensitive at order @xmath115
would then be expected to have non - perturbative corrections of order @xmath116 , where @xmath71 is the typical momentum transfer . in a few cases
, one can analyze the problem using renormalon methods , and by using the criterion of infrared sensitivity .
it is found in these cases that both methods give the same estimate for the size of non - perturbative corrections .
the total cross section for @xmath119 is obtained ( at low values of @xmath120 ) by multiplying the muon - pair cross section by the factor @xmath99 . at lowest order in qcd perturbation theory @xmath121 where @xmath122 is the electric charge of the quark of flavor @xmath123 .
the higher - order qcd corrections to this are known , and the results can be expressed in terms of the factor : @xmath124 \label{rfactor}\ ] ] where @xmath125 and @xmath126@xcite .
this result is only correct in the zero - quark - mass limit .
the @xmath127 ) corrections are also known for massive quarks@xcite .
the principal advantages of determining @xmath0 from @xmath99 in @xmath128 annihilation are that the measurement is inclusive , that there is no dependence on the details of the hadronic final state and that non - perturbative corrections are suppressed by @xmath129 .
a measurement by cleo @xcite at @xmath130 gev yields @xmath131 which corresponds to @xmath132 a comparison of the theoretical prediction of eqn .
[ rfactor ] ( corrected for the @xmath133-quark mass ) , with all the available data at values of @xmath120 between 20 and 65 gev , gives@xcite @xmath134 it should be noted that the size of the order @xmath135 term is of order 40% of that of the order @xmath136 and 3% of the order @xmath0 . if the order @xmath135 term is not included , the extracted value decreases to @xmath137 , a difference smaller than the experimental error .
measurements of the ratio of the hadronic to leptonic width of the @xmath117 at lep and slc , @xmath138 probe the same quantity as @xmath99 . using the average of @xmath139 gives @xmath140@xcite .
the prediction depends upon the couplings of the quarks and leptons to the @xmath117 .
the precision is such that higher order electroweak corrections to these couplings must be included .
there are theoretical errors arising from the values of top - quark and higgs masses which enter in these radiative corrections .
hence , while this method has small theoretical uncertainties from qcd itself , it relies sensitively on the electroweak couplings of the @xmath117 to quarks@xcite and on the ability of the standard model of electroweak interactions to predict these correctly .
the presence of new physics which changes these couplings via electroweak radiative corrections would invalidate the extracted value of @xmath55 .
since the standard model fits the measured @xmath117 properties well , this concern is ameliorated and more precise value of @xmath0 can be obtained by using a global fit to the many precisely measured properties of the @xmath117 boson and the measured @xmath141 and top masses .
this gives @xcite @xmath142 this error is larger than the shift in the value of @xmath55 ( @xmath143 ) that would result if the order @xmath144 term were omitted and hence one can conclude that it is very unlikely that the uncertainty due to the unknown @xmath145 terms will dominate over the experimental uncertainty .
the original and still one of the most powerful quantitative tests of perturbative qcd is the breaking of bjorken scaling in deep - inelastic lepton - hadron scattering .
consider the case of electron - proton scattering ( @xmath146 ) , where the cross - section can be written as @xmath147 \label{eq19}\ ] ] the variables are defined as follows ( see figure [ fig : refd ] ) : @xmath123 is the momentum of the exchanged photon , @xmath148 is the momentum of the target proton , @xmath7 is that of the incoming electron , and @xmath149 for charged current scattering , which proceeds via the exchange of a virtual @xmath141 boson between the lepton and target nucleus , there is an additional parity violating structure function @xmath150 @xmath151 for @xmath152 scattering the sign of the last ( @xmath153 ) term is reversed . in the leading - logarithm approximation , the measured structure functions @xmath154 )
are related to the quark distribution functions @xmath155 according to the naive parton model , for example @xmath156 here @xmath155 is the probability for a parton of type @xmath157 to carry a fraction @xmath158 of the nucleon s momentum .
the @xmath159 dependence of the parton distribution functions is predicted by perturbative qcd , hence a measurement of the @xmath159 dependence ( `` scaling violation '' ) can by used to measure @xmath0 . in describing the way in which scaling is broken in qcd , it is convenient to define nonsinglet and singlet quark distributions : @xmath160 f^ns = q_i - q_j f^s = _ i ( q_i + _ i ) [ fns ] @xmath160 the nonsinglet structure functions have nonzero values of flavor quantum numbers such as isospin or baryon number .
the variation with @xmath159 of these is described by the so - called dglap equations@xcite@xcite : @xmath161 where @xmath162 denotes a convolution integral : @xmath163 the leading - order altarelli - parisi@xcite splitting functions are @xmath164 + 2 \delta ( 1 - x ) \label{apsf;a } \cr p^{qg } & = \frac{1}{2 } \left [ x^2 + ( 1 - x)^2 \right ] \label{apsf;b } \cr p^{gq } & = \frac{4}{3 } \left [ \frac{1 + ( 1 - x)^2}{x } \right ] \label{apsf;c } \cr p^{gg } & = 6 \left [ \frac{1 - x}{x } + x ( 1 - x ) + \frac{x}{(1 - x)_+ } + \frac{11}{12 } \delta ( 1 - x ) \right ] \cr & \qquad -\frac{n_f}{3 } \delta ( 1 - x ) \label{apsf;d } \end{array}\ ] ] here the gluon distribution @xmath165 has been introduced and @xmath166 means @xmath167 measurement of the structure functions over a large range of @xmath158 and @xmath159 allows both @xmath0 and the parton distributions to be determined .
notice that @xmath0 _ and _ the gluon distribution can only be obtained by measuring the @xmath159 dependence .
the precision of contemporary experimental data demands that higher - order corrections also be included@xcite .
the above results are for massless quarks .
algorithms exist for the inclusion of nonzero quark masses@xcite .
these are particularly important for neutrino scattering near the charm threshold . at low @xmath159 values , there are also important `` higher - twist '' ( ht ) contributions of the form : @xmath168 -3pt@xmath169 leading twist ( lt ) terms are those whose behavior can be predicted using the parton model , and are related to the parton distribution functions .
higher - twist corrections depend on matrix elements of higher dimension operators .
these corrections are numerically important only for except for @xmath158 very close to 1 . at very large values of @xmath158 corrections proportional to @xmath170
can become important@xcite . from eqn .
[ altarelliparisi;a ] , it is clear that a nonsinglet structure function offers in principle the most precise test of the theory , since the @xmath159 evolution is independent of the unmeasured gluon distribution .
the ccfr collaboration fit to the gross - llewellyn smith sum rule@xcite which is known to order @xmath135@xcite@xcite ( estimates of the order @xmath171 term are available@xcite ) @xmath172-\delta{ht } \ ; \label{lls } \end{array}\ ] ] where the higher - twist contribution @xmath173 is estimated to be @xmath174 in @xcite@xcite and to be somewhat smaller by @xcite .
the ccfr collaboration@xcite , combines their data with that from other experiments @xcite and gives @xmath175 .
the error from higher - twist terms ( assumed to be @xmath176 ) dominates the theoretical error .
if the higher twist result of @xcite is used , the central value increases to 0.31 in agreement with the fit of @xcite .
this value extrapolates to @xmath177 .
measurements involving singlet - dominated structure functions , such as @xmath178 , result in correlated measurements of @xmath0 and the gluon structure function . a full next
to leading order fit combining data from slac@xcite , bcdms@xcite , e665@xcite and hera@xcitehas been performed @xcite .
these authors extend the analysis to next to next to leading order ( nnlo ) . in this case
the full theoretical calculation is not available as not all the three - loop anomalous dimensions are known ; their analysis uses moments of structure functions . ] and is restricted to those moments where the full calculation is available @xcite .
the nnlo result is @xmath179 .
here the first error is a combination of statistical and systematic experimental errors , and the second error is due to the uncertainties , quark masses , higher twist and target mass corrections , and errors from the gluon distribution .
if only a next to leading order fit is performed then the value decreases to @xmath180 indicating that the theoretical results are stable .
no error is included from the choice of @xmath14 ; @xmath181 is assumed .
we use a total error of @xmath182 to take into account an estimate of the scale uncertainty .
this result is consistent with earlier determinations @xcite , @xcite , and @xcite .
the spin - dependent structure functions , measured in polarized lepton nucleon scattering , can also be used to determine @xmath0 .
the spin structure functions @xmath183 and @xmath184 are defined in terms of the asymmetry in polarized lepton nucleon scattering @xmath185 where the subscript @xmath186 ( @xmath187 ) refers to the state where the nucleon spin is parallel ( anti - parallel ) to its direction of motion in the center of mass frame of the lepton - nucleon system . in both cases
the lepton has its spin aligned along its direction of motion . @xmath188 the @xmath159 evolution of the spin structure functions @xmath189 and @xmath190 is similar to that of the unpolarized ones and is known at next to leading order @xcite.here the values of @xmath191 are small particularly for the e143 data @xcite and higher - twist corrections are important . a fit@xcite using the measured spin dependent structure functions measured by themselves and by other experiments @xcite@xcite gives @xmath192 .
data from hermes @xcite are not included in this fit ; they are consistent with the older data .
@xmath0 can also be determined from the bjorken sum rule@xcite .
@xmath193 at lowest order in qcd @xmath194 .
a fit gives @xcite @xmath195 ; a significant contribution to the error being due to the extrapolation into the ( unmeasured ) small @xmath158 region .
theoretically , the sum rule is preferable as the perturbative qcd result is known to higher order , and these terms are important at the low @xmath159 involved .
it has been shown that the theoretical errors associated with the choice of scale are considerably reduced by the use of pade approximants@xcite which results in @xmath196 corresponding to @xmath197 .
no error is included from the extrapolation into the region of @xmath158 that is unmeasured .
should data become available at smaller values of @xmath158 so that this extrapolation could be more tightly constrained , the sum rule method could provide a better determination of @xmath0 than that from the spin structure functions themselves . at very small values of @xmath158 and @xmath159 , both the @xmath158 and @xmath159 dependence of the structure functions
is predicted by perturbative qcd@xcite . here
terms to all orders in @xmath198 are summed .
the data from hera@xcite on @xmath199 can be fitted to this form@xcite , including the nlo terms which are required to fix the @xmath159 scale .
the data are dominated by @xmath200 .
the fit@xcite using h1 data@xcite gives @xmath201 .
( the theoretical error is taken from @xcite . )
the dominant part of the theoretical error is from the scale dependence ; errors from terms that are suppressed by @xmath202 in the quark sector are included@xcite while those from the gluon sector are not .
typically , @xmath51 is extracted from the deep inelastic scattering data by parameterizing the parton densities in a simple analytic way at some @xmath203 , evolving to higher @xmath159 using the next - to - leading - order evolution equations , and fitting globally to the measured structure functions .
thus , an important by - product of such studies is the extraction of parton densities at a fixed - reference value of @xmath203 .
these can then be evolved in @xmath159 and used as input for phenomenological studies in hadron - hadron collisions ( see below ) .
these densities will have errors associated with the that value of @xmath0 .
a next - to - leading order fit must be used if the process being calculated is known to next - to - leading order in qcd perturbation theory . in such a case , there is an additional scheme dependence ; this scheme dependence is reflected in the @xmath204 corrections that appear in the relations between the structure functions and the quark distribution functions .
there are two common schemes : a deep - inelastic scheme where there are no order @xmath0 corrections in the formula for @xmath205 and the minimal subtraction scheme .
it is important when these next - to - leading order fits are used in other processes ( see below ) , that the same scheme is used in the calculation of the partonic rates .
most current sets of parton distributions are obtained using fits to all relevant data @xcite . in particular , data from purely hadronic initial states are used as they can provide important constraints on the gluon distributions .
experiments in @xmath118 collisions can be used to study photon - photon interactions and to measure the structure function of a photon@xcite , by selecting events of the type @xmath206 which proceeds via two photon scattering .
if events are selected where one of the photons is almost on mass shell and the other has a large invariant mass @xmath71 , then the latter probes the photon structure function at scale @xmath71 ; the process is analogous to deep inelastic scattering where a highly virtual photon is used to probe the proton structure .
the @xmath159 variation of this structure function follows that shown above ( see eq [ altarelliparisi;a ] ) .
a review of the data can be found in @xcite .
data have become available from lep@xcite and from tristan@xcite@xcite which extend the range of @xmath159 to of order 300 gev@xmath207 and @xmath158 as low as @xmath208and show @xmath159 dependence of the structure function that is consistent with qcd expectations .
experiments at hera can also probe the photon structure function by looking at jet production in @xmath209 collisions ; this is analogous to the jet production in hadron - hadron collisions which is sensitive to hadron structure functions .
the data @xcite are consistent with theoretical models @xcite .
measurements of the fragmentation function @xmath210 , the probability that a hadron of type @xmath157 be produced with energy @xmath211 in @xmath118 collisions at @xmath212 , can be used to determine @xmath0 . as in the case of scaling violations in structure functions
, qcd predicts only the @xmath213 dependence in a form similar to the @xmath159 dependence of eq [ altarelliparisi;a ] .
hence , measurements at different energies are needed to extract a value of @xmath0 .
because the qcd evolution mixes the fragmentation functions for each quark flavor with the gluon fragmentation function , it is necessary to determine each of these before @xmath0 can be extracted .
the aleph collaboration has used data in the energy range @xmath214 gev to @xmath215 gev .
a flavor tag is used to discriminate between different quark species , and the longitudinal and transverse cross sections are used to extract the gluon fragmentation function@xcite .
the result obtained is @xmath216@xcite .
the theory error is due mainly to the choice of scale at which @xmath0 is evaluated .
the opal collaboration@xcite has also extracted the separate fragmentation functions .
delphi@xcite has performed a similar analysis using data from other experiments at center of mass energies between 14 and 91 gev with the result @xmath217 .
the larger theoretical error is because the value of @xmath14 was allowed to vary between @xmath218 and @xmath219 .
these results can be combined to give @xmath220 .
an alternative method of determining @xmath0 in @xmath128 annihilation involves measuring the the topology of the hadronic final states .
there are many possible choices of inclusive event shape variables : thrust@xcite , energy - energy correlations@xcite , average jet mass , _
etc._. these quantities must be infrared safe , which means that they are insensitive to the low energy properties of qcd and can therefore be reliably calculated in perturbation theory .
for example , the thrust distribution is defined by @xmath221 where the sum runs over all hadrons in the final state and the unit vector @xmath222 is varied . at lowest order in qcd
the process @xmath223 results in a final state with back to back quarks _
i.e. _ `` pencil - like '' event with @xmath224 .
alternatively , the event can be divided by a plane normal to the thrust axis and the invariant mass of the particles in the two hemispheres is computed , the larger ( smaller ) of these is @xmath225 ( @xmath226 ) . at lowest order in qcd @xmath227 .
the observed final state consists of hadrons rather than the quarks and gluons of perturbation theory .
the hadronization of the partonic final state has an energy scale of order @xmath51 .
the resulting hadrons acquire momentum components perpendicular to the original quark direction of order @xmath51 .
this effect induces corrections to the shape variables of order @xmath228 .
a model is needed to describe the detailed evolution of a partonic final state into one involving hadrons , so that detector corrections can be applied .
furthermore if the qcd matrix elements are combined with a parton - fragmentation model , this model can then be used to correct the data for a direct comparison with the perturbative qcd calculation .
the different hadronization models that are used@xcite model the dynamics that are controlled by non - perturbative qcd effects which we can not yet calculate .
the fragmentation parameters of these monte carlo simulations are tuned to get agreement with the observed data .
the differences between these models can be used to estimate systematic errors .
in addition to using a shape variable , one can perform a jet counting experiment . at order @xmath0 the partonic final state @xmath229 appears which can manifest itself as a three - jet final state after hadronization .
every higher order produce a higher jet multiplicity and measuring quantities that are sensitive to the relative rates of two- , three- , and four - jet events can lead to a determination of @xmath0 .
there are theoretical ambiguities in the way that particles are combined to form jets .
quarks and gluons are massless , whereas the observed hadrons are not , so that the massive jets that result from combining them can not be compared directly to the massless jets of perturbative qcd
. the jet - counting algorithm , originally introduced by the jade collaboration@xcite , has been used by many other groups . here ,
particles of momenta @xmath230 and @xmath231 are combined into a pseudo - particle of momentum @xmath232 if the invariant mass of the pair is less than @xmath233 .
the process is then iterated until no more pairs of particles or pseudo - particles remain .
the remaining number of pseudo - particles is then defined to be the number of jets in the event , and can be compared to the perturbative qcd prediction which depends on @xmath234 . the durham algorithm is slightly different : in computing the mass of a pair of partons , it uses @xmath235 for partons of energies @xmath236 and @xmath237 separated by angle @xmath238@xcite .
different recombination schemes have been tried , for example combining 3-momenta and then rescaling the energy of the cluster so that it remains massless .
these varying schemes result in the same data giving slightly different values@xcite@xcite of @xmath0 .
these differences can be used to estimate a systematic error .
however , such an error may be conservative as it is not based on a systematic approximation .
the starting point for all these quantities is the multijet cross section .
for example , at order @xmath0 , for the process @xmath239:@xcite @xmath240 where @xmath241are the center - of - mass energy fractions of the final - state ( massless ) quarks .
the order @xmath136 corrections to this process have been computed , as well as the 4-jet final states such as @xmath242@xcite .
a distribution in a `` three - jet '' variable , such as those listed above , is obtained by integrating this differential cross section over an appropriate phase space region for a fixed value of the variable .
thus @xmath243 , @xmath244 and @xmath245 .
the result of this integration depends explicitly on @xmath0 but scale @xmath14 at which @xmath27 is to be evaluated is not clear . in the case of jet counting ,
the invariant mass of a typical jet ( or @xmath246 ) is probably a more appropriate choice than the @xmath118 center - of - mass energy .
while there is no justification for doing so , if the value of @xmath14 is allowed to float in the fit to the data , the fit improves and the data tend to prefer values of order @xmath247 gev for some variables @xcite@xcite ; the exact value depends on the variable that is fitted . typically experiments
assign a systematic error from the choice of @xmath14 by varying it by a factor of 2 around the value determined by the fit .
the choice of this factor is arbitrary estimates for the non - perturbative corrections to @xmath248 have been made @xcite using an operator product expansion .
@xmath249 where a and b known quantities @xcite , @xmath14 is the renormalization scale and @xmath250 is the non - perturbative parameter ( the matrix element of an appropriate operator ) to be determined from experiment .
note that the corrections are only suppressed by @xmath251 .
this provides an alternative to the use of hadronization models for estimating these non - perturbative corrections .
the delphi collaboration @xcite uses data below the @xmath117 mass from many experiments and eq .
[ eq : thrust ] to determine @xmath252 , the error being dominated by the choice of scale .
the values of @xmath0 and the non - perturbative parameter @xmath250 are also determined by a fit to using the variable @xmath253 . while the extracted values of @xmath55 are consistent with each other , the values of @xmath250 are not .
the analysis is useful as one can directly determine the size of the @xmath254 corrections ; they are approximately 20% ( 50% ) of the perturbative result at @xmath255 gev . even at @xmath215 gev the omission of these perturbative terms will cause a shift on the extracted value of @xmath0 of @xmath256 which is much larger than the quoted experimental errors .
the perturbative qcd formulae can break down in special kinematical configurations .
for example , the first term in eq .
[ eq : thrust ] contains a term of the type @xmath257 .
the higher orders in the perturbation expansion contain terms of order @xmath258 . for @xmath259 ( the region populated by 2-jet events ) ,
the perturbation expansion in @xmath0 is unreliable .
the terms with @xmath260 can be summed to all orders in @xmath0@xcite .
if the jet recombination methods are used , higher - order terms involve @xmath261 , these too can be resummed@xcite .
the resummed results give better agreement with the data at large values of @xmath262 .
some caution should be exercised in using these resummed results because of the possibility of overcounting ; the showering monte carlos that are used for the fragmentation corrections also generate some of these leading - log corrections .
different schemes for combining the order @xmath136 and the resummations are available@xcite .
these different schemes result in shifts in @xmath55 of order @xmath263 .
the use of the resummed results improves the agreement between the data and the theory .
studies on event shapes have been undertaken at lower energies at tristan , pep / petra , and cleo .
a combined result from various shape parameters by the topaz collaboration gives @xmath264 , using the fixed order qcd result , and @xmath265 ( corresponding to @xmath266 ) where the error is dominated by scale and fragmentation uncertainties .
the cleo collaboration fits to the order @xmath136 results for the two jet fraction at @xmath267 gev , and obtains @xmath268@xcite .
the dominant systematic error arises from the choice of scale ( @xmath14 ) , and is determined from the range of @xmath0 that results from fit with @xmath269 gev , and a fit where @xmath14 is allowed to vary to get the lowest @xmath270 .
the latter results in @xmath271 gev .
since the quoted result corresponds to @xmath272 , it is by no means clear that the perturbative qcd expression is reliable and the resulting error should , therefore , be treated with caution .
a fit to many different variables as is done in the lep / slc analyses would give added confidence to the quoted error .
recently studies have been carried out at energies between @xmath273130 gev@xcite and @xmath273189 gev @xcite .
these can be combined to give @xmath274 and @xmath275 .
the dominant errors are theoretical and systematic and , as most of these are in common at the different energies , these data , those at the @xmath117 resonance and lower energy provide very clear confirmation of the expected decrease in @xmath0 as the energy is increased .
a combined analysis of the data between 35 and 189 gev using data from opal and jade@xcite using a large set of shape variables shows excellent agreement with @xmath276 .
a comparison of this result with those at the @xmath117 resonance from sld@xcite , opal@xcite , l3@xcite , aleph@xcite , and delphi@xcite , indicates that they are all consistent with this value .
the experimental errors are smaller than the theoretical ones arising from choice of scale @xmath14 and modeling of non - perturbative effects , which are common to all of the experiments .
the sld collaboration @xcite determines the allowed range of @xmath14 by allowing any value that is consistent with the fit .
this leads to a larger error ( @xmath277 ) than that obtained by delphi @xcite who vary @xmath14 by a factor of 2 around the best fit value and obtain @xmath278 .
we elect to use a more conservative average of @xmath279 . at lowest order in @xmath0 ,
the @xmath280 scattering process produces a final state of ( 1 + 1 ) jets , one from the proton fragment and the other from the quark knocked out by the underlying process @xmath281 . at next order in @xmath0 ,
a gluon can be radiated , and hence a ( 2 + 1 ) jet final state produced . by comparing the rates for these ( 1 + 1 ) and ( 2 + 1 ) jet processes
, a value of @xmath0 can be obtained .
a nlo qcd calculation is available@xcite .
the basic methodology is similar to that used in the jet counting experiments in @xmath118 annihilation discussed above . unlike those measurements , the ones in @xmath282 scattering are not at a fixed value of @xmath159 .
in addition to the systematic errors associated with the jet definitions , there are additional ones since the structure functions enter into the rate calculations .
results from h1@xcite and zeus@xcite can be combined to give @xmath283 .
the contributions to the systematic errors from experimental effects ( mainly the hadronic energy scale of the calorimeter ) are comparable to the theoretical ones arising from scale choice , structure functions , and jet definitions .
the theoretical errors are common to the two measurements ; therefore , we have not reduced the systematic error after forming the average .
the coupling constant @xmath0 can be determined from an analysis of hadronic @xmath284 decays @xcite .
the quantity that will be used is the ratio @xmath285 where @xmath286 represents possible electromagnetic radiation , or lepton pairs . in the absence of radiative corrections , the ratio @xmath287 is @xmath288 where @xmath289 is the number of colors .
the experimental value @xmath290 is close to three , which is experimental evidence for the existence of three colors in qcd .
the deviation of @xmath287 from three is used to extract @xmath0 .
the weak decay lagrangian for non - leptonic @xmath284 decay is @xmath291,\end{aligned}\ ] ] where @xmath292 and @xmath293 are the ckm mixing angles .
the lagrangian eq .
( [ 9 ] ) is obtained at the scale @xmath294 by integrating out the @xmath141 boson to generate a local four - fermion operator in the effective theory below @xmath295 , and @xmath296 at @xmath294 .
the typical momentum transfer in @xmath284 decays is of order @xmath297 , so it is necessary to scale the lagrangian eq .
( [ 9 ] ) from @xmath294 to @xmath298 .
electromagnetic interactions renormalize the lagrangian . at one - loop , the renormalization from graphs shown in fig .
( [ fig : tauren ] ) produce a multiplicative renormalization of the lagrangian , and give @xmath299 the @xmath284 decay amplitude @xmath300 , where @xmath301 is the final hadronic state , can be written as @xmath302.\ ] ] squaring the amplitude , and computing the decay rate gives @xmath303\nonumber \\ & & \times ( 2\pi)^4 \delta^4(p_\tau - p_\nu - p_x ) { { \left\langle{0}\vphantom{\bar u \gamma_\nu p_l d x } \right|{\bar u \gamma_\nu p_l d } \left|\vphantom{0 \bar u \gamma_\nu p_l d } { x}\right\rangle}}{{\left\langle{x}\vphantom{\bar d \gamma_\mu p_l u 0 } \right|{\bar d \gamma_\mu p_l u } \left|\vphantom{x \bar d \gamma_\mu p_l u } { 0}\right\rangle } } , \label{13a}\end{aligned}\ ] ] where we have retained only the @xmath293 term for simplicity .
the sum on @xmath301 is symbolic for the sum over all final states , including phase space factors .
the @xmath109 function can be written as @xmath304 and the sum on @xmath301 can be written as @xmath305 the tensor @xmath306 is related to another quantity @xmath307 , defined by @xmath308 where @xmath309 . inserting a complete set of states in the time - ordered product ,
one finds that @xmath310 the tensor @xmath311 depends on the only variable , @xmath123 , and must have the form @xmath312 by lorentz invariance .
the tensor @xmath306 is then given by @xmath313 where @xmath314 if the light quark mass difference @xmath315 and @xmath316 are neglected , the hadronic currents @xmath317 and @xmath318 are conserved .
this implies that @xmath319 , so that @xmath320 .
inserting eq .
( [ 12 ] ) and eq .
( [ 11 ] ) into eq .
( [ 13 ] ) gives @xmath321 , \label{13}\end{aligned}\ ] ] where @xmath322 and we added back the @xmath292 contribution . there is no interference term ( at lowest order in the weak interactions ) , because the @xmath323 and @xmath324 currents lead to final states with different flavor quantum numbers , and the strong interactions conserve flavor .
the hadronic invariant mass distribution can then be written as @xmath325 \label{15}\end{aligned}\ ] ] the ratio of the hadronic to leptonic decay rate of the @xmath284 is given by @xcite @xmath326\ ] ] the hadronic tensors @xmath327 are analytic in the complex @xmath68 plane , except for a branch cut along the positive real axis .
the discontinuity across the cut is @xmath328 , and is the cross - section for the currents to create hadrons . clearly , the hadron production rate is sensitive to non - perturbative effects , and can not be computed reliably .
far away from the physical cut , there are no infrared singularities , and qcd perturbation theory is valid .
one can rewrite the integral eq .
( [ 20 ] ) as @xmath329,\ ] ] where @xmath330 is the contour shown in fig .
( [ fig : taucontour ] ) .
the difference of @xmath331 above and below the cut is @xmath332 , so this gives back eq .
( [ 20 ] ) . since @xmath331
have no singularities in the complex @xmath68 plane other than the branch cut , one can deform the contour @xmath333 to the contour @xmath334 , and use eq .
( [ 21 ] ) with @xmath335 .
the advantage of using eq .
( [ 21 ] ) with @xmath335 rather than @xmath330 is that one needs to know @xmath331 far away from the cut for most of the integration contour .
the contour approaches the cut at @xmath336 , but at this point , the integrand vanishes as @xmath337 , so the contribution of the region near @xmath336 to the total integral is suppressed @xcite .
@xmath338 can be computed in perturbation theory using an operator product expansion , which is valid away from the physical cut .
the perturbation theory result for @xmath338 is then substituted in eq .
( [ 21 ] ) . in practice ,
the calculation can be simplified by using the perturbation theory value for @xmath339 in eq .
( [ 20 ] ) .
we have argued above that perturbation theory is not valid for @xmath339 . nevertheless , using the perturbation theory value for @xmath339 in eq .
( [ 20 ] ) is justified because @xmath338 in perturbation theory has the same analytic structure in qcd . thus using the perturbation theory value of @xmath338 in eq .
( [ 21 ] ) is equivalent to using the perturbation theory value for @xmath339 in eq .
( [ 20 ] ) , even though the perturbative computation of @xmath339 is not valid .
the ope for @xmath338 is closely related to that for @xmath26 , which depends on the time - ordered product of two electromagnetic currents .
@xmath340 where @xmath341 are the coefficient functions , and @xmath342 are the local operators .
since the contour @xmath334 is a circle of radius @xmath343 in the complex @xmath68 plane , one expects that logarithms in the uncalculated higher order corrections are minimized if one chooses @xmath298 .
the leading order operator is the unit operator . in the limit that the light quark masses are neglected , @xmath344 vanishes , and we only need to compute @xmath345 , giving @xmath346.\end{aligned}\ ] ] the first coefficient is the well - known result that the ratio @xmath347 is @xmath348 .
the next two coefficients are @xmath349 using the @xmath15-function to write @xmath350 in terms of @xmath351 , evaluating the @xmath68 integral , and setting @xmath352 gives @xcite @xmath353\label{50}\end{aligned}\ ] ] assuming @xmath354 , the series in brackets is @xmath355 , so the terms are still numerically decreasing till order @xmath356 .
one can use the value of the last term as an estimate of the theoretical uncertainty in the perturbative value for the coefficient of the unit operator .
this also tells us that we can neglect corrections from higher dimension operator that are smaller than about @xmath357 .
the @xmath358 corrections to @xmath287 arise from the quark mass corrections to the coefficient of the unit operator in the ope . at order @xmath359 ,
the currents are no longer conserved , so one needs to compute both @xmath344 and @xmath345 .
the only light quark mass contribution of any significance is the @xmath68-quark mass correction@xcite , @xmath360.\ ] ] using @xmath361 mev as an estimate for the @xmath68-quark mass gives @xmath362 , which is smaller than the error in the perturbation series .
the @xmath363 and corrections in the ope arise from the dimension four operators @xmath364 and @xmath365 , and the @xmath366 corrections from the four - quark operators @xmath367 , where @xmath368 is some combination of @xmath369 matrices .
an analysis of these corrections , based on model estimates of the operator matrix elements indicates that these corrections are smaller than the uncertainty in the perturbation series@xcite .
the size of non - perturbative corrections can be determined directly from the experimental data . instead of considering the integral eq .
( [ 20 ] ) that gives the total hadronic width , one compares the integral of @xmath370 weighted with @xmath371 with the corresponding moment of the experimental data . by studying the moments for different values of @xmath7 and @xmath372
, one finds that the non - perturbative corrections are about 3% @xcite , and so are comparable in size to the uncertainty in the perturbation series .
the experimentally measured quantity is @xmath373
@xcite , the ratio for @xmath284 to decay into non - strange hadrons to the leptonic decay rate .
this is given by eq .
( [ 50 ] ) , dropping @xmath292 , and gives @xmath374 where we have assumed a theoretical uncertainty of 100% in the @xmath356 term .
this value corresponds to @xmath375
the strong coupling constant @xmath0 can be determined from lattice gauge theory calculations of the hadronic spectrum .
the basic procedure used is to choose a definition of @xmath0 , and measure its value on the lattice .
one then has to set the scale at which @xmath0 takes on the measured value .
the lattice scale can be normalized using the hadronic spectrum measured on the same lattice . finally , one has to convert the lattice definition of @xmath0 to the value defined in the continuum in a scheme such as .
there are several sources of systematic errors that limit the current accuracy in determining @xmath0 .
typically , @xmath0 is determined by determining the spectrum of heavy quark bound states on the lattice .
there are corrections due to the finite volume and finite lattice spacing @xmath376 .
the finite lattice spacing errors can be reduced by using improved actions , that are accurate to higher order in @xmath376 . to some extent
, one can estimate the error due to finite volume and finite lattice spacing by repeating the simulation on a larger lattice .
the dominant systematic uncertainty is due to the quenched approximation , in which light quark loops are neglected .
it is difficult to reliably estimate the systematic errors due to this approximation without doing a full simulation including dynamical fermions .
simulations with dynamical fermions are just starting to be done , and in a few years one should have more reliable estimates of @xmath0 .
there is one important advantage to using a heavy quark system such as the @xmath377 to determine @xmath0 .
the leading correction to the @xmath377 energy levels due to the light quark masses is linear in the quark masses , and can only depend on the flavor singlet combination @xmath378 .
thus the light quark mass corrections can be computed to a good approximation using three light quarks of mass @xmath379 .
this avoids having to simulate almost massless dynamical quarks , which is very difficult .
lattice calculations can also be used to test theoretical calculations , and determine the regime in which perturbation theory is applicable . in the quenched approximation
, one can study the scale dependence of the coupling constant on the lattice .
this provides a check on the perturbation theory calculation with @xmath380 .
the result is in remarkable agreement with the perturbation theory result in the regime where the coupling constant is weak @xcite .
the fermilab and scri groups use the @xmath381 and @xmath382 splittings in the @xmath377 system to determine @xmath0 .
there are some systematic deviations of the calculated numbers from their experimental values in the quenched approximation ( @xmath380 ) , which are dramatically reduced if one includes @xmath383 dynamical flavors .
the value of @xmath55 in the scheme is @xcite @xmath384 where the first error is due to discretization effects , relativistic corrections , and statistical errors , the second is due to dynamical fermions , and the third is from conversion uncertainties .
more recent computations give ( in the @xmath49 scheme ) @xcite @xmath385 and @xcite @xmath386 an average of these newer values gives @xmath387 , where we have included the difference between the two central values as an estimated additional systematic error .
heavy quark bound states such as the @xmath377 can also be used to extract a value for @xmath0 .
if the bound state is treated using non - relativistic quantum mechanics , the annihilation decays @xmath388 , @xmath389 and @xmath390 can be computed as the product of the probability to find the quark - antiquark pair at the origin , times the annihilation rate for @xmath391 at rest to decay to the final state .
the relevant feynman graphs are shown in fig .
[ fig : upsdecay ] .
the decay rate @xmath389 is the inclusive decay rate for @xmath392 , and the decay rate for @xmath390 is that for @xmath393 .
the probability to find @xmath391 at the origin , @xmath394 , is sensitive to the detailed dynamics of the @xmath391 bound state .
if one takes the ratio of decay rates , @xmath394 drops out , and the ratio of decay rates can be used to determine @xmath0 . the above qualitative discussion can be made precise using nrqcd ( non - relativistic qcd ) to calculate the properties of the @xmath377 @xcite .
nrqcd has an expansion in powers of @xmath395 , the velocity of quarks in the bound state .
the @xmath0 expansion is coupled to the @xmath395 expansion , since @xmath396 in a coulombic system .
the nrqcd approach allows one to systematically factor the decay rate into short distance coefficients that are calculable in perturbation theory , and non - perturbative hadronic matrix elements that generalize the notion of the wavefunction at the origin .
the @xmath377 wavefunction in nrqcd has different fock components .
the lowest order ( in @xmath395 ) component is @xmath397 and the first correction contains a gluon , @xmath398 . the @xmath398 is referred to as the color - octet component , because the two quarks are in a color octet state .
the nrqcd velocity counting rules show that the probability to find the @xmath377 in the octet component is of order @xmath399 .
# 11.5ex-16.5mu # 1 the decay rate for @xmath388 in nrqcd is @xcite @xmath400 where @xmath401 is the @xmath133-quark pole mass , @xmath402 and @xmath403 , and @xmath404 and @xmath405 annihilate quarks and antiquarks , respectively . the first term is the leading order contribution , and the second term is the @xmath399 correction . the coefficients @xmath406 and @xmath407 can be computed from the first graph in fig .
[ fig : upsdecay ] at lowest order in perturbation theory .
the values are @xcite @xmath408,\qquad \im g_{ee}({}^3s_1)=-{4\pi \alpha^2 e_b^2\over 9},\ ] ] where @xmath409 is the charge of the @xmath133-quark , and the radiative correction to @xmath406 has also been included .
the decay rate for @xmath410 is @xcite @xmath411 where @xmath412 is the contribution to the decay rate from the color - octet component of the @xmath377 . in nrqcd ,
the color octet decay rate is @xmath413 suppressed relative to the color singlet decay rate .
however , the color octet component can decay into two gluons , rather than three , so the color octet decay rate of order @xmath414 can compete with the relativistic correction to the color singlet decay rate of order @xmath415 .
the coefficient functions are @xcite @xmath416\nonumber \\
\im g_1({}^3s_1)&= & -{5 \over 1458}(19\pi^2 - 132)\alpha_s^3(m_b)\end{aligned}\ ] ] the @xmath135 term for @xmath417 has the given value when the scale of the @xmath136 term is the @xmath133-quark pole mass .
the decay rate @xmath418 has the form @xmath419 where the coefficient function is @xcite @xmath420\end{aligned}\ ] ] the equations of motion can be used to relate the matrix elements of the @xmath421-wave and @xmath148-wave operators @xcite , @xmath422 the matrix element of the @xmath421-wave operator is the nrqcd analog of the wavefunction at the origin in a potential model calculation .
the matrix element is non - perturbative , and can be eliminated by considering ratios of decay rates .
the matrix element of @xmath423 can be determined from eq .
( [ gk ] ) using estimates of the @xmath133-quark pole mass @xmath401 , and the measured @xmath377 mass , as was done in ref
. @xcite .
however , one can instead replace @xmath424 by @xmath425 , the lowest - order result for the binding energy for a coulomb bound state .
this reduces somewhat the uncertainty in the extraction of @xmath0 , since it eliminates any uncertainty from the pole mass .
the experimental value of the ratio @xmath426 @xcite , where @xmath427 gives @xmath428 , using the ratio of the theoretical formula for the decay widths .
the unknown octet decay rate has been estimated to be less than 9% in ref .
@xcite , and this has been included as a theoretical uncertainty .
the decay rates eq .
( [ 51])([53 ] ) have been written in terms of @xmath429 .
one can instead rewrite them in terms of @xmath27 using eq .
( [ af ] ) , extract @xmath27 , and convert this into @xmath429 using eq .
( [ af ] ) .
we have include the uncertainty of a scale change by a factor of two in the theoretical estimate .
the octet and scale uncertainties are comparable in size . the experimental value of the ratio @xmath430 @xcite can also be used to extract @xmath0 .
it is convenient to use the experimental value of @xmath431 to convert this to @xmath432 , before comparing with the theoretical results .
this eliminates the theoretical uncertainty due to the octet component in the hadronic decay rate .
the extracted value of @xmath429 is @xmath433 , where we have included a scale uncertainty as above .
averaging the two extractions gives @xmath434 which corresponds to @xmath435
there are many process at high - energy hadron - hadron colliders which can constrain the value of @xmath0 .
all rely on the qcd improved parton model , and on the factorization theorems of qcd @xcite .
the rate for any process is expressed as a convolution of the partonic scattering amplitude @xmath436 and parton distribution functions discussed in section [ sec : struc ] ; see eq [ eq : parton ] ( note that here we use @xmath30 rather than @xmath123 as the sum on @xmath437 runs over quarks and gluons . @xmath438
the factorization scale @xmath439 is arbitrary . as in the case of the scale @xmath14 used in @xmath27 ( see eq [ eq : mu ] and the surrounding discussion ) ,
the exact result can not depend on its choice .
however as the processes @xmath436 is only calculated to some finite order in perturbation theory , some residual @xmath439 dependence will remain . as in the case of @xmath14
the sensitivity to @xmath439 will be small if it is chosen to be a characteristic scale of the process ; for example , in the case of the production of a pair of jets of momentum , @xmath440 , transverse to the direction defined by the incoming hadrons , @xmath441 is a reasonable choice .
the quantitative tests of qcd and the consequent extraction of @xmath0 which appears in @xmath436 are possible only if the process in question has been calculated beyond leading order in qcd perturbation theory .
the production of hadrons with large transverse momentum in hadron - hadron collisions provides a direct probe of the scattering of quarks and gluons : @xmath442 , @xmath443 , @xmath444 , _
etc._. here the leading order term in @xmath436 is of order @xmath136
so the rates are sensitive to its value .
higher order qcd calculations of the jet rates@xcite and shapes are in impressive agreement with data@xcite .
this agreement has led to the proposal that these data could be used to provide a determination of @xmath0@xcite .
a set of structure functions is assumed and tevatron collider data are fitted over a very large range of transverse momenta , to the qcd prediction for the underlying scattering process that depends on @xmath0 . the evolution of the coupling over this energy range ( 40 to 250 gev ) is therefore tested in the analysis .
cdf obtains @xmath445 @xcite .
estimation of the theoretical errors is not straightforward .
the structure functions used depend implicitly on @xmath0 and an iteration procedure must be used to obtain a consistent result ; different sets of structure functions yield different correlations between the two values of @xmath0 .
we estimate an uncertainty of @xmath446 from examining the fits .
@xcite estimates the error from unknown higher order qcd corrections to be @xmath446 . combining these
then gives : @xmath447 qcd corrections to drell - yan type cross sections ( the production in hadron collisions by quark - antiquark annihilation of lepton pairs of invariant mass @xmath71 from virtual photons or of real @xmath141 or @xmath117 bosons ) , are known@xcite .
these processes are not very sensitive to @xmath0 as the leading piece in @xmath436 is of order @xmath448 .
the production of @xmath141 and @xmath117 bosons and photons at large transverse momentum begins at order @xmath448 .
the leading - order qcd subprocesses are @xmath449 and @xmath450 .
the next - to - leading - order qcd corrections are known@xcite@xcite for photons , and for @xmath451 production@xcite , and so an extraction of @xmath0 is possible in principle .
data exist on photon production from the cdf and d collaborations@xcite@xcite and from fixed target experiments @xcite .
detailed comparisons with qcd predictions @xcite may indicate an excess of the data over the theoretical prediction at low value of transverse momenta , although other authors @xcite find smaller excesses .
these differences indicate that while the process may be understood , no meaningful extraction of @xmath0 is possible .
the ua2 collaboration@xcite has extracted a value of @xmath452 from the measured ratio @xmath453 .
the result depends on the algorithm used to define a jet , and the dominant systematic errors due to fragmentation and corrections for underlying events ( the former causes jet energy to be lost , the latter causes it to be increased ) are connected to the algorithm .
the scale at which @xmath454 is to be evaluated is not clear .
a change from @xmath294 to @xmath455 causes a shift of 0.01 in the extracted @xmath0 , and the quoted error should be increased to take this into account .
there is also dependence on the parton distribution functions , and hence , @xmath0 appears explicitly in the formula for @xmath456 , and implicitly in the distribution functions .
data from cdf and d on the w @xmath440 distribution @xcite are in agreement with qcd but are not able to determine @xmath0 with sufficient precision to have any weight in a global average .
the production rates of @xmath133 quarks in @xmath457 have been used to determine @xmath0 @xcite .
the next to leading order qcd production processes @xcite have been used . at order @xmath0
the production processes are @xmath458 and @xmath459 result in b - hadrons that are back to back in azimuth . by selecting events in this region the next - to leading order calculation
can be used to compare rates to the measured value and a value of @xmath0 extracted .
the errors are dominated by the measurement errors , the choice of @xmath14 and @xmath439 , and uncertainties in the choice of structure functions .
the last were estimated by varying the structure functions used .
the result is @xmath460 .
the previous sections have illustrated the large number of processes where quantitative tests of qcd can be made and a value of @xmath0 extracted .
figure [ figalphas ] shows the values of @xmath55 deduced from the various processes shown above .
the consistency and precision of these results is remarkable .
figure [ runfigalpha ] shows the values of @xmath27 and the values of @xmath14 where they are measured .
this figure clearly shows the experimental evidence for the variation of @xmath27 with @xmath14 predicted by eq.[3 ] .
an average of the values in figure [ figalphas ] and in table [ tab1 ] gives @xmath461 , with a total @xmath270 of 9 for twelve fitted points , showing good consistency among the data .
the value from heavy quark systems contributes slightly more that one half of the total @xmath270 . if this result is omitted the average increases to 0.1185 .
all of the other results are within @xmath462 of the average value .
the error on the average , assuming that all of the errors in the contributing results are uncorrelated , is @xmath463 , and may be an underestimate .
we have seen that in almost all of the cases discussed , the errors are dominated by systematic , usually theoretical errors . only some of these , notably from the choice of scale , are correlated .
it is important to note that the average is not dominated by a single measurement ; there are many results with comparable small errors : from @xmath284 decay , lattice gauge , theory deep inelastic scattering and the @xmath464 width .
we quote our average value as @xmath465 , which corresponds to @xmath466 mev using eq .
the reader may wish to consult other recent articles for different opinions@xcite .
significant improvements in the precision in the near future are not likely .
the accuracy of data from lep will not improve .
it is possible that a better understanding of the jet rates in hadron - hadron colliders and a systematic treatment of the errors from the structure functions will lead to and improvement in the precision of the value of @xmath0 derived . in many cases where the data are quite precise , such as heavy quark system , theoretical uncertainties limit the precision . in the very long term precision at the 1% level
may be achievable@xcite .
this work was supported by the director , office of science , office of basic energy services , of the u.s .
department of energy under contract de - ac03 - 76sf0098 , and by doe grant doe - fg03 - 97er40546 .
299 h. d. politzer , 30,1346 ( 1973 ) ; d. j. gross and f. wilczek , 9,980(1974 ) r. d. peccei and h. r. quinn , 38,1440(1977 ) ; s. weinberg , 37,657(1977 ) w.a .
bardeen , 18,3998(1978 ) g. grunberg , 95b,70(1980 ) ; 29,2315(1984);p.m .
stevenson , 23,2916(1981 ) ; and 203,472(1982 ) ; s. brodsky and h.j .
lu , slac - pub-6389 ( nov . 1993 ) ; s. brodsky , g.p .
lepage , and p.b .
mackenzie , 28,228(1983 ) s.a .
larin , t. van ritbergen , and j.a.m .
vermaseren , 400,379(1997 ) k. g. chetyrkin , b. a. kniehl and m. steinhauser , 510,61(1998 ) k. wilson 179,1499 ( 1969 ) b. lautrup , 69,109(1977 ) g. t hooft , in _ the why s of subnuclear physics _ , ed . by a.zichichi , ( plenum , new york , 1978 ) f. david , 209,433(1982 ) , 234,237(1984 ) i.i .
bigi , m. shifman , n.g .
uraltsev and a. vainshtein , 50,2234(1994 ) m. beneke and v. braun , 426,301(1994 ) m. beneke,317,1(1999 ) t. d. lee and m. nauenberg , 133,b1549 ( 1964 ) ; t. kinoshita , j. math .
phys . * 3 * 650 ( 1962 ) a. sinkovics , r. akhoury and v. i. zakharov , 58,114025(1998 ) s.g .
gorishny , a. kataev , and s.a .
larin , 259,114(1991 ) l.r .
surguladze and m.a .
samuel , 66,560(1991 ) k.g .
chetyrkin and j.h .
kuhn , 308,127(1993 ) r. ammar , 57,1350(1998 ) d. haidt , in _ directions in high energy physics , _ vol .
14 , p. 201
p. langacker ( world scientific , 1995 ) g. quast , presented at the european physical society meeting , tampere , finland ( july 1999 ) a. blondel and c. verzegrassi , 311,346(1993 ) g. altarelli , 405,3(1993 ) j. erler and p langacker , brief review in eur.phys.j.c3:1-794,1998 .
v.n . gribov and l.n .
lipatov , 15,438(1972 ) yu.l .
dokshitzer , 46,641(1977 ) g. altarelli and g. parisi , 126,298(1977 ) g. curci , w. furmanski , and r. petronzio , 175,27(1980 ) r. petronzio , 97b,437(1980 ) ; and 11,293(1982 ) e.g. floratos , c. kounnas , and r. lacaze , 98b,89(1981 ) ; 98b,285(1981 ) ; and 192,417(1981 ) r.t .
herrod and s. wada , 96b,195(1981 ) ; and 9,351(1981 ) m. glck , 13,119(1982 ) g. sterman , 281,310(1987 ) s. catani and l. trentadue , 327,323(1989 ) ; b353,183(1991 ) d. gross and c.h .
llewellyn smith , 14,337(1969 ) j. chyla and a.l .
kataev , 297,385(1992 ) s.a .
larin and j.a.m .
vermaseren , 259,345(1991 ) a.l .
kataev and v.v .
starchenko , 10,235(1995 ) v.m .
braun and a.v .
kolesnichenko , 283,723(1987 ) m. dasgupta and b. webber , 382,273(1993 ) j. kim , 81,3595(1998 ) d. allasia , 28,321(1985 ) k. varvell , 36,1(1997 ) v.v .
ammosov , 30,175(1986 ) p.c .
bosetti , 142,1(1978 ) a.l .
kataev , hep - ph9907310l.w .
whitlow , 282,475(1992 ) a.c .
benvenuti , 223,490(1989 ) ; 223,485,(1989 ) ; 237,592(1990 ) ; and 237,599(1990 ) m.r .
adams , 54,3006(1996 ) m. derrick , 345,576(1995 ) ; 62,399(1999 ) j. santiago and f.j .
yndurain , hep - ph/9904344k .
adel , f. barriero , and f.j .
yndurain , 495,221(1997 ) w.l .
van neerven , and e.b .
zijlstra , 272,127(1991 ) ; 383,525(1992 ) s. a. larin , 427,41(1994 ) ; 492,338(1997 ) k. bazizi and s.j .
wimpenny , ucr / dis/91 - 02 m. virchaux and a. milsztajn , 274,221(1992 ) p.z .
quintas , 71,1307(1993 ) k. abe , 74,346(1995 ) ; 364,61(1995 ) ; 75,25(1995 ) b. adeva , 58,112002(1998 ) d. adams , 329,399(1995 ) ; 56,5330(1998 ) ; 58,1112001(1998 ) a. airapetian , 442,484(1998 ) , hep - ex/99 - 06035j.d .
bjorken , 148,1467(1966 ) r. mertig and w. l. van neerven , 70,637(1996 ) g. altarelli , 496,337(1997 ) , hep - ph/9803237 j. ellis , 54,6986(1996 ) a. derujula , 10,1669(1974 ) e.a .
kurayev , l.n .
lipatov , and v.s .
fadin , 45,119(1977 ) .
ya.ya . balitsky and l.n .
lipatov , 28,882(1978 ) r.d .
ball and s. forte , 335,77(1994 ) ; 336,77(1994 ) h1collaboration : s. aid , 470,3(1996 ) r.d .
ball and a. deroeck , hep - ph/9609309european physical society meeting , brussels , ( july 1995 ) * h1 collaboration : t. ahmed , 439,471(1995 ) s. catani and f. hautmann , 427,475(1994 ) h.l .
lai , hep - ph/9903282 a.d .
martin , hep - ph/9906231 e. witten , 120,189(1977 ) j. butterworth , _ international conference on lepton photon interactions _ , stanford , usa ( aug .
1999 ) k. ackerstaff , 412,225(1997 ) ; 411,387(1997 ) p. abreu , 69,223(1996 ) r. barate , 458,152(1999 ) m. acciarri , 436,403(1998 ) , l3 preprint 204 ( 2000 ) k. muramatsu , 332,477(1994 ) s.k .
sahu , 346,208(1995 ) * h1 collaboration : c. adloff , desy-98 - 205 j.breitweg , desy 99 - 057 s. frixone , 507,295(1997 ) b.w .
harris and j.f .
owens , 56,4007(1997 ) m. klasen and g. kramr , 72,107(1996 ) p. nason and b.r .
webber , 421,473(1994 ) d. buskulic , 357,487(1995 ) _ ibid . , _
erratum 364,247(1995 ) * opal collaboration : r. akers , 68,203(1995 ) * delphi collaboration : p. abreu , 398,194(1997 ) e. farhi , 39,1587(1977 ) c.l .
basham , 17,2298(1978 ) b. andersson , 97,33(1983 ) a. ali , 168,409(1980 ) a. ali and r. barreiro , 118b,155(1982 ) b.r .
webber , 238,492(1984 ) g. marchesini , phys .
* 67 * , 465 ( 1992 ) t. sjostrand and m. bengtsson , 43,367(1987 ) t. sjostrand , cern - th-7112/93 ( 1993 ) s. bethke , 213,235(1988 ) s. bethke , 370,310(1992 ) m.z .
akrawy , 49,375(1991)k .
abe , 71,2578(1993 ) ; 51,962(1995 ) p.d .
acton , 55,1(1992 ) ; 58,386(1993 ) o. adriani , 284,471(1992 ) d. decamp , 255,623(1992 ) ; 257,479(1992 ) j. ellis , m.k .
gaillard , and g. ross , 111,253(1976 ) _ ibid .
_ , erratum 130,516(1977 ) p. hoyer , 161,349(1979 ) r.k .
ellis , d.a .
ross , t. terrano , 45,1226(1980 ) z. kunszt and p. nason , eth-89 - 0836 ( 1989 ) o. adriani , 284,471(1992 ) m. akrawy , 47,505(1990 ) b. adeva , 248,473(1990 ) d. decamp , 255,623(1991 ) y.l .
dokshitzer and b.r .
webber 352,451 ( 1995 ) y.l .
dokshitzer _ et . al .
_ 511,396 ( 1997 ) y.l .
jhep 9801,011 ( 1998 ) * delphi collaboration : 456,322(1999 ) s. catani , 263,491(1991 ) s. catani , 269,432(1991 ) s. catani , b.r .
webber , and g. turnock , 272,368(1991 ) n. brown and j. stirling , 53,629(1992 ) g. catani , 269,632(1991 ) ; 295,269(1992 ) ; 607,3(1993 ) ; 269,432(1991 ) y. ohnishi , 313,475(1993 ) l. gibbons , clns 95 - 1323 ( 1995 ) * delphi collaboration : d. buskulic , 73,409(1997 ) ; 73,229(1997 ) * aleph collaboration : 99 - 023 ( 1999 ) ; * delphi collaboration : 99 - 114 ( 1999 ) ; * l3 collaboration : l3 - 2414 ( 1999 ) ; * opal collaboration , pn-403 ( 1999 ) ; all submitted to _ international conference on lepton photon interactions_,stanford , usa ( aug .
1999 ) * opal collaboration : m. acciarri , 371,137(1996 ) ; 72,191(1996 ) k. ackerstaff , 75,193(1997 ) * aleph collaboration : aleph 98 - 025 ( 1998 ) * jade and * opal collaborations cern / ep-99 - 175 .
p. abreu , 59,21(1993 ) , cern - ep/99 - 133 d. graudenz , 49,3921(1994 ) hep - ph/9708362j.g .
korner , e. mirkes , and g.a .
schuler , 4,1781,(1989 ) s. catani and m. seymour , 485,291(1997 ) m. dasgupta and b.r .
webber , hep - ph/9704297e .
mirkes and d. zeppenfeld , 380,205(1996 ) * h1 collaboration : t. ahmed , 346,415(1995 ) ; 5,575(1998 ) * zeus collaboration : m. derrick , 363,201(1995 ) s. narison and a. pich , 211,183(1988 ) e. braaten , s. narison , and a. pich , 373,581(1992 ) e. braaten , 60,1606(1988 ) t. coan ( cleo collaboration ) , 356,580(1995 ) aleph collaboration , 4,409(1998 ) opal collaboration , 7 , 571 ( 1999 ) g. de divitiis , 437,447(1995 ) m. luscher , 413,481(1994 ) j. shigemitsu , nucl .
* b * ( proc .
* 53 * , 16 ( 1997 ) c.t.h .
davies , 56,2755(1997 ) a. spitz _ et al .
_ [ sesam collaboration ] , 60,074502(1999 ) g. t. bodwin , e. braaten and g. p. lepage , 51,1125(1995 ) m. gremm and a. kapustin , 407,323(1997 ) particle data group , 3,1(1998 ) b. nemati , 55,5273(1997 ) a. h. mueller , 18,3705(1978 ) j. collins , d. soper and g. sterman , 223,81(1983 ) , 109b,388(1983 ) r. k. ellis , 152,285(1979 ) .
collins jc , soper de , sterman g. in _ perturbative quantum chromodynamics _ , edited by a. h. mueller ( world scientific singapore ) , p. 1 .
ellis , z. kunszt , and d.e .
soper , 64,2121(1990 ) f. aversa , 65,401(1990 ) w.t .
giele , e.w.n .
glover , and d. kosower , 73,2019(1994 ) s. frixione , z. kunszt , and a. signer , 467,399(1996 ) f. abe , 77,438(1996 ) b. abbott , 82,2451(1999 ) w.t .
giele , e.w.n .
glover , and j. yu , 53,120(1996 ) * cdf collaboration reported in @xcite g. altarelli , r.k .
ellis , and g. martinelli , 143,521(1978 ) p. aurenche , r. baier , and m. fontannaz , 42,1440(1990 ) p. aurenche , 140b,87(1984 ) p. aurenche , 297,661(1988 ) h. baer , j. ohnemus , and j.f .
owens , 234,127(1990 ) h. baer and m.h .
reno , 43,2892(1991 ) ; p.b .
arnold and m.h .
reno , 319,37(1989 ) f. abe , 73,2662(1994 ) s. abachi , 77,5011(1996 ) g. alverson , 48,5(1993 ) l. apanasevich , 59,074007(1999 ) ; 81,2642(1998 ) w. vogelsang and a. vogt , 453,334(1995 ) p. aurenche , 9,107(1999 ) j. alitti , 263,563(1991 ) s. abache , 75,3226(1995 ) ; j. womersley , private communication j. huston , in the _ proceedings to the 29th international conference on high - energy physics ( ichep98 ) , vancouver , canada ( 2329 jul 1998 ) hep - ph/9901352 d0 collaboration : b. abbott , hep - ex/9907009 t .
affolder , fermilab - pub-99/220 m. l. mangano , p. nason and g. ridolfi , 373,295(1992 ) c. albajar et al .
369,46(1996 ) for example see , s. bethke , in _ proceedings of the iv@xmath468 int . symposium on radiative corrections , barcelona , spain ( sept .
1998 ) , hep - ex/9812026 m davier , _ 33rd rencontres de moriond : electroweak interactions and unified theories , les arcs , france ( 1421 mar . 1998 ) p.n .
burrows , acta .
phys . pol .
* 28 * , 701 ( 1997 ) j. womersley , _ international conference on lepton photon interactions _ , stanford , usa ( aug . 1999 ) p.n .
burrows , in _ proceedings of 1996 dpf / dpb snowmass summer study _ , ed .
d. cassel , ( 1997 ) _ _ _ * * * * * * * * * * * * * * * * * | this paper presents a summary of the current status of determinations of the strong coupling constant @xmath0 . a detailed description of the definition
, scale dependence and inherent theoretical ambiguities is given .
the various physical processes that can be used to determine @xmath0 are reviewed and attention is given to the uncertainties , both theoretical and experimental .
epsf.tex # 1| # 1| # 1| # 1| = @=11 # 1#1 # 1 16citenum : undefined argument # 1 b@#1
@=12 # 10= 0=0 1= 1=1 0>1 # 1 / # 1#2 -12pt = # 1 -12pt # 1#2#3 -12pt = # 1=#1 -12pt # 1,#2(#3)_adv .
nucl .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_am .
inst .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_astrophys .
j. _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_astrophys .
j. supp .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_astrophys .
j. lett .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_am .
j. phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_ann .
phys .
( ny ) _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_ann .
phys .
( ny ) _ b#1 : # 2 ( # 3 ) # 1,#2(#3)_ann .
phys .
_ d#1 : # 2 ( # 3 ) # 1,#2(#3)_ann .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_ap .
space sci . _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ astropart .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ astron . & astrophys
. _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ ann .
rev .
astron .
astrophys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ ann .
rev .
nucl . and part .
sci . _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ ann .
rev .
nucl .
sci . _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_class .
quantum grav .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_comp .
phys .
comm .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_can . j. phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_commun . math .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ comm .
nucl . part .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ comm .
nucl . part .
phys .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_europhys .
lett .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_europhys .
j. _ c#1 : # 2 ( # 3 ) # 1,#2(#3)_gen .
rel .
grav .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_helv .
phys .
acta _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ ieee trans .
_ ns#1 : # 2 ( # 3 ) # 1,#2(#3 ) _ ieee _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ int .
j. of applied rad .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ int .
j. of applied rad . and
isotopes _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_j .
chem .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_j .
geophys .
res . _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_sov .
phys .
jetp _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ sov .
phys .
jetp lett .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_j .
phys .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_j .
phys .
_ g#1 : # 2 ( # 3 ) # 1,#2(#3 ) _ j. phys . a : math . and gen .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ j. phys .
chem . ref .
data _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_j .
phys .
soc .
jpn .
_ g#1 : # 2 ( # 3 ) # 1,#2(#3)_lett .
nuovo cimento _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_nature _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_nuovo cimento _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ nucl .
instrum .
methods _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ nucl .
instrum .
methods _ a#1 : # 2 ( # 3 ) # 1,#2(#3 ) _ nucl .
instrum .
methods _ b#1 : # 2 ( # 3 ) # 1,#2(#3)_nucl .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_mnras _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_med . phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_mod .
phys .
lett .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_nucl .
phys .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_nucl . phys .
_ b#1 : # 2 ( # 3 ) # 1,#2(#3)_nucl .
phys .
( proc . supp . )
_ b#1 : # 2 ( # 3 ) # 1,#2(#3)_pub .
astron .
soc .
pac . _
# 1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
lett .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_fortsch .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ int .
j. mod .
phys .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3 ) _ int .
j. mod .
phys .
_ e#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
lett .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
lett .
_ b#1 : # 2 ( # 3 ) # 1,#2(#3 ) _ proc .
natl .
acad .
sci .
( us)_b#1 : # 2 ( # 3 ) # 1,#2(#3)_proc . phys .
soc .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_proc .
phys .
soc .
_ b#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
rev .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
rev .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
rev .
_ b#1 : # 2 ( # 3 ) # 1,#2(#3)_phys . rev .
_ c#1 : # 2 ( # 3 ) # 1,#2(#3)_phys . rev .
_ d#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
reports _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ proc .
royal soc .
london _ a#1 : # 2 ( # 3 ) # 1,#2(#3)_phys .
rev .
lett .
_ # 1 : # 2 ( # 3 ) ps .
# 1,#2(#3)_phys .
scripta _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_prog .
theor .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ prog . in part .
. phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ prog .
theor .
phys .
supp .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_part .
world _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ pisma zh .
eksp . teor .
fiz .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_revs .
geophysics & space sci . _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_rev . mod .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_riv .
nuovo cimento _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ rept . on prog . in phys
.
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_science _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ sov . j. nucl .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ phys . atom .
nucl . _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_sov .
phys .
usp . _ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ surv . high energy physics _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_yad .
fiz .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3 ) _ zh .
eksp .
teor .
fiz . _ # 1 : # 2 ( # 3 ) # 1,#2(#3)_z .
phys .
_ # 1 : # 2 ( # 3 ) # 1,#2(#3)_z .
phys .
_ a#1 : # 2 ( # 3 ) # 1,#2(#3)_z .
phys .
_ c#1 : # 2 ( # 3 ) qcd , # 1 |
the notion of one object avoiding another has been studied in permutations , words , partitions , and graphs .
much recent work has been devoted to the study of pattern - avoiding permutations .
given permutations @xmath1 and @xmath2 , we say that @xmath3 contains @xmath4 as a ( classical ) pattern if there exist indices @xmath5 such that @xmath6 is order - isomorphic to @xmath4 ; that is , @xmath7 if and only if @xmath8 .
otherwise , @xmath3 is said to avoid @xmath4 .
for example @xmath9 contains the pattern @xmath10 as evidenced by @xmath11 , @xmath12 , and @xmath13 , but @xmath3 avoids the pattern @xmath14 because @xmath3 has no decreasing subsequence of length 3 .
one variation on this definition of pattern avoidance is to place the restriction @xmath15 on the indices for @xmath16 .
if there exists such a subsequence of @xmath3 that is order - isomorphic to @xmath4 , then @xmath3 is said to contain @xmath4 as a _ consecutive _ permutation pattern .
for each of these definitions , two natural questions arise : `` given a permutation @xmath4 , how many permutations of length @xmath17 avoid @xmath4 ? '' and `` when do two distinct permutations @xmath18 and @xmath19 yield the same avoidance generating function ? ''
patterns @xmath18 and @xmath19 with this property are said to be _ wilf - equivalent_. in this paper we consider analogous questions of pattern avoidance for plane trees .
all trees in the paper are rooted and ordered .
we will focus on full binary trees , that is , trees in which each vertex has 0 or 2 ( ordered ) children .
two children with a common parent are _ sibling vertices_. a vertex with no children is a _ leaf _ and a vertex with 2 children is an _ internal vertex_. a binary tree with @xmath17 leaves has @xmath20 internal vertices , and the number of such trees is @xmath21 ( oeis a000108 ) . for simplicity of computation
, we adopt the convention that there are zero rooted binary trees with zero leaves .
the first few binary trees are depicted in figure [ f : trees ] .
conceptually , a plane tree @xmath22 avoids a tree pattern @xmath23 if there is no instance of @xmath23 anywhere inside @xmath22 .
pattern avoidance in vertex - labeled trees has been studied in various contexts by steyaert and flajolet @xcite , flajolet , sipala , and steyaert @xcite , flajolet and sedgewick @xcite , and dotsenko @xcite .
recently , disanto @xcite studied pattern containment of caterpillar subgraphs in binary trees while khoroshkin and piontkovski @xcite considered generating functions for general unlabeled tree patterns in a different context . in 2010 , rowland @xcite explored contiguous pattern avoidance in binary trees ( that is , rooted ordered trees in which each vertex has 0 or 2 children ) .
he chose to work with binary trees because there is natural bijection between @xmath17-leaf binary trees and @xmath17-vertex trees .
his study had two main objectives .
first , he developed an algorithm to find the generating function for the number of @xmath17-leaf binary trees avoiding a given tree pattern ; he adapted this to count the number of occurrences of the given pattern .
second , he determined equivalence classes for binary tree patterns , classifying two trees @xmath24 and @xmath23 as equivalent if the same number of @xmath17-leaf binary trees avoid @xmath24 as avoid @xmath23 for @xmath25 .
he completed the classification for all binary trees with at most eight leaves , using these classes to develop replacement bijections between equivalent binary trees . in 2012 ,
gabriel , peske , tay , and the second author @xcite considered rowland s definition of tree pattern in ternary , and more generally in @xmath26-ary , trees .
after generalizing rowland s algorithmic approach to compute functional equations for the avoidance generating functions of arbitrary ternary tree patterns , they explored bijections between equinumerous sets of pattern - avoiding trees . along the way they found sets of pattern - avoiding trees whose enumeration yielded a number of well - known sequences . in this paper , we extend rowland s work in a new direction .
the work of @xcite and @xcite may be seen as parallel to the definition of consecutive permutation patterns given at the beginning of this section . in those papers , tree @xmath22
was said to contain tree @xmath23 as a ( contiguous ) pattern if @xmath23 was a contiguous , rooted , ordered , subtree of @xmath22 . in this paper , we modify the definition of tree pattern to mirror the idea of classical pattern avoidance in permutations . in the case of pattern - avoiding permutations , there are more wilf - classes for consecutive patterns of a given length than for classical patterns ( as a small example , there are 7 wilf classes of consecutive permutation patterns of length 4 , compared to 3 wilf classes for classical permutation patterns of length 4 ) .
this parallel holds true in the case of trees .
in fact , as we show in section [ s : gfs ] , there is precisely one wilf class of @xmath0-leaf patterns for any @xmath27 . as with previous work , given any binary tree pattern @xmath23 we present a technique to compute the generating function that enumerates trees avoiding @xmath23 according to the number of leaves .
this enumeration shows that there is exactly one wilf class of @xmath0-leaf patterns .
we also consider trees avoiding multiple tree patterns and explore relationships between sets of pattern - avoiding trees and pattern - avoiding permutations .
in this paper , a tree @xmath22 _ contains _ @xmath23 as a ( non - contiguous ) tree pattern if there is a tree @xmath28 , obtained from @xmath22 via a finite sequence of edge contractions , such that @xmath23 is a contiguous , rooted , and ordered subtree of @xmath28 .
conversely , @xmath22 _ avoids _
@xmath23 if there is no such @xmath28 that contains @xmath23 as a subtree .
for example , consider the three trees shown in figure [ feg1 ] .
@xmath22 avoids @xmath23 as a contiguous pattern , but @xmath22 contains @xmath23 non - contiguously ( contract all non - bolded edges ) . on the other hand
, @xmath22 avoids @xmath24 both contiguously and non - contiguously since no vertex of @xmath22 has a left child and a right child , both of which are internal vertices .
@xmath29 @xmath30 @xmath31 @xmath32 @xmath33 @xmath34 we define @xmath35 to be the set of @xmath17-leaf binary trees that avoid the pattern @xmath23 non - contiguously , and @xmath36 .
we will also write @xmath37 for the set of @xmath17-leaf binary trees that avoid @xmath23 contiguously , and @xmath38
. we will be particularly interested in determining the generating function @xmath39 for various tree patterns @xmath23 . before we explore particular binary tree patterns , we list all of the 1 , 2 , 3 , and 4-leaf binary trees .
we label trees with a double subscript notation .
the first subscript gives the number of leaves of the tree , and the second subscript distinguishes between distinct tree patterns with the same number of leaves .
we also note that if @xmath40 is the left - right reflection of tree @xmath23 , then @xmath41 by symmetry , so we omit left - right reflections .
we will use these labels throughout the remainder of the paper .
@xmath42 @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 + @xmath48 @xmath32 @xmath49 @xmath50 @xmath51 @xmath34
in this section , we find generating functions for the number of trees avoiding each of the tree patterns in figure [ f : trees ] . for each tree , we discuss the structure of trees that avoid the given tree pattern and explain how to find a recurrence and generating function from this structure .
we also compare these generating functions to previously known results for contiguous tree patterns .
clearly , a tree avoids @xmath52 if and only if it has no vertices .
this is also true for contiguous avoidance , so we have @xmath53 next , a tree avoids @xmath54 if and only if it has no vertex with two children . in other words
, only @xmath52 avoids @xmath54 .
again , we have @xmath55 finally , we observe that tree @xmath23 avoids @xmath56 if and only if @xmath23 has no vertex whose left child is not a leaf . for each @xmath57
there is precisely one such tree , so we have @xmath58 from these few cases , it may seem that non - contiguous and contiguous avoidance is the same for many trees . the reader should suspect that this similarity does not hold for larger tree patterns based on the example of figure [ feg1 ] .
this suspicion is confirmed when we consider 4-leaf tree patterns .
rowland showed that for @xmath59 @xmath60 where @xmath61 is the @xmath17th motzkin number ( oeis a001006 ) .
it turns out that non - contiguous avoidance is even more well - behaved for 4-leaf tree patterns .
we will show that for @xmath62 @xmath63 to find @xmath64 , we consider the structure of a general @xmath17-leaf binary tree @xmath22 that avoids @xmath65 .
we have two cases .
let @xmath66 be the root of @xmath22 .
in the first case , @xmath66 s left child has no children , while @xmath66 s right child is the root of an @xmath67-leaf subtree avoiding @xmath65 . in the second case , @xmath66 s left child has two children , but the leftmost of these children is a leaf .
if the right child of @xmath66 s left child is the root of a subtree with @xmath68 leaves ( @xmath69 ) , then @xmath66 s right child is the root of a subtree with @xmath70 leaves .
further , the @xmath68-leaf subtree must avoid the tree pattern @xmath56 .
in the first case , we considered @xmath71 possible trees . in the second case
, we considered @xmath72 trees .
however , since we know that @xmath73 for @xmath74 , we have @xmath75 together with the base case @xmath76 , we see that @xmath77 next , we consider the structure of a general @xmath17-leaf binary tree @xmath22 that avoids @xmath78 where @xmath66 is the root of @xmath22 . again , we have two cases . as before , in the first case , @xmath66 s left child has no children , while @xmath66 s right child is the root of an @xmath67-leaf subtree avoiding @xmath78 . in the second case , @xmath66 s left child has two children , but now the rightmost of these children is a leaf .
the subtree whose root is the left child of @xmath66 s left child must avoid the tree pattern @xmath79 .
after a nearly identical calculation to that of avoiding @xmath65 , we see that @xmath80 finally , we consider the structure of a general @xmath17-leaf binary tree @xmath22 that avoids @xmath81 where @xmath66 is the root of @xmath22 . again , we have two cases : either @xmath66 s left child has children or @xmath66 s right child has children , but not both .
there are @xmath82 trees that fall into the first case , and @xmath82 trees that fall into the second case , which yields @xmath83 we have @xmath84
we have just seen that all 4-leaf binary tree patterns produce the same avoidance sequence when considered as non - contiguous patterns .
this observation leads naturally to the main theorem of this paper , which provides a particularly clean answer to both questions stated in the introduction .
namely , `` given a tree pattern @xmath23 , how many trees with @xmath17 leaves avoid @xmath23 ? '' and `` given two distinct tree patterns @xmath23 and @xmath24 , when do @xmath23 and @xmath24 produce the same avoidance sequence ? ''
in fact , we see that not only is exact enumeration possible for _ any _ non - contiguous tree pattern , we see that all avoidance generating functions are rational and of a particularly attractive form
. let @xmath85 and let @xmath23 be a binary tree pattern with @xmath0 leaves .
then @xmath86 [ th1 ] the reader can check that this indeed matches the generating functions computed in the previous section when @xmath87 .
further , notice that the generating function given in theorem [ th1 ] depends only on the number of leaves of @xmath23 , and not on @xmath23 itself ; that is , there is precisely one wilf class of @xmath0-leaf tree patterns for each @xmath27 . as we will see in section [ s : multipatterns ] and the appendix , the analogous statement for pairs of trees is not true .
in this section we build the necessary framework to prove theorem [ th1 ] .
first , following @xcite we say that @xmath22 contains pattern @xmath88 at the root if @xmath22 contains a _ contiguous _ copy of @xmath88 where the root of @xmath88 coincides with the root of @xmath22 .
now , let @xmath89 be the generating function that enumerates binary trees avoiding tree pattern @xmath23 non - contiguously and containing the contiguous tree pattern @xmath88 at their root according to number of leaves . because all binary trees have a root vertex
, it follows that the generating function for all trees avoiding @xmath23 is given by @xmath90 .
also , let @xmath91 and @xmath92 denote the subtrees descending from the left and right children of the root of @xmath23 respectively .
since we are working with full binary trees , the root has either zero or two children .
when there are zero children , we have a 1-leaf tree .
when there are two children , we have a tree with pattern @xmath93 at the root .
such trees are enumerated with the generating function @xmath94 .
thus , we have @xmath95 next , we determine a recurrence for @xmath94 .
consider a tree @xmath22 that avoids @xmath23 and has the contiguous pattern @xmath93 at the root .
either the subtree extending from the left child of the root of @xmath22 avoids @xmath91 , the subtree extending from the right child of the root of @xmath22 avoids @xmath96 , or both .
inclusion - exclusion gives @xmath97 combining equations [ eq : t11 ] and [ eq : t21 ] yields @xmath98 now , solve for @xmath99 to obtain @xmath100 this computation yields one immediate result : for any tree pattern @xmath23 , @xmath99 is a rational function of @xmath101 .
we have already seen that @xmath99 is rational for all tree patterns @xmath23 with @xmath87 leaves .
thus , by induction and equation [ eq : genfunrec ] , the proposition follows .
equation [ eq : genfunrec ] simplifies further for one particular family of binary tree patterns .
let the @xmath0-leaf left comb be the unique @xmath0-leaf binary tree where every right child is a leaf .
write @xmath102 for the @xmath0-leaf left comb
. then @xmath103 , @xmath104 , @xmath105 , and @xmath106 .
@xmath107 for @xmath108 .
let @xmath109 in equation [ eq : genfunrec ] .
then @xmath110 and @xmath111 .
we have : @xmath112 since @xmath113 , this simplifies to @xmath114 this nice relationship between @xmath115 and @xmath116 seems natural because of the large overlap between copies of @xmath102 and @xmath117 .
our next lemma shows that there is a simple relationship between generating functions for non - comb tree patterns as well .
fix @xmath27 .
let @xmath23 and @xmath24 be two @xmath0-leaf binary tree patterns .
then @xmath118 assume the lemma holds for tree patterns with @xmath119 leaves where @xmath120 . since we suppose that all @xmath119-leaf trees have the same avoidance generating function , we have that any @xmath119-leaf tree has avoidance generating function @xmath121 .
suppose that tree @xmath122 has @xmath119 leaves to the left of its root and @xmath123 leaves to the right of its root .
similarly , suppose that tree @xmath124 has @xmath125 leaves to the left of its root and @xmath126 leaves to the right of its root .
we will show that @xmath127 . from equation [ eq : genfunrec ] , we have @xmath128 similarly , @xmath129 thus @xmath130 since this holds for @xmath131 , by transitivity , all @xmath0-leaf tree patterns have the same non - contiguous avoidance generating function . finally , to prove the main theorem
, it suffices to show that @xmath132 it is a straightforward induction proof , using the fact that @xmath133 to show that this formula holds in general .
we have now explicitly enumerated trees avoiding any single binary tree pattern non - contiguously and determined all equivalences for when two trees exhibit the same avoidance sequence .
we display the explicit generating function , first 8 sequence terms , and appropriate oeis entry for tree patterns with @xmath0 leaves where @xmath134 in table [ tableseqs ] .
rate & oeis number + 1 & 0 & @xmath135 & 0 & trivial + 2 & @xmath101 & @xmath136 & 0 & trivial + 3 & @xmath137 & @xmath138 & @xmath139 & trivial + 4 & @xmath140 & @xmath141 & @xmath142 & a000079 + 5 & @xmath143 & @xmath144 & @xmath145 & a001519 + 6 & @xmath146 & @xmath147 & @xmath148 & a007051 + 7 & @xmath149 & @xmath150 & @xmath151 & a080937 + 8 & @xmath152 & @xmath153 & @xmath154 & a024175 + 9 & @xmath155 & @xmath156 & @xmath157 & a080938 + because all generating functions @xmath99 are rational it follows that for any @xmath0-leaf binary tree @xmath23 the sequence @xmath158 satisfies a linear recurrence with constant coefficients .
in fact , when tree pattern @xmath23 has @xmath159 leaves , @xmath158 grows exponentially . because there are catalan - many binary trees with @xmath17 leaves , the growth of these sequences is bounded above by @xmath160 . in table [ tableseqs ] we display also the asymptotic growth rate for @xmath161 .
while we gave explicit combinatorial explanations for the recurrences satisfied by @xmath162 when @xmath23 has @xmath163 or @xmath164 leaves in section [ s : simple ] , the rest of our work has been algebraic .
it is possible , however , to derive the recurrences satisfied by tree enumeration sequences through other techniques . in this section ,
we give a combinatorial explanation for these recurrences by considering the structure of trees that avoid the @xmath0-leaf left comb .
first , we define a many - to - one correspondence between the set of @xmath165-leaf binary trees and the set of @xmath17-leaf binary trees . given a tree @xmath23 with @xmath166 leaves , let @xmath167 be rightmost leaf whose sibling vertex is also a leaf
. then , let the _ parent tree _ of @xmath23 be the @xmath17-leaf tree obtained by deleting @xmath167 and its sibling . if @xmath168 is the parent tree of @xmath23 , then we say that @xmath23 is a _ descendent tree _ of @xmath168 . while the parent tree of a tree is unique , a given tree may have multiple descendent trees .
for example , @xmath169 , @xmath170 , and @xmath171 are descendent trees of @xmath172 .
similarly , @xmath173 and @xmath174 are descendent trees of @xmath175 . now , for any @xmath17-leaf tree @xmath176 , we can use this parent / descendent relationship to generate a list @xmath177 where @xmath178 , @xmath179 has @xmath68 leaves for @xmath180 and @xmath181 is a descendant tree of @xmath179 .
we refer to such a sequence of trees as the _ ancestry _ of @xmath23 .
because parent trees are unique , the ancestry of tree @xmath23 must be unique .
for example , the tree @xmath182 has ancestry @xmath183 given an @xmath17-leaf tree @xmath23 , we may also generate all descendent trees of @xmath23 in a systematic way . call an internal vertex @xmath66 of @xmath23 _ closed _ if @xmath66 s right child is not a leaf .
the number of descendent trees of @xmath23 is equal to the number of leaves of @xmath23 that appear to the right of all closed vertices .
in fact , to produce a descendent tree of @xmath23 , we need only attach a pair of children to any one such leaf . the descendent tree relationship is further articulated in the following proposition .
suppose that tree @xmath23 has @xmath68 descendant trees .
then those descendant trees will have @xmath184 descendant trees respectively .
further , if @xmath23 has @xmath68 descendant trees then the leftmost vertex to which one may add children is the leftmost vertex in a copy of an @xmath68-leaf left comb .
[ prop : descs ] both claims in the proposition are consequences of the fact that the number of descendent trees of @xmath23 is equal to the number of leaves of @xmath23 appearing to the right of all closed vertices .
let @xmath185 be the @xmath186th leaf from the right of these @xmath68 leaves .
when we append two children to @xmath185 , @xmath185 s parent is closed , and this new tree has only @xmath187 descendents .
further , if @xmath23 has @xmath68 leaves to the right of all closed vertices , by definition of closed , none of these leaves parents have right children that are internal vertices . the only way to arrange a collection of vertices so that no internal vertices are right children is in the shape of an @xmath68-leaf left comb .
now , consider the descendant relation restricted to trees that avoid the @xmath0-leaf left comb .
any tree with @xmath0 descendant trees contains an @xmath0-leaf comb , so we are only concerned with trees that have at most @xmath188 descendent trees . as in proposition
[ prop : descs ] , a tree with @xmath189 descendant trees will have descendant trees with @xmath184 descendants respectively .
notice further that for a tree which would have @xmath188 descendent trees , one of these descendent trees contains the @xmath0-leaf left comb , so such a tree only has @xmath190 descendent trees that avoid the @xmath0-leaf comb .
consequently , for a tree with @xmath190 descendent trees , those descendents will have @xmath191 descendent trees respectively that avoid the @xmath0-leaf left comb .
let @xmath192 be the number of @xmath17-leaf trees that avoid the @xmath0-leaf left comb and have exactly @xmath68 @xmath165-leaf descendants , and let @xmath193 .
we have the base cases @xmath194 and @xmath195 .
more generally , for @xmath196 @xmath197 @xmath198 @xmath199 @xmath200 ultimately , we seek a recurrence for @xmath193 . adding the above equations and algebraic manipulation produces the result @xmath201 from the discussion above
, trees avoiding an @xmath0-leaf comb can be constructed with a finitely labeled generating tree ( in particular with a generating tree using precisely @xmath190 labels ) , and one may use the transfer matrix method to obtain an alternate proof of theorem [ th1 ] for the case of avoiding the @xmath0-leaf left comb .
as evidenced in the appendix , many sequences obtained by counting trees that avoid non - contiguous binary tree patterns are already known in the literature for other reasons . in this section
we present a theorem that fully explains this connection for the case of avoiding a single binary tree pattern . to this end , let @xmath202 denote the set of permutations of length @xmath17 . as in the introduction , given @xmath203 and @xmath204
we say that @xmath3 contains @xmath4 as a pattern if there exist indices @xmath205 such that @xmath206 if and only if @xmath207 .
let @xmath208 , and @xmath209 .
for example , @xmath210 for @xmath25 since the only way to avoid the pattern 12 is to be the decreasing permutation of length @xmath17 .
it is also well - known that if @xmath211 , then @xmath212 where @xmath213 is the @xmath17th catalan number .
let @xmath23 be any binary tree pattern with @xmath108 leaves .
then @xmath214 [ permpat ] it is well known that the set of binary trees with @xmath17 leaves is in bijection with the set of permutations of length @xmath20 which avoid the pattern 231 . to see this , label the root of tree @xmath23 with the label @xmath20 .
now , suppose there are @xmath68 internal vertices to the left of the root and @xmath215 internal vertices to the right of the root .
the @xmath68 vertices on the left will receive labels from the set @xmath216 and the vertices on the right will receive labels from the set @xmath217 . for each subtree , give the root the largest available label and continue recursively until each internal vertex has been labeled .
now , there is a natural left - to - right ordering of the vertices of @xmath23 . read the labels of the vertices from left to right to obtain a permutation @xmath218 .
necessarily , @xmath3 avoids @xmath219 because all labels to the left of a given vertex have smaller labels than all labels to the right .
further , we can see that the @xmath0-leaf right comb encodes the unique decreasing permutation of length @xmath188 .
it is not hard to see that if a tree avoids the @xmath0-leaf right comb , then the corresponding permutation avoids the decreasing permutation of length @xmath188 and vice versa .
we note that this correspondence between @xmath220-avoiding permutations and binary trees is not new .
if one ignores the leaves in our trees , the bijection given in the proof of theorem [ permpat ] is identical to the correspondence between postorder - labeled trees with inorder - read permutations found in @xcite .
further work connecting permutations to binary trees in the context of sorting can be found in @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
it is worth considering when theorem [ permpat ] generalizes and how .
the correspondence given in the proof of theorem [ permpat ] does not necessarily work with trees other than the right comb .
for example , while @xmath221 the permutation corresponding to @xmath65 is 123 .
however , @xmath222 ( @xmath223 ) while @xmath224 ( @xmath223 ) .
thus even if the sequence obtained from avoiding a set of tree patterns is also the avoidance sequence for some set of pattern - avoiding permutations , there may be other bijections required to demonstrate the equivalence .
it remains open to give a natural interpretation of theorem [ permpat ] for trees other than the right comb .
thus far , we have analyzed generating functions and enumeration sequences for trees avoiding a single non - contiguous binary tree pattern . we will now consider trees that avoid two tree patterns simultaneously . to this end , let @xmath225 be the generating function that enumerates trees avoiding @xmath226 and @xmath227 with pattern @xmath88 at the root according to number of leaves .
for brevity , we may abbreviate @xmath228 .
parallel to equations [ eq : t11 ] and [ eq : t21 ] we have @xmath229 and @xmath230 this latter expression follows again from inclusion - exclusion .
one can solve this pair of equations for @xmath231 to obtain a rational combination of rational generating functions and see that for any pair of binary trees , @xmath231 is indeed rational . to compactly state the equivalent expression to equation [ eq:2treerec ] for trees avoiding @xmath186 binary tree patterns ( @xmath232 ) , we need to introduce more efficient notation .
notice that all terms on the right hand sides of equations [ eq : t21 ] and [ eq:2treerec ] are of the form @xmath233 where for each tree @xmath179 @xmath234 , there are 3 possibilities : ( i ) @xmath235 and @xmath236 , ( ii ) @xmath237 and @xmath238 , or ( iii ) @xmath237 and @xmath236 .
let @xmath239 and let @xmath240 where ( i ) if @xmath241 then @xmath235 and @xmath236 , ( ii ) if @xmath242 then @xmath237 and @xmath238 , and ( iii ) if @xmath243 then @xmath237 and @xmath236 .
further , for @xmath244 , define @xmath245 . by inclusion - exclusion
we have : @xmath246 notice that this expression is linear in @xmath247 .
in fact , @xmath247 only appears in two terms : the ones corresponding to @xmath248 and @xmath249 , so we see that @xmath250 is rational for any finite set of @xmath186 tree patterns .
we used equations [ eq:2treebase ] and [ eq:2treerec ] to compute @xmath231 for any pair of binary tree patterns where @xmath226 and @xmath227 have no more than 5 leaves .
a summary of these results is given in the appendix . in light of theorem [ permpat ] one might wonder if for every set of binary tree patterns @xmath251 , @xmath252 is identical to the avoidance sequence @xmath253 for some set @xmath254 of permutation patterns .
this , however turns out not to be the case .
for example consider the sequence @xmath255 if this were the avoidance sequence for some set of permutation patterns @xmath254 , we see from the fact that @xmath256 that @xmath254 contains a pattern of length 3 .
further , since it is known that @xmath257 for any pattern @xmath211 , @xmath254 must contain precisely two patterns of length 4
. however , exhaustive checking of @xmath258 for sets of patterns @xmath254 consisting of one pattern of length 3 and two patterns of length 4 yields no match for this sequence .
throughout this paper , we have investigated non - contiguous pattern avoidance in binary trees .
unlike rowland s work for contiguous patterns , our avoidance generating functions are always rational , and the wilf classes obtained for avoidance of single patterns are easy to describe : two tree patterns are wilf - equivalent if and only if they have the same number of leaves . the results in this paper not only give a complete characterization of trees avoiding a single pattern , but we also present a computational technique to quickly determine the number of trees avoiding any finite set of non - contiguous tree patterns .
finally we explore combinatorial proofs of our results and give an explicit bijection between certain pattern - avoiding trees and pattern - avoiding permutations .
the authors thank eric rowland for assistance with generating the tree graphics for this paper and for several helpful presentation suggestions .
this appendix lists pairs of binary tree patterns each of which have at most 5 leaves , classifying them by their avoidance generating function and sequence . for each class , we give the generating function @xmath259 , and we list the first 15 terms of the corresponding avoidance sequence . if the avoidance sequence for a class is listed in the online encyclopedia of integer sequences @xcite , we include the appropriate reference
. for brevity , left
right reflections are omitted .
p. flajolet , p. sipala , and j. m. steyaert .
analytic variations on the common subexpression problem .
_ automata , languages , and programming : proc . of icalp
1990_. volume 443 of _ lecture notes in computer science _ , pages 220234 .
springer , 1990 . | in this paper we consider the enumeration of binary trees avoiding non - contiguous binary tree patterns .
we begin by computing closed formulas for the number of trees avoiding a single binary tree pattern with 4 or fewer leaves and compare these results to analogous work for contiguous tree patterns .
next , we give an explicit generating function that counts binary trees avoiding a single non - contiguous tree pattern according to number of leaves and show that there is exactly one wilf class of @xmath0-leaf tree patterns for any positive integer @xmath0 .
in addition , we enumerate binary trees that simultaneously avoid more than one tree pattern . finally , we explore connections between pattern - avoiding trees and pattern - avoiding permutations .
michael dairyko + department of mathematics + pomona college + claremont , ca 91711 , usa + + lara pudwell + department of mathematics and computer science + valparaiso university + valparaiso , in 46383 , usa + [email protected] + + samantha tyner + department of mathematics + augustana college + rock island , il 61201 , usa + + casey wynn
+ department of mathematics and computer science + hendrix college + conway , ar 72032 , usa |
despite the detection of objects up to @xmath1 and the impressive success of the @xmath2cdm scenario to account for the properties of the cosmic microwave background , one of the main and still controversial questions remains how and when galaxies assembled their mass as a function of the cosmic time .
the hierarchical scenario predicts that galaxies built up their present - day mass through a progressive assembly of smaller sub - systems driven by the merging of dark matter halos . with ground - based observations , one of the most solid approaches to address the problems of galaxy formation and evolution is to study samples of field faint galaxies selected in the near - infrared , particularly in the @xmath3-band ( 2.2@xmath4 m ) @xcite .
firstly , since the rest - frame near - ir luminosity is a good tracer of the galaxy stellar mass @xcite , @xmath3-band surveys allow to select galaxies according to their mass up to @xmath5 ( @xmath6 m ) . at higher redshifts ,
the @xmath3-band starts to sample the rest - frame optical and uv regions , and space - based observations at @xmath7 m are needed to cover the rest - frame near - infrared ( e.g. sirtf , astro - f ) .
secondly , the similarity of the spectral shapes of different galaxy types in the rest - frame near - ir makes the @xmath3-band selection free from strong biases against or in favor of particular classes of galaxies .
in contrast , the selection of high-@xmath8 galaxies in the observed optical bands is more sensitive to the star formation activity than to the stellar mass because it samples the rest - frame uv light and makes optical samples biased against old passive or weakly star - forming galaxies .
last but not least , near - infrared surveys are less affected by dust extinction than optical surveys . motivated by the above considerations ,
several near - infrared surveys have been undertaken during the last decade @xcite ( fig .
1 ) . spectroscopic surveys are particularly relevant thanks to their capability not only to derive the redshifts , but also to unveil the nature and the spectral properties of the targeted galaxies .
since near - ir multi - object spectrographs are not fully available on 8 - 10m - class telescopes , most spectroscopy of @xmath3-selected galaxies has been done in the optical .
the most crucial probes of massive galaxy evolution are galaxies with the very red colors expected in case of passive evolution at @xmath9 ( e.g. @xmath10 , usually called extremely red objects , eros ) . however , for the typical limiting fluxes of near - ir surveys ( @xmath11 ) , they have very faint optical magnitudes ( @xmath12 ) already close to the spectroscopic limits of 8 - 10m - class telescopes . in addition , for @xmath9 their main spectral features ( e.g. the 4000 break and h&k absorptions ) fall in the very red part of the observed spectra , where the strong oh sky lines and the ccd fringing and low quantum efficiency make spectroscopy even more demanding .
the k20 survey is a project aimed at investigating galaxy evolution through deep eso vlt spectroscopy of a sample of 546 objects with @xmath13 selected from two fields covering a total area of 52 arcmin@xmath14 .
the two fields are completely independent , and one of them is a sub - area of the chandra deep field south .
most spectroscopy was done in the optical with fors1 and fors2 and designed to reach the highest possible signal - to - noise ratio in the red .
isaac near - ir spectroscopy was also done for a small fraction of the sample . the multi - band ( @xmath15 ) imaging available for both fields
was used to derive high - quality photometric redshifts `` trained '' with the spectroscopic redshifts .
the final sample covers a redshift range of @xmath16 , and the current redshift completeness is 92% ( spectroscopic ) and 100% ( spectroscopic + photometric redshifts ) .
the k20 survey represents a significant improvement with respect to other surveys for faint @xmath3-selected galaxies thanks to its high spectroscopic redshift completeness ( the highest to date ) extended to very faint red objects , the larger sample , the coverage of two fields and the availability of optimized photometric redshifts .
we note here that the distribution of the sample over two independent sky fields is a significant advantage in reducing the field - to - field variation effects . for more details on the k20 survey see @xcite and _
overall , the k20 and other recent near - ir surveys show that galaxies are characterized by little evolution to @xmath18 , so that the observed properties can be mimicked by a pure luminosity evolution ( ple)-like scenario .
this is in contrast with the current @xmath2cdm hierarchical merging models where the assembly of galaxies occurs later than what is actually observed . in particular : @xmath19 for @xmath13 , @xmath20 has a median redshift of @xmath21 and a high-@xmath8 tail extended beyond @xmath0 .
current hierarchical models do not match the observed redshift distribution @xcite ( fig.2 ) , but an improved treatment of the merging processes is providing encouraging results @xcite . compared with the predictions of three hierarchical merging model ( dotted and dashed lines ) and a pure luminosity evolution ( ple ) model ( solid line ) .
the models refer to _ total _ magnitudes @xmath13 and are not corrected for the average under - estimate of 0.2 magnitudes due to the photometric selection effects evaluated in the k20 sample ( see @xcite ) . including this bias in the models results in improving the agreement with the ple models and increasing the discrepancy with the hierarchical models , as shown in the inserted plot where the model of menci et al .
( 2002 ) for @xmath22 is compared to the observed @xmath9 tail of @xmath20 ( see @xcite ) .
, scaledwidth=95.0% ] @xmath19 the rest - frame @xmath23-band luminosity function shows a mild luminosity evolution up to @xmath24 , with a brightening of about 0.5 magnitudes .
significant density evolution is ruled out up to @xmath24 @xcite .
the high - luminosity tail at @xmath18 is dominated by red early - type galaxies @xcite .
current hierarchical models fail in reproducing the shape and the evolution of the lf as they over - predict the number of sub-@xmath25 galaxies , under - predict that of luminous galaxies , and predict a strong density evolution .
@xmath19 the rest - frame @xmath23-band luminosity density evolves slowly to @xmath18 with @xmath26 , and much slower than the uv luminosity density @xcite .
@xmath19 the properties of `` old '' eros with @xmath13 at @xmath18 ( morphology , spectra , luminosities , ages , stellar masses , clustering ) imply the existence of a substantial population of old ( a few gyr ) , passively evolving and fully assembled massive spheroids which requires that major episodes of massive galaxy formation occurred at least at @xmath27 @xcite .
their number density at @xmath18 is consistent ( within @xmath28 ) with that of local luminous e / s0 galaxies .
current hierarchical models severely under - predict the number of `` old '' eros @xcite .
@xmath19 a numerous population of dusty star - forming eros with disk - like and irregular morphologies emerges at @xmath29 from near - ir surveys @xcite .
these objects are often too faint to be detected in submm surveys due to their inferred far - ir luminosities @xmath30 l@xmath31 and represent an ensemble of galaxies important to complement other star - forming systems selected with different techiniques .
they are also expected to be important contributors to the cosmic star - formation density @xcite . also
in this case , current hierarchical models under - estimate the number of `` dusty '' eros .
@xmath19 `` old '' eros seem to have much stronger clustering than `` dusty '' eros , with a comoving @xmath32 similar to that of present - day luminous ellipticals @xcite .
@xmath19 the number density of massive galaxies and the cosmic stellar mass density @xmath33 show a slow decrease from @xmath34 to @xmath18 qualitatively consistent with a hierarchical scenario , but much slower than what the current models predict @xcite .
at higher redshifts , the picture becomes more controversial : at @xmath9 the evolution of the near - ir luminosity function seems to depart from a ple - like pattern @xcite , it is still not clear if the number density of old spheroids drops at @xmath35 @xcite , and unexpected populations of massive star - forming galaxies are being found in the range of @xmath36 thanks to near - ir surveys @xcite .
moreover , the galaxy stellar mass function and @xmath37 display a fast evolution at @xmath9 @xcite , and the k20 survey results suggest a rapid increase of the stellar mass density from @xmath38% of the local value at @xmath39 to about 100% at @xmath40 @xcite .
these results suggest that @xmath41 may be the critical cosmic epoch during which most star formation activity and galaxy mass assembly took place .
however , probing the @xmath41 range is difficult due to the lack of strong spectral features and emission lines redshifted in the observed optical and to the strong oh emission sky lines severely affecting the spectra beyond @xmath178000 . due to the limited number of spectroscopically identified galaxies , this redshift range has been traditionally called `` redshift desert '' .
thanks to the very deep red - optimized optical spectroscopy and to near - ir spectroscopy , the k20 survey allowed us to populate the `` redshift desert '' and to unveil the nature of @xmath3-selected galaxies in this critical redshift range . by combining deep hst+acs imaging with k20 spectroscopy , we found that a variety of galaxy types are present at @xmath41 : spheroids at @xmath5 , disk- and spiral - like systems , and many irregular merging - like objects . particularly relevant
is a sub - sample of luminous galaxies at @xmath42 spectroscopically identified in the k20 survey @xcite .
their inferred star formation rates of @xmath43 m@xmath31yr@xmath44 , stellar masses up to @xmath45 m@xmath31 derived from the fitting of their multi - color spectral energy distributions , and their merging - like optical morphologies becoming more compact in the near - ir , suggest that these galaxies may be massive galaxies caught during their major episode of mass assembly and star - formation , possibly being the progenitors of the present - day massive spheroidal systems .
current semi - analytical models under - predict by a factor of @xmath46 the number density of such galaxies @xcite .
other near - ir surveys are independently finding candidate massive systems at @xmath47 @xcite . in order to draw a coherent picture of galaxy evolution at @xmath41
, it will be crucial to combine @xmath3-selected samples ( e.g. k20 , fires @xcite , gdds @xcite ) with uv- and optically - selected samples more biased toward star - forming objects with little dust extinction such as those resulting from the lyman - break selection at @xmath48 @xcite .
the k20 survey team includes : t. broadhurst ( huj ) , s. cristiani ( inaf - trieste ) , s. dodorico ( eso ) , e. daddi ( eso ) , a. fontana ( inaf - roma ) , e. giallongo ( inaf - roma ) , r. gilmozzi ( eso ) , n. menci ( inaf - roma ) , m. mignoli ( inaf - bologna ) , f. poli ( university of rome ) , l. pozzetti ( inaf - bologna ) , a. renzini ( eso ) , p. saracco ( inaf - brera ) , and g. zamorani ( inaf - bologna ) . | near - infrared surveys provide one of the best opportunities to investigate the cosmic evolution of galaxies and their mass assembly .
we briefly review the main results obtained so far with the k20 and other recent near - ir surveys on the redshift distribution , the evolution of the luminosity function and luminosity density , the nature of old and dusty eros , the evolution of the galaxy stellar mass function , the properties of the galaxies in the `` redshift desert '' and the nature of luminous starbursts at @xmath0 . |
an alternative model is the massive thirring model ( mtm ) with vector self - interaction @xcite , where the nonlinear term in the lagrangian is based on the scalar quantity @xmath73 built from the lorentz vector @xmath74 which represents the charge - current density .
we mention that the mtm in one spatial dimension is completely integrable , and in two spatial dimensions is equivalent to the soler model .
we demonstrate below that in two spatial dimensions the soler and the mtm models coincide . in the massive thirring model with vector self - interaction , the nonlinear term in the lagrangian
is based on the scalar quantity @xmath73 built from the lorentz vector @xmath74 which represents the charge - current density . using the dirac matrices
@xmath75 , @xmath76 , @xmath77 , and using the identity @xmath78 valid for any @xmath25 , we compute that in 2d these two scalar quantities coincide : @xmath79
prior to explaining the numerical methods used for calculating stationary solutions , we will proceed to present a summary of spectral methods needed for dealing with derivatives in continuum settings . for a detailed discussion on these methods ,
the reader is directed to @xcite and references therein .
spectral methods arise due to the necessity of calculating spatial derivatives with higher accuracy than that given by finite difference methods . to this aim
, a differentiation matrix @xmath80 must be given together with collocation ( i.e. grid ) points @xmath81 , which are not necessarily equi - spaced .
thus , if the _
spectral _ derivative of a function @xmath82 needs to be calculated , it can be cast as : @xmath83 where @xmath84 and @xmath85 . if @xmath86 $ ] and the boundary conditions are periodic , the fourier collocation can be used . in this case , @xmath87 with @xmath88 even .
the differentiation matrix is @xmath89 \dfrac{\pi}{2l}(-1)^{n+m}\cot\dfrac{x_n - x_m}{2 } & \textrm{if } n\neq m . \end{array}\right.\ ] ] notice that doing the multiplication @xmath90 is equivalent to performing the following pair of discrete fourier transform applications : @xmath91 with @xmath92 and @xmath93 denoting , respectively , the direct and inverse discrete fourier transform @xcite .
the vector wavenumber @xmath94 is defined as : @xmath95 0 & \textrm{if } n = n/2 . \end{array}\right.\ ] ] the computation of the direct and inverse discrete fourier transforms , which is useful in simulations , can be accomplished by the fast fourier transform . in what follows , however ,
the differentiation matrix is used for finding the jacobian and stability matrices .
notice that the grid for finite differences discretization is the same as in the fourier collocation ; and , in addition , there is a differentiation matrix for the finite differences method , i.e. @xmath96 with @xmath97 being kronecker s delta .
it can be observed from the above discussion that in the fourier spectral method , the banded differentiation matrix of the finite differences method is substituted by a dense matrix , or , in other words , a nearest - neighbor interaction is exchanged for into a long - range one .
the lack of sparsity of differentiation matrices is one of the drawbacks of spectral methods , especially when having to diagonalize large systems .
however , they have the advantage of needing ( a considerably ) smaller number of grid points @xmath88 for getting the same accuracy as with finite difference methods . for fixed ( dirichlet ) boundary conditions , the chebyshev spectral methods are the most suitable ones .
there are several collocation schemes , the gauss - lobato being the most extensively used : @xmath98 with @xmath88 being even or odd .
the differentiation matrix is @xmath99 \dfrac{(-1)^{n+m}}{l\cos(x_n - x_m ) } & \textrm{if } n\neq m . \end{array}\right.\ ] ] the significant drawback of chebyshev collocation is that the discretization matrix possesses a great number of spurious eigenvalues or _
outliers_. they are approximately equal to @xmath100
. these outliers also have a significant non - zero real part , which increases when @xmath88 grows .
this fact naturally reduces the efficiency of the method when performing numerical time - integration .
however , it presents a higher spectral accuracy than the fourier collocation method ( see e.g. @xcite ) .
several modifications must be introduced when applying spectral methods to polar coordinates .
they basically rely on overcoming the difficulty of not having dirichlet boundary conditions at @xmath101 and the singularity of the equations at that point .
in addition , in the case of the dirac equation , the spinor components can be either symmetric or anti - symmetric in their radial dependence , so the method described in @xcite must be modified accordingly . as shown in the previously mentioned references ,
the radial derivative of a general function @xmath102 can be expressed as : @xmath103 notice that in this case , the collocation points must be taken as @xmath104 but only the first @xmath88 points are taken so that the domain of the radial coordinate does not include @xmath101 .
analogously the differentiation matrix would possess now @xmath105 components , but only the upper half of the matrix , of size @xmath106 is used .
if the function that must be derived is symmetric or anti - symmetric , i.e. @xmath107 , with the upper ( lower ) sign corresponding to the ( anti-)symmetric function , equation ( [ supp_eq : polar1 ] ) can be written as : @xmath108\ ] ] thus , the differentiation matrix has a different form depending whether @xmath102 is symmetric or anti - symmetric : @xmath109 with @xmath110 defined as in ( [ supp_eq : polar2 ] ) .
having established the basic features of spectral methods , we have enough tools for undertaking the numerical analysis of stationary states for the nonlinear dirac equation .
we recall that the dynamical equations in polar coordinates read as : @xmath111\psi_1 , \nonumber \\
i\partial_t\psi_2 & = & -{\mathrm{e}}^{i\theta}\left(i\partial_r-\frac{\partial_\theta}{r}\right)\psi_1-[1-(|\psi_1|^2-|\psi_2|^2)^k]\psi_2.\end{aligned}\ ] ] and stationary solutions can be written in the form @xmath112 with @xmath113 \phi_2(r,\theta ) \end{array}\right ) = \left(\begin{array}{c } v(r){\mathrm{e}}^{i s\theta } \\[1ex ] i\,u(r){\mathrm{e}}^{i(s+1)\theta } \end{array}\right ) , \ ] ] where @xmath31 and @xmath32 could be chosen real - valued .
thus , the equations for the stationary solutions read : @xmath114v , \nonumber \\ \omega u & = & -\left(\partial_r-\frac{s}{r}\right ) v-[1-(v^2-u^2)^k]u , \end{aligned}\ ] ] with @xmath115 .
there are no analytical solutions available for this system . for this reason , numerical methods
must be used to this aim . among them
we have chosen to use fixed point methods , as the newton - raphson one @xcite , which requires the transformation of the set of two coupled ordinary differential equations ( [ supp_eq : stat ] ) into a set of @xmath116 algebraic equations ; this is performed by defining the set of collocation points @xmath117 , and transforming the derivatives into multiplication of the differentiation matrices @xmath118 and @xmath119 times the vectors @xmath120 and @xmath121 , respectively , being @xmath122 and @xmath123 as explained in the previous section .
thus , the discrete version of ( [ supp_eq : stat ] ) reads : @xmath124 it is important to notice that matrices @xmath118 and @xmath119 correspond to either @xmath125 and @xmath126 , depending on the symmetry of @xmath127 and @xmath128 , which , at the same time , depends on the value of the vorticity @xmath129 . if @xmath129 is even , then @xmath127 ( @xmath128 ) is ( anti-)symmetric and @xmath130 ( @xmath131 ) . on the contrary , if @xmath129 is odd , then @xmath128 ( @xmath127 ) is ( anti-)symmetric and @xmath132 ( @xmath133 ) . in order to find the roots of the vector function @xmath134 , an analytical expression of the jacobian matrix @xmath135 \dfrac{\partial \mathbf{f^{(2)}}}{\partial \mathbf{v } } & \dfrac{\partial \mathbf{f^{(2)}}}{\partial \mathbf{u } } \end{array}\right)= \left(\begin{array}{cc } ( 1-\omega)-(v^2-u^2)^{k-1}[(2k+1)v^2-u^2 ] & 2kuv(v^2-u^2)^{k-1}+d^{(2)}+\dfrac{s+1}{r } \\[3 ex ] -2kuv(v^2-u^2)^{k-1}+d^{(1)}-\dfrac{s}{r } & ( 1+\omega)-(v^2-u^2)^{k-1}[v^2-(2k+1)u^2 ] \end{array}\right)\ ] ] must be introduced , with the derivatives expressed by means of spectral methods and the matrix must be evaluated at the corresponding grid points .
the roots of @xmath136 , @xmath137 , are found by successive application of @xmath138 until convergence is attained . in our case , we have fixed as convergence condition that @xmath139 .
to find the spectrum of linearization at a solitary wave in two spatial dimensions , we consider a solution @xmath34 in the form of a perturbed solitary wave solution : @xmath140 \ !
i\big(u(r)+\xi_2(r,\theta , t)+i\eta_2(r,\theta , t)\big){\mathrm{e}}^{i(s+1)\theta } \end{array } \!\right ] { \mathrm{e}}^{-i\omega t},\ ] ] with @xmath141 and @xmath142 corresponding to a small perturbation . the linearized equation on @xmath143^t
\in\mathbb{r}^4 $ ] has the form @xmath144 with @xmath145 a matrix - valued first order differential operator @xmath146 -l_{+}&-\sigma_1\frac{{\partial}\sb\theta}{r } \end{array } \right],\ ] ] where @xmath1471&0\end{array}\right ) , \qquad l_{-}= \left ( \begin{array}{cc } f-\omega&{\partial}\sb r+\frac{s+1}{r } \\[1ex ] -({\partial}\sb r-\frac{s}{r})&-f-\omega \end{array } \right ) , \qquad l_{+}=l_{- } + 2\left ( \begin{array}{cc } v^2&-uv \\[1ex ] -uv&u^2 \end{array } \right)f',\ ] ] with @xmath148 , @xmath149 , and with @xmath150 , @xmath151 evaluated at @xmath152 . to find the spectrum of the operator @xmath153 , we consider it in the space of @xmath154-valued functions .
the key observation which facilitates a computation of the spectrum is that the explicit form of @xmath153 contains @xmath47 , @xmath48 , @xmath49 , but not @xmath50 . as a consequence
, @xmath153 is invariant in the spaces which correspond to the fourier decomposition with respect to @xmath50 , @xmath155e^{iq\theta}\right\};\ ] ] the restriction of @xmath153 to each such subspace is given by @xmath156 -l_{+}&-\sigma_1\frac{iq}{r } \end{array } \right ] , \qquad q\in{\mathbb{z}},\ ] ] and this allows us to compute the spectrum of @xmath153 as the union of spectra of the one - dimensional spectral problems : @xmath157 the spectrum @xmath158 is found by evaluating the functions appearing therein at the collocations points and substituting the partial derivatives by the corresponding differentiation matrix . at this point , one must be very cautious because , as also occurred with the jacobian , there will be two different differentiation matrices in our problem .
thus , the representation of @xmath159 matrix will be : @xmath160 -(d^{(1)}-\frac{s}{r})&-f-\omega \end{array } \right)\ ] ] in the 3d case , the nld solitary waves were initially found in @xcite to be of the form @xmath161e^{-i\omega t } , $ ] with @xmath162 and with @xmath163 , @xmath164 satisfying @xmath165 to study the linearization operator in the invariant space which has the same angular dependence as the solitary waves , we consider the perturbed solutions in the form @xmath166e^{-i\omega t}.\ ] ] the linearized equation on @xmath167 is similar to : @xmath168-l_{+}&0\end{array } \right ] , \quad l_{-}= \left ( \begin{array}{cc } f-\omega&{\partial}\sb r+\frac{2}{r } \\[1ex ] -{\partial}\sb r&-f-\omega \end{array } \right ) , \quad l_{+}=l_{- } + 2\left ( \begin{array}{cc } v^2&-uv \\[1ex ] -uv&u^2 \end{array } \right)f'.\ ] ]
the figure below shows the critical frequencies for radially - symmetric collapse , as a function of the exponent @xmath169 associated with the nonlinearity in our generalized nld model . for @xmath61 ,
the nld solitary waves are linearly unstable .
below @xmath170 the linear instability disappears . for @xmath171 , with @xmath172 being the system dimension , there is no linear instability for @xmath173 .
finally , we briefly discuss how the dynamics of eq .
( [ supp_eq : dyn ] ) is simulated . in this case ,
chebyshev spectral methods are not the most suitable ones , because of the presence of many outliers ( see @xcite ) .
in addition , as it was demonstrated in @xcite , finite difference methods also raise a number of concerns for the dirac equation .
thus , the possibility that appears to us to be optimal presently is to use fourier spectral methods , that works fairly well as long as the frequency @xmath59 is not close to zero .
consequently , periodic boundary conditions must be supplied to our problem .
this is less straightforward when working in polar coordinates in the domain @xmath174 .
for this reason , we opt to work with a purely 2d problem in rectangular coordinates in the domain @xmath175\times(-l , l]$ ] .
the simulations of the paper have been performed with a dormand - prince numerical integrator using such a spectral collocation scheme with the aid of fast fourier transforms ( [ supp_eq : fft ] ) . | we explore a prototypical _ two - dimensional _ _ massive _ model of the nonlinear dirac type and examine its solitary wave and vortex solutions .
in addition to identifying the stationary states , we provide a systematic spectral stability analysis , illustrating the potential of spinor solutions to be neutrally stable in a wide parametric interval of frequencies .
solutions of higher vorticity are generically unstable and split into lower charge vortices in a way that preserves the total vorticity .
these conclusions are found not to be restricted to the case of cubic two - dimensional nonlinearities but are found to be extended to the case of quintic nonlinearity , as well as to that of three spatial dimensions .
our results also reveal nontrivial differences with respect to the better understood non - relativistic analogue of the model , namely the nonlinear schrdinger equation . _
introduction .
_ in the context of dispersive nonlinear wave equations , admittedly the prototypical model that has attracted a wide range of attention in optics , atomic physics , fluid mechanics , condensed matter and mathematical physics is the nonlinear schrdinger equation ( nls ) @xcite . by comparison ,
far less attention has been paid to its relativistic analogue , the nonlinear dirac equation ( nld ) @xcite , despite its presence for almost 80 years in the context of high - energy physics @xcite .
this trend is slowly starting to change , arguably , for three principal reasons .
firstly , significant steps have been taken in the nonlinear analysis of stability of such models @xcite , especially in the one - dimensional ( 1d ) setting .
secondly , computational advances have enabled a better understanding of the associated solutions and their dynamics @xcite .
thirdly , and perhaps most importantly , nld starts emerging in physical systems which arise in a diverse set of contexts of considerable interest .
these contexts include , in particular , bosonic evolution in honeycomb lattices @xcite and a growing class of atomically thin 2d dirac materials @xcite such as graphene , silicene , germanene and transition metal dichalcogenides @xcite .
recently , the physical aspects of nonlinear optics , such as light propagation in honeycomb photorefractive lattices ( the so - called photonic graphene ) @xcite have prompted the consideration of intriguing dynamical features , e.g. conical diffraction in 2d honeycomb lattices @xcite .
inclusion of nonlinearity is then quite natural in these models , although in a number of them ( e.g. , in atomic and optical physics ) the nonlinearity does not couple the spinor components .
these physical aspects have also led to a discussion of potential 2d solutions of nld in @xcite .
however , a systematic and definitive characterization of stability and nonlinear dynamical evolution of the prototypical coherent structures in nld models is still lacking , to the best of our knowledge .
the present work is dedicated to offering analytical and numerical insights into these crucial mathematical and physical aspects of higher - dimensional nonlinear dirac equations bearing in mind the physical relevance and potential observability of such waveforms .
as our model of choice , in order to also be able to compare and contrast with the multitude of existing 1d results ( e.g. @xcite ) , we select the well - established soler model @xcite ( known in 1d as the gross neveu model @xcite ) , which is a dirac equation with scalar self - interaction . such self - interaction is based on including into the lagrangian density a function of the quantity @xmath0 ( which transforms as a scalar under the lorentz transformations ) : @xmath1 where @xmath2 , @xmath3 , @xmath4 , @xmath5 and @xmath6 , @xmath7 are @xmath8 dirac @xmath9-matrices satisfying the anticommutation relations @xmath10 , with @xmath11 the minkowski tensor @xcite , and @xmath12 .
( the clifford algebra theory gives the relation @xmath13}$ ] between the spatial dimension and spinor components ( * ? ? ?
* chapter 1 , 5.3 ) . )
the nonlinearity of the model is generalized in the spirit of @xcite , by using @xmath14 with @xmath15 .
the proof of existence of solitary waves in this model ( in 3d ) is in @xcite .
our results show that the nld in 2d admits different solutions involving a structure of vorticity @xmath16 in the first spinor component , with the other spinor component bearing a vorticity @xmath17 .
we identify such solutions for @xmath18 .
while prior stability results have often been inconclusive ( particularly in higher dimensions , see , e.g. , @xcite ) , our numerical computation of the spectrum of the corresponding linearization operator reveals that _ only _ the @xmath19 solutions can be spectrally stable ( the spectrum of the linearization contains no eigenvalues with positive real part ) , and that this stability takes place in a rather wide interval of the frequency of the solitary waves . on the contrary , we find that the states of higher vorticity are generically linearly unstable . complementing the stability analysis results , our direct dynamical evolution studies show that the unstable higher vorticity solutions break up into lower vorticity waveforms , yet conserving the total vorticity .
importantly , the fundamental @xmath19 solutions are found to be potentially stable in models both with a higher order ( quintic ) two - dimensional nonlinearity , as well as in higher dimensions ( 3d ) under cubic nonlinearity .
these features again reflect differences from the nls model and as such suggest the particular interest towards a broader and deeper study of nld models . an important extension of our stability findings for higher dimensional @xmath19 solutions , is that they remain valid for other types of nonlinearities .
these include non - lorentz - invariant ones such as most notably those arising in atomic @xcite , and optical @xcite problems .
the fundamental difference of those models is that they correspond to _ massless _ equations , contrary to the soler model .
for this reason , we have confirmed our stability conclusions by comparison with those emerging from the model for square binary waveguides @xcite which leads to a _ massive _ nonlinear dirac equation with the same nonlinearity as in @xcite .
_ theoretical setup .
we start from the prototypical 2d nonlinear dirac equation system , derived from the lagrangian density ( [ nld - lagrangian ] ) with @xmath20 and @xmath21 : @xmath22 where @xmath23 , @xmath24 are the components of the spinor @xmath25 and the nonlinearity is @xmath26 .
we note that is a @xmath27-invariant , translation - invariant hamiltonian system .
we simplify our analysis by using the polar coordinates , where eq .
( [ eq : dyn ] ) takes the form @xmath28 the form of this equation suggests that we look for solutions as @xmath29 with @xmath30 , \label{ansatz}\ ] ] with @xmath31 and @xmath32 real - valued .
the value @xmath16 can be cast as the vorticity of the first spinor component .
once solitary waves have been identified , we explore their stability .
this approach has been previously developed in related settings including the multi - component nls ( see e.g. @xcite ) , as well as a massless variant of the dirac equation of @xcite .
the presence of a mass in our case allows not only a direct comparison with nls ( when @xmath33 ) , but also generates fundamental differences between our results and those of @xcite , as discussed below as well . to examine its spectral stability
, we consider a solution @xmath34 in the form of a perturbed solitary wave solution : @xmath35 { \mathrm{e}}^{-i\omega t},\ ] ] with @xmath36 a small perturbation .
we consider the linearized equation on @xmath37 , @xmath38 with @xmath39 and with a matrix - valued first order differential operator @xmath40 @xcite .
if the spectrum of the linearization operator @xmath41 contains an eigenvalue @xmath42 with @xmath43 , we say that the solitary wave is linearly unstable ; in such cases , we resort to dynamical simulations of eqs .
( [ eq : dyn ] ) to explore the outcome of the unstable evolution .
if there are no such eigenvalues , the solitary wave is called spectrally stable .
a convenient feature of nls ground states is that the linearization operator at such states , albeit non - selfadjoint , has its point spectrum confined to the real and imaginary axes .
this observation is at the base of the vk criterion @xcite : a linear instability can thus develop when a positive eigenvalue bifurcates from @xmath44 .
more precisely , the loss of stability due to the appearance of a pair of positive and a pair of negative eigenvalues follows the jump in size of the jordan block corresponding to the unitary invariance ; this happens when the vk condition @xmath45 is satisfied , with @xmath46 being the charge of a solitary wave .
crucially , in the nld case , the spectrum of the linearization at a solitary wave is no longer confined to the real and imaginary axes ; the linear stability analysis requires that one studies the _ whole complex plane_. the key observation is that @xmath41 in contains @xmath47 , @xmath48 , @xmath49 , but not @xmath50 ; this allows to perform a detailed study of the spectrum of @xmath41 using the decomposition of spinors into fourier harmonics corresponding to different @xmath51 @xcite . in the 3d case , we are not yet able to perform the general spectral analysis , but we studied the part of spectrum in the invariant subspace corresponding to perturbations of the same angular structure as the solitary waves @xcite , @xmath52,iu(r)[\cos\theta , e^{i\phi}\sin\theta]]^t$ ] ; this invariant subspace seems most important since it is responsible for the linear instability in the non - relativistic limit @xmath53 which is a consequence of the instability of the 3d cubic nls . [ cols="^,^ " , ]
it is especially interesting that a wide parametric ( over frequencies ) interval of stability of solitary waves can _ also _ be observed in the quintic ( @xmath54 ) nld case ( see fig .
[ fig : spectra - n2k2-n3k1 ] ) ; while the quintic nls solitary waves blow up ( even in one dimension ) , the quintic nld solitary waves are stable even in two dimensions , except for the interval @xmath55 where the coherent structures experience the same hopf bifurcation as in the cubic case , and for @xmath56 where an exponential instability created by radial @xmath57 perturbations emerges .
perhaps even more remarkably , the right panel of the fig .
[ fig : spectra - n2k2-n3k1 ] illustrates that this stability of nld solitons against radial perturbations can be found in suitable frequency intervals _ even in 3d _ ( see @xcite for a discussion of the equations for existence and stability of radial perturbations in 3d ) .
both of the above cases ( quintic 2d and cubic 3d nld ) are charge - supercritical i.e. , the charge goes to infinity in the nonrelativistic limit @xmath58 .
contrary to the pure - power supercritical nls whose solitary waves remain linearly unstable for all frequencies , solitary waves in the soler model become spectrally stable when @xmath59 drops below some dimension - dependent critical value @xmath60 @xcite .
the relevant unstable eigenvalue ( associated with @xmath57 and radially - symmetric collapse ) is only present as real for @xmath61 , where @xmath62 .
this was identified in @xcite as the value at which both the energy and charge of solitary waves have a minimum .
hence , we indeed find that the radially - symmetric collapse - related instability ceases to be present below this critical point
. finally , as regards two dimensions , @xmath63 vortices are unstable for every @xmath59 , because of the presence in the spectrum of quadruplets of complex eigenvalues .
these quadruplets emerge ( and disappear ) for different values of @xmath64 via direct ( inverse hopf ) bifurcations ; see the right panel of fig .
[ fig : spectra ] .
the spectrum for @xmath65 vortex is quite similar to that of @xmath63 ; for this reason , we do not analyze it further . in order to analyze the result of instabilities in 2d settings
, we have probed the dynamics of unstable solutions directly ( see @xcite for details ) .
prototypical examples of unstable @xmath19 solitary waves and @xmath63 vortices for @xmath20 are shown in figs . [
fig : dyns0 ] and [ fig : dyns12 ] . as can be observed ,
the @xmath19 solitary waves spontaneously amplify perturbations breaking the radial symmetry in their density and , as a result , become elliptical and rotate around the center of the circular density of the original solitary wave in line with the expected amplification of the @xmath66 unstable eigenmode .
on the other hand , the @xmath63 vortices split into three smaller ones .
let us mention that in the latter case , the first spinor component splits into structures without angular dependence , whereas the second component splits into corresponding ones with angular dependence @xmath67 , in accordance with the ansatz of eq .
( [ ansatz ] ) .
this preserves the total vorticity across the two components , as is also shown in fig .
[ fig : dyns12 ] . along a similar vein
, the instability of an @xmath65 vortex eventually leads to the emergence of five @xmath68 pairs , again preserving the total vorticity .
finally , we have analyzed the outcome of the instabilities caused by radially - symmetric perturbations in the @xmath54 case for @xmath69 ( see fig . [
fig : dyns12 ] ) .
we can observe the typical behavior of such solutions , i.e. the density width ( and amplitude ) oscillate leading to a `` breathing '' structure , but there is no collapse .
this phenomenology is reminiscent of the 1d case @xcite . solitary wave with @xmath70 .
the soliton which initially had a circular shape becomes elliptical and rotates around the center of the original solitary wave.,title="fig:",width=264 ] solitary wave with @xmath70 .
the soliton which initially had a circular shape becomes elliptical and rotates around the center of the original solitary wave.,title="fig:",width=264 ] ( top ) and @xmath65 ( center ) vortex with @xmath20 , @xmath71 and ( bottom ) an @xmath19 solitary wave with @xmath54 and @xmath72.,title="fig:",width=264 ] ( top ) and @xmath65 ( center ) vortex with @xmath20 , @xmath71 and ( bottom ) an @xmath19 solitary wave with @xmath54 and @xmath72.,title="fig:",width=264 ] ( top ) and @xmath65 ( center ) vortex with @xmath20 , @xmath71 and ( bottom ) an @xmath19 solitary wave with @xmath54 and @xmath72.,title="fig:",width=264 ] _ conclusions and future challenges .
_ we have illustrated that solitary waves of vorticity @xmath19 in one spinor component and @xmath63 in the other are _ spectrally stable within a large parametric interval _ , suggesting their physical relevance . in that connection
, we highlight that although our models of choice may bear a particular nonlinearity , our results suggest that under different nonlinearities including the more physically relevant ones of e.g. , @xcite and massive models @xcite _ still _ bear stable solitary waves for a suitable _ wide parametric range _ of frequencies .
thus , the conclusion of higher dimensional stability is more general than the specifics of our particular nonlinearity and hence of broad interest .
we also showcased the significant difference of nld from the focusing nls equation , where solitary waves are linearly unstable in the charge - supercritical cases . when the nld solutions were found to be unstable , their dynamical evolution suggested breathing oscillations in the @xmath19 case and splitting into lower charge configurations for @xmath63 and @xmath65 .
it is of interest to extend present considerations to numerous settings . from a mathematical physics perspective
, it would be useful to explore further the 3d stability and associated dynamics .
this is especially timely given that the 3d analogue of photonic graphene has been experimentally realized very recently @xcite .
admittedly , the latter setting does not feature a mass in the model , thus the generalization of nld models such as those appearing in the works of @xcite would be particularly important there .
it would also be of interest to compare more systematically the present findings with models associated with different nonlinearities , including the case of honeycomb lattices in atomic and optical media or , e.g. , those stemming from wave resonances in low - contrast photonic crystals @xcite .
the research of a. comech was carried out at the institute for information transmission problems of the russian academy of sciences at the expense of the russian foundation for sciences ( project 14 - 50 - 00150 ) .
a.s . was supported by the u. s. department of energy .
p.g.k . gratefully acknowledges the support of the us - nsf under the award dms-1312856 , and of the erc under fp7 , marie curie actions , people , international research staff exchange scheme ( irses-605096 )
. 99 yu.s .
kivshar and g.p .
agrawal , _ optical solitons : from fibers to photonic crystals _ , academic press ( san diego , 2003 ) .
l.p .
pitaevskii and s. stringari , _ bose - einstein condensation _ , oxford university press ( oxford , 2003 ) .
c.j .
pethick and h. smith , _ bose - einstein condensation in dilute gases _ , cambridge university press ( cambridge , 2002 ) .
p.g .
kevrekidis , d.j .
frantzeskakis , and r. carretero - gonzlez , _ the defocusing nonlinear schrdinger equation : from dark solitons and vortices to vortex rings_. siam , philadelphia ( 2015 ) .
m.j .
ablowitz , b. prinari , and a.d .
trubatch , _ discrete and continuous nonlinear schrdinger systems _ , cambridge university press ( cambridge , 2004 ) . c. sulem and p.l .
sulem , , springer - verlag ( new york , 1999 ) .
j. bourgain , _ global solutions of nonlinear schrdinger equations _
, american mathematical society ( providence , 1999 ) .
f.m .
toyama , y. hosono , b. ilyas , and y. nogami , j. phys . a : math
. gen .
* 27 * , 3139 ( 1994 ) .
d. d. ivanenko , zh .
ksp .
teor .
fiz * 8 * , 260 ( 1938 ) .
r. finkelstein , r. lelevier , and m. ruderman , phys .
rev . * 83 * , 326 ( 1951 ) .
w. heisenberg , rev .
mod .
phys .
* 29 * , 269 ( 1957 ) .
w. thirring , ann . phys . * 3 * , 91 ( 1958 ) .
s.y .
lee , t. k. kuo , and a gavrielides , phys .
rev .
d * 12 * , 2249 ( 1975 ) .
n. boussad and s. cuccagna , comm .
pde * 37 * , 1001 ( 2012 ) . d.e .
pelinovsky and y. shimabukuro , lett .
math .
phys . * 104 * , 21 ( 2014 ) .
a. contreras , d.e .
pelinovsky and y. shimabukuro , comm .
pde * 41 * , 227 ( 2016 ) .
d.e . pelinovsky and a. stefanov , j. math . phys . * 53 * , 073705 ( 2012 ) . a. comech , t. phan , and a. stefanov , arxiv:1407.0606 .
n. boussad and a. comech , arxiv:1211.3336 .
j. xu , s. shao and h. tang , j. comp .
phys * 245 * , 131 ( 2013 ) .
s. shao , n.r .
quintero , f.g .
mertens , f. cooper , a. khare , and a. saxena , phys .
rev .
e * 90 * , 032915 ( 2014 ) .
f. cooper , a. khare , b. mihaila , and a. saxena , phys .
rev .
e * 82 * , 036604 ( 2010 ) .
g. berkolaiko and a. comech , math . model .
nat .
phenom .
* 7 * , 13 ( 2012 ) .
j. cuevas - maraver , p.g .
kevrekidis , and a. saxena , j. phys .
a * 48 * , 055204 ( 2015 ) .
l. haddad , k. ohara , and l.d .
carr , phys .
rev . a * 91 * , 043609 ( 2015 ) . l. haddad and l.d .
carr .
new j. phys .
* 17 * , 113011 ( 2015 ) .
t. o. wehling , a. m. black - schaffer , and a. v. balatsky , adv .
phys . * 63 * , 1 ( 2014 ) .
k. f. mak , c. lee , j. hone , j. shan , and t. f. heinz , phys .
rev .
lett . * 105 * , 136805 ( 2010 ) .
m. j. ablowitz , s.d .
nixon , and y. zhu , phys .
rev .
a * 79 * , 053830 ( 2009 ) .
m. j. ablowitz and y. zhu , phys .
rev .
a * 82 * , 013840 ( 2010 ) .
o. peleg , g. bartal , b. freedman , o. manela , m. segev , and d.n .
christodoulides , phys .
rev .
lett . * 98 * , 103901 ( 2007 ) .
m. soler , phys .
rev .
d * 1 * , 2766 ( 1970 ) .
d.j .
gross and a. neveu , phys .
rev .
d * 10 * , 3235 ( 1974 ) . b. de wit and j. smith , _ field theory in particle physics _ , north holland physics publishing ( new york , 1986 ) .
b.v .
fedosov , _ index theorems _ , in : _ partial differential equations , viii encyclopaedia math .
sci .
_ , * 65 * , 155
. springer verlag ( berlin , 1996 ) .
t. cazenave and l. vzquez , comm .
math .
phys . * 105 * , 35 ( 1986 ) .
f. merle , j. differential equations * 74 * , 50 ( 1988 ) .
m.j . esteban and e. sere ,
commun .
math .
phys . * 171 * , 323 ( 1995 ) .
a. alvarez and m. soler , phys .
rev .
lett . * 50 * , 1230 ( 1983 ) .
n.g .
vakhitov and a.a .
kolokolov , radiophys .
quantum electron . * 16 * , 783 ( 1973 ) .
t.x .
tran , x.n .
nguyen , and f. biancalana , phys .
rev . a * 91 * , 023814 ( 2015 ) .
z.h .
musslimani , m. segev , d.n .
christodoulides , and m. soljai , phys .
rev .
lett . * 84 * , 1164 ( 2000 ) ; j. yang and d.e .
pelinovsky phys .
rev .
e * 67 * , 016608 ( 2003 ) .
l. h. haddad and l.d .
carr , epl * 94 * , 56002 ( 2011 ) ; l. h. haddad and l.d .
carr , new j. phys .
* 17 * , 063034 ( 2015 ) .
see the supplementary material ` [ url ] ` , which includes refs .
@xcite , @xcite , @xcite @xcite .
m. wakano , progr .
. phys . * 35 * , 1117 ( 1966 ) .
f. merle , comm .
math .
phys . * 129 * , 223 ( 1990 ) .
j. cuevas - maraver , p.g .
kevrekidis , a. saxena , f. cooper , a. khare , a. comech , and c.m .
bender , j. sel .
top .
. electr .
* 22 * , 5000109 ( 2016 ) .
a. comech , m. guan and s. gustafson , ann . i. h. poincar an * 31 * , 639 ( 2014 ) .
j. cuevas - maraver , p.g .
kevrekidis , a. saxena , f. cooper and f.g .
mertens , in : _ ordinary and partial differential equations _ , chapter 4 .
nova science publishers ( new york , 2015 ) .
l. lu , z. wang , d. ye , l. ran , l. fu , j.d .
joannopoulos , m. soljacic , science * 349 * , 622 ( 2015 ) .
d. agueev and d. pelinovsky , siam j. appl . math . *
65 * , 1101 ( 2005 ) .
j.p .
boyd , _ chebyshev and fourier spectral methods _ ,
2nd . edition .
dover ( 2001 ) .
l.n .
trefethen , _ spectral methods in matlab_. siam , philadelphia ( 2000 ) .
g. herring , l.d .
carr , r. carretero - gonzlez , p.g .
kevrekidis , and d.j .
frantzeskakis .
phys . rev .
a * 77 * , 023625 ( 2008 ) .
w.h . press ,
s.a .
teukolsky , w.t .
vetterling and b.p .
flannery , _ numerical recipes : the art of scientific computing _ ,
3rd . edition .
cambridge university press ( 2007 ) .
* stability of solitary waves and vortices in a 2d nonlinear dirac model .
supplementary material * |
graphene is a two - dimensional ( 2d ) carbon material with tremendous applications . in 1962
it was isolated by boehm _
for the first time @xcite .
more recently , in 2004 , a team led by geim proposed a more efficient way to produce it based on exfoliation procedures .
that moment was the starting point for the rise of graphene @xcite .
the research was fueled by the extraordinary electronic transport properties of graphene which differ substantially from those in metals and semiconductors @xcite .
in fact , electrons behave like massless particles when they propagate in graphene sheets , like photons do in matter with the notable difference that they have an electric charge .
these unprecedented electronic properties make graphene a serious candidate for the material of the @xmath0 century @xcite to be used for ultrafast , low loss electronic devices because of its high conductivity and its extraordinary mechanical parameters .
likewise the optical properties of graphene attracted a great deal of research interest . in the visible , graphene acts as an absorptive dielectric .
the absorption _ a _ of a single sheet amounts to @xmath1 @xcite , with @xmath2 being the fine structure constant .
various applications , ranging from graphene based optical modulators @xcite , transformation optics @xcite to numerous other devices @xcite may be envisaged by using graphene in the optical domain .
examples have been already experimentally demonstrated and have been provided solid evidence for the versatility of graphene @xcite . at lower frequencies ,
the optical properties of graphene resemble those of a drude - type material .
such materials promote the formation of surface plasmon polaritons ( spp ) .
an spp is a collective oscillation of the charge density and light at the interface between a graphene sheet and its surrounding @xcite .
an spp supported at a graphene interface can be guided in many ways like other types of waves ( light or sound ) .
this property attracted a particular interest of the metamaterial and plasmonic community @xcite .
one of the most interesting properties of plasmonic metamaterials is the perfect absorption of light @xcite .
such functionality was demonstrated while relying on different designs and while optimizing the point of operation over large frequency ranges @xcite .
the functionality of such perfect absorbers can be understood in the framework of the more general theory of coherent perfect absorption ( cpa ) .
it is in many ways the time - reversed analogue of the lasing effect . instead of radiating a coherent flux of light in a very narrow spectral band @xcite
, cpa can occur for structures possessing intrinsic losses .
it only requires that an incoming radiation pattern is dissipated by suppressing the associated reflection and transmission channels .
in fact , the cpa effect can only occur because of the interplay of interference and absorption being achievable just for a very specific geometry and frequency range .
more recently , graphene nanodisks were suggested for this purpose @xcite .
it was shown that graphene flakes allow for the perfect absorption of thz radiation owing to their high extinction cross section ( compared to their geometrical cross section ) .
such perfect absorption has also been achieved when the discs were put in front of a mirror at some distance .
recently graphene micro - ribbons is theoretically and experimentally studied @xcite . in this paper
we propose to take advantage of the recent technological progress in fabricating nanostructured graphene layers grown on dielectric substrates @xcite .
we show that perfect absorption can be achieved by patterned graphene micro - ribbons on a thick dielectric layer deposited on top of a reflecting metal substrate .
furthermore , we show that the complete light absorption is almost independent of angle of incidence and its frequency can be easily tuned by changing the chemical potential or gate voltage .
the _ modus operandi _ of the suggested perfect absorber differs from that of the ordinary ones made of metallic nanoparticles brought in close proximity to a metallic ground plate .
there , the current induced in the ground plate via a near - field interaction is out - of - phase to the current in the nanoparticle .
this causes the resonance to be dark and radiative losses to be reduced .
if the condition of critical coupling is achieved , the incident light can be fully absorbed @xcite . in striking contrast
, the device we suggest and study here is characterized by a thick dielectric intermediate layer that prohibits any near - field interaction between the nanostructured graphene and the ground plate .
this type of perfect absorber is known as the salisbury screen or absorber @xcite . instead , perfect absorption is achieved by exploiting a perfect destructive interference of the reflected light @xcite .
transmission through the structure is totally suppressed because the thickness of the ground plate is much larger than the typical skin depth at thz frequency .
therefore , the complete electromagnetic energy gets absorbed by the graphene micro - ribbons .
we explain our findings by considering the micro - ribbons as a metasurface .
this allowed us to describe them only by their reflection and transmission coefficients and using airy formula for an asymmetric fabry perot cavity to calculate the absorption in the device @xcite .
predictions from such simple model are in excellent agreement with full - wave simulations and clearly prove that the perfect absorption is a purely coherent effect .
the structure under consideration is shown in fig . [ fig_gpa ] .
it consists of a graphene micro - ribbon array on top of a metallic ground plate separated by a thick dielectric spacer .
it is periodic in one direction ( @xmath3 ) with periodicity @xmath4 m and infinitely extended in the other one .
we assume the refractive index ( @xmath5 ) of the dielectric deposited on the metal .
the ground plate is made of gold with a conductivity @xmath6s / m which is perfectly reflecting in the frequency domain of interest ( far - infrared regime ) .
@xmath7 m , @xmath8 @xmath7 m , @xmath9 @xmath7 m and @xmath10 @xmath7 m .
the inset shows the honeycomb microscopic structure of the graphene layer arranged as micro - ribbons.,width=359 ] the graphene was numerically modeled by a thin layer ( with thickness @xmath11 nm ) of permittivity @xmath12 , with @xmath13 as the surface conductivity of the graphene sheet @xcite which can be derived using the well - known kubo formula @xcite and writes as : @xmath14 + \frac{ie^2k_\mathrm{b}t}{\pi\hbar^2(\omega+i2\gamma)}\left[\frac{\mu_c}{k_\mathrm{b}t}+2\ln(e^{-\mu_c / k_\mathrm{b}t}+1)\right ] , \label{kubo}\end{aligned}\ ] ] where @xmath15 , @xmath16 and @xmath17 are universal constants related to the electron charge , planck s and boltzmann s constant , respectively .
@xmath18 is the temperature and is fixed to 300 k. @xmath19 and @xmath20 ( @xmath21 , @xmath22 is the electron - phonon relaxation time ) are physical parameters of the graphene sheet and account for the chemical potential ( or fermi energy ) and the intrinsic losses , respectively .
we assumed @xmath23 mev in this contribution and this value is based on the theoretical estimation of maximum mobility in graphene @xcite .
furthermore , it should be noted that even for a larger scattering rate the general functionality of proposed device is not affected and all the phenomena can be observed .
the frequency of operation @xmath24 is in the far - infrared regime ( 1 - 10 thz ) .
the parameter @xmath25 was taken 1 nm in our study , and the convergence of @xmath26 was verified numerically ( by taking the limit @xmath27 ) and by reproducing some previous results @xcite .
in addition , the frequency dependent permittivity of graphene depending on the some selected values of the chemical potential are shown in fig .
[ fig_absgpa](a ) and [ fig_absgpa](b ) . ) and dielectric thickness ( @xmath28 ) for a graphene micro - ribbon perfect absorber with chemical potential @xmath29mev at normal incidence .
( d ) absorption as a function of frequency and chemical potential for graphene micro - ribbons perfect absorber with @xmath30 @xmath7 m at normal incidence .
( e ) and ( f ) show the absorption as a function of the angle of incidence with dielectric thickness @xmath30 @xmath7 m and @xmath31 @xmath7 m for the first and second mode , respectively .
the insets show the electric field distribution @xmath32 for both modes as well.,width=359 ] the geometrical parameters of our structure are selected based on experimental capabilities @xcite , a filling fraction @xmath33 was shown to be easier for fabrication ( see fig .
[ fig_gpa ] ) .
nevertheless , the very parameters are of marginal importance and variations would only cause a slight modification of the frequency of operation .
the set - up is illuminated by a transverse magnetic ( tm ) incident plane wave .
the fourier modal method ( fmm ) is used to explore the underlying physics of the absorber @xcite .
the method solves maxwell s equations for the periodic structure rigorously . in order to have convergent results ,
101 fourier orders were selected in all numerical results presented herein .
figure [ fig_absgpa](c ) shows the absorption _ a _ of the graphene micro - ribbons ( of fig .
[ fig_gpa ] ) as function of the thickness of the spacer @xmath28 and the frequency of operation .
the strong coupling of the fabry - perot resonance with the localized eigenmode of the graphene micro - ribbons is clearly visible in fig . [ fig_absgpa](c ) and this leads to an avoided crossing and a significant rabbi splitting .
the results reveal that perfect absorption ( @xmath34 ) can be achieved at various values of @xmath28 , periodically spaced between 1 and 100 @xmath7 m around a frequency of 3.6 thz for the first mode ( dipolar one ) and around 7.3 thz for the second mode , where the peaks are less pronounced ( @xmath35 ) .
however , by optimizing the geometry and tuning the chemical potential , we have verified that it is possible to achieve total absorption for both modes ( this is not included in this contribution ) . moreover
, the resonance frequency of maximum absorption is slightly changing for different dielectric spacer and this effect is more pronounced for the first mode ( see fig . [ fig_absgpa](c ) ) . in fig .
[ fig_absgpa](d ) the absorption as a function of the chemical potential @xmath19 ( ranging from 50 to 200 mev ) and the frequency is displayed .
it can be clearly seen that the resonance frequency of the absorber can be tuned over a wide range by varying @xmath19 .
it should be mentioned that by applying a gate voltage ( a static electric field ) or by means of chemical doping , the chemical potential and thus the conductivity of graphene can be controlled on purpose . in order to investigate the robustness of the graphene perfect absorber , the absorption as a function of frequency and the angle of incidence
is displayed in fig .
[ fig_absgpa](e ) and [ fig_absgpa](f ) .
the first mode is almost unaffected while varying the angle of incidence @xmath36 until 80 degrees [ fig .
[ fig_absgpa](e ) ] .
then absorption starts to decrease .
the behavior of the higher order mode ( the second mode for instance ) is quite different , as shown in fig .
[ fig_absgpa](f ) , where the absorption is strongly dependent on the angle of incidence and significantly decreases beyond 50 degrees to vanish completely at about 80 degrees .
the electric field distribution @xmath32 ( component parallel to the gap ) at plasmon resonance is displayed for both modes in the insets of fig .
[ fig_absgpa](e ) and [ fig_absgpa](f ) , respectively , and confirms the nature of the excited modes [ dipole and higher order ] .
let us focus now on the underlying mechanism of perfect absorption for graphene . to simplify the structure and to provide a physical explanation
, we shall consider the structure as an asymmetric fabry - perot cavity with two mirrors , i.e. , a graphene micro - ribbon array as the top mirror and the metallic ground plate as the bottom mirror .
the transmission channel ( @xmath18 ) of the system is completely suppressed by choosing a sufficiently thick ground plate ( @xmath37 ) . in order to achieve total absorption of the incident energy ( @xmath38 ) the reflection channel
must be closed as well ( @xmath39 and the multiple reflection coefficient @xmath40 [ as sketched in fig .
[ fig_airy&fmm](e ) ] .
the total reflection coefficient can be expressed as @xcite : @xmath41 where , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 are complex - valued transmission and reflection coefficients at both interfaces , @xmath47 is the phase accumulated upon a cavity transfer , @xmath48 is the free space wavenumber , @xmath49 is the refractive index of the dielectric , and @xmath28 is its thickness . ) as well as multiple reflection coefficient ( @xmath40 ) with dielectric thickness @xmath30 @xmath7 m and @xmath31 @xmath7 m for the first and the second mode respectively .
the red arrows represent the crossing point when real or imaginary parts of reflection coefficients has same magnitude with opposite sign ( @xmath50).(e ) schematic of the asymmetric fabry - perot cavity with the definition of the multiple reflection and transmission coefficients involved and an indication on the contribution of the directly reflected light and the reflected light due to multiple scattering inside the fabry - perot cavity .. ,width=359 ] the dependence of the absorption maxima on the dielectric spacer thickness @xmath28 calculated by using the airy formula ( eq . [ eq_airy ] ) is shown in fig .
[ fig_airy&fmm ] for the first ( a ) and the second ( b ) modes .
this formula could be used by considering the graphene micro - ribbons as a metasurface ( this could be justified by its extremely small thickness compared to other dimensions and to the wavelength ) , and it is completely defined by its reflection and transmission coefficients . these coefficients were calculated using fmm while assuming that the sufficient thin graphene metasurface is sandwiched between two semi - infinite half - spaces ; one of them being air ; the other being the dielectric . assuming an excitation with a linearly polarized incident plane wave either from the top medium ( air ) or from the bottom medium ( dielectric ) , provides unambiguously the coefficients @xmath44 , @xmath42 , @xmath45 , and @xmath43 , respectively . for comparison
, the rigorous calculations ( using fmm ) are also displayed in the same figure , and a perfect agreement between the two methods is evident for both modes [ see fig.[fig_airy&fmm](a ) and [ fig_airy&fmm](b ) ] .
in addition , the real and imaginary parts of the multiple reflection coefficient @xmath40 and the direct reflection coefficient @xmath44 are plotted in fig . [ fig_airy&fmm](c ) and [ fig_airy&fmm](d ) again for both modes .
let us concentrate for a while on the first mode [ fig .
[ fig_airy&fmm](c ) ] ; at the resonance frequency @xmath51 thz , and for a spacer thickness @xmath52 m , both coefficients @xmath40 and @xmath44 have the same phase and opposite amplitude .
hence , the resulting total reflection coefficient is completely canceled due to destructive interference ( @xmath53 ) . as a result ,
the absorption is almost unity for this frequency .
however , we expect for the second mode , resonant at @xmath54 thz that the multiple reflection coefficient can not totally eliminate the direct reflection coefficient at @xmath31@xmath7 m because the absorption is less than unity , [ see fig.[fig_airy&fmm](d ) ] . but
as mentioned earlier , this issue could be fixed by optimizing the structure which is beyond the scope of this paper .
to conclude , in this contribution we have discussed a new class of perfect absorbers in the far - infrared regime based on graphene micro - ribbons .
our structure showed complete absorption for the first mode and for different geometrical configurations .
the robustness of the mechanism was evidenced by analyzing its dependence on the angle of incidence and the chemical potential of the graphene sheets .
moreover , we utilized a semi - analytical approach based on a fabry - perot model that explains very well the observed behavior .
we believe that our study demonstrates the versatility of graphene and could be used in future for optical or thz interconnects .
this work was partially supported by the german federal ministry of education and research ( metamat and phona ) and by the thuringian state government ( mema ) . | metamaterial - based perfect absorbers promise many applications .
perfect absorption is characterized by the complete suppression of transmission and reflection and complete dissipation of the incident energy by the absorptive meta - atoms .
a certain absorption spectrum is usually assigned to a bulk medium and serves as a signature of the respective material . here
we show how to use graphene flakes as building blocks for perfect absorbers .
then , an absorbing meta - atom only consists of a molecular monolayer placed at an appropriate distance from a metallic ground plate .
we show that the functionality of such device is intuitively and correctly explained by a fabry - perot model
. 99 h. p. boehm , a. clauss , g. o. fischer , and u. hofmann , dnnste kohlenstoff - folien , " z. naturforschg .
* 17 * , 150157 ( 1962 ) .
k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , y. zhang , s. v. dubonos , i. v. grigorieva , and a. a. firsov , electric field effect in atomically thin carbon films , " science * 306 * , 666669 ( 2004 ) .
a. h. castro neto , f. guinea , n. m. r. peres , k. s. novoselov , and a. k. geim , the electronic properties of graphene , " rev .
mod .
phys . * 81 * , 109162 ( 2009 ) .
s. stankovich , d. a. dikin , g. h. b. dommett , k. m. kohlhaas , e. j. zimney , e. a. stach , r. d. piner , s. t. nguyen , and r. s. ruoff , graphene - based composite materials , " nature * 442 * , 282286 ( 2006 ) . c. berger , z. song , x. li , x. wu , n. brown , c. naud , d. mayou , t. li , j. hass , a. n. marchenkov , e. h. conrad , p. n. first , and w. a. de heer , electronic confinement and coherence in patterned epitaxial graphene , " science * 312 * , 11911196 ( 2006 ) .
r. r. nair , p. blake , a. n. grigorenko , k. s. novoselov , t. j. booth , t. stauber , n. m. r. peres , and a. k. geim , fine structure constant defines visual transparency of graphene , " science * 320 * , 1308 ( 2008 ) .
m. liu , x. yin , e. ulin - avila , b. geng , t. zentgraf , l. ju , f. wang , and x. zhang , a graphene - based broadband optical modulator , " nature * 474 * , 6467 ( 2011 ) .
a. vakil , and n. engheta , transformation optics using graphene , " science * 332 * , 12911294 ( 2011 ) .
p .- y .
chen , and a. al , atomically - thin surface cloak using graphene monolayers , " acs nano * 5 * , 58555863 ( 2011 ) .
j. christensen , a. manjavacas , s. thongrattanasiri , f. h. l. koppens , and f. j. g. de abajo , graphene plasmon waveguiding and hybridization in individual and paired nanoribbons , " acs nano * 6 * , 431440 ( 2012 ) .
a. manjavacas , p. nordlander , and f. j. g. de abajo , plasmon blockade in nanostructured graphene , " acs nano * 6 * , 17241731 ( 2012 ) .
l. ju , b. geng , j. horng , c. girit , m. martin , z. hao , h. a. bechtel , x. liang , a. zettl , y. r. shen , and f. wang , graphene plasmonics for tunable terahertz metamaterials , " nat .
nanotechnology * 6 * , 630634 ( 2011 ) .
z. fei , a. s. rodin , g. o. andreev , w. bao , a. s. mcleod , m. wagner , l. m. zhang , z. zhao , m. thiemens , g. dominguez , m. m. fogler , a. h. castro neto , c. n. lau , f. keilmann and d. n. basov , gate - tuning of graphene plasmons revealed by infrared nano - imaging , " nature * 487 * , 8285 ( 2012 ) .
j. chen , m. badioli , p. alonso - gonzlez , s. thongrattanasiri , f. huth , j. osmond , m. spasenovic , a. centeno , a. pesquera , p. godignon , a. z. elorza , n. camara , f. j. g. de abajo , r. hillenbrand and f. h. l. koppens , optical nano - imaging of gate - tunable graphene plasmons , " nature * 487 * , 7781 ( 2012 ) .
e. h. hwang , and s. das sarma , dielectric function , screening , and plasmons in two - dimensional graphene , " phys .
rev .
b * 75 * , 205418 ( 2007 ) .
f. rana , graphene terahertz plasmon oscillators , " ieee trans .
nanotechnol * 7 * , 91 - 99 ( 2008 ) .
m. jablan , h. buljan , and m. soljacic , plasmonics in graphene at infrared frequencies , " phys .
rev .
b * 80 * , 245435 ( 2009 )
. a. yu .
nikitin , f. guinea , f. j. garcia - vidal , and l. martin - moreno ,
fields radiated by a nanoemitter in a graphene sheet , " phys .
rev .
b * 84 * , 195446 ( 2011 ) .
p.tassin , t. koschny , m. kafesaki and c. m. soukoulis , a comparison of graphene , superconductors and metals as conductors for metamaterials and plasmonics , " nature photonics * 6 * , 259264 ( 2011 ) . f. h. l. koppens , d. e. chang , and f. j. g. de abajo , graphene plasmonics : a platform for strong light - matter interactions , " nano lett .
* 11 * , 33703377 ( 2011 ) .
n. i. landy , s. sajuyigbe , j. j. mock , d. r. smith , and w. j. padilla , perfect metamaterial absorber , " phys .
rev .
lett .
* 100 * , 207402 ( 2008 ) . c. m. watts , x. liu , and w. j. padilla , metamaterial electromagnetic wave absorbers , " adv
. mater .
* 24 * , op98op120 , ( 2012 ) .
t. v. teperik , f. j. g. de abajo , a. g. borisov , m. abdelsalam , p. n. bartlett , y. sugawara , and j. j. baumberg , omnidirectional absorption in nanostructured metal surfaces , " nat .
photonics * 2 * , 299-301 ( 2008 ) .
y. avitzour , y. a. urzhumov , and g. shvets , wide - angle infrared absorber based on a negative - index plasmonic metamaterial , " phys .
rev .
b * 79 * , 045131 ( 2009 ) .
k. aydin , v. e. ferry , r. m. briggs , and h. a. atwater , broadband polarization - independent resonant light absorption using ultrathin plasmonic super absorbers , " nat .
commun .
* 2 * , 517 ( 2011 ) .
x. liu , t. tyler , t. starr , a. f. starr , n. m. jokerst , and w. j. padilla , taming the blackbody with infrared metamaterials as selective thermal emitters , " phys .
rev .
lett . * 107 * , 045901 ( 2011 ) .
n. liu , m. mesch , t. weiss , m. hentschel and h. giessen , infrared perfect absorber and its application as plasmonic sensor , " nano lett .
* 10 * , 23422348 ( 2010 ) . c. wu , and g. shvets ,
design of metamaterial surfaces with broad - band absorbance , " opt . lett . * 37 * , 308310 ( 2012 ) . c. wu , b. neuner , g. shvets , j. john , a. milder , b. zollars , and s. savoy , large - area wide - angle spectrally selective plasmonic absorber , " phys .
rev .
b * 84 * , 075102 ( 2011 ) .
y. d. chong , li ge , hui cao , and a. d. stone , coherent perfect absorbers : time - reversed lasers , " phys .
rev .
lett . * 105 * , 053901 ( 2010 ) .
w. wan , y. chong , l. ge , h. noh , a. d. stone , and h. cao , time - reversed lasing and interferometric control of absorption , " science * 331 * , 889892 ( 2011 ) .
s. thongrattanasiri , f. h. l. koppens , and f. j. g. de abajo , complete optical absorption in periodically patterned graphene , " phys .
. lett .
* 108 * , 047401 ( 2012 ) .
a. yu .
nikitin , f. guinea , f. j. garcia - vidal , and l. martin - moreno , surface plasmon enhanced absorption and suppressed transmission in periodic arrays of graphene ribbons , " phys .
rev .
b * 85 * , 081405(r ) ( 2012 ) . c. wu , b. neuner , g. shvets , j. john , a. milder , b. zollars , and s. savoy , large - area wide - angle spectrally selective plasmonic absorber , " phys . rev .
b * 84 * , 075102 ( 2011 ) .
w. w. salisbury , absorbent body for electromagnetic waves , " us patent 2599944 ( 1952 ) .
h. t. chen , interference theory of metamaterial perfect absorbers , " opt .
express * 20 * , 71657172 ( 2012 ) .
l. huang , d. r. chowdhury , s. ramani , m. t. reiten , s .- n .
luo , a. k. azad , a. j. taylor , and h. t. chen , impact of resonator geometry and its coupling with ground plane on ultrathin metamaterial perfect absorbers , " appl .
. lett .
* 101 * , 101102 ( 2012 ) .
p. yeh ,
_ optical waves in layered media _ " wiley , new york , ( 2005 ) .
g. w. hanson ,
dyadic greens functions and guided surface waves on graphene , " j. appl .
phys . * 103 * , 064302 ( 2006 ) .
l. a. falkovsky , s. s. pershoguba , optical far - infrared properties of a graphene monolayer and multilayer , " phys .
rev .
b * 76 * , 153410 ( 2007 ) . e. h. hwang and s. das sarma dielectric function , screening , and plasmons in two - dimensional graphene , " phys .
rev .
b * 75 * , 205418 ( 2007 ) . v. p. gusynin , s. g. sharapov and j. p.
carbotte magneto - optical conductivity in graphene , " j. phys . condens
. matter , * 19 * , 026222 ( 2007 ) .
l. li , new formulation of the fourier modal method for crossed surface - relief gratings , " j. opt .
soc .
am .
a * 14 * , 27582767 ( 1997 ) . |
_ gravity probe b _ ( ) is the spaceborne relativity experiment developed by nasa and stanford university to test two predictions of general relativity ( gr ) .
the experiment used four superconducting gyroscopes , contained in a low - altitude , polar orbiting spacecraft , to measure the geodetic effect and the much smaller frame - dragging effect . according to gr
, each of these effects induces precessions of the gyroscopes .
the predicted precession is 6.6 arcsecyr@xmath3 for the geodetic effect and 39 milliarcsecondsyr@xmath3 ( ) for the frame - dragging effect .
was designed to measure the precessions with a standard error @xmath40.5 relative to distant inertial space .
this space may be closely linked to the extragalactic international celestial reference frame ( icrf ) , which is based on radio very - long - baseline interferometry ( vlbi ) measurements @xcite .
because of technical limitations , the spacecraft could not measure the precessions relative to such a frame directly , but rather only to an optically bright star .
the proper motion of this chosen `` guide star , '' ( = hr 8703 ; hd 216489 ; fk5 3829 ) , was independently determined using vlbi measurements made relative to a fiducial point in the quasar ( = 4c 15.76 ; qso j2253 + 1608 ) .
two other radio sources , and , are used as secondary reference sources to set a limit on the stability of the fiducial point .
the radio structures of these three reference sources and their possible temporal changes are the subject of this paper .
it is one of a series of seven papers reporting on the astronomical support for . for an overview of the paper series
see paper i @xcite .
the quasar 3c 454.3 ( @xmath5 ; * ? ? ?
* ) is an optically violent variable , and one of the brightest extragalactic radio sources ( see , e.g. , * ? ? ?
its milliarcsecond - scale structure has been extensively studied with vlbi . at radio frequencies between
1.7 ghz and 15 ghz , the structure of 3c 454.3 consists of a bright , relatively compact ( unresolved ) core region and an extended one - sided jet which is directed west - southwest near the core region and bends to the northwest @xmath25 mas from the core region ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
the @xmath25 yr vlbi observing program at 10.7 ghz of @xcite traced the motions of four compact components which emerged from the core region and moved at superluminal apparent speeds in the range 59@xmath1 along the curved path of the jet . and a friedmann - robertson - walker cosmology with @xmath6 and @xmath7 . ]
the results at 5 and 8.4 ghz by @xcite give for one particular component a mean motion between 1983 and 1991 of @xmath217@xmath1 . at frequencies of 22 and 43 ghz , the core region itself
is resolved into two principal components oriented approximately east - west and separated by @xmath20.6 mas ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
moreover , observations at 43 ghz reveal that new components are occasionally ejected from the easternmost core - region component and move toward the westernmost core - region component with superluminal speeds anywhere in the range 313@xmath1 @xcite .
the easternmost component is very compact at 43 ghz ( @xmath40.1 mas full - width - at - half - maximum : fwhm ) and is considered at that frequency to be the `` core , '' though its stationarity with respect to a celestial reference frame ( e.g. , the icrf ) has not been determined .
the stationarity of a spatially related component at 8.4 ghz has recently been determined relative to an updated celestial reference frame ( crf ) , defined by @xmath24000 compact extragalactic radio sources ( see paper iii ; @xcite ) .
the radio structure of quasar b2250 + 194 ( @xmath8 ; * ? ? ?
* ) is not nearly so well studied as that of 3c 454.3 .
in fact , only a handful of vlbi images of b2250 + 194 have been produced by others , and mostly as part of the united states naval observatory s astrometry program and the very long baseline array s ( vlba s ) calibrator survey @xcite .
these images show that the source is only partially resolved at both 2.3 and 8.4 ghz on intercontinental baselines .
the radio source b2252 + 172 is catalogued in the national radio astronomy observatory very large array sky survey ( nvss ; * ? ? ?
* ) , but has not previously been studied with vlbi . a spectrum for its optical counterpart has not yet been obtained , so its distance is not yet known . in
[ select ] , we briefly summarize the process of selecting the extragalactic reference sources for our astrometry of the guide star . in
[ obs ] , we describe our 35 sessions of vlbi observations , made between 1997 january and 2005 july . in [ reduc ] , we describe our data reduction and analysis procedures . in
[ 3c454 ] , we show the 35 images made for quasar 3c 454.3 at 8.4 ghz , use model fitting to investigate the relative motions of the observed 8.4 ghz components , and present images made from vlbi observations at 5.0 and 15.4 ghz obtained on 2004 may 18 . in [ 2250 ] , we show the 35 images made for quasar b2250 + 194 at 8.4 ghz , study its radio structure via model fitting , and present images made from vlbi observations at 5.0 and 15.4 ghz on 2004 may 18 . in [ 2252 ] , we show the 12 images made for radio source b2252 + 172 at 8.4 ghz ( which we observed starting only in 2002
november ) , investigate its radio structure via model fitting , and show the images made from the 5.0 and 15.4 ghz vlbi observations on 2004 may 18 . we discuss our results in
[ discuss ] and give our conclusions in
[ conclus ] .
the principal astrometric requirements for our vlbi reference sources are that they be ( 1 ) radio bright and compact , ( 2 ) in close proximity on the sky to i m peg , and ( 3 ) extragalactic in origin , with a core , or other structural reference point , stationary on the sky to a very high degree .
these requirements are consistent with those of targeted astrometric vlbi experiments employing phase referencing ( see , e.g. , * ? ? ?
* ) . to summarize briefly
, phase referencing requires that the reference source be detected in scans @xmath21 minute or less in length , so that the interferometric phase delay , which changes with time , often very rapidly , can be successfully interpolated to scans of the target source . for this reason
, the phase reference source must have a correlated flux density @xmath9100 mjy , i.e. , a flux density of at least this magnitude arising from a sufficiently compact emission region .
similarly , since the phase delay can change significantly with pointing direction , due mainly to radio propagation through the troposphere and ionosphere , the reference source should be located within @xmath21 of the target source .
finally , if the morphology of the reference source is variable between observing sessions , then a point must be identified within the source image at every session that may confidently be assumed to be a ( nearly ) stationary feature of the source over the time spanned by our observing program .
the quasar 3c 454.3 is the only radio source which has both a correlated flux density @xmath10100 mjy and a relatively compact , milliarcsecond - scale structure within 1 of i m peg ( see figure [ foursourcepos ] ) .
furthermore , 3c 454.3 was used as the only reference source for i m peg in the phase - referenced mapping vlbi observations made for the hipparcos frame - tie project from 1991 to 1994 ( see * ? ? ?
our continued observation of 3c 454.3 therefore provides a long time baseline for the guide - star proper - motion measurement . on the other hand ,
3c 454.3 is a classic superluminal source with , in general , moving jet components . without the benefit of a detailed and evolving model of its radio structure , the position accuracy of 3c 454.3
is limited ( in the icrf of * ? ? ?
* ) to @xmath21.5 mas , which is not quite sufficient to yield proper motions to within the @xmath20.1 standard error desired for
. however , since our astrometric vlbi observations were designed so that we could produce high quality images of our reference sources , we can identify a point in 3c 454.3 which is stationary on the sky to a much higher degree .
all things considered , 3c 454.3 was the best choice to be our phase reference and main astrometric reference source .
the quasar b2250 + 194 is the next closest radio source on the sky to i m peg which has a flux density @xmath10100 mjy and is compact on milliarcsecond scales . indeed , b2250 + 194 has a more compact structure and displays more limited structural changes than 3c 454.3 , as evidenced by the accuracy to which its position is measured in the icrf ( @xmath20.3 mas ;
see * ? ? ?
. however , at a separation on the sky from i m peg of 2.9 ( see figure [ foursourcepos ] ) , its use as the primary astrometric reference could have introduced large errors in the stellar proper - motion measurement ( see , e.g. , * ? ? ?
consequently , on the basis of its compact structure and `` reasonably '' small sky separation from i m peg , we selected b2250 + 194 to be a secondary reference source .
the radio source b2252 + 172 is very compact and favorably located within @xmath22 of , and approximately along the same north - south axis as , 3c 454.3 , b2250 + 194 , and i m peg ( see figure [ foursourcepos ] ) .
b2252 + 172 is relatively weak ( @xmath220 mjy at 8.4 ghz ) , and was discovered only recently by the nvss . in light of its low centimetric spectral index , we assume that b2252 + 172 is also extragalactic , even though its @xmath11 optical counterpart is so faint that no spectrum has yet been obtained for it .
we selected b2252 + 172 to serve as an additional reference source and astrometric check source , since its north - south alignment with 3c 454.3 , b2250 + 194 , and i m peg helps to better model propagation - medium effects .
we carried out 35 sessions of 8.4 ghz ( @xmath12 cm ) vlbi observations of the guide star i m peg between 1997 january and 2005 july using a global vlbi array of between 12 and 16 telescopes ( see below ) .
the considerations governing the scheduling of the observing sessions are outlined in paper v @xcite . for each session
, we interleaved observations ( `` scans '' ) of i m peg with those of our references sources so that we could employ the phase - referencing technique ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) to determine an accurate astrometric position for i m peg .
the phase reference source for all sessions was 3c 454.3 .
b2250 + 194 was also observed during all sessions .
b2252 + 172 was included for the final 12 sessions , starting in 2002 november .
we give the sky coordinates for each reference source in paper iii .
our global vlbi array for each set of observations included the ten 25-m vlba telescopes and , for increased sensitivity to the relatively weak i m peg ( flux density 0.280 mjy ; see paper v ) , two or more of the following large - aperture telescopes : the phased very large array ( vla ) , the 100-m effelsberg telescope , the 43-m green bank telescope , the 46-m algonquin radio observatory telescope , and the three 70-m nasa - jpl deep space network ( dsn ) telescopes .
the total observing time after data editing and calibration for each session was 1115 hr . during each session , we switched between i m peg and each of the reference sources with a cycle time of between 5.5 and 7.3 minutes , in which i m peg
was observed for a minimum of 120 seconds , 3c 454.3 and b2250 + 194 for approximately 60 seconds each , and b2252 + 172 ( when included ) for approximately 75 seconds .
the data were recorded with either the vlba or mark iii system with sampling rates of 112 , 128 , or 256 @xmath13 , and were correlated with the nrao vlba processor in socorro , new mexico . at most epochs ,
both right - hand and left - hand circular polarizations were recorded .
the specific characteristics of each set of observations are given in table [ antab ] .
for our session on 2004 may 18 , we observed i m peg and the three reference sources at 5.0 ghz ( @xmath14 cm ) and 15.4 ghz ( @xmath15 cm ) in addition to using our standard frequency of 8.4 ghz .
we observed for @xmath22 hr at 8.4 ghz at the start of the observations , alternated @xmath240 minute observing cycles at each of the three frequencies for the next @xmath28 hr , observed at 8.4 ghz again for an extended period of @xmath24 hr , and returned to the @xmath240 minute cycle of three frequencies for the remaining @xmath22 hr .
we used the vla as a stand - alone interferometer ( in addition to using it as an element of our vlbi array ) for most of our sessions in order to monitor the variability of the total flux density and total circular polarization of i m peg over the course of the 1115 hr observing sessions . to set the flux - density scale for each set of vla observations , we used one or both of the flux - density calibrator sources 3c 286 and 3c 48 , with flux densities defined on the scale of @xcite .
we assume a standard error of @xmath25% in the vla - determined flux densities at each session .
the range of values for the total flux density of i m peg for each session , as well as the radio light curves in total intensity and total fractional circular polarization for a selected sample of sessions , are presented in paper vii @xcite .
the flux densities0.2 mjy arcsecond - scale jet . ] of reference sources 3c 454.3 , b2250 + 194 , and b2252 + 172 for each session are given in table [ fluxdensities ] .
l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l@ l c c c c 1997 & jan & 16 & x & x & x & x & x & x & x & x & x & x & x & x & x & & x & x & & 14.2 & 7.1 & iii - a & r + 1997 & jan & 18 & x & x & x & x & x & x & x & x & x & x & x & x & x & x & x & x & & 14.2 & 7.1 & iii - a & r + 1997 & nov & 29 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & & 11.7 & 5.5 & 128 - 4 - 1 & r , l + 1997 & dec & 21 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & & 14.8 & 5.5 & 128 - 4 - 1 & r , l + 1997 & dec & 27 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & & 12.5 & 5.5 & 128 - 4 - 1 & r , l + 1998 & mar & 01 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & x & 14.9 & 5.5 & 128 - 4 - 1 & r , l + 1998 & jul & 12 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 1998 & aug & 08 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 1998 & sep & 16 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 1999 & mar & 13 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 1999 & may & 15 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 1999 & sep & 18 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & & x & 15.2 & 5.5 & 128 - 4 - 1 & r , l + 1999 & dec & 09 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & & 14.5 & 5.5 & 128 - 4 - 1 & r , l + 2000 & may & 15 & x & x & x & x & & x & x & x & x & x & x & & x & & & x & x & 14.6 & 5.5 & 128 - 4 - 1 & r , l + 2000 & aug & 07 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 2000 & nov & 05 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 2000 & nov & 06 & x & x & x & x & x & x & x & x & x & x & x & & & & x & & & 13.3 & 5.5 & 128 - 4 - 1 & r , l + 2001 & mar & 31 & x & x & & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.2 & 5.5 & 128 - 4 - 1 & r , l + 2001 & jun & 29 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & x & 13.9 & 5.5 & 128 - 4 - 1 & r , l + 2001 & oct & 19 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 14.9 & 5.5 & 128 - 4 - 1 & r , l + 2001 & dec & 21 & x & x & x & x & x & x & x & x & x & x & & & x & & x & & x & 15.1 & 5.5 & 128 - 4 - 1 & r , l + 2002 & apr & 14 & x & x & x & x & x & x & x & & x & & x & & x & & x & x & x & 13.9 & 5.5 & 128 - 4 - 1 & r , l + 2002 & jul & 14 & x & x & x & x & x & x & x & x & x & x & & & x & & x & x & x & 14.8 & 5.5 & 128 - 4 - 1 & r , l + 2002 & nov & 20 & x & x & x & x & x & x & x & x & x & x & & & & & x & x & & 10.8 & 7.3 & 128 - 4 - 1 & r , l + 2003 & jan & 26 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & & 14.2 & 7.3 & 128 - 4 - 1 & r , l + 2003 & may & 18 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.2 & 7.3 & 128 - 4 - 1 & r , l + 2003 & sep & 08 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.1 & 7.3 & 128 - 4 - 1 & r , l + 2003 & dec & 05 & x & x & x & x & x & x & x & x & x & x & & & x & & x & x & x & 15.2 & 11.0 & 128 - 4 - 1 & r , l + 2004 & mar & 06 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 15.2 & 11.0 & 128 - 4 - 1 & r , l + 2004 & may & 18 & x & x & x & x & x & x & x & x & & x & x & & x & & x & x & x & 15.2 & 11.0 & 128 - 4 - 1 & r , l + 2004 & jun & 26 & x & x & x & x & x & x & x & x & x & x & x & & & & x & x & x & 15.2 & 11.0 & 128 - 4 - 1 & r , l + 2004 & dec & 11 & x & x & x & x & x & x & x & x & x & x & x & & x & & & x & x & 15.1 & 11.0 & 128 - 4 - 1 & r , l + 2005 & jan & 15 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 13.4 & 11.0 & 256 - 4 - 2 & r , l + 2005 & may & 28 & x & & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 13.3 & 11.0 & 256 - 4 - 2 & r , l + 2005 & jul & 16 & x & x & x & x & x & x & x & x & x & x & x & & x & & x & x & x & 13.3 & 11.0 & 256 - 4 - 2 & r , l + l@ l@ l@r@r@r@r@l@c@c 1997 & jan & 16 & & 12.16 & @xmath16 & 0.65 & & 438@xmath1622 & + 1997 & jan & 18 & & 12.06 & @xmath16 & 0.64 & & 430@xmath1621 & + 1997 & nov & 29 & & 11.31 & @xmath16 & 0.57 & ( d ) & 469@xmath1624 & + 1997 & dec & 21 & & 10.54 & @xmath16 & 0.54 & ( d ) & 451@xmath1623 & + 1997 & dec & 27 & & 10.01 & @xmath16 & 0.51 & ( d ) & 432@xmath1622 & + 1998 & mar & 01 & & 10.84 & @xmath16 & 0.56 & & 488@xmath1624 & + 1998 & jul & 12 & & 11.80 & @xmath16 & 0.60 & & 447@xmath1622 & + 1998 & aug & 08 & & 11.49 & @xmath16 & 0.61 & & 417@xmath1624 & + 1998 & sep & 16 & & 11.46 & @xmath16 & 0.61 & & 455@xmath1623 & + 1999 & mar & 13 & & 11.16 & @xmath16 & 0.58 & ( d ) & 432@xmath1623 & + 1999 & may & 15 & & 11.09 & @xmath16 & 0.57 & ( d ) & 455@xmath1623 & + 1999 & sep & 18 & & 12.05 & @xmath16 & 0.74 & & 418@xmath1621 & + 1999 & dec & 09 & & 10.74 & @xmath16 & 0.55 & & 468@xmath1623 & + 2000 & may & 15 & & 9.86 & @xmath16 & 0.52 & & 433@xmath1622 & + 2000 & aug & 07 & & 9.13 & @xmath16 & 0.47 & ( d ) & 420@xmath1621 & + 2000 & nov & 05 & & 9.14 & @xmath16 & 0.46 & & 436@xmath1622 & + 2000 & nov & 06 & & 9.12 & @xmath16 & 0.46 & & 441@xmath1622 & + 2001 & mar & 31 & & 11.17 & @xmath16 & 0.59 & & 443@xmath1624 & + 2001 & jun & 29 & & 10.92 & @xmath16 & 0.56 & & 395@xmath1620 & + 2001 & oct & 19 & & 9.70 & @xmath16 & 0.49 & ( d ) & 380@xmath1619 & + 2001 & dec & 21 & ( * ) & 9.25 & @xmath16 & 0.93 & & 406@xmath1641 & + 2002 & apr & 14 & & 9.30 & @xmath16 & 0.47 & & 359@xmath1618 & + 2002 & jul & 14 & ( * ) & 9.32 & @xmath16 & 0.93 & & 354@xmath1635 & + 2002 & nov & 20 & ( * ) & 9.47 & @xmath16 & 0.95 & & 396@xmath1640 & [email protected] + 2003 & jan & 26 & & 10.80 & @xmath16 & 0.54 & & 421@xmath1621 & [email protected] + 2003 & may & 18 & & 10.42 & @xmath16 & 0.54 & & 428@xmath1622 & [email protected] + 2003 & sep & 08 & & 9.66 & @xmath16 & 0.48 & & 480@xmath1624 & [email protected] + 2003 & dec & 05 & ( * ) & 9.31 & @xmath16 & 0.93 & & 488@xmath1649 & [email protected] + 2004 & mar & 06 & & 8.99 & @xmath16 & 0.45 & & 512@xmath1626 & [email protected] + 2004 & may & 18 & & 8.58 & @xmath16 & 0.43 & & 533@xmath1627 & [email protected] + 2004 & may & 18 & ( c ) & 10.45 & @xmath16 & 0.53 & & 381@xmath1619 & [email protected] + 2004 & may & 18 & ( u ) & 5.94 & @xmath16 & 0.42 & & 561@xmath1639 & [email protected] + 2004 & jun & 26 & & 8.59 & @xmath16 & 0.43 & ( d ) & 525@xmath1626 & [email protected] + 2004 & dec & 11 & & 9.36 & @xmath16 & 0.47 & & 507@xmath1626 & [email protected] + 2005 & jan & 15 & & 9.58 & @xmath16 & 0.48 & & 531@xmath1627 & [email protected] + 2005 & may & 28 & & 9.41 & @xmath16 & 0.47 & & 498@xmath1625 & [email protected] + 2005 & jul & 16 & & 9.56 & @xmath16 & 0.48 & & 519@xmath1625 & [email protected] +
we carried out the reduction and analysis of the vlbi data for the three reference sources with nrao s astronomical image processing system ( aips ) .
we performed the preliminary steps in the data reduction for each session , i.e. , initial amplitude calibration ( using system - temperature measurements and antenna gain curves ) , intermediate - frequency - channel bandpass and phase - offset correction , and data editing , in the usual manner ( e.g. , * ? ? ? * ) .
we describe the remaining data reduction and analysis steps in some detail below . for each of the stronger sources , 3c 454.3 and b2250 + 194
, we fringe fit and then iteratively self - calibrated ( e.g. , * ? ? ?
* ) and imaged the data . for fringe
fitting , we found a model consisting of a single point source to be adequate for determining the phase - delay rates and group delays for all telescopes ( relative to a reference telescope ) for both 3c 454.3 and b2250 + 194 . for phase self - calibration , we found a starting model consisting of a single point source to be adequate for b2250 + 194 at all epochs , but inadequate for 3c 454.3 at most epochs . for 3c 454.3 , we used , instead , a starting model consisting of two point sources , which more accurately reflects the internal structure of the bright core region of the source ( see [ 3c454 ] ) .
( we estimated the flux densities and relative positions of the two point sources by first fitting the model to the post - fringe - fit data . ) after self - calibrating in phase , we self - calibrated the data for each of 3c 454.3 and b2250 + 194 in amplitude .
we stopped the iterative process when we achieved convergence in both the phase and amplitude self - calibrations .
all images of 3c 454.3 and b2250 + 194 were generated using the clean deconvolution algorithm ( see * ? ? ? * ) and the robust data - weighting scheme @xcite .
for our weakest reference source , b2252 + 172 , our fringe fitting failed for each session to find a reliable rate and delay solution ( with signal - to - noise ratio , snr @xmath17 5 ) on 1050% of scans for at least one , and usually several , telescopes ( relative to a reference telescope ) .
instead we used the final image of 3c 454.3 to estimate the phase component of the complex antenna gains at the position of b2252 + 172 as a function of time ( see [ 2252phaseref ] ) , and the images of both 3c 454.3 and b2250 + 194 to estimate the amplitude component as a function of time .
( note that , in contrast to the phase gains , the amplitude gains are largely unaffected by inaccuracies in the propagation - medium model . ) for each telescope , the amplitude gains for sources 3c 454.3 and b2250 + 194 were identical to within 2% , so we smoothed them together over intervals of 30 minutes to 1 hr before using them to fine - tune the amplitude calibration for b2252 + 172 . at each session , the phase components of the complex antenna gains were derived for every scan of 3c 454.3 , with a cadence of 5.57.3 minutes , and interpolated to the effective time of the b2252 + 172 scans .
we found that phase connection between successive scans of 3c 454.3 was not possible in some instances for time ranges in which at least one telescope on a baseline was viewing the source at an elevation @xmath1820 .
any time range of data for which phase connection on a given baseline was problematic was flagged but not deleted .
we fourier - inverted the unflagged b2252 + 172 data and used the clean deconvolution algorithm to produce a phase - referenced image of the source . using the phase - referenced image as a starting model , we then phase self - calibrated the full set of b2252 + 172 data at each session ( including the data previously flagged ) over time intervals from scan lengths to @xmath220 minutes , depending on the flux density of the source . the quality of the image ( judged by the ratio of peak brightness to root - mean - square , `` rms , '' background and the ratio of peak to minimum brightness ) produced after a few iterations of phase self - calibration was , for each session , significantly higher ( @xmath950% improvement in each ratio ) than that of the strictly phase - referenced image .
we also self - calibrated the b2252 + 172 data at each session in amplitude , but found little improvement ( @xmath185% improvement in each ratio ) in the resulting image over the image produced in the phase - only self - calibration process .
the final images for b2252 + 172 were produced with data that were phase self - calibrated only . as for 3c 454.3 and b2250 + 194 , all images of b2252
+ 172 were generated using the clean deconvolution algorithm and the robust data - weighting scheme . to derive quantitative results for each of the three reference sources
, we also fit simple brightness - distribution models to either the image plane data using the aips program jmfit , or the plane data using the aips program omfit .
both programs utilize weighted least - squares algorithms . in the case of the plane model
fits , we first averaged the data within each individual scan .
we describe the model used in each fit , and the process of estimating the standard errors in the fit parameters , source by source , in
[ 3c454 ] , [ 2250 ] , and [ 2252 ] . the mean ratios of vlbi - determined to vla - determined flux densities are @xmath19 for 3c 454.3 ( see table [ 3cimstat ] ) and @xmath20 for b2250 + 194 ( see table [ 2250imstat ] ) .
however , the ratios are systematically lower ( and more variable ) for epochs before 2003 january , compared to those after , likely due to improved accuracy and implementation of the system - temperature measurements and/or gain curves for the large - aperture telescopes .
we estimate that our vlbi array `` resolves out '' only @xmath24% of the total ( vla - determined ) flux density for 3c 454.3 and @xmath23% of the total flux density for b2250 + 194 . in order to simplify our quantitative analysis , we adopt , unless otherwise stated , a conservative standard error of 10% for all vlbi - determined flux densities .
we present in figure [ 3cimages ] the 8.4 ghz images of 3c 454.3 generated from each of the 35 sessions of vlbi observations made between 1997 january and 2005 july .
each image is convolved with a gaussian ( clean ) restoring beam of size and shape determined , as usual , from a fit to the main lobe of the interferometer beam for that set of observations .
the median size of the minor axis of the restoring beam is 0.67 mas ( fwhm ) , and the median ratio of major to minor axis is 2.5 .
the image characteristics are summarized in table [ 3cimstat ] .
the images in the figure are aligned on the center of the easternmost component of the radio structure observed at each epoch ( see [ implanefits ] ) .
the coordinates in each image are relative to the phase center , in the image ) refers to the assumed source position used at the correlator . during the phase self - calibration process , the offset between the true and assumed positions , for a flux - density - averaged point in the radio structure ,
is removed from the data . ]
which is generally offset from the brightness peak by @xmath40.02 mas in each of @xmath21 and @xmath22 .
l@ l@ l@r c c c c c c c c 1997 & jan & 16 & & 6.14 & @xmath23 & 0.76 & 1.68 & 0.72 & @xmath24 & 10.70 & 0.88 + 1997 & jan & 18 & & 6.05 & @xmath25 & 0.66 & 1.83 & 0.70 & @xmath26 & 10.61 & 0.88 + 1997 & nov & 29 & & 6.20 & @xmath27 & 0.80 & 1.77 & 0.93 & @xmath28 & 9.72 & 0.86 + 1997 & dec & 21 & & 6.23 & @xmath29 & 0.58 & 1.68 & 0.83 & @xmath30 & 10.15 & 0.96 + 1997 & dec & 27 & & 5.91 & @xmath31 & 0.72 & 1.75 & 0.80 & @xmath32 & 9.54 & 0.95 + 1998 & mar & 01 & & 6.00 & @xmath33 & 0.56 & 1.59 & 0.82 & @xmath34 & 9.58 & 0.88 + 1998 & jul & 12 & & 6.10 & @xmath35 & 0.71 & 1.75 & 0.64 & @xmath36 & 10.48 & 0.89 + 1998 & aug & 08 & & 6.27 & @xmath37 & 0.74 & 1.78 & 0.65 & @xmath38 & 11.04 & 0.96 + 1998 & sep & 16 & & 6.18 & @xmath39 & 0.73 & 1.80 & 0.69 & @xmath40 & 10.69 & 0.93 + 1999 & mar & 13 & & 5.40 & @xmath41 & 0.59 & 1.73 & 0.68 & @xmath42 & 9.62 & 0.86 + 1999 & may & 15 & & 5.90 & @xmath25 & 0.55 & 1.66 & 0.62 & @xmath43 & 10.41 & 0.94 + 1999 & sep & 18 & & 8.16 & @xmath44 & 1.02 & 1.76 & 0.81 & @xmath45 & 12.32 & 1.02 + 1999 & dec & 09 & & 5.04 & @xmath46 & 0.52 & 1.73 & 0.67 & @xmath47 & 9.46 & 0.88 + 2000 & may & 15 & & 3.37 & @xmath48 & 0.77 & 1.54 & 0.59 & @xmath49 & 9.30 & 0.94 + 2000 & aug & 07 & & 4.60 & @xmath50 & 0.38 & 1.87 & 0.74 & @xmath51 & 9.31 & 1.02 + 2000 & nov & 05 & & 4.14 & @xmath52 & 0.71 & 1.63 & 0.64 & @xmath53 & 8.77 & 0.96 + 2000 & nov & 06 & & 4.98 & @xmath46 & 0.69 & 1.79 & 0.99 & @xmath54 & 8.84 & 0.97 + 2001 & mar & 31 & & 5.13 & @xmath55 & 0.90 & 1.56 & 0.63 & @xmath56 & 9.65 & 0.86 + 2001 & jun & 29 & & 5.32 & @xmath57 & 0.54 & 1.78 & 0.75 & @xmath58 & 10.08 & 0.92 + 2001 & oct & 19 & & 3.78 & @xmath35 & 0.88 & 1.83 & 0.64 & @xmath59 & 9.35 & 0.96 + 2001 & dec & 21 & & 4.78 & @xmath60 & 0.51 & 1.71 & 0.89 & @xmath61 & 9.25 & + 2002 & apr & 14 & & 3.97 & @xmath62 & 0.89 & 2.73 & 0.66 & @xmath63 & 8.67 & 0.93 + 2002 & jul & 14 & & 4.18 & @xmath31 & 0.41 & 1.71 & 0.63 & @xmath64 & 9.32 & + 2002 & nov & 20 & & 4.90 & @xmath40 & 0.70 & 1.76 & 0.75 & @xmath65 & 9.47 & + 2003 & jan & 26 & & 4.93 & @xmath66 & 0.70 & 1.81 & 0.70 & @xmath67 & 10.35 & 0.96 + 2003 & may & 18 & & 3.54 & @xmath68 & 0.53 & 1.62 & 0.59 & @xmath69 & 9.80 & 0.94 + 2003 & sep & 08 & & 3.02 & @xmath70 & 0.63 & 1.67 & 0.60 & @xmath71 & 9.34 & 0.97 + 2003 & dec & 05 & & 2.90 & @xmath72 & 0.98 & 1.69 & 0.59 & @xmath73 & 9.31 & + 2004 & mar & 06 & & 2.85 & @xmath74 & 0.97 & 1.62 & 0.60 & @xmath75 & 8.97 & 1.00 + 2004 & may & 18 & & 2.73 & @xmath76 & 0.83 & 1.76 & 0.68 & @xmath77 & 8.41 & 0.98 + 2004 & may & 18 & ( c ) & 3.93 & @xmath78 & 1.81 & 3.03 & 0.99 & @xmath79 & 9.73 & 0.93 + 2004 & may & 18 & ( u ) & 1.13 & @xmath80 & 2.24 & 0.98 & 0.33 & @xmath81 & 5.82 & 0.98 + 2004 & jun & 26 & & 2.85 & @xmath82 & 0.45 & 1.61 & 0.76 & @xmath83 & 8.03 & 0.94 + 2004 & dec & 11 & & 2.78 & @xmath84 & 0.34 & 1.58 & 0.61 & @xmath85 & 8.94 & 0.96 + 2005 & jan & 15 & & 2.76 & @xmath86 & 0.41 & 1.64 & 0.61 & @xmath87 & 9.00 & 0.94 + 2005 & may & 28 & & 2.68 & @xmath88 & 0.73 & 1.62 & 0.59 & @xmath89 & 9.17 & 0.98 + 2005 & jul & 16 & & 2.50 & @xmath90 & 0.23 & 1.66 & 0.62 & @xmath91 & 9.07 & 0.95 + each of our vlbi images of 3c 454.3 exhibits a radio structure generally consistent with that observed previously for this source at frequencies between 1.7 and 15 ghz , namely a bright , relatively compact core region and a bent milliarcsecond - scale jet .
nevertheless , a number of structural changes are apparent over the @xmath28.5 yr period of our observing program .
the most significant changes take place in the core and inner - jet regions .
the core region , which we roughly define as the easternmost @xmath21 mas of the source , is visibly extended mainly east - west , ranging in appearance from a single elongated emission region to a marginally resolved double - peaked emission region .
we model the core region at each of our 35 epochs with two compact components separated by 0.450.7 mas ( see [ coreregion ] ) .
the two - component , or indeed multi - component , structure of the core and inner - jet regions of 3c 454.3 was first observed by @xcite during their @xmath25 yr vlbi program at 10.7 ghz .
their images show that the two easternmost components have a relatively constant 0.50.6 mas separation , and that others move approximately westward away from them with superluminal apparent velocities .
our images likewise show that new components occasionally emerge to the west - southwest of the core region and move approximately westward toward the extended jet .
we discuss the motions in the inner - jet region in [ innerjetregion ] .
the brightness peak of the jet , located @xmath25.5 mas west of the easternmost core - region component , also appears to move with respect to the core region , but at a lower apparent velocity , and northward rather than westward . beyond this point , the axis of the jet bends sharply to the northwest .
we discuss the motion of the jet peak , and extended jet as a whole beyond the peak , in [ outerjetregion ] . to assign a systematic error to quantities estimated in our image - plane analysis in the next section
, we also produced for each set of observations an image ( not shown ) of 3c 454.3 using a common restoring beam .
we chose a beam with major axis 2.00 mas ( fwhm ) , minor axis 0.80 mas ( fwhm ) , and position angle ( p.a . ) 0 ( east of north ) to encompass the majority of the clean restoring beams in the nominal - resolution images ( see table [ 3cimstat ] ) , and to reflect the median ratio of major to minor axis of the clean restoring beams . for 28 epochs ,
the common restoring beam fully encompassed the clean restoring beam , so we simply convolved the nominal - resolution image with the larger common restoring beam . for five of the seven remaining epochs
, we first generated an image with a clean restoring beam slightly smaller than the common restoring beam , by increasing the relative weights on the longer baselines during imaging , and then convolved the new image with the common restoring beam . for two epochs , namely 2000
november 6 and 2002 april 14 , no choice of data weights produced a beam smaller than the common restoring beam , but we still forced the clean restoring beam to have the dimensions and orientation of the common restoring beam ( resulting in a mild super - resolution ) .
we performed a detailed analysis of the changing 8.4 ghz radio structure of 3c 454.3 via model fitting in the image plane .
we used for this analysis the nominal - resolution images shown in figure [ 3cimages ] .
fitting in the image plane facilitates dissection of the complex radio structure over its full brightness range . in the plane
( see [ uvplanefits ] ) , the bright core - region components dominate the low - surface - brightness jet components . in the image plane
, we can isolate structure and investigate in turn the core and jet regions . on the other hand ,
concerns about systematic errors in the image plane analysis such as from ( clean ) deconvolution errors and the differing resolution at each epoch have to be investigated and quantified as best as possible . in the remainder of this section ,
we discuss the image - plane - based model fitting of the core , inner - jet , and extended - jet regions of 3c 454.3 , using windows allocated approximately as shown in figure [ windowandcompid ] .
for ease of reference , we also indicate in figure [ windowandcompid ] the approximate positions of the model components discussed below .
several of our images of 3c 454.3 , in particular those after 2003 january , show that the easternmost @xmath21 mas of the 8.4 ghz radio structure is marginally resolved into two components oriented approximately east - west .
we fit to the core region at each epoch a model consisting of two point sources , and , for comparison , a model consisting of a single elliptical gaussian .
the two - component model fit the image data for the core region better ( yielding @xmath92 to @xmath93 lower values of the postfit rms residuals ) at all epochs , including those in which the core region appears only slightly elongated in figure [ 3cimages ] . the results of the two - component fit are given in table [ 3ccorecomp ] .
we refer to the easternmost component as c1 and the westernmost component as c2 .
the tabulated standard errors in the flux densities and positions of c1 and c2 at each epoch include contributions , added in quadrature , from the statistical standard error of the fit and from the systematic errors associated with deconvolution ( see [ deconv ] ) and the `` choice '' of image resolution .
the last error is taken at each epoch to be the difference between the parameter estimate from the fit to the nominal - resolution image ( figure [ 3cimages ] ) and the corresponding parameter estimate from the fit to the common - resolution image .
at every epoch , the systematic errors are @xmath21020 times larger than the statistical standard errors . unless otherwise noted , the quoted errors for all model parameters are standard errors , which include the systematic - error estimate .
l@ l@ l@ c@c@c@c@c@c@c@c@c@c@c@c 1997 & jan & 16 & 3.39 & 0.21 & 0.243 & 0.019 & @xmath94 & 0.013 & 3.41 & 0.18 & @xmath95 & 0.029 & 0.013 & 0.011 + 1997 & jan & 18 & 3.41 & 0.25 & 0.239 & 0.026 & @xmath96 & 0.021 & 3.32 & 0.19 & @xmath97 & 0.039 & 0.008 & 0.014 + 1997 & nov & 29 & 5.24 & 0.34 & 0.141 & 0.017 & @xmath98 & 0.010 & 2.15 & 0.16 & @xmath99 & 0.078 & 0.046 & 0.011 + 1997 & dec & 21 & 5.04 & 0.56 & 0.082 & 0.041 & 0.004 & 0.017 & 2.61 & 0.39 & @xmath100 & 0.110 & 0.037 & 0.012 + 1997 & dec & 27 & 4.79 & 0.44 & 0.126 & 0.029 & 0.000 & 0.016 & 2.45 & 0.29 & @xmath101 & 0.092 & 0.003 & 0.018 + 1998 & mar & 01 & 4.19 & 0.49 & 0.172 & 0.035 & 0.000 & 0.010 & 3.07 & 0.36 & @xmath102 & 0.083 & @xmath103 & 0.010 + 1998 & jul & 12 & 3.04 & 0.16 & 0.353 & 0.016 & 0.000 & 0.012 & 5.24 & 0.26 & @xmath104 & 0.017 & @xmath105 & 0.010 + 1998 & aug & 08 & 3.30 & 0.18 & 0.445 & 0.015 & 0.002 & 0.010 & 5.44 & 0.27 & @xmath106 & 0.019 & @xmath107 & 0.010 + 1998 & sep & 16 & 3.39 & 0.18 & 0.341 & 0.015 & 0.006 & 0.016 & 4.97 & 0.25 & @xmath108 & 0.017 & @xmath109 & 0.010 + 1999 & mar & 13 & 3.38 & 0.20 & 0.285 & 0.016 & 0.036 & 0.010 & 4.03 & 0.20 & @xmath110 & 0.028 & @xmath111 & 0.010 + 1999 & may & 15 & 3.72 & 0.22 & 0.229 & 0.016 & 0.042 & 0.010 & 4.17 & 0.21 & @xmath112 & 0.027 & @xmath113 & 0.010 + 1999 & sep & 18 & 4.91 & 0.43 & 0.194 & 0.019 & 0.057 & 0.012 & 4.97 & 0.34 & @xmath114 & 0.047 & @xmath115 & 0.016 + 1999 & dec & 09 & 3.03 & 0.16 & 0.321 & 0.015 & 0.048 & 0.011 & 4.06 & 0.21 & @xmath116 & 0.021 & @xmath98 & 0.010 + 2000 & may & 15 & 3.48 & 0.19 & 0.214 & 0.015 & 0.012 & 0.010 & 3.15 & 0.17 & @xmath117 & 0.028 & 0.021 & 0.013 + 2000 & aug & 07 & 4.53 & 0.28 & 0.002 & 0.019 & 0.014 & 0.013 & 2.36 & 0.12 & @xmath118 & 0.076 & 0.002 & 0.017 + 2000 & nov & 05 & 4.21 & 0.24 & 0.089 & 0.015 & 0.065 & 0.012 & 1.97 & 0.10 & @xmath119 & 0.066 & @xmath120 & 0.016 + 2000 & nov & 06 & 4.63 & 0.25 & 0.139 & 0.017 & 0.024 & 0.015 & 1.87 & 0.10 & @xmath121 & 0.049 & @xmath122 & 0.013 + 2001 & mar & 31 & 3.54 & 0.23 & 0.212 & 0.021 & 0.012 & 0.010 & 3.62 & 0.21 & @xmath97 & 0.034 & @xmath123 & 0.016 + 2001 & jun & 29 & 3.82 & 0.40 & 0.285 & 0.041 & 0.029 & 0.013 & 3.82 & 0.26 & @xmath124 & 0.082 & @xmath125 & 0.019 + 2001 & oct & 19 & 3.52 & 0.21 & 0.193 & 0.017 & 0.075 & 0.011 & 3.00 & 0.15 & @xmath126 & 0.042 & 0.010 & 0.020 + 2001 & dec & 21 & 4.34 & 0.28 & 0.176 & 0.023 & 0.016 & 0.011 & 2.48 & 0.14 & @xmath127 & 0.077 & @xmath128 & 0.010 + 2002 & apr & 14 & 3.58 & 0.18 & 0.165 & 0.015 & 0.055 & 0.010 & 2.59 & 0.13 & @xmath129 & 0.016 & @xmath130 & 0.011 + 2002 & jul & 14 & 2.88 & 0.30 & 0.315 & 0.033 & 0.049 & 0.010 & 3.42 & 0.19 & @xmath131 & 0.068 & @xmath132 & 0.023 + 2002 & nov & 20 & 2.62 & 0.23 & 0.439 & 0.039 & 0.086 & 0.010 & 4.23 & 0.23 & @xmath133 & 0.048 & @xmath120 & 0.013 + 2003 & jan & 26 & 2.72 & 0.26 & 0.448 & 0.033 & 0.138 & 0.010 & 4.59 & 0.23 & @xmath134 & 0.056 & 0.012 & 0.016 + 2003 & may & 18 & 2.81 & 0.15 & 0.600 & 0.015 & 0.086 & 0.011 & 3.84 & 0.19 & @xmath135 & 0.022 & 0.022 & 0.011 + 2003 & sep & 08 & 3.02 & 0.18 & 0.648 & 0.016 & 0.106 & 0.011 & 3.26 & 0.18 & @xmath136 & 0.032 & 0.019 & 0.012 + 2003 & dec & 05 & 3.20 & 0.16 & 0.175 & 0.015 & 0.009 & 0.011 & 2.84 & 0.14 & @xmath137 & 0.019 & @xmath138 & 0.010 + 2004 & mar & 06 & 3.13 & 0.16 & 0.189 & 0.015 & 0.069 & 0.011 & 2.51 & 0.13 & @xmath139 & 0.023 & @xmath140 & 0.014 + 2004 & may & 18 & 2.58 & 0.15 & 0.041 & 0.016 & @xmath141 & 0.010 & 2.51 & 0.13 & @xmath142 & 0.039 & @xmath143 & 0.018 + 2004 & jun & 26 & 2.55 & 0.25 & 0.376 & 0.027 & 0.064 & 0.011 & 2.41 & 0.12 & @xmath144 & 0.109 & @xmath145 & 0.021 + 2004 & dec & 11 & 2.62 & 0.14 & 0.113 & 0.018 & 0.029 & 0.010 & 2.90 & 0.15 & @xmath121 & 0.032 & @xmath146 & 0.013 + 2005 & jan & 15 & 2.64 & 0.14 & 0.033 & 0.018 & 0.080 & 0.010 & 2.92 & 0.15 & @xmath147 & 0.030 & 0.028 & 0.014 + 2005 & may & 28 & 2.90 & 0.15 & @xmath148 & 0.015 & 0.060 & 0.010 & 2.79 & 0.14 & @xmath149 & 0.021 & 0.020 & 0.011 + 2005 & jul & 16 & 2.74 & 0.14 & @xmath105 & 0.015 & 0.011 & 0.010 & 2.71 & 0.14 & @xmath150 & 0.023 & @xmath109 & 0.011 + in figure [ c1c2relpos ] , we plot at each epoch the separation of c2 from c1 .
we see that this separation increased by @xmath20.25 mas over the @xmath28.5 yr period of our observing program , largely in the @xmath21 coordinate .
we determined , via a weighted - least - squares fit of a straight line to the separation at all 35 epochs , a relative proper motion for c2 with respect to c1 of @xmath151 in @xmath21 and @xmath152 in @xmath22 .
( we use weighted least - squares for all proper - motion determinations . )
the quoted uncertainties are the statistical standard errors of the fit , scaled by a common factor so that the @xmath153 per degree of freedom is unity .
we performed the same weighted fit to the separations determined for the common - resolution images , and found that the relative proper motion was 0.5@xmath154 larger in @xmath21 and 0.2@xmath154 smaller in @xmath22 , where @xmath154 is the statistical standard error of the nominal ( differing - resolution ) case .
taking the difference as an estimate of the systematic error in the determination of the relative proper motion in the nominal case , and adding in quadrature the statistical standard error for the nominal case , we get a relative proper motion ( nominal value and standard error ) for c2 with respect to c1 of @xmath155 in @xmath21 and @xmath156 in @xmath22 .
( we compute the standard errors in the relative proper motions of components in the same way throughout
[ implanefits ] . )
the two - component model described above obviously does not account for all the structure seen in , or very near , the core region at each epoch . in particular , the two - component model fails to accurately describe low - level emission extending west - southwest from c2 with surface brightness 525% of the peak brightness . at many epochs
, this emission appears to be associated with the emergence of a new component into the inner - jet region ( see , e.g. , figure [ windowandcompid ] ) .
unfortunately , our resolution is not sufficient to adequately constrain a more elaborate model for the core region . in an attempt to simultaneously describe the bright components of the core region and what we call a transition component that moves between the core and inner - jet regions , we extended at each epoch the core - region fitting window @xmath21 mas west into the inner - jet region and
fit a model consisting of three point sources . for 13 epochs ,
the fit positions and flux densities of each of c1 and c2 in the three - component and two - component models differ by no more than two times the standard error in the two - component model ( see table [ 3ccorecomp ] ) . for these epochs , we ( somewhat arbitrarily ) consider the three - component model to give an acceptable description of the two core - region components and the transition component . the fit results for the transition component for these epochs and a brief description of the results for the remaining epochs are presented in [ innerjetregion ] .
in addition to the extension due to the transition component , there appears to also be core - region emission north of c2 , especially at our later epochs .
this emission is at a very low level and likely does not affect the results from the two - component model for either c1 or c2 .
how well does our simple two - component model represent the underlying structure of the core region of 3c 454.3 ?
vlbi images at 43 ghz and 86 ghz reveal a great complexity and variability in the easternmost @xmath21 mas of 3c 454.3 . at
the extreme east is a very compact component ( @xmath180.03 mas fwhm at 86 ghz ; * ? ? ?
* ) , considered by various authors to be the vlbi `` core . ''
new components are occasionally ejected from approximately the position of this core and then move west - northwest ( along @xmath157 ) at superluminal apparent velocities in the range 313@xmath1 @xcite . at a distance of @xmath20.65
mas from the core , the moving components are thought to encounter a bend or a recollimation shock in the jet ( see * ? ? ?
* ) and brighten to produce at 43 ghz a component which is approximately stationary relative to the core at that frequency .
the separation of the core and the `` stationary '' component at 43 ghz is similar to the separation of the two easternmost components observed at 10.7 ghz by @xcite and also to the mean separation of c1 and c2 in our 8.4 ghz images .
diffuse emission appears in the 43 ghz images to the north and northwest of the `` stationary '' component at various epochs , but the moving components emerge primarily to the west or southwest of the `` stationary '' component ( e.g. , * ? ? ?
the ejection from the core of new jet components , and the motions of these components toward and beyond the `` stationary '' component , may be responsible for the apparent `` oscillations ''
we observe in the 8.4 ghz separation of c2 relative to c1 ( see figure [ c1c2relpos ] ) .
the proper motion of c2 relative to c1 could be a consequence of the westward motion of a particularly bright jet component through the position of the `` stationary '' component in the latter part of our observing period ( see [ corediscuss ] ) .
to better quantify the spatial relationship of component c1 in our 8.4 ghz images to the core as imaged at 43 ghz , and likewise that of component c2 to the `` stationary '' component , we used the results from the studies of @xcite to simulate 8.4 ghz images of the core region of 3c 454.3 .
in particular , we used model data for the core , `` stationary '' component , and moving jet components at nine epochs , each within 30 days of one of our observations ( see table [ 3csim ] ) .
we retained the nominal position of each 43 ghz component at each epoch , but modified the flux densities of the components in approximate accordance with spectra presented in the literature ( e.g. , @xcite ; @xcite ; @xcite ) .
we generated the simulated 8.4 ghz image for each of the relevant epochs by ( 1 ) fourier transforming the flux - density - adjusted 43 ghz model onto a spatial frequency `` grid '' corresponding to our 8.4 ghz coverage , ( 2 ) imaging ( i.e. , fourier - inverting and deconvolving with the clean algorithm ) the `` gridded '' data using the same restoring beam as for the actual 8.4 ghz image , ( 3 ) fitting the two - point - source model ( described previously ) to the image , and finally ( 4 ) repeating steps 13 , tweaking the flux densities of the 43 ghz model components , until the flux densities and separation of the two point sources in the simulated 8.4 ghz image agreed with those in the actual 8.4 ghz image to approximately within the errors given in table [ 3ccorecomp ] .
we present for illustrative purposes in figure [ c1c2corestat ] the simulated image and corresponding actual image for one of the nine epoch pairs .
we give the results of the two - component model for the simulated image at each of the nine epochs in table [ 3csim ] .
we refer to the simulated easternmost component as @xmath158 and the simulated westernmost component as @xmath159 .
all positions in the table are given with respect to the 43 ghz core position .
the fit position of @xmath158 ( columns 1 and 2 in table [ 3csim ] ) therefore gives explicitly the offset ( in each coordinate ) of this component from the 43 ghz core .
the mean @xmath160 displacement is @xmath0 mas towards the west , where the uncertainty represents the rms variation in the position of @xmath158 .
we note that the largest displacements occur shortly after the ejection of new jet components , when the flux density of the core is relatively low compared to that of the jet component ( see figure [ c1c2corestat ] ) .
columns 5 and 6 in table [ 3csim ] give the offset in each coordinate of @xmath159 from the 43 ghz `` stationary '' component .
the mean @xmath161-component displacement is @xmath162 mas towards the west - southwest .
the smaller value of this displacement , compared to the @xmath160 displacement , is reasonable , since jet components move both into and out of the position defined by the `` stationary '' component , whereas jet components move only westward away from the core .
the larger peak - to - peak variations in the @xmath163-component displacement , compared to the @xmath160 displacement ( see table [ 3csim ] ) , reflect the complexity of the structure seen at 43 ghz at the western end of the core region .
l@ l@ l@ l@ l@ l@ c@c@c@c@c@c 1996 & dec & 23 & 1997 & jan & 16 & @xmath164 & 0.03 & @xmath165 & 0.14 & 0.05 & @xmath166 + 1998 & mar & 25 & 1998 & mar & 01 & @xmath167 & 0.00 & @xmath168 & 0.07 & @xmath169 & 0.00 + 1998 & aug & 01 & 1998 & aug & 08 & @xmath170 & 0.11 & @xmath171 & 0.16 & @xmath172 & @xmath173 + 1998 & oct & 05 & 1998 & sep & 16 & @xmath174 & 0.08 & @xmath175 & 0.05 & @xmath176 & 0.01 + 1999 & apr & 29 & 1999 & may & 15 & @xmath177 & 0.00 & @xmath178 & 0.10 & @xmath179 & @xmath179 + 1999 & oct & 06 & 1999 & sep & 18 & @xmath180 & 0.05 & @xmath181 & 0.06 & @xmath182 & @xmath179 + 1999 & dec & 05 & 1999 & dec & 09 & @xmath183 & 0.01 & @xmath184 & 0.04 & @xmath179 & 0.00 + 2000 & jul & 17 & 2000 & aug & 07 & @xmath185 & 0.00 & @xmath186 & @xmath179 & @xmath174 & @xmath187 + 2001 & apr & 14 & 2001 & mar & 31 & @xmath188 & 0.00 & @xmath189 & @xmath187 & @xmath176 & 0.01 + our images of 3c 454.3 show that new components occasionally emerge to the west or southwest of the position approximately defined by core - region component c2 , and move through the inner - jet region toward the extended jet . at any given epoch ,
the inner - jet region itself may contain zero , one , or multiple components in transit between the core region and extended jet .
unfortunately , given the general complexity of the radio structure in the inner - jet region , the relative weakness of the components , and the resolution of our vlbi arrays , we were able to confidently `` isolate '' inner - jet - region components for only a limited number of epochs ( see below ) . to estimate the proper motion ( relative to c1 ) of components transitioning between the core region and inner - jet region , we attempted , as described in
[ coreregion ] , to simultaneously describe the core - region components and a transition component by fitting a three point - source model to the easternmost @xmath22 mas of the 3c 454.3 radio structure . for 22 of the 35 epochs
, we found the fit positions and flux densities of core - region components c1@xmath190 and c2@xmath190 in the three - component model to differ significantly ( i.e. , by more than two times the respective standard errors given in table [ 3ccorecomp ] ) from those of c1 and c2 in the two - component model .
moreover , we found that the position of c1@xmath190 for these 22 epochs did not align with the features seen in the images ; specifically , c1@xmath190 was positioned too far east to produce the sharp cutoff seen at the eastern edge of the source .
we therefore consider the three - component model unacceptable for these epochs . of the 13 epochs with acceptably fit models , 11 fall within the @xmath22.5 yr interval from 2003 january to 2005 july .
for these 11 epochs , we repeated the three - component fit but with the position of c1@xmath190 fixed to the position of c1 found in the two - component model .
we plot in figure [ d1andd2c1relpos ] the separation from c1 of the @xmath191 jy transition component , d1 .
we determined for the 11 epochs a relative proper motion for d1 with respect to c1 of @xmath192 in @xmath21 and @xmath193 in @xmath22 .
we were also able to track the motion of a second component , d2 , for seven consecutive epochs from 2004 march to 2005 july , by fitting a model consisting of a single point source ( d2 ) to the @xmath22-mas - wide portion of the inner - jet region west of d1 .
we plot in figure [ d1andd2c1relpos ] the separation from c1 of the @xmath194 jy d2 .
we determined for the seven epochs a relative proper motion for d2 with respect to c1 of @xmath195 in @xmath21 and @xmath196 in @xmath22 .
while the proper motion in @xmath21 is similar for components d1 and d2 , the difference in the proper motion in @xmath22 from a negative value for d1 to a positive value for d2 clearly demonstrates a bend in the jet between these components .
a similar bend in the inner - jet region was noted by @xcite for a component at 5 and 8.4 ghz that moved from a position ( relative to their `` core '' ) near that of d1 ( relative to c1 ) to a position beyond that of d2 .
the bend demonstrated by component motion is also clearly observed in the low - surface - brightness emission in vlbi images between 5 and 15 ghz ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
the magnitudes of the proper motions of d1 and d2 are a factor 24 lower than that seen at 5 and 8.4 ghz for a component traversing a region 1.55.0 mas west of the core between 1983 and 1991 @xcite , but consistent with that seen at 15 ghz for a component traversing a region 2.53.5 mas west of the core between 2006 and 2009 ( see * ? ? ?
* ) . beyond the brightness peak @xmath25.5 mas west of component c1 ,
the extended jet of 3c 454.3 has no clearly discernible sub - components , at least none that can be traced from epoch to epoch .
we fit to the extended - jet region at each epoch a model consisting of a point source and an elliptical gaussian , representing , respectively , the compact emission component near the ( jet ) brightness peak and the @xmath26.5 mas ( mean major - axis length ) extended jet itself .
the results of the fit are given in table [ 3cextjetcomp ] .
we refer to the compact component as j1 and the extended component as jext .
the uncertainties in the major axis , minor axis , and p.a . of jext are statistical standard errors .
l@ l@ l@ c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c 1997 & jan & 16 & 0.267 & 0.027 & @xmath197 & 0.055 & 0.329 & 0.077 & 1.32 & 0.07 & @xmath198 & 0.045 & 1.650 & 0.046 & 7.28 & 0.10 & 1.77 & 0.11 & @xmath199 & 0.7 + 1997 & jan & 18 & 0.263 & 0.026 & @xmath200 & 0.058 & 0.341 & 0.079 & 1.34 & 0.07 & @xmath201 & 0.055 & 1.674 & 0.057 & 7.43 & 0.13 & 1.78 & 0.15 & @xmath202 & 0.9 + 1997 & nov & 29 & 0.357 & 0.036 & @xmath203 & 0.054 & 0.454 & 0.073 & 1.51 & 0.08 & @xmath204 & 0.043 & 1.894 & 0.044 & 7.53 & 0.09 & 1.87 & 0.12 & @xmath205 & 0.7 + 1997 & dec & 21 & 0.332 & 0.033 & @xmath206 & 0.065 & 0.449 & 0.073 & 1.58 & 0.09 & @xmath207 & 0.066 & 1.854 & 0.061 & 7.29 & 0.15 & 1.83 & 0.19 & @xmath208 & 1.1 + 1997 & dec & 27 & 0.308 & 0.031 & @xmath209 & 0.059 & 0.443 & 0.073 & 1.47 & 0.08 & @xmath210 & 0.049 & 1.840 & 0.045 & 7.02 & 0.09 & 1.85 & 0.11 & @xmath211 & 0.7 + 1998 & mar & 01 & 0.284 & 0.028 & @xmath212 & 0.062 & 0.439 & 0.073 & 1.48 & 0.08 & @xmath213 & 0.051 & 1.695 & 0.041 & 7.09 & 0.08 & 1.74 & 0.09 & @xmath214 & 0.6 + 1998 & jul & 12 & 0.276 & 0.028 & @xmath215 & 0.055 & 0.460 & 0.086 & 1.28 & 0.07 & @xmath216 & 0.045 & 1.911 & 0.048 & 6.78 & 0.11 & 1.82 & 0.15 & @xmath217 & 1.0 + 1998 & aug & 08 & 0.275 & 0.028 & @xmath218 & 0.054 & 0.463 & 0.090 & 1.41 & 0.07 & @xmath219 & 0.042 & 1.726 & 0.045 & 6.49 & 0.10 & 1.83 & 0.14 & @xmath220 & 1.0 + 1998 & sep & 16 & 0.298 & 0.030 & @xmath221 & 0.053 & 0.473 & 0.084 & 1.46 & 0.07 & @xmath222 & 0.040 & 1.841 & 0.042 & 6.99 & 0.08 & 1.89 & 0.11 & @xmath199 & 0.7 + 1999 & mar & 13 & 0.238 & 0.024 & @xmath223 & 0.053 & 0.506 & 0.077 & 1.33 & 0.07 & @xmath224 & 0.045 & 1.885 & 0.048 & 6.45 & 0.11 & 1.77 & 0.15 & @xmath225 & 1.0 + 1999 & may & 15 & 0.246 & 0.025 & @xmath226 & 0.054 & 0.491 & 0.070 & 1.40 & 0.07 & @xmath227 & 0.047 & 1.840 & 0.049 & 6.28 & 0.12 & 1.78 & 0.15 & @xmath228 & 1.1 + 1999 & sep & 18 & 0.297 & 0.030 & @xmath229 & 0.055 & 0.507 & 0.080 & 1.54 & 0.08 & @xmath230 & 0.049 & 1.903 & 0.051 & 6.26 & 0.12 & 1.88 & 0.17 & @xmath199 & 1.3 + 1999 & dec & 09 & 0.269 & 0.027 & @xmath231 & 0.056 & 0.548 & 0.081 & 1.34 & 0.07 & @xmath232 & 0.042 & 2.028 & 0.045 & 6.31 & 0.09 & 1.91 & 0.14 & @xmath233 & 1.0 + 2000 & may & 15 & 0.226 & 0.023 & @xmath234 & 0.069 & 0.717 & 0.058 & 1.45 & 0.07 & @xmath235 & 0.038 & 1.893 & 0.039 & 6.29 & 0.06 & 1.86 & 0.09 & @xmath236 & 0.6 + 2000 & aug & 07 & 0.240 & 0.024 & @xmath237 & 0.074 & 0.721 & 0.073 & 1.43 & 0.07 & @xmath238 & 0.041 & 2.076 & 0.041 & 6.48 & 0.07 & 1.94 & 0.11 & @xmath239 & 0.8 + 2000 & nov & 05 & 0.217 & 0.022 & @xmath240 & 0.068 & 0.710 & 0.068 & 1.41 & 0.07 & @xmath241 & 0.040 & 1.926 & 0.041 & 6.07 & 0.08 & 1.84 & 0.11 & @xmath242 & 0.8 + 2000 & nov & 06 & 0.310 & 0.031 & @xmath243 & 0.057 & 0.709 & 0.080 & 1.35 & 0.07 & @xmath244 & 0.045 & 2.372 & 0.048 & 6.13 & 0.10 & 2.08 & 0.16 & @xmath245 & 1.2 + 2001 & mar & 31 & 0.243 & 0.024 & @xmath246 & 0.061 & 0.770 & 0.061 & 1.27 & 0.07 & @xmath247 & 0.043 & 2.147 & 0.043 & 6.18 & 0.09 & 1.78 & 0.12 & @xmath220 & 0.9 + 2001 & jun & 29 & 0.207 & 0.021 & @xmath248 & 0.076 & 0.811 & 0.069 & 1.29 & 0.07 & @xmath249 & 0.053 & 2.084 & 0.040 & 5.78 & 0.07 & 1.87 & 0.11 & @xmath250 & 0.9 + 2001 & oct & 19 & 0.222 & 0.022 & @xmath251 & 0.068 & 0.864 & 0.081 & 1.47 & 0.08 & @xmath252 & 0.048 & 2.275 & 0.054 & 6.34 & 0.13 & 1.98 & 0.19 & @xmath250 & 1.4 + 2001 & dec & 21 & 0.244 & 0.024 & @xmath234 & 0.063 & 0.848 & 0.077 & 1.32 & 0.07 & @xmath253 & 0.043 & 2.340 & 0.040 & 6.08 & 0.07 & 1.96 & 0.11 & @xmath254 & 0.8 + 2002 & apr & 14 & 0.241 & 0.024 & @xmath255 & 0.071 & 0.937 & 0.092 & 1.29 & 0.07 & @xmath256 & 0.040 & 2.287 & 0.042 & 5.92 & 0.09 & 1.98 & 0.15 & @xmath257 & 1.2 + 2002 & jul & 14 & 0.194 & 0.019 & @xmath258 & 0.068 & 0.876 & 0.066 & 1.43 & 0.07 & @xmath252 & 0.048 & 2.237 & 0.039 & 6.02 & 0.07 & 1.85 & 0.10 & @xmath259 & 0.8 + 2002 & nov & 20 & 0.199 & 0.020 & @xmath260 & 0.072 & 0.929 & 0.075 & 1.28 & 0.07 & @xmath261 & 0.053 & 2.379 & 0.039 & 5.95 & 0.07 & 1.84 & 0.11 & @xmath242 & 0.8 + 2003 & jan & 26 & 0.183 & 0.018 & @xmath262 & 0.079 & 0.887 & 0.077 & 1.40 & 0.07 & @xmath263 & 0.050 & 2.216 & 0.044 & 5.79 & 0.09 & 1.82 & 0.14 & @xmath199 & 1.2 + 2003 & may & 18 & 0.169 & 0.017 & @xmath264 & 0.058 & 0.983 & 0.078 & 1.30 & 0.07 & @xmath265 & 0.039 & 2.250 & 0.041 & 5.54 & 0.07 & 1.80 & 0.12 & @xmath220 & 1.0 + 2003 & sep & 08 & 0.144 & 0.015 & @xmath266 & 0.099 & 0.976 & 0.080 & 1.30 & 0.07 & @xmath267 & 0.043 & 2.261 & 0.045 & 5.56 & 0.09 & 1.73 & 0.14 & @xmath268 & 1.1 + 2003 & dec & 05 & 0.175 & 0.018 & @xmath269 & 0.070 & 0.927 & 0.076 & 1.42 & 0.07 & @xmath270 & 0.039 & 2.267 & 0.040 & 5.81 & 0.07 & 1.78 & 0.11 & @xmath268 & 0.8 + 2004 & mar & 06 & 0.176 & 0.018 & @xmath271 & 0.063 & 0.951 & 0.081 & 1.49 & 0.08 & @xmath272 & 0.040 & 2.375 & 0.042 & 6.05 & 0.08 & 1.86 & 0.12 & @xmath273 & 0.9 + 2004 & may & 18 & 0.194 & 0.019 & @xmath271 & 0.088 & 0.944 & 0.092 & 1.44 & 0.07 & @xmath274 & 0.038 & 2.441 & 0.039 & 5.86 & 0.07 & 1.91 & 0.11 & @xmath239 & 0.9 + 2004 & jun & 26 & 0.166 & 0.017 & @xmath275 & 0.084 & 0.946 & 0.072 & 1.40 & 0.07 & @xmath276 & 0.042 & 2.346 & 0.034 & 5.80 & 0.04 & 1.82 & 0.05 & @xmath228 & 0.4 + 2004 & dec & 11 & 0.176 & 0.018 & @xmath266 & 0.074 & 0.931 & 0.077 & 1.39 & 0.07 & @xmath277 & 0.036 & 2.440 & 0.033 & 5.99 & 0.03 & 1.85 & 0.05 & @xmath199 & 0.4 + 2005 & jan & 15 & 0.181 & 0.018 & @xmath278 & 0.067 & 0.936 & 0.075 & 1.36 & 0.07 & @xmath279 & 0.037 & 2.469 & 0.035 & 5.90 & 0.04 & 1.86 & 0.07 & @xmath280 & 0.5 + 2005 & may & 28 & 0.189 & 0.019 & @xmath281 & 0.066 & 0.978 & 0.076 & 1.38 & 0.07 & @xmath282 & 0.037 & 2.476 & 0.037 & 5.80 & 0.06 & 1.83 & 0.08 & @xmath239 & 0.7 + 2005 & jul & 16 & 0.187 & 0.019 & @xmath283 & 0.065 & 0.955 & 0.083 & 1.37 & 0.07 & @xmath284 & 0.034 & 2.506 & 0.033 & 5.87 & 0.03 & 1.84 & 0.05 & @xmath280 & 0.4 + in figure [ j1andextc1relpos ] , we plot the separations at each epoch of j1 from c1 and the peak of jext from c1 . we see that j1 moved @xmath20.7 mas relative to c1 over the @xmath28.5 yr period of our observing program , largely in the @xmath22 coordinate .
we determined a mean relative proper motion for j1 with respect to c1 of @xmath285 in @xmath21 and @xmath286 in @xmath22 .
the proper motion of j1 ( see figure [ j1andextc1relpos ] ) , though approximately uniform in @xmath21 , changed in @xmath22 between the first 26 epochs ( @xmath26.5 yr ) and the remaining nine epochs ( @xmath22.0 yr ) .
fitting to each time period separately , we determined the relative proper motion in @xmath22 to decrease from @xmath287 for the first 26 epochs to @xmath288 for the remaining nine .
the result for the first 26 epochs suggests that the northward bend in the jet that starts at a position @xmath23 mas west of c1 , i.e. , at a position between transient components d1 and d2 , continues to a position @xmath25.6 mas west of c1 .
the rapid change at this point in the direction of the proper motion may signify a second sharp bend in the jet .
additionally , the drastic decrease at this point in the magnitude of the proper motion could indicate a transition away from the highly collimated flow seen in the core and inner - jet regions .
figure [ j1andextc1relpos ] shows that the peak of jext moved northwest @xmath21.1 mas relative to c1 over the @xmath28.5 yr period of our observing program .
we determined a relative proper motion for the peak of jext with respect to c1 of @xmath289 in @xmath21 and @xmath290 in @xmath22 .
the proper motion of the peak of jext ( along @xmath291 ) was precisely aligned with the component s major axis ( @xmath292 ) .
this result is consistent with the 1.7 ghz space - vlbi observations of @xcite , which showed for an epoch in 1999 june that the extended jet region was composed of 3 or 4 components .
moreover , the decrease in the major - axis length ( fwhm ) of the `` moving '' jext from @xmath27.3 mas to @xmath25.9 mas ( see table [ 3cextjetcomp ] ) suggests that no `` new '' component transited from the inner - jet region to the extended - jet region during our @xmath28.5 yr observing period .
component j1 , as discussed above , may have been close to this transition near the end of our program .
we note that the major axis of jext is approximately aligned with the axis of the arcsecond - scale jet of 3c 454.3 ( see * ? ? ?
* ) , suggesting that the trajectory directions of components beyond j1 are roughly constant over @xmath25.2@xmath293 ( @xmath240 kpc ) .
deconvolution errors are known to limit the fidelity of images produced using the clean deconvolution algorithm ( see , e.g. , * ? ? ?
* ; * ? ? ?
we performed two tests to investigate the reliability of the radio structure seen in the clean deconvolved images of 3c 454.3 .
the first test was a simulation study in which we produced data on a `` grid '' corresponding to our typical ) .
we repeated the simulation study described here for non - typical instances of our coverage ( 1997 january 18 , 2002 april 14 , and 2002 november 20 ) and found flux - density and position variations for the 3c 454.3 model components consistent with the uncertainties derived here for the typical coverage . ] coverage and consisting of the fourier transforms of simple geometrical models similar to 3c 454.3 as described in
[ coreregion] [ outerjetregion ] .
each variation of the model used different positions and flux densities for the two bright core - region components ( c1 and c2 ) , and in a few cases also used different positions and flux densities for the weaker components ( d1 , d2 , j1 , and jext ) .
we fourier - inverted the simulated data for each variation and used the clean deconvolution algorithm to produce images .
we subsequently compared to the input models the resulting simulated images and new models fit to the simulated images .
( the new model , or `` output '' model , contains the same number and type of components as the input model . ) for each variation of the input model , we found the model structure to be reasonably well reproduced in the simulated images ( taking into account the size and orientation of the clean restoring beam ) , but noted that additional `` ripples '' were present in the core region with peak - to - peak amplitudes of 13% of the image peak and in the inner - jet and extended - jet regions with amplitudes of 0.21% of the image peak .
the fourier counterparts of these ripples occur at spatial frequencies just outside the bulk of the coverage .
the maximum fraction ( considering several variations of the input model ) of the rms flux density in ripples near the positions of the input - model components to the flux density in the components themselves was @xmath25% for the core - region components ( c1 and c2 ) and extended - jet component ( jext ) and @xmath210% for the compact - jet components ( d1 , d2 , and j1 ) .
when we compared the flux densities of the output - model components to those of the input - model components , we found corresponding differences of up to @xmath25% for c1 , c2 , and jext and @xmath210% for d1 , d2 , and j1 .
we therefore take these values ( @xmath25% for each of c1 , c2 , and jext and @xmath210% for each of d1 , d2 , and j1 ) as a conservative estimate of the fractional flux - density uncertainty due to deconvolution errors . to estimate the amount by which the ripples affect the positions of the model components , we compared the positions of the output and input model components for each variation .
we found the maximum difference between the output and input component positions to be @xmath20.020 mas for c1 and c2 ( @xmath20.015 mas in @xmath21 and @xmath20.010 mas in @xmath22 ) , @xmath20.100 mas for d1 ( @xmath20.080 mas in @xmath21 and @xmath20.050 mas in @xmath22 ) , @xmath20.070 mas for d2 and j1 ( @xmath20.050 mas in each coordinate ) , and @xmath20.040 mas for jext ( @xmath20.030 mas in each coordinate ) .
we take the maximum difference between the positions of the output and input model components quoted above as conservative estimates of the various position uncertainties due to deconvolution errors . as a second reliability test we used another deconvolution algorithm , namely the maximum entropy method ( mem
; e.g. , * ? ? ?
* ) , to produce images of 3c 454.3 at select epochs .
we then performed the same fits in the image plane as described in [ coreregion] [ outerjetregion ] .
we found that the differences in the position determinations of the relevant components in the mem deconvolved images compared to the clean deconvolved images were well within the limits given above .
we also performed an analysis of the changing 8.4 ghz radio structure of 3c 454.3 via model fitting in the plane .
fitting in the plane is not affected by deconvolution errors ( as described in [ deconv ] ) and is not as sensitive to differences from epoch to epoch in the precise coverage . on the other hand ,
we can not , given our coverage , simultaneously constrain the high - flux - density components in the core region and the low - flux - density or low - surface - brightness components in the inner - jet and extended - jet regions .
we fit to the data at each epoch a model consisting of three point sources , which describe , approximately , the core - region components c1 and c2 as well as the compact jet component j1 ( see [ implanefits ] ) . for comparison with the image - plane models , we plot in figure [ uvrelpos ] the separations in the -plane model of @xmath294 and @xmath295 ( top panels ) and @xmath296 and @xmath295 ( bottom panels ) .
the proper motion of @xmath294 with respect to @xmath295 is consistent with that found in the image - plane analysis .
the proper motion of @xmath296 with respect to @xmath295 differs at the 1.5@xmath154 level from that found in the image - plane analysis .
this difference may result from structural changes in the unmodeled extended jet over the @xmath28.5 yr period of our observing program .
given the limitations of the -plane - based analysis for 3c 454.3 , we believe our image - plane models ( see [ implanefits ] ) provide the better overall description of the evolving structure of this source .
we present in figure [ 3cimagescxu ] the 5.0 , 8.4 , and 15.4 ghz images of 3c 454.3 generated from our nearly simultaneous vlbi observations on 2004 may 18 ( see [ obs ] ) .
the 8.4 ghz image is the same as that presented in figure [ 3cimages ] except for the choice of contour levels .
the images are aligned on the center of the easternmost component of the radio structure ( see below ) .
the image characteristics are given in table [ 3cimstat ] .
the 5.0 and 15.4 ghz images are slightly noisier than the 8.4 ghz image , mainly because the total time on source was @xmath24 hr at each of 5.0 and 15.4 ghz compared to @xmath27 hr at 8.4 ghz .
nevertheless , the images at all three frequencies show a core region extended nearly east - west and a jet bending toward the north .
to quantitatively compare the positions and flux densities of the core- and inner - jet - region components , we fit to the combined core / inner - jet region in each of the 5.0 and 15.4 ghz images the three - component model ( c1 , c2 , and d1 ) described in [ innerjetregion ] .
similarly , to compare the positions and flux densities of the extended - jet - region component(s ) , we fit to the extended - jet region in the 5.0 ghz image the two - component model ( j1 and jext ) described in [ outerjetregion ] .
however , at 15.4 ghz we fit only a single point source ( j1 ) to the region within @xmath22 mas of the jet peak , because the surface brightness of the extended jet in this image is too low to allow us to adequately constrain an elliptical gaussian component .
the results of the fits are given in table [ 3ccompcxu ] and illustrated in figure [ 3cimagescxu ] .
the uncertainties in the flux densities and positions include contributions , added in quadrature , of the statistical standard error of the fit and an estimate ( using an analysis similar to that described in [ deconv ] ) of the systematic error associated with deconvolution errors in the corresponding image .
c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c 5.0 & 2.94 & 0.20 & 2.73 & 0.14 & @xmath149 & 0.022 & @xmath110 & 0.021 & 0.946 & 0.096 & @xmath297 & 0.081 & @xmath298 & 0.055 & 0.328 & 0.033 & @xmath299 & 0.052 & 0.878 & 0.061 + 8.4 & 2.42 & 0.15 & 2.45 & 0.13 & @xmath300 & 0.021 & @xmath301 & 0.016 & 0.621 & 0.078 & @xmath302 & 0.088 & @xmath303 & 0.053 & 0.194 & 0.019 & @xmath271 & 0.051 & 0.944 & 0.054 + 15.4 & 1.41 & 0.08 & 1.22 & 0.07 & @xmath304 & 0.018 & @xmath305 & 0.017 & 0.271 & 0.028 & @xmath306 & 0.052 & @xmath307 & 0.050 & 0.096 & 0.010 & @xmath308 & 0.036 & 0.950 & 0.045 + we choose in figure [ 3cimagescxu ] to align the images at each frequency on the ( fit - estimated ) center of component c1 .
this is a reasonable choice , since our comparison of the structure of the core region at 8.4 and 43 ghz in
[ coreregion ] shows that the 8.4 ghz c1 is located within @xmath20.2 mas of the 43-ghz - identified core .
but how well do the images at 5.0 , 8.4 , and 15.4 ghz actually line - up ? to answer this question we look at the separation of j1 and c1 at each frequency .
the jet peak is clearly identified at each frequency , and thus serves as a useful marker in the radio structure .
the largest difference between the three images in the j1c1 separation ( see the last four columns of table [ 3ccompcxu ] ) is @xmath309 mas in @xmath21 and @xmath310 mas in @xmath22 .
we therefore believe that the images at 5.0 , 8.4 , and 15.4 ghz are aligned ( relative to these internal reference points ) to within 0.10 mas in @xmath21 and 0.15 mas in @xmath22 ; i.e. , to better than @xmath215% of the resolution of the 5.0 ghz image . the separations of c2 and d1 from c1 are more variable with frequency than the separation of j1 from c1 .
this difference is not surprising considering the complexity of the underlying structure in the core and inner - jet regions .
the @xmath21.8 times higher angular resolution of the 15.4 ghz image , compared to that of the 8.4 ghz image , reveals that the core and inner - jet regions for 2004 may 18 contain more than three components .
specifically , there appears to be a component located between the positions of c1 and c2 , and another component located west - southwest of the position of c2 .
can we understand the differences in the fit - estimated c2c1 and d1c1 separations at each frequency in terms of the frequency - dependent flux densities of these unmodeled components ? generally speaking ,
the spectral index , @xmath21 ( @xmath311 ) , of a component is steeper ( i.e. , more negative ) the farther west it is from the core ( e.g. , * ? ? ?
* ; * ? ? ?
thus , relative to c1 and c2 , a component lying between them will likely have a larger fractional flux density at 15.4 ghz than at 8.4 or 5.0 ghz , while a component west of c2 will have a larger fractional flux density at 5.0 ghz than at either 8.4 or 15.4 ghz .
these flux - density differences would create at 15.4 ghz a `` plateau '' between c1 and c2 ( as observed ) and at 5.0 ghz a west - broadened c2 . for a core - region modeled only by c1 and c2
, we might then expect a c2c1 separation which is smaller at 15.4 ghz than at 5.0 ghz .
this is indeed what we find in our model fits ( see table [ 3ccompcxu ] ) . in the inner - jet region ,
the situation is reversed . here
, the component west - southwest of c2 might be expected to create at 5.0 ghz a plateau between c2 and d1 , and yield a d1c1 separation which is smaller at this frequency than at 15.4 ghz .
these expectations , too , are confirmed by our model fits .
we give in table [ 3cspeccxu ] the flux - density ratio ( @xmath312 ) and resulting point - to - point spectral index ( @xmath313 ) for each model component for each pair of frequencies .
we also give for comparison the total ( clean ) flux - density ratio ( using the values in table [ 3cimstat ] ) and resulting point - to - point spectral index for each pair of frequencies .
note that to compute the standard errors in the flux densities , we added in quadrature to the other contributions a 5% uncertainty representative of the standard error in the vlbi flux - density scale . we adopt a 5% uncertainty for this epoch rather than the nominal 10% ( see [ vlbiamp ] ) , because the vlbi- and vla - determined flux densities at each frequency differ by only 27% ( see table [ 3cimstat ] ) . the slightly lower ratio of vlbi - determined to vla - determined flux density at 5.0 ghz compared to those at 8.4 and 15.4 ghz is likely due to the insensitivity of our vlbi observations to low - surface - brightness ( and steep - spectral - index ) emission in the extended jet at distances @xmath915 mas northwest of the core ( see * ? ? ?
c@ c@ c@ c@ c@ c@ c@ c@ c@ c@ c 8.4/5.0 & @xmath314 & @xmath315 & @xmath316 & @xmath317 & @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath322 & @xmath323 + 15.4/8.4 & @xmath324 & @xmath325 & @xmath326 & @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & @xmath332 & @xmath333 + 15.4/5.0 & @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath343 + the spectral indices for c1 , c2 , and d1 in table [ 3cspeccxu ] should be viewed with some caution . as noted above
, the structure in the core and inner - jet regions is more complicated than that described by our three - component model .
moreover , the flux density in each unmodeled component is probably allocated differently to c1 , c2 , and d1 at each frequency due to differences in resolution and opacity .
nonetheless , we can still make some general comments and compare our results with previous multi - frequency studies : ( 1 ) estimates of the spectral index of the 43- and 86-ghz - identified core range between @xmath344 and @xmath345 ( @xcite ; @xcite ; @xcite ; @xcite ) .
our estimate of the spectral index of c1 between 5.0 and 8.4 ghz ( @xmath346 ) is at the low end of this range , and consistent with an isolated core several months removed from its most recent ejection @xcite .
our spectral - index estimate for c1 between 8.4 and 15.4 ghz ( @xmath347 ) suggests that the component between c1 and c2 is contributing fractionally more flux density to c1 at 8.4 ghz than at 15.4 ghz ( where it is partially resolved ) .
( 2 ) the `` stationary '' component @xmath20.65 mas from the core is modeled by @xcite for their 1999 may epoch as a synchrotron self - absorbed source , with a turnover frequency between 5.0 and 8.4 ghz .
the flat spectral index we estimate for c2 between 5.0 and 8.4 ghz ( @xmath348 ) is consistent with this interpretation .
our spectral - index estimate for c2 between 8.4 and 15.4 ghz ( @xmath349 ) suggests that the component west - southwest of c2 is contributing fractionally more flux density to c2 at 8.4 ghz than at 15.4 ghz .
( 3 ) @xcite find for a component located @xmath21.5 mas west - southwest of the core a spectral index of @xmath350 .
our spectral - index estimates for d1 ( @xmath21 @xmath351 @xmath352 to @xmath353 ) are roughly consistent with this value , but again suggest that components between c2 and d1 are contributing fractionally more flux density to d1 at 5.0 and 8.4 ghz than at 15.4 ghz .
in contrast to the core and inner - jet regions , the extended - jet region appears to be reasonably well represented by our two - component model at 5.0 and 8.4 ghz , and likewise the ( immediate ) jet - peak region by our one - component model at 15.4 ghz .
our estimates of the spectral index of j1 ( @xmath21 @xmath351 @xmath354 to @xmath355 ) are self - consistent and also consistent with the spectral index ( @xmath350 ) estimated by @xcite for a component approximately @xmath25.5 mas west of the core .
we note that the lower - level structure extending west of j1 in the 15.4 ghz image is likely real , and not a noise spike or sidelobe of one of the bright core - region components .
its appearance is robust against changes ( e.g. , to the positioning of clean windows ) in the imaging process .
the east - west extension at j1 at all three frequencies is , perhaps , a consequence of a near superposition of jet components undergoing the change in trajectory described in [ outerjetregion ] .
can we determine the ejection epoch for the component seen between c1 and c2 in our 15.4 ghz image for 2004 may 18 , and perhaps also the time of transit of the component from c1 to c2 ?
our 8.4 ghz image for 2003 september 8 shows a `` sudden '' brightening of c1 relative to c2 not seen 2003 may 18 or earlier . on 2003 december 5
, the 8.4 ghz brightness peak is very nearly aligned with the center of c1 . over the next several epochs , the brightness peak moves westward , and by 2005 january 15 is very nearly aligned with the center of c2 . if we assume that the brightening in 2003 september 8 temporally coincides ( within @xmath212 months ) with the ejection of a new component from the `` core , '' then we may interpret the subsequent shifts in the position of the brightness peak as roughly tracking the motion of the new component between the positions defined by c1 and c2 .
the total transit time , using 2003 september 8 as the epoch of ejection and 2005 january 15 as the epoch of arrival , is @xmath21.4 yr . the 2004 may 18 epoch represents the approximate mid - point of this interval .
we present in figure [ 2250imagesplot ] the 8.4 ghz images of b2250 + 194 generated from each of the 35 sessions of vlbi observations made between 1997 january and 2005 july .
the image characteristics are summarized in table [ 2250imstat ] .
the images reveal a relatively compact source . near the brightness peak
, the source shows no deviation from the restoring beam , down to the @xmath210% level . below this level , i.e. , at the brightness level of the outermost five or six contours , the source shows a distinct `` kidney - bean '' shape : the northern portion of the source bends @xmath230 to the west , while the southern portion of the source bends @xmath210 to the west .
other than the @xmath225% increase in the peak brightness over the @xmath28.5 yr period of our observing program , there is no significant indication of radio structure evolution in the images . the low - level kidney - bean structure may trace ( 1 ) continuous emission along a single , bent jet , ( 2 ) emission from well - separated regions in a single , bent jet viewed nearly straight on , or ( 3 ) emission from a pair of bent ( or at least not quite counter - aligned ) jets .
if scenario 1 were correct , then the core of the quasar could be located at either the north or south end of the source , and the centrally - located brightness peak would be a bright jet component . if , on the other hand , scenario 2 or 3 were correct , then the brightness peak could be nearly aligned with the core .
we elaborate more on these options in [ 2250uvplanefits ] and [ 2250canduimages ] .
l@ l@ l@r c c c c c c c c 1997 & jan & 16 & & 318 & @xmath356 & 0.18 & 1.60 & 0.73 & @xmath357 & 407 & 0.93 + 1997 & jan & 18 & & 314 & @xmath358 & 0.14 & 1.71 & 0.70 & @xmath359 & 396 & 0.92 + 1997 & nov & 29 & & 356 & @xmath178 & 0.18 & 1.65 & 0.91 & @xmath360 & 415 & 0.89 + 1997 & dec & 21 & & 383 & @xmath361 & 0.14 & 1.60 & 0.82 & @xmath362 & 457 & 1.01 + 1997 & dec & 27 & & 367 & @xmath175 & 0.16 & 1.70 & 0.70 & @xmath363 & 429 & 0.99 + 1998 & mar & 01 & & 401 & @xmath364 & 0.15 & 1.51 & 0.79 & @xmath365 & 468 & 0.96 + 1998 & jul & 12 & & 360 & @xmath366 & 0.15 & 1.51 & 0.61 & @xmath367 & 435 & 0.97 + 1998 & aug & 08 & & 373 & @xmath368 & 0.12 & 1.68 & 0.64 & @xmath61 & 438 & 1.05 + 1998 & sep & 16 & & 360 & @xmath165 & 0.14 & 1.81 & 0.61 & @xmath369 & 440 & 0.97 + 1999 & mar & 13 & & 313 & @xmath370 & 0.12 & 1.61 & 0.66 & @xmath75 & 401 & 0.93 + 1999 & may & 15 & & 347 & @xmath371 & 0.09 & 1.59 & 0.62 & @xmath372 & 428 & 0.94 + 1999 & sep & 18 & & 382 & @xmath373 & 0.14 & 1.67 & 0.80 & @xmath374 & 445 & 1.06 + 1999 & dec & 09 & & 360 & @xmath375 & 0.12 & 1.64 & 0.66 & @xmath376 & 433 & 0.93 + 2000 & may & 15 & & 357 & @xmath368 & 0.15 & 1.43 & 0.58 & @xmath377 & 441 & 1.02 + 2000 & aug & 07 & & 346 & @xmath171 & 0.12 & 1.73 & 0.69 & @xmath378 & 412 & 0.98 + 2000 & nov & 05 & & 354 & @xmath379 & 0.10 & 1.81 & 0.59 & @xmath380 & 429 & 0.98 + 2000 & nov & 06 & & 341 & @xmath189 & 0.17 & 2.12 & 1.00 & 15.75 & 406 & 0.92 + 2001 & mar & 31 & & 331 & @xmath381 & 0.14 & 1.56 & 0.61 & @xmath382 & 403 & 0.91 + 2001 & jun & 29 & & 320 & @xmath383 & 0.08 & 1.72 & 0.74 & @xmath23 & 401 & 1.02 + 2001 & oct & 19 & & 269 & @xmath364 & 0.19 & 1.79 & 0.55 & @xmath384 & 347 & 0.91 + 2001 & dec & 21 & & 321 & @xmath385 & 0.11 & 1.69 & 0.84 & @xmath386 & 406 & + 2002 & apr & 14 & & 276 & @xmath387 & 0.13 & 2.56 & 0.57 & @xmath388 & 339 & 0.94 + 2002 & jul & 14 & & 264 & @xmath389 & 0.11 & 1.64 & 0.62 & @xmath390 & 354 & + 2002 & nov & 20 & & 314 & @xmath391 & 0.16 & 1.74 & 0.81 & @xmath392 & 396 & + 2003 & jan & 26 & & 312 & @xmath393 & 0.17 & 1.75 & 0.70 & @xmath394 & 401 & 0.95 + 2003 & may & 18 & & 330 & @xmath395 & 0.10 & 1.58 & 0.59 & @xmath396 & 409 & 0.96 + 2003 & sep & 08 & & 390 & @xmath397 & 0.12 & 1.60 & 0.60 & @xmath62 & 468 & 0.98 + 2003 & dec & 05 & & 408 & @xmath358 & 0.15 & 1.64 & 0.60 & @xmath398 & 488 & + 2004 & mar & 06 & & 437 & @xmath399 & 0.13 & 1.63 & 0.54 & @xmath400 & 516 & 1.01 + 2004 & may & 18 & & 456 & @xmath361 & 0.16 & 1.71 & 0.67 & @xmath398 & 529 & 0.99 + 2004 & may & 18 & ( c ) & 290 & @xmath366 & 0.18 & 3.04 & 0.96 & @xmath401 & 367 & 0.96 + 2004 & may & 18 & ( u ) & 502 & @xmath402 & 0.61 & 0.98 & 0.35 & @xmath403 & 567 & 1.01 + 2004 & jun & 26 & & 423 & @xmath399 & 0.10 & 1.88 & 0.67 & @xmath404 & 511 & 0.97 + 2004 & dec & 11 & & 418 & @xmath405 & 0.10 & 1.55 & 0.53 & @xmath406 & 494 & 0.97 + 2005 & jan & 15 & & 431 & @xmath168 & 0.08 & 1.63 & 0.54 & @xmath407 & 507 & 0.96 + 2005 & may & 28 & & 421 & @xmath408 & 0.09 & 1.62 & 0.53 & @xmath409 & 495 & 0.99 + 2005 & jul & 16 & & 425 & @xmath410 & 0.09 & 1.67 & 0.55 & @xmath411 & 494 & 0.95 + to more quantitatively investigate changes in the radio structure of b2250 + 194 , we fit to the data , not including baselines to the 70-m dsn tidbinbilla ( ti ) telescope ( see below ) , a model consisting of a single elliptical gaussian .
the results for the fits are given in table [ 2250model ] and illustrated in figure [ 2250modelplot ] .
the uncertainty in each parameter at each epoch is a standard error ; it includes contributions , added in quadrature , from both the statistical standard error of the fit and the systematic error associated with the particular coverage . to estimate the systematic error we performed the fit at each epoch an additional five or six times , using for each additional fit only a subset ( 812 telescopes ) of the full vlbi array ( minus
we take the systematic error for a given parameter at a given epoch to be the rms variation of the subset values about the value derived from all of the available data at that epoch .
the uncertainty in the tabulated flux density at each epoch includes also the 10% standard error in the vlbi flux - density scale .
l@ l@ l@ c@c@c@c@c@c@c@c 1997 & jan & 16 & 357 & 36 & 0.75 & 0.04 & 0.31 & 0.02 & @xmath412 & 3 + 1997 & jan & 18 & 346 & 35 & 0.75 & 0.04 & 0.29 & 0.02 & @xmath413 & 3 + 1997 & nov & 29 & 395 & 40 & 0.62 & 0.05 & 0.21 & 0.02 & @xmath412 & 4 + 1997 & dec & 21 & 436 & 44 & 0.69 & 0.04 & 0.22 & 0.02 & @xmath412 & 3 + 1997 & dec & 27 & 413 & 41 & 0.67 & 0.04 & 0.20 & 0.02 & @xmath412 & 3 + 1998 & mar & 01 & 455 & 46 & 0.68 & 0.04 & 0.20 & 0.02 & @xmath414 & 2 + 1998 & jul & 12 & 413 & 41 & 0.63 & 0.04 & 0.21 & 0.02 & @xmath415 & 3 + 1998 & aug & 08 & 419 & 42 & 0.59 & 0.03 & 0.21 & 0.01 & @xmath416 & 3 + 1998 & sep & 16 & 423 & 42 & 0.71 & 0.04 & 0.26 & 0.02 & @xmath414 & 2 + 1999 & mar & 13 & 386 & 39 & 0.78 & 0.03 & 0.29 & 0.01 & @xmath412 & 2 + 1999 & may & 15 & 413 & 41 & 0.72 & 0.03 & 0.25 & 0.01 & @xmath417 & 2 + 1999 & sep & 18 & 425 & 42 & 0.61 & 0.03 & 0.19 & 0.01 & @xmath418 & 3 + 1999 & dec & 09 & 416 & 42 & 0.66 & 0.03 & 0.24 & 0.01 & @xmath419 & 2 + 2000 & may & 15 & 417 & 42 & 0.64 & 0.03 & 0.20 & 0.01 & @xmath413 & 2 + 2000 & aug & 07 & 392 & 39 & 0.66 & 0.03 & 0.22 & 0.01 & @xmath417 & 2 + 2000 & nov & 05 & 415 & 41 & 0.70 & 0.03 & 0.24 & 0.01 & @xmath412 & 2 + 2000 & nov & 06 & 393 & 39 & 0.74 & 0.06 & 0.29 & 0.03 & @xmath420 & 3 + 2001 & mar & 31 & 394 & 39 & 0.69 & 0.03 & 0.26 & 0.01 & @xmath419 & 2 + 2001 & jun & 29 & 386 & 39 & 0.80 & 0.03 & 0.29 & 0.01 & @xmath412 & 2 + 2001 & oct & 19 & 364 & 36 & 0.79 & 0.04 & 0.31 & 0.02 & @xmath419 & 4 + 2001 & dec & 21 & 385 & 39 & 0.83 & 0.03 & 0.30 & 0.01 & @xmath421 & 3 + 2002 & apr & 14 & 327 & 33 & 0.76 & 0.03 & 0.29 & 0.01 & @xmath412 & 3 + 2002 & jul & 14 & 327 & 33 & 0.82 & 0.03 & 0.27 & 0.01 & @xmath412 & 2 + 2002 & nov & 20 & 359 & 36 & 0.73 & 0.04 & 0.19 & 0.01 & @xmath415 & 2 + 2003 & jan & 26 & 395 & 39 & 0.83 & 0.07 & 0.28 & 0.02 & @xmath416 & 2 + 2003 & may & 18 & 393 & 39 & 0.70 & 0.03 & 0.25 & 0.01 & @xmath415 & 2 + 2003 & sep & 08 & 449 & 45 & 0.63 & 0.03 & 0.22 & 0.01 & @xmath413 & 2 + 2003 & dec & 05 & 471 & 47 & 0.64 & 0.03 & 0.23 & 0.01 & @xmath417 & 2 + 2004 & mar & 06 & 501 & 50 & 0.61 & 0.02 & 0.22 & 0.01 & @xmath415 & 2 + 2004 & may & 18 & 512 & 51 & 0.60 & 0.03 & 0.21 & 0.01 & @xmath413 & 3 + 2004 & jun & 26 & 496 & 50 & 0.68 & 0.05 & 0.25 & 0.02 & @xmath416 & 2 + 2004 & dec & 11 & 475 & 47 & 0.59 & 0.03 & 0.20 & 0.01 & @xmath413 & 2 + 2005 & jan & 15 & 490 & 49 & 0.62 & 0.03 & 0.20 & 0.01 & @xmath417 & 2 + 2005 & may & 28 & 477 & 48 & 0.59 & 0.03 & 0.20 & 0.01 & @xmath412 & 2 + 2005 & jul & 16 & 478 & 48 & 0.59 & 0.02 & 0.19 & 0.01 & @xmath415 & 2 + figure [ 2250modelplot ] shows that the size of the gaussian ( parametrized by major and minor axis lengths ) is variable from epoch to epoch , and that there is an anti - correlation between the size and flux density of the gaussian . in contrast , the orientation of the gaussian is approximately constant ( p.a . of the major axis @xmath422 ) .
while the apparently random epoch - to - epoch fluctuations in the size of the gaussian may well be largely due to limitations of the one - component model , the anti - correlation between size and flux density suggests that activity is at least mainly confined to a relatively compact region , namely the core or a shocked region of the jet .
we did not include data from baselines to ti in our above analysis since ti was not used consistently over the @xmath28.5 yr period of our observing program , and because including these data led to systematic variations in the results for the size and orientation of the elliptical gaussian .
( because their radio structures are extended largely east - west , the results presented for sources 3c 454.3 and b2252 + 172 are far less sensitive to the inclusion of the long , north - south oriented , ti baselines . )
the major - axis length ( fwhm ) of the gaussian is significantly smaller and less variable when ti baselines are included than when they are left out : @xmath423 mas instead of @xmath424 mas ( for the same epochs ) .
this comparison suggests that the ti baselines further constrain the compact central region .
additionally , the major axis of the gaussian rotates when ti baselines are included : @xmath425 p.a .
becomes @xmath426 .
the @xmath427 estimate plausibly better represents the alignment of the jet axis near the core ( see [ 2250canduimages ] ) .
we attempted to separate the changes in the compact central region from changes in the lower - level structure to the northwest and south - southwest by also fitting to the data a model consisting of three point sources .
unfortunately , we were unable to adequately separate the three components : we found the same anti - correlation between overall source size and total flux density with the three - component model as with the one - component model ; i.e. , we found that during periods of increased flux density ( 1 ) the flux densities of all three components increased , and ( 2 ) the northwest and southwest components moved closer to the central component ( which we fixed to the position of the brightness peak at each epoch ) .
we also found no significant change with time of the angles of the position vectors drawn from the center component to the northwest component or from the center component to the southwest component .
we present in figure [ 2250imagescxu ] the 5.0 , 8.4 , and 15.4 ghz images of b2250 + 194 generated for the vlbi observations on 2004 may 18 .
the 8.4 ghz image is the same as that presented in figure [ 2250imagesplot ] except for the omission of the lowest contour level .
the image characteristics are given in table [ 2250imstat ] .
the images at all three frequencies show a compact emission region near the brightness peak and low - level emission to the northwest and south or south - southwest . the upper limit on the angular size of the compact emission region
is set approximately by the angular resolution of our vlbi array at 15.4 ghz ( see below ) . between 5.0 and 15.4 ghz ,
the peak flux density in the images increases @xmath273% , and the total clean flux density increases @xmath255% .
when taken together with our observation that the flux density at 8.4 ghz is at or near a maximum on 2004 may 18 ( see top panel in figure [ 2250modelplot ] ) , the small size and inverted spectrum ( between 5.0 and 15.4 ghz ) of the compact central region suggest that the brightness peak in the image at each frequency nearly coincides with the core .
scenario 1 above ( see [ 2250imagestext ] ) , which places the core at either the north or south end of the source , thus seems unlikely .
we fit to the 5.0 and 15.4 ghz data the same one - component model described in [ 2250uvplanefits ] .
the results are given in table [ 2250compcxu ] .
the uncertainties are standard errors ( see [ 2250uvplanefits ] ) , with the flux - density uncertainties including also a 5% uncertainty in the vlbi flux - density scale for 2004 may 18 .
we adopt a ( conservative ) 5% uncertainty for this epoch rather than the nominal 10% ( see [ vlbiamp ] ) , because the vlbi- and vla - determined flux densities at each frequency differ by only 14% ( see table [ 2250imstat ] ) .
the results at each frequency for the flux density , size , and orientation of the gaussian are consistent with a compact emission region near the brightness peak that is progressively more dominant , relative to the surrounding low - level emission , as the frequency increases from 5.0 to 15.4 ghz .
we estimate , using the @xmath428 upper limit on the major - axis length ( fwhm ) of the gaussian at 15.4 ghz , a maximum diameter for the compact emission region of @xmath429 mas . the orientation of the gaussian at 15.4 ghz ( @xmath430 ) is consistent with that at 8.4 ghz when ti baselines are included in the fit .
this consistency supports the accuracy of the results obtained by including ti baselines at 8.4 ghz , and provides additional evidence for the underlying orientation of the jet axis near the core .
( note that telescope ti did not observe at either 5.0 or 15.4 ghz . )
the decrease in the spectral index from @xmath431 between 5.0 and 8.4 ghz to @xmath432 between 8.4 and 15.4 ghz is consistent with the spectrum for the compact emission region peaking at a frequency near 15.4 ghz .
c@c@c@c@c@c@c@c@c 5.0 & 342 & 19 & 0.81 & 0.13 & 0.51 & 0.02 & @xmath421 & 5 + 8.4 & 512 & 27 & 0.60 & 0.03 & 0.21 & 0.01 & @xmath413 & 3 + 15.4 & 549 & 32 & 0.31 & 0.03 & 0.12 & 0.02 & @xmath433 & 4 +
we present in figure [ 2252imagesplot ] the 8.4 ghz images of b2252 + 172 generated from each of the 12 sessions of vlbi observations made between 2002 november and 2005 july .
the image characteristics are summarized in table [ 2252imstat ] .
the images reveal a source consisting of a relatively bright , compact emission region and extended emission at or just above the noise level . near the brightness peak
, the source shows no deviation from the restoring beam , down to the @xmath210% level . below this level , i.e. , at the brightness level of the outermost three or four contours
, the source extends west of the brightness peak . for our last three epochs , a very weak emission component ,
itself extended east - west , appears ( above the new peaks in the background ) @xmath22.6 mas west - southwest of the brightness peak .
this component is likely real , since its appearance is robust for these epochs against changes ( e.g. , to the positioning of clean windows ) in the imaging process .
the flux density of this component is @xmath20.7 mjy .
variations in the quality of the images from epoch to epoch ) . ]
make it difficult to trace any motion of this component relative to the brightness peak .
however , there is evidence for the evolution of the low - level emission @xmath20.5 mas to @xmath21.5 mas west of the brightness peak .
the peak flux density in the image for 2003 may is at or near the maximum we observed during the @xmath22.7 yr span of our observations of this source .
the low - level emission at this 2003 may epoch shows minimal separation from the brightness peak . in subsequent epochs ,
the low - level emission extends increasingly westward toward the @xmath20.7 mjy component .
this evolution is consistent with the ejection and subsequent motion of a new jet component .
the brightness peak in each image near the easternmost edge of the source and the apparent westward motion of the low - level emission suggests that the brightness peak nearly coincides with the core .
l@ l@ l@r c c c c c c c c 2002 & nov & 20 & & 14.5 & @xmath379 & 0.10 & 1.88 & 0.71 & @xmath434 & 18.0 & + 2003 & jan & 26 & & 14.3 & @xmath178 & 0.12 & 1.88 & 0.67 & @xmath435 & 20.6 & 0.82 + 2003 & may & 18 & & 18.7 & @xmath392 & 0.08 & 1.44 & 0.48 & @xmath436 & 22.9 & 0.85 + 2003 & sep & 08 & & 15.9 & @xmath437 & 0.06 & 1.69 & 0.55 & @xmath438 & 18.7 & 0.87 + 2003 & dec & 05 & & 16.4 & @xmath392 & 0.08 & 1.66 & 0.53 & @xmath439 & 19.1 & + 2004 & mar & 06 & & 12.4 & @xmath371 & 0.08 & 2.03 & 0.67 & @xmath440 & 15.2 & 0.89 + 2004 & may & 18 & & 9.7 & @xmath441 & 0.15 & 1.79 & 0.67 & @xmath442 & 12.5 & 0.78 + 2004 & may & 18 & ( c ) & 10.2 & @xmath443 & 0.17 & 3.20 & 0.95 & @xmath444 & 14.3 & 0.83 + 2004 & may & 18 & ( u ) & 9.3 & @xmath445 & 0.60 & 1.16 & 0.59 & 9.62 & 10.2 & 0.89 + 2004 & jun & 26 & & 8.9 & @xmath446 & 0.10 & 1.92 & 0.67 & @xmath447 & 13.0 & 0.82 + 2004 & dec & 11 & & 11.6 & @xmath448 & 0.09 & 1.87 & 0.53 & @xmath449 & 15.5 & 0.85 + 2005 & jan & 15 & & 11.6 & @xmath183 & 0.05 & 1.94 & 0.55 & @xmath450 & 15.4 & 0.88 + 2005 & may & 28 & & 16.8 & @xmath185 & 0.05 & 1.93 & 0.54 & @xmath451 & 20.8 & 0.90 + 2005 & jul & 16 & & 16.9 & @xmath452 & 0.04 & 1.98 & 0.56 & @xmath453 & 21.0 & 0.80 + to more quantitatively investigate changes in the radio structure of b2252 + 172 near its brightness peak , we fit to the data a model consisting of a single elliptical gaussian .
our results ( not tabulated ) show that the emission region near the brightness peak is elongated along @xmath454 and has major - axis length ( fwhm ) @xmath455 mas .
the quoted parameter values and uncertainties are the mean and rms variation about the mean over our 12 observing sessions .
the orientation of the gaussian suggests that the jet axis may bend slightly from @xmath456 near the position of the brightness peak to @xmath457 at distances @xmath91 mas from the brightness peak ( see also figure [ 2252imagesplot ] ) .
no systematic variations in either the orientation or the size of the gaussian are evident over our 12 epochs .
we also fit to the data a two - point - source model in an attempt to characterize the changes observed in the low - level emission immediately west of the brightness peak .
unfortunately , the snr of the phase - referenced data was not sufficiently high to adequately constrain both the relatively bright component near the brightness peak and the weak secondary component .
we present in figure [ 2252imagescxu ] the 5.0 , 8.4 , and 15.4 ghz images of b2252 + 172 generated for the vlbi observations on 2004 may 18 . the 8.4 ghz image is the same as that presented in figure [ 2252imagesplot ] except for the omission of the lowest contour level .
the image characteristics are given in table [ 2252imstat ] .
the images at 5.0 and 8.4 ghz each show a compact emission region near the brightness peak and low - level emission to its west . the image at 15.4 ghz shows that the compact emission region is marginally extended northeast - southwest .
the component @xmath22.6 mas west - southwest of the brightness peak , identified in the 8.4 ghz images only at later epochs , already appears prominently in the 5.0 ghz image for 2004 may 18
. the greater prominence at 5.0 ghz of this component is consistent with steep - spectral - index emission from one or more jet components perhaps several years after ejection .
the evolving low - level emission observed at 8.4 ghz closer to the brightness peak ( see [ 2252imagestext ] ) may derive from a more recent ejection event .
we fit to the 5.0 and 15.4 ghz data the same one - component model described in [ 2252uvplanefits ] . at 5.0 ghz ,
the best - fit gaussian is elongated along @xmath458 and has major - axis length ( fwhm ) @xmath459 mas . at 15.4 ghz ,
the gaussian is elongated along @xmath460 and has major - axis length ( fwhm ) @xmath461 mas .
the uncertainties are standard errors , for which the systematic contributions were estimated using a procedure similar to that described in [ 2250uvplanefits ] .
the results for the orientation are consistent with the mean value ( @xmath454 ) found at 8.4 ghz for the 12 epochs we observed this source . using the @xmath428 upper limit on the major - axis length ( fwhm ) of the gaussian at 15.4 ghz , we estimate a maximum diameter of @xmath462 mas for the compact emission region about the brightness peak .
the flat spectrum of the emission between 5.0 and 15.4 ghz ( @xmath463 ) suggests that the brightness peak in the image at each frequency nearly coincides with the core .
our 35 images at 8.4 ghz , obtained from data gathered between 1997 january and 2005 july , show that the core region ( easternmost @xmath21 mas ) of 3c 454.3 is quite variable . for each epoch , we fit to the core region in the image a model consisting of two point sources .
we refer to the eastern component as c1 and the western component as c2 .
our 8.4 ghz two - component model only approximately represents the complex underlying structure seen in vlbi images at higher frequencies ( namely 43 ghz and 86 ghz ) .
nevertheless , we determined via a simulation study that c1 and c2 are closely aligned with the two principal 43 ghz components identified by @xcite : c1 is located @xmath0 mas west of the 43 ghz core and c2 is located @xmath162 mas west - southwest of the `` stationary '' component located @xmath20.65 mas west of the 43 ghz core . since component c1 is the more nearly aligned with the 43 ghz core , we consider it to be the 8.4 ghz core .
the uncertainties quoted above represent real variations in the positions of c1 and c2 , which appear to be largely caused by the westward motion of new components along the jet axis .
the 43 ghz images of @xcite show the passage of four new jet components between the core and `` stationary '' component between 1997 january and 2001 march ( the time period in common between our and jorstad et al.s vlbi programs ) .
if we consider a simplified picture for the core region consisting of the core , new jet component , and the `` stationary '' component , then we can interpret the position changes of c1 and c2 as follows : after its ejection , the jet component moves west away from the core . since the 8.4 ghz images do not resolve the core and jet component , c1 shifts first slightly west .
as the jet component continues westward , separating itself further from the core , c1 eventually `` relaxes '' back east toward the core .
the 0.06 mas value of the rms scatter suggests that the forward - and - back `` motion '' of c1 spans not much more than 0.12 mas .
a similar scenario occurs as the jet component approaches and subsequently passes the `` stationary '' component . here , c2 shifts first east , then west , and finally back to its starting point .
the rms scatter ( 0.08 mas ) suggests that the back - forward - back `` motion '' of c2 spans not much more than 0.16 mas .
the proper motion of c2 with respect to c1 between 1997 january and 2005 july was @xmath155 in @xmath21 and @xmath464 in @xmath22 .
this relative proper motion corresponds to an apparent transverse velocity of @xmath20.9@xmath1 ( see figure [ 3ccompmotions ] ) .
this result indicates that over the @xmath28.5 yr period of our observing program either c1 `` moved '' systematically east - northeast and/or c2 `` moved '' systematically west - southwest .
given the proximity of c1 to the 43 ghz core , the latter is more likely .
moreover , vlbi images at 43 and 86 ghz produced from observations between 2005 july and november @xcite show the presence of a bright , extended jet component with peak @xmath20.4 mas west of the `` stationary '' component . a plausible westward motion of this component at earlier epochs , first toward , and then beyond
, the nominal position of the `` stationary '' component , would have caused our c2 to `` move '' approximately westward away from c1 as we observed .
the various component models of 3c 454.3 for the time period of our vlbi program appear to be able to accommodate all of the features and changes we observed at 8.4 ghz in the core region of this source . based on its position near the easternmost edge of the 8.4 ghz radio structure , close spatial association with the 43 ghz core , and relatively flat spectrum , we believe c1 is the best choice for the ultimate reference point for the guide star . the general westward motion away from c1 of the other 8.4 ghz components ( see figure [ 3ccompmotions ] ) further supports this choice . in paper iii
, we estimate the proper motion of c1 relative to the updated crf .
we can trace in our 8.4 ghz images the motions of several jet components at varying distances from c1 .
we determined the proper motions ( relative to c1 ) of d1 , d2 , and j1 , respectively located @xmath21.7 , @xmath23.7 , and @xmath25.5 mas west of the core .
the proper motion of d1 was @xmath465 ( @xmath24@xmath1 ) directed west - southwest , and that of d2 was @xmath466 ( @xmath25@xmath1 ) directed northwest ( see figure [ 3ccompmotions ] ) .
the difference in direction is consistent with the jet having a bend at a distance from the core of @xmath23 mas .
the proper motion of j1 ( see figure [ 3ccompmotions ] ) changed from a northwestward @xmath467 masyr@xmath3 ( @xmath23@xmath1 ) between 1997 january and 2003 may to a westward @xmath468 ( @xmath20.8@xmath1 ) between 2003 september and 2005 july .
the change in direction and magnitude of the proper motion of j1 may signify a second bend in the jet axis , and possibly a transition away from the highly collimated flow seen in the easternmost @xmath25.5 mas .
we observed `` bulk motion '' ( @xmath467 directed northwest , see figure [ 3ccompmotions ] ) in the low - surface - brightness emission beyond j1 , but did not identify individual components in this extended region .
our 35 images at 8.4 ghz of b2250 + 194 reveal a source consisting of a bright , compact emission region and low - level emission which extends both northwest and south - southwest of the brightness peak .
we were unable using a three - component model to discern any relative motions in the radio structure .
however , we determined using a model consisting of a single elliptical gaussian that the size and flux density of this source were anti - correlated over the course of our observing program .
this result indicates that the compact central region was likely more dominant during periods of increased flux density . by also fitting the elliptical gaussian model to our 5.0 and 15.4 ghz data separately for 2004 may 18
, we determined that near a flux - density maximum the compact central region ( 1 ) had an inverted spectrum , ( 2 ) had a diameter ( at 15.4 ghz ) @xmath40.40 mas , and ( 3 ) was elongated ( at 15.4 ghz ) along @xmath469 .
the brightness peak in the image for 2004 may 18 for each frequency likely nearly coincides with the core .
if periodic flux density increases were accompanied by the ejection and subsequent motion of new components , then we may expect the 8.4 ghz brightness peak to have shifted approximately north - northwest or south - southeast on the sky between each flux - density maximum and the subsequent minimum .
since the overall 8.4 ghz structure of b2250 + 194 appears approximately static over the course of our observing program , it is likely that any shifts in the position of the brightness peak were small ( peak - to - peak variations @xmath180.2 mas ) . in paper
iii , we bound the proper motion of the 8.4 ghz brightness peak of b2250 + 194 relative to the updated crf .
our 12 images at 8.4 ghz of b2252 + 172 reveal a source consisting of a relatively bright , compact emission region and low - level emission which extends west - southwest at least @xmath22.6 mas from the brightness peak . using a model consisting of a single elliptical gaussian
, we determined that the compact emission region was elongated along @xmath454 and had a diameter @xmath470 mas ( fwhm ) . by also fitting the elliptical gaussian model to our 5.0 and 15.4 ghz data separately for 2004 may 18
, we determined that ( at this epoch ) the spectrum of the compact emission region between these frequencies was nearly flat .
the brightness peak in the image for 2004 may 18 for each frequency likely nearly coincides with the core . given the observed evolution of the low - level emission immediately west of the compact emission region
, we may expect the 8.4 ghz brightness peak to have shifted slightly westward between 2002 november and 2005 july , as a possible new component was first ejected ( on or near 2003 may ) and then moved away from the core .
any such shift was likely small ( peak - to - peak variations @xmath180.2 mas ) , since the low - level emission is @xmath1810% of the brightness peak in the 8.4 ghz images . in paper iii
, we bound the proper motion of the 8.4 ghz brightness peak of b2252 + 172 relative to the updated crf .
here we give a summary of our results and conclusions ( all quoted uncertainties are estimated standard errors , including allowance for systematic errors ) : \1 .
each of our 8.4 ghz images of quasar 3c 454.3 , obtained from data at 35 epochs spread between 1997 january and 2005 july , shows a bright , relatively compact core region and a bent milliarcsecond - scale jet .
the images reveal significant structural changes in the core region ( easternmost @xmath21 mas ) , and three compact , moving jet components at various distances from the core .
we modeled the structure of the 3c 454.3 core region with two point sources oriented approximately east - west .
we refer to the eastern component as c1 and the western component as c2 .
we determined a relative proper motion for c2 with respect to c1 of @xmath471 in @xmath21 and @xmath156 in @xmath22 , an apparent transverse velocity of @xmath20.9@xmath1 .
c1 and c2 are closely aligned with the two principal core - region components identified in vlbi images at 43 ghz by @xcite : c1 is located @xmath0 mas west of the 43 ghz core and c2 is located @xmath162 mas west - southwest of a component observed at 43 ghz to be located @xmath20.65 mas west of the core . since component c1 is most nearly aligned with the 43 ghz core , we consider it to be the 8.4 ghz core .
the `` oscillations '' we observed in the 8.4 ghz separation of c2 relative to c1 probably reflect the motions of new jet components periodically ejected from near the position of the 43 ghz core .
the proper motion of c2 relative to c1 could be a consequence of the westward motion of a bright jet component , which was visible at the westernmost edge of the core region in 43 and 86 ghz images made 14 months after our last observing session .
we determined the proper motions relative to c1 of the 3c 454.3 jet components d1 , d2 , and j1 , respectively located @xmath21.7 , @xmath23.7 , and @xmath25.5 mas west of the core .
the proper motion of d1 was @xmath472 ( @xmath24@xmath1 ) towards the west - southwest , and that of d2 was @xmath466 ( @xmath25@xmath1 ) towards the northwest .
the change in the direction presumably demonstrates a bend in the jet axis @xmath23 mas from the core .
the proper motion of j1 changed from a northwestward @xmath467 ( @xmath23@xmath1 ) between 1997 january and 2003 may to a westward @xmath473 ( @xmath20.8@xmath1 ) between 2003 september and 2005 july .
the change in direction and magnitude of the proper motion of j1 may signify a second bend in the jet axis @xmath25.5 mas from the core .
we observed `` bulk motion '' ( @xmath467 directed northwest ) in the low - surface - brightness emission beyond j1 .
\5 . our 15.4 ghz image of 3c 454.3 for 2004 may 18 shows a `` new '' jet component between c1 and c2 .
our 8.4 ghz images suggest that this component was ejected from near the position of c1 on or about 2003
september and arrived near the position of c2 @xmath21.4 yr later .
we fit an elliptical gaussian to the radio structure of b2250 + 194 .
we found the size and flux density of the gaussian to be anti - correlated , suggesting that the variability of the source was greatest near the brightness peak .
the 8.4 ghz flux density of b2250 + 194 reached a maximum in our observations on or about 2004 may .
our 5.0 , 8.4 , and 15.4 ghz elliptical - gaussian models for 2004 may 18 show that the compact central region ( 1 ) had an inverted spectrum , ( 2 ) had a diameter ( at 15.4 ghz ) @xmath40.40 mas , and ( 3 ) was elongated ( at 15.4 ghz ) along @xmath474 .
the brightness peak in the image for 2004 may 18 for each frequency likely nearly coincides with the core .
we fit an elliptical gaussian to the radio structure of b2252 + 172 and found the gaussian to be elongated along @xmath475 and to have a diameter @xmath476 mas ( fwhm ) .
our 5.0 , 8.4 , and 15.4 ghz elliptical - gaussian models for 2004 may 18 show that the compact emission region had a flat spectrum .
the brightness peak in the image for 2004 may 18 for each frequency likely nearly coincides with the core .
\11 . based primarily on its close spatial association with the 43 ghz core
, we believe 3c 454.3 component c1 to be the best choice for the ultimate reference point for the guide star . the inverted to flat spectra and small bounds for any changes near their brightness peaks make b2250 + 194 and b2252 + 172 useful secondary reference sources . acknowledgments .
we thank the anonymous referee for a constructive review of the paper and for comments helpful in the preparation of the final manuscript .
this research was primarily supported by nasa , through a contract with stanford university to the smithsonian astrophysical observatory ( sao ) , and a subcontract from sao to york university .
the national radio astronomy observatory ( nrao ) is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
the dsn is operated by jpl / caltech , under contract with nasa .
we thank s. g. jorstad for providing the 43 ghz models used in the simulation study of 3c 454.3 . we have made use of the united states naval observatory ( usno ) radio reference frame image database ( rrfid ) and nasa s astrophysics data system abstract service , developed and maintained by sao .
, d. s. , schwab , f. r. , & sramek , r. a. 1999 , in astronomical society of the pacific conference series , vol .
180 , synthesis imaging in radio astronomy ii , ed .
g. b. taylor , c. l. carilli , & r. a. perley , 127 , i. i. k. 1998 , in asp conference series , vol .
164 , asp conf .
ser . 144 : iau colloq . 164 : radio emission from galactic and extragalactic compact sources , ed .
j. a. zensus , g. b. taylor , & j. m. wrobel ( san francisco : asp ) , 75 | we used 8.4 ghz vlbi images obtained at up to 35 epochs between 1997 and 2005 to examine the radio structures of the main reference source , 3c 454.3 , and two secondary reference sources , b2250 + 194 and b2252 + 172 , for the guide star for the nasa / stanford relativity mission _ gravity probe b _ ( ) . for one epoch in 2004 may , we also obtained images at 5.0 and 15.4 ghz . the 35 8.4 ghz images for quasar 3c 454.3 confirm a complex , evolving , core - jet structure .
we identified at each epoch a component , c1 , near the easternmost edge of the core region .
simulations of the core region showed that c1 is located , on average , @xmath0 mas west of the unresolved `` core '' identified in 43 ghz images .
we also identified in 3c 454.3 at 8.4 ghz several additional components which moved away from c1 with proper motions ranging in magnitude between 0.9@xmath1 and 5@xmath1 .
the detailed motions of the components exhibit two distinct bends in the jet axis located @xmath23 and @xmath25.5 mas west of c1 .
the spectra between 5.0 and 15.4 ghz for the `` moving '' components are steeper than that for c1 .
the 8.4 ghz images of b2250 + 194 and b2252 + 172 , in contrast to those of 3c 454.3 , reveal compact structures .
the spectrum between 5.0 and 15.4 ghz for b2250 + 194 is inverted while that for b2252 + 172 is flat .
based on its position near the easternmost edge of the 8.4 ghz radio structure , close spatial association with the 43 ghz core , and relatively flat spectrum , we believe 3c 454.3 component c1 to be the best choice for the ultimate reference point for the guide star . the compact structures and inverted to flat spectra of b2250 + 194 and b2252 + 172 make these objects valuable secondary reference sources . |
the understanding of the transition between the confinement and the higgs phase of asymptotically free nonabelian gauge theories in presence of higgs fields is a relevant and yet not fully unraveled topic .
needless to say , such theories are the building blocks of the standard model
. numerical studies of the lattice version of these models @xcite have already revealed a rich structure of the corresponding phase diagram , as well as mean - field approaches to the underlying lattice models as in @xcite .
what emerges from the lattice works is that the transition between the confinement and the higgs phase strongly depends on the representation of the higgs field . in the case of the fundamental representation ,
it turns out that these two phases can be continuously connected , _
i.e. _ they are not completely separated by a transition line . for the benefit of the reader ,
it is worth spending a few words on this statement .
we follow here the seminal work by @xcite , where the lattice version of the model has been considered for a higgs field with fixed length , _
i.e. _ by freezing its radial part
. this amounts to keep the higgs quartic self - coupling very large , not to say infinite , so that the higgs field is frozen to its _ vev_. the resulting theory has two basic parameters , the gauge coupling @xmath3 and the _ vev _ @xmath4 of the higgs field . in the plane
@xmath5 , the higgs phase corresponds to the region of weak coupling , _
i.e. _ small @xmath3 and sufficiently large @xmath4 . in this region
, the wilson loop exhibits a perimeter law and the force between two static sources is short ranged .
instead , the confining phase corresponds to the strong coupling region in the @xmath5 plane , characterized by large values of @xmath3 and sufficiently small @xmath4 . here , the wilson loop gives rise to a linear potential over some finite distance region , followed by string breaking via scalar particle production .
it turns out that these two regions are smoothly connected .
more precisely , from e.g. fig.1 of @xcite , one learns that the confining and higgs phases are separated by a first order line transition which , however , does not extend to the whole phase diagram .
instead , it displays an endpoint .
this implies that the higgs phase can be connected to the confining phase in a continuous way , _
i.e. _ without crossing the transition line .
even more striking , one observes a region in the @xmath5 plane , called analyticity region @xcite , in which the higgs and confining phases are connected by paths along which the expectation value of any local correlation function varies analytically , implying the absence of any discontinuity in the thermodynamical quantities . according to @xcite , the spectrum of the theory evolves continuously from one regime to the other .
said otherwise , the physical states in both regimes are generated by suitable gauge invariant operators which , in the higgs phase , give rise to massive bosons while , in the confining phase , to a meson like state ,
_ i.e. _ to a bound state of confined excitations
. we shall stick to the use of the word `` phase '' here , although it is clear there can not exist a local order parameter serve as an order parameter since the presence of a fundamental higgs field breaks the center symmetry in a hard way , giving anyhow a nontrivial expectation value to @xmath6 . ] discriminating between the higgs or confinement behaviour . in the adjoint higgs case ,
the breaking or not of center symmetry and associated polyakov loop expectation value can in principle be used to discriminate between phases , leading to ( an ) expected phase transition(s ) between confining and higgs behaviour , as already put forward in @xcite and tested in e.g. @xcite . moreover , as speculated in @xcite , the appearance of an in some cases expected coulomb phase with long - range correlations was not reported . over the years , most attention has however been paid to the fundamental higgs , due to its more direct physical relevance for the electroweak physics .
the attentive reader will have noticed that most of the cited works either concern lattice simulations of the gauge higgs systems , or analytical semiclassical mean - field analysis of the lattice model . there is a good reason for this as the probed physics is usually of a nonperturbative nature . some preliminary functional dyson - schwinger approaches to gauge field theories supplemented with
scalar fields have appeared nonetheless @xcite , however these do not really touch upon the supposed phase diagram and the analyticity results of fradkin & shenker .
direct computation of phase diagram related quantities as polyakov loops or even the vacuum ( free ) energy are noteworthy hard to access at the functional level .
let us by the way notice here that the proof of the fradkin - shenker analyticity result , inspired by @xcite , heavily relies on the lattice formulation of the problem .
we are unaware of a strict continuum version of the theorem , although one might expect it to hold as well .
it is in any case a strong property of the theory if it would be analytical in a certain region of the phase diagram . in some cases ,
the fradkin & shenker result has been related to the lee - yang theorem that also handles analyticity properties of statistical systems @xcite .
one can imagine a lot of subtleties making any concrete analytical study of the higgs phase diagram troublesome . to name only one ,
renormalization will require to consider the gauge coupling constant at a particular renormalization scale and the issue of strong versus weak coupling is directly set by the size of this scale .
we will come back to this issue later .
the aim of this work is that of investigating the transition between the higgs and the confinement phase within a continuum quantum field theory .
the set up which we shall follow is that of taking into account the nonperturbative effect of the existence of the gribov copies @xcite which are unavoidably present in the gauge fixing quantization procedure . in particular , as already shown in the case of three - dimensional gauge theories @xcite , this framework enables us to encode nonperturbative information about the phases of the theory in the two - point correlation function of the gauge field in momentum space .
the transition between one phase to another is detected by the corresponding change in the pole structure of the gluon propagator .
more precisely , a gluon propagateor of the gribov type , _
i.e. _ displaying complex conjugate poles , has no particle interpretation , being well suited to describe the confining phase . on the other hand , a propagator of the yukawa type , _
i.e. _ displaying a real pole in momentum space , means that the theory is located in the higgs phase . by quantizing the theory in the landau gauge and by performing the restriction to the so called gribov region @xmath7 @xcite in order to take into account the effect of the gribov copies
, we shall be able to discuss the changes in the gluon propagator when the parameters @xmath5 are varied from the higgs weak coupling region to the confining strong coupling region . as we shall see in details , in the case of the fundamental representation we shall be able to detect a region in the @xmath5 plane , characterized by the line stands for the energy scale which shows up in the renormalization of ultraviolet divergent quantities . in the present case , dimensional regularization in the @xmath8 scheme is employed . ]
@xmath9 , with @xmath10 the higgs region will correspond to values of @xmath5 for which @xmath11 , while the confining phase will be located in the region @xmath12 .
moreover , the gluon propagator evolves in a smooth continuous way from the confining to the higgs region , namely we observe a continuous evolution of the poles of the propagator from a gribov type to the yukawa behavior .
in addition , we shall be able to evaluate the vacuum energy in gribov s semiclassical approximation and check explicitly that it is a continuous function , together with its first and second derivative , of the parameter @xmath13 across the transition line @xmath14 .
the third derivative of the vacuum energy displays a discontinuity across @xmath14 .
however , we shall be able to check that the line @xmath14 is not located in the fradkin - shenker analyticity region , something which we can not access within the present approximation as @xmath14 corresponds to a region wherein we can not really trust the made approximations .
though , we are able to confirm within a continuum field theory that the transition between the higgs and the confining phase occurs in a smooth way , something which is far from being trivial . another relevant result of our analysis is that , in the higgs phase , there is no need to implement the restriction to the gribov region @xmath7 . said otherwise , in the higgs phase , the values of the parameters @xmath5 are such that the theory automatically lies within the gribov region @xmath7 , so that the gribov horizon is never crossed .
this is a quite relevant observation which has an important physical consequence .
it ensures that , in the weak coupling region , the standard higgs mechanism takes place without being affected by the restriction to the gribov region .
we have thus a truly nonconfining theory whose asymptotic states are the massive gauge bosons .
when the higgs field is in the adjoint representation , our results indicate that things change drastically , as also expected from lattice investigations . here , the phase structure looks much more intricate as well as the evolution of the two - point correlation function of the gauge field .
in addition of the confining phase , in which the gluon propagator is of the gribov type , we shall find what can be called a @xmath15 confining phase , in which the third component @xmath16 of the gauge field displays a gribov type correlation function , while the remaining off - diagonal components , @xmath17 , @xmath18 , are of the yukawa type . it is worth noticing here that this phase has been reported in lattice investigations of the georgi - glashow model @xcite , _
i.e. _ of three dimensional gauge theories with a higgs field in the adjoint .
unfortunately , till now , we are unaware of lattice studies of the gluon propagator in four dimensions with higgs field in the adjoint . so far ,
only the case of the fundamental representation has been addressed @xcite . from that point of view
, we hope that our results will stimulate further studies of the gluon propagator on the lattice in order to confirm the existence of the @xmath15 confining phase in four dimensions , to the extent that the diagonal gauge boson propagator is not of the massless type for finite value of the higgs condensate .
a lattice study of the adjoint higgs phase diagram was presented in @xcite , giving evidence of a massless coulomb phase in the limit of infinite higgs condensate @xmath4 .
more precisely , the theory was shown to reduce to a compact @xmath15 model with its confinement - deconfinement transition .
we do however not expect such a transition in the continuum version of qed .
we shall see that in the limit @xmath19 we can however recover also a massless photon .
this is not as trivial a result as it might appear since it involves a delicate cancelation between diverging higgs condensate and vanishing gribov parameter .
the paper is organized as follows . in sect . 2 we discuss the restriction to the gribov region in the case of the fundamental representation .
the behavior of the gluon propagator and of the vacuum energy are discussed within gribov s approximation . in sect .
3 we address the more intricate case of the higgs field in the adjoint representation .
4 collects our conclusion .
let us consider first the case of @xmath1 yang - mills theories interacting with higgs fields in the fundamental representation . this will be the most interesting case for future reference as well , when the physical case of @xmath20 will be analyzed , _
i.e. _ the electroweak theory . working in euclidean space and adopting the landau gauge , @xmath21 ,
the action of the current model is specified by the following expression @xmath22 where the covariant derivative is defined by @xmath23 the field @xmath24 stands for the lagrange multiplier implementing the landau gauge , @xmath21 , while @xmath25 are the faddeev - popov ghosts .
the indices @xmath26 refer to the fundamental representation , and @xmath27 , are the pauli matrices . the vacuum configuration which minimizes the energy
is achieved by a constant scalar field parameterized as @xmath28 it will be understood that we work in the limit @xmath29 for simplicity , _ i.e. _ we have a higgs field frozen at its vacuum expectation value , as in @xcite .
all components of the gauge field acquire the same mass @xmath30 .
in fact , for the quadratic part of the action we have now @xmath31 as already mentioned in the introduction , the aim of the present work is that of analyzing the possible nonperturbative dynamics of the model by taking into account the gribov copies . in the landau gauge gauges , which have proven their usefulness as being renormalizable and offering a way to get rid of the unwanted propagator mixing between ( massive ) gauge bosons and associated goldstone modes , @xmath32 .
the latter terms indeed vanish upon using the gauge field transversality .
the upshot of specifically using the landau gauge is that it allows to take into account potential nonperturbative effects related to the gauge copy ambiguity . ] , this issue can be faced by restricting the domain of integration in the path integral to the so called gribov region @xmath7 @xcite , defined as the set of all transverse gauge configurations for which the faddeev - popov operator is strictly positive , namely @xmath33 the region @xmath7 is known to be convex and bounded in all directions in field space . the boundary of @xmath7 , where the first vanishing eigenvalue of the faddeev - popov operator appears , is called the first gribov horizon . a way to implement the restriction to the region
@xmath7 has been worked out by gribov in his original work .
it amounts to impose the no - pole condition @xcite for the connected two - point ghost function @xmath34 , which is nothing but the inverse of the faddeev - popov operator @xmath35 .
one requires that @xmath36 has no poles at finite nonvanishing values of @xmath37 , so that it stays always positive . in that way
one ensures that the gribov horizon is not crossed , _
i.e. _ one remains inside @xmath7 .
the only allowed pole is at @xmath38 , which has the meaning of approaching the boundary of the region @xmath7 .
following gribov s procedure @xcite , for the connected two - point ghost function @xmath36 at first order in the gauge fields , one finds @xmath39where use has been made of the transversality condition @xmath40 .
taking into account that all masses of the gauge field are degenerate in color space , eq .
, we introduce the ghost form factor @xmath41 as @xmath42the quantity @xmath41 turns out to be a decreasing function of the momentum @xmath43 @xcite .
thus , the no - pole condition for the ghost function @xmath44 is implemented by imposing that @xcite @xmath45where @xmath46 is given by @xmath47 this expression is obtained by taking the limit @xmath48 of eq . , and by making use of the property @xmath49 which follows from the transversality of the gauge field , @xmath50 .
also , it is useful to remind that , for an arbitrary function @xmath51 , we have @xmath52where , upon contracting both sides of eq . with @xmath53 , @xmath54 in order to ensure the restriction to the gribov region @xmath7 in the functional integral , we encode the information of the no - pole conditions into a step function @xcite : @xmath55\delta ( \partial a)\ ; \det ( -\partial d^{ab})\ ; \theta ( 1-\sigma(0;a))\ ; e^{-s_{ym}}. } \label{im}\ ] ] though , as our interest for now lies only in the study of the gauge boson propagators , we shall consider here the quadratic approximation for the partition function , namely @xmath56\ ; e^{\vartheta ( 1-\sigma ( 0,a ) ) } \ ; e^{-\frac{1}{4}\int d^{4}x(\partial _
{ \mu } a_{\nu } ^{a}-\partial _ { \nu } a_{\mu } ^{a})^{2}-\frac{1}{2\xi } \int { d^{4}x(\partial _ { \mu } a_{\mu } ^{a})^{2}-}\frac{{g^{2}\nu ^{2}}}{4}\int { d^{4}xa_{\mu } ^{a } a_{\mu}^{a } } } \ ; , \label{zf}\ ] ] where use has been made of the integral representation @xmath57 the extension of this work beyond the semiclassical approximation can be worked out using the tools of @xcite .
we shall also not dwell upon renormalization details here , we expect that the proof of the pure gauge case as in @xcite can be suitably adapted .
after simple algebraic manipulations , one gets for @xmath58 \ ; e^{-\frac{1}{2 } \int \frac{d^{4}q}{(2\pi ) ^{4}}a_{\mu } ^{a } ( q)\mathcal{p}_{\mu \nu } ^{ab } a_{\nu } ^{b } ( -q ) } , \label{zq1f}\ ] ] with @xmath59 the parameter @xmath60 is understood to be zero at the end in order to recover the landau gauge . evaluating the inverse of the expression above and taking the limit @xmath61 , for the gluon propagator one
gets @xmath62 it remains to find the gap equation for the gribov parameter @xmath63 , enabling us to express it in terms of the parameters of the starting model , _
i.e. _ the gauge coupling constant @xmath3 and the _ vev _ of the higgs field @xmath4 .
in order to accomplish this task we follow @xcite and evaluate the partition function @xmath64 in the semiclassical approximation .
first , we integrate out the gauge fields , obtaining @xmath65 making use of @xmath66 for the determinant in expression ( [ zq2f ] ) we get @xmath67 \;.\ ] ] therefore , @xmath68 where , in the thermodynamic limit in expression can be neglected in the derivation of the gap equation , eq . , when taking the thermodynamic limit , see @xcite for details . ] , @xmath69 expression ( [ zf ] ) can be now evaluated in the saddle point approximation @xcite , _
i.e. _ @xmath70 where the parameter @xmath71 is determined by the stationary condition @xmath72 which yields the following gap equation @xmath73 notice also that the function @xmath74 has the meaning of the vacuum energy @xmath75 of the system .
more precisely @xmath76 as it is apparent from the expression of the partition function @xmath64 , eq .. to discuss the gap equation , we decompose the denominator according to @xmath77 with @xmath78 making use of the @xmath8 renormalization scheme in @xmath79 and of the standard integral @xmath80 for the gap equation we consequently get @xmath81 in order to analyze this equation we rewrite it in a more suitable way , _
i.e. _ @xmath82 so that @xmath83 whose final form can be written as @xmath84 where we have introduced the dimensionless variables @xmath85 with @xmath86 in order to have two real , positive , distinct roots @xmath87 , eq .. it is worth to underline that the renormalization scale @xmath88 could be exchanged in favour of the invariant scale @xmath89 , defined at one - loop as @xmath90 with @xmath91 given by @xcite @xmath92 where @xmath93 is the casimir of the representation of the higgs field equaling @xmath94 , resp .
@xmath95 , for the fundamental , resp .
adjoint , representation of @xmath1 . for @xmath96 , the roots @xmath87
become complex conjugate , and the gap equation takes the form @xmath97 moreover , it is worth noticing that both expressions , involve only one function , _
i.e. _ they can be written as @xmath98 where for @xmath99 we might take @xmath100 which is a real function of the variable @xmath101 .
expression is easily obtained from by rewriting it in the region @xmath102 . in particular ,
it turns out that the function @xmath103 for all @xmath101 , and strictly decreasing . as consequence ,
for each value of @xmath12 , equation has always a unique solution with @xmath104 .
moreover , it is easy to check that @xmath105 .
therefore , we can distinguish ultimately three regions , namely * when @xmath106 , eq .
has no solution for @xmath107 . as the gap equation has been obtained by acting with @xmath108 on the expression of the vacuum energy @xmath109 , eq .
, we are forced to set @xmath110 .
this means that , when @xmath106 , the dynamics of the system is such that the restriction to the gribov region can not be consistently implemented . as a consequence ,
the standard higgs mechanism takes place , yielding three massive gauge fields , according to @xmath111 for sufficiently weak coupling @xmath112 , we underline that @xmath13 will unavoidably be larger than @xmath113 .
* when @xmath114 , equation has a solution for @xmath115 . in this region ,
the roots @xmath87 are real and the gluon propagator decomposes into the sum of two terms of the yukawa type : @xmath116 where @xmath117 moreover , due to the relative minus sign in eq . only the component @xmath118 represents a physical mode . * for @xmath119 , equation has a solution for @xmath102 .
this scenario will always be realized if @xmath112 gets sufficiently large , i.e. at strong coupling . in this region
the roots @xmath87 become complex conjugate and the gauge boson propagator is of the gribov type , displaying complex poles . as usual , this can be interpreted as the confining region . in summary , we clearly notice that at sufficiently weak coupling , the standard higgs mechanism , eq .
, will definitely take place , as @xmath106 , whereas for sufficiently strong coupling , we always end up in a confining phase because then @xmath12 .
having obtained these results , it is instructive to go back where we originally started .
for a fundamental higgs , all gauge bosons acquire a mass that screens the propagator in the infrared .
this effect , combined with a sufficiently small coupling constant , will lead to a severely suppressed ghost self energy , i.e. the average of ( to be understood after renormalization of course ) . if the latter quantity will a priori not exceed the value of 1 under certain conditions , the theory is already well inside the gribov region and there is no need to implement the restriction . actually , the failure of the gribov restriction for @xmath106 is exactly because it is simply not possible to enforce that @xmath120 .
perturbation theory in the higgs sector is _ in se _ already consistent with the restriction within the 1st gribov horizon .
let us verify this explicitly by taking the average of with as tree level input propagator a transverse yukawa one with mass @xmath121 , cfr .
using that there are 3 transverse directions transverse polarizations in @xmath122 dimensions .
positive powers in @xmath123 can ( and will ) combine with the divergences in @xmath124 to change the finite terms . however , as already pointed out before , a careful renormalization analysis of the gribov restriction is possible , see e.g. @xcite and this will also reveal that the `` 1 '' in the gribov gap equation will receive finite renormalizations , compatible with the finite renormalization in e.g. @xmath125 , basically absorbable in the definition @xmath13 . the main results of our current paper thus remain correct and we leave the full renormalization details for later . ] in @xmath0 , we easily get @xmath126 introducing @xmath13 as in , we may reexpress the latter result as @xmath127 for @xmath106 , the logarithm is positive and it is then evident that @xmath125 will not cross @xmath128 , indicating that the theory already is well within the first gribov horizon .
another interesting remark is at place concerning the transition in terms of a varying value of @xmath13 .
if @xmath13 crosses @xmath129 , the imaginary part of the complex conjugate roots becomes smoothly zero , leaving us with 2 coinciding real roots , which then split when @xmath13 grows . at @xmath14 , one of the roots and its accompanying residue vanishes , to leave us with a single massive gauge boson .
we thus observe all these transitions are continuous , something which is in qualitative correspondence with the theoretical lattice predictions of the classic work @xcite for a fundamental higgs field that is `` frozen '' ( @xmath29 ) . concerning the somewhat strange intermediate phase , _ i.e. _ the one with a yukawa propagator with a negative residue , eq .
, we can investigate in future work in more detail the asymptotic spectrum based on the brst tools developed in @xcite when the local action formulation of the gribov restriction is implemented .
let us look at the vacuum energy @xmath75 of the system , which can be easily read off from expression , namely @xmath130 where @xmath71 is given by the gap equation .
making use of @xmath131it is very easy to write down the vacuum energy : * for @xmath12 , we have @xmath132 where @xmath107 is obtained through eqs . , .
let us also give the expressions of the first two derivatives of @xmath133 with respect to @xmath13 . from the gap equation , we easily get @xmath134 therefore @xmath135 & = & \frac{1}{64\pi^2 } \frac{1}{a } ( 2-\zeta ) \ ; , \nonumber \\ \frac { \partial^2 } { \partial a^2 } \left[\frac{8}{9 g^4\nu^4}\ ; { \cal e}_v \right ] & = & \frac{1}{64\pi^2 } \frac{1}{a^2 } \left(\zeta - 2 - \frac{2}{g'({\zeta } ) } \right)\;. \label{d12}\end{aligned}\ ] ] * for @xmath106 , @xmath136 owing to the fact that @xmath137 , it turns out that vacuum energy @xmath133 is a continuous function of the variable @xmath13 , as well as its first and second derivative .
the third derivative develops a jump at @xmath14 .
we might be tempted to interpret this is indicating a third order phase transition at @xmath14 .
the latter value actually corresponds to a line in the @xmath138 plane according to the functional relation .
however , we should be cautious to blindly interpret this value .
it is important to take a closer look at the validity of our results in the light of the made assumptions .
more precisely , we implemented the restriction to the horizon in a first order approximation , which can only be meaningful if the effective coupling constant is sufficiently small , while simultaneously emerging logarithms should be controlled as well . in the absence of propagating matter ,
the expansion parameter is provided by @xmath139 as in pure gauge theory .
the size of the logarithmic terms in the vacuum energy ( that ultimately defines the gap equations ) are set by @xmath140 and @xmath141 . a good choice for the renormalization scale would thus be @xmath142 : for ( positive ) real masses , a fortiori we have @xmath143 and the second log will not get excessively large either because @xmath144 gets small and the prefactor is thus small , or @xmath144 is of the order of @xmath145 and the log itself small . for complex conjugate masses , the size of the log
is set by the ( equal ) modulus of @xmath146 and thus both small by our choice of scale .
let us now consider the trustworthiness , if any , of the @xmath14 phase transition point . for @xmath147 , we already know that @xmath148 , so a perfect choice is @xmath149 .
doing so , the @xmath13-equation corresponds to @xmath150 so that @xmath151 . evidently , this number is thus far too big to associate any meaning to the `` phase transition '' at @xmath14 .
notice that there is no problem for the @xmath13 small and @xmath13 large region .
if @xmath152 is sufficiently large and we set @xmath153 we have a small @xmath154 , leading to a large @xmath13 , i.e. the weak coupling limit without gribov parameter and normal higgs - like physics .
the logs are also well - tempered . for a small @xmath152
, the choice @xmath155 will lead to @xmath156 so that a small @xmath13 can now be compatible with a small @xmath154 , leading to a gribov parameter dominating the higgs induced mass , the `` small number '' corresponds to @xmath157 . due to the choice of @xmath158 , the logs are again under control in this case . within the current approximation ,
we are thus forced to conclude that only for sufficiently small or large values of the parameter @xmath13 we can probe the theory in a controllable fashion .
nevertheless , this is sufficient to ensure the existence of a higgs - like phase at large higgs condensate , and a confinement - like region for small higgs condensate .
the intermediate @xmath13-region is more difficult to interpret due to the occurrence of large logs and/or effective coupling .
notice that this also might make the emergence of this double yukawa phase at @xmath159 not well established at this point .
let us face now the case in which the higgs field @xmath160 transforms according to the adjoint representation of @xmath1 .
for the action , we now have @xmath161 where the covariant derivative is defined by @xmath162 the vacuum configuration which minimizes the energy is achieved by a constant scalar field satisfying @xmath163 setting @xmath164 for the quadratic part of the action involving the gauge field @xmath165 , one gets @xmath166 from which one notices that the off - diagonal components @xmath167 acquire a mass @xmath168 . we again take @xmath29 . in order to implement the restriction to the gribov region @xmath7
, we start from the expression of the two point ghost function @xmath36 of eq .. in contrast with the case of the fundamental representation , in order to correctly take into account the presence of the higgs vacuum , eq .
, we must decompose @xmath36 into its diagonal and off - diagonal components , according to @xmath169 where @xmath170 \mathcal{g}_{diag}(k;a ) & = & \frac{1}{k^{2}}\left ( 1+g^{2}\frac{k_{\mu } k_{\nu } } { k^{2}}\int \frac{d^{4}q}{(2\pi ) ^{4}}\frac{1}{(q - k)^{2}}\left ( a_{\mu } ^{\alpha } ( q)a_{\nu } ^{\alpha } ( -q)\right ) \right ) \nonumber \\ &
\equiv & \frac{1}{k^{2}}\left ( 1+\sigma _ { diag}(k;a)\right ) \approx \frac{1}{k^{2}}\left ( \frac{1}{1-\sigma _ { diag}(k;a)}\right ) \label{gdiag } \;.\end{aligned}\]]the quantities @xmath171 turn out to be decreasing functions of the momentum @xmath43 @xcite .
thus , in the case of the adjoint representation , the no - pole condition for the ghost function @xmath172 is implemented by demanding that @xcite @xmath173where @xmath174 are given by @xmath175 where use has been made of eqs .
, , . to implement the restriction to the gribov region @xmath7 in the functional integral , we proceed as before and encode the no - pole conditions , eqs .
, into step functions @xcite , obtaining @xmath176\;e^{\beta ( 1-\sigma _ { diag}(0,a))}e^{\omega \left ( 1-\sigma _ { off}(0,a)\right ) } \nonumber \label{zq } \\ & \times & e^{-\frac{1}{4}\int d^{4}x(\partial _ { \mu } a_{\nu } ^{a}-\partial _ { \nu } a_{\mu } ^{a})^{2}-\frac{1}{2\xi } \int { d^{4}x(\partial _ { \mu } a_{\mu } ^{a})^{2}-}\frac{{g^{2}\nu ^{2}}}{2}\int { d^{4}xa_{\mu } ^{\alpha } a_{\mu } ^{\alpha } } } \;.\end{aligned}\ ] ] notice that , in the case of the adjoint representation , two gribov parameters @xmath177 are needed in order to implement conditions , see also the three dimensional case recently discussed in @xcite .
after simple algebraic manipulations , we get @xmath178[da^{3}]\;e^{-\frac{1}{2}\int \frac{d^{4}q}{(2\pi ) ^{4}}a_{\mu } ^{\alpha } ( q)\mathcal{p}_{\mu \nu } ^{\alpha \beta } a_{\nu } ^{\beta } ( -q)-\frac{1}{2}\int \frac{d^{4}q}{(2\pi ) ^{4}}a_{\mu } ^{3}(q)\mathcal{q}_{\mu \nu } a_{\nu } ^{3}(-q ) } , \label{zq1}\ ] ] with @xmath179 where again @xmath180 is understood to recover the landau gauge . evaluating the inverse of the expressions above and taking the limit @xmath61 , for the gluon propagators one gets @xmath181 to establish the gap equations for the gribov parameters @xmath177
, we evaluate the partition function @xmath64 in the semiclassical approximation . as done before , we integrate out the gauge fields , obtaining @xmath182 where the determinants in expression ( [ zq2 ] ) are given by @xmath183 \ ; , \nonumber \\
\left ( \det \mathcal{p}_{\mu \nu } ^{\alpha \beta } \right ) ^{-\frac{1}{2 } } & = & \exp \left [ -3\int { \frac{d^{4}q}{(2\pi ) ^{4}}\ln \left ( ( q^{2}+g^{2}\nu^{2})+g^{2}\left ( \frac{\beta } { 2}+\frac{\omega } { 4}\right ) \frac{1}{q^{2}}\right ) } \right ] \;.\end{aligned}\ ] ] therefore , @xmath184 where , in the thermodynamic limit @xcite @xmath185 we again proceed by evaluating expression ( [ zq3 ] ) in the saddle point approximation @xcite , _
i.e. _ @xmath186 where the parameters @xmath187 and @xmath188 are determined by the stationary conditions @xmath189 which yield the following gap equations : @xmath190 allowing us to express @xmath191 in terms of the parameters @xmath192 .
let us start by considering the second gap equation , eq .
, and decompose the denominator according to @xmath193 with @xmath194 let us discuss first the case of two real , positive , different roots , namely @xmath195 . from , we reexpress the gap equation as @xmath196 introducing now the dimensionless variables . ]
@xmath197 \xi & = & \frac{4\tau}{g^4\nu^4 } \geq 0\ ; , \qquad 0 \le \xi < 1 \ ; , \label{vb}\end{aligned}\ ] ] and proceeding as in the previous case , equation can be recast in the following form @xmath198 or compactly , @xmath199 with the same function as already defined in the fundamental sector , eqns . , .
also here , the equation remains valid also for complex conjugate roots , viz .
we are then easily lead to the following cases . using the properties of @xmath202
, it turns out that eq
. admits a unique solution for @xmath60 , which can be explicitly constructed with a numerical approach .
more precisely , when the mass scale @xmath168 is sufficiently smaller than @xmath203 , _
i.e. _ @xmath204 we have what can be called a @xmath15 confined phase .
in fact , in this case , the gap equations and can be rewritten as @xmath205 moreover , making use of @xmath206 the gap equation gives @xmath207 therefore , for @xmath201 , the @xmath16 component of the gauge field gets confined , exhibiting a gribov - type propagator with complex poles , namely @xmath208 relying on eq .
we can then distinguish the two phases : * when @xmath209 , equation has a single solution with @xmath210 . in this region ,
the roots @xmath211 are thus real and the off - diagonal propagator decomposes into the sum of two yukawa propagators @xmath212 where @xmath213 however , due to the relative minus sign in eq . , only the component @xmath214 can be associated to a physical mode , analogously as in the fundamental case . also here
, a further study using brst tools will be devoted to this . due to the confinement of the third component
@xmath16 , this phase is referred as the @xmath15 confining phase .
it is worth observing that it is also present in the @xmath215 case , with terminology coined in @xcite , see also @xcite . * for @xmath216 , equation has a solution for @xmath200 . in this region
the roots @xmath211 become complex conjugate and the off - diagonal gluon propagator is of the gribov type , displaying complex poles . in this region
all gauge fields display a propagator of the gribov type .
this is the @xmath1 confining region .
similarly , the above regions are continuously connected when @xmath217 varies . in particular , for @xmath218 , we obtain @xmath219 as solution .
let us consider now the case in which @xmath220 . here , there is no solution of the equation for the parameter @xmath60 , as it follows by observing that the left hand side of eq .
is always positive , while the right hand side is always negative .
this has a deep physical consequence .
it means that for a higgs mass @xmath221 sufficiently larger than @xmath203 , _
i.e. _ @xmath222 the gap equation is inconsistent .
it is then important to realize that this is actually the gap equation obtained by acting with @xmath223 on the vacuum energy @xmath224 , eq .. so , we are forced to set @xmath225 , and confront the remaining @xmath226-equation , viz .
@xmath227 which can be transformed into @xmath228\equiv h(\xi ) \label{gapddn}\end{aligned}\ ] ] after a little algebra , where @xmath229 .
the behaviour of @xmath230 for @xmath231 is more complicated than that of @xmath202 .
because of the @xmath232 contribution , @xmath230 becomes more and more positive when @xmath60 approaches zero .
in fact , @xmath230 strictly decreases from @xmath233 to @xmath234 for @xmath60 ranging from 0 to @xmath233 .
it is interesting to consider first the limiting case @xmath235 , yielding @xmath236 .
so , there is a discontinuous jump in @xmath60 ( i.e. the gribov parameter for fixed @xmath237 ) when the parameter @xmath217 crosses the boundary value @xmath113 .
we were able to separate the @xmath220 region as follows : * for @xmath238 , we have a unique solution @xmath200 , i.e. we are in the confining region again , with all gauge bosons displaying a gribov type of propagator with complex conjugate poles . * for @xmath239 , we have a unique solution @xmath240 , indicating again a combination of two yukawa modes for the off - diagonal gauge bosons .
the `` photon '' is still of the gribov type , thus confined .
completely analogous as in the fundamental case , it can be checked by addressing the averages of the expressions that for @xmath220 and @xmath226 obeying the gap equation with @xmath241 , we are already within the gribov horizon , making the introduction of the second gribov parameter @xmath242 obsolete .
it is obvious that the transitions in the adjoint case are far more intricate than in the earlier studied fundamental case .
first of all , we notice that the `` photon '' ( diagonal gauge boson ) is confined according to its gribov propagator .
there is never a coulomb phase for @xmath243 .
the latter finding is can be understood again from the viewpoint of the ghost self - energy . if the diagonal gluon would remain coulomb ( massless ) , the off - diagonal ghost self - energy , cfr .
eq . , will contain an untamed infrared contribution from this massless photon .
] , leading to an off - diagonal ghost self - energy that will cross the value 1 at a momentum @xmath244 , indicative of trespassing the first gribov horizon .
this crossing will not be prevented at any finite value of the higgs condensate @xmath4 , thus we are forced to impose at any time a nonvanishing gribov parameter @xmath226 . treating the gauge copy problem for the adjoint higgs sector will screen ( rather confine ) the a priori massless `` photon '' .
an interesting limiting case is that of infinite higgs condensate , also considered in the lattice study of @xcite .
assuming @xmath19 , we have @xmath245 according to its definition . expanding the gap equation around @xmath246 , we find the limiting equation @xmath247 , or equivalently @xmath248 .
said otherwise , we find that also the second gribov parameter vanishes in the limit of infinite higgs condensate . as a consequence ,
the photon becomes truly massless in this limit .
this result provides in our opinion a kind of continuum version of the existence of the coulomb phase in the same limit as in the lattice version of the model probed in @xcite .
it is instructive to link this back to the off - diagonal no pole function , see eq .
, as we have argued in the proceeding paragraph that the massless photon leads to @xmath249 upon taking averages .
however , there is an intricate combination of the limits @xmath19 , @xmath250 preventing such a problem here .
indeed , we find in these limits , again using dimensional regularization in the @xmath8 scheme , that @xmath251 the latter quantity is always smaller than @xmath128 for @xmath112 positive , meaning that we did not cross the gribov horizon .
this observation confirm in an explicit way the intuitive reasoning also found in section 3.4 of @xcite , at least in the limit @xmath19 .
the subtle point in the above analysis is that it is not allowed to naively throw away the 2nd integral in the first line of for @xmath19 .
there is a logarithmic @xmath252 ( @xmath19 ) divergence that conspires with the @xmath253 ( @xmath254 ) divergence of the 1st integral to yield the final reported result .
this displays that , as usual , certain care is needed when taking infinite mass limits in feynman integrals .
as done in the case of the fundamental representation , let us work out the expression of the vacuum energy @xmath75 , for which we have the one loop integral representation @xmath255 where @xmath256 based on and on @xmath257 we find @xmath258 we can introduce @xmath217 via its definition to write after simplification @xmath259 with @xmath260 first , for @xmath261 , we can use the @xmath226-gap equation to establish @xmath262 , and thus @xmath263 where @xmath60 is determined by the equations , .
for @xmath264 , we can remember that @xmath241 and thus @xmath265 , in which case we can easily obtain @xmath266 with @xmath60 now given by eq
.. it is worth noticing that the discontinuity in the parameter @xmath60 directly reflects itself in a discontinuity in the vacuum energy , as is clear from the plot of fig.[bs - d4 ] . .
the discontinuity at @xmath267 is evident .
, width=302 ] investigating the functional in terms of @xmath60 and @xmath268 , it is numerically ( graphically ) rapidly established there is always a solution to the gap equations @xmath269 for @xmath201 , but the solution @xmath270 is pushed towards the boundary @xmath219 if @xmath217 approaches @xmath113 , to subsequently disappear for @xmath220 . in that case , we are forced to return on our steps as in the fundamental case and conclude that @xmath241 , leaving us with a single variable @xmath265 and a new vacuum functional to extremize .
there is a priori no reason why these 2 intrinsically different vacuum functionals would be smoothly joined at @xmath267 .
this situation is clearly different from what happens when a potential has e.g. 2 different local minima with different energy , where at a first order transition the two minima both become global minima , thereafter changing their role of local vs. global .
evidently , the vacuum energy does not jump since it is by definition equal at the transition . nevertheless , a completely analogous analysis as for the fundamental case will learn that @xmath267 is beyond the range of validity of our approximation the gribov masses will dominate over the higgs condensate and we can take @xmath88 of the order of the gribov masses to control the logs and get a small coupling . for large @xmath217 , we have @xmath225 and a small @xmath188 : the first log will be kept small by its prefactor and the other logs can be managed by taking @xmath88 of the order of the higgs condensate . ] .
the small and large @xmath217 results can again be shown to be valid , so at large @xmath217 ( @xmath271 large higgs condensate ) we have a mixture of off - diagonal yukawa and confined diagonal modes and at small @xmath217 ( @xmath271 small higgs condensate ) we are in a confined phase .
in any case we have that the diagonal gauge boson is _ not _ coulomb - like , its infrared behaviour is suppressed as it feels the presence of the gribov horizon .
in this work we have attempted at studying the transition between the higgs and the confinement phase within a continuum quantum field theory .
the problem has been addressed by restricting the domain of integration in the functional integral to the so called gribov region @xmath7 , which enables us to take into account the nonperturbative effect of the gribov copies .
this framework allows us to discuss the transition between the higgs phase and the confinement phase by looking at the pole structure of the two - point gluon correlation function .
both fundamental and adjoint representation for the higgs field have been considered .
the output of our investigation reveals that the case of the fundamental representation is different from that of the adjoint representation , a feature in agreement with the results of numerical lattice simulations @xcite . in the case of the fundamental representation
, the gluon propagator evolves in a continuous way from a confining propagator of the gribov type to a yukawa type propagator describing the higgs phase .
again , this feature is in qualitative agreement with lattice results @xcite , which show that the transition between the higgs and the confining phase occurs in a continuous way .
moreover , we have been able to show that , in the weak coupling region , _
i.e. _ in the higgs phase , there is no need to implement the restriction to the gribov region @xmath7 .
said otherwise , in this region , the presence of the higgs field automatically ensures that the theory lies within the gribov region , so that the gribov horizon is not crossed .
this is a relevant result , implying that , at weak coupling , the usual higgs mechanism takes place , being not affected by the existence of the gribov copies .
a safely asymptotic non - confining theory can be introduced , with massive gauge bosons as asymptotic states . in the adjoint representation things
look quite different . besides the confining phase , in which the gluon propagator is of the gribov type , our results indicate the existence of what can be called a @xmath15 confining phase for finite values of the higgs condensate .
this is a phase in which the third component @xmath16 of the gauge field displays a propagator of the gribov type , while the remaining off - diagonal components @xmath272 , @xmath18 , exhibit a propagator of the yukawa type .
interestingly , this phase has been already detected in the lattice studies of the three dimensional georgi - glashow model @xcite .
a second result of our analysis is the absence of the coulomb phase for finite higgs condensate . for an infinite value of the latter
, we were able to clearly reveal the existence of a massless photon , in agreement with the lattice suggestion of @xcite .
summarizing , it seems safe to state that the results we have obtained so far can be regarded as being in qualitative agreement with the lattice findings and worth to be pursued .
let us end by giving a preliminary list of points for future investigation : * in the pure gauge case , it has been shown in past years that the gribov theory dynamically corrects itself via the condensation of the auxiliary fields arising when the restriction to the gribov region is implemented in a local and renormalizable way .
this has led to the refined gribov - zwanziger ( rgz ) theory @xcite , resulting in propagators and dynamics in very good agreement with lattice investigations @xcite .
it would be worth to discuss the transition between the higgs and confining phase within the rgz framework
. this might give more reliable quantitative results to be compared with the lattice data .
* within the present approximation , we were not able to address in a concrete and detailed way the issue of the characterization of the phase diagram in the @xmath5 plane . to that aim
, it might be interesting to embed in our framework the polyakov loop @xmath6 as order parameter of the higgs - confinement transition , as studied in @xcite . in @xcite ,
the pure gauge thermodynamics at finite @xmath93 was investigated by using the rgz gluon and ghost propagators .
in particular , an approximate effective action for @xmath273 was constructed .
we can try to couple @xmath273 to the gribov effective action and investigate the interplay of both gribov mass and @xmath273 in terms of a varying higgs condensate @xmath168 , the latter quantity playing the role of temperature as in @xcite .
* certainly , the extension of the present investigation to the case of the gauge group @xmath274 @xcite has an apparent interest , due to its relationship with the electroweak theory .
since the latter is based on a fundamental higgs and since we have shown in this paper that in the `` perturbative regime '' ( sufficiently large higgs condensate and small effective coupling constant ) in the fundamental case the gribov dynamics becomes trivial , we can expect to recover at least a massless photon for the qed part of the @xmath274 gauge theory in the same region of ( physically relevant ) parameter space .
our preliminary findings @xcite do support this naive extrapolation of the here presented work .
the concrete analysis is however rather cumbersome , details are thus to be reported at a later stage in @xcite . *
finally , we hope that our results will motivate further lattice investigations on the behavior of the gluon and ghost propagators in presence of higgs fields .
although these studies have been already started in the case of the fundamental representation @xcite , it would be quite interesting to have at our disposal also data for the case of the adjoint representation .
the conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq - brazil ) , the faperj , fundao de amparo pesquisa do estado do rio de janeiro , the latin american center for physics ( claf ) , the sr2-uerj , the coordenao de aperfeioamento de pessoal de nvel superior ( capes ) are gratefully acknowledged .
d. d. is supported by the research - foundation flanders .
we thank .
olejnk for discussions .
e. h. fradkin and s. h. shenker , phys .
d * 19 * ( 1979 ) 3682 .
c. b. lang , c. rebbi and m. virasoro , phys .
b * 104 * ( 1981 ) 294 .
w. langguth , i. montvay and p. weisz , nucl .
b * 277 * ( 1986 ) 11 .
v. azcoiti , g. di carlo , a. f. grillo , a. cruz and a. tarancon , phys .
b * 200 * ( 1988 ) 529 .
w. caudy and j. greensite , phys .
d * 78 * ( 2008 ) 025018 [ arxiv:0712.0999 [ hep - lat ] ] . c. bonati , g. cossu , m. delia and a. di giacomo , nucl .
b * 828 * ( 2010 ) 390 [ arxiv:0911.1721 [ hep - lat ] ] .
a. maas , eur .
j. c * 71 * ( 2011 ) 1548 [ arxiv:1007.0729 [ hep - lat ] ] .
a. maas and t. mufti , arxiv:1211.5301 [ hep - lat ] .
a. m. horowitz , nucl .
b * 235 * ( 1984 ) 563 .
p. h. damgaard and u. m. heller , phys .
b * 164 * ( 1985 ) 121 .
r. baier and h. j. reusch , nucl .
b * 285 * ( 1987 ) 535 .
l. fister , r. alkofer and k. schwenzer , phys .
b * 688 * ( 2010 ) 237 [ arxiv:1003.1668 [ hep - th ] ] .
v. macher , a. maas and r. alkofer , int .
j. mod .
a * 27 * ( 2012 ) 1250098 [ arxiv:1106.5381 [ hep - ph ] ] . k. osterwalder and e. seiler , annals phys .
* 110 * ( 1978 ) 440 .
t. d. lee and c. n. yang , phys .
* 87 * ( 1952 ) 410 .
v. n. gribov , nucl .
b * 139 * ( 1978 ) 1 .
r. f. sobreiro and s. p. sorella , arxiv : hep - th/0504095 . n. vandersickel and d. zwanziger , phys .
* 520 * ( 2012 ) 175 [ arxiv:1202.1491 [ hep - th ] ] .
m. a. l. capri , d. dudal , a. j. gomez , m. s. guimaraes , i. f. justo and s. p. sorella , arxiv:1210.4734 [ hep - th ] .
s. nadkarni , nucl .
b * 334 * ( 1990 ) 559 .
a. hart , o. philipsen , j. d. stack and m. teper , phys .
b * 396 * ( 1997 ) 217 [ hep - lat/9612021 ] .
r. c. brower , d. a. kessler , t. schalk , h. levine and m. nauenberg , phys .
d * 25 * ( 1982 ) 3319 .
j. greensite , s. olejnik and d. zwanziger , phys .
d * 69 * ( 2004 ) 074506 [ hep - lat/0401003 ] .
d. dudal , s. p. sorella and n. vandersickel , eur .
j. c * 68 * ( 2010 ) 283 [ arxiv:1001.3103 [ hep - th ] ] .
d. j. gross and f. wilczek , phys .
d * 8 * ( 1973 ) 3633 .
a. g. m. pickering , j. a. gracey and d. r. t. jones , phys .
b * 510 * ( 2001 ) 347 [ erratum - ibid .
b * 535 * , 377 ( 2002 ) ] [ hep - ph/0104247 ] .
d. dudal and s. p. sorella , phys .
d * 86 * ( 2012 ) 045005 [ arxiv:1205.3934 [ hep - th ] ] .
d. dudal , s. p. sorella , n. vandersickel and h. verschelde , phys .
d * 77 * ( 2008 ) 071501 [ arxiv:0711.4496 [ hep - th ] ] .
d. dudal , j. a. gracey , s. p. sorella , n. vandersickel and h. verschelde , phys .
d * 78 * ( 2008 ) 065047 [ arxiv:0806.4348 [ hep - th ] ] .
d. dudal , s. p. sorella and n. vandersickel , phys .
d * 84 * ( 2011 ) 065039 [ arxiv:1105.3371 [ hep - th ] ] .
j. a. gracey , phys .
d * 82 * ( 2010 ) 085032 [ arxiv:1009.3889 [ hep - th ] ] .
d. dudal , o. oliveira and n. vandersickel , phys .
d * 81 * ( 2010 ) 074505 [ arxiv:1002.2374 [ hep - lat ] ] .
a. cucchieri , d. dudal , t. mendes and n. vandersickel , phys .
d * 85 * ( 2012 ) 094513 [ arxiv:1111.2327 [ hep - lat ] ] .
o. oliveira and p. j. silva , arxiv:1207.3029 [ hep - lat ] . | we present an analytical study of continuum @xmath0 @xmath1 gauge higgs models with a single higgs field with fixed length in either the fundamental or adjoint representation .
we aim at analytically probing the renowned predictions of fradkin & shenker on the phase diagram in terms of confinement versus higgs behaviour , obtained for the lattice version of the model .
we work in the landau version of the t hooft @xmath2 gauges in which case we can access potential nonperturbative physics related to the existence of the gribov copies . in the fundamental case , we clearly show that in the perturbative regime of small gauge coupling constant @xmath3 and large higgs vacuum expectation value @xmath4 , there is a higgs phase with yukawa gauge boson propagators without gribov effects . for a small value of the higgs _ vev _ @xmath4 and/or large @xmath3 , we enter a region with gribov type propagators that have no physical particle interpretation : the gauge bosons are as such confined .
the transition between both behaviours is found to be continuous . in the adjoint case , we find evidence of a more drastic transition between the different behaviours for the propagator of the off - diagonal gauge bosons , whereas the `` photon '' , _
i.e. _ the diagonal component of the gauge field , displays a propagator of the gribov type . in the limit of infinite higgs condensate , we show that a massless photon is recovered .
we compare our findings with those of fradkin & shenker as well as with more recent numerical lattice simulations of the fundamental higgs model .
we also carefully discuss in which region of the parameter space @xmath5 our approximations are trustworthy . |
the decay of @xmath6 charmonium states , like @xmath2 and @xmath3 , to @xmath7 is a cp violating process . however , since cp violations in both the @xmath8 and @xmath9 systems have been well established @xcite , it is of interest to search for them in other possible processes .
furthermore , it has been shown @xcite that the @xmath7 system can be used to test the einstein - podolsky - rosen ( epr ) paradox versus quantum theory .
the space - like , separated coherent quantum system may yield some @xmath0 decays if epr s locality is correct while quantum theory forbids this process .
mark - iii searched for decays of @xmath0 with 2.7 million @xmath2 events @xcite .
no signal was observed , and the upper limit on the decay branching ratio was determined to be @xmath10 at the 90% confidence level ( c. l. ) .
the same search for @xmath3 decays has not been performed before . in this letter , we report on a search for @xmath0 using a sample with 20 times more statistics than before , and on the first search for @xmath1 .
the data samples used for the analyses are taken with the beijing spectrometer ( besii ) detector at the beijing spectrometer ( bepc ) storage ring at a center - of - mass energies corresponding to @xmath11 and @xmath12 .
the data samples contain @xmath13 @xmath2 events and @xmath14 @xmath3 events , as determined from inclusive hadronic decays @xcite .
bes is a conventional solenoidal magnet detector that is described in detail in ref .
@xcite ; besii is the upgraded version of the detector @xcite . a 12-layer vertex chamber ( vc ) surrounding the beam pipe
provides trigger information .
a forty - layer main drift chamber ( mdc ) , located radially outside the vc , provides trajectory and energy loss ( @xmath15 ) information for charged tracks over @xmath16 of the total solid angle .
the momentum resolution is @xmath17 ( @xmath18 in @xmath19 ) , and the @xmath15 resolution for hadron tracks is @xmath20 .
an array of 48 scintillation counters surrounding the mdc measures the time - of - flight ( tof ) of charged tracks with a resolution of @xmath21 ps for hadrons .
radially outside the tof system is a 12 radiation length , lead - gas barrel shower counter ( bsc ) .
this measures the energies of electrons and photons over @xmath22 of the total solid angle with an energy resolution of @xmath23 ( @xmath24 in gev ) .
outside of the solenoidal coil , which provides a 0.4 tesla magnetic field over the tracking volume , is an iron flux return that is instrumented with three double layers of counters that identify muons of momentum greater than 0.5 gev/@xmath25 .
a monte carlo ( mc ) simulation is used for the determination of mass resolutions and detection efficiencies . for the signal channels , @xmath0 and @xmath1 ,
the angular distribution of one @xmath26 is generated with the same distribution as for the cp allowed channels @xmath27 and @xmath28 , namely as @xmath29 , where @xmath30 is the polar angle of the @xmath26 in the laboratory system .
only @xmath31 is generated , and the @xmath26 is allowed to decay in the detector according to its lifetime . the simulation of the detector response is done using a geant3 based program with detailed consideration of the detector performance ( such as dead electronic channels ) .
reasonable agreement between data and monte carlo simulation has been observed in various channels tested , including @xmath32 , @xmath33 , @xmath34 and @xmath35 @xmath36 . for this study ,
10,000 events for each signal channel are produced .
for the decay channels of interest , the candidate events are required to satisfy the following selection criteria : 1 .
the number of charged tracks is required to be four with net charge zero .
each track should satisfy @xmath37 , where @xmath30 is the polar angle in the mdc , and should have a good helix fit so that the error matrix of the track fitting is available for secondary vertex finding .
the tracks are assumed to be either @xmath38 or @xmath39 .
the higher momentum positive track and the lower momentum negative track are assumed to come from one of the @xmath26 decays , and the remaining two tracks are assumed to come from the other @xmath26 decay .
the intersection of the track - pairs near the interaction point are determined and are regarded as the @xmath26 vertices .
3 . each @xmath26 candidate is required to have a decay length in the transverse plane ( @xmath40 ) greater than 3 mm and an invariant mass within twice the mass resolution as estimated from monte carlo simulations ( the mass resolution in @xmath2 decays is 7.1 mev/@xmath41 and in @xmath3 decays 7.9 mev/@xmath41 ) . for @xmath3 decays , backgrounds from @xmath42 with @xmath2 decaying into two charged tracks
are removed by further requiring the mass recoiling against the lower momentum positive and negative tracks , @xmath43 , to be less than 3.0 gev/@xmath41 .
the comparison of the recoil mass distributions of signal and background events from monte carlo simulation is shown in figure [ mrec ] .
it can be seen that this requirement is very efficient in removing background , and the loss of efficiency is small ( @xmath44 4% ) . another possible background is from @xmath45 @xmath46 with @xmath47 .
figure [ pmiss ] shows @xmath7 missing momentum distributions for the signal and background channels , all simulated by monte carlo .
we make an additional requirement @xmath48 gev/@xmath25 for @xmath3 selection .
this removes 94% of @xmath49 background and almost all the @xmath50 background , while the efficiency for @xmath1 is about 92% according to monte carlo simulation . for @xmath2
decays , since @xmath51 is forbidden by parity conservation , no requirement is applied .
the momenta of the two @xmath26 tracks for events passing the above selection are plotted in figures [ pkspksj ] and [ pkspks ] , for @xmath2 and @xmath3 decays , respectively , together with the regions predicted for signal events by the monte carlo simulation .
the two circles correspond to @xmath52 where @xmath53 and @xmath54 are the momenta of the two @xmath26 candidates in the event ; @xmath55 is the @xmath26 momentum resolution , which is 28 mev/@xmath25 in @xmath2 decays and 34 mev/@xmath25 in @xmath3 decays , as determined from monte carlo simulations ; and @xmath56 is the expected @xmath26 momentum in @xmath2 ( 1.466 gev ) or @xmath3 ( 1.775 gev ) two - body decays .
the inner circles correspond to @xmath57 , and the outer circles are for @xmath58 .
it can be seen that there are no candidate events in either @xmath2 or @xmath3 decays within the one @xmath55 circle , and there is only one candidate in each case lying between the one and two @xmath55 circles . using the 2@xmath55 circle for our event selection
, one event is obtained each for @xmath0 and for @xmath1 .
the corresponding efficiencies , @xmath59 , are 20.74% and 19.18% for @xmath2 and @xmath3 decays , respectively , as estimated from monte carlo samples .
the backgrounds remaining in both @xmath2 and @xmath3 decays after the @xmath26 requirement are studied with monte carlo simulations .
the main background channel in @xmath2 decays is from @xmath60 is estimated to be @xmath61 events , which is obtained using a monte carlo sample six times as large as the @xmath2 sample .
the main backgrounds channels in @xmath3 decays are @xmath62 , @xmath63 and @xmath64 . a monte carlo simulation estimates the total background to be @xmath65 events , with about 80% due to @xmath63 .
the errors on the estimated backgrounds are from the uncertainties of the branching ratios used and the limited statistics of the monte carlo samples .
the two candidate events are investigated further , and it is found that the @xmath0 candidate is most likely a @xmath60 , with @xmath66 misidentified as @xmath31 because no particle identification is used in the event selection , while the @xmath1 candidate looks like a @xmath62 , @xmath63 event with an isolated photon with bsc energy of 126 mev .
due to the long decay length of the high momentum @xmath26 and the requirement of hits in the vc , the trigger efficiency of @xmath7 events is lower than for normal hadronic events .
since the trigger system is not included in the monte carlo simulation , the trigger efficiency is measured with data by comparing events within and beyond the outer radius of the vc using @xmath67 events , which yields trigger efficiencies of @xmath68 and @xmath69 for a single @xmath26 in @xmath2 and @xmath3 decays , respectively . for @xmath7 events ,
the trigger efficiency , @xmath70 , is calculated to be @xmath71% in @xmath2 decays and @xmath72% in @xmath3 decays .
the secondary vertex finding algorithm is checked using @xmath73 events .
it is found that the monte carlo simulates data fairly well , and extrapolating the difference between data and monte carlo simulation to the @xmath26 momentum range under study and correcting by the polar angle dependence of the efficiency , correction factors of @xmath74 and @xmath75 are obtained for a single @xmath26 in @xmath2 and @xmath3 decays , respectively . for @xmath7 decays , the correction factors and errors to the monte carlo efficiencies , @xmath76 , for @xmath0 and @xmath1 are @xmath77 and @xmath78 , respectively .
the total systematic error on the branching ratio measurement comes from all sources listed in table .
the simulation of the tracking efficiency agrees with data within 1 - 2% for each charged track as measured using channels like @xmath79 and @xmath80 , @xmath81 , and 8% is taken conservatively as the systematic error for the channel of interest .
.summary of the systematic errors . [ cols="<,^,^",options="header " , ] [ br ]
the cp violating processes @xmath0 and @xmath1 are searched for using the besii 58 million @xmath2 event and 14 million @xmath3 event samples . the upper limits on the branching ratios are determined to be @xmath82 and @xmath83 at the 95% c. l. the former is much more stringent than the previous mark - iii measurement @xcite , and the latter is the first search for this channel in @xmath3 decays .
current bounds of the production rates still beyond the sensitivity for testing the epr paradox @xcite .
we wish to thank dr .
h. b. li for helpful comments .
the bes collaboration thanks the staff of bepc for their hard efforts .
this work is supported in part by the national natural science foundation of china under contracts nos .
19991480 , 10225524 , 10225525 , the chinese academy of sciences under contract no .
kj 95t-03 , the 100 talents program of cas under contract nos .
u-11 , u-24 , u-25 , and the knowledge innovation project of cas under contract nos .
u-602 , u-34(ihep ) ; by the national natural science foundation of china under contract no .
10175060 ( ustc ) ; and by the department of energy under contract no .
de - fg03 - 94er40833 ( u hawaii ) . a summary on the cp violation in kaon and b decays can be found in k. hagiwara _ et al .
_ ( particle data group ) , phys . rev .
* d66 * , 010001 - 118 ( 2002 ) , and references therein . m. roos , `` test of einstein locality '' , hu - tft-80 - 5 ( revised ) , nov .
1980 , and references there in . | the cp violating processes @xmath0 and @xmath1 are searched for using samples of 58 million @xmath2 and 14 million @xmath3 events collected with the beijing spectrometer at the beijing electron positron collider .
no signal is observed , and upper limits on the decay branching ratios are determined to be @xmath4 and @xmath5 at the 95% confidence level . |
the astrophysical observations of supernovae of type ia give a convincing evidence of a universe undergoing an accelerated phase of its expansion history ; rather than going to deceleration as was expected theoretically @xcite .
similar conclusion has been deduced from the observations of anisotropies in cosmic microwave background by the wmap @xcite which favors a low density , spatially flat ( @xmath4 ) universe filled with an exotic vacuum energy containing the maximum portion of the total energy density , @xmath5 @xcite .
this mysterious dark energy is represented by a barotropic equation of state ( eos ) @xmath6 , where @xmath7 and @xmath8 is the pressure and the energy density of dark energy , with @xmath9 .
if @xmath3 , it is called ` cosmological constant ' , @xmath10 is dubbed ` quintessence ' , while it is called ` phantom energy ' if @xmath11 @xcite .
all these candidates have some fundamental problems : the former one presents discrepancy of @xmath12 orders of magnitude between theoretical and empirical results @xcite , while quintessence requires fine tuning of cosmological parameters for a suitable choice of the potential function @xcite . also the phantom energy yields eccentric predictions like ` big rip ' and ripping apart of gravitationally bound objects @xcite .
moreover , the variations in @xmath13 suggest that there is no consensus on the actual eos of dark energy and one can only deal with the upper and lower bounds on @xmath13 @xcite . although the dark energy is generically considered to be a perfect fluid yet it need not to be perfect if it experiences perturbations @xcite .
moreover , the dark energy also may not be completely ` dark ' especially if it gets coupled with matter and energy exchange takes place ; consequently the matter evolution becomes modified @xcite .
it is now obvious that this sudden transition to acceleration from the earlier deceleration phase is rather recent with the corresponding redshift @xmath14 @xcite .
it has been proposed that the standard model of cosmology may not be sufficient to explain this exotic phenomenon and hence significant modifications are proposed like a modified friedmann equation @xmath15 , where @xmath16 is an arbitrary function of @xmath8 and @xmath17 is the hubble parameter @xcite and adding a cardassian term in the friedmann equation @xcite .
other possible explanations proposed are dark energy arising from tachyonic matter @xcite , van der waals fluid @xcite , geometric dark energy @xcite , dvali - gabadadze - porrati ( dgp ) gravity model @xcite and randall - sundrum brane world model @xcite are most prominent .
although the nature of dark energy is not clear but its thermodynamical properties suggest a universe filled with it becomes hotter with time @xcite .
one of the most interesting problems in the present cosmology is the cosmic coincidence ( or cosmic conundrum ) problem which naively asks : why the energy densities of matter and dark energy are of the same order or the corresponding dimensionless ratio is closer to 1 , at current time @xcite .
this problem can be posed in terms of the eos parameter @xmath13 : if the parameter @xmath18 in the past and @xmath11 in the future then why we are observing @xmath3 at present time . in recent years , this problem is addressed using a notion of interacting dark energy model in which both the interacting components i.e. dark energy and matter exchange energy to keep the density ratio close to 1 .
this model has some intrinsic problems that are still unresolved : the microphysics of energy transfer is not exactly understood i.e. the particles that can mediate the interaction are not pointed out .
moreover , the coupling function ( or the decay rate ) for the required interaction is chosen quite arbitrarily @xcite and also the coupling constant involved is not yet properly constrained theoretically or observationally @xcite . in this paper , we address this problem using a simplified approach by considering the constants of physics to evolve over cosmic time .
we here take three ansatz for scale factor and analyze the behavior of parameter @xmath13 .
curiously , the parameter @xmath19 evolves from positive to negative values and finally converges to @xmath20 .
this result turns out to be consistent with the observations .
hence our model fairly addresses the cosmic coincidence problem and practically alleviates it .
due to multitude of uncertainties in the determination of @xmath13 observationally and other intrinsic theoretical problems ( as discussed above ) with it , we here proceed with an eos commonly called the chaplygin gas ( cg ) represented by @xcite @xmath21 where @xmath22 is a constant parameter .
the cg effectively explains the evolution of the universe from the earlier deceleration ( matter dominated era ) to the later acceleration phase ( dark energy dominated ) as is manifested in the following equation @xmath23 here @xmath24 is a constant of integration parameter . for small @xmath25 ,
it gives @xmath26 while for large @xmath25 , we have @xmath27 .
therefore models based on cg are also called dark energy - matter unification models @xcite . due to its this effectiveness ,
several generalizations of cg are proposed ( see e.g. @xcite ) .
the chaplygin gas arises from the dynamics of a generalized @xmath28-brane in a ( @xmath29 ) spacetime and can be described by a complex scalar field whose action can be written as a generalized born - infeld action @xcite .
the plan of the paper is as follows : in the second section , we shall present the model of our system . in third section ,
we determine the cosmological parameters for different choices of the scale factor parameter .
the last section is devoted for the conclusion of our paper .
the friedmann - robertson - walker ( frw ) metric which satisfies the cosmological principle is specified by @xmath30.\ ] ] here @xmath31 is the scale factor that determines the expansion of the universe . also the parameter @xmath32 is the curvature parameter determining the spatial geometry of the frw spacetime . it can take three possible values @xmath33 which correspond to spatially closed , flat and open universe respectively or geometrically spherical , minkowskian and hyperbolic spacetime respectively .
the equations of motion corresponding to frw metric are @xmath34 above @xmath17 is the hubble parameter .
note that we have assumed @xmath35 in the above equations which is favored by the observational data . here
@xmath0 is called the cosmological constant with dimensions of @xmath36 .
note that eq .
( 5 ) shows that accelerated expansion of the universe @xmath37 is possible if the strong energy condition @xmath38 is violated and also it is independent of the choice of @xmath32 .
the energy conservation equation for the above system is @xmath39 the cherished constants of physics that describe the universe need not to be constant but can vary with respect to other parameters .
for instance , the cosmological constant @xmath40 which is constant at zeroth order approximation but is really a time dependent function at higher order approximations .
note that accelerated expansion of the universe follows from @xmath41 .
the cosmic history of @xmath0 shows that it was large in the past while it is small at present and will continue to decrease , hence it gives a parametrization @xmath42 , @xmath43 and @xmath44 @xcite .
a variable cosmological constant also arises in theories of higher spatial dimensions like string theory and manifests itself as the energy density for the vacuum @xcite and it can also addresses the cosmic age problem effectively @xcite .
similarly , there is some evidence of a varying newton s constant @xmath1 : observations of hulse - taylor binary pulsar b@xmath45 gives a following estimate @xmath46 @xcite , helioseismological data gives the bound @xmath47 @xcite ( see ref @xcite for various bounds on @xmath48 from observational data ) . the variability in @xmath1 results in the emission of gravitational waves .
dimensional analysis also shows that the time dependent parameter @xmath0 to be decreasing with time @xmath49 @xcite . in another approach
, it is shown that @xmath1 can be oscillatory with time @xcite .
it is recently proposed that variable cosmic constants are coupled to each other i.e. variation in one leads to changes in others @xcite .
a variable gravitational constant also explains the dark matter problem as well @xcite .
also discrepancies in the value of hubble parameter can be removed with the consideration of variable @xmath1 @xcite . due to these reasons
, we shall take @xmath0 and @xmath1 to be time dependent quantities i.e. @xmath50 and @xmath51 .
hence eqs .
( 4 ) and ( 5 ) yield @xmath52 using eqs . ( 6 ) and ( 7 ) , we can write @xmath53 taking the ansatz for cosmological constant as @xcite @xmath54 where @xmath55 and @xmath56 are constant parameters .
note that this is a general ansatz and can reduce to chakraborty and debnath @xcite if @xmath57 .
using the modified chaplygin gas ( mcg ) eos given by @xcite @xmath58 where @xmath59 and @xmath60 are constant parameters and @xmath61 .
thermodynamical analysis of mcg show that the values @xmath62 and @xmath63 are consistent with the phenomenological results @xcite .
it is also shown that the recent supernovae data favors @xmath64 values @xcite .
the mcg reduces to generalized chaplygin gas ( gcg ) if @xmath65 while it gives cg if further @xmath66 . while a barotropic eos is obtained if @xmath67 .
( 10 ) is a combination of a barotropic and gcg eos .
precisely , the observations of cosmic microwave background gives the constraint @xmath68 at @xmath69 confidence level @xcite .
analysis of various cosmological models show that models based on chaplygin gas best fit with supernova data @xcite . using eq .
( 10 ) in ( 6 ) , we get the density evolution of mcg as @xmath70 where @xmath71 and @xmath72 making use of eqs .
( 9 ) and ( 11 ) in ( 8) , the parameter @xmath1 is determined to be @xmath73,\ ] ] where @xmath74 the parameter @xmath0 can alternatively be written as @xmath75
to analyze the behavior of the above cosmological parameters , we will consider three cases : 1 .
@xmath76 2 .
@xmath77 , 3 .
@xmath78^{-\frac{1}{p}}.$ ] here @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 and @xmath85 are constant parameters .
also @xmath86 is the dimensionless time parameter with @xmath87 is the current age of the universe .
we consider power law form of the scale factor @xmath88 where @xmath85 and @xmath84 are arbitrary constants . for this choice ,
it is possible to get the accelerated expansion of the universe if @xmath89 .
now , all the physical parameters will take the following forms as : @xmath90,\\ \lambda&\simeq&3\beta\left(x+\frac{c}{(a_ot^n)^{y}}\right)^{\frac{-\gamma}{1+\alpha}},\\ p&\simeq&a\left[\left ( x+\frac{c}{(a_ot^n)^{y } } \right)^{\frac{1}{1+\alpha}}\right]-\frac{b}{\left[\left ( x+\frac{c}{(a_ot^n)^{y } } \right)^{\frac{1}{1+\alpha}}\right]^\alpha}.\end{aligned}\ ] ] the cosmological parameters obtained in this section are plotted in figures 1 to 4 against parameter @xmath91 .
16 ) is plotted against time @xmath91 .
model parameters are fixed at @xmath92 , @xmath93 , @xmath94 and @xmath95 .
, title="fig : " ] + ( see eq .
17 ) is plotted against time parameter @xmath91 for different choices of @xmath96 .
model parameters are fixed as taken in fig.1,title="fig : " ] + ( see eq .
18 ) is plotted against time parameter @xmath91 .
model parameters are fixed as taken in fig.1,title="fig : " ] + ( see eq . 16 and 19 ) is plotted against the time parameter @xmath91 .
model parameters are fixed as taken in fig.1,title="fig : " ] + in this case , we consider constant deceleration parameter model defined by @xmath97 here the constant is taken to be negative i.e. it is an accelerating model of the universe @xcite .
the solution of equation ( 20 ) is @xmath98 where , @xmath82 and @xmath83 are integration constants .
this equation implies , the condition of expansion is @xmath99 .
now , all the physical parameters will take the following forms as : @xmath100,\\ \lambda&\simeq&3\beta\left(x+\frac{c } { ( ut - v)^{\frac{y}{(1+q ) } } } \right)^{\frac{-\gamma}{1+\alpha}},\\ p&\simeq&a\left[\left ( x+\frac{c } { ( ut - v)^{\frac{y}{(1+q ) } } } \right)^{\frac{1}{1+\alpha } } \right]-\frac{b}{\left[\left ( x+\frac{c } { ( ut - v)^{\frac{y}{(1+q ) } } } \right)^{\frac{1}{1+\alpha}}\right]^\alpha}.\end{aligned}\ ] ] the cosmological parameters obtained in this section are plotted in figures 5 to 8 against parameter @xmath91 .
22 ) is plotted against time @xmath91 .
model parameters are fixed at @xmath101 , @xmath102 , @xmath103 , @xmath104 and @xmath95.,title="fig : " ] + ( see eq .
23 ) is plotted against time parameter @xmath91 for different choices of @xmath96 .
other model parameters are fixed as in fig.5,title="fig : " ] + ( see eq .
24 ) is plotted against time parameter @xmath91 .
other model parameters are fixed as in fig.5,title="fig : " ] + ( see eq . 22 and 25 ) is plotted against the time parameter @xmath91 .
other model parameters are fixed as in fig.5,title="fig : " ] + in this case , we consider specific form deceleration parameter model defined by @xcite @xmath105 here @xmath80 is is a constant .
this choice of deceleration parameter provides an early deceleration and late time acceleration of the universe .
the solution of equation ( 26 ) is @xmath106^{-\frac{1}{p}},\ ] ] where @xmath79 is an integration constant . for negative values of @xmath80 ,
we always get accelerated expansion of the universe .
now , all the physical parameters will take the following forms as : @xmath107^{-\frac{y}{p } } } \right)^{\frac{1}{1+\alpha}},\\ g&\simeq&\frac{-3\beta\gamma c}{8\pi ( \alpha+\gamma+2 ) } [ \left(x+\frac{c } { [ e^{-dpt } - 1 ] ^{-\frac{y}{p}}}\right)^{-\delta}\left(1+x\frac { [ e^{-dpt } - 1 ] ^{-\frac{y}{p}}}{c}\right)^\delta\nonumber\\ & \;&\times { _ { 2}f_{1}}\left(\delta,\delta,1+\delta ,- x\frac { [ e^{-dpt } - 1 ] ^{-\frac{y}{p}}}{c } \right)],\\ \lambda&\simeq&3\beta \left ( x+\frac{c } { [ e^{-dpt } - 1 ] ^{-\frac{y}{p } } } \right)^{\frac{-\gamma}{1+\alpha}},\\ p&\simeq&a\left ( x+\frac{c } { [ e^{-dpt } - 1 ] ^{-\frac{y}{p } } } \right)^{\frac{1}{1+\alpha}}-\frac{b}{\left[\left ( x+\frac{c } { [ e^{-dpt } - 1 ] ^{-\frac{y}{p } } } \right)^{\frac{1}{1+\alpha}}\right]^\alpha}.\end{aligned}\ ] ] the cosmological parameters obtained in this section are plotted in figures 9 to 12 against parameter @xmath91 .
28 ) is plotted against time @xmath91 .
model parameters are fixed at @xmath101 , @xmath108 , @xmath109 and @xmath110,title="fig : " ] + ( see eq .
29 ) is plotted against time parameter @xmath91 for different choices of @xmath96 .
other model parameters are fixed as in fig.9,title="fig : " ] + ( see eq .
30 ) is plotted against time parameter @xmath91 .
other model parameters are fixed as in fig.9,title="fig : " ] + ( see eq .
28 and 31 ) is plotted against the time parameter @xmath91 .
other model parameters are fixed as in fig.9,title="fig : " ] +
in this section , we discuss the results of our paper .
all the cosmological parameters with the exception of @xmath13 are plotted in figures in 1 to 12 in logarithmic scale against dimensionless time parameter @xmath91 .
the parameters in section 3.1 are shown in figures 1 to 4 .
the cosmological energy density decreases with time and then remains constant in far future .
the cosmological constant was large in the past which resulted in inflation while now it is small to produce current accelerated expansion .
the newton s gravitational constant steadily increased which caused structures to form .
also the dimensionless parameter @xmath13 varies from the positive to negative values and converging to @xmath20 at current time @xmath111 or @xmath112 showing that @xmath13 is inherently evolving over cosmic history , with @xmath113 corresponds to the big bang epoch ( units are chosen to be meter , kilogram , sec ) .
a similar behavior is obtained for parameters of section 3.2 and 3.3 shown in figures 5 to 8 and 9 to 12 , respectively . in figures ( 13 ) , ( 14 ) and ( 15 ) , we have plotted the same parameters against redshift @xmath114 .
the problem attempted in this paper can also be looked in the context of bulk viscous cosmology .
the anisotropic stresses can be important at large scale and hence they should be incorporated in the mcg equation of state ( see @xcite for the basic formalism ) .
it would also be interesting to extend our model using the modified @xmath115 gravity theory as well @xcite .
99 s. perlmutter et al , ap .
j. 517 ( 1999 ) 565 .
riess et al , ast .
j. 116 ( 1998 ) 1009 .
tonry et al , ap
. j. 594 ( 2003 ) 1 .
y. wang and m. tegmark , phys .
92 ( 2004 ) 241302 .
spergel et al , ap .
j. supp . 148
( 2003 ) 175 .
bennett et al , ap .
j. supp . 148
( 2003 ) 1 .
s. perlmutter et al , phys .
83 ( 1999 ) 670 .
caldwell , phys .
b 545 ( 2002 ) 23 . r.r .
caldwell et al , phys .
d 73 ( 2006 )
023513 . v. sahni , astro - ph/0403324 . i. zlatev et al , phys .
. lett . 82 ( 1999 ) 896 .
e. babichev et al , phys .
93 ( 2004 ) 021102 .
m. jamil et al , eur .
j. c 58 ( 2008 ) 325 ; astro - ph/0808.1152v1 a. upadhye , nuc . phys .
b ( proc . suppl . ) 173 ( 2007 ) 11 . j.c .
fabris and j. martin , phys .
d 55 ( 1997 ) 55 .
linder , astro - ph/0610173v2 .
s. nojiri and s.d .
odintsov , hep - th/0505215v4 . r.a . daly and s.g .
djogovski , ap .
j. 597 ( 2003 ) 9 .
y. gong and c.k .
duan , mon . not .
r. astron .
( 2004 ) 847 .
k. freese and m. lewis , phys .
b 540 ( 2002 ) 1 .
t. padmanabhan , phys .
d 66 ( 2002 ) 021301 .
gonzalez - diaz , hep - th/0408225v1 g.m .
kremer , phys .
d 68 ( 2003 ) 123507 . e.v .
linder , new astron .
49 ( 2005 ) 93 .
dvali et al , phys .
b 485 ( 2000 ) 208 .
l. randall and r. sundrum , phys .
83 ( 1999 ) 4690 .
lima and j.s .
alcaniz , phys .
b 600 ( 2004 ) 191 .
s. del campo et al , phys .
d 78 ( 2008 ) 021302(r ) .
n. dalal et al , phys .
87 ( 2001 ) 141302 .
s. dodelson et al , phys .
85 ( 2000 ) 5276 .
m. quartin et al , astro - ph/0802.0546 m. jamil and m.a .
rashid , eur .
j. c 58 ( 2008 ) 111 ; astro - ph/0802.1146v3 m. jamil and m.a .
rashid , eur .
j. c 56 ( 2008 ) 429 ; astro - ph/0803.3036v3 m. jamil , gr - qc/0810.2896v2 m. jamil and m.a .
rashid , astro - ph/0802.1144v3 a. dev et al , phys .
d 67 ( 2003 ) 023515 .
a. kamenshchik et al , phys .
b 511 ( 2001 ) 265 .
n. bili@xmath116 et al , phys .
b 535 ( 2002 ) 17 . m.c .
bento et al , phys .
d 70 ( 2004 ) 083519 .
s. chattopadhyay and u. denath , gr - qc/0805.0070v1 .
u. debnath , gr - qc/0710.1708v1 .
w. chakraborty , gr - qc/0711.0079v1 m.r .
setare , phys .
b 648 ( 2007 ) 329 . m.r .
setare , phys .
b 654 ( 2007 ) 1 . z.k .
guo and y.z .
zong , astro - ph/0506091v3 .
a.a . sen and r.j .
scherrer , phys .
d 72 ( 2005 ) 063511 .
h. zhang and z.h .
zhu , astro - ph/0704.3121v2 .
m.c . bento et al , phys .
d 66 ( 2002 ) 043507 .
u. mukhopadhay et al , gr - qc/0711.4800v1 .
fomin et al , gr - qc/0509042v1 .
h. liu and p.s .
wesson , gr - qc/0107093v1 . s. ray and u. mukhopadhyay , astro - ph/0411257v2 .
kogan , gr - qc/0511072v3 .
guenther , phys .
b 498 ( 1998 ) 871 . s. ray and u. mukhopadhyay , astro - ph/0510549v1 .
j.a . belinchon and i. chakrabarty , gr - qc/0404046v1 .
belinchon , gr - qc/809.2412v1 .
a. pradhan et al , gr - qc/0608107v1 .
vishwakarma , gr - qc/0801.2973v1 .
i. goldman , phys .
b 281 ( 1992 ) 219 .
o. bertolami et al , phys .
b 311 ( 1993 ) 27 . a.i .
arbab , hep - th/0711.1465v1 .
w. chakraborty and u. debnath , gr - qc/0705.4147v1 . h. jing et al , chin .
lett . 25 ( 2008 ) 347 .
bedran et al , phys .
b 659 ( 2008 ) 462 . o. bertolami et al , mon . not .
( 2004 ) 329 .
bento et al , phys
d 71 ( 2005 ) 063501 .
liu and x.z .
li , astro - ph/0501115v1 .
g. panotopoulos , phys .
d 77 ( 2008 ) 107303 .
f. rahaman et al , astrophys .
space sci .
299 ( 2005 ) 211 . n. banerjee and s. das , astro - ph/0505121 .
mak et al , gr - qc/0110119v1 .
p.s . denath and b.c .
paul , gr - qc/0508031v2 . | the standard model of cosmology is investigated using time - dependent cosmological constant @xmath0 and newton gravitational constant @xmath1 .
the total energy content is described by the modified chaplygin gas equation of state .
it is found that the time - dependent constants coupled with the modified chaplygin gas interpolate between the earlier matter to the later dark energy dominated phase of the universe .
we also achieve a convergence of the parameter @xmath2 , almost at the present time . thus our model fairly alleviates the cosmic - coincidence problem which demands @xmath3 at the present time .
_ keywords _ : chaplygin gas ; coincidence problem ; cosmological constant ; dark energy ; newton s gravitational constant . |
the low - temperature physics of amorphous and disordered solids has been a subject of great interest for more than four decades .
below about @xmath0k the acoustic and thermodynamic properties of a large variety of glasses are not only qualitatively different compared to their crystalline counterparts , but show a remarkable degree of universality.@xcite this universal behavior was explained by the existence of low - energy excitations with two - level structure , known as tunneling two - level systems ( tlss).@xcite in the standard tunneling model ( stm ) , each tls is characterized by the energy bias @xmath1 between its two basis states , the tunnel splitting @xmath2 between both , and its coupling @xmath3 to the strain field @xmath4 .
thus the hamiltonian of each tls in the stm is@xcite @xmath5 where @xmath6 and @xmath7 are the pauli matrices in the local states basis .
the central idea of the stm is that the tlss are independent and their energy bias and tunnel splitting are randomly distributed , with a universal distribution function @xmath8 , where @xmath9 is a material dependent constant.@xcite the role of the coupling to the strain field [ third term in eq .
( [ eq : stm ] ) ] is threefold .
first , via this channel the tls couples to phonons , which gives rise to the energy relaxation of the tls .
second , by applying mechanical stress externally , one can control and manipulate the tls.@xcite third , this coupling is at least partially responsible for the coupling between the tlss . although the stm has originally considered independent tlss , it is now well established that tlss interact via elastic ( phonon - mediated ) or electric ( photon - mediated ) dipole interactions,@xcite described by the low - energy effective hamiltonian @xmath10 where @xmath11 and @xmath12 are the pauli matrices that represent the tlss at sites @xmath13 and @xmath14 , and the interaction coefficients are @xmath15 the parameter @xmath16 in general is a complicated function of the polar angle between the tlss at sites @xmath13 and @xmath14 , but frequently can be treated as a normal - distributed random variable , @xmath17 is the distance between the tlss , and @xmath18 characterizes the specific interaction . here we will focus on elastic interactions for which @xmath19 where @xmath20 and @xmath21 are the mass density and sound velocity , respectively .
our results can be straightforwardly generalized to the case where electric interactions are present . with the advent of superconducting quantum bits ( qubits ) ,
the investigation of the low - temperature properties of amorphous solids has gained further interest@xcite for two main reasons .
first , ensembles of tlss are believed to be a major source of fluctuations leading to energy relaxation and decoherence in superconducting qubits and microwave resonators .
specifically , the coherence times of superconducting qubits incorporating josephson junctions were found to be strongly affected by the presence of tlss located in the tunnel barriers of the junctions @xcite which are typically made of amorphous aluminum oxide with a thickness of 2 - 3 nm
. a deeper understanding of the nature of tlss and their role as sources of decoherence is necessary in order to achieve high - fidelity operation of superconducting qubits .
second , it was established that superconducting phase qubits can be utilized to study the properties of individual tlss.@xcite the qubit spectrum exhibits an avoided level crossing at bias values for which the qubit is in resonance with a certain tls . in addition
this resonant coupling allows one to manipulate the quantum state of the tls directly by coherent single - pulse resonant driving @xcite as well as to study the longitudinal ( @xmath22 ) and transverse ( @xmath23 ) relaxation times of individual tlss.@xcite the individual tlss that are being probed and manipulated by qubits are characterized by the level splitting of the order of that of the qubit ( @xmath24ghz ) , which is substantially higher than the temperature in the experiment . to distinguish the ( high - frequency ,
probed ) tls from , e.g. , the tlss with energy splittings smaller or of the order of the temperature ( thermal tlss ) , we introduce for the former the subscript @xmath25 and rewrite its hamiltonian ( [ eq : stm ] ) as@xcite @xmath26 the bias @xmath27 is controlled by the applied strain @xmath28 as written in eq .
( [ eq : stm ] ) , i.e. @xmath29 .
recently , a static strain tuning of the energy bias of individual tlss in a superconducting phase qubit was demonstrated.@xcite the results support the assumption of the stm that the coupling of tlss to strain fields occurs mainly through a change of the asymmetry bias @xmath27 , whereas changes of @xmath30 are negligible . in ref .
the strain dependence of the relaxation and coherence times of individual high - frequency tlss was measured , revealing the following interesting features : 1 .
the pure dephasing rate @xmath31 of the probed tls , as extracted from the ramsey ( free induction ) protocol , is linear with respect to the asymmetry bias @xmath32 in the vicinity of the symmetry point @xmath33 . outside the window @xmath34ghz around the symmetry
point,@xcite the ramsey dephasing rate deviates from the linear behavior , exhibiting small changes of slope or irregularities . 2 .
the pure dephasing rate @xmath35 , as extracted from the hahn - echo protocol , is much smaller than the ramsey dephasing rate ( with @xmath36 at @xmath37ghz ) , demonstrating the efficiency of the echo protocol in filtering noise at frequencies @xmath38 .
furthermore , the echo dephasing rate grows quadratically with @xmath39 in the whole measured range of @xmath39 .
this implies the presence of a white - noise environment inducing energy fluctuations on a time scale @xmath40 .
the starting point of the theoretical description is the following : the individual high - frequency probed tls with energy splitting @xmath41 suffers from decoherence due to its interaction with an ensemble of thermal tlss with energy splittings lower than the temperature . such thermal tlss undergo random transitions ( telegraph noise ) between their two eigenstates , thereby causing fluctuations in the energy splitting @xmath42 of the probed high - frequency tls .
this is the mechanism of spectral diffusion observed in hole - burning experiments .
@xcite as we will show in the following this picture needs to be refined in several directions .
in this paper we provide a detailed theoretical analysis of the coherence properties of an individual high - frequency tls coupled to an ensemble of thermal tlss .
the qubit used to probe serves merely as a tool and is not part of the modeling . after providing a general theoretical description of the spectral diffusion we analyze the decoherence caused by an ensemble of thermal tlss with parameters taken from the stm ( we call these tlss `` @xmath43-tlss '' ) .
a short description of this theory was provided in ref . .
we show that by assuming elastic dipole - dipole interactions as in eqs .
( [ eq : int1])-([eq : int3 ] ) , with a bath of thermal @xmath43-tlss , one may explain very well both the qualitative dependence on strain as well as the order of magnitude of the ramsey dephasing rate .
this demonstrates that the tlss introduced in the stm form the main low - frequency noise source in superconducting phase qubits .
moreover , the suggested theory predicts the temperature dependence of the ramsey dephasing rate ( at a fixed strain ) . for the echo dephasing rate , on the other hand
, the theory at this level predicts that the contribution of @xmath43-tlss is orders of magnitude smaller than the experimental result .
in addition , the strain dependence of the echo dephasing rate predicted by the theory is in disagreement with the experiment .
we further show that the experimental results on the echo dephasing rate could be potentially explained by the presence of a few fast thermal tlss.@xcite these tlss are characterized by parameters very different from those of the stm ( @xmath43-tlss ) .
in particular they have a much stronger coupling to the strain , which leads to the enhanced relaxation ( switching ) rate . from the point of view of the probed tls
the fast thermal fluctuators produce gaussian noise , since their switching rate is larger than their coupling to the probed tls .
this noise is white at relevant frequencies ( frequencies lower than the typical switching rate of the fast thermal tlss ) .
this explains the quadratic dependence of the echo dephasing rate on the applied strain .
such strongly interacting tlss ( henceforth denoted as @xmath44-tlss ) were recently introduced within a model suggesting a possible explanation for the universality of acoustic and thermodynamic properties in disordered systems at low temperatures.@xcite we estimate the order of magnitude for the contribution of the @xmath44-tlss to the echo dephasing rate and discuss its relevance in accounting for the experimental findings .
we emphasize that in all our considerations we assume the probed tls to be a standard ( @xmath43-)tls , since these are the prevalent tlss at low temperatures .
the paper is organized as follows : in sec .
[ sec 1 ] we first define the spin - fluctuator model,@xcite and then describe its results for the ramsey and echo decay signals of the probed tls in different regimes of relaxation rates and tls - tls couplings ( sec .
[ sec 1a ] ) .
we then discuss the important distinction between average and typical results , which arises in situations in which a quantity is dominated by contributions from a small number of fluctuators ( sec .
[ sec 1b ] ) .
the model is applied in sec .
[ sec 2 ] to the case of an environment consisting of weakly interacting tlss as postulated in the stm .
we first describe the characteristic relaxation rates and tls - tls couplings of such tlss ( sec . [ sec 2a ] ) , and then derive the results for the ramsey ( sec .
[ sec 2b ] ) and echo ( sec.[sec 2c ] ) dephasing rates . in sec .
[ sec 3 ] we generalize the model to the model with two types of tlss introduced in ref . .
we first review the two - tlss model ( sec .
[ sec 3a ] ) , estimate the characteristic relaxation rates and tls - tls couplings of strongly interacting @xmath44-tlss ( sec .
[ sec 3b ] ) , and then study the contribution of @xmath44-tlss to the echo dephasing rate ( sec .
[ sec 3c ] ) .
our conclusions are discussed and summarized in sec .
[ sec 5 ] .
we consider a single high - frequency probed tls interacting with a set of thermal tlss via the hamiltonian ( [ eq : int1 ] ) .
thus , the hamiltonian reads @xmath45 the first term of eq .
( [ eq : hpj ] ) describes the probed tls .
the second term describes the set of thermal tlss . the coupling constants @xmath46 denote the coupling strength ( [ eq : int2 ] ) between the thermal tls at site @xmath14 and the probed tls .
the term @xmath47 describes the internal interactions between the thermal tlss , as well as their coupling to phonons . in the eigenbasis of the tlss
, the system is described by the hamiltonian @xmath48 here @xmath49 are the pauli matrices in the eigenbasis of the probed tls and @xmath50 are the pauli matrices in the eigenbasis of the thermal tls at site @xmath14 .
the first term of eq .
( [ eq:3 ] ) describes the probed tls with energy @xmath51 and @xmath52 , @xmath53 .
all these are strain - dependent via @xmath54 .
the second term describes the set of thermal tlss with energies @xmath55 , and the third term describes the interactions between the probed tls and the set of thermal tlss , with the operator @xmath56 given by @xmath57 here @xmath58 and @xmath59 .
the operator @xmath56 fluctuates since each thermal tls , i.e. tls for which @xmath60 , switches randomly between its two eigenstates .
such random transitions are driven by the emission or absorption of phonons,@xcite or by internal interactions between the tlss.@xcite the fluctuations in @xmath56 add a random contribution to the bare energy @xmath61 of the probed tls through the longitudinal coupling ( @xmath62 ) in the third term of eq .
( [ eq:3 ] ) .
this results in a dephasing of the probed tls .
henceforth , we will simplify the model by making the following assumptions : 1 .
we consider random telegraph noise of thermally excited tlss due to relaxation ( @xmath22 ) processes .
tlss with energy larger than the temperature are frozen in their ground state and are thus irrelevant for the dephasing of the probed tls .
we will thus restrict the sum in eq .
( [ eq:4 ] ) to thermal tlss .
further , since the probed tls and the thermal tlss are non - resonant , we can neglect the transverse term ( @xmath63 ) of @xmath56 in eq .
( [ eq:4 ] ) .
2 . we will treat the environment - induced noise classically , i.e. , we replace the operator @xmath50 by the classical random variable @xmath64 ( sometimes called a fluctuator ) which switches between @xmath65 .
accordingly , @xmath66 is the corresponding sum ( [ eq:4 ] ) of such random variables .
such a classical treatment is allowed provided that the frequencies of the random transitions , @xmath67 , are much smaller than the temperature @xmath68 and the probed tls energy splitting @xmath69.@xcite having modelled the environment of tlss as a classical noise , we now consider the simplified model @xmath70 with @xmath71 here @xmath72 is the effective coupling between the thermal tls at site @xmath14 and the probed tls , given by @xmath73 with @xmath74 . this model is often referred to as the spin - fluctuator model.@xcite let us now consider the effect of the classical noise ( [ eq:6 ] ) , which consists of the environment of thermal tlss , on the ramsey and echo decay signals.@xcite the decay of a ramsey signal is given by @xmath75 , where @xmath76 is the random phase accumulated at time @xmath77 , @xmath78 in an echo experiment , the phase acquired is the difference between the two free evolution periods , @xmath79 and the corresponding decay function is @xmath80 . since we consider thermal tlss with @xmath60 , the transition rates in both directions can be assumed to be equal . for a single fluctuator , characterized by switching rate @xmath81 and coupling @xmath21 , an average over the switching history yields@xcite @xmath82\ , , \end{aligned}\ ] ] where @xmath83 .
the decay function due to a set of fluctuators is the product of individual contributions of the form ( [ eq:10 ] ) , that is @xmath84 .
it is useful to present the main limits of eqs .
( [ eq:10 ] ) , as shown in table [ tbl ] .
.[tbl ] the different limits of the ramsey and echo decay signals , eqs .
( [ eq:10 ] ) , due to a single fluctuator , characterized by switching rate @xmath81 and coupling @xmath21 . [ cols="^,^,^,^,^ " , ] 0.3 cm one can clearly see here the effect of the echo protocol . in the short time
limit , @xmath85 and @xmath86 , the initial parabolic decay @xmath87 is replaced by the much slower decay @xmath88 . in this regime ,
the decay due to a single fluctuator is small for both ramsey and echo protocols , but having many fluctuators might make this decay relevant ( see below ) . for the regime @xmath89 and @xmath90 the oscillating entry @xmath91
$ ] requires a separate discussion .
these oscillations were discovered in ref . as an example of a prominent non - gaussian effect .
the origin of the oscillations is the interference between two `` paths '' in which the equally probable @xmath92 state of the fluctuator never changes during a single experimental run .
therefore , these oscillations are almost completely removed by the echo protocol , as can be seen in table [ tbl ] .
if several fluctuators in this regime contribute , the ramsey decay function will be a product of several incommensurate oscillating functions and , thus , will be strongly suppressed . as we will see below ,
this is the regime of the ramsey decay , which occurs at times longer than the typical dephasing times . finally , in the regime @xmath93 and @xmath94 the echo protocol is inefficient and the result coincides with the one given by the golden rule ( see more details in sec .
[ sec 3c ] ) .
there are three parameters that control the thermal tlss : i ) the energy @xmath95 , ii ) the normalized tunnel splitting @xmath74 , and iii ) the distance to the probed tls @xmath96 . in a concrete sample at a given strain
we have a unique realization of the parameters @xmath97 characterizing the tlss , and the total ramsey / echo decay signal is given by @xmath98 we first estimate @xmath99 , where @xmath100 denotes the averaging over the disorder configurations of the tlss .
since we are dealing with quenched disorder , the question of self - averaging arises and one should examine if the result for @xmath99 corresponds to a typical situation.@xcite we obtain @xmath101 where @xmath102 is the spatial dimension of the amorphous tunnel barrier and @xmath103 is derived from the standard tunnelling model distribution function @xmath104 using the relations @xmath105 and @xmath106 ( see ref . ) . in eq .
( [ eq : lnfgeneral ] ) @xmath107 are given by eqs .
( [ eq:10 ] ) and @xmath108 is a lower cutoff for the parameter @xmath109 . it should be emphasized that @xmath110 depends on @xmath69 and @xmath109 [ see eq .
( [ eq:18 ] ) ] , whereas @xmath111 depends on @xmath112 and @xmath109 [ see eq .
( [ eq:7 ] ) ] . according to table
[ tbl ] , it is convenient to divide the integration into three domains : a ) @xmath113 and @xmath114 ; b ) @xmath89 and @xmath115 ; c ) @xmath116 and @xmath117 , as illustrated in fig .
[ fig : fig1 ] .
thus , for the ramsey decay signal we obtain @xmath118\nonumber\\ & + \int d\omega\int^{\infty}_{r^{}_{0 } } r^{d-1}dr\int^{1}_{u^{}_{\mathrm{min}}}du\int^{t}_{0}de\ , p(e , u ) \theta(\gamma t-1)\theta(\gamma - v)\,\frac{v^{2}t}{\gamma}\ , .\end{aligned}\ ] ] for the echo decay this would give @xmath119 in the next section we use these expressions to study the average ramsey and echo decay signals that arise from the coupling of the probed tls to thermal @xmath43-tlss .
we show that for the ramsey dephasing there is no self - averaging , i.e. the average result differs markedly from the typical one . for the echo dephasing , on other hand ,
the result is self - averaging and the dephasing rate calculated from eq .
( [ eq:15 ] ) coincides with the typical case .
( color online ) the different domains in the @xmath21-@xmath81 plane corresponding to the various limits of a ramsey and echo decay signals due to a single fluctuator , as described in table [ tbl ] .
the general expressions ( [ eq:14 ] ) and ( [ eq:15 ] ) for the averaged ramsey and echo decay signals over an ensemble of tlss are obtained by dividing the integrations into the various domains and setting the appropriate limits for @xmath120 in each domain .
the shaded blue region shows the relevant domain for @xmath43-tlss , @xmath121.,scaledwidth=52.0% ]
according to table [ tbl ] , the effect of an environment of thermal tlss on the coherence properties of the probed tls depends on the typical couplings and relaxation ( switching ) rates of the thermal tlss . to estimate the relaxation rates of @xmath43-tlss
, we assume that the random transitions of each tls are mainly due to its coupling to phonons
. we will thus neglect interactions between the tlss comprising the noisy environment .
additional relaxation mechanisms possible in superconducting circuits , such as relaxation by quasiparticle excitations , are negligible at @xmath122mk.@xcite therefore , the rate of random transitions of a single tls is the relaxation rate due to one - phonon processes,@xcite @xmath123 where @xmath124 and @xmath125 are the sound velocities of the longitudinal and transverse modes , respectively . for a given energy splitting @xmath69 ,
the maximum relaxation rate is obtained for @xmath126 . moreover , for thermal tlss with @xmath127 , eq .
( [ eq:16 ] ) yields the maximum relaxation rate @xmath128 the relaxation rate ( [ eq:16 ] ) for thermal tlss ( @xmath129 ) can be written in terms of @xmath69 and @xmath109 as @xmath130 where we replaced @xmath131 in eq .
( [ eq:16 ] ) by @xmath132 and used eq .
( [ eq:17 ] ) . to estimate the maximum relaxation rate for @xmath43-tlss , we assume @xmath133ev.@xcite furthermore , since the dielectric layer thickness is much smaller than the relevant phonon wavelength , the sound velocity in eqs .
( [ eq:16 ] ) and ( [ eq:17 ] ) is set by the aluminum layers of the electrodes.@xcite at @xmath134mk ( the temperature corresponding to the measurements of ref . )
we obtain @xmath135 or equivalently , @xmath136ms .
we next estimate the typical coupling between the probed tls and its nearest thermal tls . the typical distance between the probed tls and its nearest thermal @xmath43-tls , @xmath137 ,
can be estimated using the relations @xmath138 where @xmath108 is a lower cutoff for the parameter @xmath109 and @xmath139 is the tunnel barrier thickness ( @xmath140 nm ) . here
we provide the estimates in three and two dimensions ( 3d and 2d , respectively ) due to the quasi-2d structure of the amorphous tunnel barrier . in the framework of the stm
, one usually assumes a constant dos , namely @xmath9 in eq .
( [ eq:13 ] ) is a constant .
the central dimensionless parameter of the stm is the tunneling strength @xmath141,@xcite where @xmath142 and @xmath143 are the typical distance and typical interaction between nearest neighbors @xmath43-tlss , respectively .
substituting eq .
( [ eq:13 ] ) into eq .
( [ eq:20 ] ) , we readily obtain @xmath144 with @xmath145 . inserting these distances in the interaction coefficients [ eqs .
( [ eq : int2 ] ) and ( [ eq : int3 ] ) ] , we find the typical coupling strength between the probed tls and its nearest thermal @xmath43-tls , @xmath146 where we assumed the usual values @xmath147 and @xmath148 .
for the 2d estimate we also used @xmath149 and @xmath150ghz . as shown in ref .
, the 2d estimate of @xmath151 is in good agreement with the observed magnitude of the ramsey dephasing rate ( see more details below ) , whereas the 3d estimate yields somewhat larger rates than observed experimentally .
this agrees with the quasi-2d structure of the junction barrier and suggests that tlss reside only in the amorphous tunnel barrier .
comparison of eqs .
( [ eq:19 ] ) and ( [ eq:21 ] ) reveals that thermal @xmath43-tlss satisfy @xmath152 .
we recall eq .
( [ eq:7 ] ) and take into account that typically , @xmath153 .
therefore , in a typical sample the most significant thermal @xmath43-tlss ( i.e. those located at @xmath154 ) are characterized by @xmath155 , except in the very close vicinity of the symmetry point of the probed tls , @xmath33 ( hence @xmath156 ) , which can not be distinguished within the resolution of @xmath39 in the experiment .
therefore , in the case of @xmath43-tlss we should study the @xmath39-dependence of the dephasing rates assuming the presence of fluctuators which are slow and strongly coupled to the probed tls , as sketched schematically in fig . [
fig : fig2 ] .
( color online ) schematic view of tlss ( circles ) in the aluminum oxide tunnel barrier ( of width @xmath139 ) of a josephson junction .
the probed tls is shown in blue and non - thermal tlss are shown in black .
thermal tlss are denoted in red and the typical distance @xmath137 between the probed tls and its nearest thermal tls is shown by a green arrow .
thermal tlss are typically in the regime @xmath155 .
as discussed in the text , ramsey dephasing is dominated by few thermal tlss located at @xmath157 whereas echo dephasing is caused by a large number of thermal tlss.,scaledwidth=40.0% ] as explained above , the contribution of @xmath43-tlss to the ramsey and echo decay signals should be treated within the limit @xmath155 . under these circumstances
the third line of eqs .
( [ eq:14 ] ) and ( [ eq:15 ] ) can be disregarded .
furthermore , we examine now the regime of purely static dephasing , @xmath158 , which is the relevant regime for the ramsey dephasing studied in ref . .
since the observed dephasing times @xmath159 are on the order of microseconds , the condition @xmath158 indeed holds at the relevant times @xmath160 . in this case the average ramsey decay signal due to @xmath43-tlss can be estimated as @xmath161 where @xmath162 [ see eqs .
( [ eq : int2 ] ) , ( [ eq : int3 ] ) , ( [ eq:7 ] ) ] . using eq .
( [ eq:13 ] ) and introducing an auxiliary variable @xmath163 we obtain for @xmath164 @xmath165 where @xmath166 . setting @xmath150ghz ,
we observe that at all relevant times ( of the order of the dephasing time @xmath167 ) @xmath168 , except in a very small vicinity of the symmetry point @xmath169 .
thus we take the limit @xmath170 in eq .
( [ eq:23 ] ) .
the integral over @xmath171 converges , whereas integration over @xmath109 gives a result of the order of @xmath145 .
thus we obtain ( for @xmath164 ) @xmath172 a similar calculation for @xmath173 gives @xmath174^{2/3}\ , t^{2/3}\ .\end{aligned}\ ] ] the coupling strength between the probed tls and its closest thermal tls , @xmath151 , is defined in eqs .
( [ eq:21 ] ) for both @xmath164 and @xmath173 . the results ( [ eq : lnfav3d ] ) and ( [ eq : lnfav2d ] ) turn out to be non - self - averaging .
indeed , in a typical situation there is a closest thermal tls with maximum coupling @xmath175 . at very short times , @xmath176 ,
one obtains @xmath177 . since @xmath178 the sum is dominated by the few closest thermal tlss and in both @xmath164 and @xmath173 @xmath179 ^ 2\ , t^2\ .\ ] ] both the average [ eqs .
( [ eq : lnfav3d ] ) and ( [ eq : lnfav2d ] ) ] and the typical [ eq .
[ eq : lnftyp ] ) ] results give the dephasing rate@xcite @xmath180 yet with a very different functional time dependence . at short times , @xmath181
, the typical result gives a much weaker ramsey decay than would be naively expected from the average result . at longer times , @xmath182 , the difference between the average and the typical results is even more striking .
the typical decay function @xmath183 at such times oscillates between positive and negative values,@xcite which means that the envelope of ramsey fringes has points of zero amplitude where phase slips occur .
in contrast , the average result is monotonically decaying .
this situation is demonstrated in fig .
[ fig : fig3 ] .
( color online ) comparison of @xmath184 obtained by averaging over an ensemble of slow fluctuators with @xmath185 as given by eq .
( [ eq : lnfav3d ] ) ( dashed - dotted blue line ) , and of @xmath186 for a specific realization of six thermal fluctuators with couplings @xmath187mhz and relaxation rates @xmath188 ( solid black line ) . at short times , @xmath189s ,
the ramsey decay is well approximated by a typical gaussian decay as given in eq .
( [ eq : lnftyp ] ) ( dashed red line).,scaledwidth=52.0% ] as demonstrated experimentally in refs . and
, the energy bias of the probed tls changes linearly with an external strain ( which , in turn , is proportional to a piezo voltage @xmath190 ) .
therefore , close to the symmetry point @xmath33 , we can write @xmath52 as @xmath191 combining eqs .
( [ eq:25 ] ) and ( [ eq:26 ] ) , one finds @xmath192 , i.e. the ramsey dephasing rate is expected to change linearly with @xmath39 . however , this linear dependence is expected to hold as long as the ramsey dephasing is dominated by the same decohering thermal tlss .
as the strain is varied the asymmetry bias @xmath193 of the decohering tlss will change as well , similarly to the probed tls . as a result , at some point the energy of a decohering tls will exceed the thermal energy . at the same time ,
other tlss which were non - thermal at lower strains will become thermal . assuming that thermal tlss are coupled to the strain similarly as the probed tls
, one expects that once @xmath39 is changed by @xmath194ghz , the dominant decohering tlss will be different ones .
this gives rise to a non - regular behavior , reflected in a change of slope or small irregularities in the ramsey dephasing rate as a function of strain .
such features are indeed observed in the @xmath39-dependence of the ramsey dephasing rate@xcite and are strongly suggestive of the predominance of single thermal tlss in the ramsey dephasing . as a result
, we expect the pure linear behavior to be relevant at a window @xmath195ghz around the symmetry point of the probed tls . indeed , for the four tlss studied in ref .
, the estimate for @xmath151 in 2d is in excellent agreement with the fitted slope of the linear line in the region @xmath195ghz .
equations ( [ eq:21 ] ) and ( [ eq:25 ] ) also allow us to predict the temperature dependence ( for low temperatures , @xmath122mk ) of the ramsey dephasing rate . at a fixed strain
, these equations predict @xmath196 for the 3d case and @xmath197 for the 2d case . for an energy - dependent dos of the form @xmath198,@xcite the corresponding temperature dependence is @xmath199 and @xmath200 , for 3d and 2d , respectively .
to estimate the contribution of @xmath43-tlss to the ensemble average of the echo decay signal , we have to calculate the first two integrals of eq .
( [ eq:15 ] ) . for simplicity
, we perform the integration in 3d .
the results in 2d are qualitatively similar and given in appendix [ sec appa ] .
the contribution to the first integral comes from tlss for which @xmath201 . since for @xmath43-tlss @xmath155 ,
the condition @xmath202 is automatically satisfied .
hence , the contribution to this integral comes from tlss which are located at distances @xmath203 from the probed tls , where @xmath204 . therefore , @xmath205 where @xmath151 is given by the first of eqs .
( [ eq:21 ] ) . the main contribution to
the second integral comes from the close tlss , for which @xmath206 .
again , the condition @xmath155 for @xmath43-tlss ensures that the second step function is always unity .
thus , @xmath207 neglecting a numerical prefactor of order unity , the sum of eqs .
( [ eq:27 ] ) and ( [ eq:28 ] ) gives @xmath208 let us check if this result is self - averaging .
to this end , we estimate the number of thermal tlss , @xmath209 , in a sphere of radius @xmath210 at time @xmath211 , with @xmath212 the echo dephasing rate @xmath213 this yields @xmath214 therefore , except for strains very close to the symmetry point ( which are outside the resolution of the experiment ) , a sphere of radius @xmath210 contains many thermal tlss .
the contribution of @xmath43-tlss to the echo decay signal is thus self - averaging , and the echo dephasing rate is given by eq .
( [ eq:30 ] ) . since thermal @xmath43-tlss
fluctuate very slowly , we expect the echo dephasing rate to be much smaller than the ramsey dephasing rate .
using eqs .
( [ eq:25 ] ) and ( [ eq:30 ] ) , we find @xmath215 thus , the echo protocol is very efficient in suppressing the noise caused by fluctuations of thermal @xmath43-tlss .
this contradicts the experimental observation , which shows that the echo dephasing rate is much larger than the contribution of @xmath43-tlss , eq .
( [ eq:30 ] ) . furthermore , the @xmath216 dependence of the echo dephasing rate on the asymmetry bias , predicted by eq .
( [ eq:30 ] ) , is also in disagreement with the experimental result , which shows a quadratic dependence . by repeating the calculations in 2d
one finds @xmath217 ( see appendix [ sec appa ] ) , which also contradicts the experiment . in fact , among the three integrals of eq .
( [ eq:15 ] ) only the third one can explain the quadratic dependence of the echo dephasing rate .
the contribution to this integral comes from fluctuators with @xmath218 and @xmath116 , namely from fast fluctuators that are weakly coupled to the probed tls .
such fluctuators are in the opposite regime of that corresponding to @xmath43-tlss . in the next section
we show that @xmath44-tlss , proposed in ref .
, obey the above conditions and form a possible source of such a noise .
we estimate the order of magnitude of the echo dephasing rate due to such tlss and examine their relevance to the results of ref ..
the stm assumes that the coupling of tlss to the strain is unique or taken from a narrow distribution . here
we consider the possibility that tlss with much stronger coupling to the strain generically exist in amorphous solids , as has recently been proposed in ref . .
we show that the existence of a few or even a single thermal and strongly coupled to strain tlss within the amorphous barrier produces echo dephasing rate which is in agreement with both the magnitude and the strain dependence found experimentally . the theory presented in ref .
suggests the existence of two types of tlss , characterized by different coupling strengths to strain fields .
weakly and strongly coupled tlss are characterized by coupling strength @xmath219 and @xmath220 to strain fields , respectively .
the small dimensionless parameter of the theory is then @xmath221 , which is expected @xcite to lie in the range @xmath222 .
the bimodality of the coupling strengths of tlss to strain fields also modifies the phonon - mediated interactions between the tlss.@xcite the hamiltonian ( [ eq : int1 ] ) is then generalized to@xcite @xmath223 where @xmath11 and @xmath224 are the pauli matrices that represent the @xmath43- and @xmath44-tlss at sites @xmath13 and @xmath14 , and the interaction coefficients are similar to those of eqs .
( [ eq : int2 ] ) and ( [ eq : int3 ] ) , @xmath225 with @xmath226 and @xmath142 being the distance between nearest neighbor tlss .
the implications of the interactions ( [ eq:1 ] ) between the two types of tlss on their single particle density of states ( dos ) have been studied in detail in refs . and .
it was found that a pseudo - gap exists in the dos of @xmath44-tlss at low energies @xmath227ghz . in contrast
, the @xmath43-tls dos is well described by a gaussian with width @xmath228 .
thus , at low energies @xmath229k , @xmath44-tlss are much more scarce than @xmath43-tlss . as a consequence , most low - temperature properties , such as specific heat and thermal conductivity ,
are dominated by the weakly interacting @xmath43-tlss , which can be identified as the well - known tlss of the stm .
we estimate the maximum relaxation rate of thermal @xmath44-tlss and their typical coupling to the probed @xmath43-tls by repeating the calculations of sec .
[ sec 2a ] .
the relaxation rate is given by eqs .
( [ eq:16])-([eq:18 ] ) , with @xmath219 replaced by @xmath230 .
this yields the maximum relaxation rate @xmath231 where we used eq .
( [ eq:19 ] ) and @xmath222 . in appendix [ sec appb ]
we derive the typical distance between the probed tls and its nearest thermal @xmath44-tls and conclude that it is somewhat larger than the size of the amorphous tunnel barrier used in ref .
( @xmath232 nm ) .
we therefore assume the existence of one or few @xmath44-tlss in the amorphous layer , relying also on the possibility of an enhanced @xmath44-tlss dos on the surface.@xcite thus , the typical distance between the probed tls and its nearest thermal @xmath44-tls is of the order of the amorphous layer size which gives a typical interaction @xmath233 therefore , one finds that for @xmath44-tlss @xmath234 .
it thus follows that in a typical sample the thermal @xmath44-tlss are characterized by @xmath93 , as shown schematically in fig .
[ fig : fig4 ] ( cf .
fig.[fig : fig2 ] ) .
as argued above , such fluctuators give rise to a quadratic dependence of the echo dephasing rate on the asymmetry bias @xmath39 .
we now study in more detail the effects of @xmath44-tlss on the ramsey and echo decay signals .
( color online ) schematic view of a bath of thermal @xmath44-tlss ( green circles ) and @xmath43-tlss ( red circles ) coupled to the probed tls ( blue circle ) .
non - thermal tlss are shown in black circles .
thermal @xmath44-tlss are much more scarce than thermal @xmath43-tlss and thus located much farther from the probed tls ( the typical distance @xmath137 is shown by a green arrow ) .
thus , in contrast to @xmath43-tlss , @xmath44-tlss are fast and weakly coupled to the probed tls , i.e. they are typically in the regime @xmath218 and @xmath93.,scaledwidth=40.0% ] for @xmath44-tlss , which satisfy @xmath93 , we neglect the second integral of eqs .
( [ eq:14 ] ) and ( [ eq:15 ] ) . at short times ,
@xmath235 , only the first integral of eqs .
( [ eq:14 ] ) and ( [ eq:15 ] ) contributes .
similarly to eq .
( [ eq : lnftyp ] ) , the typical ramsey decay signal is governed by the nearest thermal @xmath44-tlss , i.e.@xmath236^{2}t^{2}\ , .\end{aligned}\ ] ] this quantity is very small for @xmath237 , so dephasing has not yet occurred at such short times .
the typical echo decay signal is of the order @xmath238^{2}\gamma^{(s)}_{1,\text{max}}t^{3}\ , , \end{aligned}\ ] ] which is also very small .
it should also be emphasized that in contrast to @xmath43-tlss , the contribution of @xmath44-tlss to the average echo decay signal is not self - averaging , since the first integral of eq .
( [ eq:15 ] ) is dominated by rare samples in which the nearest thermal @xmath44-tls is located at @xmath239 . as @xmath77 becomes longer than @xmath240 , both ramsey and echo decay signals
are dominated by the third integral of eqs .
( [ eq:14 ] ) and ( [ eq:15 ] ) .
since this integral also involves @xmath241 , the main contribution comes from the thermal @xmath44-tls for which the quantity @xmath242 is maximum .
thus , in a typical sample we expect the order of magnitude of the contribution of @xmath44-tlss to the ramsey and echo decay signals to be @xmath243^{2}}{\gamma^{(s)}_{1,\text{max}}}t\ , .\end{aligned}\ ] ] the ramsey and echo dephasing rates in 2d are thus @xmath244 where in the last step we used eqs .
( [ eq:33 ] ) and ( [ eq:33b ] ) .
close to the symmetry point eq .
( [ eq:39 ] ) predicts a quadratic dependence of the echo dephasing rate on the asymmetry bias , in agreement with the experiment .
moreover , the order of magnitude of the echo dephasing rate is also in line with the experimental result .
we note that the above result holds provided that at least one @xmath44-tls exists within the amorphous layer .
otherwise , we expect the echo dephasing to be negligible .
this is indeed observed in one out of four tlss in the experiment in ref . .
the result ( [ eq:39 ] ) for the ramsey and echo dephasing rates due to fast fluctuators can also be understood in terms of the gaussian approximation , since in the regime @xmath116 the results of the spin - fluctutator model are expected to coincide with the gaussian approximation.@xcite within the gaussian approximation , where @xmath245 in eqs .
( [ eq:6 ] ) , ( [ eq:8 ] ) and ( [ eq:9 ] ) is assumed to have a gaussian statistics , one ends up with @xmath246 where @xmath247 and @xmath248 is the noise spectral density , i.e. the fourier transform of the correlation function @xmath249 . if @xmath248 is approximately constant for @xmath250,@xcite namely @xmath245 is a white noise , then @xmath251 where @xmath252 using eq .
( [ eq:6 ] ) and assuming independent fluctuators , one finds@xcite @xmath253 where @xmath254 is the noise spectral density of a single fluctuator , @xmath255 equation ( [ eq:41 ] ) then yields @xmath256 for an interaction which falls off as @xmath257 , the sum is dominated by a few closet fluctuators and the result ( [ eq:44 ] ) is of the same order of magnitude as ( [ eq:39 ] ) .
we have investigated the strain - dependent dephasing rates of individual tlss in the amorphous tunnel barrier of a superconducting phase qubit , as recently studied in ref .. the interaction between the probed tls and thermal standard ( @xmath43- ) tlss described in the framework of the stm is capable of explaining the observed ramsey dephasing rate .
this includes : 1 . the linear dependence on the applied strain and hence on the bias energy of the probed tls , @xmath192 , near the symmetry point , @xmath33 .
it is a consequence of the small relaxation rate of thermal @xmath43-tlss compared to their interaction with the probed tls .
the deviation from linear behavior and the appearance of irregularities when @xmath32 is varied by an amount of the order of the thermal energy @xmath258ghz .
such features are expected since the theory predicts the ramsey dephasing to be dominantly caused by a small number of tlss .
as the strain is varied , some of these tlss are no longer thermal while other tlss will become so and contribute to the ramsey dephasing .
3 . the magnitude of the ramsey dephasing rate . using standard estimates for the coupling of @xmath43-tlss to strain fields , @xmath259ev , and for the dimensionless `` tunneling strength '' @xmath260 , it is shown that the order of magnitude of the ramsey dephasing rate is in excellent agreement with the experimental result .
the estimate for a quasi-2d configuration is in better agreement than that in a 3d configuration which is consistent with the experimental setup .
although we considered elastic interactions between @xmath43-tlss , it should be noted that our analysis and results can be carried through to the case where electric dipole interactions are present between @xmath43-tlss . for typical values of the dipole moment , of the order of @xmath261debye,@xcite electric dipole interactions
are expected to be of the same order of magnitude as elastic interactions between @xmath43-tlss and the results are the same .
furthermore , we predict the dependence of the ramsey dephasing rate on temperature for both 2d and 3d geometries for an arbitrary power - law energy dependence of the tls dos at low energies . for the echo dephasing , our analysis shows that the contribution of @xmath43-tlss is very much reduced ( @xmath262 for the typical strains studied in ref . ) and the bias - dependence is predicted to be @xmath263 ( or @xmath264 in 3d ) .
such small dephasing rates are observed only in one out of four tlss studied in ref . .
in the other tlss the echo dephasing rate is higher and varies quadratically with the applied strain , implying the presence of energy fluctuations on a time scale @xmath40 .
we suggest that tlss that are strongly coupled to strain@xcite and therefore fast fluctuating may be the dominant source for noise and governing echo dephasing at low temperatures .
the properties of such tlss , in particular their scarcity compared to the standard tlss of the stm , result in both the order of magnitude and the strong fluctuations of the echo dephasing as seen in the experiment @xcite .
we acknowledge fruitful discussions with j. lisenfeld .
this work was supported by the german - israeli foundation ( gif ) and by the dfg research grant scho 287/7 - 1 , sh 81/2 - 1 .
we give here the results of the calculations performed in sec .
[ sec 2c ] for the 2d case .
the integrals [ eq:27 ] and [ eq:28 ] in 2d give @xmath266^{2/3}\gamma^{}_{1,\text{max}}\ , , \end{aligned}\ ] ] @xmath267^{2/3}\gamma^{}_{1,\text{max}}\ , , \end{aligned}\ ] ] where @xmath268 is given by the first of eqs .
( [ eq:21 ] ) .
the sum of eqs .
( [ eq : appa_eq1 ] ) and ( [ eq : appa_eq2 ] ) yields [ cf .
( [ eq:29 ] ) ] @xmath269^{2/3}\nonumber\\ & \times\gamma^{}_{1,\text{max}}\ , , \end{aligned}\ ] ] where a numerical coefficient of order unity is omitted .
the echo dephasing rate is thus @xmath270^{3/5}|\cos\theta^{}_{p}|^{2/5}\ , .\end{aligned}\ ] ] equation ( [ eq:31 ] ) takes the form @xmath271^{2/3}\nonumber\\ & \approx\pi\left[\frac{d^{3/2}\xi j^{}_{t,3d}}{r^{3/2}_{0}\gamma^{}_{1,\mathrm{max}}}\sqrt{\frac{j^{}_{t,3d}}{j^{}_{0}}}|\cos\theta^{}_{p}|\right]^{2/5}\nonumber\\ & \approx 100|\cos\theta^{}_{p}|^{2/5}\ , .\end{aligned}\ ] ] therefore , except for strains very close to the symmetry point ( which are outside the resolution of the experiment ) , a circle of radius @xmath210 contains many thermal tlss .
the contribution of @xmath43-tlss to the echo decay signal in 2d is thus also self - averaging , and the echo dephasing rate is given by eq .
( [ eq : appa_eq4 ] ) .
the ratio between the echo dephasing rate and the ramsey dephasing rate is @xmath272 which suggests that the echo protocol is very efficient also in 2d .
to estimate the typical distance @xmath137 between the probed tls and its nearest thermal @xmath44-tls , we use the relations [ cf.eqs .
( [ eq:20 ] ) ] @xmath273 where @xmath274 is the dos of @xmath44-tlss . for @xmath275 , the interaction between the tlss [ eqs .
( [ eq:1 ] ) and ( [ eq:2 ] ) ] gives rise to a power - low pseudo - gap in the dos of @xmath44-tlss , namely @xmath276.@xcite moreover , the dos of @xmath44-tlss is related to that of @xmath43-tlss via @xmath277 at @xmath278 , where @xmath279k is the temperature below which universality in amorphous and disordered systems is observed.@xcite since the dos of @xmath43-tlss is assumed to be constant , @xmath280 , it follows that @xmath281 . using eqs .
( [ eq:34 ] ) , we readily obtain @xmath282 setting @xmath283mk , @xmath279k and @xmath284k , we obtain @xmath285 ( we note that this result is weakly dependent on @xmath286 ) . assuming @xmath287 nm we find @xmath288 m , somewhat larger than the sample size . for completeness
we give here the typical coupling at this distance . inserting eq .
( [ eq:34b ] ) in the interaction coefficients ( [ eq:2 ] ) , we find the typical coupling strength between the probed tls and its nearestc thermal @xmath44-tls , @xmath289^{3/2}j^{\tau\tau}_{t,2d}\sim 2\pi\cdot g^{2}\,\mathrm{mhz } \;\ : \left(\mathrm{2d}\right)\ , , \end{aligned}\ ] ] where the coupling strength between the probed tls and its nearest thermal @xmath43-tls , @xmath290 , is defined in eqs .
( [ eq:21 ] ) for both @xmath164 and @xmath173 .
r. c. zeller and r. o. pohl , phys .
b * 4 * , 2029 ( 1971 ) .
j. f. berret and m. meissner , z. phys .
b : condens . matter * 70 * , 65 ( 1988 ) .
r. o. pohl , x. liu , and e. thompson , rev .
phys . * 74 * , 991 ( 2002 ) .
p. w. anderson , b. i. halperin , and c. m. varma , philos . mag . * 25 * , 1 ( 1972 ) .
w. a. phillips , j. low - temp .
* 7 * , 351 ( 1972 ) .
w. a. phillips , rep .
. phys . * 50 * , 1657 ( 1987 ) .
j. l. black and b. i. halperin , phys .
b * 16 * , 2879 ( 1977 ) .
a. l. burin , d. natelson , d. d. osheroff , and y. kagan , in _ tunneling systems in amorphous and crystalline solids _ , edited by p. esquinazi ( springer , berlin , 1998 ) .
m. schechter and p. c. e. stamp , j. phys .
matter * 20 * , 244136 ( 2008 ) .
e. paladino , y. m. galperin , g. falci , and b. l. altshuler , rev .
phys . * 86 * , 361 ( 2014 ) .
r. w. simmonds , k. m. lang , d. a. hite , s. nam , d. p. pappas , and j. m. martinis , phys .
* 93 * , 077003 ( 2004 ) .
j. m. martinis , k. b. cooper , r. mcdermott , m. steffen , m. ansmann , k. d. osborn , k. cicak , s. oh , d. p. pappas , r. w. simmonds , and c. c. yu , phys . rev .
lett . * 95 * , 210503 ( 2005 ) .
m. neeley , m. ansmann , r. c. bialczak , m. hofheinz , n. katz , e. lucero , a. oconnell , h. wang , a. n. cleland , and j. m. martinis , nature phys .
* 4 * , 523 ( 2008 )
. y. shalibo , y. rofe , d. shwa , f. zeides , m. neeley , j. m. martinis , and n. katz , phys . rev
* 105 * , 177001 ( 2010 ) .
j. lisenfeld , c. mller , j. h. cole , p. bushev , a. lukashenko , a. shnirman , and a. v. ustinov , phys .
lett . * 105 * , 230504 ( 2010 ) . as mentioned above , the strain field in the absence of any external perturbation is also responsible for the relaxation of the probed tls , discussed in ref . , and
for the tls - tls interactions [ see eqs .
( [ eq : int1])-([eq : int3 ] ) and the third term of eq .
( [ eq : hpj ] ) ] . in eq .
( [ eq : probedh ] ) we focus only on the applied strain @xmath28 , whose sole effect is to externally tune the energy bias of the probed tls .
g. j. grabovskij , t. peichl , j. lisenfeld , g. weiss , and a. v. ustinov , science * 338 * , 232 ( 2012 ) .
j. lisenfeld , a. bilmes , s. matityahu , s. zanker , m. marthaler , m. schechter , g. schn , a. shnirman , g. weiss , and a.v .
ustinov , arxiv:1601.03213 .
we set @xmath291 and measure energies in units of frequencies or temperatures .
w. arnold and s. hunklinger , solid state commun . * 17 * , 833 ( 1975 ) . w. arnold , c. martinon , and s. hunklinger , j. physique colloq . * 39 * , c6 - 961 ( 1978 ) .
an alternative explanation was given in s. zanker , m. marthaler , g. schn , arxiv:1601.03861 . m. schechter and p. c. e. stamp , phys
b * 88 * , 174202 ( 2013 ) .
e. paladino , l. faoro , g. falci , and r. fazio , phys .
rev . lett . * 88 * , 228304 , ( 2002 ) .
y. m. galperin , b. l. altshuler , and d. v. shantsev , in _ fundamental problems of mesoscopic physics _ , edited by i. v. lerner et .
( kluwer , dordrecht , 2004 ) ; cond - mat/0312490 . j. bergli , y. m. galperin , and b. l. altshuler , new . j. phys . *
11 * , 025002 ( 2009 ) .
j. jckle , z. phys .
* 257 * , 212 ( 1972 ) .
a. l. burin and yu .
kagan , jetp * 79 * , 347 ( 1994 ) .
y. m. galperin , b. l. altshuler , j. bergli , and d. v. shantsev , phys .
* 96 * , 097009 ( 2006 ) .
g. ithier , e. collin , p. joyez , p. j. meeson , d. vion , d. esteve , f. chiarello , a. shnirman , y. makhlin , j. schriefl , and g. schn , phys . rev .
b * 72 * , 134519 ( 2005 ) .
j. schriefl , y. makhlin , a. shnirman , and g. schn , new j. phys .
* 8 * , 1 ( 2006 ) .
s. hunklinger , cryogenics * 28 * , 224 ( 1988 ) .
j. lisenfeld , private communication .
we use the term `` rate '' to refer to the inverse of the time for which the signal has decayed to @xmath292 .
the decay , however , might not be an exponential one .
j. burnett , l. faoro , i. wisby , v. l. gurtovoi , a. v. chernykh , g. m. mikhailov , v. a. tulin , r. shaikhaidarov , v. antonov , p. j. meeson , a. y. tzalenchuk , and t. lindstrm , nat . commun . * 5 * , 4119 ( 2014 ) .
l. faoro and l. b. ioffe , phys .
b * 91 * , 014201 ( 2015 ) .
a. churkin , s. matityahu , a. l. burin , and m. schechter , to be published .
a. gaita - ario and m. schechter , phys .
lett . * 107 * , 105504 ( 2011 ) .
a. churkin , d. barash , and m. schechter , j. phys .
matter * 26 * , 325401 ( 2014 ) .
the low - energy effective hamiltonian includes also a random field term of the form @xmath293 , where the random fields @xmath294 and @xmath295 have typical strengths @xmath296 and @xmath297 . such a term has no effect on the dynamics of the tlss and can be absorbed into the energy bias @xmath1 .
a. churkin , d. barash and m. schechter , phys .
b * 89 * , 104202 ( 2014 ) .
a. churkin , i. gabdank , a. burin , and m. schechter , arxiv:1307.0868 . l. faoro and l. b. ioffe , phys .
109 * , 157005 ( 2012 ) .
this condition is equivalent to @xmath298 where @xmath299 is the time over which the correlation function @xmath300 decays .
a. p. paz , i. v. lebedeva , i. v. tokatly , and a. rubio , phys .
b * 90 * , 224202 ( 2014 ) .
m. s. khalil , s. gladchenko , m. j. a. stoutimore , f. c. wellstood , a. l. burin , k. d. osborn , phys .
b * 90 * , 100201(r ) ( 2014 ) .
a. l. burin , s. matityahu , and m. schechter , phys .
b * 92 * , 174201 ( 2015 ) . | we study the dephasing of an individual high - frequency tunneling two - level system ( tls ) due to its interaction with an ensemble of low - frequency thermal tlss which are described by the standard tunneling model ( stm ) .
we show that the dephasing by the bath of tlss explains both the dependence of the ramsey dephasing rate on an externally applied strain as well as its order of magnitude , as observed in a recent experiment [ j. lisenfeld _ et al . _ ] .
however , the theory based on the stm predicts the hahn - echo protocol to be much more efficient , yielding too low echo dephasing rates , as compared to the experiment .
also the strain dependence of the echo dephasing rate predicted by the stm does not agree with the measured quadratic dependence , which would fit to a high - frequency white noise environment .
we suggest that few fast tlss which are coupled much more strongly to the strain fields than the usual tlss of the stm give rise to such a white noise .
this explains the magnitude and strong fluctuations of the echo dephasing rate observed in the experiment . |
consider the following class of regression models , with response variable denoted by @xmath0 and @xmath1-dimensional covariate vector by @xmath2 , @xmath3 where @xmath4 is the unknown parameter vector , @xmath5 is the unobserved error term that is independent of @xmath2 with a completely unspecified distribution , and @xmath6 is a monotone increasing , but otherwise unspecified function .
it is easily seen that this class of models contains many commonly used regression models as its submodels that are especially important in the econometrics and survival analysis literature .
for example , with @xmath7 , becomes the standard regression model with an unspecified error distribution ; with @xmath8 ( @xmath9 ) , the box - cox transformation model ( box and cox , 1964 ) ; with @xmath10 $ ] , the binary choice model ( maddala , 1983 ; mcfadden , 1984 ) ; with @xmath11 $ ] , a censored regression model ( tobin , 1958 ; powell , 1984 ) ; with @xmath12 , the accelerated failure times ( aft ) model ( cox and oakes , 1984 ; kalbfleisch and prentice , 2002 ) ; with @xmath5 having an extreme value density @xmath13 , the cox proportional hazards regression ( cox , 1972 ) ; with @xmath5 having the standard logistic distribution , the proportional odds regression ( bennett , 1983 ) . a basic tool for handling model is the maximum rank correlation ( mrc ) estimator proposed in the econometrics literature by han ( 1987 ) .
because both the transformation function @xmath6 and the error distribution are unspecified , not all components of @xmath4 are identifiable . without loss of generality
, we shall assume henceforth that the last component , @xmath14 .
let @xmath15 be a random sample from .
han s mrc estimator , denoted by @xmath16 , is the maximizer of following objective function @xmath17}i[{{{\bf x}}_i'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})>{{\bf x}}_j'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})}],\ ] ] where @xmath18 $ ] denotes the indicator function , @xmath19 the transpose of @xmath2 , and @xmath20 the first @xmath21 components of @xmath4 , i.e. @xmath22 . han ( 1987 ) proved that the mrc estimator @xmath16 is strongly consistent under certain regularity conditions .
an important subsequent development is due to sherman ( 1993 ) , who made use of the empirical process theory and hoeffding s decomposition to approximate the objective function , viewed as a u - process .
he showed that @xmath16 is , in fact , asymptotically normal under additional regularity conditions .
estimation of the transformation function @xmath6 was studied by chen ( 2002 ) , who constructed a rank - based estimator and established its consistency and asymptotic normality .
in addition to the econometrics , model also encompasses the main semiparametric models in survival analysis , where right censoring is a major feature . under the right censorship , there is a censoring variable @xmath23 and one observes @xmath24 and @xmath25 .
khan and tamer ( 2007 ) constructed the following partial rank correlation function as an extension of the rank correlation objective function , @xmath26}i[{{{\bf x}}_i'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})>{{\bf x}}_j'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})}].\ ] ] they showed that the resulting maximum partial rank correlation estimate ( prce ) @xmath27 , as the maximizer of @xmath28 , is consistent and asymptotically normal .
crucial for the statistical inference of based on @xmath29 is the consistent variance estimation . in standard objective ( loss ) function derived estimation ,
the asymptotic variance is usually estimated by a sandwich - type estimator of form @xmath30 with @xmath31 being the second derivative of the objective function and @xmath32 an estimator of the variance of the first derivative ( score ) .
the challenge here , however , is that @xmath33 itself is a ( discontinuous ) step function that precludes automatic use of differentiation to obtain @xmath31 .
furthermore , @xmath32 is also difficult to obtain since the score function can not be derived directly from @xmath33 via differentiation .
sherman(1993 ) suggested using numerical derivatives of first and second orders to construct @xmath31 and @xmath32 .
his approach requires bandwidth selection for the derivative functions .
it is unclear how stable the resulting variance estimator is .
alternatively , one may resort to bootstrap ( efron , 1979 ) or other resampling methods ( e.g. jin et al .
these approaches require repeatedly solving the maximization of , which is discontinuous and often multidimensional when @xmath34 .
the computational cost could therefore be prohibitive . in this paper
, we develop a self - induced smoothing method for rank correlation criterion function so that the differentiation can be performed , while bypassing the bandwidth selection .
both point and variance estimators can be obtained simultaneously in a straightforward way that is typically used for smooth objective functions .
the new method is motivated by a novel approach proposed in brown and wang ( 2005 , 2007 ) , where an elegant self - induced smoothing method was introduced for non - smooth estimating functions . although our approach bears similarity with that of brown and wang ( 2005 ) , it is far from clear why such self - induced smoothing is suitable for the discrete objective function ( rank correlation ) .
in fact , undersmoothing would make the hessian ( second derivative ) unstable while oversmoothing would introduce significant bias . through highly technical and tedious derivations , we will show that the proposed method does strike a right balance in terms of asymptotic unbiasedness and enough smoothness for differentiation ( twice ) .
the rest of the paper is organized as follows . in section 2 ,
the new methods are described and related large sample properties are developed . in particular , we give construction for simultaneous point and variance estimation and show that the resulting point estimator is asymptotically normal and the variance estimator is consistent . in section 3 , the approach , along with the algorithm and large sample properties , is extended to handle survival data with right censoring .
simulation results are reported in section 4 , where application to a real data set is also given .
section 5 contains some concluding remarks .
additional technical proofs can be found in the appendix .
in this section we develop a self - induced smoothing method for the rank correlation criterion function defined by .
it is divided into three subsections , with the first introducing the method and the algorithm , the second establishing large sample properties and the third covering proofs .
since mrc estimator @xmath29 is asymptotically normal ( sherman , 1993 ) , its difference with the true parameter value , @xmath35 , should approximately be a gaussian noise @xmath36 , where @xmath37 is a @xmath21-dimensional normal random vector with mean @xmath38 and covariance matrix @xmath39 .
assume that @xmath40 is independent of data and let @xmath41 denote the expectation with respect to @xmath40 given data .
a self - induced smoothing for @xmath33 is @xmath42 .
the self - induced smoothing using the limiting gaussian distribution was originally proposed by brown and wang ( 2005 ) for certain non - smooth estimating functions . to get an explicit form for @xmath43 ,
let @xmath44 be the standard normal distribution function , @xmath45 , @xmath46 where @xmath47 denotes the first @xmath21 components of @xmath48 .
then , it is easy to see that @xmath49\phi\left(\sqrt{n}{{\bf x}}_{ij}'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})/\sigma_{ij}\right).\ ] ] we shall use @xmath50 to denote the corresponding estimator , which will be called the smoothed maximum rank correlation estimator ( smrce ) . here and in the sequel , @xmath51 denotes the parameter space for @xmath20 .
smoothing is an appealing way for a simple solution to the inference problem associated with the mrce .
if @xmath43 were a usual smooth objective function , then its first derivative would become the score function and its second derivative could be used for variance estimation .
speficically , if we use @xmath52 to denote the limiting variance of the score scaled by @xmath53 and @xmath54 the limit of the second derivative , then the asymptotic variance of the resulting estimator , scaled by @xmath55 , should be of form @xmath56 . a consistent estimator could then be obtained by the plug - in method , i.e. replacing unknown parameters by their corresponding empirical estimators .
it is unclear , however , whether or not the self - induced smooth will provide a right amount of smoothing , even in view of the results given in brown and wang ( 2005 ) . with over - smoothing
, @xmath57 may be asymptotically biased , i.e. the bias is not of order @xmath58 ; with under - smoothing , the `` score '' function ( first derivative of @xmath43 ) may have multiple `` spikes '' and thus the second derivative matrix ( hessian ) of @xmath43 may not behave properly and certainly can not be expected to provide a consistent variance estimator . in subsection 2.2
, we show that the self - induced smoothing here does result in a right amount of smoothing in the sense that the bias is asymptotically negligible and the hessian matrix behave properly . before starting the theoretic developments ,
we first describe our method .
we first differentiate the smoothed objective function @xmath43 to get score @xmath59 where @xmath60 .
this is a u - process of order 2 with kernel @xmath61 where @xmath62 denotes the pair @xmath63 . by hoeffding s decomposition ,
the asymptotic variance of @xmath64 is approximated by @xmath65\right\}^{\otimes2},\\ \end{split}\end{aligned}\ ] ] where , for a vector @xmath66 , @xmath67 .
thus , @xmath68 is used to estimate @xmath52 , the middle part of the `` sandwich '' variance formula discussed in remark 1 .
as for @xmath54 , we differentiate @xmath69 to get @xmath70^{\otimes2}\right\},\ ] ] where @xmath71 is the derivative of @xmath72 .
although the self - induced smoothing was motivated earlier with @xmath39 being the limiting covariance matrix of the estimator , we will show later that for any positive definite matrix @xmath39 , @xmath73 converges to @xmath54 .
note that the above discussions about @xmath54 and @xmath52 are not mathematically rigorous .
this is because the kernel function for the score process is sample size @xmath53-dependent .
the usual asymptotic theory for the u - process is not applicable
. indeed , our rigorous derivations , to be given in subsection 2.3 , are quite tedious , involving many approximations that are quite delicate .
let @xmath74 if @xmath20 is the true parameter value , then @xmath75 converges to the limiting covariance matrix , which is the desired choice for @xmath39 in the self - induced smoothing .
therefore , leads to an iterative algorithm of form @xmath76 ; see also brown and wang ( 2005 ) . specifically , we propose the following algorithm : ( smrce ) 1 .
compute the mrc estimator @xmath16 and set @xmath77 to be the identity matrix .
2 . update variance - covariance matrix @xmath76 .
smooth the rank correlation @xmath78 using covariance matrix @xmath79 .
maximize the resulting smoothed rank correlation to get an estimator @xmath80 .
repeat step 2 until @xmath80 converge .
this subsection is devoted to the large sample theory .
the main results are : 1 .
the smoothed mrc estimator ( smrce ) is asymptotically equivalent to the mrc estimator ; 2 .
the proposed method leads to a consistent variance estimator ; and 3 .
the iterative algorithm for point and variance estimation converges numerically .
we first introduce notation as well as assumptions , which are similar to those in sherman ( 1993 ) for the mrc estimator .
let @xmath81}i_{[({{\bf x}}-{{\bf x}})'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})>0]}+ i_{[y < y]}i_{[({{\bf x}}-{{\bf x}})'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})<0]}\right],\ ] ] which is the projection of the kernel of u - process @xmath78 .
the expectation is taken for @xmath82 .
also let @xmath83 the following assumptions 1 and 2 are used in han ( 1987 ) ( see also sherman , 1993 ) to establish consistency for the mrc estimator . for asymptotic normality ,
we need an additional regularity condition ( assumption 3 ) given in sherman ( 1993 ) .
the true parameter value @xmath84 is an interior point of @xmath51 , which is a compact subset of the @xmath21-dimensional euclidean space @xmath85 .
the support of @xmath2 is not contained in any linear subspace of @xmath86 .
conditional on the first @xmath21 components of @xmath2 , the last component of @xmath2 has a density function with respect to the lebesgue measure .
there exists a neighborhood , @xmath87 , of @xmath84 such that for each pair @xmath88 of possible values of @xmath89 , \(i ) the second derivatives of @xmath90 with respect to @xmath20 exist in @xmath91 .
\(ii ) there is an integrable function @xmath92 such that for all @xmath20 in @xmath91 , @xmath93 \(iii ) @xmath94 \(iv ) @xmath95 \(v ) the matrix @xmath96 is strictly negative definite .
( sherman , 1993 ) assume that assumptions 1 - 3 hold .
we have , uniformly over any @xmath97 neighborhood of @xmath84 , @xmath98 where @xmath99 , @xmath100 and @xmath101 . consequently , for the mrc estimator @xmath16 , @xmath102 where @xmath103 , @xmath104')$ ] and @xmath105 . because of the standardization , the rank correlation criterion function @xmath33 is bounded by 1 .
it is not difficult to establish a uniform law of large numbers @xmath106 where @xmath107 is the expectation of @xmath78 ; cf .
han ( 1987 ) and sherman ( 1993 ) .
likewise , we can show that such uniform convergence also holds for @xmath43 , i.e. @xmath108 note that the limit @xmath109 remains the same . in the following theorem
, we claim that the estimate obtained from maximizing the smoothed rank correlation function is also asymptotically normal with the same asymptotic covariance matrix as han s mrce .
[ thy : normality ] for any given positive definite matrix @xmath39 , let @xmath110 be defined as in and @xmath111 . then , under assumptions 1 - 3 , @xmath112 is consistent , @xmath113 a.s . and asymptotically normal , @xmath114 where @xmath115 is defined as in proposition 1 .
in addition , @xmath112 is asymptotically equivalent to @xmath16 in the sense that @xmath116 .
recall that defines the sandwich - type variance estimator by pretending that @xmath117 is a standard smooth objective function .
theorem [ thy : var formula ] below shows that is consistent .
[ thy : var formula ] let @xmath118 be the mrc estimator and @xmath119 be defined by . then , for any fixed positive definite matrix @xmath39 , @xmath120 converges in probability to @xmath115 , the limiting variance - covariance matrix of @xmath121 .
the self - induced smoothing uses the limiting covariance matrix @xmath115 as @xmath39 . in practice
, we may initially choose the identity matrix for @xmath39 , which is the same way as the initial step in algorithm i. by theorem 2.1 , we know that the one - step estimator @xmath122 in algorithm i converges in probability to the true covariance . however , this one - step estimator depends on the initial choice of @xmath39 .
algorithm 1 is an iterative algorithm with the variance - covariance estimator converging to the fixed point of @xmath123 .
convergence of algorithm 1 is ensured by the following theorem . for notational simplicity , we let @xmath124 be the vectorization of matrix @xmath125 . for any function @xmath126 of @xmath39 , @xmath127 where @xmath128 denotes the @xmath129 entry of @xmath39 .
[ thy : alg converg ] let @xmath130 be defined as in algorithm 1 .
suppose that assumptions 1 - 3 hold .
then there exist @xmath131 , @xmath132 , such that for any @xmath133 , there exists @xmath134 , such that for all @xmath135 , @xmath136 for a fixed @xmath53 , @xmath137 represents the fixed point matrix in the iterative algorithm .
the above theorem shows that with probability approaching 1 , the iterative algorithm converges to a limit , as @xmath138 , and the limit converges in probability to the limiting covariance matrix @xmath115 .
the speed of convergence of @xmath130 to @xmath137 is faster than any exponential rate in the sense that @xmath139 for any @xmath140 .
this can be seen from step 2 of algorithm 1 in subsection 2.1 and below , @xmath141_{r , s}=o_p(1),\ ] ] which will be proved in the appendix . here
@xmath142 is a small neighborhood of @xmath115 and @xmath39 is a positive definite matrix . in this section
, we provide proofs for ( 1 ) asymptotic equivalence of smrce to mrce , ( 2 ) consistency of the induced variance estimator and ( 3 ) convergence of algorithm 1 . some of the technical developments used in the proofs will be given in the appendix .
[ proof of theorem [ thy : normality ] ] without loss of generality , we assume @xmath143 as in subsection 2.1 , let @xmath40 be a @xmath21-variate normal random vector with mean @xmath144 and covariance matrix @xmath39 .
define @xmath145 let @xmath146 and @xmath147 . define @xmath148 = { \rm argmax}_{{\mbox{\boldmath$\theta$ } } } { \widetilde\gamma_n}({\mbox{\boldmath$\theta$}}).\ ] ] let @xmath149 $ ] , where @xmath150 then @xmath151 due to the gaussian tail of @xmath40 .
since @xmath152 and @xmath153 , @xmath154\}|\leq p({\mbox{\boldmath$\omega$}}_n)=o(n^{-2}).\ ] ] by the cauchy - schwarz inequality , @xmath155\}=o(n^{-2 } ) \mbox { and } e_{{{\bf z } } } \{|{{\bf z}}|^2 i[{\mbox{\boldmath$\omega$}}_n]\}=o(n^{-2}).\ ] ] by , uniformly over @xmath156 neighborhoods of @xmath144 , @xmath157\ } = ( 1/2)e_{{{\bf z } } } \{({\mbox{\boldmath$\theta$}}+{{\bf z}}/\sqrt{n})'{{\bf a}}_0({\mbox{\boldmath$\theta$}}+{{\bf z}}/\sqrt{n})i[{\mbox{\boldmath$\omega$}}_n^c]\ } \\
& + ( 1/\sqrt{n})e_{{{\bf z } } } \{({\mbox{\boldmath$\theta$}}+{{\bf z}}/\sqrt{n})'{\bf w}_ni[{\mbox{\boldmath$\omega$}}_n^c]\ } + o_p(e_{{{\bf z } } } \{|{\mbox{\boldmath$\theta$}}+{{\bf z}}/\sqrt{n}|^2i[{\mbox{\boldmath$\omega$}}_n^c]\}+\frac{1}{n}).\end{aligned}\ ] ] note that @xmath158\ } \leq 2 ( e_{{{\bf z}}}|{\mbox{\boldmath$\theta$}}|^2 + e_{{{\bf z}}}|{{\bf z}}|^2/n)=o(|{\mbox{\boldmath$\theta$}}|^2 + 1/n).\ ] ] therefore , uniformly over @xmath156 neighborhoods of @xmath144 , we have @xmath159 replacing @xmath20 in with @xmath160 and subtracting it from @xmath161 , we have @xmath162 combining ( 15 ) with lemma 1 in the appendix , we get , @xmath163 therefore , from and , we have @xmath164 finally , strong consistency of @xmath112 follows the uniform almost sure convergence of @xmath165 as stated in .
this completes the proof . for notational simplicity ,
we assume throughout the proof that @xmath39 is the identity matrix .
the same argument with modifications to include constants for up and lower bound may be applied to deal with a general covariance matrix @xmath39 .
we first show @xmath166 by definition , @xmath167_{r , s}=\partial^2\tilde q_n({\mbox{\boldmath$\theta$}})/(\partial\theta_r\partial\theta_s)$ ] .
as defined in @xmath168 has the following integral representation , @xmath169 by change of variable @xmath170 , @xmath171 @xmath172 . from , @xmath173 and @xmath174 where @xmath175 and @xmath176 .
in view of , to show , it suffices to prove @xmath177_{r , s } + o_p(1)\ ] ] uniformly over @xmath178 . to show
, we define @xmath179 by lemma 2(i ) and the boundedness of @xmath180 , we have , @xmath181 where @xmath182 for set @xmath183 denotes its complement .
therefore , reduces to @xmath184_{r , s } + o_p(1).\ ] ] to show , we establish a quadratic expansion of @xmath180 for @xmath185 . since @xmath186 for @xmath185 and @xmath187 , it follows that @xmath188 .
therefore , by , @xmath189 therefore , the left hand side of equals @xmath190 , where @xmath191\ddot{k}_{n , r , s}({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}},\\ \bf{ii}&=q_n({\mbox{\boldmath$\theta$}}_0)\times\int_{{\mbox{\boldmath$\omega$}}_{n , r}\cap{\mbox{\boldmath$\omega$}}_{n , s}}\ddot{k}_{n , r , s}({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}},\\ \bf{iii}&=\frac{{\bf w}_n'}{\sqrt{n}}\times\int_{{\mbox{\boldmath$\omega$}}_{n , r}\cap{\mbox{\boldmath$\omega$}}_{n , s}}({{\bf t}}-{\mbox{\boldmath$\theta$}}_0)\ddot{k}_{n , r , s}({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}},\\ \bf{iv}&=\frac{1}{2}\int_{{\mbox{\boldmath$\omega$}}_{n , r}\cap{\mbox{\boldmath$\omega$}}_{n , s}}({{\bf t}}-{\mbox{\boldmath$\theta$}}_0)'{{\bf a}}({\mbox{\boldmath$\theta$}}_0)({{\bf t}}-{\mbox{\boldmath$\theta$}}_0)\ddot{k}_{n , r , s}({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}}. \end{split}\end{aligned}\ ] ] by the definition of @xmath192 , @xmath193 by lemma 2(ii ) , @xmath194 .
furthermore , @xmath195 due to lemma 2(iii ) .
note that @xmath196 where the last equality follows from the fact that @xmath192 and @xmath197 are symmetric at @xmath20 and @xmath198 is an odd function of @xmath199_{r}$ ] for @xmath200 . combining this with lemma 2(i ) , we have @xmath201 .
again by symmetry , @xmath202 by lemma 2 ( i ) and ( iv ) , @xmath203_{r , s } + o(n^{-1/2})$ ] . combining the approximations for @xmath204
, we get .
next we prove @xmath205 by showing , componentwise , @xmath206_{r , s } = [ { { \bf v}}({\mbox{\boldmath$\theta$}}_0)]_{r , s } + o_p(1)\ ] ] uniformly over @xmath178 for @xmath207 .
define @xmath208}i_{[({{\bf x}}-\tilde { { \bf x}})'{\mbox{\boldmath$\beta$}}>0 ] } + i_{[y<\tilde y]}i_{[({{\bf x}}-\tilde { { \bf x}})'{\mbox{\boldmath$\beta$}}<0]},$ ] where @xmath209 and @xmath210 .
in addition , let @xmath211 where @xmath212 is the empirical distribution for @xmath213 s . by definition , @xmath214_{r , s } & = \frac{1}{n}\sum_{i=1}^n \left[\frac{\partial}{\partial\theta_r } \int \tau_n({{\bf u}}_i,{\mbox{\boldmath$\theta$}}+\frac{{{\bf z}}}{\sqrt{n } } ) ( 2\pi)^{-\frac{d}{2}}e^{-\frac{\|{{\bf z}}\|_2 ^ 2}{2}}d{{\bf z}}\right]\\ & \quad \quad \times \left[\frac{\partial}{\partial\theta_s } \int \tau_n({{\bf u}}_i,{\mbox{\boldmath$\theta$}}+\frac{\tilde { { \bf z}}}{\sqrt{n } } ) ( 2\pi)^{-\frac{d}{2}}e^{-\frac{\|\tilde { { \bf z}}\|_2 ^ 2}{2}}{d\tilde{{\bf z}}}\right ] .
\end{split}\end{aligned}\ ] ] letting @xmath170 and @xmath215 , we have @xmath214_{r , s } & = \frac{1}{n}\sum_{i=1}^n \frac{\partial}{\partial\theta_r } \int \tau_n({{\bf u}}_i,{{\bf t}})k_n({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}}\times \frac{\partial}{\partial\theta_r } \int \tau_n({{\bf u}}_i,{\mbox{\boldmath$\omega$ } } ) k_n({\mbox{\boldmath$\omega$}},\theta)d{\mbox{\boldmath$\omega$}}\\ & = \frac{1}{n}\sum_{i=1}^n \int \tau_n({{\bf u}}_i,{{\bf t}})\dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})d{{\bf t}}\times \int \tau_n({{\bf u}}_i,{\mbox{\boldmath$\omega$}})\dot{k}_{n , s}({\mbox{\boldmath$\omega$}},{\mbox{\boldmath$\theta$}})d{\mbox{\boldmath$\omega$}}\\ & = \int g_n({{\bf t}},{\mbox{\boldmath$\omega$ } } ) \dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})\dot{k}_{n , s}({\mbox{\boldmath$\omega$}},{\mbox{\boldmath$\theta$}})d{{\bf t}}d{\mbox{\boldmath$\omega$ } } , \end{split}\end{aligned}\ ] ] where @xmath216 , which is bounded by 0 and 1 . by lemma 2 ( vii ) ,
@xmath217_{r , s}&=o(n^{-\frac{1}{2}})\\ & + \int_{{\mbox{\boldmath$\omega$}}_{n , r}\times{\mbox{\boldmath$\omega$}}_{n , s } } g_n({{\bf t}},{\mbox{\boldmath$\omega$ } } ) \dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})\dot{k}_{n , s}({\mbox{\boldmath$\omega$}},{\mbox{\boldmath$\theta$}})d{{\bf t}}d{\mbox{\boldmath$\omega$}}\end{split}\end{aligned}\ ] ] uniformly over @xmath218 .
let @xmath219 and @xmath220 , the symmetrized @xmath221 . by definition , @xmath222 clearly @xmath223 is a third - order u - statistics and @xmath224 is a second - order u - statistics . applying hoeffding s decomposition ( van der vaart , 1998 , section 12.3 ) , @xmath225 where @xmath226 is a u - statistics of order @xmath227 ( @xmath228 ) and defined as @xmath229.\ ] ] here , adopting the notations from van der vaart ( 1998 , section 11.4 ) , we define @xmath230 $ ] as a projection of @xmath231 such that @xmath232 - e f^ * , \\
p_{\{i , j\ } } f^ * & = e [ f^ * | { { \bf u}}_i , { { \bf u}}_j ] - e [ f^ * | { { \bf u}}_i ] - e [ f^ * | { { \bf u}}_j ] + e f^*,\\ p_{\{1,2,3\ } } f^ * & = e [ f^ * | { { \bf u}}_1 , { { \bf u}}_2 , { { \bf u}}_3 ] - \sum_{i\neq j } e [ f^ * | { { \bf u}}_i , { { \bf u}}_j ] + \sum_{i=1,2,3}e [ f^ * | { { \bf u}}_i ] - e f^*. \end{split}\end{aligned}\ ] ] we know from hoeffding s decomposition that @xmath233 and @xmath234 are second- and third - order degenerated u - statistics with bounded kernels and thus of order @xmath235 and @xmath236 ; see sherman ( 1992 , corollary 8) . therefore , by lemma 2(vi ) , @xmath237 replacing @xmath226 by @xmath238 in also results in @xmath97 .
then combining this and with and , reduces to @xmath239_{r , s } & = 3 \times\int_{{\mbox{\boldmath$\omega$}}_{n , r}\cap{\mbox{\boldmath$\omega$}}_{n , s } } u_{n,1}\times\dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})\dot{k}_{n , s}({\mbox{\boldmath$\omega$ } } , { \mbox{\boldmath$\theta$}})d{{\bf t}}d{\mbox{\boldmath$\omega$}}\\ & + \int_{{\mbox{\boldmath$\omega$}}_{n , r}\cap{\mbox{\boldmath$\omega$}}_{n , s } } ef\times\dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})\dot{k}_{n ,
s}({\mbox{\boldmath$\omega$}},{\mbox{\boldmath$\theta$}})d{{\bf t}}d{\mbox{\boldmath$\omega$}}+ o_p(1 ) . \end{split}\ ] ] let @xmath240 $ ] , @xmath241 $ ] and @xmath242 $ ] .
we define @xmath243 . by the definitions of @xmath244 and @xmath245
, we have @xmath246 . by lemma 3 and applying integration by parts twice , @xmath247 where @xmath248 . by lemma 3 , @xmath249_{r , s}+o_p(1)\ ] ] uniformly over @xmath250
. therefore , @xmath251_{r , s}+o_p(1).\ ] ] similarly , applying integration by parts and by lemma 3 and 2(vi ) , we have @xmath252_{r , s}+o_p(1),\ ] ] @xmath253_{r , s}+o_p(1),\ ] ] @xmath254\times \dot{k}_{n , r}({{\bf t}},{\mbox{\boldmath$\theta$}})\dot{k}_{n , s}({\mbox{\boldmath$\omega$}},{\mbox{\boldmath$\theta$}})d{{\bf t}}d{\mbox{\boldmath$\omega$}}= [ { { \bf v}}({\mbox{\boldmath$\theta$}}_0)]_{r , s}+o_p(1).\ ] ] hence the right hand side of is @xmath255_{r , s}+o_p(1)$ ] , which gives . from and ,
@xmath256 .
from theorem , we know that @xmath257 and @xmath258 . by the mean value theorem , @xmath259_{r , s } & = [ \hat{\mbox{\boldmath$\sigma$}}_n(\hat{\mbox{\boldmath$\theta$}}_n , \hat{{\mbox{\boldmath$\sigma$}}}_n^{(1 ) } ) - \hat{\mbox{\boldmath$\sigma$}}_n(\hat{\mbox{\boldmath$\theta$}}_n , { { \bf d}}_0)]_{r , s } + [ \hat{\mbox{\boldmath$\sigma$}}_n(\hat{\mbox{\boldmath$\theta$}}_n , { { \bf d}}_0 ) - { { \bf d}}_0]_{r , s}\\ & = \left[\frac{\partial}{\partial{\mbox{\boldmath$\sigma$}}}[\hat{{{\bf d}}}_n]_{r , s}\bigg|_{{\mbox{\boldmath$\sigma$}}={\mbox{\boldmath$\sigma$}}^*}\right ] ' \times vech(\hat{{\mbox{\boldmath$\sigma$}}}_n^{(1)}-{{\bf d}}_0 ) + [ \hat{\mbox{\boldmath$\sigma$}}_n(\hat{\mbox{\boldmath$\theta$}}_n , { { \bf d}}_0 ) - { { \bf d}}_0]_{r , s } , \end{split}\end{aligned}\ ] ] where @xmath260 and thus @xmath261 . in view of lemma 4 and @xmath258 , @xmath262 . again by the mean value theorem , @xmath263_{r , s}= \left[\frac{\partial}{\partial { \mbox{\boldmath$\sigma$}}}[\hat { { { \bf d}}}_n]_{r , s}\bigg|_{{\mbox{\boldmath$\sigma$}}={\mbox{\boldmath$\sigma$}}^*}\right ] ' \times vech(\hat{{\mbox{\boldmath$\sigma$}}}_n^{(k)}-{{\bf d}}_0),\ ] ] where @xmath264 . then by lemma 4 and mathematical induction ,
we know that for any @xmath133 and @xmath140 , there exist @xmath265 and @xmath134 , such that for any @xmath135 and @xmath266 , @xmath267_{r , s}\big| \leq \eta \times \big|[\hat{\mbox{\boldmath$\sigma$}}_n^{(k)}-\hat{\mbox{\boldmath$\sigma$}}_n^{(k-1)}]_{r , s}\big| , \mbox { for all } k > k\right)>1-\epsilon,\ ] ] where @xmath268 .
note that the inequality inside the above probability implies that @xmath79 converges as @xmath269 and the limit @xmath137 satisfies @xmath270 and @xmath271 .
in this section , we extend the approach to the partial rank correlation ( prc ) criterion function @xmath272 , defined by ( 3 ) , of khan and tamer ( 2007 ) for censored data . under the usual conditional independence between failure and censoring times given covariates and additional regularity conditions , khan and tamer ( 2007 ) developed asymptotic properties for prce that are parallel to those by sherman ( 1993 ) .
the same self - induced smoothing can be applied to partial rank correlation criteria function to get @xmath273\phi\left(\sqrt{n}{{\bf x}}_{ij}'{\mbox{\boldmath$\beta$}}({\mbox{\boldmath$\theta$}})/\sigma_{ij}\right ) .
\end{split}\end{aligned}\ ] ] we define its maximizer , @xmath274 , as the smoothed partial rank correlation estimator ( sprce ) .
let @xmath275^{\otimes2}\right\},\ ] ] @xmath276\right\}^{\otimes2 } , \\
\end{split}\end{aligned}\ ] ] @xmath277^{-1}\times \hat { { \bf v}}_n^*({\mbox{\boldmath$\theta$ } } , { \mbox{\boldmath$\sigma$}})\times [ \hat { { \bf a}}^*_n({\mbox{\boldmath$\theta$ } } , { \mbox{\boldmath$\sigma$}})]^{-1},\ ] ] where @xmath278-\delta_i \times i[\tilde y_j > \tilde y_i]$ ] . based on @xmath279 , we have the following iterative algorithm to compute the sprce and variance estimate simultaneously .
( sprce ) 1 .
compute the prc estimator @xmath280 and set @xmath77 to be the identity matrix .
2 . update variance - covariance matrix @xmath281 .
smooth the partial rank correlation @xmath282 using covariance matrix @xmath283 . maximize the resulting smoothed partial rank correlation to get an estimator @xmath284 .
repeat step 2 until @xmath284 converge .
in addition to assumptions 1 - 3 , khan and tamer ( 2007 ) added the following assumption for the consistency of prce .
let @xmath285 be the support of @xmath286 , and @xmath287 be the set @xmath288 then @xmath289 .
similar to the rank correlation function , it can be shown that under assumptions 1 - 4 , and still hold for partial rank correlation function @xmath290 .
therefore , theorems 1 - 3 in section 2 continue to hold when replacing the point and variance estimators for smoothed rank correlation by the corresponding ones for the smoothed partial rank correlation .
specifically , for any positive definite matrix @xmath39 , under assumptions 1 - 4 , we have 1 .
the sprce @xmath291 is asymptotically equivalent to the prce @xmath27 in the sense that @xmath292 , and , therefore , @xmath293 where @xmath294 is the limiting variance - covariance matrix of @xmath27 .
variance estimator is consistent : @xmath295 .
algorithm 2 converges numerically in the sense that there exist @xmath131 , @xmath132 , such that for any @xmath133 , there exists @xmath134 , such that for all @xmath135 , @xmath296 the proofs are similar to those of theorems 1 - 3 in section 2 , and are , therefore , omitted .
in this section , we first apply the proposed self - induced smoothing method to analyze the primary biliary cirrhosis ( pbc ) data ( fleming and harrington , 1990 , appendix d ) and compare the result with that using the cox regression .
we then report results from several simulation studies we conducted using the method .
we applied smoothed prce to the survival times of the first 312 subjects with no missing covariates in the pbc data .
we included two covariates albumin and age50 ( age divided by 50 ) .
we reparameterized the transformation model ( 1 ) by setting @xmath297 as 1 , and estimated @xmath298 by sprce .
we also calculated prce for @xmath298 and fitted the standard cox model .
for the cox regression , the ratio @xmath299 is the estimate of @xmath298 .
the results are summarized in table 1 .
table 1 + regression analysis of pbc data [ cols="^,^,^,^,^,^",options="header " , ] we conducted simulation studies for a number of cases . in the first case ( design i )
, we generated @xmath2 from a bivariate normal distribution with mean @xmath300'$ ] and a covariance matrix diag@xmath301 .
we set @xmath302 $ ] and generated @xmath303 from the probability density function @xmath304 .
we set the transformation @xmath305 as @xmath306 .
this is indeed a weibull proportional hazard model .
the sample sizes were @xmath307 and the numbers of replications were @xmath308 .
the smrce , mrce and cox model were used to estimate @xmath309 , and the standard error of smrce was computed by algorithm 1 .
the mean(mean ) , bias(bias ) and root mean square error(rmse ) for each method as well as mean of standard error(se ) and coverage of @xmath310 confidence interval for the smrce are reported in table 2 .
the second case ( design ii ) is similar to the first one except that @xmath0 is censored by a random variable @xmath23 , which is independent of @xmath2 and normally distributed with mean @xmath311 and variance @xmath312 .
the sample sizes were @xmath313 and the numbers of replications were @xmath308 .
this design is similar to that in grgens and horowitz ( 1999 ) .
the sprce , prce and cox model were used to estimate @xmath309 , and the standard error of sprce was computed by algorithm 2 .
the resulting estimates are summarized in table 3 where we also report bias(bias ) , root mean square error(rmse ) , mean of standard error(se ) , and coverage of @xmath310 confidence interval . in the third case ( design iii ) , we generated @xmath314'$ ] by two steps .
we first generated @xmath315'$ ] from a bivariate normal distribution with mean @xmath316'$ ] and an identity covariance matrix .
we then generated @xmath317 as 0 or 2 with equal probability .
we set @xmath318 $ ] and generated @xmath303 from a normal distribution with @xmath319 and @xmath312 .
we set the transformation @xmath320 .
the sample sizes were @xmath321 and the numbers of replications were @xmath308 .
the smrce , mrce and least squared method were applied to estimate @xmath322 and @xmath323 , and the standard error of smrce was computed by algorithm 1 .
table 4 reports the mean ( mean ) , bias ( bias ) and root mean square error ( rmse ) for each method as well as mean of standard error ( se ) and coverage of @xmath310 confidence interval for the smrce . from tables 2 , 3 and 4
, we find that ( 1 ) the root mean squared error is close to the mean standard error for the smrce ( sprce ) ; ( 2 ) as the sample size increases , the bias reduces and the coverage of @xmath310 confidence interval converges to the nominal level . these show that the proposed variance estimator is accurate and algorithms 1 and 2 work well
this paper provides a simple yet general recipe for smoothing the discontinuous rank correlation criteria function .
the smoothing is self - induced in the sense that the implied bandwidth is essentially the asymptotic standard deviation of the regression parameter estimator .
it is shown that such smoothing does not introduce any significant bias in that the resulting estimator is asymptotically equivalent to the original maximum rank correlation estimator , which is asymptotically normal .
the smoothed rank correlation can be used as if it were a regular smooth criterion function in the usual m - estimation problem , in the sense that the standard sandwich - type plug - in variance - covariance estimator is consistent .
simulation and real data analysis provide additional evidence that the proposed method gives the right amount of smoothing .
because of the family of transformation models contains both the proportional hazards and accelerated failure time models as its submodels , the new approach may be used for model selection .
the specification test commonly used in the econometrics literature ( hausman , 1978 ) may also be used for testing a specific semiparametric model assumption .
in addition , the smoothed objective function also makes it possible to fit a penalized regression by introducing a lasso - type penalty .
the method and theory developed herein can easily be extended to other problems of similar nature , i.e. discontinous objective functions with associated estimators being asymptotically normal .
in particular , we can apply the self - induced smoothing to the estimator introduced by chen ( 2002 ) for the transformation function @xmath6 to obtain a consistent variance estimator .
lemma 1 below is due to sherman ( 1993 , theorem 2 ) .
we denote @xmath324 as general objective functions which are centered and satisfies the same regularity conditions as in sherman ( 1993 ) .
suppose @xmath325 is consistent for @xmath84 , an interior point of @xmath51 .
suppose also that uniformly over @xmath97 neighborhoods of @xmath84 , @xmath326 where @xmath54 is a negative definite matrix , and @xmath327 converges in distribution to a @xmath328 random vector .
then @xmath329 recall in theorem 2 , we define @xmath330 and its first and second partial derivatives with respect to @xmath20 as @xmath331 also recall @xmath332 then we have the following lemma . uniformly over @xmath333 , \(i ) @xmath334 + ( ii)@xmath335 . + ( iii)@xmath336
. + ( iv ) @xmath337_{r , s}+o(n^{-1}).$ ] + ( v)@xmath338 .
+ ( vi)@xmath339 .
+ ( vii)@xmath340 + @xmath341 + @xmath342 let @xmath343 , and divide its complement into @xmath344 and @xmath345 . we prove ( i)-(iv ) for @xmath346 . for @xmath347 ,
the proofs are similar and omitted . for ( i ) , note that @xmath348 since @xmath349 and @xmath350\geq0 $ ] , @xmath351 where the last equality follows from @xmath352 .
similarly , @xmath353 for ( ii ) , by definition , @xmath354 \bigg|_{t_r=1/\sqrt{n}}^{2\sqrt{\frac{\log n}{n } } } + \left[n t_r e^{-\frac{nt_r^2}{2}}\right ] \bigg|_{t_r=1/\sqrt{n}}^{0}\right ) = o(1 ) , \end{split}\end{aligned}\ ] ] where the inequality follows from @xmath355 . for ( iii ) , by definition , @xmath356 + @xmath357 , where the last equality follows from @xmath352 .
iv ) , by definition and applying integration by parts twice , @xmath358 where @xmath359 with @xmath360 entry being 1 , and the last equality follows from the gaussian tail probability . for ( v ) , we know , by definition , @xmath361 . by symmetry , @xmath362 where the inequality follows from @xmath352
similarly , @xmath363 for ( vi ) , by definition , @xmath364 where the second equality is due to symmetry and the third equality follows from @xmath352 . to prove ( vii ) , without loss of generality , we assume @xmath365 . we denote @xmath366 as @xmath192 and @xmath367 its complement .
then , @xmath368 where @xmath369 and @xmath370 are chosen from @xmath371 . then by ( v ) and ( vi ) , @xmath372 uniformly over @xmath373 such that @xmath374 and @xmath375 , we have @xmath376 = e\left[\frac{\partial\tau({{\bf u}},{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_r } \tau({{\bf u}},{\mbox{\boldmath$\theta$}}_0)\right]+o_p(1),\ ] ] @xmath377 = e\left[\frac{\partial\tau({{\bf u}},{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_r } \frac{\partial\tau({{\bf u}},{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_s}\right]+o_p(1).\ ] ] we sketch the main steps of the proof below .
first of all , observe that @xmath378 & -\frac{1}{n}\sum_{i=1}^n \left[\frac{\partial\tau({{\bf u}}_i,{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_r}\tau({{\bf u}}_i,{\mbox{\boldmath$\omega$}})\right]\bigg| \\ & \leq\frac{1}{n}\sum_{i=1}^n \left[\bigg|\frac{\partial\tau({{\bf u}}_i,{{\bf t}})}{\partial\theta_r } -\frac{\partial\tau({{\bf u}}_i,{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_r}\bigg| \times
\big|\tau({{\bf u}}_i,{\mbox{\boldmath$\omega$}})\big|\right ] \\ & \leq \frac{1}{n}\sum_{i=1}^n m_2({{\bf u}}_i)\times|{{\bf t}}-{\mbox{\boldmath$\theta$}}_0| , \end{split}\end{aligned}\ ] ] where @xmath379 is an integrable function .
the last inequality is due to assumption 3 and @xmath380 .
since @xmath379 is integrable , by the law of large numbers , the left hand side of above inequality is thus @xmath97 . by a similar argument
, we can show that + @xmath381= \frac{1}{n}\sum_{i=1}^n \left[\frac{\partial\tau({{\bf u}}_i,{\mbox{\boldmath$\theta$}}_0)}{\partial\theta_r } \tau({{\bf u}}_i,{\mbox{\boldmath$\theta$}}_0)\right]+o_p(1).\ ] ] by the law of large numbers , we get .
the proof of is similar .
let @xmath382 and @xmath383 be the same as those in .
then , for @xmath384 , we have @xmath385_{r , s}=o_p(1),\ ] ] @xmath386_{r , s}=o_p(1),\ ] ] where @xmath142 is a small neighborhood of @xmath115 and @xmath39 is a positive definite matrix .
we now extend the definition of kernels in lemma 2 for any covariance matrix @xmath39 as follows , @xmath387 where @xmath388 is the determinant of @xmath39 .
then the first and second derivatives of @xmath389 with respect to @xmath20 become @xmath390{\bf e_s}\\ & \qquad\qquad\qquad\qquad\qquad\times e^{-\frac{n}{2}({{\bf
t}}-\theta)'{\mbox{\boldmath$\sigma$}}^{-1}({{\bf t}}-\theta)}. \end{split}\end{aligned}\ ] ] we partition @xmath85 into @xmath192 and its complement @xmath391 , where @xmath392
. furthermore , we define @xmath393 note that @xmath394 is bounded for @xmath395 .
similar to the proofs of theorem 2 and lemma 2 , we can get @xmath396_{r , s}}{\partial { \mbox{\boldmath$\sigma$}}}= [ { { \bf a}}({\mbox{\boldmath$\theta$}}_0)]_{r , s}\int_{\tilde{\mbox{\boldmath$\omega$}}_{n , r } } \frac{\partial k_n({{\bf t}},{\textbf 0},{\mbox{\boldmath$\sigma$}})}{\partial { \mbox{\boldmath$\sigma$}}}d{{\bf t}}+o_p(1),\ ] ] @xmath397_{r , s}}{\partial { \mbox{\boldmath$\sigma$}}}= [ { { \bf v}}({\mbox{\boldmath$\theta$}}_0)]_{r , s}\int_{\tilde{\mbox{\boldmath$\omega$}}_{n , r } } \frac{\partial k_n({{\bf t}},{\textbf 0},{\mbox{\boldmath$\sigma$}})}{\partial { \mbox{\boldmath$\sigma$}}}d{{\bf t}}+o_p(1),\ ] ] uniformly over @xmath398 such that @xmath399 and @xmath400 . likewise , we have @xmath401 , which , combined with @xmath402 , implies @xmath403 this completes the proof . for @xmath384
, we have + @xmath404_{r , s}=o_p(1)$ ] . first , by theorem 2 , lemma 4 and the mean value theorem
, we can show that @xmath405_{r , s } = [ \hat{{{\bf a}}}_n({\mbox{\boldmath$\theta$ } } , { \mbox{\boldmath$\sigma$}})-\hat{{{\bf a}}}_n({\mbox{\boldmath$\theta$ } } , { { \bf d}}_0 ) + \hat{{{\bf a}}}_n({\mbox{\boldmath$\theta$ } } , { { \bf d}}_0)]_{r , s}=[{{\bf a}}({\mbox{\boldmath$\theta$}}_0)]_{r , s}+o_p(1)$ ] . by matrix differentiation , @xmath406 .
thus @xmath407 , where @xmath101 . the rest of the proof is straightforward and thus omitted .
[ thy : fast convergence ] for @xmath408 , we have @xmath409_{r , s}=o_p(1).\ ] ] the result follows immediate from lemma 4 and corollary 1 .
suppose @xmath221 is the joint density for @xmath410 and @xmath411 is the conditional density of @xmath412 given @xmath413 and @xmath414 .
suppose @xmath415 is the conditional density of @xmath416 given @xmath414 and @xmath417 is the conditional density of @xmath416 given @xmath413 and @xmath414 .
by change of variable , @xmath418 . therefore , where @xmath423 is the marginal distribution of @xmath0 .
therefore if the conditional density @xmath424 has bounded derivatives up to order three for each @xmath425 in the support of space @xmath426 , it is not difficult to show that assumption 3 is satisfied .
the sufficient condition can be easily verified in certain common situations such as when the conditional density @xmath427 is normal . | this paper deals with a general class of transformation models that contains many important semiparametric regression models as special cases .
it develops a self - induced smoothing for the maximum rank correlation estimator , resulting in simultaneous point and variance estimation .
the self - induced smoothing does not require bandwidth selection , yet provides the right amount of smoothness so that the estimator is asymptotically normal with mean zero ( unbiased ) and variance - covariance matrix consistently estimated by the usual sandwich - type estimator .
an iterative algorithm is given for the variance estimation and shown to numerically converge to a consistent limiting variance estimator .
the approach is applied to a data set involving survival times of primary biliary cirrhosis patients .
simulations results are reported , showing that the new method performs well under a variety of scenarios . , , + |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.